ESTERFIP, A TRANSESTERIFICATION PROCESS TO PRODUCE BIO-DIESEL FROM RENEWABLE ENERGY SOURCES

ESTERFIP, A TRANSESTERIFICATION PROCESS TO PRODUCE BIO-DIESEL FROM RENEWABLE ENERGY SOURCES A. Hennico, J. A. Chodorge and A. Forestikre INSTITUT FRANCAIS DU PETROLE RUEIL-MALMAISON (92500). FRANCE Keywords : Transesterification, Vegetable Oils, Bio-Diesel 1 - INTRODUCTION Vegetables oils and products synthesized from natural raw materials (either of vegetable or animal origin) are having a strong "come back" in the recent decades. One of the major reasons for the increased utilization of fatty chemicals for indusmal use has been the ability to tailor the products to specific needs. This trends is clearly indicated in Table 1 that gives an estimate of the world fat production in millions tons and in the case of vegetable oils, the yields per unit area (hectare) per year. End uses of upgraded products or derivative compounds are extremely numerous but usually highly specialized. Major areas of applications are : Food industry, soap and detergents, cosmetics, pharmaceuticals, textile and paper industry, oild field chemicals, fat based emulsifiers, synthetic lubricants, metal working fluids and last but not least introduction into the automative fuel sector. This last application will be the subject of this presentation. In the early days of diesel engines, vegetable oils were tested (their original compositions unchanged) as a possible motor fuel but the idea never took hold owing to incompatibility problems such as deterioration of the oil with time, high viscosity, and fouling of the engine. Recently the bio-diesel route has been reactivated for a number of xasons as outlined h- ereafter : It has been found that vegetable oil can be transformed via esterification into a product which is much more adequate as a diesel fuel than the original oil itself. A wide variety of vegetable oils can be used as raw material for transesterification; this has led to the idea that bio-diesel production could be a way to extend the role of agriculture (more jobs created and reduced financial burden for petroleum imports in developing countries, slow-down in the current reduction of cultivated surfaces for developed countries like those of the European community). 2 - THE ESTERFIP PROCESS DEVELOPED BY IFP' FOR THE TRANSESTERIFICATION OF VEGETABLE OILS Transesterification of natural glycerides with methanol to methylesters is a technically important reaction that has been used extensively in the soap and detergent manufacturing industry. IFP has done extension R and D work in the transesterification field with the aim of creating a product that would be suitable as an excellent substitute for djesel fuel. As a result, a new process called ESTERFIP was developed that allows the elimination of certain impurities from the product that otherwise would be detrimental to classical diesel engines. The ESTERFIP process was developed by IFP first on a laboratory scale, then tested in a pilot plant (1987) and demonstrated in a commercial plant that is operating satisfactorily since 1992 (capacity 20 000 t/yr). Originally the design was developed for batch operation which is very suitable for small capacities and then further upgraded to continuous operadon, an economically dictated choice for intermediate and large capacities. 2 - 1 Chemistry Involved The reaction of transesterification involves the reaction of methanol with the s glycerides of the rapeseed oil to form the corresponding methylesters and glycerine as indicated on the following reaction scheme : Jointly with Sofiproteol (France) 763 VegetableOil + Methanol -+ Esters + Glycerine (Triglyceride) This global stoichiometry is of course an oversimplification as we are in presence Of a three-step reversible reaction with di - and monogycerides as intermediate products. The reaction takes place in presence of a catalyst that is most commonly sodium hydroxide, potassium hydroxide or sodium methylate. In the case of bio-diesel manufacturing, the main objective is to achieve the maximum possible conversion towards methylester (in excess of 97 %). This aim puts certain specific constraints on the reaction scheme, such as long hold-up time or eventually unreacted feed components recycling. involving a difficult separation between reactants and product. Furthermore to avoid operating problems in the ESTERFIP process the vegetable oil used as feedstock should be partially refined to eliminate phospolipides, gummy substances, free acid and water. T- ypical feed specifications are : * Acidityindex: . 1 maximum The situation is also complicated by solubility problems. For example in the present case neither methanol is soluble in the starting material triglyceride nor the end products glycerine and fatty acid methyl esters are miscible, whereas methanol is soluble in fatty acid methyl esters. We can therefore expect different time dependent situations - at the beginning a two-phase system, followed by an almost complete solution. Then as soon as a considerable amount of glycerine is formed, a new two phase system will again prevail. 2 - 2 Composition of Fatty Acids in three common Vegetable Phosphorous content : 10 ppm wt maximum Water content : 0,l wt % maximum Oils Whereas in Europe methylesters from rapeseed oil and sunflower oil are the most common feedstocks for bio-diesel the US leans heavily upon soybean oil as raw feedstock. The Table 2 gives the composition of three of the most common renewable vegetable sources that are used in the preparation of bio-diesel. Although the feed composition is quite different, a careful selection of operating conditions (t. p) and amount of catalyst used permits the production of a bio-diesel that satisfies the most shingent specifications required by the automobile industry. It is however important here to stress the importance of experimental data checking and unit modeling based upon practical experience, before undertaking the conceptual design of a large size industrial unit. 2 - 3 ESTERFIP Process Description (continuous scheme) A complete block flow scheme is given on Figure 1. The sequence of processing steps is as follows: * Transesterification of the vegetable oil by dry methanol in presence of a basic catalyst. Decantation to completely separate methyl esters from glycerine. The ester phase is water-washed and purified in a continuous operation in order to eliminate the last traces of catalyst particles. This step is very critical to avoid harmful deposits during the combustion in the diesel engine. Vacuum evaporation of the methyl ester product to recover traces of methanol and water. The raw glycerine recovemd in the settler is evaporated (the main methanol removal step), neutralised, decanted to separate fatty acids, and finally completely freed from methanol. * 2 - 4 Overall Material Balance (Rapeseed Oil Case) Refer to Figure 2. 764 2 - 5 Product Properties Bio - Diesel (Methyl esters) Glycerine (by -product) Specific gravity 0.88 Glycerine content, wt % > 80 Cetane number 49 Other organic compounds, < 2.5 Flash point, O C 55 Ash content, wt % < 10 mini Wt % CFPP, O C - 12 Methanol content, wt % < 0.2 Viscosity (cSt 20°C) 1.52 Water content < 10 2 - 6 Bio-Diesel based Commercial Fuels in France In the diesel fuel application two main blends of methyl esters are currently commercialised i-n France, namely : A 5 % mixture of bio-diesel in conventional diesel which is for sale to the public in service stations (without distinctive labelling obligation) * 30 to 50 % mixtures of bio-diesel for use in bus fleets run by municipalities. The estimated tonnage of bio-diesel commercialised in France for the total year 1994 is 150,000 Tons. 2 - 7 Environmental Advantages of Bio-Diesel The main distinctive features of bio-diesel versus conventional diesel fuel are : * No sulphur * Noaromatics * Renewable energy. The engine emissions are sulphur free and the other exhaust components are given (on a comparative basis with conventional diesel) in Figure 3. Presence of oxygen in the molecular composition 3 - CONCLUSIONS Bio-diesel is at present the most attractive market among the non-food applications of vegetable oils. The different stages in the production of rapeseed methyl ester generate byproducts which offer further outlets. Oil cake, the protein rich fraction obtained after the oil has ken extracted from the seed is used for animal feed. Glycerol, the other important by-product has numerous applications in the oil and chemical industries such as the cosmetic. pharmaceutical, food and painting industries. New applications are under investigations. The bio-diesel market in the European Union has a very strong potential growth position due to special fiscal measures that are already applied in several counnies and under serious considerations in others. 765 TABLE 1 - ESTIMATED WORLD VEGETABLE OIL + FAT PRODUCTION I I Prduction(106T) (1) I Yield,metrict/ha soybean Rapeseed (canola, colza) per year 1980 1990 2000 14.4 16.9 23.2 0.2 - 0.6 3.4 8.1 10.7 1.5 - 2 IPaim 1 E 1 10.; 1 1;:; 1 5 - 8 I Sunflower 1 - 1.5 Coconut 3 .O 3 .O 3.3 3 - 4 Sesame 2.1 0.2 Others 11.4 12.7 15.3 Total 43.2 60.6 81.9 Animal fat 16.1 18.6 21.5 FATTY ACIDS mecyl ester oil methyl ester Raw& oil methyl ester Soybean oil methylester Sunflower TABLE 2 - COMPOSITION OF FATTY ACIDS AND METHYL ESTERS * C160 Palmitic % 5 IO 7 C18:O stearic 90 2 - - 4 4 CIS:] Oleic % 59 - - 23 C18:Z Linoleic % 21 - 53 65 CJ 8:3 Linolenic 9b 9 8 < 0,s c200 h h i d i C 9% c 0.5 c 0.5 - c 2 0 1 Gadoleic 90 1 < 0.5 < 0.5 C220 Behenic 90 < 0.5 - < 0,5 < 0,s c22: I Enric *Cx : y : hydrocarbon chain with X = a number of carbon atoms and Y the number of double bonds. (1) : A.J. Kaufman + R.J. Ruebusch, J. Amer. Oil Chemist's Soc. - Inform 1, 1034 (1990) 766 ! Figure 1 Esterfip Process-Block Diagram Methanol Salts Fatty Acids, Esters 1 Figure 2 Overall Material Balance (Rapeseed Oil Case) Figure 3 Exhaust Emissions Compared: Bio-Diesel vs Diesel 0Blo -dlesel Diesel 80 0 05 21 Carbon Monoxide H$$g& Nitrogen Oxides Eosch (CO), glkWh g,kWh (NOx). glkWh Smoke Index s''fur' wt% 767 INVESTIGATIONS ON REDUCING THE BENZO(A)PYRENE CONTENT OF COALTAR PITCH Janusz Zielinski, Blandyna Osowiecka Technical University of Warsaw Institute df Chemistry ul. Lukasiewiwa 17 09400 Plock, Poland Jerzy Polaczek Research Institute of Industrial Chemistry ul. Rydygiera 8 01-793 Warsaw, Poland George Gorecki Brent America, Incorporated 921 Shewood Drive Lake Bluff, IL 60044 Keywords: benzo(a)pyrene, coal-tar pitch. Introduction Bitumens, like coalderived tars and pitches, as well as petroleum asphalts, have been widely used in many branches of industry and economy [l]. A dramatic limitation of the application areas for bitumens of coal origin is currently observed, due to the carcinogenic action of some bitumen-wntaining polycyclic aromatic hydrocarbons, especially benzo(a)pyrene (BAP). This hazardous condition was the reason for shutting down plants involved in the coking of coal-tar pitch in Poland and Germany [2,3]. As a result, many research studies on decreasing BAP content in bitumen materials have been performed. According to literature reviews [4]. a considerable reduction in BAP content could be achieved by changing the conditions under which coal-tar pitch is manufactured, especially by decreasing the coal coking temperature [5]. Other workers [6,7] have attained lower BAP concentrations by modifying the pitch properties through oxidation, ' ultraviolet irradiation 181, or by extraction with low-boiling solvents [8,9]. Polymers not only improve the properties and applicability of bitumencontaining materials [l], but also can play an important role in decreasing their carcinogenicity. The current work studies how the properties of coal-tar pitch are affected by specific high molecular weight substances at elevated temperatures. Experimental The following materials were used: Polish coal-tar pitch (R & 8 softening point, 68.5OC; toluene insolubles, 17.2% w/w; BAP content, 1.83% wlw), suspension-grade polyvinyl chloride (PVC, molecular weight, 139,000; Fikentcher number, 66.9), polystyrene (PS, molecular weight, 304,000; Vicat softening point, 103OC), polyethylene terephthalate waste (PET) and unsaturated polyester resin (UPR, 40-50% styrene solution). The study was performed stepwise. In the first step, the pitch was heated at 150 to 43OoC for 6 h to determine the effect of temperature on the pitch properties. The procedure was executed both with and without removal of distillate. In the second step, the molten pitch was blended with the various polymers: with PVC from between 120 and 350°C for 0.5 to 4 h, with PS from between 240 and 35OoC for 0.5 to 4 h, with PET from between 260 and 350°C for 1 to 6 h, and with UPR at 16OoC for 3 to 5 h. The products were analyzed for softening and dropping points, penetration (temperature relationship), as well as for BAP content and the amount of toluene-insoluble material. The BAP content was determined using the UV-VIS spectroscopic method [lo]. 768 Results and Discussion The results (Table I) show that the structural changes in the heated pitch are demonstrated by a decrease in penetration and increases in softening point, dropping Point and toluene-insolubles content. Changes in these properties became substantial In systems whose temperature was greater than 38OoC. The observed decrease in BAP content from 1.83% to 1.48% wlw was not caused by its evaporation because no BAP was found in the distillate fractions. There was very little change in the BAP content for pitch mixtures heated at temperatures below 38OoC. As a result, the changes in BAP content in this temperature range can be explained only by chemical interactions between the polymer and the pitch. It has been found that homogeneous pitch-polymer blends can be obtained under the following conditions: , / t 1 - an anthracene oil or dibutyl phthalate-plastified PVC up to 10% wlw and below 1 3OoC, - PS up to 10% wlw and below 31OoC, - PET and UPR, each up to 30% wlw and below 26OoC, - UPR up to 30% w/w and at llO°C, and after subsequent crosslinking at 140 to 1 6OoC. An individual selection of blending parameters, however, was necessary for each polymer. Temperature was an especially important property. It can be assumed that the elevated temperature contributes to an increase in the amount of toluene-insoluble material. This is due to a simultaneous destruction of polymer molecules and the polycondensation of pitch components, which is also evidenced by an increase of softening point and a decrease of penetration. No correlation, however, between this occurrence and a change in BAP content has been obsetved. The largest reductions of BAP content were achieved with pitch-polymer blends containing either PET at 30%; UPR at 30%; or a system comprised of PVC at 4.76%, anthracene oil at 22.63% and butadiene-styrene copolymer latex at 4.76%. The corresponding decreases in BAP content were 72%, 80-90%, and 46%, respectively. Amounts of polyester additive and the effect on BAP content in coal-tar pitch are presented in Fig. 1. The polyester resin used in these compositions was modified additionally by initiators: naphthenate cobalt and hydroperoxide of methyl ethyl ketone. The substantial decrease in BAP content in the case of UPR modification was independent of crosslinking of the resin. The changes in BAP content are likely connected to some chemical interactions between the pitch and the polymer. It has also 8 been found that the applications, such as building industry [l]. plastified PVCcontaining pitches can be -used in many the manufacture of insulating and sealing materials for the This investigation was sponsored by the Scientific Research Committee and realized as Project No. 7 S203 009 05. References 1. J. Zielinski, G. Gorecki, "Utilization of Coal-Tar Pitch in Insulating-Seal Materials," ACS Division of Fuel Chemistry, Chicago, 1993, p. 927. 2. J. Jastrzebski, et. a/., Koks, Smola, Gaz, 1985, 30, 5. 3. G. Nashan, Erdol, Erdgas, Kohle, 1993,109,33. 4. J. Zielinski, et. a/., Koks, Smola, Gaz, in press. 5. W. Boenigk, J.A. Stadelhofer, "Coal-Tar Pitches with a Reduced Low Molecular PAH Content," 5th International Carbon Conference, Essen, 1992, p. 33. 6. E.A. Sukhorukova, et. a/., Koks, Khlm., 1984, 7, 36. 769 7. W.A. Lebedev, et. ai., Carbon, 1988,8,36. 8. G.K. Low, G.E. Batley, J. Chromat, 1987, 392, 193. 9. R. Rajagopalan, et. a/., Sci. Total Environm. 1983, 27, 1. 10. A. Labudzinska, et. a/., lCRl Annual Report, 1993. p. 84. Softening Dropping Penetration Toluene BAP Content Point (°C>Point ('C) (x 10.' m, 50'C) insoiubles (% Wh) (% w/w) Original Pitch 68.5 82.0 8.3 f 1.5 11.20 1.03- Plch after 6 h of healing without removal of distillate at CC) 150 12.0 85.5 9.3 f 1.4 18.04 1.81 250 78.0 88.0 4.5 f 1.1 18.88 1.82 300 75.0 87.5 5.3 t 0.5 20.50 1.77 350 77.0 88.0 5.3 f 0.6 23.68 1.19 380' I 83.01 97.01 1.3 f 0.5 1 27.841 1.71 Pitch after 8 h of heating with distillate removal at (" C) 350-400 I 88.01 102.51 - I 32.20 I 1.64 L400-430 I 111.01 130.01 - 52.591 1.48 " 1 Table I. Properties of thermally treated coal-tar pitch '4h. -*. in terms of 100 g of pitch. distributed into acetone-solubles (1.72%), acetone-insolubles (0.08%), and tolueneinsolubles (0.03%). 2.50 - 2.00 E 3: z 8 1.50 1.00 P 0.50 3 0.00 1.88 PAK PES5 PES-10 PES-15 PES-25 PES-27 PES30 Fig. 1. Benzo(a)pyrene content in coal-tar pitch modified by polyesters. PES-5 relates to a composition of coal-tar pitch containing 5% w/w polyester resin. 110 THE PRODUCTION OF CHARS BY SUPERCRITICAL FLUID EXTRACTION Edwin S. Olson and Ramesh K. S h a m University of North Dakota Energy & Environmental Research Center PO Box 9018 Grand Forks, ND 58202-9018 (701) 777-5000 Key words: Supercritical fluid extraction; SFE; chars ABSTRACT Novel techniques were explored for developing larger micropore structure in the chars prepared by supercritical fluid extraction of low-rank coals. Extractions were carried out with 2-butanone at various temperatures and pressures above the critical point, and experiments were performed to maintain the highsurfacearea char structure as the pressure was released. The temperatures and pressures were then brought down close to the critical point, and then the pressure was released very slowly while keeping the temperature constant. This aerogel method gave higher surface areas than the method in which temperature or pressure was abruptly lowered, but ultraporous materials were not obtained. The introduction of pillaring reagents under supercritical conditions to preserve the expanded pore structure was also attempted. These experiments were again only partially successful in increasing the surface area of the char. INTRODUCTION Supercritical fluid extraction (SFE) of volatile material from coal offers an alternative to coal pyrolysis for production of chars. Previous efforts with low-rank coals at the University of North Dakota gave chars with relatively low surface areas. However, x-ray scattering experiments in an aluminum-beryllium high-pressure high-temperature extraction cell showed that very large surface areas ( > 2000 mz/g) are present during SFE of Wyodak subbituminous coal with an organic solvent, but the pores collapse during the reversion back to subcritical conditions (I). New techniques were explored to attempt to maintain the ultraporous structure that develops in the low-rank coals under supercritical conditions. Following supercritical solvent extraction of some of the coal material, attempts were made to stabilize the highly porous structure so that it did not undergo the collapse normally observed when the pressure is brought back to ambient. The techniques involve careful release of pressure at the critical point of the solvent as in the preparation of aerogel precursors and introduction of a stabilizing agent under pressure with a high-pressure liquid chromatograph (HPLC) injection device. The stabilizing agents were boron, silicon, and titanium compounds that could decompose to oxide clusters which could pillar the micropore structure. EXPERIMENTAL Wyodak (Clovis Point) subbituminous coal, Gascoyne (Kmfe River) lignite, and Velva lignite were used for the supercritical extractions. These coals were ground to -60-mesh size and dried in an oven at ll0"C for several hours. The samples were then stored under argon in plastic containers until used. 2- Butanone and ethanol were used as solvents. Tetraethyl orthosilicate (TEOS), titanium tetraisopropoxide (TIP), and tributyl borate (TBB) were added to the coal to stabilize the micropores generated during extraction. An HPLC column (Supelco, 250-mm long, 8.5-mm i.d. X 12.5-mm 0.d.) was used for supercritical extraction of coal because it could withstand the high pressure and temperature (up to 2500 psi and 350"C, respectively). The supercritical fluids (2-butanone or ethanol) were introduced into the stainless steel reactor via an ISCO LC-5000 syringe pump (ISCO, Lincoln, NE, USA), an injector (Rheodyne, Cotati, CA, USA), and a 2-m long (1/16-in.-o.d. x 0.02-in.4.d.) stainless steel preheating coil. The reactor and the preheating coil were placed inside a gas chromatograph (GC) oven (Varian, Aerograph series 1400 GC) to control the extraction temperature. A fluid flow rate of approximately 1-2 mL/min (measured at the pump) was achieved using a needle valve and a I-m X 0.1-mm silica capillary restrictor attached to the outlet of the extraction tube. The reactor was packed, with 5 g of desired coal and placed in the oven. After the extraction apparatus was assembled, the reactor was filled with 5 mL of the solvent under static conditions (no flow out of the cell) while the oven was heated to desired temperature. The dynamic extraction (constant fluid flow) was then started and was continued for the desired time period. The extract was collected in an Erlenmeyer flask placed in a hood. At the end of the extraction, solvent flow was stopped, and residual solvent in the reactor was slowly released (requiring about 10 min.). Thereafter, the oven was cooled to ambient temperature. and the reactor was detached from the extraction line. The residue from the reactor was collected, dried at IIO°C, weighed, and analyzed for surface area using American Society for Testing 171 and Materials (ASTM)-D4607 (iodine number) and by the percent iodine sorption method used by Sutcliffe c o p . RESULTS AND DISCUSSION Effects of Process Variables Supercritical extraction of Wyodak coal with 2-butanone at 350°C (980 psi) for 5 min followed by extraction at 265°C for 25 min (640 psi) gave a char with a relatively low iodine number (IN) of 177 mg/g, when the temperature and pressure were dropped to ambient immediately after the extraction time. This value is just a little higher than that of the original coal (162), and indicates that the pores collapse rather quickly as a result of capillary movement of metaplast material, even at this relatively low temperature. Only 10% of the coal was extracted or volatilized in the experiment. The experiment performed under similar conditions, but with a very slow pressure release at constant temperature (265"C), gave a char with significantly higher area (IN = 243), although the amount of material extracted was about the same (8%). Further improvements in the surface area were obtained by increasing the initial extraction period at 350°C to 20 and 40 min before dropping the temperature and pressure to 265°C and 640 psi. By maintaining the temperature while slowly releasing the pressure, chars with INS of 267 and 309, respectively, were obtained, and extraction yields of 12% for both runs were obtained. The 30-min extraction at 350°C (IO00 psi) followed by slow pressure release at 265°C gave a char with an intermediate surface area (IN = 283) and the same yield of 12%. Thus, the surface area appears to be directly related to the extraction time at 350°C, but the time at 265°C prior to slow pressure reduction may not be important. At a somewhat higher pressure (1250 psi) and higher solvent flow rate (2 mllmin), the 350°C. 30-min experiment gave a higher extraction (16%), but a lower area (IN = 254) was obtained. Although SFE yields are usually greater at the higher pressures (I), the surface area generated in the char is not directly related to the extract yield. Experiments conducted with Gascoyne lignite gave chars with generally higher surface areas than those from the Wyodak subbituminous coal. When Gascoyne was extracted for 30 min at 350°C and subjected to rapidly decreasing temperature and pressure, the resulting char had an IN of 256. The corresponding experiment at 350°C (1250 psi) with slow pressure release gave a char with the IN = 361 and a similar extraction yield (12%). Increasing the pressure during the extraction (2500 psi) gave a higher extraction as expected (19%), and the IN of the char was again lower (301). Another solvent, ethanol, was also investigated. Extraction with ethanol at 350°C (1500 psi) with slow pressure release gave a low extraction (8%) and a low surface area (IN = 137). Previous work demonstrated that the char surface is highly alkylated during SFE in alcohol (2). The alkylated metaplast may have a lower viscosity and undergo more extensive collapse. A trial with the high-calcium Velva lignite gave a lower-area char (IN = 323) than the Gascoyne lignite under similar conditions (35OoC, 1250 psi), although a higher extraction yield was obtained (25%). This could be attributed to increased solubility of the decomposing coal materials (metaplast) because of calcium-catalyzed decarboxylation. Normally, only partial decarboxylation occurs at 350°C. Effects of Pillaring Additives To stabilize the high surface areas that develop during SFE, solutions of various alkoxides were introduced under supercritical conditions following the extraction. It was anticipated that the alkoxides would decompose on the coal surface to form metal oxide clusters that would serve as stabilizing pillars to keep the pores from collapsing. Three of these organometallic agents were investigated for their effects in modifying the porosity of the supercritical chars. Addition of TEOS to char produced by SFE of Wyodak coal at 350°C for 5 rnin (1050 psi) gave a modified char with a higher surface area (IN = 293) than that produced without the TEOS (IN = 243). Titanium isoproxide addition under the same conditions gave a slightly lower area char (IN = 238). Addition of TEOS to the char obtained by extraction of Wyodak at 350°C for 20 rnin also gave a modified char with higher area (IN = 281). but this showed less of an increase. When less TEOS (113 of the previous amounts) was added to the 20-min SFE char, the increase in area was greater (IN= 297). When TEOS and TIP were added to Wyodak extracted for 30 rnin, the INS were similar to those for the 20-rnin runs. Addition of TBB to the 30-min char gave a significantly higher area char (IN = 328). Similar experiments with Gascoyne lignite were inexplicably not effective in promoting the surface area and, instead, decreased it substantially. Tributyl borate gave a char with IN = 266, compared with the original at IN = 361. Addition of a thiol to capture radicals generated during thermal reactions of the coal also gave a low-area char. The chars produced by this treatment still contain substantial amounts of coal "volatile" material that can be released by further heating at higher temperatures. Devolatilization of the supercritical chars at 750°C and 30 min gave carbons with very low surface areas, however. 772 CONCLUSIONS Several modified chars were prepared by SFE of low-rank coals to develop a large micropore strucnire. Pressure was released slowly at the supercritical temperature to maintain a more porous St~Clure. Tetraethylorthosilicate, titanium isoproxide, and tributyl borate were introduced under the Supercritical conditions to attempt to stabilize the micropore structure by forming pillaring clusters. v ACKNOWLEDGMENTS The support of the U.S. Department of Energy is gratefully acknowledged REFERENCES . 1. Olson, E.S.; Diehl, J.W.; Home, D.K.; Bale, H.D. Prepr, Pap.-Am. Chem. Soc., Div. Fuel Chem. 1988, 33, 826. 2. Olson, E.S.; Swanson, M.L.: Olson, S.H.; Diehl, J.W. Prepr. Pap.-Am. Chem. SOC., Div. Fuel Chem. 1986, 31, 64. Y 773 Table 1. Extractions of Wyodak Reanion Conditions Flow, Temp., Time, Pressure, coal Yield, %' Solvent mWmin 'C min psi Pw IN wyodak 10 2-BU' 1 350 5 980 Fast 177 265 25 640 265 25 620 WYW 8 2-Bu I 350 5 920 Slow 243 W Y U 12 2-Bu I 350 ul 980 Slow 267 365 10 630 WY- 12 2-Bu 1 350 40 la00 Slow 309 365 20 640 wyodak 16.2 2-Bu 2 350 30 1250 Slow 254 WY& 8 EIOH' 1 246 30 la00 Fast 137 ' Extraction w. coal (mf) - wt. char (mf)/w. coal (mf) x 100. mf refers to moisture free. ' 2-Bumone. ' Ethanol. Pressure drop. Table 2. Extractions of Gascoyne Reaction Conditions Flow, Temp.. Time, Pressure. Coal Yield, %' Solvent mWmin 'C min psi PD' IN Gascoyne 13.5 2-gU3 1 350 30 1250 Fast 256 Gascoyne 13.2 E ~ O H ~ I 350 30 1500 Slow 280 Gascoyne 11.6 2-Bu I 350 30 1250 Slow 361 Gascoyne 19.3 2-Bu 1 3% 30 2% Slow 301 ' Extraction wt. coal (mf) - wl. char (mf)/wt. coal (mf) X 100. mf refers to mOiSNre free. ' Pressure drop. ' 2-Butanone. ' Ethanol. Table 3. Extractions of Wyodak with Stabilizer Addition' conditions Temp., Time, Pressure. Additive coal Yield, % "C min psi (4) IN wyodak 8 350 5 920 None 243 265 25 620 Wycdak 13.4 350 5 1050 TEOS 293 265 25 620 (300) wyodak 8 350 5 265 25 wyodak I2 350 20 265 10 wyodak 12 350 20 265 IO Wy& 12 350 20 265 10 Wycdak 12.5 350 30 wyodak 13 33.3 30 Wyodak 13.8 350 300 920 650 980 630 loo0 620 980 640 Loo0 loo0 500 1200 TIP 238 (300) None 267 TEOS 281 (m) TEOS 297 (300) None 283 TEOS 299 TIP 28 1 wyodak 9 350 30 1250 TBB 328 ' Solvent = 2-butanone, flow rate = 1 mllmin. pressure drop = slow. Table 4. Extractions of Gascoyne with Stabilizer Added' Reaction Conditions Temp., Time, Ressure, Additive Coal Extraction, I 'C min psi ( t L ) IN Gascoyne 11.6 350 30 1250 None 361 G~WYW 14.4 350 30 1250 TBB (300) 266 Gascoyne 10.2 350 30 1250 p-Thiocresol (300) 259 ' Solvent = methylethyl ketone. flow rate = I mUmin. STRUCTURAL AND THERMAL BEHAVIOR OF COAL COMBUSTION AND GASIFICATION BY-PRODUCTS: SEM, FTIR, DSC, and DTA Measurements P. S. Valimbe', V. M. Malhotra', and D. D. Banerjee'. 1, Department of Physics, Southern Illinois University, Carbondale, Illinois 62901-4401 2. Illinois Clean Coal Institute, Carterville, Illinois 6291 8-0008. Keywords: Coal combustion residues, scrubber sludge, thermal and spectroscopic characterization ABSTRACT The pulverized coal combustion fly ash, fluidized bed combustion fly ash, fluidized bed combustion spent bed ash, and scrubber sludge samples were systematically characterized using scanning electron microscopy (SEM), differential scanning calorimetry (DSC), differential thermal analysis (DTA), and transmission Fourier transform infrared (FTIR) techniques. Our spectroscopic results indicated that the scrubber sludge is mainly composed of a gypsum-like phase whose lattice structure does not exactly match either conventional gypsum (CaS0,.2&0) or hannebachite (CaSO,.OS%O). SEM images suggested that unlike PCC fly ash particles, which were mainly spherical, the FBC fly ash and FBC spent bed ash particles were irregularly shaped and showed considerable fusion, FE3C fly ashes were mainly composed of anhydrite, lime, portlandite, calcite, hematite, magnetite, and various glass phases. The DTA and DSC data presented evidence implying that the PCC fly ash is thermally stable at 30°C < T < 1100°C. However, this was not the case for FBC ashes. INTRODUCTION More than 800 million tons of coal per year are burned in the United States, producing approximately 10 % of the coal burned as combustion residues in the form of solids. These solids, which are largely noncombustible, are classified as "fly ash" and "bottom ash". The fly ash particles are fine materials which are mostly captured in precipitators and in bag houses. The bottom ash term is used for those materials which settle or flow as melt to the bottom of the boiler. If the boiler is designed to use pulverized coal, then the coal combustion by-products are called "pulverized coal combustion" (PCC) fly ash and PCC bottom ash. The midwestern USA coals are high in sulfur content. The suhr in coal is in the form of inorganic minerals (chiefly pyrite) and is also organically bound Therefore, environmental concerns require that the sulfur content of the coal be reduced if this abundant resource is to be continuously utilized. A two prong approach is being developed to mitigate the sulfur problem. In the first, physical, chemical, and microbiological coal cleaning techniques have been and are being developed to reduce the sulfur content of midwestern coals. In the second, technologies have been developed and are being perfected to capture sulfur-containing combustion gases during coal combustion. One such clean coal technology is fluidized bed combustion (FBC)','. The advantage of the FBC technology is that it affords a large reduction of SO, from the combustion gases. The sorbents, like calcium carbonate (CaCO,) and calcium oxide (CaO), are injected along with the coal into FBC combustor. As SO, is produced, it reacts with the sorbent and is captured in the form of anhydrous calcium sulfate (CaSO,)'.'. There are also reports in the literature which suggest the formation of sulfides4. Just like for conventional combustors, two types of solid residues are produced, e g , FBC fly ash, which leaves the combustor at the top, and FBC spent bed ash, which is left at the bottom of the combustor. Wet scrubber processes are extensively used in flue gas desulfurization (FGD) technology. The major waste products produced are gypsum, calcium sulfite (CaSO,), fly ash, and excess reagent?. Calcium sulfite may be oxidized to calcium sulfate which in combination with water forms gypsum, It is generally believed that the calcium sulfate purity of residues from wet scrubber technology using lime or limestone ranges between 95 YO to R? %. It is estimated that by the turn of century about 200 million tons of coal combustion residue will be produced annually. With the current cost of residue disposal expected to rapidly escalate, the economic stakes for the coal utilization industry are substantial. Consequently, the technologies which can convert combustion residues into high value, but economically sound, materials are of utmost importance. Presently, only about 25% of the combustion residues generated are utilized', with the rest going to landfill or surface impoundments. Therefore, efforts are underway to find alternative usage69 of the combustion residues, e g . , ultra-lightweight aggregates for insulation industry, Portland cement-based FBC mixes, highway and street construction, construction bricks or tiles, roofing or paving tiles, pipe construction, and ashalloys. We have recently initiated research in our laboratory in which we are attempting to form advanced composite materials from coal combustion residues obtained from Illinois utilities. However, the successful utilization of coal 776 combustion and gasification residues requires a thorough physical and chemical characterization of these ashes, EXPENMENTAL TECENIQUES For our characterization studies, we examined four samples, i.e., PCC fly ash (Baldwin), FBC fly ash (ADM), FBC spent bed ash (ADM),a nd scrubber sludge (CmP ) . The residue samples were obtained from the sample bank established at the Mining Engineering Department of Southern Illinois university at Carbondale. The magnetic content of PCC fly ash, FBC fly ash, and FBC spent bed ash Was extracted from the as-received ashes by applying a magnetic separation technique. Microscopic studies of the coal combustion residues were accomplished using a Hitachi S570 scanning electron microscope. The samples were mounted on the SEM sample stubs using sticky tabs. The mounted samples were then cured at 60°C for 24 hours to ensure that the ash particles would not detach from the stub while under the electron beam. Mer curing, the samples were sputter coated with 40 nm of gold layer to help eliminate the problem of sample charging. The SEM data were collected using an accelerating voltage of 20 kV, except for FBC spent bed ash whose SEM pictures were acquired at an accelerating voltage of 10 kV to reduce sample charging. The structural characteristics of the combustion ashes and scrubber sludge were probed by recording their FTIR spectra. We used KBr pellet technique to collect the infrared spectra on a IBM IR44 FTIR spectrometer. Since the as-received scrubber sludge sample was wet, i.e., had substantial amount of moisture in it, the sludge was dried at 100°C prior to making its KBr pellets. One hundred scans were acquired at a 4 cm" resolution. . The thermal behavior of coal combustion residues and scrubber sludge were obtained using DSC and DTA techniques. The DSC data were recorded on PCC fly ash, FBC fly ash, FBC spent bed ash, and scrubber sludge using a well calibrated'"12 Perkin-Elmer DSC7 system interfaced with a 486 PC computer. The procedures adopted for the calibration of the temperature and of the specific heat have been described elsewhere',. Our calibrated DSC system had a temperature precision of k 1 K. The thermal characteristics of the residues using DSC technique were ascertained at 30°C < T < 600°C. We used a heating rate of 2O0C/min under a controlled N, purge environment (30 cm3/min) to collect our DSC data. The thermal stability of fly ashes, spent bed ash and scrubber sludge at 50°C < T < 1010" was examined by acquiring DTA data using a Perkin-Elmer DTA7 system. The samples were heated from 50T to 1000°C under a nitrogen gas environment. The heating rate used was 20"C/min. RESULTS AND DISCUSSION Microscopic Studies: Figures 1, 2, 3, and 4 reproduce the SEM micrographs of PCC fly ash, FBC fly ash, FBC spent bed ash, and scrubber sludge, respectively. The PCC fly ash particles were mainly composed of spherically-shaped particles whose sizes ranged from 0.2 mm to 15 mm. The spherical particles were usually hollow. It should be noted from Fig. 1 that small spherical particles of PCC fly ash were attached to bigger fly ash particles giving the appearance of agglomerates. Our SEM data on PCC fly ash did show some irregularly shaped particles in the ash, but predominantly particles were spherical. From the SEM micrographs of FBC fly ash, it appears that this ash had small particles of the range 0.1 mm to 1 mm, which had fked together to form agglomerates of the size ranging from 2 mm to 100 mm. Our SEM micrographs also indicated that the FBC fly ash contained very little spherical particles unlike PCC fly ash. The lack of the presence of spherical particles in FBC fly ash may be due to the lower combus!ion temperatures'.' for FBC combustor (around 850'C) than for PCC combustor (around 1150°C). The microscopic analysis of the FBC spent bed ash exhibited three distinct types of particles in this ash material. The first type of particles had a smooth surface to which smaller particles @e., 2 nun - 10 mm) were fused. These smooth particles lacked any pore structure. The second type of particles showed varying shapes and sizes but generally was around 750 mm. The third type of particles in this ash had a glass-like structure, These particles had an extensive pore structure, as can be seen in Fig. 3, and their sizes ranged from 250 mm - 300 mm. Figure 4 reproduces the SEM micrographs of scrubber sludge particles which were dried at room temperature. Generally, the sludge particles had a whisker-like shape, ranging from 50 mm to 400 mm in length, and were about 50 mm thick. In addition to the whisker-like particles, the sludge had some agglomerated parficles whose average size was about 100 mm. Thermal Behavior: The thermal stability of the PCC fly ash, FBC fly ash, FBC spent bed ash, and scrubber sludge was probed by recording their DSC and DTA data. The main thermal events observed from our DSC results are summarized in Table 1. The high temperature thermal stability of the combustion residues was ascertained by collecting DTA data at 50°C - 1000°C. We summarize our DTA results in Table 2, and Fig. 5 depicts typical DTA curves obtained from the combustion residues and scrubber sludge. The thermal data can be summarized as follows: (a) The PCC fly ash I17 Sample PCC Fly Ash FBC Fly Ash FBC Spent Bed Ash Scrubber Sludge The additional endothermic event at 674°C for FBC fly ash strongly suggested the presence of hematite (a-Fe,O,) in this ash. It should be noted from Fig. 5 that this thermal event was absent from the spent bed ash's DTA curve. The weak endothermic peak could be assigned to the magnetic transformation of hematite". (c) The DSC and DTA curves for the scrubber sludge showed a strong endothermic peak at 180°C and a weak exothermic peak at 380°C. The endothermic peak at 180°C suggested the dehydration ofthe gypsum, i t . , CaS0,.2qO+ CaSO, + 2qO (vapor). From their thermogravimetric experiments, Dorsey and Bueckerl' suggested the presence of calcium sulfite in their sample of scrubber sludge. They reported weight loss at 408°C < T < 452°C from their sample and associated this weight loss with the dehydration of hemihydrate (CaSO,.CaSO,. 1/2H,O), is., Thermal Event Temperature ("C) % Weight Loss on Heating the sample to 580°C 4.4 Endothermic 420 2.7 Endothermic 420 1 Endothermic 141 17.2 Endothermic 176 Exothermic 380 CaSO,. CaSO,. 1 /2%0+ CaSO,. CaSO, + 1 /29O(vapor) As listed in Table 2, the exothermic peak at 380°C for our scrubber sludge sample began at around 348°C and terminated at around 493°C. Therefore, one may argue that the exothermic peak at 380°C could be assigned to the dehydration of hemihydrate. The FTR spectrum of our scrubber sludge sample did not show any oscillators at 970 and 945 cm.' due to sulfite ions. Moreover, dehydration should produce an endothermic peak. It has been reported in the literatureI6 that on heating gypsum it undergoes a polymorphous transformation at 370°C < T < 460°C which results in a weak, exothermic peak . Therefore, we assigned the exothermic peak at 380°C for our scrubber sludge to this polymorphous transition. Spectroscopic Characterization: The spectroscopic studies of various combustion residues were undertaken to characterize the mineral and glass phases of the PCC fly ash, FBC fly ash, FBC spent bed ash, and scrubber sludge. In Fig. 7 we have reproduced the transmission-FTIR spectrum of scrubber sludge particles which were air dried prior to recording their spectrum. Three very strong bands were observed at 1154, 1126, and 1105 cm". In addition, a doublet having frequencies 662 and 602 cm-l was observed. In the water's stretching region, two distinct vibrational modes could be Seen at 3617 and 3559 cm.'. In the water's bending region only a single oscillator was observed at 778 \ I 1620 Cm', It is generally believed that the FGD residue, e.g., scrubber sludge, contains calcite (taco,), hannebachite (C&O3.O.54O), gypsum (CaS0,.2&0), quartz (SiOJ, and troilite (FeS). The absence of any vibrational bands below 450 cm" led us to discount the presence of troilite in our sample. Since we did not observe any band in the FTIR spectrum of the scrubber sludge at around 1430 Cm", we could also rule out the presence of calcite particles in our sludge sample. The argument that quartz may be present in our sample was discarded because the diagnostic bands for it at around 1050 and 472 cm'l were not observed in our FTIR spectrum. However, our tranSmission-FTR data did suggest the presence of gypsum. The vibrational bands at 1154, 1126, and 1105 cm-' could be assigned to v3 of sulfate of gypsum, while the oscillators at 662 and 602 cm-' could be attributed to v, of sulfate ions. The presence of two vibrational modes in the water's stretching region implied that there are two types of hydrates in our scrubber sludge. A comparison Of a Commercially available gypsum's FTIR spectrum, see Fig. 7, with our scrubber sludge spectrum indicated that gypsum formed in the FGD residue had a lattice structure which was different from that of commercial gypsum. It is worth pointing out that the FTIR spectrum of bassanite (CaSO,.0.5~0) shows a vibrational mode at about 3615 cm" and we observed a band at 3617 cm.'. However, we could not assign 3617 cm.' band ro bassanjte because the accompanying water band at 3465 cm.' was absent in our spectrum. We also ruled out the presence of hannebachite because of two reasons, i.e., (a) we did not observe the expected strong bands at 975 and 940 cm" of SO, ion in our FTIR spectrum of the scrubber sludge, and (b) we did not see any rectangular crystals, which could be associated with hannebachite, in our SEM images of the sludge. In TABLE 2 The Thermal Characteristics of the Combustion Residues as determined by DTA at 50°C < T < 1100°C. 275 348 493 Peak Temperature CC) 155 185 574 438 674 205 442 180 380 view of the discussion presented above we argue that scrubber sludge is mainly composed of gypsum. However, its lattice structure is not identical to the lattice structure of conventional gypsum. The transmission-FTIR spectrum of PCC fly ash, FBC fly ash, and FBC spent bed ash is reproduced in Fig. 7, and the observed frequencies are listed in Table 3. Based on the observed FTIR spectrum, it is argued that PCC fly ash is largely composed of various oxides. The strongest bands in our transmission-FTIR spectrum of PCC fly ash originated from quartz. The transmission-FTIR spectrum of as-received FBC fly suggested the presence of quartz , anhydrite (CaSO,), lime (CaO), portlandite (Ca(OH)J, calcite, hematite (Fe,O,), magnetite (Fe,O,), and glass phases. From the observed infrared frequencies of F8C spent bed ash, which are listed in Table 3, the following minerals have been identified, i.e., anhydrite, lime, portlandite, calcite, periclase, hematite, and magnetite. It is also generally reported that spent bed ash contains Cas. The formation of Cas is believed to occur for circulating fluidized bed combustion (FBC) via the following reaction, i.e., CaO + 4s 3 Cas + H,O. However, it is difficult for us to confirm the presence of Cas in our FBC spent bed a h as Cas produces no infrared bands. ACKNOWLEDGMENTS This research was supported by grants made possible by the U. S. Department of Energy Cooperative Agreement Number DE-FC22-92PC9252 1 and the Illinois Department of Energy through the Illinois Coal Development Board and the Illinois Clean Coal Institute. Neither the authors nor the U. S. Department of Energy, Illinois Clean Coal Institute, nor any person acting on behalf of either: (A) Make any Warranty of representation, express or implied, with respect to the accuracy, completeness, or usefulness of the information contained in this report, or that the use of 719 any information, apparatus, method, or process disclosed in this report may not infringe privately-owned rights; or (B) Assume any liabilities with respect to the use of, or for damages resulting from the use of, any information, apparatus, method or process disclosed in this report. References herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the U. S. Department of Energy. The views and opinions of authors expressed herein do not necessarily state or reflect those of the U. S. Department of Energy. REFERENCES 1. M. Valk, "Fluidized Bed Combustors" in 'Fluidized Bed Combustion', M. Radovanovic, Ed., 2. L. Yaverbaum, 'Fluidized Bed Combustion of Coal and Waste Materials', Noyes Data Co. (1977). 3. R. J. Collins, J. Testing and Evaluation 8, 259 (1980). 4. E. E. Berry, R. T. Hemmings, B. J. Cornelius, and E. J. Anthony, "Sulphur Oxidation States in Residues From a Small-Scale Circulating Fluidized Bed Combustor" in 'Fly Ash and Coal Conversion By-products Characterization, Utilization and Disposal V', R. T. Hemmings, E. E Berry, G. J. McCarthy, and F. P. Glasser, Eds., Materials Res. SOC. Procd. 136, 9 (1989). Inter. Pittsburgh Coal Conf., S-H Chiang, Ed., pp 561-566 (1993). Hemisphere Publishing Co., Washington (1 986). 5. L. B. Clarke, "Management of FGD Residues: An International Overview", Procd. 10th Annual 6. D. Golden, EPRI Journd, JanuaryEebruary, pp 46-49 (1994). 7. C. A. Holley, Preprints, Am. Chem. SOC. Fuel Div. 36(4), 1761 (1991). 8. 0. P. Modi, A. K. Singh, A. H. Yegneswaran, and P. K. Rohatgi, J. Materials Science lett. 11, 9. S. Ciby, B. C. Pai, K. G. Satyanarayana, V. K. Vaidyan, and P. K. Rohatgi, J. MaterialsEng IO. S. Jasty, P. D. Robinson, and V. M. Malhotra, Phys. Rev. B 43, 13215 (1991). 11. R. Mu, and V. M. Malhotra, Phys. Rev. B 44,4296 (1991). 12. S. Jasty, and V. M. Malhotra, Phys. Rev. B 45, 1 (1992). 13. R. C. Mackenzie and G. Berggren in 'Differential Thermal Analysis. Volume 1: Fundamental 14. 0. Chaix-Plucher, J. C. Niepce, and G. Martinez, J. Materials Sci. Lett. 6, 1231 (1987). 15. D. L. Dorsey and D. Buecker, Res. & Develop., May (1986). 16. D. N. Todor, 'Thermal Analysis of Minerals', Abacus Press, Kent (1976) TABLE 3 This table summarizesthe observed infrared bands for PCC fly ash, FBC fly ash, and FBC spent bed ash. The observed frequencies are in cm". 1466 (1992). Performance 2,353 (1993). Aspects', R. C. Mackenzie, Ed., Academic Press, New York (1970). PCC Fly Ash I FBC Fly Ash I FBC Spent I Comments 1 . Assignment Bed Ash 3,642 3,448 1,63 1 1,072 794 778 694 613 593 560 462 3,642 3,462 1,449 1 I44 1 1 1 1 1,011 885 795 68 1 616 602 515 462 3,642 3,448 1,624 1,448 1154 1122 945 920 680 615 605 595 560 462 sharp 0-H stretch [Ca(OH),] broad 0-H stretch, adsorbed water sharp, weak H-0-H bend of water broad, medium asymmetric C0;stretch broad, strong SO;' stretch broad, strong Si-0 stretch broad, weak CaSO, sharp, weak 0-H bend [Ca(OH),] sharp, weak CaCO, [CaGO,I broad, strong [Cas041 sharp, medium sharp, weak sharp, weak sharp, weak sharp, weak sharp, weak sharp, weak broad, weak broad, weak broad, medium quartz quartz quartz Anhydrite [CaSO,] Anhydrite [CaSO,] Anhydrite [CaSO,] Anhydrite [CaSO,] Fe,O, and Fe,O, oxides auartz 180 \ Figure 1 SEM photo ofPCC fly ash showing the spherical nature of the particles -28 w - * " Figure 2 SEM photo of FBC fly ash 61.5pm Figure 3. SEM photo of FBC spent bed ash -113-.2 p m Figure 4. SEM photo of scrubber Sludge 781 h v o 0 v + a -5 FBC Spent Bed Ash BC Spent Bed Ash \I 200 400 600 800 1000 TEMPERATURE (O C) Figure 5. Differential thermal analysis @TA) of combustion residues and scrubber sludge. 1 3000 2000 1000 FREQUENCY (cm-1) Figure 6. FTIR spectrum of combustion residues I I I 3000 2000 1000 FREQUENCY (crn-l) Figure 7. FTJR Spectrum of scrubber sludge and a commercial gypsum sample. 782 I i A COMPARISON OF ZEOLITE AND DOLOMITE AS GASIFICATION TAR-CRACKING CATALYSTS and Brian C. Young Energy & Environmental Research Center University of North Dakota Grand Forks, ND 58202-9018 Keywords: tar-cracking catalysts, gasification, zeolite, dolomite ABSTRACT Unconverted liquid products produced during steam gasification of coal are heavy tars. The object of this study was to compare a zeolite with dolomite as tar-cracking catalysts. Up to 75% of the tars from a lignite and a subbituminous coal were cracked to lower molecular weight compounds by use of a heated catalyst bed. Collection of the tars downstream of the catalyst bed resulted in approximately 50% less tar from the test with dolomite as the catalyst than with zeolite. Simulated distillations of the tars showed more effective cracking with the dolomite than with the zeolite. INTRODUCTION Tar produced in the gasification of coal is deleterious to the operation of downstream equipment, including fuel cells, gas turbines, hot-gas stream cleanup filters, and pressure-swing adsorption systems. Catalytic cracking of tars to smaller hydrocarbons can be an effective means of removing these tars from gas streams and, in the process, generating useful products, e.g., methane gas, which is crucial to operation of molten carbonate fuel cells. The need for on-line cracking of gasification tars is common to many processes involving gas stream cleanup. Aerosol tars are not readily removed from gas streams by conventional means and, as a consequence, often result in plugged filters or fouled fuel cells, turbines, or sorbents. Catalytic cracking of tars to molecular moieties of C,, or smaller would prevent these problems. As an example, the moving Bourdon (fixed-bed) gasifier by virtue of its efficient countercurrent heat exchange and widespread commercial use may offer the lowest-cost IGCC system, provided tar generation and wastewater contamination can be minimized. This study involved catalytic tar cracking to evaluate the potential of selected catalysts to minimize tar accumulation and maximize char conversion to useful liquid and/or gaseous products. EXPERIMENTAL Two low-rank coals (LRC) were chosen for testing the tar-cracking propensity of dolomite and a zeolite (Engelhard X-2388). The proximate analyses of the Beulah West Pit lignite and Beluga Alaskan subbituminous are shown in Table 1. Pyrolysis and steam gasification were carried out in the integrated bench-scale gasifier (IBG). A ' module for containing a catalyst bed was fabricated and connected by flange to the top of the IBG reactor. The module was heated through contact with the reactor (conduction) and flow-through of gases from the reactor (convection). Operated in the fluidized mode, the fully instrumented IBG was used to pyrolyze and gasify coal. The gas and tar produced exited the reactor through the catalyst module containing a hot catalyst (dolomite or zeolite) bed, passed through two water-cooled condensers, and was analyzed by on-line Fourier Transform infrared spectrometry (FT-IR). Trapped liquids were collected in two water-cooled condensers connected in series and were saved for later analysis. In addition, the product gas was sampled periodically by collecting samples in gas bags for later analysis by gas chromatography (GC). IBG The IBG is a small batch process gasifier, with a charge capacity of nominally 70 g of coal. This unit provides data on the effects of bed fluidization, conversion of feedstock, reaction rate response to temperature, pressure, catalyst and feed gas composition and flow rate, and gaseous products, while providing sufficient quantities of conversion products for subsequent analysis. The top of the reactor has been fitted with a catalyst module through which the hot exhaust gas must pass before entering the series of two condensers. Although the module has no heaters of its own, it receives heat from the reactor and tends to remain predictably within 5O0-IO0"C of the reactor. A typical catalyst charge to the module is 30-50 g. Gas flows uninterrupted through the system and through the heated ET-IR cell. Gas exiting the second condenser flows through the cell where it is analyzed. The data obtained indicate the effect on the tar by noting the levels of methane in the gas stream. In this study, dolomite and zeolite were tested for their effect on the pyrolysis tar. RESULTS Gas Production During Steam Gasification of Beulah Lignite Table 2 shows the operating parameters for steam gasification of Beulah West Pit lignite in the IBG. The temperatures at which the Beulah lignite was gasified were selected on the basis of potential operating temperatures of various gasifiers. Beluga subbituminous coal was gasified at only one temperature, Le., 800°C. The conversions shown are based on maf proximate analysis values for volatiles and fixed carbon in raw coal sample. There was a clear conversion trend with temperature, with 90 wt% conversion or above occurring at or above 700°C. Each reaction was carried out at the gasification temperature indicated until the production of CO, as monitored by IR spectrometry became negligible, generally 1 to 3 hours. The dolomite tended to decrepitate, producing fines, some of which blew over into the primary trap. The quantities of dolomite blown over did not correlate with temperature, but rather the fines tended to blow over with the occasional random increases in gas flow resulting primarily from uneven steam flow. The methane content of the gaseous product normalized to the volatiles content of the coal from tests at each of the five gasification temperatures are shown in Figure I . Pyrolysis methane is produced initially at temperatures above 500°C and drops off after about 25 minutes into the run. Methane continues to be produced as a result of methanation reaction and catalytic cracking. Methane is not a product of the reaction carried out at 250°C. but substantial methane is produced at each of the other temperatures. Indeed, at 700°C, more methane relative to the volatiles content was produced than at 850°C. Tar Production During Steam Gasification of Beulah Lignite The tar collected from each of the tests listed in Table 2 was analyzed following collection of the tar from the tubing and extraction from the liquids collected in the condensers. Table 3 shows the tar content collected following the catalyzed tar-cracking experiments (Tar,) at each of the five temperatures, relative to the tar content collected following the uncatalyzed tar-cracking experiments (Tar,) at each of the same temperatures. During each of these experiments, the catalyst was heated by the reactor and the flowing gases and was approximately 50°-100"C cooler than the reactor temperature shown in Table 3. Dolomite decrepitation and powder carryover contributed to the unexpectedly high dolomite tar recovery at 700°C. Care was taken following during the remaining tests to ensure that this effect was minimized. Examination of the tar recovered from the 800+"C test showed a small amount of particulate material, probably dolomite in origin. Noticeable amounts of particulates were not found in the remaining tar samples. Characterization of the tar was by simulated distillation. This technique was carried out on a Hewlett- Packard 5890 gas chromatograph equipped with a flame ionization detector (FID). A column and a hydrogen carrier gas flow at - I cm'lmin was used to separate the components. The temperature ramp was 2"C/min to 40°C. then 8"C/min to 320°C. The 1 pl injection was split 150. Peak area % was calculated from area counts exclusive of solvent peak and was used to approximate relative component concentrations. No attempt was made to identify individual components. but they were assumed to be primarily hydrocarbons. Plots of chromatography cumulative peak area % versus retention time for simulated distillations of tars collected from the tar cracking tests with and without catalyst at the temperatures of the tests are shown as Figures 2-4. Tar produced during steam gasification at 800+ "C undergoes some thermal cracking without benefit of cracking catalyst, as shown in Figure 2. Lighter organics with boiling points of approximately 150°C and 175°C constituted >50 area% of gas chromatographic components of the tar produced at 800+"C. The component distributions of the tars produced at 400". 550". and 700°C as determined by area% were not readily distinguishable. The lighter organics with boiling points of approximately 150" and 175°C made up approximately 40 area% of the gas chromatographic components produced at 400", 550". and 700°C. The effluent gas stream passed through a dolomite bed contained few tar components that boiled in the range 200" to 375°C. as shown by Figure 3. This compares with 25-30 area% of the tar over the same temperature range when not subjected to contact with a catalyst bed. The tars from each of the four tests with dolornitc show a large area% for components boiling at a temperature >375"C. The rest of the components have boiling points below approximately 225°C. Tar produced at 550°C and passed through a zeolite bed at approximately 45O0-5O0"C had approximately 55% of its organic components in the boiling point range equal to or less than 175"C, as shown by Figure 4. Tar produced at 700°C and passed through a zeolite bed at approximately 60O0-650"C had greater than 60 area% representing components with boiling points of approxirn?ely 175°C or less. Tar components in the same boiling point range produced during an 80O1.T gasification .test and passed through a bed of zeolite at 70Oo-750"C were represented by less than 50 area%. Components boiling at <270°C were represented by 60 area% of the 800+"C tar plot. 184 Tar Production During Steam Gasification of Beluga Subbituminous Coal The reduction in tar quantity by zeolite and dolomite tar-cracking catalysts was determined from data obtained at 800°C gasification of Beluga subbituminous coal using the IBG. Dolomite was shown to be more effective in cracking the tar than the zeolite. The bulk tar collected after cracking with zeolite Was 58 gas chromatographic area% and approximately one-half the weight of the tar collected with no cracking catalyst, as compared with 28 gas chromatographic area% and approximately one-fourth the weight'for cracking with dolomite. In addition, the chromatograms showed greater total components with dolomite than with zeolite. CONCLUSIONS 50% or more of tar produced during steam gasification of Beulah lignite at temperatures of 400"-800+"C is cracked by either dolomite or zeolite where the temperature of the catalyst is 50"-100°C below that of the reactor. Dolomite decrepitated during heating, especially at the tempe~atures > 550°C. resulting in loss to downstream collection devices. Overall, dolomite was more effective in the lignite tar cracking. but the X-2388 zeolite appeared to give slightly better results with the very heavy ends (tars) produced at the higher temperatures. ACKNOWLEDGMENTS The authors wish to thank the Engelhard Corporation for providing the zeolite catalyst and the U.S. Department of Energy and the Morgantown Energy Technology Center for the support to carry out this work. Thanks is also extended to Ron Kulas and Jerry Petersburg for their assistance with the laboratory and IBG work. 140 120 8g 100 2 f 80 5 .-- - 2 60 0 40 20 .g 3.0 -al E 2.5 c- 2.0 1.5 1.0 6 0.5 0.0 ._ c I c EERCRT11218CO No. of Carbons 5 E 7 -8 - 9- -10 11 13 141._5- - -.1 8.1. 9 21 23 25 I 36 98 128 151 195 234 270 317 357- '402 - - - - - - 50 100 Time, min -0.5 I 1 +85O'C +25O'C t 5 5 0 " C +7OO"C -0-4OOT 0 Figure 1. Methane concentration relative to tar production at each of five gasification temperatures 0 10 20 30 40 0' GC Retention Time, min Figure 2. Simulated distillation of tar collected during gasification of Beulah lignite at 400". 550", and 700" and 800 +"C. 785 120 r.n""n - No. of Carbons 5 6 7 8 9 10 11 13 1415 1819 21 23 25 36 98 126 151 195 234 270 317 357 402 - - - -- . _---_ .-_ - - - . .. , - - . n-Alkanes, B.P, 'C - 60 40 20 ... -~ . 0 , .. . ,.. ,, .____.I.___.__... 1 0 10 20 30 40 50 GC Retention Time, min ------------- "I_:"""[ ...... 700% .- 800+' - - - Figure 3. Simulated distillation of dolomite-cracked tar collected during gasification of Beulah lignite at 400". 550". 700". and 800+"C. EERCRTllllOCDi No. of Carbons 5 6 7 8 9 10 11 13 1415 1819 21-73. 25 36 98 126 151 195 234 270 317 357 402 - - - - - - .. - - -. . . n-Alkanes, B.R, "C - - - 400'C ..'.. 550°C - 700°C c &)O+"C I 0 10 20 30 40 E GC Retention Time, min I Figure 4. Simulated dis:illation of zeolitecracked tar collected during gasification of Beulah lignite at 400". 550". 700". and 800+"C. TABLE 1 Proximate Analysis of Beulah West Pit Lignite and Beulah Subbituminous Moisture. AR. Volatiles. mf. Fixed Carbon. Ash, mf. / Coal Wt % wt% mf, wt% Wt % Beulah West Pit 30.2 44.8 47.6 7.6 Beluga 22.3 45.0 44.5 10.5 TABLE 2 Conversion of Volatiles and Fixed Carbon to Tar and Gaseous Products under Steam Gasification Conditions at Different Temperatures Run Temperature, Am., glmin Conversion No. Coal "C Catalyst at 50 psig wt%, mf IBGl I Beulah West 800+ Dolomite Steam, 3-4 97 IBGll IBG12 IBG12 IBG12 IBG12 IBG12 IBG12 IBG12 IBG12 IBG12 1BG12 IBG13 IBG13 IBG 13 IBG13 IBG13 IBG13 IBG13 Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beulah West Beluga Beluga Beluga 250 550 700 400 250 700 700 550 800 + 400 800 + 250 700 550 400 800 800 800 Dolomite Dolomite Dolomite Dolomite Zeolite Zeolite Zeolite Zeolite Zeolite Zeolite None None None None None None Dolomite Zeolite Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 Steam, 3-4 17 48 86 31 12 82 86 44 98 34 98 11 86 45 28 88 90 87 TABLE 3 Uncracked Tar (TarJTarJ 100%) Temp., "C Dolomite Zeolite 250 97 106 400 550 700 24 12 47 50 35 351 800 + 19 32 * Average of two tests. 787 HYDROCRACKING OF POLYOLEFINS TO LIQUID FUELS OVER STRONG SOLID ACID CATALYSTS K. R. Venkatesh, J. Hu, J. W. Tierney and I. Wender Department of Chemical and Petroleum Engineering University of Pittsburgh, Pittsburgh, PA 15261 Keywords: polymer hydrocracking, solid acid, synthetic fuel INTRODUCTION Post-consumer plastic makes up about 13 wt% of the 48 million tons of total packaging wastes generated annually'. Plastics are non-biodegradable, constitute a increasingly large volume of solid wastes (20 vol. % in 1990),, and are not being recycled to a significant extent'. Pyrolysis, as an alternative for plastic waste recycling, usually results in unsaturated and unstable oils of low yield and value. Significant amounts of char are formed on pyrolyzing plastic wastes. Liquefaction of plastic wastes could be a useful way of producing desirable liquid transportation fuels. Thermoplastics such as polyethylene (PE), polypropylene (PP) and polystyrene (PS) make up the bulk of plastic wastes'. The liquid products obtained from them are likely to have a high volumetric energy content because of their relatively high (HIC) atomic ratio. It is known that strong liquid superacids such as Magic acid (HSO,F:SbF,) are effective in converting paraffinic wax to t-butyl cations at room temperature; however, the stability of these liquid acids is poor at the high temperatures and reducing environments* needed for improving the H/C ratio of the products. In this paper, we discuss the results obtained from the hydrocracking of high density polyethylene (HDPE), PP and PS over metalpromoted sulfated zirconia catalysts, viz., Pt/Zr02/S04 and Ni/ZrO,/SO,. Strong solid acids such as these and other anion-modified metal oxides are active in a variety of acid-catalyzed hydrocarbon reaction^'^^^^*^*^; they are environmentally benign, non-corrosive (unlike strong liquid acids) and are easily separated from product streams. They are also characterized by long-term activity in the presence of hydrogen" in reactions such as n-butane isomerization. EXPERIMENTAL The sulfated zirconium oxides were prepared as described in a previous publication". Incorporation of Ni on to sulfated zirconia was achieved using wet impregnation of Ni(NQ), followed by drying at 110°C overnight and calcination at 600°C for three hours. The amounts of Pt and Ni promoted onto ZrOJSO, were 0.5 wt% and 2.0 wt% respectively, based on the final weight of the catalyst. HDPE (density 0.95, M,v.=125,000), PP (isotactic, density 0.85, M,.=250,000) and PS (M,.=280,000) were obtained from Aldrich Chemicals Inc. and were used as received. A Pt/A1,03 catalyst (1 wt% Pt) was purchased from Aldrich and was activated at 450°C in air for one hour before use. All of our polymer reaction studies were conducted in a 27 cc stainless steel microautoclave attached to a 15 cc reactor stem. The catalysts were activated at 450°C in air for one hour before use; to minimize exposure to moisture, they were then charged immediately into a dried (1 10°C) reactor which was then quickly sealed. After cooling to room temperature, the reactants were added through the reactor stem. The feed to catalyst ratio was 5: 1 by weight in all experiments. The reaction pressure (initial) at 325°C and 1200 psig (cold) H, were 1835 psig; initial reaction pressures at 375°C experiments for 1200 psig (cold) H, and 750 psig (cold) H, were 1930 psig and 1210 psig respectively. The reaction products were analyzed using a GC-MS (Hewlett Packard 5970B) and a gas chromatograph (Hewlett Packard 5890 11) with an FID detector. Simulated distillations of products obtained were conducted using an HP 5890 Series I1 gas chromatograph (with a TCD detector) controlled by a HP 3396A integrator which is programmed to run the ASTM D2887 distillation method. The entire product mixture is dissolved in carbon disulfide (CS,) to form a homogeneous mixture; CS, is not detected by the TCD detector of the simulated distillation unit. The result is given as a series of boiling points, one after every 5 wt% of the sample is eluted. Sulfur analyses of the catalyst samples were performed by Galbraith Laboratories, Inc. RESULTS AND DISCUSSION Hydrocracking of HDPE at 325°C (1200 psig (cold) H,, 60 min.) over the Pt/ZrOJSO., I 788 catalyst gave a 25 wt% conversion, mainly to gases (C,-C, alkanes), perhaps due to poor mas transfer during reaction so that liquid products from the initial cracking of HDPE underwent multiple cracking to gases. When HDPE was reacted at 375°C and 1200 psig (cold) Hz (for 25 min.) over the same catalyst, more than 99 wt% of HDPE could be converted to liquids (69 wt%) and C,-C, gases (- 30 wt%) (Table I). Total conversion was based on solid recovered which likely consisted of polyethylene molecules of shorter chain length than the starting HDPE. When the same reaction was conducted with a Ni/ZrO,/SO, catalyst, HDPE conversion exceeded 96 wt% with slightly different liquid and gas yields (Table I). Table I1 lists the detailed product distribution of the liquid products formed from the reaction of HDPE over pt/zfl,/So, and Ni/ZrO,/SO, catalysts, for the results shown in Table I. Large amounts of isoparaffins are obtained for each carbon number, close to an order of magnitude higher than their Shght-chain counterparts. The high iso-/normal alkane ratios obtained at these temperatures is due to a kinetic rather than a thermodynamic effect. The more stable branched carbenium ions could abstract a hydride ion from an oligomeric fragment or react with hydride ions formed from the dissociation of molecular hydrogen over the metal as suggested by Iglesia et al." and are thus easily desorbed from the catalytic sites before an equilibrium is reached. Impregnation of Ni on ZrO,/SO,resulted in a higher iso/normal ratio of C,-C, alkanes from HDPE than that obtained with Pt. This may be due to the lower hydrogenation activity of Ni (based on n-hexadecane hydrocracking experiments") resulting in correspondingly lower concentration of hydride ions on the catalyst surface; the adsorbed carbocations could undergo a higher degree of skeletal transformation before desorption from the active sites by hydride transfer. Hydrocracking of HDPE over a Pt/ZrQ/SO, catalyst was conducted with a lower hydrogen pressure (750 psig (cold)) with other conditions the same as in Table I. The same total conversion (99 wt%) was obtained in this reaction but with a higher yield of liquid products (79.8 wt%) and a correspondingly lower yield of gases (19.2 wt%). Comparison of the liquid products from HDPE reactions at both values of hydrogen pressure are given below. Reactions with PP were conducted under the comparatively milder temperature of 325°C in the presence of 1200 psig (cold) H,; at these conditions, PP was converted almost entirely to C,-C, gases for a reaction time of one hour. When the reaction time was reduced to 20 minutes, PP conversion was - 100 wt% with about 90 wt% yield of liquid products. Product analysis showed (Table 111) that 78.6 wt% of C5-C,, gasoline range compounds were present in the liquid products together with 11.4 wt% products in the diesel range (C,&,,). Similar results were obtained with a Ni/ZrO,/SO,catalyst in the reaction of PP (Table 111). It appears that at sufficiently high temperatures, the hydrogenation activity of Ni approaches that of Pt in these reactions. It was found earlier that, in hydrocracking of n-hexadecane at milder conditions (16O"C, 350 psig (cold) H,), Ni/ZrO,/SO, showed little activity whereas high conversions were obtained with a Pt/ZQ/SO, catalyd4. A cheaper, non-noble metal such as Ni can be effective in these reactions but requires a higher temperature for activation by ' hydrogen. As was the case with HDPE, high ratios of isohormal paraffins was obtained. We found that PS could also be converted to benzene, alkylated aromatics and bicyclics at 300°C with 1200 psig (cold) H,. The reaction of polypropylene at 325°C and 1200 psig (cold) H, over a ZrO,/SO, (in the absence of either Pt or Ni) catalyst which has strong acidity but no hydrogenation function, resulted in no appreciable conversion of polypropylene. The white ZrO,/SO, catalyst turned black during reaction indicating deactivation, possibly by coking. This result confirms the finding by others5.'* that the presence of a hydrogenation metal on ZrO,lSO, provides stability to the catalyst by resisting coke formation in a variety of hydrocarbon reactions. On the other hand, a one wt% Pt supported on y-A1203 catalyst (strong hydrogenation function but weak acidic function) also gave almost no conversion of PP at 325°C and 1200 psig (cold) H,. It appears that at the conditions employed for hydrocracking of these polymers, both strong acid and hydrogenation functions are required for high yields of low molecular weight branched alkanes in the gasoline range. We conducted simulated distillations of the product mixtures from polymer hydrocracking reactions to analyze their boiling point characteristics. The boiling point distribution Of the liquid prodUCtS obtained from the hydrocracking of HDPE (Table 3) at 3 7 5 " ~fo r 25 minutes over Pt/ZrO,/SO, under two different initial hydrogen pressures are shown in Figure 1. Reactions under 750 psig (cold) H, and I200 psig (cold) H, seem to have only a marginal effect on the boiling ranges of the liquid products obtained. More than 90 wt% of the products are in the gasoline (C5-C12r)a nge (Le., between 90°F (32.2"C) and 421°F (216.1OC)) indicating the possibility of converting HDPE to a high quality liquid fuel. A similar boiling point curve was also obtained from the simulated distillation of the products from PP over Pt/ZrO,/SO,; more than 70 wt% of the products obtained boil in the gasoline range. A strong tendency towards isomerization over these metal-promoted sulfated zirconia catalysts was observed for both HDPE and PP hydrocracking; this is reflected by the similar boiling point curves obtained from products of both polymers. Despite the high activity of the sulfate-modified zirconia catalysts in these reactions, they have questionable long-term stability at these severe reducing conditions. Sulfur analyses of the catalysts after hydrocracking of HDPE revealed that the catalysts lost about 34 wt% of their starting sulfur contents during reactions at 375°C and in the presence of high Hz pressures. Since the presence of SO,'- anions on the catalyst surface is responsible for the strong acidity of these catalysts, loss of sulfur during reaction implies loss of activity for longer periods of time. CONCLUSIONS Sulfate-modified metal oxides promoted by a hydrogenation metal exhibit high activities for the hydrocracking of HDPE, PP and PS. While HDPE and PP are cracked predominantly to gasoline range branched alkanes (C5-C,J, PS is hydrocracked to benzene, alkylated aromatics and bicyclic compounds. Impregnation with a non-noble metal such as Ni, which showed little activity in alkane hydrocracking at milder conditions (160°C and 350 psig (cold) H2) resulted in high activity for polymer hydrocracking at 325"C+, indicating that activation of Ni occurs at higher temperatures. The long-term stability of these catalysts for these reactions is in doubt due to their loss of sulfur. Novel catalyst formulations which have higher stability under severe reducing conditions are currently being investigated. ACKNOWLEDGMENTS The financial support of this work by the U.S. Department of Energy (Grant No. DEFG22- 93PC93053) is gratefully acknowledged. REFERENCES I. R. J. Rowatt, "The Plastics Waste Problem", Chemtech, Jan. 1993, 56-60. 2. U.S. Environmental Protection Agency (EPA) Office of Solid Waste. "Characterization of Municipal Solid Waste in the United States: 1990 Update" Publ. No. EPA1530-SW-90- 042A. Washington, D.C., 1990. 3. S.G.Howell, J. Hazard. Matl., 29 (1992) 143-164. 4. G.A. Olah, G.K. Surya Pmkash, and J. Sommer, Superacids, John Wiley & Sons, New 5. K. Tanabe and H. Hattori, Chem. Lett., (1976) 625. 6. M. Hino and K. Arata, Chem. Lett., (1981) 1671. 7. K. Arata and M. Hino, React. Kinet. Catal. Lett., (1988) 1027. 8. H. Matsuhashi, M. Hino and K. Arata, Chem. Lett., (1988) 1027. 9. K. Ebitani, H. Konno, T. Tanaka, H. Hattori, J. Catal., 135 (1992) 60-67. 10. Tanabe, K., Hattori, H. and Yamaguchi, T., Crit. Rev. Surf. Chem., l(1) (1990) 1-25. 11. M.Y. Wen, I. Wender and J.W. Tiemey, Energy & Fuels, 4 (1990) 372-379. 12. E. Iglesia, S.L. Soled and G.M. Kramer, J. Catal., 144 (1993) 238. 13. K.R. Venkatesh, J.W. Tiemey and I. Wender, unpublished results. 14. W. Wang, Ph.D. dissertation, University of Pittsburgh, 1994. York, 1985. 790 PRODUCT Conversion Gases (Cl-C6) Liquids: C4-Cl2 C 13-C20 C21 and above Yield (wt%) obtained with wzlQ/so4 NilZrOJSO, 99 wt% 98 wt% 30.0 28.0 68.7 67.6 0.3 2.4 trace trace 791 PRODUCT Gases (Cl-C6) Liquids: C4-Cl2 C13-C20 ' C21 and above Yield (wt%) obtained with Pt/Zrq/SO, NilZrO,lSO, 10.0 14.5 78.6 76.0 11.4 9.5 trace trace Wt% distilled Figure 1 Boiling point curves obtained from the simulated distillation of products from HDPE hydrocracking over a Pt/ZrO,/SO, catalyst at 375°C for 25 minutes under two different initial H, pressures. 192 TRACE EMISSIONS FROM COAL COMBUSTION: MEASUREMENT AND CONTROL Lesley L Sloss IEA Coal Research, Gemini House, 10-1 8 Putney Hill, London SW 15 6AA Keywords: trace emissions, measurement, control MTRODUCTION Combustion of coal is a potential source of emissions of many trace elements and organic compounds to the atmosphere. It is important that emissions of potentially toxic air pollutants from sources such as Coal combustion are measured and, if necessary, controlled in order to limit any environmental effects. Increasing concern about the effects of trace pollutants in the environment may lead to the introduction of emission standards for some of these species. If such emission standards are adopted they must be supported by commercially available equipment which can measure and monitor the emissions with enough accuracy to ensure compliance. Efficient coal combustion is not a significant source of emissions of organic compounds and therefore these compounds are not discussed further here. However, since there is increasing concern over emissions of mercury from coal combustion, specific attention will be paid to this particular trace element. Several reviews have been published by IEA Coal Research on emissions from coal combustion. These include the halogens (Sloss, 1992). trace elements (Clarke and Sloss, 1992). organic compounds (Sloss and Smith, 1993), and mercury (Sloss, 1995). A complementary report has also been published on sampling and analysis of emissions of these compounds from coal-fired power station stacks (Sloss and Gardner, 1994). This paper draws together the conclusions from these reports. EMlSSIONS OF POTENTIAL AIR TOXlCS Various estimates have been published which attempt to evaluate coal combustion as a sourcc of potential air toxic emissions. Data from Nriagu and Pacyna (1988) indicate the importance of coal as a source of some trace elements on a global scale. For example, coal combustion may be responsible up to 21% of Sb emissions, around 18% of Ni and Se emissions and 15% of Cr emissions Emissions of Cu, Sn, TI and Zn from coal combustion are also thought to contribute between 5 and 20% of global emissions of these elements. Estimates from the early 1980s (Pacyna and others, 1993) indicate that coal combustion may be responsible for up to 25% oftotal global mercury emissions to the atmosphere. There are a few more recent estimates from some countries in the 1990s, for example, coal's contribution to mercury emissions from human activities is 23% in Finland, 27% in the former FRG, about 10% in the Netherlands, 45% in the UK, and 16% in the USA (Sloss, 1995). Estimation of global and regional budgets is difficult. Emission factors commonly based on a relatively small amount of actual measured data. The wide variation in the composition of coals, in combustion conditions, and in pollution control equipment need to be taken into account when estimating emission factors. Furthermore, many of the techniques used for the measurement of emissions of trace species, and thus for the estimation of emission factors, are still under development and are known to have serious limitations. Estimates for global and even regional emissions of trace species from most sources can therefore be considered as no more than educated guesses. LEGISLATION Concern over the emissions of potential air toxics from all sources and their possible effects in the environment has lead to the introduction of legislative controls in several countries. Legislation specific to the emission of individual trace elements has been specified in Austria. Germany and certain states in Australia (AHC, 1992; Maier, 1990; Nilsson, 1991). This legislation is summarised in Table 1 The 1990 Us Clean Air Act requires the evaluation of emissions of several trace elements with a view to the possible introduction of relevant legislation in the future (Chow and others, 1990). Legislation for power stations is also being considered in Canada and the Netherlands. Although no specific emission standards apply in Sweden, electrical utilities are required to fit best available technologies. These include particulate controls and FGD processes and therefore result in a substantial reduction in the emissions of most trace elements (Clarke and Sloss, 1992). 4' 793 MEASUREMENT OF EMISSIONS Sampling and analysis techniques for the measurement of trace species at the concentrations emitted from coal-fired power plants are still under development. Countries such as Germany, Japan and the UK have published guidelines for sampling and analysis of some trace pollutants. In the USA, the methods are specified by law within the Code of Federal Regulations. However, many of these methods are known to have inherent problems and are still subject to review. The maiority of sampling techniques are based, initially, on the separation of gases from particles on filters. in cyclones or in cascade impactors. Each ofthese techniques are known to have problems such as clogging and irreversible adsorption (Masterson and Bamert-Wiemer, 1987). Gaseous species may be analysed directly by analytical instruments, but such instruments are rarely portable. Samples are more commonly transported to the laboratow for analysis. Some vapour-phase species may be reduced to liquid form simply by condensation in cooled chambers. Other species are captured in a series of impinger bottles containing solutions which selectively solubilise the species of interest. Activated cabon can be used to capture volatile trace metals such as mercury. Although solid sorbents have the advantage of allowing volatile species to be trapped and transported in a stable form, some have problems with background contamination and decomposition products (Sloss and Gardner, 1994). The development of sampling and analysis techniques for mercury is proving to be a particularly challenging problem. The speciation of mercury, as oxidised forms such as mercury chloride, or in the elemental form. determines its behaviour in pollution control equipment and in the environment. However, mercury emissions cannot be speciated with the standard methods currently available for sampling emissions of trace metals. New techniques based on sorbents such as activated carbon appear to be the most promising methods (Sloss and Gardner. 1995: Sloss, 1995). Sampling and analysis techniques are not at the stage where they are accurate enough to produce a single value which would be considered representative. From what is already known of the behaviour of potential air toxics in coal-fired systems, their emissions are never constant, they vary with coal tvpe, combustion conditions, pollution control systems and wen depend on the concentration,of other pollutants within the flue gas with which they may react. Continuous emissions monitors produce virtually real-time data, avoiding transport and handling errors. and providing true representation of potential air toxic concentrations over time. However, continuous emissions monitors are not currently available for air toxics. Several systems, such as those based on FTIR, are under development (Sloss and Gardner, 1994). EFFECTS OF EMISSION CONTROL TECHNOLOGIES Currently there are no widely available control technologies designed specifically for the removal or trace elements from coal-fired power stations. However, technologies for the removal of particulates. such as electrostatic precipitators (ESP) and fabric filters, and control technologies for SO, and NO,, may affect emissions of potential air toxics. Particulate control systems capture any pollutants which are associated with the particles retained. The capture of individual air toxics thus depends upon their volatility. Most trace elements are not especially volatile and are captured efficiently by particulate controls. for example only 2% of Cd in the flue gas passes ESP uncaptured. However, B and Se are slightly more volatile and between 20 and 30% of these elements may pass uncaptured. Unless lime or a similar sorbent has been used in the boiler. virtually all the halogen gases pass through particulate controls (Clarke and Sloss, 1992; Sloss, 1992). The capture of mercury by particulate control devices dependsupon its speciation. Mercury in the particulate form (G%)is captured efficiently. Oxidised mercury may also be associated with fly ash or can adsorb onto particles already associated with baghouses. Average mercury capture efficiencies in ESP and baghouses are around 35-40%. Since mercury speciation is temperature dependant, the capture of mercury in particulate control devices can be optimised by keeping temperatures as low as possible (<15OoC) to increase the proportion of mercury in the oxidised form (Sloss, 1995). Wet and dry flue gas desulphurisation (FGD) systems, required in many countries to remove SO,, incidentally remove some amounts of potential air toxics. For example, Figure 1 shows the average removal ofvolatile elements in wet-lime FGD systems in the Netherlands (Clarke and Sloss, 1992). Some FGD systems remove around 50% of the remaining B and Se in the flue gas. Reductions of over 90% for all the halogens have been achieved in such systems (Clarke and Sloss, 1992; Sloss, 1992). 794 / Wet and dry FGD systems have wide ranges of efficiency for mercury capture from 20% up to 90%. Mercury capture in FGD depends upon its speciation. Up to 95% of oxidised mercury can be removed in spray dry scrubbers whereas elemental mercury passes through uncaptured. Capture of mercury in FGD systems can be maximised by increasing the proportion of oxidised mercury in the flue gas (S~OSS, 1995). Combustion modifications for NO, control may lead to increased concentrations of unburned carbon in flue gases. It is not clear to what extent this unbumt carbon may affect the distribution and behaviour of potential air toxics. NO, control systems do not appear to reduce or increase trace or minor element emissions. However, high dust SCR systems can oxidise up to 95% of the mercury in flue gas, enhancing the capture of mercury in FGD systems downstream (Sloss, 1995). SPECIFIC CONTROL OF POTENTIAL AIR TOXICS There is currently no requirement for the specific removal or abatement of potential air toxics from the flue gas of coal-fired power stations. However, in the future, legislation on air toxics emissions is likely to become more stringent. Some specific technologies for the capture of potential air toxics are already under development and some are commercially available for use on waste incineration units. Concentrations of the more harmful air toxics. such as mercury, may be several orders of magnitude higher in flue gas emissions from waste incinerators than from cod-fired power plants. Work has already been started in several countries to reduce emissions of air toxics from waste incinerators. Some of the technologies used in waste incinerators may be applicable, with modification, to coal-tired units (Clarke and Sloss, 1992). Sorbents which are available for the removal of heavy metals, such as mercury. from flue gases, include those based on activated carbon, zeolites, siliceous materials, alumina, and calcium compounds. Up to 100% of oxidised mercury and 60% of elemental mercury in flue gas may be captured with activated carbon Sulphur impregnated activated carbons can capture over 9096 of total mercury emissions and iodine impregnated activated carbons are reported to capture up to 100°/,. Figure 2 shows mercury removal by different types of activated carbon injected upstream of a spray dry scrubber and a baghouse. The use of some sorbents in coal-fired power stations may be limited due to low operating temperatures, harmful secondaw effects and the high cost of some sorbents (Mojtahedi and Mroueh, 1989). CONCLUSIONS Coal combustion is an important source of some trace elements to the environment. Existing legislation for the control of particulate emissions effectively controls emissions of the ma-iority of trace elements. Flue gas desulphurisation technologies may efficiently capture many of the remaining vapour phase pollutants. Over 90% of the halogens and 40-50% of the B and Se may be captured by this means. The speciation of mercury determines the emissions and effects of mercury from coal combustion Particulate control devices may capture up to 40% of mercury emissions, and FGD systems commonly up to 7Ph These efficiencies may be enhanced by maximising the proportion of mercury present in the flue gas in the oxidised state. More research is required in order to understand mercury speciation and to use this information to determine the most appropriate control strategies. Emission standards are becoming more stringent and, in the future. it is likely that emission limits for air toxics will be introduced more widely for sources such as coal-fired power plants. However. emission standards are worthless if the emission concentrations they specify cannot be measured accurately and on a regular basis by operators and regulatory authorities. REFERENCES AHC (I 992), Acrs/ruliuir High ('tnnrnis.sioi~. London, UK, Per.wrml conrrnriiiicu/iorr Baek S 0, Field R A, Goldstone M E, Kirk P W, Lester J N, Perry R (1991) A review of atmospheric PAH: sources, fate and behaviour. Wafer, Air aid Soil f+dh/ifJiJ, 60; 279-300 Chow W, Miller M J, Fortune J, Behmns C, Rubin E (1990) Managing Air Toxics. X3rd A M M Meeling. Pittsburgh, USA, 1990. 15 pp Clarke L B, Stoss L L ( 1992) 7rocr elernriif ernissI~~~i.s,fcrronorl c~~rnhi~s~aIioidir~ u.~(ficafi~~ii. IEA CW49. London. UK. LEA Coal Research. 1 1 I pp Maier H (1990) Emission of volatile and filter-penetrating heavy metals from lignite-fired plants. VGH Krufm~erk.s/rchnik7, 0 (IO); 749-755 ' 795 Masterson T, Barnert-Wiemer H (1987) GithaIion qf rn0s.s balance Ini~e.sligo1ion.isn c0al:fired powerp/an/s. Juel--2160. Germany, Diisseldorf, Kernforschungsanlage Julich GmbH, 54 pp Meij R, Spoelstra H, Waard F J de (1989) The determination of gaseous inorganic trace compounds in flue gases from coal-fired power plants. In: Man and his eco.\yslem. Proceedings qf /he 81h world clean air c~~iigre1s9s8 9. The Hague, the Netherlands, 1 1-1 5 Sep 1989. Amsterdam, the Netherlands, Elsevier Science Publishers BV, vol3, pp 7 17-722 Mojtahedi W, Mroueh U M (1989) Tract? elemenls remow1,from hor,flflL. gases. VTT-TUTK-663, Valtion teknillinen tutkimuskeskus, Espoo, Finland. Nilsson K (1991) Swedish emission standards for waste incineration. IINKP /ndtf.Wy and k,it~~Ironmenp/p. 73-74 Nriagu J 0, Pacyna J M (1988) Quantitative assessment of world-wide contamination of air, water and soils by trace metals. NuItfre, 333; 134-1 39 Pacyna J M, Voldner E, Bidelman T, Evans G, Keeler C J (1993) Emissions, atmospheric transport and deposition of heavy metals and persistant organic pollutants. In: f'rficeednigs uf the Is/ workshop on emt.w;om atd m~nlellittgq fa /mo.spheric /rampor/p f persis~enlo rganic polltt/an/.sa nd hemy rnrla1.v. Durham, NC, USA, 6-7 May 1993. vp Sloss L L (1992) Halogen emis.yions,from coal comhmliotr. IEA CW45. London, UK, IEA Coal research. 62 pp Sloss L L (1995) Merc7~yem i.s.siom mid efleffecls: the role efc oal. London, UK, IEA Coal Research. In draft. Sloss L L, Gardner C (1994) Sampling m7d anut'ysis of /race emis.sion.s,from c0al;firrd p/an/s. London, UK, IEA Coal Research. I05 pp (in druft) Sloss L L, Smith I S (1993) Organic emi.~?on.s.fiomco n1 t~lili.~lionIE. A CW63. London, UK, IEA Coal Research. 69 pp Wiederkehr P (1991) Control of hazardous air pollutants in OECD countries a comparative policy analysis. Mmqing Hazardmrs Air Polltrlattls: Stale of /he Arl. EPRl Conference, Washington 199 1. 16 PP Table 1 COUNTRY AIR TOXIC LIMIT (mdm') National legislation for air toxic emissions from coal-fired power plants Australia Austria As, Cd, Hg, Ni, Pb. Sb, V Varies between States and Territories Cr, Pb, Zn 2.0 (total of all three) As, Co, Ni 0.5 (total of all three) Cd and Hg 0.05 (total separately) Germany Inorganic dust Category I Cd, Hg, TI 0 2 (total of all three) Category I1 As, Co, Ni, Se, Te 1 .O (total of all three) Category 111 Cr, Cu, Mn, Pb, Pd, ' Pt, Sb, Sn, V 5 0 (total of all three) Organic substances Category I Category 11 Category 111 20 (total) 100 (total) 150 (total) Carcinogenic substances Category 1 (including BaP) 0.1 (total) Category 111 hydrazine etc 5.0 (total) Category 11 As, Co etc 1.0 (total) Planned legislation Canada, the Netherlands, the US4 '196 100 80 s gn 60 2 40 -c E" t 20 0 L Se n-l F I Figure 1 Average removal of volatile elements in wet-lime FGD systems In the Netherlands 1w 90 80 70 - g9 6 0 5 50 $ 40 30 0 20 io Iodine impregnated carbon Sulphur Impregnated carbon I I I I I I 0 1 2 3 4 5 L Relalive carbon injection rate Figure 2 Influence of active carbon type on mercury removal 797 TRACE METAL CONTENT OF COAL AND ASH AS DETERMINED USING SCANNING ELECTRON MICROSCOPY WITE WAVELENGTH-DISPERSIVE SPECTROMETRY Karen A. Katrinak and Steven A. Benson Energy & Environmental Research Center University of North Dakota Grand Forks, ND 58202-9018 Keywords: scanning electron microscopy, trace metals, coal analysis ABSTRACT Scanning electron microscopy with wavelength-dispersive spectrometry has been used to measure trace metals in coal and ash. Hg, As. Ni, and Se have been detected in individual pyrite grains in Illinois #6 coal at levels up to 2680 ppm, 410 ppm, 320 ppm, and 880 ppm, respectively. These elements were present in fewer than half the grains analyzed. Cr has been detected at up to 950 ppm in half of clay mineral grains analyzed in Illinois #6 coal. The same trace metals were detected in pyrite and clay grains from Pittsburgh #8 coal. Ash samples show a similarly heterogeneous distribution of trace metals. Hg has been detected at up to 700 ppm in 24% of aluminosilicate. particles analyzed in ash from Absaloka coal, a subbituminous Montana fuel. These data confrm that coal cleaning processes which remove pyrite are likely to be suitable for trace metal emissions control. In addition, back-end control devices which target specific types of ash particles may be helpful for control of air toxics emissions. INTRODUCTION Scanning electron microscopy (SEM) is one of the analytical tools available for determining the abundance of trace metals in coal and ash samples. This information is important in predicting and evaluating the behavior of these substances in combustion processes, a topic which is of increased importance in recent years as stricter regulation of trace metal emissions from coal-fired power plants is under consideration. Although scanning electron microscopy is not routinely applied to detection of trace quantities of metals, the use of a wavelength-dispersive spectrometer attachment makes such analyses possible. Scanning electron microscope techniques differ from traditional trace metal analysis techniques in that SEM provides information with high spatial resolution, compared with the bulk compositions obtained through atomic absorption and other widely-used methods. High-spatialresolution data concerning trace metal distribution in coal and ash is important for two reasons. First, ash behavior in fossil fuel combustion systems is best understood in terms of the behavior of individual particles. Knowledge of the bulk composition of an ash deposit frequently is not sufficient in determining what caused that deposit to have its particular physical characteristics such as size, &ability, crystallinity. and density. Information concerning the chemical and mineralogical composition of individual ash particles can provide insight into how particles interact and transform to produce a deposit. Methods for obtaining this information using SEM with energy-dispersive x-ray spectrometry (EDS) have become widely available (1-3). The SEMEDS technique provides data for major elements only, with detection limits of approximately 0. I wt%. In order to obtain similar information concerning trace elements, SEM with wavelengthdispersive spectrometry (SEM-WDS) must be used. The SEM-WDS technique has detection limits of approximately 100 ppm (0.01 wt%) for most metals. Although it is time-consuming, the SEM-WDS method is valuable because it provides a means for acquiring single-particle trace element data for coal and ash particles, information that is essential in understanding how best to control the emission of trace elements from combustion sources. Trace metal emissions from coal-fired power plants may be subject to increased regulation; thus knowledge of how best to control them is vital. A second reason for investigating the distribution of trace metals at high spatial resolution is that this information is helpful in understanding potential health effects of these substances. Trace metals can occur as coatings on airborne particles, and frequently are found in particles in the respirable size range (4.5); in these instances, the toxicity of the trace metals is greater than if those elements were distributed evenly throughout a particle, or were present in larger, nonrespirable particles. For the purposes of assessing potential health impacts of trace metal emissions, it is important to know whether these elements are distributed homogeneously throughout an ash sample, or whether their distribution varies on an individual-particle basis. 798 METBODS Samples were mounted in epoxy, cross-sectioned, polished, and coated with carbon to improve conductivity. Analyses were conducted on a JEOL 35C scanning electron microscope equipped with two JEOL wavelength-dispersive spectrometers with xenon-filled proportional COUnters, and a Noran Instruments energy-dispersive spectrometer. The analytical capabilities of the microscope are controlled by a Noran Instruments Voyager 2 computer system, which can wordinate simultaneous EDS and WDS. The microscope was operated at an accelerating voltage of 25 kV with a beam current of 8 nA. Wavelength-dispersive spectral peaks were counted for 100 s; the total energy-dispersive live time per spectrum was 3 s. Certified standards were used for calibration. The data were subjected to ZAF corrections following collection. Individual coal mineral grains and ash particles as small as 5 p n in biameter were analyzed. Under the more commonly used SEM-EDS analysis conditions, it is possible to analyze volumes as tittle as 1 pm in diameter, but the more intensely energetic conditions required of SEM-WDS make it impossible to analyze these smaller quantities without exciting the surrounding area (6). RESULTS AND DISCUSSION Coal and ash samples were analyzed for trace metal content using SEM-WDS. EDS was also used to determine the major element composition of each coal mineral grain or ash particle. Ash Analyses. A sample of Absaloka ash was inspected for Hg content using SEM-WDS. Iodated activated carbon sorbent had been added to this Montana subbituminous coal. Ash particles analyzed ranged from 5 to 20 pm in diameter. As shown in Table I, Hg was detected in six particles (21% ofthe total analyzed), in amounts ranging from 100 to 700 ppm (0.01 to 0.07 wt%). These Hg-bearing particles are mostly Ca- and AI-bearing silicates, with some S present. Results for the 22 non-Hg-bearing particles analyzed in the same ash sample are shown in Table 2. The major element composition of these particles is similar to that of the Hg-bearing particles listed in Table 1, suggesting that the occurrence of Hg in these ash particles is not related to any compositional parameter. Another sample of Absaloka ash, produced from coal to which a non-iodated activated carbon sorbent had been added, did not have any detectable Hg in individual particles. The ash particles in this sample were predominantly Ca- and AI-bearing silicates, as in the sample produced using iodated carbon sorbent, but with little S present. Coal Analyses. In a sample of Illinois #6 bituminous coal, individual mineral grains were selected for trace metal analysis. Table 3 shows results for pyrite grains in Illinois #6 coal. Hg, As, Ni, and Se are present in individual grains at levels up to 2680 ppm, 410 ppm, 320 ppm, and 880 ppm. respectively. These trace metals were present in fewer than half of the pyrite grains analyzed. Clay mineral grains from the Illinois #6 coal sample were examined for Cr content; this element was detected at up to 950 ppm in half of the grains analyzed. These results show the heterogeneous distribution of these trace metals in coal mineral grains. Similar results are evident for Pittsburgh #8 bituminous coal. Table 4 shows the distribution oftrace metals in pyrite grains. As and Hg values for individual grains range up to close to 3000 ppm; Cd was detected in amounts less than 100 ppm only, Ni ranges up to approximately 1300 ppm; and Se values are as high as almost 2000 ppm, In clay mineral grains from Pittsburgh #8 coal, Cr ranges up to 377 ppm in 27 individual grains, including six grains with Cr not detected. The average value for Cr in the Pittsburgh #8 clay mineral grains is 75 ppm. CONCLUSIONS This study has shown the varied distribution of trace metals in coal and ash samples, The relative abundance of Hg and other trace metals in pyrite grains suggests the effectiveness of coalcleaning processes in helping to reduce toxic emissions from power plants. Further investigation of the distribution of trace elements in ash particles of different compositions may lead to the development of emissions control devices tailored for removal of specific metals. 799 REFERENCES 1. 2. 3. 4. 5. 6. Zygarlicke, C.J.; Steadman, E.N. “Advanced SEM Techniques to Characterize Coal Minerals,” Scanning Microscopy 1990, 4(3), 579-590. Casuccio, G.S.; Gruelich, F.A.; Hamburg, G.; Huggins, F.E.; Nissen, D.A.; Vleeskens, J.M.“C oal Mineral Analysis: A Check on Inter-Laboratory Agreement,’’ Scanning Microscopy 1990, 4(2), 227-236. Straszheim, W.E.; Yousling, J.G.; Younkin, K.A.; Markuszewski, R. “Mineralogical Characterization of Lower Rank Coals by SEM-Based Automated Image Analysis and Energy-Dispersive X-Ray Spectrometry,” Fuel, 1988.67, 1042-1047. Natusch, D.F.S.; Wallace., J.R. “Urban Aerosol Toxicity: The Influence of Particle Size,” Science, 1974, 186, 696-699. Natusch, D.F.S.; Wallace, J.R. “Toxic Trace Elements: Preferential Concentration in Respirable Particles,” Science, 1974, 183. 202-204. Galbreath, K.C.; Brekke, D.W. “Feasibility of Combined WavelengthlEnergy-Dispersive Computer-Controlled Scanning Electron Microscopy for Determining Trace Metal Distribution,’’ Fuel Processing Technology, 1994, 39, 63-72. Table 1. Composition of Hg-Bearing Particles in Absaloka Ash (With Iodated Activated Carbon Sorbent Added) Particle number Hg Na Mg AI Si S K Ca Ti Fe 1 0.04 0.0 11.0 25.4 10.1 1.8 0.0 49.0 1.7 1.1 2 0.02 0.0 0.0 10.7 12.0 0.0 0.5 1.0 0.0 76.2 3 0.01 0.0 7.4 14.7 25.6 0.0 0.0 52.4 0.0 0.0 I 4 0.03 0.0 4.2 26.7 28.1 4.3 0.0 36.7 0.0 0.0 5 0.03 0.0 10.5 17.0 12.5 6.5 0.0 52.0 0.8 0.7 6 0.07 0.0 6.1 11.9 37.8 0.0 0.0 44.1 0.0 0.0 Normalized composition (wt%, C- and 0-free basis) Avg. 0.03 0.0 6.5 17.7 21.0 2.1 0.1 39.2 0.4 13.0 800 Table 2. Composition ofNon-Hg-Bearing Particles in Absaloka Ash (With Iodated Activated Carbon Sorbent Added) Particle number Normalized composition (wt%, C- and 0-fiee basis) Na Mg Al Si S K Ca Ti Fe 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 0.0 0.0 0.0 2.2 0.0 0.0 1.7 4.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0' 0.0 4.4 0.0 0.0 0.0 0.0 10.5 3.4 6.8 5.7 1.6 2.9 4.3 1.2 6.2 4.2 0.0 0.0 0.0 7.5 6.2 6.9 10.1 2.9 9.7 11.7 10.6 26.7 0.0 20.4 41.9 0.0 9.4 19.3 0.0 19.0 42.6 0.4 12.4 27.7 1.8 20.7 21.8 2.6 27.6 42.2 0.0 18.7 40.7 0.0 13.5 15.0 0.0 22.3 31.6 8.2 20.6 12.1 10.6 18.9 37.9 0.0 33.9 57.2 0.0 31.9 59.8 0.0 0.0 100.0 0.0 27.8 18.3 0.0 14.9 29.5 2.3 14.8 17.3 4.5 18.2 17.2 6.4 21.7 43.1 2.2 11.1 10.3 7.6 14.7 10.1 0.0 0.0 0.0 0.0 0.7 0.0 0.0 1.8 0.8 0.0 2.6 0.0 0.0 1.6 1.4 0.0 0.0 1.2 0.0 0.0 1.3 0.0 0.0 62.7 0.0 37.8 0.0 58.5 1.1 29.5 1.3 49.6 1.8 46.0 0.7 23.1 1.0 21.4 0.0 64.9 1.6 29.4 0.0 50.4 0.0 35.5 1.7 5.9 1.5 4.7 1.0 0.0 0.0 43.3 1.6 41.5 2.6 54.4 0.0 48.1 0.0 22.0 2.4 61.4 0.0 63.6 0.0 0.0 0.0 1.2 0.9 0.0 2.5 1.0 11.4 0.8 4.7 0.0 1.9 0.0 1.2 0.0 1.5 1.8 2.1 0.0 0.0 0.0 0.0 Avg. 0.6 4.6 18.3 32.8 2.1 0.5 38.8 0.8 1 4 ~ Table 3. Trace Element Content of Pyrite Grains in Illinois #6 Coal Values in ppm Mean Range Element As 310 210-410 Cd ND __- Hg 2680* 2680' Ni 210 140-320 Se 760 530-880 ND = not detected ~g values are for a single pyrite grain 801 Table 4. Trace Element Content of Pyrite Grains in Pittsburgh #8 Coal Particle Values in ppm number As Cd Hg Ni Se 1 2 3 4 5 6 7 8 9 10 273 ND ND 2030 123 532 2900 ND 575 146 ND ND ND 17 ND ND 13 ND ND ND 2660 103 ND 1240 1870 974 ND ND 1040 ND 459 797 1330 149 285 ND ND 90 ND ND ND ND 1810 730 13 1220 1280 , 1950 1120 1650 ND = not detected 802 MORATORY LEACHING BEHAVIOR OF ENVIRONMENTALLY SENSITIVE TRACE ELEMENTS FROM FLY ASH AND BOTTOM ASH SAMPLES C.A. Palmer', R.B. Finkelman', M.R. Krasnow'. C.F. Eble*'. 'US. Geological Survey, National Center, M.S. 956, Reston VA 22092; .* Kentucky Geological Survey, 228 Mining and Minerals Bldg.. Lexington, KY 40506 Keywords: Fly ash, Bottom ash, Trace elements INTRODUCTION The distribution of trace elements in coal combustion residues such as fly ash and bottom ash have received considerable attention.',' Several studies of fly ash have concentrated on relationships of trace elements to fly ash particle ~ i z e ~ *S. ~tu.d ies related to etching6, mineralogical transformation during combustion' and leaching have also been reported. Dudas' Conducted long-term leachability studies. Grisafe et al.' examined leachability of fly ash as a source of Se contamination. Fernandez-Turiel et a1.l' have looked at the mobility of heavy metals from coal fly ash. The objectives of these studies were primarily to understand Potential problems associated with the storage or disposal. To meet these objectives, the Solvents used in these studies were chosen to emulate conditions in nature. The leaching study presented in this paper differs from previous leaching studies because the primary objective was to obtain information on modes of occurrence of trace elements in the fly ash and bottom ash and provide data which could be compared to previous studies on the leaching behavior on whole coal samples" . Although preliminary data for 29 elements in the fly ash and bottom ash are available at this time, only results for environmentally sensitive trace elements and other related elements will be discussed in this paper. These elements include those identified in 1990 Clean Air Act Amendments: Co, Cr. Ni. Sb. and radionuclides (Th and U). Fe was also studied because of its importance to coal cleaning and S removal, and Zn because of its relationship to Cd. EXPERIMENTAL The samples were collected from an electric utility power plant having boilers burning high sulfur (3.3 weight percent total sulfur) and low sulfur (0.9 weight percent total sulfur) coal. Approximately 10 grams from each of two fly ash samples and two corresponding bottom ash samples were subjected to sequential leaching. In this procedure each sample was automatically shaken for 18 hours, centrifuged, and the leachate filtered. The samples were first leached with 1N ammonium acetate (NH,CzH,02) . A representative 0.5 gram split of each of the leached samples was reserved for analysis by instrumental neutron activation analysis. This procedure was repeated using 2N hydrochloric acid (HCI), concentrated (48 to 51 %) hydrofluoric acid (HF) and 1.5 N nitric acid (HNO,) and a representative 0.5 gram split was obtained for INAA from the material leached by each solvent . All resulting splits and representative samples of the original material were irradiated for 8 hours at a neutron flux of about 2 x lo'* neutronslcm'sec' using instrumental neutron activation analysis (INAA) procedures similar to those of Palmer." The data was calculated using the SPECTRA program.', The mass of each of the splits used to calculate percent material leached and the concentrations for each of the splits determined by INAA were used to calculate the percent of each element leached by each solvent. The proportion of an element leached by a specific solvent is an indicator of the elements' mode of occurrence. In Contrast to coal, which is primarily an organic matrix not leachable to a significant extent by most inorganic solvents, the bottom ash and fly ash are mainly silicates which are leachable to a large degree by inorganic solvents, particularly by HF. In addition, because of the high temperature of combustion (-1500 "C) phases present in the coal such as clays, carbonates, and sulfides have also been transformed to silicates and oxides. Table 1 shows the percent of the material leached by each of the solvents used in this study. The total amount of material leached ranged from 78 to 99 percent, with 97 percent or more leached from the fly ashes. Seventy to seventy-nine percent of all samples was leached by HF. Clearly a large percentage of the fly ash and bottom ash are in the silicate phases. Generally less than 5 percent of the fly ash and bottom ash is ammonium acetate soluble (probably water soluble as well). Less than 5 percent Of the bottom ash and fly ash is HCI soluble. About 5 to 15 percent of the material was leached by nitric acid. Because sulfides are not likely to be present in the fly ash or the bottom ash (as discussed above) it is not clear which mineral forms were leached by nitric acid. It is possible that species soluble in the nitric acid, unleached by HF, and encased in the Silicates during combustion could have been leached only after the destruction of the silicates. It should be noted that the fly ash is generally more soluble in the solvents used in this study than is the bottom ash. This trend may be explained in part by the presence of a larger proportion of unburned carbon in the bottom ash than the fly ash. Preliminaty results from CHN analyses and ash determinations showed that up to 18 percent unburned carbon was found in the bottom ash in BA3. RESULTS AND DISCUSSION 803 The percentage of some environmentally important elements leached differed from that of the bulk material indicating that their modes of occurrence were clearly different from those of the bulk material. More than 80 percent of the As in the fly ash samples and about 45 percent of the As in one bottom ash sample were leached with HCI. Davidson et ai., suggest that As, as well as some other elements, may be volatilized during combustion and recondensed on the surface of the particles as they cool in the stack. Turner" and EPRI" suggest that As may exist as a metal arsenate, such as Ca,(AsO,), or Ba,(AsO,),. These suggestions explain why As was leached to a large degree by HCI. The behavior of As in BA3 is different from the other bottom ash sample and from the fly ash samples. Condensation of volatile species such as As is unlikely to occur in bottom ash samples. Significant quantities of Sb (Figure 2) are leached by HCI in the two fly ash samples; although the amounts are not as large as those for As. Results from a comparison of magnetic and nonmagnetic fractions16 show similarities in behavior between Sb and As. The results of this study however, suggest that Sb and As behave differently. A few elements, such as U and Th, are leached only to a small degree (as little as 20 percent leached by all solvents). This behavior may be due to their association with minerals such as zircon which are inert and are not significantly altered by either combustion or leaching. Once again these elements are significantly more soluble in fly ash (especially FA1) than in the bottom ash, and U is more soluble than Th. The data for U in fly ash suggests that it may exist in several modes of occurrence because there is roughly equal leaching by HCI and HF in both fly ash samples and equal leaching by HNO, in FA1. Figure 3 shows the percentage of these elements leached by each solvent. Most of the other elements studied show leaching behavior similar to the bulk material. Figure 4 shows the percentage leached for Fe, Ni, Co, and Cr in the bottom ash and the fly ash. In all cases, the majority of these elements are leached by HF. which indicates that they are concentrated in the glassy or crystalline silicates. Most of these elements showed a small amount (e20 percent) of material leached by HCI. Any oxides present are probably locked in the matrix and not exposed until HF destroys the silicates. Figure 5 shows the percent Zn leached (likely an indicator of Cd behavior). The leaching behavior of Zn is similar to the leaching behavior of the bulk material (Table 1). However, there is a significant fraction of Zn leached by HCI in sample FA3. In addition, about 20 percent Zn was leached by ammonium acetate in sample BA3. In summary. most. but not all. elements studied behave similarly to the bulk material and are probably associated with the glassy or silicate portions of the fly ash and bottom ash. Because As, U, Th and possibly Sb (in the fly ash) display behavior significantly different than that for the bulk sample , it can be inferred that they are associated with different minerals or chemical forms than the major elements. Other minor differences in the leaching behavior may indicate that small amounts of that element are associated with minor phases in the ash. Some of these minor phases may be material which has not been completely combusted. REFERENCES (1) Keefer, R.F.; Sajwan, K.S.. Trace elements in coal and coalcombustion Residues; Lewis Pub..:Boca Raton, 1993 308 pages. (2) Eary. L.E.; Rai. D.; Mattig0d;S.V.; Ainsworh, C.C., J. of Environ Qual., 1990,19(2) 202-214. (3) Davidson, R.L.; Natusch. D.F.S.;Wallace. J.R.; Evans, C.A., Jr., fnviron. Sci. andTech., 1974-. 8 (.11), 11 07-1113. (4) Hansen, L.D.; Silberman, D.; Fisher, G.L.; Eatough, D.J. fnviron. Sci Technobl984 18 (3) 181-186 (5) Furuya, K.; Yoshihiro M.; Chiba, T.; Kikuchi. T.. €nviorn. Sci. Techno/. 1987 21 898-903 (6) Heulett. L.D.; Weinberger, A.J. Environ. Sci Technol.1980 14 (8) 965-969 (7) Chinchon, J.S.; Querol. X.; Fernandez-Turiel, J.L.; Lopez-Soler, A.. fnviron. Geol. Sci, (8) Dudas, M.J., Environ. Sci. and Tech., 1981,15 (7) 840-843. (9) Grisafe, D.A.; Angino, E.E.. Smith, S.M., Appl. Geochem, 1988, 3601-608 (10) Fernandez-Turiel, J.L.; de Carvahalho. W.; Cabanas, M.; Querol, X.; Lopez-Soler, A,, hviron. Geol., 1994 23 264-270 (I 1) Palmer, C.A. Krasnow, M.R.. Finkelman. R.B. and D'Angelo, W.M. J. Coal Qual. 1993, (12) Palmer, C.A., Energy and Fuels, 1990,4 (5), 436439 (13)Badeckec P.A.; Grossman, J.N.. The SPECTRA program library: A PC based system for gamma-ray spectra analysis and INAA data reduction, U.S. Geological Survey Open File Rep. 94-168, 1994 47 pages. (14) Turner, R.R.. Environ. Sci. and Tech.. 1981, I 5 (9) 1062-1066 (1 5) Electric Power Research Institute, EPRl TR-104614-V2 Project 3081, 1994 p.Gl-G4. (16) Palmer, C.A.; Finkelman. R.B.; Krasnow. M.R., unpublished data. 1991,18 (1) 11-15 lZ(4) 135-141. 804 Table 1. Weight percentage of material leached by solvents used in this study. Solvent 1 NH~C~H,O, 1 1 5 3 HCI 2 1 5 3 HF 70 71 7a 79 Total 86 78 99 97 BAl BA3 FA1 FA3 HNO, 14 5 10 13 As 100 5 80 2 60 5 40 U a, tu v) c 0 $ 20 a n- BAl BA3 FA1 Sample Leached NH4C2H302 HCI IHF 0H N03 FA3 Figure 1. Percent As leached in bottom ash samples (BAI and BA3) and fly ash samples (FA1 and FA3) by solvents used in this study. Sb 100 80 60 - 40 a U 13 cn c 0 a2, 20 0 BAl BA3 FA1 FA3 Samples Leached NH4C2H302 0 HCI HF 0 HN03 Figure 2. Percent Sb leached in bottom ash samples (BAI and BA3) and fly ash samples (FA1 and FA3) by solvents used in this study. 805 U 120 I I I I I I EA3 FA1 FA3 Samples Leached Th BAI BA3 FA1 FA3 Samples Leached N H ~ C Z H0~ OH ~CI IHF 0H N03 Figure 3. Percent U and Th leached in the bottom ash samples (BAI an the fly ash samples (FA1 and FA3) in this study. Fe co Ni BA3) and - BAl 8A3 FA? FA3 Sern~lsL eached Figure 4 Percent Fe, Co, Ni and Cr leached in the two bottom ash samples and the two fly ash samples by the samples in this study. 806 Zn 120 I 73 1 " BAI BA3 FA1 FA3 Samples Leached NH4CZH302 0 HCI HF 0 HNO3 I Figure 5. Percent Zn leached in the bottom ash (BAl and BA3) and fly ash (FA1 and FA3) by solvents used in this study. / 807 DETERMINATION OF CHROMIUM OXIDATION STATES IN COAL COMBUSTION PRODUCTS BY XAFS SPECTROSCOPY Mohammad Najih, Frank E. Huggins, and G. P. HufFman Department of Chemical and Materials Engineering 341 Bowman Hall University of Kentucky Lexington, KY 40506 Keywords: XAFS spectroscopy, chromium speciation, hazardous air pollutants. ABSTRACT Chromium XAFS spectroscopy has been used to determine the relative amounts of Cr(V1) and Cr(Il1) in ash samples obtained from coal combustion. The method, which is based on the relative heights of the pre-edge peaks for the different Cr oxidation states in XANES spectra, can be used to speciate as little as 50 ppm of chromium in ash. The results indicate that the fraction of Cr(VI) oxidation state present in combustion ash from commercial combustion plants is typically at or close to the detection limit (approx. 3% of the total chromium). Such findings provide justification for a reappraisal of whether or not chromium should be considered a significant HAP in coal combustion. INTRODUCTION Chromium is listed as one of eleven inorganic hazardous air pollutants (HAPS), the so-called "airtoxics", in Title I11 of the 1990 Amendments to the Clean Air Act (I), largely because of the well-known toxicological and carcinogenic properties of the hexavalent oxidation state of chromium (2). This oxidation state is virtually always found in nature and the environment in the form of chromate (CrOa-) or dichromate (Cr20$-) oxoanions. The other common oxidation state of chromium, Cr(IlI), is generally of much less concern to human health, and may in fact be essential in small quantities to mammals. Hence, in assessing potential health hazards posed by chromium in industrial emissions and wastes, it is clearly important that the chromium oxidation state be identified and determined quantitatively. The different oxidation states of chromium in solids or any other state of matter can be readily distinguished in chromium X-ray absorption fine structure (XAFS) spectra by the intensity of the pre-edge peak (3,4,5). The pre-edge feature is generally very weak (typically less than 0.05 times the edge step) for trivalent chromium in an octahedral crystal-field of oxygen anions, whereas for hexavalent chromium oxoanions, the pre-edge peak is usually almost as intense as the edgestep. In this paper, a method for determining the oxidation states of chromium directly in solids is developed based on this difference in pre-edge peak intensity in chromium K-edge XAFS spectra and then the method is applied to the determination of chromium oxidation states in flyash and other products of coal combustion. Owing to the huge tonnages of coal used for electricity generation worldwide, coal combustion is viewed as a major potential source of release of many inorganic HAPS, including chromium, to the environment (6). EXPEFUMENTAL Chromium K-edge XAFS spectroscopy: XAFS spectroscopy is a synchrotron-based technique that provides information about the local structure and bonding around the absorbing element in a material from analysis of the fine structure associated with one of the clement's characteristic X-ray absorption edges (7). For this study, experimental measurements were made at the chromium K-edge at both the National Synchrotron Light Source at Brookhaven National Laboratory, New York, and at the Stanford Synchrotron Radiation Laboratory, Stanford University, CA. Similar experimental practices were used at both synchrotrons. To record the chromium K-edge XAFS spectra, the monochromator was stepped from about 100 eV below the edge to as much as 1,000 ev above the edge and the intensity of the monochromatic x-ray beam before and after absorption by the sample was measured as a function of energy. All spectra were calibrated with respect to the first inflection point in the absorption spectrum of a thin chromium metal foil. This calibration point, which occurs at 5,989 eV, defines the zero-point of energy in the XAFS spectra shown in Figure I and other figures in this paper. The absorption spectra were measured in three different ways depending on the concentration of chromium in the material under investigation. For concentrated samples (Cr > lOwt%), measurements were made in absorption geometry, in which the intensity of the X-ray beam after attenuation by the absorption process in the sample was compared to the incident X-ray intensity These measurements were made with ion chambers. For more dilute samples with chromium contents less than 5 wt% but more than about 0.1 wt% (1000 ppm), the intensity of the fluoresccnt X-rays emitted by the sample in response to the absorption process was measured with a Lytlc detector (8). Finally, for chromium in ash samples, in which the concentration of chromium is very dilute (typically between 50 and 500 808 PPm), measurements were made using a 13-element germanium detector that collected X-rays only in a electronically gated energy interval set for fluorescent chromium X-rays (9). For the fluorescent measurements, a vanadium filter was normally used in association with seller slits to enhance the signalhoise ratio. Spectral scans of about 30 mins were sufficient for most samples, except those for which the chromium content was much less than 500 ppm. Depending on the actual chromium concentration of such dilute samples, up to IO separate scans were accumulated and summed to give a single spectrum. The spectra have been analyzed in a conventional manner that is well described in the literature (7). Basically, the spectra are split into two distinct regions: a near-edge region that includes the fine Structure associated with the edge itself, and an extended fine-structure region that consists ofthe weak oscillatory structure that may persist to as much as 1,000 eV above the edge. These two regions give rise to the X-ray absorption near-edge structure (XANES) spectra and the extended X-ray absorption fine structure (EXAFS) spectra, respectively. The XANES spectrum is generally used as a "fingerprint" to idcntify the form or forms of the element in the material under investigation, whereas the EXAFS region can be mathematically manipulated to yield a "radial structure function" (RSF) from which the local structure around the absorbing element may be inferred. In this work, only the XANES spectra will be discussed fwther as the EXAFS region of the spectrum is not used to determine the chromium oxidation states. Deleminative Melhod: A calibration method for the XANES pre-edge peak was developed by measuring the XANES spectra of carefully prepared mixtures of potassium chromate (KzCrOd) and potassium ehromium(II1) alum sulfate (KCr(S0,),.12H20). Except for different standards in the mixtures, the current method is similar that described by Bajt et al. (4). The mixtures were prepared so that Cr(VI) constituted 0%. 5%. 10%. 15%. 20%. 25%, 50%, 75%, and 100% of the total chromium in the samples. In addition, the total chromium contents of all mixtures were reduced to 4.0 wt% by dilution of the mixtures in boric acid (HBO,). Edge-step normalized XANES spectra of chromium in the boric acid pellets are shown in Figure 1 for all nine calibration mixtures. The spectra are offset vertically to highlight the systematic changes that occur with increasing Cr(V1) content. The pre-edge feature between 0 and IO eV is the spectral feature that shows the most change and it is also the easiest to quantify. As shown by most Cr(1ll) standards, thc pre-edge feature of the end-member K-Cr alum sulfate consists of two peaks: a weak peak at about 1.5 - 2.0 eV and a second, even weaker peak at about 4.0 eV. The chromate pre-edge peak consists of a single intense peak at about 4.0 eV. To quantify these changes, a least-squares iterative fitting program was used that fits the peaks to a mixed lorentzian-gaussian line shape and the background to an arctangent function. This program returns information on the intensity, width, and position of the peaks once the least-squares fitting has converged. These data are summarized in Table 1 for the pre-edge regions shown in Figure 1 and calibration curves were then prepared from the data for the peak at 4.0 eV. The variation of the normalized height of the pre-edge feature with Cr(V1) content was linear with a correlation coefficient (8) in excess of 99% (Figure 2). TABLE 1 Results from Least-Squares Fitting of Calibration Data Peak at 2.0 eV Peak at 4.0 eV Cr(VI)/Total Cr Height Width Area Height Width Area 0 5 IO 15 20 25 50 75 100 0.035 0.033 0.032 0.037 0.034 0.036 0.024 _-- -- 1.855 1.974 2.306 2.527 2.000 2.167 2.000 _- _-- 0.064 0.065 0.073 0.093 0.068 0.078 0.048 -- -_ 0.013 0.042 0.086 0.116 0.170 0.199 0.404 0.620 0.823 1.855 1.983 2.058 1.968 2.070 2.087 2.200 2.200 2.280 0.024 0.083 0.176 0.228 0.35 1 0.415 0.880 1.364 1.876 It should be understood that although the derived calibration curve has an extremely small standard error associated with it (4%Cr ), there are other significant sources of uncertainty that need to be addressed. These include possible variation of the pre-edge intensity with site distortion (IO), thick absorber effects (7), dead-time corrections in the 13-Ge element detector 809 (9). and appropriateness of K2Cr04 as the standard for Cr(Vl) in ash. Such factors were not explicitly considered in the method described by Bajt et al. (4). To circumvent all of these sources of uncertainty, it was decided to use the Cr(ll1) pre-edge peak at lower energy (1.5 ev) to calibrate any possible peak intensity enhancement due to these effects. This pre-edge peak is approximately three times the intensity of the pre-edge peak at a b u t 3.5 - 4.5 eV for most Cr" materials. This relationship can then be used to define a zero Cr(W) baseline that allows for possible experimental saturation effects and site distortion phenomena and, hence, for more precise estimation of the Cr(V1) content. By using this approach for defining the intercept from the normalized height, h,, of the peak at about 1.5 - 2.5 eV, a generalized equation can be derived for the relationship between the normalizcd height, h4, of thc peak at about 3.5 - 4.5 eV and the concentration of Cr(V1) in a sample, as follows: %Cr(VI) = 110 (ha - hJ3) (1) The slope is derived not only from the linearity of the calibration data presented in Table 1, but is an average value that also takes into account the variation in pre-edge peak height exhibited by different chromate compounds. Consequently, any value of Cr(V1) determined from this equation has an uncertainty of up to *IO%, because the probable forms of Cr(V1) in combustion ash samples are likely not well represented by any one chromate compound. RESULTS AND DISCUSSION: We have applied the above equation to measurements made on the Cr XANES spectra to estimate approximate values for Cr(V1) in various ash samples derived from coal combustion. Figure 3 shows the chromium XANES spectra for three commercial and one laboratory ash samples. Parameters (normalized height, width, area, position) for the pre-edge peaks were quantified by least-squares fitting using the program EDGFIT. Examples of the least-squares fitting are shown in Figure 4. The percentage of Cr(VI) present in the sample was estimated from the preedge peak heights using the above relationship (equation 1). The results of such fitting are summarized in Table 2 for all ash and slag samples examined. Based on the spectra shown in Figures 3 and 4 and the results listed in Table 2 derived from least-squares fitting of the pre-edge peak, none of the fly-ash or bottom ash samples from either commercial coal combustion plants or laboratory experiments appears to contain significant C O I ) present in the samples, with the possible exception of the Pittsburgh drop-tube sample. All the results showcd that the determined Cr(V1) content was typically around I - 5% of the total chromium. However, there is an estimated experimental uncertainty of *3 - 5% in such determinations from uncertainty in the determined heights of the peaks in the fitting procedure. Hence, 0% Cr(V1) is almost as significant a result as the determined value in many instances. TABLE 2 Results from Least-Squares Fitting of Cr XANES of Combustion Ashes Peak at 2.0 eV Peak at 4.0 eV Estimated Content of Ash Sample Height Posh Height Pos'n %Cr(VI) Commercial: Cooper FA 0.044 1.5 0.030 3.8 2 NIST SRM 1633b 0.044 1.4 0.038 3.8 3 LET FA-I 0.050 2.4 0.034 4.1 2 LET BA-I 0.054 1.5 0.055 3.9 4 LET BA-2 0.040 1.6 0.035 3.3 3 LGE FA 0.042 1.9 0.054 3.9 5 LGE BA 0.053 2.1 0.033 4.1 1 5 1 -14-Come 0.063 2.3 0.061 4.8 3 5 I-14-Med 0.045 2.0 0.041 4.4 3 5 I - 14-Fine 0.040 1.8 0.046 4.2 4 47-07-Come 0.039 1.9 0.020 4.4 1 Pittsburgh Ash 0.031 1.7 0.091 3.6 9 Illinois #6 Ash 0.036 1.9 0.047 3.7 4 Laboratow: Univ. Arizona Cornbuster: PSIT Drop Tube: 810 h e Pittsburgh drop-tube sample (DECS-12) exhibits a value for Cr(VI), 9 * 3%, that is Significantly higher than those determined for other ash samples. It should be noted that the Cr UNES spectrum (Fig. 3) for this sample was also one of the best quality so that this bigher value can not be due to larger than normal experimental uncertainty. It is likely that this result Can be attributed to the fact that the drop-tube experiments are normally carried out in excess air in comparison to conditions in the larger-scale combustcrs. Hence, a slight enhancement in the cr(VI)/Cr(llI) ratio may not be unusual given such circumstances. However, this observation Would also appear to imply that conditions of coal combustion are not far removed from those that could result in significant formation of Cr(V1): Unusual furnace conditions (eg. low temperatures, high oxygen fugacity) or possibly unusual slag chemishy may yet be found that result in the formation of significant Cr(V1) in combustion ash materials. CONCLUSIONS A direct and nondestructive method has been developed for speciating chromium in solid samples based on the normalized peak-height of the pre-edge feature in chromium XANES spectra. The method is capable of determining the relative percentages of the two major chromium oxidation States, with an uncertainty of *IO%, down to as little as 5-10 ppm of chromium in relatively Xmy transparent solids such as combustion ash or coal. The only preparation necessary is to ensure that the sample has a particle size less than about 200 mesh (0.075 mm top size) and that it is representative over an X-ray beam spot size of between 4 and IO mm’. No chemical Separation is done on the sample nor is any method of prc-concentration used prior to analysis. This spectroscopic method shows that the Cr(V1) content of all commercial ash samples so far examined is at or below the detection limit for Cr(VI), estimated to be about 3 - 5% of the total chromium, depending on concentration. These results are in agreement with data for fly-ash samples determined by ICP-AES. in which the Cr(V1) is complexed and extracted by ammonium hydroxide (1 I). Such findings imply that the behavior of chromium in coal combustion should be re-examined carefully to assess whether or not this element is a significant HAP. However, as the current results indicate for ash samples from small-scale laboratory combustion experhents, typical combustion conditions appear. to be quite close to those that could promote formation of significant Cr(V1). ACKNOWLEDGEMENTS: This rcsearch was supported in part by the Electric Power Research Institute, Palo Alto, CA, and in part by the U.S. Department of Energy and State of Kentucky, through the DOE/KY/EPSCoR program. One of us (M.N.) would like to thank BAPPENAS for providing financial support. We also acknowledge Drs. 1. Zhao and N. Shah (Univ. Kentucky) for assistance with the XAFS measurements, Dr. John Wong (Univ. Louisville) for providing a sample of the NlST fly-ash SRh4 1633b, and the US. Department of Energy for its support of synchrotron facilities in the U.S., without which this work would not have been possible. REFERENCES: United States Public Law 101-549, Nov. IS, 1990; Superintendent of Documents, U.S. Government Printing Office: Washington, DC, 1990, 314 pp. Tietz, N., (Ed.) Clinics[ Guide ID Laboraiory Tests, 2nd. ed.; Saunders: Philadelphia, PA, 1990; Baruthio, F. Eiol. Trace Elem. Res., 1992, 32, 145-153. Huggins, F. E.; Shah, N.; Zhao. J.; et al. Energy & Fuels, 1993, 7, 482-489. Bajt, S.; Clark, S. B.; Sutton, S. R.: et al. Anal. Chem., 1993, 6s. 1800-1804. Lytle, F. W.; Greegor, R. B.; Bibbins, G. L.; et al. Corr. Science 1995, 37, 349-369. Clark, L. B.; Sloss L. L. Trace Elements; IEA Coal Research Report, lEACW49, London, 1992. Lee, P. A.; Citrin, P. H.; et al. Rev. Mod, Phvs. 1981,53,769-806; Konigsberger, D. C.; Prim, R. X-ray Absorption Spectroscopy; J. Wiley & Sons: New Yo& W, 1988. Lytle, F. W.; Greegor, R. B.; et al. Nucl. Instrum. Meth. 1984, A226, 542-548. Cramer, S. P.; Tench, 0.;et al. Nuel. Instrum Meth. 1988, A266, 584-591. Wong, J.; Lytle, F. W.; Messmer, R. P.; Maylotte, D. H. Ph.vs. Rev. E. 1984, 30, 5596- 5610; Waychunas, G. A. Amer. Mineral., 1987, 72, 89-101. Hwang, J. D.; Wang, W-J. Appl. Spectrosc., 1994, 48, 11 11-1 117. 811 Energy, eV Figure 1 : Chromium XANES spectra for mixtures of K-Cr(1ll) alum sulfate and K2Cr(V1)04. 3 , 0 -20 0 20 40 60 Energy (eV) Cr(Vl)/[Cr(Vl)+Cr(lll)J Figure 2: Calibration curve based on normalized peak height of least-squares fitted pre-edge peaks in Figure 1 Illinois #6 PSI Drop-Tube c 0.2 L1 2 .- e 4 N 0.1 .-- 2 zB 0 :4 -2 0 2 4 6 8 Energy, eV Figure 3: Chromium K-edge XANES spectra Figure 4: Examples of least-squares fitting for three commercial ash products and a of the pre-edge peak present in Cr XANES laboratory drop-tube ash. spectra of ash samples. 812 SELENIUM SAMPLING AN0 ANALYSIS IN COAL COMBUSTION SYSTEMS Matthew S. DeVito and Rachel J. Carlson CONSOL Inc., Research & Development 4000 Brownsville Road Library, PA 15129 Keywords: Selenium, Coal Trace Analysis BACKGROUND The Clean A i r Act Amendments of 1990 (CAAA) i d e n t i f i e d 189 elements and compounds that are classified by the U.S. EPA as hazardous a i r pollutants (HAPS): Among these are eleven inorganic trace elements found in coal. A provision o f the CAAA required EPA t o conduct a study of the health and environmental impacts o f HAP emissions from e l e c t r i c u t i l i t y generating units. EPA has completed a number of draft documents i n compliance with t h i s mandate. For trace element emission estimates, they have r e l i e d on a number o f f i e l d tests which were conducted by a variety of organizations including the U.S. Department o f Energy (DOE). The DOE program u t i l i z e d the EPA Method 29 sampling t r a i n t o measure the emissions Of trace elements including Se. EPA Method 29 i s validated f o r municipal waste combustor sampling but not f o r coal - f i r e d combustion sources. The DOE program involved measurements at eight coal-fired u t i l i t i e s selected t o represent a cross-section o f the coal-fired u t i l i t y industry i n regard to fuels and furnace configurations. A l l o f the test teams reported low material balance closures f o r Se.' CONSOL R&D participated at two o f these test sites: Minnesota Power Clay Boswell and I l l i n o i s Power Baldwin stations. The Se balfnce closures f o r the Boswell plant ranged from 12% t o 21% and averaged 18.5%. The Se balance closures f o r the Baldwin plant ranged from 30% to 60% and averaged 50%. Selenium i s the only element that showed a material balance closure problem f o r both t e s t sites, indicating e i t h e r a sampling or analytical e r r o r . A t the t h i r d DOE A i r Toxics Working Group Meeting, the poor Se balances obtained from the e i g h t s t a t i o n t e s t s were discussed, but there were no clear answers as t o the cause. The fact that a l l of these programs showed low Se balance closures i s evidence o f a sampling or analytical problem. After reviewing these results, CONSOL R&D conducted a sampling and analytical program t o determine the reasons f o r the poor Se material balances. This program focused on two areas: 1) the accuracy o f sampling and analytical procedures f o r measuring Se i n solids, and 2) the potential f o r Se losses w i t h i n the combustion or sampling system. Selenium Properties Among the eleven trace elements l i s t e d as HAPS, Se has unique v o l a t i l i t y characteristics that could r e s u l t i n sampling problems. A l l o f these eleven elements except mercury (Hg) and Se are predominantly (>99%) i n the s o l i d phase at coal-fired f l u e gas For these non-volatile elements, f l u e gas sampling i s not required t o complete a material balance. Because o f i t s vapor pressure, almost a l l o f the Hg released during combustion should be present as a vapor. The equilibrium vapor pressure curve (Figure 1) f o r Se (as SeO,) indicates that t h i s element can be przsent in both the gas and solid phases at normal u t i l i t y flue gas temperatures. The curve shows that there can be a large change i n the p a r t i t i o n i n g o f SeO, between the gas and s o l i d phases i n the temperature range of 200 'F t o 300 'F. This temperature range i s important because it encompasses the typical flue gas exhaust temperature for u t i l i t i e s (-280 'F t o 300 'F) and the operating temperature of the EPA Method 29 probe and f i l t e r (258 'F f20 *F). The Se content in the I l l i n o i s coal f i r e d at the Baldwin plant was 4 ppm (whole coal basis). If a l l the Se i n the coal v o l a t i l i z e d during combustion, t h i s would result in a gas phase Se concentration o f approximately 97 ppbv. As the f l u e gas cools, some fraction of the gas phase Se would condense. Table 1 shows the theoretical d i s t r i b u t i o n of Se between the vapor and condensed phases at various temperatures. Selenium i s the only Clean A i r Act trace element that undergoes t h i s phase t r a n s i t i o n i n t h i s temperature window. The implication o f t h i s phenomenon on Se sampling results i s discussed below. Five sampling teams performed the testing. Selenim i n U.S. Coals There i s a l i m i t e d amount of information on the Se contents o f commercial (i.e., as-fired) coals. CONSOL has collected trace element data on over 250 coal samples representing a wide cross-section of U.S. coal production. This database shows a Se-in-coal concepration range of 0.5 t o 6.5 ppm (whole-coal basis) with an average o f -1.5 ppm. The recent DOE program involved nine coals with Se concentrations between 0.85 ppm and 3.25 ppm. I n a DOE-sponsored coal analysis 813 round robin study conducted by CONSOL R&D, Se determinations f o r a NIST reference coal ranged from 0.75 ppm t o 1.52 ppm compared to a c e r t i f i e d value o f 1.29 ppm. Accuracies ranged from 42% low to 15% i i g h . Only one o f the ten reported values was within 10% o f the c e r t i f i e d value. The d i f f i c u l t y in obtaining an accurate Se-in-coal determination a t concentrations typical f o r coal i s certainly a contributing factor t o the uncertainty i n material balance closures. Trace element emission factors f o r combustion sources are developed by using the trace element concentration in the fuel and calculating a maximum uncontrolled emission rate. This value then i s adjusted t o account f o r bottom ash-to-fly ash p a r t i t i o n i n g , p a r t i c u l a t e - to-gas p a r t i t i o n i n g , and removal in control devices. In many cases these p a r t i t i o n i n g factors are estimated from the best available t e s t data. If possible, the estimated emission factor i s compared with emission measurements. The phase d i s t r i b u t i o n o f Se makes estimation o f p a r t i t i o n i n g and removal factors d i f f i c u l t and uncertain. The d i f f i c u l t y i n closing Se balances around coal-fired power plants leads to uncertainty i n the v a l i d i t y o f the measured emissions and estimated emission factors based on these measurements. The accuracy o f emission estimates i s important because they ultimately w i l l be used i n r i s k assessments. RESULTS AND DISCUSSION This research program was focused on two areas o f concern: Analysis of selenium i n process stream samples, Se losses i n the flue gas ducts and EPA Method 29 sampling t r a i n . Analysis of Selenium i n Process Stream Samples There are three factors t h a t contribute t o qood material balance closures: obtaining a representative sample, accurately measuring the process stream flowrate, and an accurate chemical analysis. Assuming that the f i r s t two conditions are met, the chemical analysis becomes the most important step. However, the determination o f selenium in process stream samples can be d i f f i c u l t . Table 2 shows the r e s u l t s of Se analyses conducted on a NIST coal ash standard. These data show that the digestion step outlined i n Method 29 procedures may not be suitable f o r a l l s o l i d materials. The Method 29 digestion (SW 846) involves the digestion o f -0.5 g of s o l i d s w i t h 6 mL o f concentrated HNO and 4 mL of concentrated HF and e i t h e r conventional heating in a Parr Bomb a% 285 'F ( s i x hours) or microwave heating. This digestion showed a low recovery f o r Se and f o r a l l of the HAPS elements. The CEM microwave procedure involves a multi-stage digestion using the same acids outlined i n the Method 29 technique, but with larger volumes and longer digestion times. This technique showed a very good Se recovery. The open-vessel technique showed low recoveries f o r Se, although previous analyses o f t h i s ash standard have shown excellent recoveries f o r Se and the non-volatile trace elements. The low Se recoveries specific t o t h i s determination are thought t o be a r e s u l t o f uncontrolled fluctuations in the temperature used i n the digestion. Because o f the low results f o r Se by open vessel digestion, CONSOL R&D analyzed a variety of solids f o r Se by f i r s t preparing the sample using hydropyrolysis. I n t h i s procedure, the solids are pyrolyzed in a stream o f excess a i r and steam. The v o l a t i l e Se i s passed through a condenser and then i n t o a NaOH scrubber solution f o r Se capture. This solution i s analyzed by ICP-MS. The efficiency o f t h i s procedure has been v e r i f i e d by the analysis of SARM, NIST, and NBS standards. The open-vessel digestion technique has several advantages. It i s safer than the microwave technique, more t i m e - e f f i c i e n t than the other procedures, and provides excellent elemental recoveries for most o f the trace elements of interest (Hg determinations are obtained using a separate sample preparation technique). This work indicates that Se may be l o s t during the open vessel digestion step and additional work i s being completed t o determine the c r i t i c a l digestion temperature for t h i s procedure f o r a variety o f coal ash matrices. Conclusions drawn from these data are that the Method 29 procedure does not provide a s u f f i c i e n t l y rigorous digestion f o r coal ash samples. Typical coal fly ash has a strong c l a y - s i l i c a t e matrix which requires either a more rigorous digestion o r l a r g e r q u a n t i t i e s o f the acids. The same c r i t i c i s m applies t o the analysis o f the Method 29 s o l i d fraction. These data indicate the f r o n t - h a l f f i l t e r analysis can be biased low, which would lead t o inaccurate material balance closures. Selenium Losses i n the Flue Gas Ducts and Samplins Train Because the Se analyses o f the coal, ash, and Method 29 f r o n t - h a l f samples could be i n error, Se material balances from the sampling programs at the Baldwin' and Boswel13 plants were recalculated based on analyses obtained using the hydropyrolysis digestion techniques for the process stream (coal and ash) samples. The Method 29 samples were not available f o r repeat analyses. The ash samples showed somewhat higher Se concentrations, but the increase had only a small 814 effect on the Se balances. The selenium balances f o r the Baldwin testing are shown i n Table 3. These data indicate the Se material balance closures are low by -5M. The Se input value i s based on the Se in the coal which averaged 3.73 ppm (whole-coal basis) f o r these tests. This analysis was v e r i f i e d as part o f the DOE round robin which involved a comparative analysis by f i v e labs. The Se values i n the ESP ash samples were v e r i f i e d through replicate analyses and comparison with standard reference materials. The temperature o f the f l u e gas entering the ESP was -340 'F and -330 'F at the sampling location. The vapor pressure curve f o r SeO, at these temperatures indicates that a l l o f the available Se should have been present i n the vapor state. This i s supported by the low level o f Se i n the ESP ash samples. The Method 29 procedure c a l l s f o r a f r o n t - h a l f (probe and f i l t e r ) temperature o f 258 'F t20 'F. The vapor pressure curve a t 250 'F predicts a gas phase Se concentration o f 8 ppbv. This value i s very close to the observed values (4, 6, and 7 ppbv). A possible explanation f o r the poor Se balance f o r t h i s u t i l i t y i s t h a t a t the Method 29 f r o n t - h a l f sampling temperature (258 'F i20 'F), the equilibrium between gas phase and s o l i d phase Se i s shifted t o the s o l i d phase. I n reviewing the f i e l d sampling sheets, it was noted that the normal variations i n the heater box gave temperatures as low as 240 'F. As shown i n Figure 1, the selenium vapor pressure at 240 'F corresponds t o a gas phase Se concentration o f only 3.5 ppbv, which i s well below what would be expected at the flue gas temperature. The speciation becomes more severe at lower temperatures and could be aggravated by i n s u f f i c i e n t heat t o the sampling probe. If condensation occurs, the measurement o f the Se emissions becomes a function o f the accuracy o f the f r o n t - h a l f (solid) fraction. For t h i s program, the f r o n t - h a l f analyses were found t o be ~ n r e l i a b l e , ' , ~ and it was assumed that the particulate phase Se was represented by the ESP hopper ash samples. However, the ESP solids were collected a t a point in the gas stream where the gas temperature i s -340 'F. A t t h i s temperature, almost a l l o f the Se i s i n the gas phase. It i s l i k e l y that a s i g n i f i c a n t f r a c - t i o n o f the gas-phase Se condensed in the f r o n t - h a l f o f the Method 29 sampling t r a i n and was unaccounted f o r due to the i n a b i l i t y t o obtain an accurate f r o n t - h a l f (particulate) Se analysis. CONSOL Pilot-Scale Selenium SamDlina Results CONSOL R&D conducted a series o f 12 Se measurements on the f l u e gas from a 1.5 MM Btu/hr p i l o t - s c a l e coal combustor (Figure 2). A l l measurements were taken under t i g h t l y controlled combustion conditions using a constant coal source. The only variable was the flue gas temperature. The gas phase emission results from t h i s test and the associated gas and sampling temperatures were compared. The test with the lowest flue gas temperature (200 'F) also showed the lowest concentrat i o n o f gas phase Se (2.9 ppbv). The test with the highest f l u e gas temperature (335 'F) resulted i n the highest gas phase Se concentration (9.3 ppbv). The percent o f the available Se found i n the gas phase ranged from 11% t o 34% and t h i s value was dependent on the temperature o f the flue gas and sampling equipment. Vapor pressure has an exponential dependence on temperature. However, because the temperatures are w i t h i n a narrow range, a l i n e a r correlation analysis was conducted on the data t o assess the co-variance o f gas phase Se concentrations with f l u e gas and sampling temperatures. The following correlations were obtained from t h i s data set: ~ Gas Phase Se Concentration Correlated to: rz Duct Temperature 0.77 Probe Temperature 0.51 F i l t e r Temperature 0.23 These data show that the gas phase Se i s moderately well correlated with the temperature of the flue gas and (more weakly) with the temperature at which the solids are collected i n the Method 29 t r a i n . These data suggest that the p a r t i t i o n i n g between gas and solid phase i s influenced by these temperatures and supports the mechanisms previously discussed. The data also show that cold spots i n the f l u e gas handling system or the sampling probe can decrease the apparent gas phase Se concentration (Figure 2). A decrease i n temperature between one sampling position t o the next, in the temperature window o f 200 'F t o 300 'F could deplete the vapor phase Se by deposition on the sidewalls or on fly ash solids. CONCLUSIONS * The Method 29 analytical procedure (including SW 846 digestion) shows a low bias f o r most trace elements commonly found i n coal ash, including Se. 815 * Analytical bias (due t o Se v o l a t i l i z a t i o n ) can occur during the sample preparation (digestion) stage. * Se p a r t i t i o n i n g i s influenced by the gas and sampling temperatures. * The Method 29 Sampling procedure can s h i f t the apparent speciation between gas phase and s o l i d phase Se. * Material balance closures can be affected if vapor phase and s o l i d phase samples are taken at d i f f e r e n t f l u e gas temperatures. The simultaneous sampling and analysis o f Se i n conjunction with the other elements as described i n EPA Method 29 may lead t o an inaccurate Se determination. RECOMMENDATIONS This work represents an i n i t i a l step t o a more complete understanding o f Se sampling in coal combustion systems. There are a number o f research areas that should be further investigated to improve t h i s understanding and improve emission measurements. Recommendations f o r future research are as follows: * Conduct comparative M-29 sampling with the f r o n t - h a l f temperature at 258'F and at the actual duct temperature. * Analyze M-29 f r o n t - h a l f Se concentrations by both the 511-846 technique and the hydropyrolysis method. * Investigate more e f f e c t i v e digestion techniques f o r Se analysis o f s o l i d samples . * Conduct a Se balance program around a well-controlled system using the suggested modifications. REFERENCES 1. Personal communication (2/7/95) with Tom Brown, DOE/PETC, Pittsburgh, PA, regarding data presented by the DOE project leaders at the Tenth Annual Coal Preparation, Uti1 ization, and Environmental Control Contractors Conference, 7/18-21/94, Pittsburgh, PA. 2. Final Report f o r Clay Boswell Power Station - Unit 2 (Minnesota Power Company) f o r the Comprehensive Assessment o f Toxic Emissions from Coal- Fired Power Plants, Prepared by Roy F. Weston, Inc., f o r DOE/PETC, DOE Contract No. DE-AC22-93PC93255, July 1994. Final Report f o r Baldwin Power Station - Unit 2 ( I l l i n o i s Power Company) f o r the Comprehensive Assessment o f Toxic Emissions from Coal-Fired Power Plants, Prepared by Roy F. Weston, Inc., f o r DOE/PETC, DOE Contract No. 4. E l l i o t , T.C., (and editors o f Power Magazine), Standard Handbook of PowerDlant Engineering, McGraw H i l l , New York, NY, 1989. 5. Sloss, L. L., and Gardner, C. A., Samolina and Analvsis o f Trace Emissions from Coal-Fired Power Stations, IEA Coal Research Draft Report, 9/94. 6. Perry, G., Chemical Engineer's Handbook, Sixth Edition, McGraw H i l l , New York, NY, 1984. 7. Obermiller, E. L.; Conrad, V. 6.; and Lengyel, J. "Trace Element Content o f Commercial Coals", EPRI Symposium on Managing Hazardous A i r Pollutants: State of the A r t , Washington, DC, 11/4-6/91. 8. Rosendale, L. W.; DeVito, M. S. "Interlaboratory V a r i a b i l i t y and Accuracy of Coal Analysis i n the U.S. Department o f Energy U t i l i t y A i r Toxics Assessment Program", A(LWMA Annual Meeting and Exhibition, Cincinnati, OH, Private communication discussing trace element analysis o f DOE Method 29 s o l i d samples f o r the Baldwin and Boswell t e s t s i t e s from Barry Rayfield (Triangle Labs) t o Barry Jackson (Roy F. Weston, Inc.), 10/12/93. 3. OE-CZ2-93PC93255, July 1994. 6/19-24/94. 9. Table 1. Theoretical Phase Distribution for Se Emissions Solid Phase (Fly ash) Vapor Phase TemPerature,'F ppmwt (a) % of Total 96% 220 130 240 127 260 117 87% 280 90 67% 33% 300 33 24% 74 76% - (a) Based on 10% ash i n coal, 70% bottom ash - 30% overhead ash r a t i o , and no Se i n bottom ash 816 Table 2. Comparison o f Se Results on NIST 1633a *Designates informational values a) b) c) Digestion and analytical procedure described i n M-2g9 Digestion and analytical procedure developed by CEM Corporation' Digestion and analytical procedure developed by CDNSOL R&D2p3 Table 3. Selenium Mass Flowrates f o r the Baldwin Process Streams (unit i s lb/hr*) * The values inside the parentheses indicate the theoretical vapor phase ** concentration i n ppbv i f a l l o f the Se present was v o l a t i l i z e d Values obtained from the Method 29 sampling t r a i n - - 0 A. 200 210 220 250 2.0 250 264 270 2110 290 300 310 320 Gas Temperafurc. Y Figure 1. Vapor Pressure Curve f o r SeD,. A CONTINUOUS EMISSIONS MONITOR FOR TOTAL, ELEMENTAL, AND TOTAL SPECIATED MERCURY Andrew D. Sappey, Ph.D., Kevin G. Wilson, Ph.D., Richard J. Schlager, Gary L. Anderson, and Douglas W. Jackson ADA Technologies, Inc. 304 lnvemess Way South, Suite 1 IO Englewood, CO 801 12 (303) 792-5615 Key Words: Elemental Mercury, Speciated Mercury, Continuous Emissions Monitor ABSTRACT ADA Technologies, Inc., is developing a continuous emissions monitoring system that measures the concentrations of mercury and volatile mercury compounds in flue gas. These pollutant species are emitted from a number of industrial processes. The largest contributors of these emissions are coal and oil combustion, municipal waste combustion, medical waste combustion, and the thermal treatment of hazardous materials. It is dificult, time consuming, and expensive to measure mercury emissions using current, manual sampling test methods. Part of this difficulty lies in the fact that mercury is emitted from sources in several different forms, such as elemental mercury and mercuric chloride. The ADA analyzer measures these emissions in real time, thus providing a number of advantages over existing test methods: 1) it will provide a real-time measure of emission rates, 2) it will assure facility operators, regulators, and the public that emissions control systems are working at peak efficiency, and 3) it will provide information as to the nature of the emitted mercury (elemental mercury or speciated compounds). This paper presents an overview of the CEM and describes features of key components of the monitoring system-the mercury detector, a mercury species converter, and the analyzer calibration system. THE NEED FOR A MERCURY CEM Future strategies for controlling hazardous air pollutants will involve the use of continuous emissions monitoring systems. These systems provide a real-time measure of pollutants being emitted from sources and are needed in terms of assuring compliance with emissions regulations. They can also be used to help facilities operate pollution control equipment at peak efficiencies. Mercury is a pollutant that has been receiving much attention in terms of monitoring and control strategies. The toxicity of mercury has prompted industry and regulators to develop methods to minimize its release to the environment. Continuous monitoring systems will play a key role in assuring that emissions of this hazardous material are minimized. Mercury is emitted from industrial sources in a variety of chemical forms depending on the specific process and flue gas conditions. For example, mercury is known to exist as elemental mercury [Hgo] and as mercuric chloride [HgCI2] in most industrial flue gases that contain mercury.' A knowledge of the relative concentrations of th- . - : c v d m s farm of mercury wil: be ieqliiied fcr gr pollilticn control devices to operate effectively. An example of this principle is given in Table 1 for a coal-fired power plant.2 Current standard testing techniques rely on manual "grab samples" where flue gas is drawn through a series of impinger solutions to collect elemental and speciated forms of niercury.) The collected samples are analyzed in an analytical chemistry laboratory using complex techniques and instrumentation. These field sampling and analytical techniques are cumbersome, labor intensive, and expensive. A I-week comprehensive sampling program can cost in the range of $25,000-$50,000, 818 A continuous mercury monitoring system should address the following needs: Since the optimum control device depends on the specific chemical form of the mercury, an analyzer that can distinguish between the chemical forms is needed to assure effective operation of the APCD. An analyzer is needed that will assure the APCD is working properly An analyzer is needed that can be used to control the feed rate of a process generating the mercury emission. An analyzer is needed that will help assure the public and regulatory agencies that facilities which produce mercury emissions are in compliance with regulatory limits. DESCRIPTION OF CEM In response to the need for monitoring mercury emissions in real-time, ADA Technologies has developed a continuous emissions monitoring system that is capable of measuring total mercury, elemental mercury, and (by difference) total speciated mercury. The system features a sensitive mercury detector, a mercury species converter, and a calibration system. Figure I shows the components in a typical CEM arrangement. The "converter.' is used to change speciated mercury compounds to elemental mercury. When the sample gas is placed through the converter, a measure of the total mercury content of the flue gas is obtained. When the converter is bypassed, only elemental mercury is measured in the gas sample. The difference between the two measurements is the concentration of total speciated mercury content. A heated, non-reactive sample transport line is used to convey the gas sample to the analyzer. Calibration gas is introduced to the end of the sample line in order to assure that the entire sampling system and the analyzer are calibrated as a single unit. DESCRIPTION OF COMPONENTS Mercurv Detector The analyzer uses a unique ultraviolet absorption technique to quantify the mercury. Proprietary optical components are incorporated that provide a measurement sensitivity below I p8/m3 (less than approximately IS0 ppt vlv). The analyzer has a linear response to a concentration of greater than 100 p8/m3. The optical design also eliminates the effects of interfering gases such as sulfur dioxide, Figure 2 shows the analyzer response when elemental mercury was introduced at a concentration of 2.6 p8/m3 (390 ppt v/v). Also shown in the figure is the signal when zero gas was introduced into the analyzer. Based on the peak-to-peak noise level observed, a minimum level of detection (defined as 2x noise level) of 0.39 pgim3 (58 ppt v/v) is calculated. Operation with detection limits as low as 45 ppt have been observed under ideal conditions. 7 he ADA analyzer incorporates a unique optical design that eliminates the effects of interfering gases such as sulfur dioxide. Figure 3 shows the response of the detection system when measuring mercury at a concentration of 8.1 pgim3 (1.2 ppb v/v) in the presence of sulfur dioxide at a concentration of 1000 ppm. Within the uncertainty of the measurement, the analyzer corrects for the SO2 absorption perfectly. l'igure 4 shows the response of the analyzer over a concentration range of 0 to 6 ppb v/v). This range is expec!ed to cover most concentrations expected in coal-fired and municipal solid waste generated flue gases. A dilution probe is used on the analyzer for situations in which high concentrations of mercury are present, such as when monitoring uncontrolled emissions ahead of an APCD. 819 Converter A mercury species converter is another key component of the CEM system. The converter is used to distinguish between concentrations of elemental and "total" mercury found in the flue gas. Since the mercury detector measures only elemental mercury, the converter is needed to change speciated forms of mercury present in the flue gas to elemental mercury. Total mercury is, therefore, measured by passing the flue gas sample through the converter. Elemental mercury concentrations are measured by the CEM when the flue gas sample bypasses the converter. Total speciated mercury is then determined as the difference between the measured total mercury concentration and the elemental mercury concentration. The converter uses a unique design that eliminates the need for expendable chemicals to reduce the speciated forms of mercury. Figures 5 and 6 show the response of the analyzer to two surrogate speciated mercury compounds--mercuric chloride and dimethyl mercury. The test sequence repeatedly injected the mercury species through the converter to allow measurement of the resulting elemental mercury and then bypassed the converter to demonstrate the converter effectiveness. Note the slower response time of the analyzer for the speciated forms relative to elemental mercury. This is due to adsorption of the "sticky" speciated forms of mercury on the walls of the tubing and gas cells. Heat tracing of the gas handling system mitigates this problem to some extent. This becomes important in the final analyzer system as it will limit how quickly one can calibrate the analyzer for speciated forms of mercury and therefore how often this procedure can be done. Qlibrator ADA Technologies developed a calibrator for use with the mercury CEM. The calibrator is based on the use of permeation tubes to provide known and accurate concentrations of elemental mercury and mercuric chloride. These devices are considered primary standards for calibrating continuous monitors and they are used to calibrate ambient air analyzers. ADA has developed a two-channel calibrator--one channel is used to calibrate the elemental mercury detector and the other is used to calibrate the converter. REFERENCES I . Niclscn (IYY3). "Air 'Inxics Control by Spray Dryer Absorption Systems," presented at the EPRl S c c d International Conference on Managing Hazardous Air Pollutants, Washington, D.C., July 13-15. 2. Mcllvaine Company (1992). "Mercury Speciation is lmportant and Doable," Air Pollution Monitoring and Sampling Newsletter. No. 147. January. 3. 'l'iiney. M.I.. (1993). "Update on EPA's Experience with Method 301: Field Validation of Emission Concentrations." I'apcr No. 93-RP-145.01, presented at the 86th Annual Meeting of the Air & Waste Man:igemcnt Associalion, Denver. CO, June 13-18. TABLES Table 1. Mercury Removal Under Different Process Conditions Ash Loading Coal % Mercury Plant to Spray Dryer CI Removed B High Low 23 C High Low 6 G High Low 16 A High Low i4 E Low High 55 H LOW ' High 44 F Medium High 89 D High High 96 820 FIGURES 30- 2.5- 2.0- 1 1 . 5 - j u" 1.0- I" 0.5 - / Elemental Mercury Analyzer Figure 1. Mercury CEM arrangement. /L 0.0 0 10 20 30 40 50 60 70 Time (minutes) Figure 2. Response of the analyzer to 2.6 @m3 of mercury. 12 10 % 6 : 4 I" 2 0 10 al 30 40 Time (minutes) Figllre 3. hlerciiry detector response when measuring mercury in the presence of stdrltr dioxide. 821 Analyzer Response 0 1 2 3 4 6 6 Mercury Concentration (ppb v/v) Figure 4. Linearity of the mercury detector. , Through Converter 33 pg/m3 (5 ppb vlv) Analyzer Response Time Figure 5. Mercuric chloride being converted to elemental mercury. Through Converter A rialyzer Res po tise L 822 PRECOMBUSTION CONTROL OPTIONS FOR AIR TOXICS David J. Akers and Clark D. Harrison CQ Inc. One Quality Center Homer City, Pennsylvania 15748 Keywords: Trace Elements, Coal Cleaning, and Air Toxics Control INTRODUCTION Coal cleaning rcducw the ash and sulfur content of coal by removing ash-forming and sLIlfur-bcaring minerals. Coal cleaning can also rcduce the concentration of most of the clemcnrs named as hazardous air pollutants in the 1990 Amendmcnts to the Clean Air Act lxcause many of thcsc elcmcnts arc associated with mincral mattcr. For example, arscnic is commonly associated with pyritc; cadmium with sphaleritc; chromium with clay mincrals; mcrcury with pyritc and cinnahar; nickcl with millcritc, pyrite, and othcr sulfides; and sclcnium with lcad sclcnidc, pyrite, and othcr sulfides (Finkclman, 1980). Thcrc arc also casw in which some of these clcments arc organically bound. Just as both organic and pyritic sulfur can lx found in the samc coal, thc samc tracc clcmcnt may be both organically bound and present as part of a mineral in the samc coal. Organically hound trace elements are riot removcd by currcntly uscd methods of clcaning coal. Trace clcments removcd by coal cleaning will not Ix rckascd into the atmosphere during combustion. Also, coal cleaning reduces the ash coiitcnt of thc coal and increases the hcating value, reducing transponation costs and increasing lniler ctticicncy. Finally, coal clcaning providcs othcr cnvironmcntal Ixncfits by rcducing the sulfilr dioxidc cmissions potcntial of die coal and thc amount of ash for collection and disposal. As an air tonics control measure, coal clcaning otfers several advantages to utilities. Because physical coal clcaning is a rclativcly incxpnsivc tcchnology, it may prove to be the lowest-cost control option in many cascs. Also, coal cleaning is currcntly thc only cotnmcrcially availnhlc control tcchnology for the highly volatile trace clcment mercury. Finally, removing tracc clements bcforc co~nhustiorir educcs the concentration of thrsc clemcnts in utility solid wastes, rcducing possible long-term cnvironmrntal liahility. TRACE ELEMENT REDUCTION BY CONVENTIONAL CLEANING In tlic US, work hy CQ Inc., Southcrn Company Serviccs, Iiic. (SCS), Consolidation Coal Company (CONSOL), and Bituminous Coal Research Inc. (BCR) has dcmonstratcd that conventional mrthods of coal cleaning can produce largc reductions in the concentration of many trace clcments (Akcrs and Dospoy, 1993; CQ Inc. and SCS, 1993; DcVito et al., 1993; and Ford and Price, 1982). Combincd, thcsc sources providc tracc clrmcnt reduction data from 16 commercial and tcn commercial-scale cleaning tests. This data is summarizcd for arsenic and mercury in Tablc 1. As no attempt was madc to enhance removal of any tracc clcment, thesc results arc reprcscntativc of tracc clcmcnt reductions that occur as a by-product of cleaning for ash and sulfur reduction. The data in Tahlc 1 demonstrate that physical coal cleaning is ctfcctive in reducing the concentration of. thesc two tracc elcments, although the dcgrce of- ctfectivencss varies. For example, arsenic reduction varies 'from 20 to 85 prcent and mcrcury reduction from -191 (an incrcase) to 78 prccnt. Part of the olwxvcd variability in trace clement rcduction is caused by poor analytical prccision. Thc accurate measurement of clcments present in tracc conccntrations in coal is challenging and cvcn wcll qualiticd laboratories can produce faulty rcsults (Akcrs ct al., 1990). Howcver, most of the variability appcars to rclatc to the interactions lxnvcen the total amoiint of mincral mattcr rcmoved by cleaning, the m~thod hy which the coal is cleaned, and the mode of occurrcnce of the tracc clrmcnt txaring-mineral matter. Thc primary cconomic motivc for ckaning coal is to cemovc ash-forming mineral matter to rcduce coal transportation costs, lower ash collection, handling and disposal costs, and increase combustion efficiency. Coals are clcaned to a varicty of ash levcls to mcet local and regional market dcmalids. Thc ash reduction achievcd by a cleaning plant is dircctly related to thc total amount of mincral matter removcd. Not surprisingly, tracc clcmcnt reduction tends to increase with ash rcduction. Howcver, factors other than ash reduction impact the reduction of many elements including thc dcgrcc of lilxration of the tracc clement lxaring mineral and the ability of thc coal clcaning equipment utilized to removr the mineral. 823 Mineral matter occurs in coal in a variety of forms. For example, pyrite, the most studied coahssociated mineral, can occur as anything from a massive frachire till SCVeral centimeters in sizc to discrete euhedrnl crystals a few microns in six. Some conventional coal cleaning oprrations crush the raw coal lxfore cleaning to protect equipment from oversized material and to liberate ash- or sulfur-bearing minerals. While cnishing is minimizcd to avoid producing excess tines, it can lilxrate larger minerals forms. It can also lilxrate trace element-hearing miiieral matter. CQ Inc. performed a washability study of Kentucky No. 11 Seam coal. During this study, a comparison was made of tincnished coal with coal cnished to 9.5 mm topsizc. In this case, additional arsenic lilxration occurs when the raw coal is crushed to a topsizc of 9.5 mm. For example, cleaning the uncnished coal at 90 percent energy recovery produces an 86 percent arsenic reduction, while cleaning the cnlshed coal at the same energy recovery produces a 97 yrcent arsenic reduction. In this example, cnlshing increased the lilxration of the arsenic-lxaring mineral(s) in the coal allowing additional quantities to Ix removed without any sacrifice of energy recovery. The t y of~ eq uipment used in a cleaning plant can also atfrct trace element reduction. Tahle 2 contains a comparison of a heavy-media cyclone and froth tlotation for trace element reduction. In this case, Pratt Seam coal from Alahama was cleaned by both technologies. Here, chromium reduction is roughly proprtional to ash reduction for Imth cleaning devices; however, while mercury is reduced by the heavy-media cyclone, it is increased by froth tlotation. The comparison of froth tlotation to heavy-media cycloning illustrates the concept that physical cleaning processes do not remove trace elements as such, hut rather remove trace element-lxaring minerals. Mercury commonly occurs in coal within the stnicnire of the mineral pyrite. As pyrite is a ver)' dense mineral, it is easily removed by a density-based process such as a heavymedia cyclone. However, cleaning processes such as froth tlotation remove minerals based on surface characteristics. Because coal and pyrite have similar surface characteristics, convcntianal froth tlotation may not provide high reductions of either pyrite or pyrite-associated trace elements such as mercury. TRACE ELEMENT REDUCTION BY ADVANCED CLEANING Advanced coal cleaning techno~ogies may 0 t h advantagcs over conventional technologies in reducing trace elements. Advanced processes typically involve cnishing coal to increase the chance of liberating sulfur-hearing and ash-forming mineral matter, possibly also lilxrating trace element-hewing mineral matter. Also, advanced processes are specifically desiglied to clean tine-sized coal, making them more efficient than conventional processes in removing mineral matter from this material. In aIi evaluation of Sewickley Seam coal, CQ Inc. compared an advanced coal cleaning process developed by Custom Coals International to conventional coal cleaning techniques (Akers and Dospy, 1993). The Custom Coals' process is characterized by several innovative components including a tine-coal heavy-media cyclone separation circuit. A conventional coal cleaning plant using heavy-media v c d s and water-only cyclones was used for comparison. As part of this evnluntion, enensive washability and lilxration tests were perfon+ on the coal. CQ Inc. engineers d~velopcd computer models of a conventional coal cleaning plant and a plant using the advanced process with middlings crushing for lilxration. This information was used to produce a lahoratory-simulated clean coal by combining the appropriate size and density fractions of the raw coal in the proportions predicted by the models to produce both the conventional and the advanced clean coal. The results of this evaluation are prcsenied in Table 3. Conventional cleaning techniques reduced the concentration of antimony, arsenic, chromiiim, colialt, lead, mercury, and nickel and advanced techniques provided a further reduction in all cases except mercury. For example, conventional cleaning reduced the arsenic concentration of the coal from 14 to 7 ppm, while advanced cleaning provided a further reduction to 4 ppm. CONCLUSIONS Coal cleaning techniques are etfcctive in removing ash-forming mineral matter along with many mineral-associated trace elements from coal. Data gathered from commercial and commercial-scale cleaning tests indicate that trace element reduction tends to increase as ash reduction increases. However, factors such as the mode of occurrence of the trace 824 elemcnt-karing mineral and the t y of~ cle aning equipment employed also atfcct trace element reduction. Furthermore, there is some evidence that advanced coal cleaning P ~ C a S C csa n provide higher reductions of some trace elements than conventional ProCCSSCs. Knowledge of the interplay lxtween the characteristics of the trace elementbearing mineral and various types of coal cleaning equipment can be wed to enhance trace e h e n t removal during coal cleaning. REFERENCES f i e r s , D. and Dospoy, R., "An Overview of the use of Coal Cleaning to Reduce Air Toxics", Minerals and Metalltirgical Procrssing, Published by the Society for Mining, Metallurgy, and Exploration, Littleton, Colorado, Vol. 10, No. 3, pp 124-127, August 1993. Akers, D., Strcib, D., and Hudyncia, M., Lalmratory Guidelines and Procedures: Trace Elements in Coal, Volume 5: Analytic Procedures for Trace Elements, EPlU CS-5644, Val. 5, Novemlwr 1990. CQ Inc. and Southern Company Services, Inc., Engineering Development of Selective &glomeration: Trace Element Removal Study, Final Report for DOE Contract No. DEAC22- 89PC88879, September 1993. DeVito, M., Rosendale, L., and Conrad, V., "Comparison of Trace Element Contents of Raw and Clean Commercial Coals," Presented at the DOE Workshop on Trace Elements in Coal-Fired Power System, Scottsdale, AZ, April 1993. Finkelman, R.R., "Modes of Occurrence of Trace Elements in Coal," Ph.D. Dissertation, University of Maryland, College Park, MD, 1980. Ford, C. and Price, A., "Evaluation of the Effects of Coal Cleaning on Fugitive Elements: Final Report, Phase 111," DOE/EV/04427-62, J ~ l y19 82. 825 Table 1. Trace Element Reduction by Conventional Coal Cleaning Seom Centrol App. A Centrol App. E Illinois No. 6 Pittsburgh - A Pittsburgh - B Pittsburgh - C Pittsburgh - D Pittsburgh - E Pittsburgh Upper Freeport Lower Kittonning Sewickley Pittsburgh Pittsburgh Illinois No. 6 Kentucky No. 9814 Pratt/Utley Prott Utley Prott Upper Freeport Upper Freeport Illinois 2,3,5 Illinois 2.3.5 Kentucky No. 11 Kentucky No. 11 Doto CONSOL CONSOL CONSOL CONSOL CONSOL CONSOL CONSOL CONSOL scs scs ECR BCR BCR BCR BCR BCR CQ Inc. CQ Inc. CQ Inc. CQ Inc. CQ Inc. CQ Inc. CQ Inc. CQ Inc. CQ Inc. CQ Inc. Ash Reduction (%) 87 88 87 52 79 82 76 78 a4 24 74 65 69 34 57 51 75 66 43 75 83 86 -61 57 06 90 Arsenic Reduction (%) 58 49 62 68 74 75 83 63 El 40 72 51 61 30 20 46 43 42 29 28 83 85 39 54 66 43 - CONS01 - Consolidation Cool Company SCS - Southern Company Services, Inc. BCR - Bituminous Cool Research App - Appolochion Mercury Reduction (%) 22 39 60 33 50 30 12 41 42 -191 38 25 27 14 12 24 39 22 26 45 78 76 28 50 48 826 -.... Table 2- Equipment Performance Comparison (Percent Reductions) Heavy-Media Cyclone Froth Flotation Ash 70 62 Chromium 63 56 Mercury 26 -20 Toble 3. Conventional and Advanced Cleaning (ppm except where noted) Ash Content (Wt %) Antimony Arsenic Cadmium Chromium Cobalt Lead Mercury Nickel Selenium 29.2 0.80 14.0 0.20 16.07 0.27 14.73 0.16 13.39 1.14 Conventional Cleaning 15.2 0.48 7.2 0.63 8.35 0.24 6.96 0.14 9.13 1.54 Custom Cool Advanced Process 14.0 0.26 3.5 0.34 8.22 0.22 6.16 0.14 8.21 1.24 827 POTENTIALLY HAZARDOUS TRACE ELEMENTS IN KENTUCKY COALS Lori J. Blanchard, J. David Robertson, S. Srikantapura, B. K. Parekh, Frank E. Huggins Department of Chemistry and Center for Applied Energy Research University of Kentucky Lexington, KY 40506-0055 Keywords: Trace elements, coal Cleaning, elemental partitioning INTRODUCTION The minor and major trace elemental content of coal is of great interest because of the potentially hazardous impact on human health and the environment resulting from their release during coal combustion. Of the one billion tons of coal mined annually in the United States, 85-90% is consumed by coal-fired power plants. Pot.entially toxic elements present at concentrations as low as a few pgfg can be released in large quantities from combustion of this magnitude. The 1,990 Amendments to the Clean Air Act listed 12 elements found in coal as being potentially subject to control: Sb, As, Be, C1, Cd, Co, Cr, Pb, Hg, Mn, Ni, and Se. In this study the partitioning of these and other elements during coal combustion and advanced cleaning processes has been investigated. Elemental concentrations were measured in the fractions obtained before and after combustion or cleaning using external beam particle induced X-ray emission (PIXE). PIXE is a rapid, instrumental technique that, in principle, is capable of analyzing all elements from sodium through uranium without chemical interference effects. In practice more than 20 elements are routinely determined with sensitivities as low as 1 pg/g. KwERIl@aTAL Sample Preparation S;pmbustion stud ie. Samples of feed coal, fly ash, and bottom ash were collected from two western Kentucky coal-fired power plants (Plants A and B). Each sample was ground to -225 mesh and dried at 105OC overnight. The ash samples were mixed with dried, high-purity graphite to obtain -30% by weight of ash. Each coal and ash/graphite sample was pressed into a 1 nun x 19 mm pellet. Goal c l e w studies. A sample of run of mine coal from the Kentucky #9 seam was collected at the mine site, and split into subsamples as needed. Each subsample was ground to -325 mesh and a 5% (w/v) slurry was prepared. The slurry was subjected to Denver floatation, and the float fraction was further subjected to hydrothermal leaching using either a NaOH or HN03 solution. temperature, and pressure of the leaching process were varied to ascertain their influence, if any, on the removal of trace elements. The clean coal was dried at 50°C overnight, and pressed into a pellet as described above. Experimental Setup The samples were irradiated with an external 1.6 MeV and 2.1 MeV proton beam. surface, was swept over the target to irradiate a 16 mm diameter area. The sample chamber was flushed with helium at atmospheric pressure to reduce sample heating and charging. X-rays were detected with a Si(Li) detector (FWHM resolution of 160 eV at 5.90 keV) placed at an angle of 45' relative to the incident beam. The irradiation time for each sample was 15 minutes. Figure 1. Similar spectra are obtained from the analysis of fly ash and bottom ash. software. The duration, The beam, at an angle of 23' relative to the sample A typical PIXE spectrum of a coal sample is shown in Data analyses were performed using the GUPIX* PC-based RESULTS 6 DISCUSSION m u s t i o n stUd.i€S. Enrichment factors, shown in Figures 2 and 3, are used to illustrate the partitioning behavior of elements during coal combustion. The enrichment factor, EF, for element X is given by: EF" ash' feed coal [A' ash' [*I feed coal The ratio of the concentration of X in the ash and feed coal is calculated relative to the ratio of the concentration of A1 in the same 828 1E 4 183 1E2 1R1 1E 0 0 200 4 0 0 600 BOO 1000 C h a n n e l Figure 1. Typical PIXE s p e c t r u m o€ coal. A1 Si S Cl. K Ca Ti V Cc Mrl Fe Ni. CIA Zn Ga Ge As Bc Sc F i g u r e 2. Enrichment f a c t o r s tor Plant A. AlSi S C1 KCaTIVCrMnFeNiCuZnGaGeAsBrSr ~ i 3 . ~E n r i c h~me n t ~factoer s fo r P l a n t B. 829 ash and feed coal samples because A1 is known to partition equally between the fly ash and bottom ash. observed in Plant A is consistent with accepted partitioning behavior. Figure 3 illustrates that the majority of the elements were more enriched in the bottom ash than in the fly ash: Ca, Ti, Cr, Mn, Fe, Ni, Cu, Ga, As, and Br. This unusual enrichment in the bottom ash was thought to be due to the addition of tailings from the coal cleaning processes to the bottom ash. Plant operators later confirmed the use of this practice at the plant. Coal clean ina studies. Concentration factors were used to evaluate the effectiveness of the hydrothermal leaching coal cleaning process. The concentration factor, CF, for element X is given by: The partitioning of elements However, different results were obtained from Plant B samples. CF= clean coal (hydrothermal leaching) float fraction (Denver floatation) Thus, a CF < 1 indicates a reduction in the concentration of that element as a result of hydrothermal leaching. A comparison of the CFs obtained using NaOH and HN03 as the chemical leaching agents is shown in Figure 4. The increase observed in the concentration of some elements (i.e. CF > 1) could be the result of these elements being leached from reactor components. The increase may also be due to a contaminated leaching solution. It should be noted, however, that the elements whose concentration did increase are not of significant environmental concern. concentrations for a l l elements except V and Ga. When HN03 was the leaching agent most elements were removed very efficiently (CF < 0.5). The degree to which elements are removed by coal cleaning processes depends to a great extent on their mode of occurrence or chemical association in the coal. Although the exact composition can vary greatly from one coal to the next, generalizations have been made concerning common modes of occurrence for trace elements in coal. 3 , 4 . 5 Mg, Ca, Mn, and Sr have a carbonate association in some coals. This would explain their efficient removal since the solubilities of carbonates increase in acidic solutions. Elements known to have an association with pyrite, Fe, S, As, Zn, Ni, and Ga, all show a significant decrease in concentration. Similarly, a considerable reduction in elements known to be strongly associated with silicates, Si, Al. Mg, and K, was observed. The reduction in Cl and Br concentrations by both NaOH and HN03 treatment could indicate they are present as soluble salts. removed less efficiently by HN03. HN03 was more effective than NaOH in reducing elemental Elements thought to have a significant organic association were In these samples, those elements were Figure 4. Concentration factors for NaOH and HNO3. 830 vn Cr, Ti, and Cu. X-ray absorption fine structure SpeCtrOSCopY of Kentucky #9 coal has indicated a partial organic association for V, Cr. and Ti. Although the association of cu has not been determined in these samples, cu is known to have partial organic associations in other coals. are illustrated in Figures 5-7. temperature of the hydrot.herma1 leaching process showed essentially no improvement in the reduction of elemental concentrations for Some ehnents and only slight improvements for others. Thus, it appears these Variables have minimal impact on the effectiveness of this coal cl.eaning process. The effect of other variab1.e~ in the hydrothermal leaching process Increases in the duration, pressure and SUmaRY The partitioning of elements during coal combustion is influenced by t.he mode of occurrence of the elements in the feed coal, hoiler characteristics, and the volatility of the species present. Therefore it is not unusual to observe differences in the partitioning of elements - . M g A l - S i ' S 'Cl' K 'Ca.Ti'V C r ' ~ . P e . N i ' C u ' Z r ; G a ' A s B r ' S ~ Figure 5 . Concentration factors for djfferent leaching time periods. __...^.. ___ _ ._ . MgAlSl S C1 K .. ... ____.... Ti V CrMnFeNICuZnGaAsBrS. Figure 6. Concentration factors for different pressures. 831 e Sr Figure 7. C o n c e n t r a t i o n factors for d i f f e r e n t t e m p e r a t u r e s . a t d i f f e r e n t c o a l - f i r e d power p l a n t s using d i f f e r e n t feed c o a l . Nevertheless, t h e d i f f e r e n c e s observed i n t h i s study a r e more l i k e l y caused by t h e a d d i t i o n of wastes from coal c l e a n i n g p r o c e s s e s to t h e bottom ash of Plant 8. The v a r i a b l e with t h e g r e a t e s t impact on hydrothermal leaching appears t o be t h e 1.eaching chemical i t s e l f . A s i g n i f i c a n t reduction i n the c o n c e n t r a t i o n of mauy elements was observed with the iise of HN03 as the l e a c h i n g a g e n t . P r e s e n t data suggests ot.her v a r i a b l e s i n the process have on1.y s l i g h t impact on t h e removal ot hazardous elements i n c o a l . Work is ongoing to optimize the overall system t u obtain t h e .Lowest p o s s i b l e elemental. conccntrati.ons. C1, Cr, Mn, Ni, and As were analyzed i n t h i s work. The remaining seven elements not analyzed were p r e s e n t a t levels below the s e n s i t i v i t y oi our experimental system, however f u t u r e work on t h e s e samples w i l l i n c l u d e a n a l y z i n g f o r t h e s e e l e m e n t s u s i n g n e u t r o n a c t i v a t i o n a n a l y s i s . Of t h e 12 “ a i r t o x i c s ” l i s t e d i n t h e 1990 Clean A i r Act. Amendments AcKu- 9 . This work was supported by t h e U. S. DOE and the Kentucky EPSCoR Program. 1. 2. 3. 4. 5. REFERENCES 1992 Keystone Coal I n d u s t r y Manual, MacLeari Hunter P u b l i c a t i o n s , Chicago, IL (1992). Maxwell, J. A.; Campbell, J. L.; Teesdale, W. , J . Nuclear .Instrumentation and Methods 1989, 843, 218. Finkelman, R. B. In Atomic a n d N u c l e a r Methods i n F o s s i l Eneigy Research; Filby, R. H.; Carpent.er, B. S.; Ragaini, R. C., Eds.; Plenum Press Corporation: New York, 1982; 141-149. Finkelman, R. 8. Fuel P r o c e s s j n g Technology 1994, 39, 21. Conrad, V. 8.: Krotcheck. L). S. I n EemenLai Amlysis of Coa.2 and Its By-products; Vourvopoulos, G. G . , Ed.; World S c i e n t i f i c : Singapore, 1992; 97-123. 832 SCREENING OF CARBON-BASED SORBENTS FOR THE REMOVAL OF ELEMENTAL MERCURY FROM SIMULATED COMBUSTION FLUE GAS Brian C. Young and Mark A. Musich University of North Dakota Energy & Environmental Research Center PO Box 9018 Grand Forks, ND 58202-9018 (701) 777-5000 Keywords: Sorbents, Mercury, Flue Gas ABSTRACT A fixed-bed reactor system with continuous Hg" analysis capabilities was used to evaluate commercial carbon sorbents for the removal of elemental mercury from simulated flue gas. The objectives of the program were to compare the sorbent effectiveness under identical test conditions and to identify the effects of various flue gas components on elemental mercury sorption. Sorbents tested included steam-activated lignite, chemically activated hardwood, chemically activated bituminous coal, iodated steam-activated coconut shell, and sulfur-impregnated steam-activated bituminous coal. The iodated carbon was the most effective sorbent, showing over 99% mercury removal according to U.S. Environmental Protection Agency (EPA) Method IOIA. Data indicate that adding 0, at 4 vol% reduced the effectiveness of the steam-activated lignite, chemically activated hardwood, and sulfurimpregnated steam-activated bituminous coal. Adding SO, at 500 ppm improved the mercury removal of the sulfur-impregnated carbon. Further, the presence of HCI gas (at 50 ppm) produced an order of magnitude increase in mercury removal with the chemically activated and sulfur-impregnated bituminous coal-based carbons. I INTRODUCTION Coal combustion and gasification processes together with industrial and commercial operations, such as waste incineration, emit significant quantities of trace elements to the atmosphere each year (I). The 1990 Clean Air Act Amendments have identified eleven trace elements (beryllium, chromium, manganese, cobalt, nickel, arsenic, selenium, cadmium, antimony, lead, and mercury) for control because of their potential harmful effects to the ecosystem. Mercury (along with arsenic and selenium) is of particular concern because it can occur in vapor or submicron fume form, and as such conventional collection devices (precipitators and baghouses) are marginally effective for its removal (2). Trace element control strategies have recently focused on disposable or regenerable sorbents (activated carbons, coke, limestone) that can be injected as powders directly into flue gas streams or utilized in fluid-bed or fixed-bed reactors. However, homogeneous or heterogeneous reactions with other flue gas constituents (HCI. 03 can occur. Identifying and controlling these reactions are important in determining the effectiveness of sorbents to capture particular species, e&, metallic mercury, mercuric chloride, or mercuric oxide. Further, other gases such as carbon monoxide, nitrogen dioxide, and sulfur dioxide have the potential to interfere with the effective sorption of mercury species. The overall objective of the ongoing project is to identify the conditions (temperature and flow rates) and the controlling processes (mercury species and concentration, flue gas components) for the most effective capture of trace elements by carbon sorbents in combustion and gasification systems. EXPERIMENTAL Apparatus and Procedure The mercury sorbent test apparatus consists of four main subsystems: I) flue gas generation, 2) mercury injection, 3) sorbent-flue gas contactor, and 4) effluent gas mercury analysis (with data logging). A diagram of the test apparatus is presented in Figure 1. The simulated flue gas, which can contain N,, O,, CO,, SO,. HCI, and NO,, is generated in a manifold system; rotameters provide volume flow control. Elemental mercury vapors are generated with a permeation tube(s). The permeation tube mercury desorption rate, and consequently, the simulated flue gas mercury concentration, is a function of the permeation tube's N, sweep gas equilibrium temperature. permeation tube temperature control. to within 0.1 "c Of setpoint. is provided by a condensor heated with circulating heat-transfer fluid. A u.S. Environmental Protection Agency (EPA) Method 5 in-stack particulate sampling filter is used as a sorbent bed containment device. The interior of the filter assembly, including filter support grid, and all other components in contact with the mercury-laden gas are Teflon-coated. The filter assembly and influent tubing are electrically heated to maintain the desired temperature and prevent condensation. A downflow Configuration iS used to minimize entrainment of powdered sorbents. The filter static and 833 differential pressures are monitored using pressure gauges. The filter assembly can be equipped with a thermocouple to measure the flowing gas temperature. The elemental mercury concentration in the simulated flue gas stream is continuously monitored using a DuPont Model 400 ultraviolet (253.7 nanometer) photometric analyzer. A Buck Scientific Model 400 cold-vapor ultraviolet analyzer has also been used to monitor the filter inlet mercury concentration. Mercury concentration values from the analyzers are continuously logged to a chart recorder; a data acquisition unit coupled with a lap-top computer has been used to log mercury analyzer output data and select system temperatures. Diaphragm-type and bubble-type gas meters have been used to measure the total gas rate. Sorbents The following five commercial actiiated carbons were evaluated as elemental mercury sorbents: 1) chemical-activated hardwood (ACl), 2) steam-activated lignite ( A a ) , 3) 5% sulfur-impregnated steam-activated bituminous coal (AC3), 4) chemically activated bituminous coal (AC4), and 5) 10% iodine-impregnated steam-activated coconut shell (AC5). The activated carbons were tested as powders; the sulfur and iodine impregnated carbons were obtained in granular form and then comminuted to a nominal 200-mesh (75-micron) top size. Tests Performed Twenty-seven tests were performed using the five sorbents. Test variables included sorbent type, 0, concentration (0 or 4 vel%), SO, concentration (0 or 500 ppm), and HCI concentration (0 or 50 ppm). Common test parameters were as follows: a nominal mercury concentration of 100 pglm', gas rate of 26 scfh, filter assembly gas temperature of 150°C (300"F), and sorbent mass of 0.20 g. The tests are summarized below. Six tests, one each with ACI and AC2 and two each with AC3 and AC5, used 100,vol% N, as the simulated flue gas Five tests, one each with ACl, AC2, AC3, AC4, and AC5, used 4 vol% O,, 96 vol% N, as the simulated flue gas Thirteen tests, two each with ACI and AC2 and three each with AC3, AC4. and AC5, used a simulated flue gas composed of 4 ~ 0 1%0, .9 6 vol% N,, plus 500 ppm SO, Three tests, two with AC3 and one with AC4, used 4 vol% 0,. 96 vol% N,, plus 500 ppm SO, and 50 ppm HCI as the simulated flue gas EPA Method IOlA (3) was applied to the filter assembly influent and effluent simulated gas streams during one test with the AC5 (iodated carbon) in the presence of 0, + SO,. This test was performed to quantify total mercury removal by the carbon and to compare the result against the general trend of the ultraviolet analyzer output. Further, the test was applied to assess if elemental mercury was being converted to an oxidized form in the presence of AC5, and thus not adsorbed by the carbon or detected by the ultraviolet analyzer, but collected by the permanganate solution of EPA Method IOIA. Similarly, EPA draft Method 29 (4) was applied to the filter effluent scream in the single test with the AC4 (chemically activated bituminous coal) in the presence of 0, + SO, + HCI. Similarly to the test with AC5, this test was applied to assess if elemental mercury was being converted to oxidized andlor chloride forms in the presence of AC4. With this test, chloride forms of mercury would be collected in the peroxide solution. The test duration for each EPA method was one half-hour. The EPA Method lOlA permanganate solution and draft Method 29 peroxide and permanganate solutions were analyzed by cold-vapor atomic adsorption using a Leeman Labs PS200 automated mercury analyzer. RESULTS AND DISCUSSION The effluent gas from the filter assembly was monitored for < 100% mercury capture and 0% mercury capture (breakthrough). Adsorption curves, which show the mercury removal efficiency as a function of gas-sorbent contact time, are presented in Figures 2 and 3 for tests conducted with 0 and 4 ~ 0 1%0, . respectively. Tests at 0 vol% 0, indicated that ACI and AC3 each exhibited an instantaneous lowering of Hg" removal efficiency to 46% and 10%. respectively. Breakthrough with these respective carbons was achieved in approximately 4 and 24 minutes. The AC2 exhibited a slower loss of mercury removal efficacy, achieving breakthrough in approximately 30 minutes. Tests with 4 ~ 0 1%O 2 indicated that ACI. AC2. and AC3 showed similar instantaneous losses of mercury removal efficiency but with more rapid attainment of breakthrough, 0.5, 18, and 10.5 minutes, respectively, than tests without oxygen. The AC4, first used in tests with 02.ex hibited superior mercury removal efficiency relative to the AC1. Ac2, and AC3 sorbents, achieving breakthrough after 94 minutes. 834 The AC5 icdated carbon appeared to be vastly superior to the other carbons in tests with and without 0,. The baseline analyzer output indicates that elemental mercury was 100% adsorbed when using 0 vol% oxygen after over 20 hours; a replicate test produced identical results. As was observed with AC5 in tests without O,, elemental mercury was 100% adsorbed even after 112 minutes, a test duration almost 20 minutes longer than the next most effective sorbent. The addition of SO, appeared to have a selective influence on mercury removal efficiency relative to that Of 0,. A plot of the sorbent contactor effluent gas mercury concentration is shown in Figure 4 for tests performed using 4 ~ 0 1%0 , and 500 ppm SO, combined. Trends are similar to those from tests performed without SO, in that ACl and AC2 are the least effective sorbents, showing an instantaneous loss in removal efficiency and the most rapid attainment of breakthrough. Similarly to tests without SO2, AC5 (iodated carbon) retained essentially 100% removal efficiency. However, AC3 showed a slower loss of effectiveness relative to ACI and AC?, with a breakthrough time 50% longer than that with AC2. The EPA Method lOlA test using AC5 indicated that elemental mercury was removed by this carbon at a high level of effectiveness. The mass concentration of mercury in the effluent and influent permanganate solutions, 78 pg and 0.2 pg per one-half liter, respectively, indicated that mercury removal was over 99 wt%, agreeing well with analyzer output data. However, the sorption data or its analyses do not provide evidence of any conversion of mercury to oxidized form. The results of tests performed with AC3 and AC4 using 50 ppm HCI indicated evidence of interaction or reactions that enhance mercury removal efficiency. A monitoring plot of effluent gas mercury concentration as a function of gas contact time (or total mercury flowed) is presented in Figure 5 for a test performed with AC3. The sawtooth curve shows the change in mercury concentration, and presumably mercury removal efficiency, effected by starting and stopping the HCI gas flow. The straight baseline, which indicated nearly 100% mercury removal with flowing HCI, contrasts with the curve for AC3 in Figure 4, generated without HCI. During the periods without HCI injection, the mercury concentration curve exhibited a similar, slow degradation in mercury removal as seen in Figure 4. Upon injection of HCI, the return to essentially 100% mercury removal was immediate. A replicate test with AC3 and a single test with AC4 using 50 ppm HCI produced similar results. The EPA (draft) Method 29 with AC4 showed that elemental mercury was removed at a high level of effectiveness. The mass concentration of mercury in the effluent peroxide and permanganate solutions, 0.3 and 1.9 pg per one-half liter, indicated that mercury removal was over 97 wt%, agreeing with the analyzer output. The quantitation of mercury in the peroxide trap funher suggests that chloride forms of mercury were produced, and, as such, were removed by the AC4. CONCLUSIONS The AC5 (iodated) activated carbon appeared to be the consistently superior sorbent regardless of the simulated flue gas atmosphere; ACI (chemically activated hardwood) was consistently the least effective. Adding Oz at 4 ~ 0 1%ap parently reduced the effectiveness of all carbons except the iodated carbon. The effect of adding SOz. however, appeared to more selective, increasing the effectiveness of the sulfurimpregnated carbon relative to the other carbons. Adding HCI at 50 ppm had the apparent effect of enhancing the mercury removal efficiency of the sulfur-impregnated and chemically activated bituminous coals to a level comparable to the iodated carbon. REFERENCES 1. Lead, Mercury, Cadmium. and Arsenic in the Environment; Hutchinson, T.C.; Meema, K.M., Eds.; 2. Hall, B.; Schager, P.; Lindquist, 0. "Chemical Reactions of Mercury in Combustion Flue Gases," J. Wiley: Chichester. 1987. Water, Air, and Soil Pollution 1991, 56, 3-14. 3. Method IOlA - Determination of Particulate and Gaseous Mercury Emissions from Sewage Sludge Incinerators; EMTIC EPA, April 12, 1991. 4. Method 29 - Determination of Metals Emissions from Stationary Sources; Federal Register, Vol. 56, ACKNOWLEDGMENTS The authors wish to acknowledge the assistance of the U.S. Department of Energy Morgantown Energy Technology Center, the U.S. Environmental Protection Agency, the Energy & Environmental Research Center's (EERC) Center for Air Toxic Metals, as well as Mr. Grant Schelkoph and Mr. Tim Kujawa of the EERC. No. 137, pp. 32705-32720, July 17, 1991. , 1 835 3 $ .a-, E W Hg Metal Permeation Tube EERC MMI IM2.COR b GkJ ESoxtrebrennat llCy oHnetatcetodr t I Heat-Transfer fluid t I ? Figure 1. Mercury sorbent test apparatus. I . . . . I . . . . I . . . . I . . . . 0 15 30 45 60 75 90 105 120 Sorbent Contact Time, min 0 vol% 0,. Figure 2. Mercury removal efficiency curve, 105 120 Sorbent Contact Time, min Figure 3. Mercury removal efficiency curve, 4 vol% or . 836 *I E m i e . 1 .- I I g 0 $ f r 0 E a c- L . 01 .- I c al C 8 $ f 2 100 - 80 - 60 - 40 - 20 - ACl AC3 AC4 0 ' I I I I I I I I 0 10 20 30 40 50 60 70 80 Sorbent Contact Time, min Figure 4. Comparative effectiveness of activated carbons for elemental mercury sorption; 4 vol% 02 and 500 ppm S02. 80 - 70 ~ 60 - 50 - 40 - 30 - 20 - 10 - 0 20 40 60 80 100 120 Test Duration, min Figure 5. Mercury sorptisn by sulfur-impregnated steamactivated bituminous coal (AC3); 4 vol% 02, 500 ppm SOP. 50 ppm HCI. a31 PRODUCTION OF ACTIVATED CHAR FOR CLEANING FLUE GAS FROM INCINERATORS Carl W. Kruse, Anthony A Lizzio, Joseph A DeBarr and Suhhash B. Bhagwat Illinois State Geological Survey, 615 E. Peabody Dr., Champaign, IL 61820 Keywords: Incinerator flue gas cleaning, activated carbon, dioxins and furans ABSTRACT A granular activated coal char suitable for removing dioxins, furans, mercury, particulate matter, HCI, HF and SO, from incinerator flue gas has been produced from the Colchester (Illinois No. 2) coal. Tests with 250 kilogram of this adsorbent on flue gas from a commercial incinerator in Europe demonstrated that its efficiency for removing dioxins and furans was 99.72% to 99.98%. Mercury concentration in the flue gas after the adsorber was too low to be detected; an el'ficienq for mercury removal could not be calculated. This adsorbent was produced in three steps from 1 mm by 6.4 mm coal obtained from a commercial Illinois washing plant. The projected cost for manufacturing the adsorbent is lower than that of carbon adsorbents commercially available in the United States (U.S.). The estimated break even cost for the adsorbent from an 80,000 ton/yehr plant is $326/ton with a 20% return on investment and a cash flow for 20 years discounted 20% annually. INTRODUCTION The U.S. is expected to follow ;he European lead by imposing low limits for several pollutants from incinerators. The legal emission limits for European waste incinerators were tightened significantly in the late 1980's. Current regulations impose a drastic lowering of the emissions of HCI, HF, SO2, and mercury on all new incinerators (Table 1) [l]. These regulations apply to existing incinerators in Germany and the Netherlands now, and in Austria by 1996. For the first time emissions of dioxins and furans have been targeted, and are not to exceed 0.1 ng Toxicity Equivalents (TE)/m3 Standard Conditions (SC). An activated carbon process, developed in the 1970's by STEAG AG of Essen, Germany, to eliminate SO. and reduce NO. emissions, has also shown high removal efficiencies for inorganic and organic compounds such as HCI, heavy metals, dioxins and furans (Table 2) [l]. The first commercial plant using STEAGs activated carbon technology (/a/c/t") for cleaning flue gas from a waste incinerator began operation in 1991. Other medical, hazardous and municipal waste incinerators with flue gas outputs of 6,500 m3/h SC to 155,000 m3/h SC have been equipped with this process. Four plants were operating in Europe in 1993 using STEAGs /a/c/t"-process and three more will be on line cleaning an additional 1.3 MM m3/h by year's end. STEAG has begun licensing its process in the U.S. Staff members of the Illinois State Geological Survey (ISGS) became aware in 1993 that a U.S. supplier of a carbon adsorbent suitable for STEAGs /a/c/tm-process was needed 111. To be acceptable the adsorbent must pass a NO. self heating test. In this test, the adsorbent is saturated with nitric oxide and the gas flow is discontinued. The temperature rise due to the heat of reaction is measured. It is typically related to the surface area. STEAG's European licensees use adsorbent called German Herdofenkoks that is manufactured from a German brown coal (lignite). This adsorbent has a surface area less than 300 m'/g (N, BET). Carbons commercially available in the US. have higher surface areas and are reported to fail this test. Not only this safety problem but also the prices for U.S. carbons preclude their use in STEAGs once-through process. The Herdofenkoks adsorbent was reported to sell in Europe for the equivalent of $300(U.S.)/ton in May 1995. Mild gasification (MG) of Illinois coals and research on the char that accompanies this process have been active areas of research at the ISGS since the early 1980s [2]. Recent results show that, by selecting appropriate conditions during MG and following MG with a low temperature (< 475°C) oxidation step, a high-sulfur Illinois coal that emits more than 5 Ibs SOJMMBtu can be converted to char that emits less than 2.5 Ibs SOJMMBtu [3,4,5,]. This partial gasification by low temperature oxidation not only lowers the sulfur content but also activates the chars providing a product that has as much as 300 mz/g N, BET surface area [6]. These results encouraged ISGS researchers to believe an adsorbent satisfactory for a STEAG /a/cit" type process could be produced from an Illinois coal without the extensive cleaning to remove ash, extensive preoxidation times ("baking") and the bnquetting that are a part of costly steps required to make high surface area activated carbon. EXPERIMENTAL -Coal Freeman United Coal Company has a size consist of zero by 6.4 mm at one point in its plant that cleans the Colchester seam coal mined near Industry, Illinois. Twelve barrels of this size consist was made available by Freeman United Coal Company for the work described herein. It was spread at about two to four inches depth on the floor to dry overnight before removing the minus 16 mesh (< 1 mm) material by screening. A typical analysis for the Colchester coal appears in Table 4. h-uioment Pound quantities of activated char were produced in a Model RT-472-104c ontinuous feed rotary tube kiln (CFRK), manufactured by the Pereny Equipment Company, Inc., of Columbus, Ohio. The CFRK consists of a 10.2 cm ID, 1.83 m long rotating tube of HX alloy. The center portion of the tube (1.4 m) is heated by three separate electrically heated furnaces. The sample is introduced into the tube using a screw type feeder. 838 A continuous feed ehamng oven (CFCO) was also used in the pyrolysis step because it was better suited to handle large amounts of tar evolved in longer runs. The coal sample was conveyed through a 15.2 cm x 15.2 cm x 69 cm oven on a belt of close-fitting, overlapping, stainless-steel trays (12.7 cm wide) attached to links of a chain drive and heated above and below by tubular electric heating elements. A hopper that controls the bed depth fills the trays as they enter the oven. Evolved gases are removed counter to the direction of the coal and are drawn out an exhaust pipe where volatile materials are burned before being released into the fume hood. All areas outside the heating mne are enclosed in a reasonably tight sheet-metal housing which may be purged with nitrogen to exclude air and avoid loss of char due to burning. Chars were prepared in the CFCO using feed rates of 1-4 pounds per hour, bed depths of 8-20 mm, temperatures of 400-500°C and residence times of 0.25- 0.75 h. A 50 cm LD. x 1.22 m long (19.5" x 4') (active heated length), stainless steel shell, batch rotary kiln (BRK) manufactured by AMs was heated by natural gas burners. It had internal material bed disturbers, bed and gas thermocouples, system thermocouples, and off-gas combustion chamber. Nitrogen or steam purging and a nitrogen cooled sampling probe were available. The maximum operating temperature was 1OOO"C. A 10% loading required 0.83 cubic feet of coal. A 20 cm x 0.91 m (8x36") continuous feed rotary kiln (CFRK) was heated by 7 gas burners monitored by three external and two internal thermocouples. The feeder was an AMs fashioned volumetric belt that fed 13.2-15.4 kg (6-7#/h) and the cooler a 35.5 cm x 2.29 m long (14" x 7.5') kiln. The off-gas combustor was the one described for the BRK A 48 cm x 3 m (18x10), indirect fired, continuous feed rotary kiln (CFRK) had a variable rotational speed, adjustable slope and a high velocity pulsating burner system (4 burners in the first two zones, 3 in the last zone). The Inconel 601 shell, No. 10 gage thickness, had six 2.54 cm high anti-slide bars. Auxiliary equipment included a 38 cm x 3.7 m (lS'x12') rotary cooler (direct air or indirect water), secondary combustion chamber, and a 24 point continuous data logger. This CFRK could accommodate about 200 kgb (9O#/h) feed rate. Single-point BET surface areas of prepared chars were determined from N, (77 K) adsorption data obtained a t a relative pressure (PP,)of 0.30 with a Monosorb flow apparatus (Quantachrome Corporation). The kinetics of SO, adsorption on selected chars was determined using a Cahn TG-131 thermogravimetric analyzer (TGA) system. In a typical run, a 30-50 mg char sample was placed in a platinum pan and heated at 20"C/min to 120°C in flowing nitrogen. Once the temperature stabilized, the nitrogen flow was switched to one containing 5% O2 10% HzO and the balance nitrogen. Once the weight stabilized, the Solwas added in concentrations representative of a typical flue gas (e.g. 2500 ppmv SOz). The weight gain versus time was recorded by a computerized data acquisition system. DISCUSSION OF RESULTS Colchester coal, a high volatile C bituminous coal with a free swelling index of 3 or more, swells, melts, and agglomerates while being charred if it has not been air oxidized. An oxidation step was necessary (1) to maintain approximately the same particle size in the final product as that existing in the 1 mm by 6.4 mm starting material and (2) to retain sufficient initial pore structure to produce the porosity needed in the final product by partial gasification of the char. Surface area is difficult to develop if coal has passed through a melting stage, but loss of particle strength occurs if too much preoxidation occurs. To achieve the desired result in as short a residence time as practical, as high an air oxidation temperature as possible was selected while allowing a margin of safety in avoiding a loss of control due to burning. Conditions of time and temperature during oxidation were selected that gave a preoxidized coal that could be pyrolyzed at 450 to 500°C to provide char that could be activated at 850°C in CO, to increase surface area to 150-250 m2/g (N, BET)(Tahle 3) [7,8]. Observe that decreasing the amount of oxidation either by lowering the temperature or decreasing the time resulted in less surface area. The selection of carbon dioxide for developing added porosity at 850°C was influenced by previous experience [9]. This gas provided a flexibility in temperature control not available with steam. After demonstrating the three-step preparation (preoxidation, pyrolysis and activation) at the ISGS, Allis Mineral Systems (AMs)w as engaged to scale up production in equipment located at its Process Research and Test Center in Oak Creek, Wisconsin [lo]. The first successful production level at Oak Creek was preoxidation in a 50 cm diameter batch rotary kiln (BRK) with pyrolysis and activation in a 20 cm CFRK. The largest scale production of adsorbent at Oak Creek involved performing all three steps in succession in the 48 cm diameter CFRK which accommodated a feed rate as high as 90 Ibs/h. An attempt to conduct all three steps in the batch kiln was unsuccessful; the tar was not removed at a rate sufficient to avoid agglomeration. The importance of removing volatile components as quickly as possible was reinforced during the transition from the preoxidation step to the pyrolysis step in the 48 cm CFRK About 90 pounds of preoxidized coal remained in the kiln when the kiln temperature was increased rapidly to the pyrolysis temperature. Some agglomeration occurred which may have approached 7 wt% of the final product. The Oak Creek kilns had dams at the exit end that maintained a bed occupying 10% of the cross sectional area. The ISGS kiln did not have a dam and the bed occupied considerably less than 10% of the cross sectional area. The thicker bed depth decreased the amount of solids exposed to 839 reaction gas in the preoxidation and activation steps. This meant that each progression to a larger kiln required more time to achieve equivalent products. A comparison of properties of the German Herdofenkoks and the ISGS Colchester adsorbent is presented in Table 5. The lower density of the ISGS Colchester adsorbent reflects a lower level of preoxidation than ideal (some swelling). The lower surface area, 110 m2/g, than those obtained in the lab scale of operation, 151-236 m*/g (Table 3) reflect a combination of less preoxidation and less activation in the larger kiln. The SO, adsorption profiles shown in Figure 1 confirm that the scale up fell short of reaching the full potential of SO, capacity possible with Colchester coal. Additionally, this profile shows a much higher rate of SO, adsorption in the first few hours. The initial SO, uptake is best correlated with active sites, those responsible for the oxidation of SO, to SO3 [ll]. Another factor to be considered in the difference in filling rate is the pore structure. There is reason to believe that the pore structure of the Colchester adsorbent differs significantly from that of the Herdofenkoks adsorbent [ I l l . It remains to be shown how much of this difference in SO, adsorption behavior is due to active sites and how much is related to a special network geometry in which outer macro pores feed into interior micro pores. Of practical importance for use in a STEAG /a/c/t”-process, is the lower rate of uptake of nitric oxide (Figure 2) by the ISGS Colchester adsorbent. The preliminary effort to establish a beak even cost was based on the flow diagram shown in Figure 3. While the estimate of $326/ton (Table 7) for product Cram an 80,000 tonlyear plant is very preliminary, it is encouraging for at least one reason. The flow diagram does not reflect economies in residence time under conditions of better gasholid contact that are known to accompany the use of modified kilns or other types of equipment. The data (Table 6) in STEAGs pilot scale adsorber, including passing the NO. self heating test (not shown), qualify this material for use in STEAGs /a/c/tm-process. CONCLUSIONS Research has demonstrated that Colchester (Illinois No. 2) coal is a promising feedstock for producing an activated char adsorbent for removing pollutants from incinerator flue gas. Not only are the properties of this adsorbent those desired, its estimated break even cost in a dedicated commercial facility, $326/ton, could make it highly competitive for use in cleaning incinerator flue gas in the US. The ISGS Colchester adsorbent is the first adsorbent to be made in the US. from a domestic coal that meets requirements of the STFAG /a/c/t”-process. ACKNOWLEDGEMENT & DISCLAIMER This report was prepared by Carl Kruse of the ISGS with support, in part, by grants made possible by US. Department of Energy (DOE) Coopcrative Agreement Number DE-FC22-92PC92521 and the Illinois Coal Dcvelopment Board (ICDB) and the Illinois Clean Coal Institute (ICCI) and managed by F.I. Honea and H. Feldmann. Neither Carl Kruse and the ISGS nor any of its subcontractors nor thc US. DOE, Illinois Department of Energy & Natural Resources, ICDB, ICCI, nor any person acting on behalf of either assumes any Liabilities with respect to the use of, or for damages resulting from the use of, any information, apparatus, method or process disclosed in this report. The views and opinions of authors expressed herein do not necessarily state or reflect those of the US. Department of Energy. Information provided by Ray Kucik and John Lees of Allis Mineral Systems and Dick Eggers and Dennis Dare of Illinois Power Company were used in the preliminary cost estimate. Laboratory assistance by Gwen Donnals, and conversations with Massoud Rostam-Abadi of the ISGS were most helpful. REFERENCES 1. Brueggendick, H. 1993 Operating Erperience Wth STEAG’J Activated Carbon Processes - lalc/tm - in European Waste Incineration Plants Proceedings of the Tenth Annual International Pittsburgh Coal Conference, Pittsburgh, PA, September 20-24, 1993 pp 787-794. Kruse, C.W. and N.F. Shimp 1981 Removal of manic Sulfur by Low-temperature Carbonization of Illinois Conk Coal Processing Tech. 7 pp 124-134. Alvin, M.A, D.H. Archer and M.M. Ahmed 1987 Pyr0tysi.s of Coal for Production of Low-mlfur Fuel EPRI Final Report AD-5W5, Project No. 2051-2. Hackley, KC, R.R. Frost, C-L. Liu, S.J. Hawk and D.D. Coleman 1990 Srudy of Sulfur Behavior and Removal During Thermal Demlfirization of Illinois Coals Illinois State Geol. Survey Circ. #545. DeBarr, J.A, M. Rostam-Abadi, R.D. Harvey, C. Feizoulof, S. A Benson, and D.L. Toman 1991 Reactivity and Combuslion Propenies of Coal-char Blend Fuels Final Technical Report to the Center for Research on Sulfur in Coal. September, 1991. DeBarr, J.A 1993 Integrated Methods for Production of Clean Char and Its Combustion Pmpmies Final Technical Report to the Illinois Clean Coal Institute for September 1, 1992, through December 31,19m. Lizzio, AA, J.A DeBarr and C.W. Kruse 1995 Development of Low Surface Area Char for Cleanup of Incinerator Flue Gas Abstract of papers at 22nd Biennial Conference on Carbon, The University of California at San Diego, July 16-21. Lizzio, AA, J.A DeBarr, C.W. Kruse, M. Rostam-Abadi, G.L. Donnak, and M.J. Rood 1994 production and Use of Activated Char for Combined S02/NOz Removal Final Technical Report to the IUinois Clean Coal Institute. 9. Lizzio, AA 1990 The Concept of Reactive Surface Area Applied to Uncatatyed and Catalyzed Carbon (Char) Gasification in Carbon Dwxide and Oxygen Ph.D. Thesis, The Pennsylvania State University. 10. Kruse, C.W., AA Lizzio, M. Rostam-Abadi, J.A DeBarr, J.M. Lytle, S.B Bhagwat 1995 Producing Activated Char for Cleaning Flue Gasfmm Incinerarors Final Technical Report to the Illinois Clean Coal Institute. 11. DeBarr, J.A 1995 The Role of Free Sites in the Removul ofS02from Simulated Flue Gasses by Activated Char M.S. Thesis, The University of Illinois, Urbana. 2. 3. 4. 5. 6. 7. 8. 840 Table 1. Emission Limits for European Waste Incinerators [l] Germany Netherlands Austria One day One hour mean values mean values mean values Half hour ~ ~~ 10 5 15 10 10 10 50 40 so 2001 70 100 1 1 0.7 0.05 0.05 0.05 0.1 0.1 0.1 ~ Federal standards, local standards generally at 100 mg/m' Table 2. Experience with STEAGs /a/c/tP-Process [l] Medical Medical Hazardous Waste Waste Waste Incinerator Incinerator Incinerator Germany Netherlands Netherlands Total Dust (mg/m') < 2 1 < 0.5 HCI (mg/m') < 1 < 2.2 < 0.19 HF (mg/m3) < 0.05 < 0.05 < 0.05 SO, (mdm') < 2 < 0.6 < 6 NO. (mg/m') 65 3431 177l Hg (mg/m') < 0.01 < 0.00031 < 0.002 Dioxins & Furans (ng TE/m') 0.003 0.00031 0.002 Not equipped with SCR Table 3. Lab Scale Adsorbent Preparation at the ISGS / Run Preoxidation # in the CFRK 11 Air,3WaC,2h 12 Air, 280°C, 0.5 h 13 Air, 330"C, 0.75 h 14 Air, 22O"C, 0.75 h 15 Air, 220'C, 1.5 h 16 Air, 22O"C, 0.75 h PvTOlvSis Activation Surface Area in the CFRK Nz BET (m*/g) CFRK, C02, 450"C, 1 h CFRK, Nz, 410"C, 0.5 h CFRK, COz, 475"C, 1 h CFRK, COz, 475"C, 1 h CFCO, N2, 475'C, 0.75 h CFCO, N2, 475'C. 0.75 h coz, 850"C, 1 h 235 coz, 850"C, 1 h 151 cox, 850°C 1 h 236 coz, 850'C, 1 h 179 CO,, 850"C, 0.75 h 230 Coz, 850°C. 0.75 h 180 J I Table 4. Typical Analysis Colchester Coal Moisture 14.4% Vol. Matter 39.9% Fixed Carbon 53.3% H-T Ash 6.8% Carbon 74.3% Nitrogen 1.4% oxygen 8.9% Total Sulfur 3.3% Sulfatic 0.1% Pyritic 2.2% Organic 1.1% Btdb 13,645 Hydrogen 5.3% FSI 3.8% Table 5. Comparison of Properties (Yield: 48% [277 kg from 581 kg coal]) German ISGSIAMS Herdofenkoks Colchester Property sorbent sorbent Bulk density lbslft (kg/m3) 29.8 (413) 23.8 (378) PROXIMATE (moisture free) Hi-Temp. Ash, wt% 8.68 8.27 Fixed Carbon, wt% 83.61 86.98 Volatile Matter, wt% 7.71 4.74 N1 BET Surface Area, m'/g 277 110 4 841 Table 6. Results of Testing ISGS Colchester Adsorbent in STEAG's Test Module Before After Adsorption Reactor Reactor Efficiency Total Dioxins & Furans Test 1 (ng/m3) Test 2 (ng/m3) Test 3 (ng/m3) Cd+Ti Test 1 (mg/m3) Test 2 (mg/m3) Test 3 (.me_/m') . kt2 Test 1 (mg/m3) Test 2 (melm3) Test 3 (mg/m3j Sb. As. Pb. CI. Co. Cu. Mn. N. V. Sn Test 1 (me/m3) Test 2 (mg/m3) Test 3 (mg/m3) 333.3 337.9 282.3 0.0140 0.0062 0.0052 0.0177 0.0384 0.0223 0.2698 0.0805 0.0634 0.062 0.052 0.789 0.0012 0.m12 0.ooo4 below det. limit below del. limit below det. limit 0.0744 0.0347 0.0185 Table 7. Estimated Break Even Cost in an 80,000 TonNear Plant' 99.98 % 99.98 % 99.72 % 91 % 81 % 92% 72 % 57 % 71 % Land purchase price (S) Building cost ($)* Equipment (S)* Installation (S) Carbon production (S) Percent yield Coal input (Uyr) Coal price f.0.b. mine, Coal transportation cost (Ut) sized 1 mm by 6.4 mm (Ut) 1m,000 5,000,000 16,000,000 8,500,wO 80,ooo 45 177,778 34 8 Cast of coal (S/y) Cost of Natural Gas ($&) Cast of Electricity ($/y) Cost of lime (to neutralize 502) Water (2.45MM gal @ SI/Mgal) Labor cost Maintenance cost (5% of $2lMM)3 Real estate taxes ($) BREAK EVEN COST(S/ton)' 7,466,667 2,626,000 526,000 300,000 2,450 1,540,000 1,050,flW 153,000 326 ' Assume it takes one year to construct the plant, and the plant produces a1 design capacity thereafter. Depreciation (7 Yr st. line) on S21MM. b 0) Figure 1. SO2 Adsorption ' Operaling Msts are raised 3% per year. ' Inlerest an undepreciatcd value at 20%&r, net present value at 24l% discount rale, 20 years of operation. OXIDATION 2500 C max 2 hr reoldonce Umo I I I STACK AIR + 3 10 3.0 wt X oxygen on coal (molsmre h e ) Bulk denolty: 9298 Ib/cu R PRODUCT 1 mm by 0.4 mm PYROLVS S h ACTIVATION 2 hi nsldonce l l m(~to tal) Zono 1: 376 - 46dC Zone 2 826-878C ACTIVATED CARBON 46% Yleld 6816ulated COOLER (mm molelun h e coal Bulk denslty: 26-30 lblw it mlnus 1 mm mmerlcd Figure 3. Flow Diagram for a Plant Producing Adsorbent from Colchcster (Illinois No. 2) Coal 842 PILOT PLANT STUDY OF MERCURY CONTROL IN FLUE 6AS FROH COAL-FIRED BOILERS Joseph T. Maskew, William A. Rosenhoover, Mark R. Stouffer, Francis R. Vargo and Jeffrey A. Withum CONSOL Inc. Research a Development 4000 Brownsville Road library, PA 15129 Keywords: BACKGROUND Mercury control technology options for coal-fired boilers are ill-defined. Commercial development of mercury emissions control technologies has centered on high concentrations of mercury compared to the levels present in the flue gas from coal combustion, typically 5 to 10 pg/m3. In addition, most mercury in these commercial applications (medical waste and municipal sol id waste incinerators) 1*2,’0 is in the form of HgC1,; flue gas from coal-fired units contains both ionic and elemental mercury. Reaction mechanisms may be different for these two species. Development work at the lower concentrations has centered on small scale, fixed-bed, laboratory s t u d i e ~ . ~ * ~ *Re’c~e nt tests at coal combustion sources with sorbents such as and activated carbon4 have shown some mercury removal. However, neither the laboratory nor combustion tests completely address process design issues. In the laboratory studies, the actual process conditions are very different from those with coal; while in the combustion tests, it is difficult to vary the conditions. In addition, data reliability is poor because of the difficulty of mercury sampling and analysis. Mercury control, coal-fired boilers, flue gas analysis Development of mercury control technology for coal-fired flue gas requires: 1. Accurate and reliable sampling and analytical techniques, including speciation o f mercury, 2. A thorough understanding of the effects of the combustion conditions and of the speciation of mercury on mercury removal, 3. Identification of sorbents and process configurations for removal of mercury at the low levels present in coal-fired flue gas, and 4. Waste management studies and economic evaluation of control technologies. Each of these factors is important in developing a process to control mercury emissions. To this end, a 0.2 MWe equivalent, continuous flow pilot plant was constructed at CONSOL R&D to evaluate the efficiency and cost of sorbent injection technology for mercury control, and to verify mercury sampling and analysis techniques. DESCRIPTION OF THE FACILITY The 0.22 Nm3/s (500 scfm) pilot plant i s of sufficient size to provide a realistic process simulation while maintaining the capability to study the effect of potentially important variables such as sorbent/flue gas residence time, fly ash loading, and mass transport phenomena. It provides accurate and independent control of key process variables, including mercury concentration and speciation. The flue gas mercury concentration can be varied between 2 and 20 pg/m’, a range typical of coal combustion. By adding actual coal fly ash, the physical and chemical fly ash/ sorbent interactions are realistically simulated. Because the pilot plant is a flow system, the mass transfer conditions, temperature/time history, and gas/solid interactions can be varied to simulate conditions in a coal-fired power plant. The sorbent injection pilot plant accurately simulates flue gas downstream of the air preheater in a coal-fired boiler. The plant was designed to simulate a wide range of site-specific conditions by burning natural gas and by injecting the deficient components such as fly ash, CO , SO, and mercury compounds. Independent control of the temperature (38-265 ‘C, 100-400 O F ) , humidity, sorbent injection and sorbent recycle rate is maintained. The pilot plant was proven to be a reliable, accurate tool for desulfurization studies when its results for the Coolside processlwre scaled up to a 105 MWe demonstration at the Ohio Edison Edgewater plant. Figure 1 is a schematic of the 0.22 Nd/s (500) scfm sorbent injection pilot plant. Originally used in the development of the Coolside and Advanced Coolside desulfurization p r o ~ e s s e s , ’i~t was modified for mercury control studies. The plant consists Of a flue gas generation system, a flue gas conditioner for temperature and humidity control, a mercury spiking system, fly ash and sorbent injection systems, a sorbent recycle system, flue gas duct work, particulate removal systems (cyclones and a baghouse), a waste handling system, and flue gas 843 analysis systems. The p i l o t plant provides accurate and independent control of flue gas temperature and composition. Accurate control o f mercury concentration and speciation i n the simulated f l u e gas i s maintained independently of the bulk f l u e gas composition. The feed and effluent sorbent streams and f l u e gas stream can be sampled. The p i l o t plant i s instrumented and automated for. process control and data c o l l e c t i o n . A natural gas combustor, a steam i n j e c t i o n system and the f l u e gas conditioner are used t o control f l u e gas humidity and temperature independently. Control loops on these systems allow f l u e gas temperature t o be maintained automatically between 38 and 205 'C w i t h i n i0.5 'C (100 and 400 'F w i t h i n i1 'F) and the approach to adiabatic saturation t o be controlled w i t h i n i0.5 'C (tl O F ) . The feed system for elemental mercury consists o f mercury-containing permeation tubes, a constant temperature bath and an i n e r t c a r r i e r gas. The tubes are commercially available, and are an accurate, reproducible method f o r feeding mercury. The temperature of the tubes i s controlled to w i t h i n iO.01 'C (f0.02 'F) by a constant temperature bath. The permeation rate f o r the tubes i s calibrated by weighing the tubes over a known period. I n long-term tests, weight loss o f the tubes i s used t o v e r i f y the mercury material balance. Mercuric chloride i s fed by a separate, similar subsystem. Similar calibrations were carried out on the HgC1, feed system. t o the flue gas independently, the amount and speciation of mercury are controfled to w i t h i n 5%. The solids are collected using a cyclone or a baghouse. The sorbent collected by the cyclone i s almost instantaneously removed from contact with the flue gas stream. This allows solids t o be collected a f t e r a short, well-controlled contact time with the f l u e gas (1-3 sec). With two p a r a l l e l p a r t i c u l a t e collecting devices, in-duct removal can be measured separately from baghouse removal. The in-duct mercury removal allows estimation of the Hg removal i n an ESP-equipped unit. Recycling the flue gas reduces reagent costs and assists i n maintaining a consistent f l u e gas composition. A large fixed-bed carbon f i l t e r prevents recycle o f Hgo or HgC1, not removed by the sorbent. For a l l the f l u e gas sampling t e s t s , the simulated flue gas contained 1000 ppmv SO, 10% 0, and 1% CO, and had a saturation temperature o f 52 'C (125 'FJ. The flue gas flow was accurately controlled and monitored with a thermal dispersion mass flowmeter, and checked by standard manual procedures ( p i t o t tube/differential pressure gauge). The gas sampling was conducted in a section o f the p i l o t plant duct located approximately 16.8 m (55 ft) downstream o f the mercury i n j e c t i o n point. There are a gas d i s t r i b u t i o n plate in the duct j u s t downstream o f the i n j e c t i o n point and several direction changes o f the f l u e gas (90' bends) p r i o r t o sampling t o d i s t r i b u t e mercury i n the flue gas. TEST PROGRAI I n i t i a l ODerations V e r i f i c a t i o n and, i f necessary, improvement of sampl ing/analytical techniques i s the f i r s t task i n the experimental program. Accuracy and r e l i a b i l i t y are c r i t i c a l f o r measuring flue gas mercury concentration, f o r determining speciation, t o provide r e l i a b l e data for process development, and f o r scale-up to commercial application. Because the mercury concentration and speciation are accurately controlled i n the p i l o t plant, any error i n the sampling/analytical methods can be determined. Sorbent Evaluation/OeveloDment I d e n t i f i c a t i o n of an inexpensive, effective sorbent i s a primary objective of this work. Understanding the effects of temperature, humidity and mercury speciation on sorbent performance i s c r i t i c a l f o r designing a viable process. To achieve this, s t a t i s t i c a l l y designed screening tests w i l l be performed on each candidate sorbent. For candidate sorbents, s i g n i f i c a n t process variables w i l l be explored i n more d e t a i l . Steady-state t e s t s , with sorbent recycle, w i l l be made with the most cost-effective sorbents. These runs w i l l l a s t two to three days, u n t i l steady-state conditions are demonstrated by sol i d analysis. By adding Hg' and HgCl Waste Wanaaement Studies Several important technical issues involve waste management. These include mercury leaching, r e v o l a t i l i z a t i o n and the impact o f mercurv on ash u t i l i z a t i o n . However, u t i l i z a t i o n o f s o l i d waste i s preferable t o dispoG1 and can accelerate commercialization. The program w i l l evaluate options f o r waste u t i l i z a t i o n , with emphasis on the high volume use o f the material i n construction. A successful approach t o eliminate or reduce the need f o r waste disposal represents a substantial improvement i n the state o f sorbent injection processes. Economics Engineering and economic studies w i l l be conducted to determine the f e a s i b i l i t y o f process operations. Sorbent i n j e c t i o n processes have inherently low capital costs; therefore, sorbent cost i s a key issue. Hydrated l i m e may be e f f e c t i v e a44 for removing Some ionic mercury and is low in cost; however it may not be effective in removing elemental Hg. High surface area activated carbons are expensive ($0.50-0.80/kg ($450-1000/ton)), and chemically impregnated sorbents are even more expensive by a factor of five. The minimum amount of sorbent required is not known and likely will vary among applications. The potential of recycle to increase sorbent utilization also will be addressed. lntegration of FercurY control with other flue gas treatment systems represents a significant improvement in the process economics. Process economic studies also will allow research to focus on areas of the most potential benefit to process economics. INITIAL RESULTS Mercury Feed s tern Calibration ofYshe elemental mercury (Hg') and mercuric chloride (HgC1,) feed System showed a high degree of accuracy and precision. In replicate tests of weight loss vs time, the variation from the amount of Hg' or HgC1, fed at a particular calibration condition was &4% or less for Hg and i6% for HgC1,. Figure 2 shows the Hg' calibration data. In these tests, the weight loss of several of the commercially available Hg' permeation tubes was measured as a function of temperature. In these calibration tests, emphasis was placed on 110 and 114 'C, the typical temperatures of the Hg' feed system pilot plant operations. Six calibration runs were made at each of these two temperatures. Similar precision was obtained in the calibration of the HgC1, feed subsystem. Figure 3 shows the amount of HgC1, evolved at three different calibration conditions. The data represent four to six replicate tests at each calibration condition. i- made in which Hg' and/or HgCl were added to the pilot plant flue gas, and the gas sampled using EPA Method 28, followed by cold vapor atomic absorption (CVAA) analysis of the impingers solutions.14 The fluegas mercury concentration in these tests was 4 to 24 pg/m3, typical of concentrations found downstream of a coal-fired boiler. Figure 4 shows that in tests with only Hg' addition, there was very good agreement between the Method 29 gas sampling/analysis results and the amount of Hgo fed to the flue gas via the feed system. The mercury concentration in the flue gas based on Method 29 results was 10 to 12.5 pg/n?, compared to 9 to 9.5 pg/d based on feed system calibration. In the HgC1, tests, the flue gas mercury concentration based on sampl ing/analysis was, on average, 30% lower than that based on the feed system calibration (Figure 4 and Table I). It appears that the ionic mercury present in the pilot plant flue gas was not entirely recovered and/or detected by the Method 29 sampling train and analytical procedures. The accuracy of the mercury feed rates were further confirmed by injecting a large excess of activated carbon at low temperature (c93 'C or 200 'F), and measuring the mercury captured by analysis of the sorbent recovered from the baghouse. Table 1 shows that the ionic mercury (HgC1 ) was in general evenly distributed between the front impingers containing nitr%c acid and peroxide and the back set of impingers containing permanganate and sulfuric acid. This was true in several tests in which the mercury concentration in the flue gas was varied. These results are contrary to reported assumptions that ionic mercury is primarily captured in the front i m p i n g e r ~ . ~ * ~ ,A"l l the elemental mercury was captured in the back set of impingers (permanganate), which agrees with reported assumptions. 4'6''1 Additional testing will be done to further investigate mercury capture and speciation by Method 29. REFERENCES lue Gas Sam lin and A alvsis 1. 2. 3. 4. 5. Brna, T. G., Kilgroe, J. D., Miller, C. A. "Reducing Mercury Emission from Municipal Waste Combustion with Carbon Injection into Flue Gas", U.S. Environmental Protection Agency Report EPA/600/A-92/134, 1992. Brown, B., Felsvang, K. "Control of Mercury and Dioxin Emissions from United States and European Municipal Solid Waste Incinerators by Spray Dryer Absorption Systems", Proceedings, Second A&WMA International Conference on Municipal Waste Combustion, Tampa, Florida, April 1991, Brown, B., Felsvang, K. "High SO Removal Dry FGD Systems", Presented at the 1991 SO, Control Symposium, Wasiington, D.C., December 1991. !hang, R., Bustard, C. D., Schott, G., Hunt, T., Noble, H., Cooper, J. Pilot Plant Evaluation of Activated Carbon for the Removal of Mercury at Coal-Fired Utility Power Plants", Presented at the Second International Conference on Managing Hazardous Air Pollutants, Washington, DC, July 1993. Felsvang, K., Gleiser, R., Juip,,,G., Nielsen, K. K. "Air Toxics Control by Spray Dryer Absorption Systems , Presented at the Second International Conference on Managing Hazardous Air Pollutants, Washington, DC, July 1993. p. 675-705. 845 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. Felsvang, K.; Gleis:r, R.; Juip, G.; Nielsen, K. K. "Control o f A i r Toxics by Dry FGD Systems , Proceedings, Power-Gen '92 Conference, Orlando, FL, November 1992. Gleiser, R.; Nielsen, K. K.; Felsvang, K. "Control o f Mercury from MSW Combustor: by Spray Dryer Absorption Systems and Activated Carbon Injection , Third International Conference On Municipal Waste Combustion, Williamsburg, VA, March 30-April 1, 1993. Gullett, B. K. and Jozewicz, W. "Bench-Scale Sorption and Desorption O f Mercury With Activated Carbon", Third International Conference On Municipal Waste Combustion, Williamsburg, VA, March 30- A p r i l 1, 1993. Gullett, E. K. and Krishnan, S. !. "Sorbent I n j e c t i o n For Dioxin/Furan Prevention and Mercury Control , Multipollutant Sorbent Reactivity Workshop, Research Triangle Park, NC, July 1994. kilgroe, J. 0.. Brna, T. G., White, D. M., Kelly, r. E., Stucky, M. J. Camden County MWC Carbon Injection Test Results , Proceedings, 1993 International Conference on Municipal Waste Combustion, Williamsburg, VA, March 1993. Laudal,,,D. L.; M i l l e r , S. J. "Evaluation o f Sorbents f o r Enhanced Mercury Control , Proceedings, Tenth Annual Coal Preparation, U t i l i z a t i o n , and Environmental Control Contractors Conference, Pittsburgh, PA, July 1994. Livengood, C. D.; Huang, H. S.; Wu, 3. M. "Experimental Evaluation of Sorbents f o r the Capture of Mercury i n Flue Gases", Proceedings, 87th Annual Meeting o f the A i r & Waste Management Association, Cincinnati, OH, June 1994. McCoy, D. C.; Scandrol, R. 0.; Statnick, R;, M.; Stouffer, M. R.; Winschel R. A.; Withum, J. A.; Wu, M. M.; yoon, H. The Edgewater Coolside Process Demonstration: A Topical Report , DOE Cooperative Agreement DE-FCZZ- 87PC79798, February 1992. Nott,"E. R.; Huyck, k. A.; DeWees, W.; Prestbo, E.; Olmez, I ; Tawney, C. W.; Evaluation and Comparison o f Methods f o r Mercury Measurement in U t i l i t y Stack Gas", A i r & Waste Management Association, 87th Annual Meeting and Exhibition, June 19-24, 1994, Cincinnati, OH. Withum, J. A.; Maskew, J. T.; Rosenhoover, W. A.; Stouffer, M. R.; Wu, M. M. "Deve1,opment of the Advanced Coolside Sorbent I n j e c t i o n Process f o r SO Control , Proceedings, 1995 SO, Control Symposium, Miami, FL, March 1955. TABLE 1. CAPTURE OF ELEMENTAL AND IONIC MERCURY IN METHOD 29 SAMPLING TRAIN Test A B Mercury Recovered i n Impingers, % o f Total Mercury Fed Species Fed KMnOJH$O, 48 34 38 10 38 <4 <4 39 42 24 31 29 134 106 846 / I I I Figure 1. Schematic of CONSOL Sorbent Injection Pilot Plant Figure 2. Calibration o f Elemental Figure 3. Calibration of Mercuric Mercury Feed System. Chloride Feed System. 0 2 4 6 8 10 12 14 16 18 20 22 24 Flue Gas Mercury by Feed Callbralion pg/m' Figure 4. Comparison of Flue Gas Mercury Concentration Based on Method 29, with Concentration Based on Mercury Feed System Calibration. 841 DEVELOPMENTOFMERCURYCONTROL TECHNOLOGY FOR COAL-FIRED SYSTEMS C. David Livengood Hann S. Huang Marshall H. Mendelsohn liann M. Wu ArgOMe National Laboratory 9700 South Cass Avenue Argonne, Illinois 60439 Keywords: Toxics, Mercury Control, Coal INTRODUCTION The emission of hazardous air pollutants (air toxics) from various indusvial processes has emerged as a major environmental issue that was singled out for particular attention in the Clean Air Act Amendments of 1990. In particular, mercury emissions are the subject of several cumnt EPA studies because of concerns over possible serious effects on human health. Some of those emissions originate in the combustion of coal, which contains lrace amounts of mercury, and are likely to be the subject of control requirements in the relatively near future. Data collected by the Department of Energy (DOE) and the Elecuic Power Research Institute (EPRI) at operating electric-power plants have shown that conventional flue-gas cleanup (FGC) technologies are not very effective in controlling emissions of mercury in general. and are particularly poor at controlling emissions of elemental mercury. 7his paper gives an overview of research being conducted at Argonne National Laboratory on improving the capture of mercury in flue gas through the use of dry sorbents and/or wet scmbbers. BACKGROUND Mercury emissions from coal combustion have been shown to vary considerably from site to site. Those emissions depend not only on the composition of the coal, but also upon the type of boiler, the operating conditions, and the FGC system. Mercury belongs to a group of elements/compounds denoted as Class 111. which remain primarily in the vapor phase within the boiler and subsequent FGC system. However, that state can be influenced by reactions with other elements. such as chlorine, and by fly-ash characteristics that affect adsorption processes. The concentration of mercury in the flue gas from typical coal combustors ranges from less than 10 to more than 50 pg/Nm3. Few reliable data on mercury control have been available for FGC technologies used on coal-fired systems. Large variations in reported removals have been typical. duc both to differences in coal and operating characteristics and to inaccuracies in sampling/analytical procedures.' Paniculate-matter collectors. such as electrostatic precipitators (ESPs) and baghouses. can be effective for mercury control to !he extent that mercury is adsorbed on the fine particulate matter (fly ash) in the gas stream or is converted to another chemical form that can be collected as particulate matter. Recent data on mercury removals for ESPs range from about 15 to 75%. while very limited removal data for baghouses range from IO up to 70%. Mercury removal in wet flue-gas desulfurization (FGD) systems is also quite variable, with values ranging from near zero to about 50%? Much of that variation may be caused by differences in the chemical form of the mercury, inasmuch as the chloride is much more easily captured than the elemental form. Most available information on mercury conmol technologies for combustion sources has originated in work with waste incinerators. In such cases, activated carbon has been shown to be an effective sorbent for mercury. However, flue-gas conditions at incinerators are much different in temperature and composition than those found at coal-fired utility boilers, and the performance/economics of activated carbon can be expected to vary as well. In addition, the presence of wet FGD systems at many utility boilers presents a considerably different set of conditions and problems/opportunities that need to be evaluated. RESEARCH PROGRAM Based on an initial survey of published information, a number of chemical additives and sorbents with the potential for enhancing the capture of elcmental mercury in dry or wet/dry FGC systems were selected for laboratory investigation. The study of dry sorbents was chosen for several reasons. Many existing coal-fired plants have only particulate-matter control, usually in the form of ESPs, and these could be well suited to duct- or fumace-injeclion of mercury sorbents. Also, European experience with the addition of sorbentskhemicals to spray-drycr systems on municipal waste incinerators has indicated that greatly enhand mercury removals arc possible. A more extensive discussion of this research can be found in Reference 3. The research program also includes investigation of mercury removal in wet scmbbing. The iniual study found very little information regarding potential performance enhancements for scrubbers operating on coal-fired systems, although some work has been done for applications in other industries.' To date. the research has focused on physical modifications designed to improve the absorption of mercury I Work suppomd by the US. Depanment of Energy, Assistant Secretary for Fossil Energy, under contract W-31-1 09-ENG-38. 840 by the Scrubber liquid, on the testing of chemical agents selected for their potential to react with mercury. and On Process modifications designed to combine gas-phase and liquid-phase reactions. EXPERWENTAL FACILITIES Argome’s FGC-laboratory facilities include a fixed-bed reactor system for studying dry sorbents. a COmpkte wet scrubber system. and a spray-dryedfabric-filter system. Supporting facilities include a system that can provide known concentrations of elemental mercury in a gas stream, a gas-supply system capable of blending synthetic flue gas from bottled gases, on-line gas-analysis packages. and data loggers. The following sections briefly describe the key systems. More detailed descriptions of all of the SyStemS Can be found in References 5 and 6. Mercury Supply and Analysis The feed-gas preparation system consists of a mercury-containing permeation tube, a constant-temperature water bath, and a carrier-gas supply. The design capacity of the system is 20 L/min of gas with mercury concentrations of up to 100 pg/m’. Mercury measurements are made using a gold-film mercury-vapor analyzer. The range of the instrument is 0 to 999 pg/m’ with a sensitivity of 3 pg/m’ and an accuracy of -+ 5% at 100 pg/m’. Fixed-Bed Reactor The fixed-bed reactor vessel, which is constructed of glass, is 4 cm in diameter and 14 cm in height. A glass frit is positioned in the lower section to support materials placed inside the reactor. To avoid fluidization of the bed materials, the feed gas enters the reactor from the top and exits at the bonom. During shakedown and baseline tests, the reactor was packed with either silica sand (120 g) or a mixture of silica sand and hydrated lime (Ca(0H)d in a weight ratio of 40:l. The Ca(0Hk has been employed because it is a common sorbent for SO2 in FGC systems. The large amount of sand is used to avoid channeling caused by lime agglomeration. For additive/soltKnt testing. small amounts of material being studied are added to the sand/Ca(OHk bed material. To maintain a uniform temperature during experiments. the reactor is immersed in a fluidized-bed, constant-temperature sand bath. To preheat the incoming feed gas to a temperature equal to that maintained in the fixed-bed reactor, the gas-transfer line is wrapped with heating tapes. Wet Scrubber AU of the principal vessels in the wet-scrubber system are constructed of glass. The scrubber column has an inside diameter of 7.6 cm and an active height of nearly 53 cm. It is normally operated in a countercurrent mode with the flue gas entering at the bottom. The scrubber is constructed of several interchangeable sections so that it can be configured as a flooded column (no intemals). a four-stage disc and donut column, or an intermediate combination. For most of the experiments described here. the combination mode was used with the lower part of the column left open to accommodate packing. The scrubber liquor drains into a holding tank from which it is recirculated to the top of the scrubber. The temperature of the liquor can be adjusted by heating the holding tank with heat tapes. The pH of the liquid in the tank is sampled continuously and can be adjusted either manually or automatically by adding reagent from a chemical feed tank. EXPERIMENTAL RESULTS Experiments with Dry SorbentdAdditives Following initial shakedown tests that verified that neither the sand nor the lime in the fixed bed gave any measurable mercury removal. a variety of dry sorbents were studied. Various sorbents and chemical additives for mercury removal have been reponed in the literature. These include activated carbon, activated carbon impregnated with various chemicals (notably sulfur and iodine), modified zeolites, glass fibers coated with special chemicals, and pure chemicals (such as sulfur, selenium, and ferrous sulfide and sulfate). In addition to comparing the performance of different types of sorbents/additives, the research program has included investigation of the effects of varying process parameters, such as sorbent particle size, sorbent loading in the reactor, reactor/gas temperature. and mercury concentration.’ For most of the tests, the amounts of sorbent added ranged from 1 to IO wt% (relative to the lime). Three fixed-bed reactor temperatures were evaluated: 55.70, and 90°C. Target mercury concentrations in the nitrogen feed gas of either 44 or 96 pp/m’ were used, and the feed-gas flow rate was fixed at IO L/min. By far the best removal results in the initial tests were obtained with an activated carbon that was commercially treated with about 15 wt% sulfur. The success of the sulfur-treated carbon is thought to be based on a combination of physical adsorption and chemical reactions that produce mercury sulfide. This suggests that chemical additives producing other compounds. such as mercury chloride. might also be beneficial for removals. To explore this possibility, another carbon sample that previously gave essentially no removal was treated with calcium chloride (CaCIJ in the ratio of about 6:l by weight. The treated carbon gave excellent removals and actually performed better than the sulfur-treated carbon. Recently. the research has been focused on the development and testing of lower-cost alternatives to activated carbon. Several high-surface-area or low-cost mineral substrates have been identified and samples have been obtained. Tests of the materials in the as-received condition gave moderate mercury removals for a molecular sieve sorbent and essentially no removals for pumice and vermiculite samples. In current research, the samples are being treated with chemical additives shown to be effective with activated carbon and tests are king run at various additive concentrations. mercury concentrations, and flue-gas temperatures. Figure 1 gives the results of experiments with volcanic pumice treated with 849 potassium iodide, CaQ. or sulfur. The untreated pumice was ineffective for mercury removal, but the sulfur-treated sorbent gave 100% removal for over an hour, while the iodide-impregnated sorbent gave 100% removal for a few minutes followed by a decrease in removal that appeared to level out at about 30%. In order to explore the effects of temperature on the treated sorbents. additional tests were run at a temperature of 150°C. As shown in Figure 2, the iodide-impregnated sample behaved very similarly at the two temperatures. However, the sulfur treatment that was so effective at the lower temperature was found to be totally ineffective at the higher temperature. This may be due to a change in the form of the sulfur, but this issue is still under study and has not yet been resolved. Experiments with Wet Scrubbing Preliminary data from field-sampling campaigns have indicated that elemental mercury is not appreciably removed in typical wet-scrubber systems. This is not surprising given the very low solubility of mercury in the elemental form. Initial experiments were conducted using the scrubber as described above, no packing, and various degrees of "flooding" in the lower part of the column to promote gas-liquid contact. The scrubbing liquors tested were distilled water, a Saturated Ca(OH), solution, and a Ca(0Hh solution with loo0 ppm of potassium polysulfide. The polysulfide has been claimed to promote mercury removal in other research.' The mercury inlet concentration was about 40 pg/m', the liquid height in the column was vaned up to 43 cm. and the temperature was varied between 22 and 5 0 T No mercury removal was detected under any of these conditions. The addition of ceramic-saddle packing to the column produced removals of 3 to 5% with distilled water at 22OC. and removals of 6 to 7% were obtained when the temperature was raised to 55°C. However, tests involving polysulfide addition had to be terminated when reactions with the ceramic saddles produced hydrogen sulfide (H,S) that interfered with the operation of the mercury analyzer. In earlier research on mercury capture, stainless steel packing was found to promote mercury ~apture.~ Therefore, the ceramic saddles were replaced by 0.61-cm stainless-steel packing, which gave the rather unexpected result of 11% removal with no liquid in the column. Removals with water in the column ranged from 15 to 20%. Addition of polysulfide to the scrubber produced a noticeable increase in removal up to about 40%. It appears that there is a positive synergistic effect on removal involving the combination of polysulfide and stainless steel. It should be noted that this additive requires a very high pH to maintain its stability and this may preclude its use in most FGD systems. In an effort to promote greater mercury capture through changing its chemical form. tests were conducted with several additives that combine strong oxidizing properties with relatively high vapor pressurcs. Tests with minimal gas-liquid contacting yielded mercury removals as high as 100%. and indicated that the removal reactions were. occumng in the gas phase above the scrubber liquor. However, tests with the addition of SO2 to the gas stream showed the additives to be very reactive with that species as well, which could result in excessively high additive consumption in order to realize effective mercury control. Recently, tests with a new combination of oxidizing chemicals, NOXSORBTM. which is a product of the Olin Copration, have indicated promise for integrated removal of several flue-gas species including mercury. Preliminary data from those tests are shown in Figure 3. Funher tests are exploring the effects of different additive concentrations, the relationship between NO/SO, removal and mercury removal. and possible process configurations and economics. CONCLUSIONS The results and conclusions to date from the Argonne research on dry sorbents can be summarized as follows: - Lime hydrates, either regular or high-surface-area, are not effective in removing elemental mercury. Mercury removals are. enhanced by the addition of activated carbon. Mercury removals with activated carbon decrease with increasing temperature, larger particle size. and decreasing mercury concentration in the gas. Chemical pretreament (e.g.. with sulfur or CaCIJ can greatly increase the removal capacity of activated carbon. Chemically treated mimnl substrates have the potential to be developed into effective and economical mercury sorbents. Sorbents treated with different chemicals respond in significantly different ways to changes in flue-gas temperature. - - * * Preliminary results from the wet scrubbing research include: - No removal of elemental mercury is obtained under normal scrubber operating conditions * Mercury removal is improved by the addition of packing or other techniques to increase the gas-liquid contact area. 850 * Stainless steel packing appears to have beneficial properties for mercury removal and should be investigated funher. Beneficial synergisms with polysulfide solutions have been observed. Oxidizing additives may be used in conjunction with wet scrubbing to greatly enhance removals. Selectivity is required to avoid excessive additive consumption from competing reactions. * ACKNOWLEDGMENTS The authors gratefully acknowledge the guidance and support for this research provided by Perry Bergman. Charles Schmidt. and Charles Drummond of the Pittsburgh Energy Technology Center. Appreciation is also exended to Sherman Smith for his many contributions to the laboratory operations. REFERENCES 1. 2. 3. 4. 5. 6. Huang, H.S., C.D. Livengood, and S. bomb. 1991, Emissions of Airborne Toxics from Coal-Fired Boilers: Mercury. ArgoMe National Laboratory report ANUESDITM-35. Schmidt, C.E.. and T.D. Brown, 1994, Resulrs from the Department of Energy’s Assessment of Air Toxics Emissions from Coal-Fired Power Plants, presentation at Illinois Coal Development Board Program Committee Meeting. Nov. 15. Livengood, C.D., H.S. Huang, and 1. M. Wu, 1994. Experimental Evaluation of Sorbents for the Caphrre of Mercury in Flue Gases, Proc. 87th Annual Meeting & Exhibition of the Air & Waste Management Association, Cincinnati, Ohio, June 19-24. Yan. T.Y.. 1991. Reaction of Trace Mercury in Natural Gas with Dilute Polysurfide Solutionr in a Packed Column, Industrial & Engineering ChemisVy Research, 30(12):2592-2595. Livengood, C.D., M.H. Mendelsohn, H.S. Huang. and J.M. Wu. 1995. Development of Mercury Control Techniques for Ufilify Boilers, h c . 88th Annual Meeting & Exhibition of the Air & Waste Management Association, San Antonio, Texas, June 18-23. Mendelsohn, M.H., and J.B.L. Harkness. 1991. Enhanced Flue-Gas Denitrification Using FerrowEDTA and a Polyphenolic Compound in an Aqueous Scrubber System, Energy & Fuels. 5(2):244-247. 1208 sand + 3g &(OH)* + l20g sand + 2g Ca(OH), + 0.3g sorbenr-IB KI -10 0 10 20 30 40 50 60 Time (min) Figure 1. Effects of chemical pretreatment on an inen substrate at 7WC. 851 f c0.3g sorbent-i% suirur. temperature= 150°C 12Og sand + 2g Co(OH), +0.5g sorbenr-l% KI: tempemturc=7O0C 12Og sand + 2g Ca(OH), temperature= 150°C lZOg sand + 2g Ca(OH)2 + +0.5g sorbenl-I% KI; 40 h 35 2E 30 .5- 25 c! g 20 8B 15 II v c c 0 -10 -5 0 5 10 15 20 25 30 Time (min) Figure 2. Effects of different temperatures on a chemically pretreated inert substrate 100- 80 - -20 -10 0 10 20 30 40 Time (rnin) Figure 3. Removals of Hg. NO, and SO2 in h e wet scrubber with a 4% NOXSORBTM solution. 852 I STRENGTH ENHANCEMENT OF CONCRETE CONTAINING MSW INCINERATOR ASH James T. Cobb, Jr., Daniel J. Reed and James T. Lewis III Department of Chemical & Petroleum Engineering University of Pittsburgh Pittsburgh, Pennsylvania 15261 Keywords: municipal solid waste, incinerator ash, concrete ABSTRACT In previous work [I] pretreatment of fresh municipal solid waste (MSW) incinerator ash With a novel type of additive, which was not identified chemically in that paper, was shown to markedly increase the compressive strength of portland cement concrete using the MSW ash as fine aggregate. A rmnt study has shown that, at lower levels of additive, aged MSW ash does not demonstrate the Same enhancement. This presentation will provide additional information concerning the previous study, give the results of the current one and discuss the implications of both. INTRODUCTION Much work is being conducted to find beneficial uses for the solid residues from energyconversion process, such as coal-fired electric power plants and combustors of municipal solid waste (MSW). Landfill caps and liners, grouts, structural fills, artificial aggregate for road bases, concretes of various types, and additives for cement have all been examined as outlets for these energy-related wastes. For all of the uses just listed, solidification is the principal goal, while stabilization of the eight RCRA metals (arsenic, barium, cadmium, chromium, lead, mercury, selenium and silver) is an important secondary consideration. Two residues obtained from the most thorough of the MSW combustors - the O’Conner rotary burner - fail to meet the following specifications for Class C Fly Ash - moisture content, ignition loss, pozzolanic activity index and fineness. [l] In addition they fail the EP toxicity test for cadmium and lead [I], as shown in Table 1. Thus, they cannot be used as cement additives and they must be stabilized when included in any of the other beneficial uses listed in the last Paragraph. An earlier paper [l] described a study in which a combined MSW ash from the O’Conner combustor was used as fine aggregate in portland cement concrete. In that paper it was pointed out that using MSW ash for this purpose substantially degrades the strength of the concrete so produced, but it reported that a novel additive had been discovered which gives early indications of economically restoring concretes containing MSW ash to their normal strengths. However, the authors of that earlier paper did not reveal the chemical composition of the novel additive. At that time they were exploring the possibility of obtaining a patent on the use of the additive. Since then, they have decided that, such a patent being essentially be unenforceable if awarded, the nature of the additive should be disclosed. One purpose of this paper, then, is that disclosure, along with some additional information obtained during the last few months of the Westinghouse-sponsored project, which came to its conclusion shortly after that paper was written. [2] Subsequently, another graduate student conducted a brief examination of this topic and found some interesting differences between the behavior of fresh and aged MSW ash. [3] The second purpose of this paper, then, is to report his findings. SOLIDIFICATION ENHANCEMENT USING NOVEL ADDITIVES The novel additive is a common acid. Two different acids have been tested - hydrochloric acid and acetic acid. The method of introduction of the acid may best be shown by giving the procedure (based upon ASTM C192-88) for mixing a batch of concrete in which it is included. The specific batch described is Batch 32: 0 Add 17.0 pounds of coarse aggregate and 25.8 pounds of MSW ash to a small cement mixer and commence rotation. Add 500 ml of 12 normal hydrochloric acid and mix for several minutes. 853 0 Add 33.5 pounds of Cement and 12.0 pounds of water (enough to provide a slump of 1.5 to 2.5 inches) to the mixer in equal proportions, one after the other, in three or four different intervals. when the mix is ready for molding, fill twenty three inch by six inch cylindrical cardboard molds and place them in a curing room. After each period of three, seven, fourteen, twenty-eight and ninety days, test four cylinders for compressive strength, reporting the average strength of the strongest three cylinders. 0 0 Figure 1 provides a record of the 28-day average compressive strengths for twelve concretes prepared with MSW ash as fine aggregate and with varying amounts of either hydrochloric acid or acetic acid. The abscissa is structured in units of gram moleslpounds MSW ash. For comparison, the 28-day strength of a concrete made with no additive (Batch 5) is shown. Nearly a 300% increase in compressive strength (1400 psi to 5500 psi) is achieved by Batch 30, made with 0.25 gram moles of hydrochloric acid per pound of MSW ash. The results of the 90-day compressive strengths are confused. These batches were made near the end of the project and 90-day strengths were not obtained for Batches 44 through 49. In addition, the cylinders for Batches 27, 35, 36 and 37 deteriorated such that no strength could be measured. The 90-day compressive strengths for Batches 30, 31,32 and 42 are 4940, 4380, 6200 and 1618 respectively. [It should be noted that no 28day strength for batch 42 was measured. The value in Figure 1 is the 90-day strength.] Thus, the compressive strength of Batch 30 decreased somewhat after Day 28, that for Batch 31 rose very slightly and that for Batch 32 increased dramatically. The addition of these two acids may be affecting the crystallography of the cementitious portion of the concretes. Intermediate amounts of acid appear to increase strength without degradation, while larger amounts cause deterioration. Much work needs to be conducted to understand the causes and effects of strength enhancement by acid addition. METAL STABILIZATION IN MSW ASH-CONTAINING CONCRETE Samples of the first nine concretes containing MSW ash, made in this project, were extracted by the project team according to the EP toxicity method and the concentrations of the eight RCRA metals in the extracts were measured by Geochemical Testing of Somerset, Pennsylvania. The results of these tests are given in Table 1. The two metals, cadmium and lead, which caused the MSW ash to fail the EP toxicity test, have been well stabilized in all nine concretes. COMPARISON OF BEHAVIOR OF FRESH AND AGED ASH This portion of the study was conducted two years after the earlier portion. Aged ash was drawn from the fourth (and final) batch of ash that had been collected several years previously. Fresh ash was obtained from the Dutchess County MSW Incinerator. It was drawn from the ash conveyor prior to lime addition. This portion of the study utilized mortar, rather than full concrete containing coarse aggregate. The method of mortar production, based upon ASTM C109, was as follows: 0 0 Add water and mix. 0 0 Place the ash into a mixing bowl. Add hydrochloric acid (if it is to be used) and mix, Add cement (to a water/cement ratio of 0.81) and mix. Fill six plastic two-inch molds with mortar, place them in a curing room for 24 hours, break them from the molds, and continue curing for six more days. Measure the compressive strength of each of the six cubes, using a universal testing machine; calculate the average strength of the four strongest cubes. Figure 2 provides a record of the compressive strengths for eight aged ash-containing moms, six prepared with varying amounts of hydrochloric acid and two with no acid. Figure 854 ? EPMcity Cmaumt Madurn Auaabla Umt ksenlc 500 &durn io0.m W 500 MsrmlY azo Gelenlum 1.00 Silver 50.00 c8dmlUm 1.00 chmmlum Lao I Primary PIaln Ave Mar A q Mmt Mnidng ca#rt*, Conuem Qncmm &h Ash WaBl conpol Bemplea Bem@m 8nmF4Ia samples 8tandarda M . 1 tog lto9 l a d 2 lend2 am awe o.oiez 0.05 aoit 0.018 1 .am 1.09 0.7WO 1.38 ales 0.240 am 0.01 0 . 1 5 ~ 0.n am 7.480 aoio am 0 . m 0.02 am 0.m 0.010 0.01 0.1061 0.47 1.375 1.m 0.050 am 0.- a09 0.0) 0.050 am a m 0 . m 0.004 0.001 0.001 am 0.01 0.0167 0.M 0.025 0.m 3 provides a record of the compressive strengths for three fresh ash-containing mortars, two Prepared With varying amounts of hydrochloric acid and one with no acid. For comparison of Figures 2 and 3 with Figure 1, it may be noted that 100 mmol of acid in Figures 2 and 3 corresponds to 0.032 gram moles acidlpound of ash in Figure 1. First, it may be observed that all of the mortars were prepared with relatively low amounts of acid. The largest amount of acid, about 0.05 gram moles acidlpound of ash, was used in Batch 6. This corresponds to the amount used in Batches 31 and 37 of the earlier portion of the study. Thus, the increases in compressive strength with increasing amounts of acid, observed in Figures 2 and 3 to be under 100%. are as expected, based upon the experience recorded in Figure 1. From a comparison of Figures 2 and 3 it is clear that the compressive strength of mortar made from fresh ash is over six times that of mortar made from aged ash. Fresh ash has a certain amount of pozzolanic character which is lost as it ages. It is also clear from this comparison that acid addition is much more effective in increasing the compressive strength of mortar containing fresh ash than for that containing aged ash. Mortar with fresh ash doubles in strength with the addition of about 60 mmols of acid, while mortar with aged ash may require 120 mmols or more of acid for the same effect. Thus, there may be a phenomenological linkage between the strength enhancement caused by acid addition and the pozzolanic nature of the ash. CONCLUSIONS The addition of common acids, such as hydrochloric and acetic acids, to mortars and concretes containing MSW incinerator ash, increases the compressive strength of the final product. The increase is more pronounced when the ash is fresh. Aging of ash degrades the final strength of the mortar and also reduces the effect to be expected by acid addition. It should be noted that the results of this study are quite preliminary in nature. Much more work needs to be done to verify and quantify the trends and to ascertain their causative mechanism. REFERENCES [I] [2] [3] Cobb, J. T., Jr., et al., Clean Enercy from Waste& Coal, M. R. Khan, ed., ACS Symposium Series, 515, 1993, p. 264. Reed, D. I., MS Thesis, University of Pittsburgh, 1992. Lewis, J. T., 111, MS Thesis, University of Pittsburgh, 1994. Table 1. EP Toxicity Tests on Nine Concretes Containing MSW Ash 855 Figure 1 . Effect of Acid Concentration on Compressive Strength of Concrete Containing MSW Ash 0 P6 8 16 18 u) la, 1% Amount of Acid (mmol) Figure 2. Effect of Acid Concentration on Compressive Strength of Mortar Containing Aged MSW Ash 856 1 1100 1000 0 10 m 30 Amount of Acid (mmol) Figure 3. Effect of Acid Concentration on Compressive Strength of Mortar Containing Fresh MSW Ash 857 INVESTIGATION OF THE CO, ABSORPTION CAPACITY OF DRY FGD WASTES Taulbee, D.N., Graham, U., Rathbone, R.F., and Robl, T.L. University of Kentucky-Center for Applied Energy Research 3572 Iron Works pike Lexington, KY 4051 1 Keywords: CO, or carbon dioxide, Rue-gas desulfurization or FGD, natural gas, Calcium Oxide ABSTRACT Numerous utility boilers and tail-gas desulfurization units utilize lime or limestone-based sorbents to remove sulfur oxides generated during the combustion of fossil fuels. Such units generate about 20 million tons of flue-gas desulfurization (FGD) wastes annually in the US., the bulk of which (-95%) is discarded in landfills or holding ponds.' Thus. commercial utilization of FGD wastes would benefit from both a plentiful low-cost raw material as well as a signifcant savings in disposal. One such use may be for the reduction of CO, in multi-component gas streams. During the removal of SO,, the lime added to or generated in the desulfurization unit, is not iully utilized. That is, a portion of the Ca fed to the unit is not sulfated (remains as CaO or Ca(OH),). In some FGD wastes, the fraction of available Ca is quite high (> In), paniculariy for dry wastes. When hydrated, such wastes exhibit a strong aftinity to absorb CO, at ambient temperature. Funher, both the kinetics and extent of absorption are favorable as CO, initially at -2.5 volume% was rapidly reduced to below the detection limit of the measurement device (ppm range) used in this study. Leaching behavior and changes in the mineralogy of the FGD samples exposed to CO, are also discussed. INTRODUCTION. Over the past decade, numerous FGD units have been added to existing utility boilers in an effort to satisfy federal mandates on SO2 emissions. Such units are usually classified as either wet or dry &pending on whether the absorbent is used in a slurry (wet) or as a hydrated solid. Wet scrubbers capture sulfur chiefly as gypsum (CaS04R,0) with some sulfite formation (e.g., CaS03.2H,0). Dry technologies such as AFBC produce a dry product in which sulfur is captured mostly as anhydrite-CaS04 or for the dry tail-gas units such as spray drier and duct-injection, sulfur is captured as gypsum, anhydrite or hemi-hydrate (CaSO,H,O). Dry FGD by-products also differ from their wet-scrubber counterparts in that a significant portion of the calcium in the dry waste remains unsulfated. This Ca is present as either calcium oxide, CaO, or as slaked lime, &(OH),. Because FGD wastes, particularly dry FGD wastes represent relatively new materials, <6% of the -20 M tons of FGD wastes generated in 1993 are currently finding commercial uses.' The work described here represents a preliminary examination of the capacity of dry-FGD wastes to remove CO, from multi-component gas streams. Such an absorbent may have numerous commercial uses, e.g., gas purification, removal of C02 during H2 production, etc. However, the current study focused on the potential to reduce CO, in simulated natural-gas streams. As a rule of thumb, the costs associated with available COz-removal technologies (wet scrubbers, molecular sieves, membranes) are prohibitive for gas wells that produce less than about 100,OOO SCF/day? This effectively eliminates commercial production from low-porosity, carbonate-conta@ing strata common to many gas-producing deposits. Thus, a low-cost CO, absorbent that can be safely disposed or marketed (road base or fertilizer) may have applications in the natural-gas industry. In this study, CO, absorption capacity was evaluated for waste samples generated from different utility boilers, one demonstration plant, and tests conducted under four sets of conditions in a single pilot plant. With the exception of a utility-derived fly ash used as a control, all samples examined are dry-FGD waste materials. Absorption capacity was examined for both hydrated samples as well as aqueous slm'es. As of this writing, only gas blends containing inert gases and CO2 or inert gas and COJCH4 have been tested. Additional tests are planned to evaluate absorption behavior during exposure of hydrated FGD-wastes to a gas blend containing 3s. CO,, and CH,. EXPERIMENTAL. Absorption Reactor. A schematic of one configuration of the reactor used to measure CO, absorption for the hydrated samples is shown in Figure 1 (shown with 9" X 1/4"-i.d. tube reactors). This is essentially the reactor described in earlier adsorption/cracking studies of liquid hydrocarbons3 with some modification. The more significant modifications include the inucduction of standard gases containing COdAr or COdCHdAr via the entry line in which pure Ar was pfiviously metered, plugging of the liquids inlet, use of 4 X 3/8"-i.d. reactors in addition to the 9" x 1/4" reactors (most of the hydrated-sample tests), and placement of the 4" reactors in a vertical alignment to provide a more uniform flow of gases through the hydrated samples. Essentially the Same measurement system was used to measure absorption of CO, by the water/sample slurries except that a pair of 250-mLcapacity gas scrubbers were substituted for the ss tube reactors. Samples. Many of the study samples examined were obtained from commercial utilities that 858 I preferred to remain unidentified. Thus, only cursory descriptions of the samples will be given and some producers will remain anonymous. A total of 11 samples were examined. A very brief description along with the identification label used in this repon is given in Table I. The fly ash utilized as conml &FA) is a Class F fly ash from a pulverizedcoal-combustion (F'CC) utility boiler operating on bituminous coal. The fluidized-& combustion materials (FU-FA/BA and CC-FA/BA) were derived from circulating or entrained flow units operating on high-sulfur bituminous coal. The coarse material @A-bed ash) was drained from the bed while the finer material (FA) represents cyclone and baghouse catch. These samples differ primarily in panicle size and relative proportions of free lime. Two types of dry post-combustion flue-gas material were utilized in the study, a spray-dryer ash from a large industrial boiler in the Midwest, and materials from the Coolside duct-injection technology. The Coolside materials include a sample (CS) from Ohio =son's 1990 demonsmation of the technology at its Edgewater power plant4 as well as materials derived from the CONSOL'S Coolside pilot plant in Library, PA (PPl-PP4)S Run Conditions and Procedures. All absorption measurements were made at ambient temperature. Nominal gas flows of 100 mumin (ambient temperature) for the hydrated-solids tests and 150 mumin for the slurry tests were metered through each reactor. The gas streams were comprised solely of N, in the bypass line and a standard-gas blend (either Ar/CO,/He- 7.5/2.5/90.0 ~01%o; r Ar/COZ/cH,-30.4/49.6/20.1 ~01%i)n the absorbent line. Argon was included as a tracer gas to eliminate measurements problems associated with minor leaks or instrumental drift. Hydrated samples of known water content were obtained by careful blending of distilled water and dry waste. Between 2 and 5 g of the hydrated samples were packed into the absorbent reactor between quartz-wool plugs. The bypass reactor was packed with 6 g of Ottawa sand. For the slurry absorption experiments, -5 g of dry sample was added to 200 mL of distilled water in a 250-mL gas scrubber. The slurry was stirred with a magnetic stir bar for the duration of the experiment. Gas concentrations in the combined sample/bypass exit stream were determined with a VG-quadrupole mass spectrometer (QMS). This unit was operated in a selected-ionmonitoring mode in which intensities for m/e 18-H20+, 20-d'. 28-N2+, 40-h'. 44-C02+, and 15- CH3+ (for methane) were recorded at approximately 1-second intervals. For both the hydrated-solids and slurry tests, data collection was initiated with the switching valve in the bypass position, i.e., with the CO, stream passing through the bypass reactor. After a minimum of 150 data points were collected (usually 2-4 minutes), the valve was rotated so that the C02 stream was switched to the absorbent reactm and the N, stream was simultaneously switched to the bypass reactor. After a selected exposure time, the valve was returned to the initial position to reestablish the QMS baseline. Following data collection, the QMS data were imported to a spreadsheet where the molecular-ion signal for CO, (m/e-44) was ratioed to the Ar-ion signal (m/e-40). The curves described by the COgAr ratio were then numerically integrated over the interval during which CO, was routed to the absorbent reactor to determine the fraction of the 0,abs orbed. The fraction of CO, absorbed was calculated to an absolute basis then to SCF of C02 absorbed per ton of waste. Several of the hydrated samples were retained in sealed vials for post-run XRD analysis to investigate changes in mineralogy resulting from CO, absorption. Likewise, selected slurry waters were retained in sealed containers for ICP analysis of heavy metals/cations. RESULTS A plot of the C02/Ar ion-intensity ratios is shown in Figure 2. In this run, the Ar/C02 blend (2.5% CO,) was initially flowed through the sand-packed bypass reactor, switched to the absorbent bed packed with hydrated FU-fly ash at 3 min, returned to the bypass reactor at 53 min, then again to the absorbent reactor 3-min later. This particular plot demonsuates both the rapid kinetics and the extent to which CO, was absorbed in the 9 reactor as well as provides an indication of the reproducibility of the QMS response during the two bypass- and expose-mode intervals. A more complete run, also conducted in the 9" reactor with 2.5% CO,. is shown in Figure 3. This latter plot demonstrates how the QMS response collected as the CO, passes through the bypass reactor (before and after the valve switch) provides a suitable baseline for integration of the ion intensities recorded during passage of CO, through the absorbent bed. Absorption by hydrated solids. Absorption of COz is shown in Figure 4 as a function of water content. These plots were prepared from runs in which 2-5 g (dry basis) of hydrated sample were exposed to flowing 0 2 (4 9.6%; -100 mumin) in the 4" X 3/8"-i.d. reactors. In dry form, none of the wastes examined showed a smng affinity for COT However, with addition of H,O, the absorption capacity increased rapidly until the water content was sufficiently high to create a mudlike texture in the waste samples. At the highest moisture levels, absorption capacities declined, presumably limited by sample permeability. Maximum absorption ranged from -1,700 SCF/ton for the FU-FA to -300 SCF/ton for the conml fly-ash sample (L-FA). Limited testing in the 9 reactor showed absorption in excess of 2,000 SCF/ton for the FU-FA sample. 859 Absorption by watedwaste slurries. For the final phase of the study, the ss tube reactors were replaced with a pair of 250-mL gas bubbler/scrubbers. As described earlier, -5 g of solid waste were added to 200 mL of distilled water in the absorbent scrubber (bypass scrubber contained 200 mL of distilled water). The gas blend containing C02, Ar, and CH, (-50:30:20) was bubbled through the water in the bypass reactor during the initial bypass interval then switched to the absorbent sluny for up to one hour before returning to bypass. The QMS data collected during the slurry tests was processed the same as those collected during the hydration studies. Results from the slurry tests are shown in Figure. 5. Absorption ranged from less than 1.OOO SCF/ton for the L-FA control sample to more than 3,500 SCF/ton for the FU-FA and PP4 samples. These results generally correlate with the free lime data in Table I with the exception of the two samples of bed ash. The significantly larger particle size of the bed ash samples likely limits diffusion of CO, into the particle and explains their lower than expected absorption capacity. Although removal of CO, was greater in the slurry tests on an absolute basis, neither the rate or level of maximum absorption was as great as measllred for'the equivalent hydrated samples. Slurry runs typically required 10-20 minutes before CO, response returned to 95% of the original level. Further, at maximum absorption, CO, was typically reduced from 49.6% in the feed stream to around 12-15% in the exit stream for the slurry tests as compared to 1% or less in the exit stream for the hydrated-waste tests. However, the shape of the adsorption curves obtained from the slurry tests is thought to be more of a reflection of scrubber design rather than absorption kinetics. It is believed that both kinetics and the maximum level of absorption can be markedly improved with a more efficiently designed bubbledscrubber (smaller bubbles, longer contact time). Post-run analysis of hydrated solids and slurry waters. Selected samples from the hydration tests were examined by XRD and compared to similar analyses of unexposed samples. The XRD results indicate that the only significant change in mineralogy following absorption was an increase in calcium carbonate (CaC03). There was also a minor increase in etningite, a hydrous calcium sulfoaluminate phase that can substitute carbonate for sulfate in its structure. However, since the samples remained moisturized following exposure. (i.e., they may continue to react), it is possible that these minor changes occurred after the absorption run and before the XRD analysis. Regardless, it appears that the CO, reacts primarily with available Ca (CaO or Ca(OH),) to form carbonate. Two of the water samples retained from the sluny tests were analyzed for metalkations content (Table 11). Elemental concentrations are in large part controlled by pH which was >12 for these samples. At such high pH, most transition metals are relatively insoluble. This likely explains why none of the elements tested were detected at levels sufficient to suggest unreasonable disposal problems due to the leaching of toxic elements from the waste samples into the slurry water. SUMMARY The results obtained in this study clearly show that when hydrated, FGD wastes exhibit a high affinity for CO,, ranging as high as 3,600 SCF/ton. Further, there are significant differences in the capacity of FGD wastes generated in different plants to absorb COT With the exception of the larger particle-size bed-ash samples, these differences appear to be controlled by the available lime content of a given waste. This is supported by the free-lime data in Table I and XRD analysis which indicated that the absorbed C02 reacts with free lime to form CaCO,. Thus, dry wastes from less efficient utility scrubbers should produce higher-capacity CO, absorbents. Finally, analysis of the slurry waters suggests that process waters that may be used in a liquid scrubber can be safely disposed following contact with FGD wastes. ACKNOWLEDGEMENT Jefferson Gas Transmission for their assistance with this project. REFERENCES. The authors would like to thank K. Saw and B. Schram of the UK-CAER and K. Baker of 1. Coal Combustion Byproduct (CCB); Production & use: 1966-1993. Report for Coal Burning Utilities in the United States. American Coal Ash Association. 1995, Alexandria, VA, 68p. Personal communications with Wiiliam Johnson and associates at NIPER in Bartlesville OK and Tom Cooley with Grace Membrane Systems, Houston, TX. Taulbee, D. N., Prepr., ACS Div. of Fuel Chem., 1993, 38, #I, American Chemistry Society, Washington, DC, 324-329. K a n q , D.A., R.M. Stanick. H.Yoon, et& 1990, Coolside Process Demonstration at the Ohio Edison Company Edgewater Plant Unit 4-Boiler 13. Proceedings, 1990 SO, Control Symposium 3, EPRVU.S. EPA, Session 7A,, New Orleans, LA, 1%. Withum, J.A., W.A. Rosenhoover and H. Yoon, 191 Proceedings, 5& Pittsburgh Coal Conf. University of Pittsburgh, 84-96. 2. 3. 4. 5. 860 J Table I. Waste samples examined Table IL Concentration (ppm) of cation/metals in the waters retained from slurry-absorption tests. ,_______________________________.___...______.____ Metering Valves -=- Flow check j Switching valve( I I Heated valve OW* -> i... .. . . .__ .. _ _._ ... . . . .. .. . .. . .. .. . . . _ ._ ~ _ _ __. . Reactor furnace (above ambient tmp Only) .........‘4 ................................................... ...__.._ Adsorbent reactor I 1 QMS j ._....._B_y-p_as_s. _re.a.c_to.r. ...~~......~.........-..-..~~.....--.~~......~.~... Figure 1. Schematic of the absorption reactor used for the hydrated samples. 861 I 1 0.4 Time (Min) Figure 2. CO-JAr ion-ratio curve showing C02/Ar ratios as gas blend is routed through a) bypass reactor, b) absorbent reactor, c) rem to bypass reactor, and d) return to absorbent reactor. 2000 L-Fly Ash .............. ................................................. - ........ -. .Y) ;0 3 ............................................................. ......... 0 o,2 .r ......................................................................................... lo g o., OO 40 80 120 Time (Min) Figure 3. CO,/Ar ion-ratios during run with FU-fly ash in 9" reactors (1.5 g dry FU-FA; 0.58 g H,O; 2.5 mL/min CO,). oi . IO . io . 30 ' i o Addcd WaIer ( ~ 1 % ) 2o00]FU-Eed Ash Addcd water ( ~ 1 % ) Figure 4. Absorption of CO, by hydrated wastes as a function of water content. PP3 a-300 PFZ PP4 SD FU-FA CC-FA Figure 5. CO, absorption by watedwaste slurries. a) Coolside wastes; b) all others. 862 TECHNOLOGIES OF COAL FLY ASH PROCESSING INTO METALLURGICAL AND SILICATE CHEMICAL PRODUCTS Solomon Shcherhan, Int. Assn. of Science, Inc., 1 IO Bennett Ave., 3H New York, NY 10033; Victor Raizman, Assn. of Engineers & Science, New York; lliya Pevzner, Coral1 Co. of St.Petersburg Engineer Academy, St.Petersburg Keywords: coal fly ash, recovery of metals and silica, utilization ABSTRACT A study and industrial testing have made for the recovery of aluminum, iron and silica from coal ash, produced by utilities. Alkaline technologies for coal fly ash processing were developed that made it possible to separate the main components of fly ash (SiO, , AI,O,, Fe,O,) and utilize them separately, producing a large variety of useful products. Some of these technologies have already been successfully tested in pilot programs. INTRODUCTION The problem of effective utilization of solid waste from coal-fired power plants is of great importance to many countries. The coal burning utilities of the former Soviet Union generate more than 100 million tons of solid combustion by-products each year. Approximately 1 billion tons of solid waste from utilities is placed in storage and disposal areas. The combustion of coal by utilities in the United States results in the production of over 80 million tons of solid by-products each year yet less than a quarter of coal ash is presently being utilized [l]. The various fields of fly ash application are known [I-31. In the former Soviet Union much attention has been given to the area of research that is called 'High Technology Ash Application' in the United States [I]. This research focuses on the development of technologies for ash processing with recovery of valuable minerals and metals in particular for the recovery of aluminum. The necessity of this research is caused by the need to find new ways for the utilization of fly and bottom ash and simultaneously to solve the problem of expanding the source of raw materials used in aluminum industry. Ash contains approximately 1.5-2 times less aluminum oxide than common aluminum raw materials (20-35% A120, in ash as compared to 50-62% A1203 in bouxite). The high level of silica in ashes (40-65% SO,) makes it impossible to process them by the easiest and the most economical Bayer method and by the other methods of direct alkaline alumina extraction. Therefore for ash processing other methods are studied: acid, thermal, thermal reducing, electrothermal melting, new alkaline methods. This paper is dedicated to the development of alkaline methods of ash processing. The laboratory research of alkaline methods of fly ash processing have been done at the Problem Laboratory of Recovery of Light and Rare Metals (Kazakh Politechnical University, Alma-Ata). Largescale testing of the alkaline technologies has been conducted at the pilot plants of the All-Union Aluminum-Mapesium Institute (VAMI, St.Petersburg), State Research and Designed Cement Institute (GIPROCement, St.Peterburg), and the Institute of General and Inorganic Chemistry (Erevan). EXPERIMENTAL Chemical and Mineralogical Description of Ash Samples Chemical analysis of typical fly ash derived from Ekibastuz coal are given in Table 1 Table 1 CHEMICAL ANALYSIS (Wt.%) OF FLY ASH power C o n s t i t u e n t plant SiO, AI,03 Fe,O, Ti 0, CaO MgO Na20 K20 LO1 Total pavlodar 59.82 27.79 5.48 1.65 1.20 0.72 0.40 0.62 4.50 97.68 Emak 60.50 27.20 5.05 1.90 1.60 0.58 0.30 0.60 4.00 97.73 Troitsk 58.48 30.21 4.78 1.95 1.12 0.66 0.30 0.55 0.80 98.05 The major constituents of the Ekibastuz ashes are silicon dioxide, aluminum oxide and iron oxide which represent about 9044% of the total. Ekibastuz fly ash is characterized by low content ofNa20 and K,O (<= 1%) and CaO and MgO (1.78-2.2). The mineral part of Ekibastuz coal is represented by kaolinite (60-68%), quartz (27-30%), sider- / 863 ite (34%), calcite (2.5-3%), magnesite (l-l.5%) and gyps (0.3-0.5%). At burning of coal the mineral part of it is subjected to a short termed influence of high temperatures, which results in kaolinite decomposition, formation of mullite and glassy phase, thermal1 dissociation of carbonate, polymorphous conversion of quartz into high temperature modification of silica. All these transformations predetermine the mineralogical composition of ash (Table 2). Table 2 MINERALOGICAL ANALYSIS OF SILICON AND ALUMINUM CONTAINING MINERALS IN EKIBASTUZ ASH Mineral Mass. Yo A n a l y s i s Name in Ash Crystal Optics X-Ray Diffraction IR-spectroscopy d,A v, cm-' ~ ~~ ~~~ Mulite 30-35 n,=1.666, n,=1,654 5.45; 3.41; 3.36; * a - np= 0.012 2.88; 2.55; 2.21; Glassy 48-51 I . N=1.503 Amorphous 1100-1050; Phase 2. N=1.534-1.539 Amorphous 780; 475; Quartz 2-10 Ng=1.544; Np=1.531 4.27; 3.78; 2.44 I 160; 1095; 2.28; 2.23; 1.82 800-790; 465; Righ 0-45 No=1.486; Ne=1.454 4.09 (intensive); 1160; 1100; Temperature No - Ne = 0.002 2.51; 2.88 975; 820; Silica - -~ -_ -. .. __-- ~ __ * The absorption regions of mullite appear after dissolving the uncombined silica of ash. As it follows from Table 2 Ekibastuz ash basically consists of glassy phase (SiO, and SiO, with admixtures), mullite (3 AI,0,.2SiO2) and quartz (SO,). Well calcinated ash (Troitsk) contains a high temperature crystalline modification of silica with properties close to crystobalite. Most important for the alkaline methods of ash processing is the process ofkaolinite decomposition with the formation of mullite and the isolation of the most part of silica in an uncombined (free) form. This process is described by the summary reaction: 3[A12Si,0,(OH),]------->3 A1,03.2Si0,+4Si02+6H,0 kaolinite mullite silica As a result of this reaction aluminum is concentrated in mullite and about 67% of the kaolinite silica is isolated in a free form. Together with silica of quartz and its high-temperature modifications about 70-80% of silica is contained in the ash in the free form. This creates the necessary prerequisites for aluminum oxide and silicon dioxide separation. Stated phase separation of aluminum oxide and silicon dioxide in ashes came to be a basis for research and the development of alkaline methods for ash processing [4]. Interaction between Ash Minerals and Alkaline Solutions According to the data of mineralogical analysis (Table 2) mullite, glassy phase, quartz and its high temperature modification-crystobalite are the main aluminum and silicon containing ash minerals. The comparison of these minerals dissolubility in the alkaline solution is shown in Figure I . The comparison of the curves (Figure I ) shows that mullite ( I ) and quartz (2) have a small dissolubility in alkaline solution while crystobalite ( S ) dissolves practically completely after 4 hours of alkaline treatment at I05C [5]. The glassy phase (3) also has good disolubility in alkaline solution. Its presence and dissolution are determined by a comparison of IR-spectra of ash (Figure 2) and its residue after alkaline treatment. IR-spectra of ash (Figure 2.1) contain the absorption regions 1100 and 800-780 cm-' which are the characteristic regions of silicates like quartz, crystobalite, and amorphous silica with threedimensional tetrahedrons of SiO, frame. In the IR-spectra of residue after ash alkaline treatment (2) the regions of quartz, glass and crystobalite have completely or partially disappeared and absorption regions of mullite (1 180,970-920, 880-850 cm" ) have appeared. XRD analysis of residue after alkaline ash treatment reveals the complete disappearance of 864 Figure 1. Interaction between alkaline solution and ash compounds: I- mulite; 2 - quartz; 3 - ash glassy phase; 4 -ash cristobalite; 5 - synthesized cristobalite. no NRS Figure 2. Infra-red Spectra: I - ash; 2 - its residue; 3 - ash calcinated at 1250°C; 4 - its residue. BO 60 40 0 t 2 . ? 4 5 ,4 OUR5 Figure 3. Silica extraction from Fly Ash 865 crystobalite maxima d,A: 4.09-4.10; 4.52; 3.51 which is well coordinated with the data of crystobalite alkaline dissolubility (Figure I). The aforementioned data of chemical, XRD, and IR-spectroscopy research shows that uncombined (free) ash silica is extracted from ash by the alkaline solution. Hydroalkaline Recovery of Silica from Fly Ash The influence of various factors on the percentage of silica extraction from fly ash is shown in Figure 3. The data in Figure 3 shows that free silica is extracted from ash at low rates (temperature 105.C for a duration of 3-4 hours). The process can be realized at atmospheric pressure. The essential augmentation of silica recovery has been reached by means of ash activation which increased the eficiency of silica extraction by 12-20% (Figure 3, curves 6-9). The intermediates after ash hydralkaline treatment were the silica alkaline solution (SAS) and the solid residue enriched by aluminum oxide (concentrate of alumina). Chemical composition of SAS , gdm-': Na,0=160-220; Si0,=100-250; AI,03=2-7; Fe,O,=O. 1-0.9. Concentrate of alumina included %: 44-55 AI,O,; 30-27 SO,; 5.5-10 Fe,O,. Recovery of Iron from Fly Ash by Magnetic Separation In a number of studies, magnetic separation was applied as a pre-stage before the main operations of ash treatment[6]. Ekibastuz ash consists of 4-lO% Fe (as Fe,O,). The possibility of recovering the magnetic fraction from Ekibastuz ash and its classified fractions was shown in [7]. The mabmetic fractions after raw ash magnetic separation were rich in Fe (60-62% as Fe,O,). Classified fractions contained 57.6-66.4% Fe as Fe,03. Output of the magnetic fractions was 2.12-5%. The non-magnetic residues were depleted of Fe and contained 2.6-3.6% Fe as Fe,O,. Technology of Alkaline Fly Ash Processing. The Principle Process Flow Sheet The described findings of hydroalkaline recovery of silica were taken as a basis for the design of the process flow sheet of fly ash processing into metallurgical, silicate chemical products and building materials. The principle flow sheet (Figure 4) includes the hydroalkaline silica extraction from fly ash. This operation allows one to extract the good part of ash silica (60-77%) into the alkaline solution and then to process it into various silicate chemical products (sodium and calcium metasilicates, sodium-silicate mixtures, amorphous and crystalline silica and others). The solid intermediates from ash extraction-alumina concentrate-can be processed into alumina, aluminum, and aluminates by thermal or hydrochemical alkaline methods or can be used for aluminum-silicon alloys, refractories, and concrete production. Mud of the alumina production is a valuable raw material for cements. i p t Fe - CONCENTRATE HYDR~ALKKIN E SILICA EXTRACTION F ALUMINA SILICATE ALKALINE SOLUTION M PROCESSING INTO REGENERATION PURE SILICATE PRODUCTS: I SILICATES, SILICA, ZEOLITES, ALUMINA WHITE SOOT, COMPONENTS PRODUCTION FOR GLASSES, CERAMICS, I CEMENT AND OTHER variants - invariable operation ALLOYS REFRATORIES BUILDING MATERIALS ALUMINA CE~ENT Figure 4. Flow sheet for alumina, silica and iron recovery from ash 866 Large -Scale Testing of the Alkaline Technologies Practically all of the main technological operations of the fly ash processing have been tested in pilot programs: ash activation, hydroalkaline silica extraction, settling and filtration of ash pulp, washing of the alumina concentrate, processing it into alumina, producing of portland cement from mud, silica alkaline solutions processing into sodium and calcium metasilicates. Alumina Output was made up of 86% A1,0, (90-91.7% at the standard leaching). REFERENCES I. Golden D.M. 'Research to Develop Coal Ash Uses.' Ninth International Ash Use Symposium. Proceedings. Orlando, Florida, January 22-25, 1991. . 2. Shpirt M.Y. 'Nonwaste Technology: Utilization of Waste from Mining and Processing of Solid Combustible Fossils. Moscow: Entrails' 1986. 3. 'The Combined utilization of Coal Ash of the SSSR in the National Economy', Abstracts, Meeting, Irkutsk, Russia, 11-13 July, 1989. 4. Shcherban SA., Nurmagambetova S.Kh. 'The Combined Utilization of Coal Ash of the SSSR in the National Economy', Abstract, Meeting, Irkutsk, Russia, 11-13 July, 1989, p.91 5. Suliaieva N.G., Shcherban S.A., Tazhibaeva S.Kh, Romanov L.G. 'Combined Using of Mineral Row Materials' Periodical. 1982, #3, pp.62-66, Alma-Ata (Russian) 6. Hemmings R.T., Beny E.E. and Golden D.M. 'Direct Acid Leaching of Fly Ash: Recovery of Mettals and the Use of Residues as Fullers'. Eight International Coal Ash Utilization Symposium, Washington D.C., October 29-31, 1987, p.38-A 7. Shcherban S.A., Sadykov Zh.S.,Pustovalova L.S., Ergaliev G.B., Fridman S.E. 'Ekibstuz Ash Processing into Alumina with Iron Concentrate Production'. Combined Using of Mineral Raw Materials. Periodical, 1985, N4, pp.68-71, Alma-Ata (Russian) I J I 867 CHARACTERIZATION OF PYROLYSIS OILS OBTAINED FROM NON-CONVENTIONAL SOURCES J. Houde Jr., J.-P. Charland, Energy Research Laboratories, CANMET, Natural Resources Canada, 555 Booth St., Ottawa, Ont., Canada, KIA OGI. E-mail: jean.houdeQx400.emr.ca Keywords: pyrolysis oil, automobile shredder residue, pulp and paper sludges INTRODUCTION The effluents of pulp bleaching are the main problem of wastewater disposal faced by the pulp industry because of their non-biodegradability. Today the demand for quality discharges requires better methods than conventional biological processes. The changes recently proposed to the federal regulations for controlling discharges from pulp and paper industry operations in Canada have required many operations to install secondary biological effluent treatment process. Such treatment often produce sludges that must be removed from the system and disposed of routinely, usually daily or weekly. At present, Ontario and Quebec have the strictest solids disposal regulations in Canada, with leachate critena that approximate the U.S. Environmental Protection Agency (EPA) toxicity characteristic leaching procedure (TCLP) standard [l]. Most pulp and paper mills in the U.S. have some form of biological treatment, the majority having their own treatment plants, but some are tied into publicly owned treatment plants. Of those which have their own treatment plants, two thirds have aerated stabilization basins and one third have activated sludge [2]. Industrial wastewater secondary treatment using activated sludge techniques has gained increased acceptance in the paper industry. The advantages of activated sludge treatment over conventional aerated lagoons are less real estate requirement, less odour emissions, lower capital cost and higher sludge treatment efficiency [3]. One of the main disadvantages is the production of a large amount of sludge which is difficult to dewater and costly to dispose of. The Canadian pulp and paper industry produces about 2,200 t/d of sludge from wastewater treatment operations. Most of this sludge is produced in wood room or primary clarifiers treating total mill effluents. Approximately 54% of this total is incinerated, with most of the balance being landfilled [4]. Long term environmental uncertainties associated with landfilling, as well as increasing costs and a drive to greater energy efficiency, nuke it preferable to use the sludge. When old cars and trucks are sent to scrap yards for shredding to recover ferrous and nonferrous metals, large quantities of non-metallic waste, referred to as autofluff, are generated. Autofluff is a lightweight mixture of plastics, textile, glass, rubber, foam, paper and rust. Also, this material is contaminated with oils, lubricants and other fluids used in automobiles. The trend to substitute lightweight materials for iron and steel reduces the proportion of recycled metals and increases the amount of waste produced [5-81. The economics of the shredding industry relates to the recovery and resale of the ferrous metal which is used to produce high quality steel. Over the years, the use of ferrous metals in automobiles has declined whereas that of plastics and nonferrous metals has increased. There is a clear economic and environmental advantage to salvaging cars, since metals can be utilized that would otherwise end up as trash. In 1992, in Canada, one million cars and trucks were sent to scrap yards, while in the U.S., 11 million vehicles were taken off the road [71. Autofluff production is estimated at 1,80O,O00 and 2,860,000 t/a, respectively for Canada and the U.S., most of which ends up in landfill sites. Of the various disposal alternatives, conversion of these materials by pyrolysis or other proven technology to possible value-added products would reduce the use of costly landfill sites for disposal and utilize this potentially valuable resource. The Wastewater Technology Centre (WTC) of Environment Canada has been developing one such technology since 1982. The thermoconversion process involves low temperature treatment of materials such as sludge from the pulp and paper industry or autofluff, to produce liquid'and solid fuel products. A key technical feature of this conversion is the formation of a byproduct oil referred to as pulp and paper sludge derived oil (PPSDO) or autofluff oil. The thermal conversion process has been extensively described elsewhere [9-111. In 1992, Enersludge Inc., WTC and CANMET's Energy Research Laboratory (ERL) of Natural Resources Canada undertook a joint R&D program. ERL investigated pyrolysis oils obtained from autofluff and pulp and paper mill sludges. .Analytical results are presented as well as a comparison of these oils with those obtained from tires and municipal sewage sludge. EXPERMENTAL A set of samples was received for each pyrolysis experiment (PPSDO and autofluff oils). The first samples received included compounds with boiling points up to 150°C (-150°C) whereas the second samples contained compounds boiling above 150°C (+150°C). The +150°C samples 868 I / were funher distilled to yield three additional fractions each. Fractionation was performed using automated ASTM D-1160 short path distillation apparatus. Fraction cuts were selected to reflect conventional cut points from the petroleum industry: b 15O0C-35O"C - typical cut point for middle distillates w 350°C-525"C - typical cut point for heavy gas oils b +525"C - usual distillate-residue cut point Physical and chemical analyses were performed according to appropriate ASTM methods. The 'H, DEFT I3C and I3C data were acquired on a VARIAN XL300 operated at 300 MHz h the 'H mode and 75 MHz in the 'F mode. The pulse sequence in DEm experiments transfers the polarization of the hydrogen to the carbon nucleus to selectively increase its signal. The polarization transfer effect is dependent on the number of hydrogens bonded to a given carbon nucleus. This technique is used to distinguish between primary, secondary, tertiaty and quaternary carbon atoms. The NMR spectra are presented in Fig. 1. Infrared spectra were obtained using a PC-driven BOMEM MBlOO Fourier transform infrared (WR)sp ectrometer fitted with a standard sample mounting device. The IR spectra from liquids were collected using *a liquid cell fitted with a 13-mm diam circular window. The liquid cell windows are made of KBr and are separated by 0.02 mm. The spectra of the 350°C-525"C colloidal fractions were collected using the smearing technique on conventional 13-mm circular KF3r discs. The IR spectra are presented in Fig. 2. GUMS work was performed on a Hewlett Packard 5890 GC coupled to a medium resolution mass spectrometer (MS). Chromatographic separation of the sample was done using a Hewlett Packard HP-1 30 m long methyl-silicon bonded fused silica capillary column of medium resolution fitted on the GC. This column is used for separating molecular components in a mixture based on their boiling point. The samples were injected in the GC at 35°C then heated to 200°C at 5"CImin. The temperature was maintained for 10 min at the end of the temperature profile. The chromatographed compounds were identified through a MS data library search. RESULTS AND DISCUSSION The +150"C liquids resembled light molasses, similar to a vacuum tower gas oil from a petroleum refinery. The -150°C materials were brown liquids. All samples had an odour characteristic of burnt organic matter. The lower boiling products had the strongest odour. The -150°C oils and the 15O0C-35O0C fractions were characterized by IR, 'H & I 3 C NMR and GCIMS. Table 1 summarizes the fractionation results for the PPSDO and autofluff samples. For comparison, literature data for a tire oil are also included. With an initial boiling point (IBP) of 155"C, the fractionation of the + 150°C autofluff sample yielded 90 wt % of distillate composed of 37 wt % in the middle distillate range and 53 wt % in the heavy gas oil range. This range is similar to that of a typical petroleum sample. The PPSDO + 150°C sample had an IBP of 102°C and 76 wt % distillate of which 49 wt % was in the middle distillate range and 27 wt % in the heavy gas oil range. Fractionation of the autofluff sample produced more distillate of a heavier nature than the PPSDO. When compared to tire oil, these oils had higher IBPs because they were condensed at a set temperature whereas the tire oil was not. Therefore we cannot compare the distillate yields further. During fractionation of the PPSDO +15O"C sample, the maximum distillate temperature was 501°C due to the limitation of the pot temperature (4OOOC) from the automatic apparatus used for the distillation. The material loss was 6.0 wt % due to distillate that was trapped in the column and in the condenser. The trapped distillate was so waxy that heating the condenser to a maximum temperature of 80°C in order to recover some distillate was unsuccessful. The samples were also analyzed using a series of tests commonly used for fuel analysis. Table 2 shows the results as well as literature data for sludge derived oil (SDO), tire oil, No2 diesel fuel and No.6 fuel oil. The heat of combustion values for the sludge derived materials namely, PPSDO and SDO, are significantly lower, and their densities at 15°C are higher than the corresponding values for the autofluff oil, tire oil and the two fuels. Table 2 also gives elemental analysis of the samples. Clearly, PPSDO and SDO produced oils having a lower carbon content. However, their HIC ratios are still comparable to the other data mainly due to their lower hydrogen content. Heteroatom levels, particularly N and 0, are very high when compared with the other oils. Pour points of these pyrolysis oils fall in the fuel oil range and are much higher than diesel oils. Figure 1 shows the proton-decoupled I3C NMR spectra of two autofluff and two PPSDO oil fractions. The spectrum of the -150°C autofluff fraction exhibited a signal at 45 ppm associated with the -CH,-O- group, an assignment confirmed by I3C DEFT NMR (not shown). This 869 functional group is absent in the -150°C PPDSO fraction. The spectrum of the -150°C PPSDO fiaction shows a signal at 182 ppm assigned to COOH carbons. No carbonyl signal was observed in the C=O region of the -150°C autofluff fraction spectrum. A comparison of the NMR spectra of the high boiling point fractions in Fig. 1 revealed significant differences. The spectrum of the PPSDO fraction shows many signals in the 170- 180 ppm region due to various -COO(R,H) carbons. No signal was observed in the -COO- region of the autofluff fraction spectrum. Signals in the 150-160 ppm range on both spectra are due to the oxygen-bonded aromatic carbon in phenols. Signals in the 110-115 ppm region on the autofluff spectrum were assigned to terminal =CH,'s in olefinic structures by I3C DEW NMR. Figure 2 shows the infrared spectra of selected autofluff and PPSDO fractions. The PPSDO spectra display absorptions in the 3200-3600 cm-' region which are more intense than in the autofluff fractions. This intensity also is accompanied by stronger and more complex carbonyl vibration band patterns in the PPSDO's than in the autofluff spectra. This indicates and confirms the presence of carboxylic acids suggested by NMR. Another difference between these two types of oils can be observed in the autofluff fraction spectra which display weak but well resolved olefinic and aromatic =C-H stretching and bending mode bands. This suggests the presence of significantly higher amounts of aromatic and olefinic compounds in the autofluff oil fractions. Table 3 lists GUMS derived compound type distributions in the low and high temperature PPSDO and autofluff oil fractions. Table 3 indicates that: 1) PPSDO fractions contained carboxylic acids; 2) autofluff fractions are more aromatic and olefinic than PPSDO fractions; 3) low temperature fractions are more aromatic than high temperature fractions; 4) high temperature fractions contain more alcohols than low temperature fractions; 5) high temperature PPSDO fraction contains a significant amount of nitriles. CONCLUSIONS Our study has shown that pyrolysis oils and their derived fractions are very complex mixtures of compounds including significant proportions of aromatics and olefins as well as nitriles, alcohols and ketones. In addition, carboxylic acids were found in PPSDO. Pyrolysis oil's heat of combustion and density values fall within the normal fuel oil range. The high content of olefins and aromatics of these oils and their high HIC ratios would suggest possibilities as feedstocks for low cost, large volume surfactant utilization. The surfactants could be produced by sulphonation or sulphation reactions. A large-scale use would be for enhanced oil recovery for both conventional and heavy oil/bituminous sands and for cleaning heavy bunker oil pipelines. Also for PPSDO, polymerization in asphalt could be performed to improve asphalt cement quality for adhesion to aggregates due to its high nitrogen content. The low sulphur content and the high heating value of the autofluff sample suggest it could be utilized as a liquid fuel, possibly by blending it with fuels of petroleum origin in order to lower the sulphur level. While the autofluff oil cannot be considered as a diesel fuel, it could be considered as a blending agent for use with a No.6 fuel oil. The autofluff characteristics resemble those of the No.6 fuel oil more than those of diesel. ' ACKNOWLEDGEMENTS The authors wish to thank WTC for its financial and technical support and the assistance in interpreting the results of this work. The authors acknowledge the technical assistance and contributions of the staff of the Fuel Quality Assessment Section and Dr. Heather Dettman and Mr. Gary Smiley for providing the NMR and GUMS spectra, respectively. Federal support of this work was provided through the Federal Program on Energy Research and Development (PERD). REFERENCES I. Crawford, G.V., Black, S., Miyamoto, H. and Liver, S., "Equipment selection and disposal of biological sludges from pulp and paper operations", PUID & Paoer Canada 94:4:37-39, April 1993. Springer, A.M. "Bioprocessing of pulp and paper mill effluents - past, present and future", Paoeri ia DUU DaDer and timber 75:3:156-161, 1993. Gamer, J.W. "Activated sludge treatment gains popularity for improving effluent, Salib, P. "Evaluation of circulating fluidized bed combustion of pulp and paper mill sludges", Contract mort Bioenerm DeveloDment Proeram DSS Contract No.: 23216-8-9056/01SZ. RPC, Fredericton, NB, Canada, October 1991. Day, M. "Auto Shredder Residue - Characterization of a solid waste problem", NRCC Soecial Reoort EC123992.3, NRC Institute for Environmental Chemistry, 1992. 2. 3. 4. 64:2:158-163, February 1990. 5. 870 a 6. Voyer, R. "Etude technico-iconomique et environnementale des pmCdis de traitement de risidus des carcasses d'automobiles", CRIO Report VPOIT-91-098, Centre de recherche industrielle du Quibec, 1992. Pritchard, T. "The over 75% solution", Globe and Mail Reoort on Business, December 1992 issue. Braslaw, J., Melotik, D.J., Gealer, R.L. and Wingfield, Jr, R.C. "Hydrocarbon generation during the inert gas pyrolysis of automobile shredder waste", Thermochimica Acta 186: 1:l- 18, 1991. Martinoli, D.A. "Converting sewage sludge into liquid hydrocarbon - the OFS process", Hydrocarbon Residues and Waste: Conversion and Utilization Seminar, Edmonton, Alberta, Canada, September 4-5, 1991. 10. Campbell, H.W. and Bridle, T.R. "Sludge management by thermal conversion to fuels", Proc. New Directions and Research in Waste Treatment and Residuals Management, Vancouver, 1985. 11. Campbell, H.W. and Martinoli, D.A. "A status report on Environment Canada's oil from sludge technology", hoc. Status of Municipal Sludge Management for the 1990% WPCF Specialty Conference, New Orleans, 1990. 12. Mirrniran, S., Ph.D. Thesis, Department of Chemical Engineering, UniversitC Laval, Quebec, Canada, 1994. 13. Kriz J.F. et al. "Characterization of sludge derived oil and evaluation of utilization options", Division Report ERL 90-50 (CF), November 1990. 14. Kirk-Othmer Encyclopedia of Chemical Technology, 3rd Edition, vol. 11, John Wiley & 15. Personal communication, September 1993. 7. 8. 9. sons, p. 357. Table 1 - D-I160 fractionation results of pyrolysis oil samples. Fraction PPSDO A u t o f l u f f T i r e o i l ' IBP ("C) 102 155 37 1BP-350"C (wt X ) 49 37 56 350°C-5250Cz ( w t X ) 27 53 37 Total (IBP-525"C) (wt %) 76 90 93 Residue (+525"C) (wt X) 18 8 7 Loss ( w t X ) 63 2 nla' Water (wt I) trace 0 n/a Table 2 - Physical and chemical analysis data of pyrolysis oil samples and petroleum fuels. Analysi S PPSDO Autofluff SLN5 Tire o i l ' No.2 No.6 Calorific value (MJIk9) 32.1 41 8 35.3 43 44.9 42.3 (1000 btu/lb) 13.8 18.0 15.2 18.5 19.3 18.2 Density @ 15°C (k91m3) 1073.2 931.1 24' 0 -6 -40 0 ' Data from reference [12] * End-point for distillation of PPSDO: 501°C and for tire oil: 469°C Large loss due to distillate trapped in column and condenser nla: not available Data from reference [I31 Data from references [14-15] as > 2 4 T ' Film formed on top preventing pouring of sample. Thus the pour point was reported 811 Y Table 3 - Compound type distribution in selected pyrolysis oil fractions by GUMS. Compound type -150°C -150°C 102-350°C 150-350°C PPSDO AUTOFLUFF PPSDO AUTOFLUFF Alcohols 8.8 10.8 14.7 12.6 A1 dehydes 3.7 0.5 1.3 Amines 2.4 3.2 0.9 Amides 3.1 2.7 1.1 Aromatics 10.2 48.7 5.2 18.2 Carboxylic acids 27.8 6.0 Epoxides 0.6 Esters 0.5 0.2 ' 0.4 Ethers 0.8 1.2 0.5 Heterocyclics 5.7 7.1 2.3 Ketones 2.8 6.8 4.4 5.1 Nitro 0.1 Olefins 11.7 14.9 6.6 24.0 Paraffins 14.6 9.7 8.9 29.5 Total 89.5 100.1 78.2 98.7 N i t r i l e s 1.6 5.0 17.4 2.2 , , I &-I5OoC(A) Fig. 1 - "C NMR regions (0-60 and 100-200 ppm) of selected autofluff (A) & PPSDO (p) fractions. Y 4000 3500 3000 2500 2000 1500 1000 5001 Wovenumbers (cm- 1 ) Fig. 2 - IR spectra of selected autofluff (A) & PPSDO (p) fractions a n CATALYTIC PYROLYSIS OF AUTOMOBILE SHREDDER RESIDUE / I Gregory G. Arzoumanidis. Michael J. McIntosh, Eric J. Steffensen. Matthew J. McKee, and Timothy Donahugh Energy Systems Division, Building 362 Argonne National Laboratory Argonne, Illinois 60439 Keywords: Catalytic Pyrolysis, Plastics, Automobile Recycling MTRODUCTION In the United States, approximately 10 million automobiles are scrapped and shredded each year. The mixture of plastics and other materials remaining after recovery of the metals is known as Automobile Shredder Residue (ASR). In 1994, about 3.5 million tons of ASR was produced and disposed of in landfills. However, environmental. legislative, and economic considerations are forcing the industry to search for recycling or other alternatives to disposal (1.2). Numerous studies have been done relating the ASR disposal problem to possible recycling treatments such as pyrolysis, gasification, co-liquefaction of ASR with coal, chemical recovery of plastics from ASR (3). catalytic pyrolysis (4). reclamation in molten salts (5). and vacuum pyrolysis (6). These and other possibilities have been studied intensively, and entire symposia have been devoted to the problem (3). Product mix, yields. toxicology issues, and projected economics of conceptual plant designs based on experimental results are. among the key elements of past studies. Because the kinds of recycling methods that may be developed, along with their ultimate economic value, depend on a very large number of variables, these studies have been open-ended. It is hoped that it may be useful to explore some of thesd previously studied areas from fresh perspectives. One such approach, currently under development at Argonne National Laboratory, is the catalytic pyrolysis of ASR. EXPERIMENTAL METHOD To eliminate variability due to nonuniform sampling, testing was begun using a "synthetic ASR" made up from pure materials on the basis of a best estimate of the inert-free ASR composition (Table 1). For the catalytic studies, pyrolysis occurs in a ceramic tube reactor inserted into the 30-cm heating zone of an elecmc furnace. The ASR is positioned in the reactor by means of an inconnel sample holder. Tiiidtenipenture profile is controlled by microprocessor. The majority of experinients are conducted using the profile shown in Figure 1. Faster rdtes of pyrolysis correspond to the profile in Figure 2. Liquids are collected in a series of three condensers, the first water-cooled, the others cooled by glycol at -20°C. Liquids are analyzed by GC/MS. Product gas samples are collected on-line in steel containers and analyzed by comparison with GC standards, using a 25-m X 0.53- mm fused silica column coated by Poraplot U, available from Chrompack. The gas mixtures are further analyzed by FTIR. For kinetic studies, a smaller tube and a furnace with a 15-cm heating zone are used. The ASR sample is usually 10 gm. 'Ihe reactor weight is continuously monitored by a sensitive transducer. The weight increase of a condenser is monitored by a second transducer. The time/iemperature profile inside the reactor can be more accurately controlled in this smaller system. The profile, giving the results presented below, was ramped from ambient to 700°C in 60 min and maintained at 700°C for 100 min. With this equipment, a continuous time/mass profile is obtained of the reactor residue, liquids condensed, and gases produced. RESULTS Catalydc pyrolysis of synthetic ASR was conducted in the presence of several oxides, monrmorillonite, and ASR char (Table 2). The amount of gases produced (determined by difference) varied within a rather narrow range. The amount of CO, in the sample was determined by using an ascarite-filled trap. The total amount of CO, from uncatalyzed reactions is usually 6- 9% of the synthetic ASR weight. In one case, 18% CO, was obtained, but the pyrolysis was run in the presence of Fe,O,, and oxidation of carbon by the metal oxide is likely. Three gas samples were obtained at three different temperatures during each experiment. The first sample was collected at 400-440"C pyrolysis temperature, the second at 500"C, and the third at 650°C. GC analyses of the first samples gave the distribution of gases shown in Table 3. The relative amounts of each separate gaseous hydrocarbon in the table vary widely, showing the large effect of catalyst on gas yields. Despite this variability, certain trends are clear. Most interesting is the large amount of CH$I formed in the fust sample, in one case as high as 90% of the gases. Because a preliminary survey of the pyrolysis literature yielded no reports of this phenomenon. we proceeded with caution despite positive GC and FTIR identification of CH,CI. Accordingly, experiments were conducted on synthetic ASR, with its only chlorine-containing 873 material, PVC, removed. These tests produced essentially no CH,CI. In these experiments, CH, in the frst gas sample was at its highest level (23%). It was also found that addition of NaCO, to the pyrolysis reaction reduces CH,CI formation and increases the level of CH, in the fmt stage. Given these results, the information in Table 3 offers several clues concerning the mechanism and kinetics of ASR pyrolysis. These are discussed below. Interestingly, the relative amounts of the pyrolysis gases from the second (500°C) and third (650°C) samples do not change appreciably, as indicated in Table 4. Changes in the relative amounts of the gases during the duration of the pyrolysis are estimated in Figure 3. This illustrates the large effect of the pyrolysis timd temperature profile on gas product distribution. Several investigators have noticed different product dish-ibutions under a variety of experimental conditions (2a). However, a direct effect of the profile is now clearly seen in the figure. This result raises the possibility that the distribution and composition of gaseous and liquid products can be manipulated by variation of the time/temperature profile. This possibility, which could be of economic importance, is now under investigation. As shown in Table 2, the residual solids (char) yield varied with catalyst type but remained in the range of 23-33% of the initial ASR weight. Lower levels of char yield translate into higher amounts of liquids, possibly an economically desirable effect. A number of binary oxides containing ZnO as one of the catalyst components appear to reduce the formation of char. Other additives (magnesium titanate, zirconate. Fe,O,, montmorilonite) do not yield results very different from the control run without catalyst. The effects of CuO and TiO,.SiO, also are marginal. In evaluating a commercial ASR recycling process, low cost is of foremost importance. Expensive catalysts, such as the binary oxides of Table 2, are not likely to prove directly useful. The incentive for studying these more expensive materials is to gain an understanding of possible effects, which may aid in the development of cheaper catalysts. Two classes of liquids are formed by the pyrolysis of ASR. The organic class contains over 50 organic compounds, as analyzed by GC/MS, and the aqueous class contains water and water-soluble oxygenates, primarily acids and alcohols. Water may be physically present in the ASR or may be. produced chemically by primary or secondary pyrolysis reactions (see discussion). Chlorine-containing compounds could not be detected by GC/MS in either liquid class. Production of chlorine-free liquid is a desirable feature of a commercial ASR reclamation process if the liquid is to be used as fuel. In this case, it is also desirable to increase the yield of the organic class of liquid. The effects of various catalysts and pyrolysis conditions on maximizing organic liquids are currently under investigation. DISCUSSION The pyrolysis of each separate ASR component has ken extensively studied by numerous investigators, and mechanisms have been proposed (7). These mechanisms are polymer-structure-dependent and may differ within the same class of polymers. For example, thermal degradation of polyurethanes may occur by three different types of hydrogen transfer: NH, a-CH, and p-CH, depending on the exact monomeric and polymeric structure of the pyrolyzing material (8). It is generally recognized that pyrolysis in a highly reducing environment proceeds via a radical mechanism. Detailed discussion of these mechanisms is beyond the scope of this paper. The mechanism of ASR pyrolysis is very complex. Single products (e& methane and other aliphatic hydrocarbons) may be formed via different mechanisms, depending on the rype and structure. of the polymer of origin. Therefore, it is difficult to present a unified mechanism by observation of pyrolysis products alone. An important consideration, however, is the range of different temperatures at which degradation begins for each type of polymer. Polyurethanes may start degrading just above 200°C (7). Removal of HCI from PVC takes place at a relatively low temperature, and it is completed almost before the degradation of the hydrocarbon backbone begins (9). Similar observations may be made for other reactions, such as decarboxylation of nylon, polyesters, polyurethanes and acrylics, formation of chemical water from wood and paper, etc. Thus, it is reasonable to assume that each polymer begins the pyrolysis process individually, based on its own structural and thermodynamic character. One of the key roles of catalysts is to lower the decomposition temperature by lowering the activation energy for some reactions. A single catalyst will not cause the same decrease of activation energy for all reactions of all ASR components. It is likely the main reason for the wide variation of the reIative amounts of products in the fmt gas sample (Table 3) is he variability of catalyst effects as the temperature reaches a level where the most facile reactions are fairly rapid. However, most of the ASR polymers, after losing such weak-link components as CO HCI, and bo. revert to mostly hydrocarbon backbones, which likely are very similar. This progably is the reason for the limited variations in the product distribution of the second and third gas samples (Table 4). From the above discussion, at least two distinct stages in the pyrolysis of ASR are hypothesized. The fust stage ends at about 300°C. and the second. in our case, continues up to \ a74 I 700°C. Preliminary experiments suggest that most of the CO,, CO. H,O, CH,CI, and possibly some nitmgen-containing components are being released in the first stage. 'Significantly, these materials carry most of the hetero-atoms that may interfere with the overall quality of useful ASR PyrOlYSiS products. The formation of CH,CI in relatively high concenmtions during the fust stage offers an indirect view of the reactions occurring during synthetic ASR pyrolysis. It is postulated that H a released from PVC attacks N-containing polymers, such as polyurethane, to form quaternary cationic nitrogen species; this is followed by scission of the polymer chain t h u g h CO, elimination, with subsequent formation of an olefinic end-group and an amine, as described earlier (8). The amine, most likely containing an N-CH, moiety (S), is further a m k d by HCI to fom a second, low-MW quaternary salt that decomposes to yield CH,CI. To test the two-stage pyrolysis hypothesis, separate first- and second-stage pyrolysis experiments are. under way. The products from each stage are recovered, and the residue from the first pyrolysis stage is used as starting material for the second. Preliminary results indicate that the fmt residue is about 75% by weight of the synthetic ASR charged. The first-stage liquids azz mostly of the aqueous class: about SO-%% of the total CO, is released, and more than 90% of the CH,CI is released. Only very low levels of gaseous hydrocarbons form during the fust stage. Most of the oxygen is IWKIV~~ in the fust stage in the form of CO,, CO, and H,O, so seconday reactions of these inorganics with the residue are. minimized (10). Therefore, second-stage pyrolysis yields primarily organic liquids and a gas rich in olefinic and paraffinic hydrocarbons. A conceptual, two-stage ASR pyrolysis process that segregates the products from the two stages is envisioned It could produce commodity methyl chloride in the first stage and valuable feedstock chemicals in the second stage. The potential for producing products from ASR pyrolysis more valuable than liquid fuel may thus be possible. The above possibilities suggest a need to determine ASR pyrolysis kinetics experimentally. The smaller reactor system described in the experimental section was developed and operated for this purpose. In a universal reaction scheme that seems to fit the kinetic data, six universal reactions and five universal reactants and products in ASR pyrolysis are assumed: fresh ASR 0, gaseous products (G),u nvaporized liquids (L), solid residue (S) , and condensed liquids (C). The sum of F, L, and S (denoted as R) is retained in the reactor and monitored by the fust transducer, Cis monitored by the second uansducer, and the weight of gases is obtained by difference. The results of an early ASR pyrolysis run are graphically presented in Figure 4. These results are correlated by the following simple scheme of universal reactions: G +--de f 4 F r- 1.25 A), while the latter has a high efficiency for collecting hard X rays ( A < 1.25 A). Wavelength dispersive x-ray fluorescence spectroscopy was used to measure the abundances of the inorganic species in the tire before processing and to analyze the removal efficiency of the different liquids for these inorganics. RESULTS AND DISCUSSION The resulting TDP's had surface areas in 0.1-2.0 mm2 range. ANALYSIS OF THE TIRE CHUNKS. The wavelength dispersive XRF spectrum (using Cr radiation and the gas proportional counter) of an untreated tire chunk is shown in Figure 1. The spectrum contains peaks due to zinc (AK, = 1.436 A) calcium (AK, = 3.359 A), and sulfur (AK, = 5.373 A). The chromium peak (A = 2.290 A) in the WDXRF spectrum is due to the.use of chromium as the exciting radiation (tube) for the experiments. Chromium produces "soft" X rays which do not penetrate deeply into the rubbery portion of the scrap tire. Consequently, the x-ray peak due to iron (AK, = 1.937 A) is barely discernible in the spectrum. Shown in Figure 2 is the WDXRF spectrum of the same tire chunk using a scintillation counter rather than a gas proportional counter for x-ray detection and a molybdenum (AK, = 0.711 A) radiation source. A small peak due to bromine (AK, = 1.041 A) is clearly discernible, along with the Mo peak. There is, of course, no chromium peak in this spectrum. When the steel belts were removed from the rubbery section of the tire, ground to a powder, and then submitted to our XRF analysis, the peaks due to iron, copper and zinc are clearly discernible. TREATMENT WITH THE PROCESS LIQUIDS. The four liquids had different effects on the tire chunks. The n-methyl pyrrolidinone is absorbed into the tire chunk, causing the rubbery portion of the tire to swell. NMP cleaves the adhesion between the rubbery portion of the tire and the steel belts, while the'tire chunk becomes very brittle and easily grindable. It proved difficult to recover the NMP from the tire chunks. Concentrated nitric acid degrades the tire chunk into particles& dissolves the steel belts. The WDXRF spectrum of the resulting TDP is shown in Figure 3. Comparison of the WDXRF spectra of the untreated tire to that of the TDP indicates that the zinc, calcium, and sulfur abundances have been drastically reduced in the TDP by the nitric acid treatment at ambient conditions. Lengthening the time of treatment results in complete removal of the unwanted inorganics. Figure 4 shows the WDXRF spectrum of the residue produced by evaporating the nitric acid filtrate to dryness. The characteristic peaks due to the metal species are provided in this spectrum, verifying the absence of the unwanted inorganics in the TDP. The large iron peak is due to the fact that the nitric acid dissolves the steel belts. Comparison of the intensities of the zinc peaks in Figures 1,3, and 4 indicates that the mass balance for zinc in these three samples is not well established and/or that the enhancement/absorption effects for the Zn peaks cannot be ignored in these samples. Treatment with 50% hydrogen peroxide at ambient conditions does not degrade the tire chunks nearly as rapidly as does the nitric acid treatment. This treatment also extracts the inorganic% which are subsequently found in the residue evaporated from the filtrate. This method also attacks the steel belts, as evidenced by the large iron peaks in the WDXRF of the residue from the evaporate. 880 1 ' Concentrated sulfuric acid degrades the tire chunks almost as well as the nitric acid at ambient conditions but does not dissolve the steel belts. The steel belts may, then be removed easily (and essentially in tact) from the rubbery portion of the tire and collected on the stir bar. Comparison of the WDXRF intensities indicates that the sulfuric acid did not remove the zinc, calcium, or sulfur at ambient conditions. A summary of current results is presented in Table I. Additional results will be discussed. CONCLUSIONS Tires chunks may be treated at ambient conditions with different liquids, producing different effects. The unwanted inorganics can be extracted from the tire chunks, leaving a TDP with a high carbon and hydrogen content and a greatly reduced surface area. Wavelength dispersive x-ray fluorescence spectroscopy may be used to monitor the reduction in abundances of each of the unwanted inorganics. Altering the conditions of the WDXRF experiment provides different information about the distribution of inorganics in the tire chunk and the TDP's. REFERENCES 1. 2. 3. 4. 5. 6. Ray US. Environmental Protection Agency. Scrap Tire Handbook Effective Management Alternatives to Scrap Tire Disposal in Illinois, Indiana, Michigan, Minnesota, Ohio, and Wisconsin. 1994. Transportation Research. Uses of Recycled Rubber Tires in Highways. 1994. Liu, Z., Zondlo, J.W., and Dadyburjor, D.B., Energy and Fuels, 1994 (8) 607. Sopek, D.J. and Justice, A.L. in "Clean Energy from Waste & Coal", M.R. Khan, Am. Chem. SOC., 1991. Jenkins, R., "X-Ray Spectrometry", John Wiley & Sons, NY, 1988. Rousseau, R. M., "A Practical XRF Calibration Procedure", 43"' Annual Denver XConference, Steamboat Springs, CO. 1994. I DOE-EPSCoR Graduate Fellow. TABLE I. EFFECTS OF LIQUIDS OF SCRAP TIRE PARAMETERS. LIQUID INORGANIC FATE OF CONDITION USED REMOVAL STEEL BELT OF LIQUID Zn Ca S ADHESIONS NMP no effect cleaved absorbed into rubber nitric acid E E E dissolved recyclable sulfuric acid no effect cleaved recyclable hydrogen peroxide E E E dissolved recyclable / I E Extracted by the liquid. 881 I 150000 - 10 20 30 40 50 60 70 80 90 100 110 BRBPHITE MONOCHROMATOR ANBLE FIGURE 1. WDFF SPECTRUM OF UNTREATED TIRE CHUNKS; Cr/GPC. 150000 - m OFT X-RAY CUT-OFF 0 1 1 1 1 1 1 1 l 10 20 30 40 50 60 70 80 90 100 110 ERAPHITE MONOCHROMATOR ANELE FIGURE 2. WDXRF SPECTRUM OF UNTREATED TIRE CHUNK; Mo/SC. 150000 - 100000 - I 10 20 30 40 50 60 70 80 90 100 110 ERAPHITE HONOCHROMATOR ANELE FIGURE 3. WDXRF SPECTRUM OF TDP FROM THE NITRIC ACID TREATMENT, CrIGPC. 882 150000 - 0 50000 - 10 20 30 40 50 60 70 80 90 100 110 BRAPHITE MONOCHROMATOR ANBLE FIGURE 4. WDXRF SPECTRUM OF RESIDUE FROM NITRIC ACID TREATMENT; Cr/GPC. 250000 - 200000 - 150000 - .n 10 20 30 40 50 60 70 80 90 100 110 BRAPHITE MONOCHROMATOR ANBLE FIGURE 5. WDXRF SPECTRUM OF THE FILTRATE RESIDUE FROM THE 50% HYDROGEN PEROXIDE TREATMENT, Cr/GPC. f 883 THE GROWING NEED FOR RISK ANALYSIS C.C. Lee and G.L. Huffman Risk Reduction Engineering Laboratory U.S. Environmental Protection Agency Cincinnati, OH 45268 Keywords: Risk analysis, Risk assessment, Risk management ABSTRACT Risk analysis has been increasingly receiving a t t e n t i o n i n making environmental decisions. For example, i n i t s May 18, 1993 Combustion Strategy announcement, EPA required t h a t any issuance o f a new hazardous waste combustion permit be preceded by the performance o f a complete ( d i r e c t and i n d i r e c t ) r i s k assessment. This new requirement i s a major challenge t o many 'engineers who are involved i n waste i n c i n e r a t i o n a c t i v i t i e s . This Paper presents the h i g h l i g h t s o f what i s required f o r ' a r i s k analysis from a p r a c t i c a l engineering point o f view. It provides t h e regulatory basis f o r it to provide the r a t i o n a l e as t o why r i s k analysis i s needed. INTRODUCTION "Nothing would be done at a l l i f a man waited till he could do it so well t h a t no one could f i n d f a u l t with i t " - - C a r d i n a l Jewman. Cardina; Newman's statement i s very pertinent t o the subject o f Risk Analysis. The assessment o f environmental r i s k s posed t o human health i s an i n c r e d i b l y complex undertaking. Because of t h i s complexity, it i s very d i f f i c u l t t o do much more than i d e n t i f y sources and effects o f p o t e n t i a l concern and, i n a rough manner, t o quantify transport along major pathways (Martin-86). I n addition, r i s k analysis has been b a s i c a l l y developed by s c i e n t i s t s . This makes it more d i f f i c u l t f o r zngineers t o apply the r i s k models developed by these s c i e n t i s t s t o t h e i r real-world'' waste treatment problems, because o f the d i f f e r e n t d i s c i p l i n e s and terminologies involved. B a s i c a l l y , t h e authors used the documents contained i n the "References Section" t o derive the information f o r t h i s Paper. The objective o f t h i s Paper i s t o summarize the h i g h l i g h t s o f what i s required f o r a r i s k analysis from a p r a c t i c a l engineering point o f view. It emphasizes the documentation o f r i s k analysis requirements from various environmental statutes. The purpose i s t o establish the regulatory basis r e l a t i v e t o why r i s k analysis i s needed, and how r i s k analysis should be conducted. It i s believed t h a t the understanding of the s t a t u t o r y provisions i s important and that the only way t o formulate the proper r i s k analysis approach i s t o comply with the regulatory requirements under t h e s p e c i f i c environmental laws t h a t apply. For example, i n the past, r i s k assessments were not required f o r obtaining a hazardous waste i n c i n e r a t i o n permit. However, i n i t s May 18, 1993 Combustion Strategy announcement, EPA required t h a t any issuance o f a new hazardous waste combustion permit be preceded by a complete d i r e c t and i n d i r e c t r i s k assessment (EPA-93/5). This new requirement i s a major challenge t o many engineers who are involved i n waste i n c i n e r a t i o n a c t i v i t i e s . REGULATORY BASIS FOR RISK ASSESSMENT Risk i s the p r o b a b i l i t y o f i n j u r y , disease, or death under s p e c i f i c circumstances (Lee-92/6). Risk assessment i s a cornerstone o f environmental decision-making. EPA defines r i s k assessment as: (1) the determination o f the kind and degree of hazard posed by an agent (such as a harmful substance); ( 2 ) t h e extent t o which a p a r t i c u l a r group o f people has been or may be exposed t o the agent; and (3) the present or p o t e n t i a l health r i s k t h a t e x i s t s due t o the agent (Lee-92/6). Risk assessment i s a complex process by which s c i e n t i s t s determine the harm t h a t an i n d i v i d u a l substance can i n f l i c t on human health or the environment. For human h e a l t h r i s k assessment, the process takes place i n a series o f four major steps as follows (EPA-90/6; NAC-83): (1) Hazard i d e n t i f i c a t i o n : In i d e n t i f y i n g hazards, two kinds o f data are gathered and evaluated: (A) data on the types o f health i n j u r y or disease that may be produced by a chemical; and (B) data on the conditions of exposure under which i n j u r y or disease i s produced. The behavior o f a chemical w i t h i n the body and the i n t e r a c t i o n s it undergoes with organs, c e l l s , or even parts of c e l l s may also be characterized. Such data may be of value i n answering t h e u l t i m a t e question o f whether the forms of t o x i c i t y known t o be produced by a substance i n one population group or i n experimental settings are also l i k e l y t o be produced i n humans. (2) Dose-response assessment: The next step i n r i s k assessment describes the r e l a t i o n s h i p between the amount of exposure t o a substance and the extent of t o x i c i n j u r y or disease. Even where good epidemiological studies have been conducted, r e l i a b l e q u a n t i t a t i v e data on exposure i n humans are r a r e l y available. Thus, in most cases, dose-response r e l a t i o n s h i p s must 884 f r' f be estimated from studies i n animals, which immediately raises three Serious problems: (A) animals are usually exposed at high doses, and effects at low doses must be predicted by using theories about the form Of the dose-response relationship; (B) animals and humans o f t e n d i f f e r i n s u s c e p t i b i l i t y (if only because o f differences i n size and metabolism); and (C) the hu an population i s heterogeneous, so some individuals are l i k e l y t o be (3) Human exposure assessment: Assessment o f human exposure requires estimation o f the number o f people exposed and the magnitude, duration, and timing of t h e i r exposure. The assessment could include past exposures, current exposures, or exposures a n t i c i p a t e d i n t h e f u t u r e . I n some cases, measuring human exposure d i r e c t l y , e i t h e r by measuring levels of the hazardous agents i n the ambient environment or by using personal monitors, i s f a i r l y straightforward. I n most cases, however, detailed knowledge i s required of t h e f a c t o r s t h a t control human exposure, including those factors that determine the behavior o f the agent a f t e r i t s release i n t o the environment. (4) Risk characterization: The f i n a l step i n r i s k assessment combines the information gained and analysis performed during t h e f i r s t three stepes to determine the l i k e l i h o o d that humans w i l l experience any o f the various forms of t o x i c i t y associated with a substance. The r i s k characterization then becomes one o f the factors considered i n deciding whether and how the substance w i l l be regulated. I n the 198Os, as health r i s k assessment became more widely used across U.S. €PA programs, the need f o r consensus and consistency i n the areas o f hazard i d e n t i f i c a t i o n and dose-response assessment became clear. I n 1986, €PA work groups were convened t o establish consensus positions on a chemical -by-chemical basis f o r those substances o f common i n t e r e s t and t o develop a system f o r communicating t h e positions t o €PA r i s k assessors and r i s k managers. This e f f o r t resulted i n the creation o f EPA's Integrated Risk Information System (IRIS) i n 1986. I n 1988, the I R I S was made available t o the public. I R I S c u r r e n t l y contains summaries o f EPA human health hazard information that support two o f the four steps--hazard i d e n t i f i c a t i o n and dose-response evaluation--of the r i s k assessment process. It c u r r e n t l y contains information on approximately 500 s p e c i f i c substances. Questions such as "what i s the potential human health hazard o f exposure t o benzene?" and "what are the possible cancer and/or non-cancer e f f e c t s ? " can f i n d answers from IRIS (EPA- 93/1). A key factor a f f e c t i n g t h e regulatory coverage o f a statute i s the d e f i n i t i o n of the substances subject t o regulation. The statutes use several des;riptive terms, n o t necessarily having the same meaning, t o i d e n t i f y harmful substances." These include p o l l u t a n t , t o x i c pol 1 utant, hazardous substance, contaminant, hazardous material, and hazardous waste. The Toxic Substances Control Act (TSCA), for example, defines "chemical substances" and "mixtures" subject t o regulation i f certain c r i t e r i a are met; the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA) and the Marine Protection, Research, and Sanctuaries ,pet (MPRSA) specify categories o f substances [FIFRA defining "pesticides, and MPRSA "materials"]. I n general, three aspects of r i s k are addressed i n each statute. They are: type o f harm, type of r i s k , and required considerations between the chemical and the harm that may r e s u l t (Martin-86). TvDe o f Harm: The type of harm i s usually e x p l i c i t l y described by terms t h a t d e f i n e the chemicals or the substances to be addressed (e.g., a hazardous substance t h a t may cause i n j u r y t o health or the environment). The harm components of a statute's r i s k d e f i n i t i o n generally consist of a d e s c r i p t i o n o f an undesired outcome (death, i n j u r y ) and/or a d e s c r i p t i o n of the population (public, w i l d l i f e ) a t r i s k or t h e objective o f t h e r e g u l a t i o n , such as protection o f the environment. TvDe o f Risk: Considering r i s k when developing f e d e r a l regulations encompasses the p r o b a b i l i t y of harm occurring. The p r o b a b i l i t y of harm presented by a chemical may be considered zero, i n s i g n i f i c a n t , or s i g n i f i c a n t . A term such as " s i g n i f i c a n t r i s k " w i l l then be addressed by the rule-making process. Reouired Considerations: The statutory language may guide the designation and s e t t i n g o f technical or control standards for e x p l i c i t l y specifying a basis f o r making regulatory decisions and also f o r i n d i c a t i n g what factors must be, may be, or may not be considered when developing regulations. The statutes discuss the amount of protection o r r i s k reductions t o be addressed through the issuance o f standards "necessary, 885 , 7o re suscept ible than the average. "adequate," or " s u f f i c i e n t " t o p r o t e c t h e a l t h or the environment by providing detailed guidance (e.g., ample margin o f safety) and/or by prescribing p a r t i a l factors (e.g., r i s k and cost) that must be considered. REGULATORY BASIS FOR RISK MANAGEMENT EPA i s responsible f o r implementing environmental statutes. Although, the s t a t u t e s g e n e r a l l y do not prescribe r i s k assessment methodologies, many environmental laws do provide very s p e c i f i c r i s k management d i r e c t i v e s , and these d i r e c t i v e s vary from statute t o statute. EPA defines r i s k management as the process o f evaluating a1 t e r n a t i v e r e g u l a t o r y and non-regulatory responses t o r i s k and selecting among them. The selection process necessarily requires the consideration o f l e g a l , economic and social factors (Lee-92/6). Statutory r i s k management mandates can be roughly c l a s s i f i e d i n t o three categories: (1) pure r i s k ; (2) technology-based standards; and (3) reasonableness o f r i s k balanced with benefits (EPA-93/1). (1) Pure-Risk Standards Pure-risk standards are, sometimes, termed z e r o - r i s k standards. This category allows an adequate margin o f safety, however, requires the protection o f public health without regard to technology or cost factors. For example, the National Ambient A i r Q u a l i t y Standards (NAAQS) o f the Clean A i r Act belong t o t h i s category. (2) Technology-Based Standards Technology-based environmental standards focus on the effectiveness and costs o f a l t e r n a t i v e control technologies rather than on how control actions could a f f e c t r i s k s . For example, i n d u s t r i a l water p o l l u t i o n standards, where the i n s t a l l a t i o n of a s i n g l e c o n t r o l system can reduce r i s k s from a variety o f d i f f e r e n t pollutants, belong t o t h i s category. Consider the several technology-based standards i n the Clean Water Act. The Act requires industries to i n s t a l l several levels of technology-based controls for reducing water p o l l u t i o n . These include best practicable control technology, best conventional technology, and best available technology economically achievable f o r e x i s t i n g sources. New sources are subject t o the best demonstrated control technology. Total costs, age o f equipment and f a c i l i t i e s , processes involved, engineering aspects, environmental factors other than water q u a l i t y , and energy requirements are to be taken i n t o account i n assessing technology-based controls. (3) No Unreasonable Risk This category c a l l s f o r the balancing o f r i s k s against benefits i n making r i s k management decisions. The following are two examples i n t h i s category: 0 The Federal Insecticide, Fungicide, and Rodenticide Act requires EPA t o r e g i s t e r (license) pesticides which, i n addition t o other requirements, it finds w i l l not cause unreasonable adverse effects on the environment. The phrase refers t o any unreasonable r i s k s to man or the environment taking i n t o account the economic, social, and environmental costs and benefits o f the use of any pesticide. Under the Toxic Substances Control Act, EPA i s mandated to take action i f it finds that a chemical substance presents or w i l l present an unreasonable r i s k o f i n j u r y t o health or the environment. This includes considering the effects of such a substance on health and the environment and the magnitude of the exposure of human beings and the environment t o such a substance; the benefits o f such a substance f o r i t s various uses and the a v a i l a b i l i t y of substitutes f o r such uses; and the reasonably ascertainable economic consequences o f the r u l e , a f t e r considerat i o n o f the effect on the national economy, small businesses, technological innovation, the environment, and public health. THE ROLE OF COMPARATIVE RISK ANALYSIS EPA's support f o r using comparative r i s k analysis to help set i t s regulatory p r i o r i t i e s has been no secret. Unlike r i s k assessment, which f o r years has provided r e g u l a t o r s t h e basis for deciding whether or not an individual substance needs t o be controlled, comparative r i s k analysis and i t s derivative, r e l a t i v e r i s k , have arrived on the scene only recently. Very simply described, comparative r i s k analysis i s a procedure f o r ranking environmental problems by t h e i r seriousness ( r e l a t i v e r i s k ) for the purpose of assigning them program P r i o r i t i e s . Typically, teams of experts put together a l i s t o f problems; then, 886 I they sort the problems by types of risk--cancer, non-cancer health, materials damage, ecological effects, and so on. The experts rank the problems w i t h i n each type by measuring them against such standards as t h e s e v e r i t y of effects, the l i k e l i h o o d of the problem occurring among those exposed, the number of People exposed, and the l i k e . The r e l a t i v e r i s k o f a problem i s then used as a factor i n determining what p r i o r i t y the problem should receive. Other factors include s t a t u t o r y mandates, public concern over the problem, and the economic and technological f e a s i b i l i t y o f c o n t r o l l i n g it. EPA'S Science Advisory Board urged the Agency t o order i t s p r i o r i t i e s on the basis of reducing the most serious r i s k s . The Board argued, i n part ... There are heavy costs involved if society f a i l s t o set environmental p r i o r i t i e s based on r i s k . If f i n i t e resources are expended on lower p r i o r i t y problems at the expense of higher p r i o r i t y r i s k s , then society w i l l face needlessly high r i s k s . If p r i o r i t i e s are established based on the greatest opportunities t o reduce r i s k , t o t a l r i s k w i l l be reduced i n a more e f f i c i e n t way, lessening threats t o both p u b l i c health and local and global ecosystems.. . . (EPA-93/I). THE ROLE OF RISK COMMUNICATION Basically, r i s k communication deals with the approaches t o communicate with the pub1 i c on various environmental issues. Some recommended communication checklist items are provided as follows (EPA-90/6): (1) Be prepared; (2) Review the facts; (3) Anticipate l i k e l y questions; and (4) Consider what the audience wants t o know. THE ROLE OF RISK UNCERTAINTY Uncertainty means the q u a l i t y or state o f having possible v a r i a t i o n s . I n r i s k analysis, uncertainty f a c t o r s include: (1) the v a r i a t i o n i n s e n s i t i v i t y among the members o f the human population ; (2) the uncertainty i n extrapolating animal data t o the case o f humans; (3) the uncertainty in e x t r a p o l a t i n g from data obtained i n a study that i s o f l e s s - t h a n - l i f e t i m e exposure; (4) the i n a b i l i t y o f any single study t o adequately address a l l possible adverse outcomes i n humans. SUMMARY By way o f summarizing, t h e f o l l o w i n g key questions concerning the abovedescribed " r i s k " terms might be asked: (1) Risk assessment: What do we know about r i s k ? or how r i s k y i s t h i s s i t u a t i o n ? (2) Risk management: What do we wish t o do about r i s k ? or what shall we do about i t ? (3) Comparative r i s k : What i s the ranking ( p r i o r i t y ) o f the various r i s k s ? (4) Risk communication: What and how should a r i s k assessor communicate with the public on r i s k analysis? (5) Risk uncertainty: What i s the q u a l i t y o r state o f having possible v a r i a t i o n s i n conducting r i s k analysis? REFERENCES (EPA-89/3), "Risk pessment Guidance f o r Superfund: Volume I 1 - Environmental Evaluation Manual, EPA540-1-89-001, March 1989. (EPA-90/1), "Methodology f o r Assessing Health,,Risks Associated w i t h I n d i r e c t Exposure t o Combustor Emissions, Interim Final, EPA600-6-90-003, January 1990. (EpA-90/6), "Risk Assessment, Management and Communication o f Drinking Water Contamination," EPA625-4-89-024, June 1990. ( ~ ~ ~ - 9 1 / 1 2")R, i sk Assessment Guidance f o r Superfund: Volume I - Human Heal th Evaluation Manual (Part B, Development o f Risk-based Preliminary Remdiation Goals), EPA540-R-92-003, Publication 9285.7-018, December 1991. (EpA-93/1), "The ABCs o f Risk Assessment," EPA Journal, EPA175-N-93-0014, Volume 19, Number 1, January/February/March 1993. (EPA-93/5), "Environmental News: EPA Administrator Browner Announces New Hazardous Waste Reduction and Combustion Strategy," May 18, 1993. (EpA-94/4), "Exposure Assessment Guidance for RCRA Hazardous Waste Combustion F a c i l i t i e s , " EPA530-R-94-021, April 1994. (Lee-92/6), "Environmental Engineering Dictionary," C.C. Lee, June 1992. 887 (Martin-86), "Hazardous Waste Management Engineering," Edward J. Martin and James H. Johnson, J r . ) , Van Nostrand Reinhold Company, New York, 1986. (Martin-86/2), "1n;inerator Risk Analysis Presentation to: Incinerator Permit Writer's Workshop, Edward J. Martin, Peer Consultants, Inc., February 4,5, and 6, 1986. (NAC-83), "Risk Assessment i n the Federal Government: Managing the Process," Committee on the I n s t i t u t i o n a l Means f o r Assessment o f Risks to Public Health, Commission on L i f e Sciences, National Research Council, Established by National Academy Council (NAC), National Academy Press, 1983. (NIOSH-84/10), "Personal Pr;tective Equipment f o r Hazardous Materials Incidents: A Selection Guide, U.S. Department o f Health and Human Services, Pub1 i c Health Service, Centers f o r Disease Control, National I n s t i t u t e f o r Occupational Safety and Health (NIOSH), Division o f Safety Research, Morgantown, West V i r g i n i a 26505, October 1984. (NRT-87/3), "Hazardous Materials Emergency: Planning Guide," National Release Team (NRT), NRT-1, March 1987. 888 i b EVALUATION OF A BIOMASS DERIVED OIL FOR USE AS ADDITIVE4N PAVING ASPHALT 1. Jr., I. Clelland, H. Sawatzky, Energy Research Laboratories, CANMET, Natural Resources Canada, 555 Booth St., Ottawa, Ont., Canada, KIA OGI. E-mail: jean.houde~x400.emr.ca Keywords: paving asphalt, additive, antistripping INTRODUCTION Treatment and disposal costs of sewage sludge can represent up to 50% of a municipality’s annual wastewater treatment budget. Sewage sludge (30% solids) accounts 5% of Canadian landfill by weight, and the ever increasing volume of sludge coupled with the decreasing options available for disposal creates a growing problem for major municipalities. Current disposal , options are agricultural application, incineration and landfill. Concern about heavy metal . migration and public pressure to find a local solution has severely curtailed the spreading of sludge on agricultural land. Incineration is the major option for larger centres but the relatively high cost fof incineration, ranging from $350 to $1000/t dry sludge, has caused a great deal of interest in methods of improving the cost effectiveness of incineration or in new equivalent technologies. The high cost and more stringent environmental regulations for incinerating municipal sludges have led to developing more efficient sludge management technologies that are not agricultural based. The Wastewater Technology Centre of Environment Canada has been developing one such technology since 1982. The thermoconversion process shown in Fig. 1 involves low temperature ,treatment of sludge to liquid and solid fuel products (1). A key technical feature of the sludge conversion is the formation of a byproduct oil (2) referred to as sludge derived oil (SDO). In 1989, Enersludge Inc., Wastewater Technology Centre and CANMET’s Energy Research Laboratories (ERL) ,of Natural Resources Canada undertook a joint R&D program to be conducted at ERL to investigate promising utilization options for the SDO. SDO is a black, viscous high-boiling ( > 150’C) organic liquid with a characteristic odour. Initial characterization tests led to investigating the use of SDO as a feedstock material for introduction into a refmery stream. Its average chemical structure is that of a large complex molecule, with a hydrocarbon skeleton and functional groups containing nitrogen (pyrroles, amides) and oxygen (esters). Such structures, which indicate protein origins, tend to be very polar. The relatively high concentration of polar groups in SDO, especially the abundance of nitrogenous groups, and its incompatibility with most distillate hydrocarbons except heavy aromatic gas oils, as discovered in more extensive characterization testing, indicated a more appropriate role as an asplialt additive. This paper describes the work done at ERL to develop SDO for antistripping applications. COMPATIBILITY The affinity of SDO for heavy petroleum derived materials was initially investigated by blending equal amounts of SDO in each of ROSE” (residual oil supercritical extraction) residue (3) and CANMET hydrocracking pitch (4). Both were observed to be completely miscible and formed stable viscous blends. The same was then observed with SDO and pentane-precipitated Athabasca bitumen asphaltenes. Asphalt cement (AX),a petroleum product made from the fraction that has a boiling point greater than approximately 350°C. contains many polar components including sulphur and nitrogen containing compounds. SDO was found to be compatible with A/C as indicated by the high ductility of SDO and commercial asphalt blends. Incompatibility in an asphalt blend causes a drastic decrease in ductility that is easily detected in comparison to commercial A/C. Ductility and rate of elongation until it breaks. The results of this preliminary investigation as well as the characterization indicated SDO was compatible with NC. Further, the SDO showed evidence of strong affinity for asphaltenes in asphalt. The high nitrogen content of SDO is desirable for improving of adhesion to aggregate. ROAD ASPHALTS l is determined by measuring the distance a futed shape of AIC will stretch at a fixed temperature Asphaltic concrete road pavement is made from a mix of aggregate (sand, gravel and crushed stone) held together by 5% to 10% on a weight basis of A/C. Government transportation agencies have developed road pavement specifications and are also the largest buyers of road pavement. Their specifications include the hardness of the A/C, its ductility, viscosity, flash point, resistance to stripping and performance after simulated road paving and handling evaluations. Table 1 lists the American Society for Testing and Materials (ASTM) test methods typically used to assess commercial AIC’s. Currently in Canada, A/C is primarily graded on its hardness as measured by penetration, reported as the measured penetration by a needle into a sample of A/C of specified temperature 889 for controlled time and force (weight) on the needle. kphalt of 85 dmm (tenths of millimetres) to 100 dmm, i t . , 85/100 penetration, is considered hard, whereas 150/200 penetration is soft. ANTISTRIPPING ADDITIVE The resistance to stripping of the A/C from the surface of the aggregate is an important specification for the performance of asphalt pavement and is monitored by the buyers of asphalt pavement. Since NC does not adhere well to certain aggregates, transportation agencies specify the use of antistripping agents when using these particular aggregates. Stripping involves complex processes which are still not fully understood. Several factors influence the sensitivity of asphalt concrete mix to stripping [5]. For this phenomenon to occur, free water must be present. Jamieson aggregate (Ontario, Canada) which is prone to stripping and listed by the Ministry of Transport of Ontario (MTO) as requiring at least 1 % antistripping agent in the A/C, was chosen to evaluate the antistripping performance of SDO in A/C blends. ASTM test method D-1664', the test method for evaluating coating and stripping of bitumen-aggregate mixtures, requires a 95% coverage in static immersion tests to meet specifications. To demonstrate the stripping resistance of SDO in A/C blends, three commercial NCs and various concentrations of SDO were used with Jamieson aggregate as shown in Fig. 2. The results are an average from the visual stripping evaluations of a panel of five evaluators. The coverage of the aggregate by the A/C increased from about 60% (for the first two A/Cs) to above 90% as SDO was added. The retained coverage of the third AIC, with no SDO, increased from about 40% to above 90% as in the above case. Further, the performance of an A/C with one of the commercial antistripping agents, Alkazine-0, recommended by the MTO is shown for comparison (80% retained coating at the 1 wt % level). Under these conditions, SDO has antistripping performance equal to or in excess of that given by at least one commercial agent, and can be used to have coverage in excess of MTO specifications. Sources of SDO other than the Atlanta SDO (undigested sewage sludge derived oil) were also evaluated for performance as antistripping agents. These samples were also obtained from bench-scale experiments using Highland Creek (Toronto, Canada) sewage sludge, an undigested sludge, and Hamilton sewage sludge, a digested sludge, supplied by the Wastewater Technology Centre. A digested sludge is one that has been subjected to anaerobic bacterial digestion. An SDO from digested sludge has a reduced nitrogen content. In Fig. 3, several concentrations of these two SDOs are compared with commercial antistripping agents: Alkazine-0, Redicote AP and Redicote 82s. Some of these commercial agents are as effective as the SDO at approximately half the concentration. Comparison of Highland Creek SDO and Hamilton SDO (undigested versus digested sludges) as antistripping agents shows a parallel, but slightly less effective performance curve for the digested SDO, indicating the significance of the nitrogen content of SDO as a factor in adhesion to aggregate. The use of antistripping additive may cause significant modifications to other performance properties of asphalt cement. The other properties of asphalt cement susceptible to modifications were also assessed. Candidate blends of asphalt cement with SDO and commercial antistripping agents were evaluated for asphalt cement performance specifications in Table 2. Blends of 2% Redicote 82s were compared with SDO blends as well as several other commercial antistripping agents. These results indicated the performance of SDO at 5% does modify the asphalt cement test results, in particular the loss of volatiles in the thin film oven test and the penetration. The changes to the penetration of the asphalt cement can be modified by a change in the distillation temperature of the SDO fraction or a change in the consistency of the NC. However, there may be a limit to the acceptability of a change in penetration caused by an additive unless the AIC is very hard. It should be noted that the use of only 2% SDO with the AIC3 in Fig. 2, was succeSSful for all specifications monitored. Further, the viscosity of the SDO containing asphalt cement easily met the MTO criteria. . In an effort to further define the antistripping performance curve, SDO was again tested by the stripping immersion method including the 3 wl % additive level. The results in Fig. 4 show that the 3 wt % additive level was as effective as the 5 wt % additive level in both 85/100 and 150/200 penetration A/C4. The 3 wt % additive level resulted in less change of the AIC's consistency, as shown in Table 3. In another phase of testing, a one year old SDO sample was compared to a freshly obtained sample. There was no significant difference in effectiveness as an antistripping agent between the SDO samples, indicating the stability of the product. 1 ~~~ ' Summary of ASTM D-1661 : The selected and prepared aggregate is coated with the bitumen at a specified temperature appropriate to the grade of bitumen used. The coated aggregate is immersed in distilled water for 16 to 18 h. At the end of the soaking period, and with the bitumen-aggregate mixture under water, the total area of the aggregate on which the bituminous film is retained is estimated visually as t 95%. 890 1 I MARSHALL TEST The next logical step was to test how SDO as an A/C additive compared to a commercial additive in the asphalt concrete. One of the standard testing methods commonly used is known as the Marshall test. The A/C and selected aggregate are mixed hot to produce an asphalt concrete which is compacted by a laboratory compactor, simulating the roller compaction the concrete would receive in the field. The cooled asphalt concrete sample is tested for strength and resistance to plastic flow with an applied lateral force. A standard HL3 mix design was used for manufacturing all specimens to be evaluated. The ratio of coarse to fine aggregate was 40/60. Each specimen contained 5% AIC. Specimens were manufactured according to MTO LS-261 and tested according to MTO LS-263 and LS-264 procedures for resistance to plastic flow using Marshall apparatus and theoretical maximum relative density, respectively. Table 4 gives the results for the Marshall stability of 5% SDO and 2% Redicote 82.5 blended in 85/100 penetration AIC. Three duplicate determinations were done on each sample. The results show that the SDO blend performed as well as the commercial antistripping agent, Redicote 82s. The SDO addition did not lower the Marshall stability of the samples studied. RECYCLED ASPHALT When asphalt pavement is exposed to atmospheric conditions for several years, it degrades by becoming harder and more brittle. For this reason, asphalt pavement must be replaced or drastically repaired at the end of its typical age of 12 years. One notable effect of aging on asphalt cement is the increase in the asphaltene content. Attempts at softening aged asphalt cement by adding low viscosity high-boiling petroleum oils were unsuccessful because these oils do not accommodate the increase in asphaltene content. A successful agent for softening or rejuvenating aged asphalt cement must be able to disperse or peptize the asphaltenes. In general, very soft asphalts, referred to as fluxes, are used to rejuvenate aged asphalts because they can absorb the effect of the increased asphaltenes. Given the excellent ability of SDO to dissolve the CANMET residue, the ROSE" residue, and the Athabasca asphaltenes, it was expected that SDO would rejuvenate aged asphalt. Further, the softening and reduction in penetration observed in blends of SDO and virgin asphalt cement are also desirable. A sample of aged asphalt cement was obtained from a local asphalt pavement replacement operation. Milled asphalt pavement was extracted with toluene and the aged asphalt cement was recovered by evaporation. The aged asphalt cement was blended with SDO. The penetrations and kinematic viscosities are shown in Table 5. Further, Bow River crude asphalt of 454°C was blended with the aged asphalt. Bow River asphalt is very soft and is considered to be a high quality flux, which is ideal for rejuvenating asphalt. The results indicate that hard, aged asphalt cement of 30 dmm penetration can be softened to make the equivalent of 85/100 penetration grade asphalt cement. Only 12% SDO was required to make this penetration whereas more than 22% Bow River flux is required. However, the viscosity of 22% Bow River flux in asphalt cement is just above the MTO kinematic viscosity minimum of 280 cSt. If more Bow River were added to make the 85/100 penetration grade of asphalt cement, it would probably just fail the viscosity specification. SDO performance in hot mixed recycling was investigated using the thin film oven test. The results in Table 6 show that 9% SDO passes the TFOT weight loss specification, as well as improves the viscosity of recycled asphalt cement. While the penetration is below the 85/100 specification, it is possible that either a soft flux can be used, or further addition of SDO can be utilized to soften to a penetration of 85 dmm. One additional benefit of using SDO for rejuvenation is its property as an antistripping agent. A 9% blend of SDO into aged asphalt cement had a static immersion coverage of 100% on Jamieson aggregate. The unmodified, aged asphalt cement had a coverage of only 42%. The results of SDO rejuvenated asphalt cements in the thin film oven test are also encouraging. As shown in Table 6, when 9% SDO was added to an aged asphalt cement and subjected to the thin f i oven test, a 60% retention of the penetration occurred along with a weight loss of 0.60%. Both meet the ASTM D-946 specification. Not only does it perform well for rejuvenation, softening, and asphaltene compatibility, it also improves the stripping resistance of the rejuvenated asphalt cement. This increases its value over other agents to recyclers of asphalt pavement. Further, it should be noted that the amounts of SDO used are reported as a fraction of the aged asphalt cement. While this may represent a small amount on the scale of recycled pavement, methods to contact the aged asphalt cement with SDO must be considered. These results show the performance of SDO as a rejuvenating agent. 891 CONCLUSIONS It has been demonstrated by extensive antistripping studies conducted on the different SDO samples supplied to ERL that the SDO antistripping properties are independent of the process used to produce the sludge derived oil (bench scale versus pilot scale), and relatively dependent on the sludge type (digested versus undigested). The digestion process removes some materials that would be converted to SDO in the oil from sludge process. The SDO from digested sludge also is less effective as an antistripping agent relative to the SDO from undigested sludge. A 3% SDO concentration was found to be as effective as 5% SDO concentration for stripping inhibition. Furthermore, when SDO was used at 3% concentration, the AIC's properties (penetration, viscosity and weight loss after TFOT) were not modified as much as when a 5% concentration was used. If one would like to use 5% SDO, the starting asphalt cement should be in the 60/70 penetration grade to produce a f d A/C in the 85/100 penetration range. A comparison of the performance of the year old SDO sample with a fresh sample at the 5% additive level showed no major difference confirming that the SDO is very stable over time and does not lose its beneficial properties. The addition of SDO can be done either by the wet mode (standard method) in which the additive is blended with the A/C or by the dry mode in which the aggregates are prewetted with SDO before adding the A/C. In the latter case, asphalt engineering/performance tests would be required to confirm that this method does not affect the performance. This is being addressed in road pavement test strips. The SDO Marshall stability (strength test) was found to be comparable to that of a commercial additive although a greater concentration of SDO is needed. Neither of these additives showed improvement or loss of stability over the commercial virgin asphalt used as control indicating no negative effect on the strength of the asphalt concrete. Annual demand for antistripping agents for road asphalts in Ontario is valued at approximately $1 million (6). It is estimated that up to lo00 t/a of SDO could be used in such an application. Initial experimentation has also shown that SDO shows promise as a rejuvenant for aged asphalt cement. Not only does SDO perform well for rejuvenation, softening, and asphaltene compatibility, it also improves the stripping resistance of the rejuvenated asphalt cement. This increases its value over other agents to recyclers of asphalt pavement. Recycling asphalt pavement is not a mature technology. Many opportunities exist to advance this technology. As the aggregate resources close to major centres become depleted, this technology will undoubedly receive more attention. ACKNOWLEDGEMENTS The authors wish to thank P. Mourot and D. Martinoli of Enersludge Inc. and H. Campbell of WTC for their financial and technical support and their assistance in interpreting the results of this work. The authors also wish to thank J. Odgren of the Regional Municipality of Ottawa- Carleton's Material Testing Laboratory for technical support and interpretation of the Marshall Test portion of the research. Federal support of this work was provided through the Federal Program on Energy Research and Development (PERD). REFERENCES 1. Campbell, H.W. and Bridle, T.R. "Sludge management by thermal conversion to fuels", Roc. of New Directions and Research in Waste Treatment and Residuals Mananement, Vancouver, 1985. Campbell, H.W. and Martinoli, D.A. '"A status report on Environment Canada's oil from sludge technology", Proc. of Status of MuniciDal Sludne Manaeement for the 199Os, WPCF Specialty Conference, New Orleans, 1990. Gearhart, J.A. and Nelson, S.R. "Upgrading heavy residuals and heavy oils with ROSE": Energy Procrssing/Canada, May-July , 1983. Pokier, M.A. and Sawatzky, H. "The utilization of process residues for the production of road asphalt cements", Proc. of 30th Annual Conference of the Canadian Technical Asohalt Association, Moncton, 30, 42-58 (1990). Stolle D.F.E. "Silane coupling agents to reduce moisture susceptibility of asphalt concrete", MTO Reuort #PAV-90-04, November 1990. Kennepohl, G.J., MTO, personal communication, 1990. 2. 3. 4. 5. 6. 892 Table 1 - ASTM asphalt cement specification tests ASTM method Test description ASTM D-5 ASTM D-92 ASTM D-113 ASTM D-1754 ASTM D-2052 ASTM D-2170 ASTM D-2171 Penetration of bituminous materials Flash Point, Cleveland Open Cup Ductility of bituminous materials Effect of heat and air on asphaltic materials Solubility of asphalt materials in trichloroethylene Kinematic viscosity of asphalts Viscosity of asphalts by vacuum capillary viscometer ,y,$s~o 7, 62, 135 282 1557 3288 *150 88.66 ~%SDO 11,104,204 244 1006 2584 +150 8880 5 s s ~ o 11,128,238 226 7684 2272 135 8889 ,-,%~~Ooid 8, 86. 158 324 1573 3321 +150 88.83 5 y s ~ o o l d 13,156. 168 248 IOU 226 +I50 8885 Table 2 - Atlanta SDO in asphalt cement 000 58 71 3574 461.5 +I50 072 82 60 2456 3838 '150 124 88 54 2236 3546 r15O 002 53 55 3178 446.2 +I50 0.81 75 48 1756 340.3 +I50 WLSDO 38 8.86.158 324 1573 3321 -1% 8883 55SOO 96 13.1yI.168 248 IOU 2280 +I% 8985 ZXRdbo(olU6 74 13.121.144 Ya 1019 ,2760 ,150 89S3 1% Nddad S4 0 ~ s ~ )12 .141.253 288 7162 2370 148 88.81 3y.s~O 15.173.307 288 5106 1822 126 8883 5%sm 17.227.372 232 4202 1885 115 88.68 o * s ~ o o l d 15.174.218 316 4761 181 6 130 88.85 5%~DOald 19,242,301 284 3151 141.8 +I50 8885 &!cxlwm DI~SDO m ?5.174.216 316 4701 %si8 130 8985 5WSDO 85 18.242.xIl 2W 3151 1418 1150 8885 ZURsdWlr82S 85 16.1B4.282 YY 3884 1688 +l% 8898 1% Naldad 87 1 2 w n e o 75 u!xlsum WSDO 39 8.83 382 0 22SDO 85 8.81 288 1292 327.0 1XmeLarinsO 60 - 8YlW *95. lSDRW f85. 'SpsC*sTMOl~ '*Spas MTO c202 ,232 ,280" .,w r88 ,220 >lW r88 001 78 56 1426 3164 ++so 0.00 88 57 1421 282.8 +I50 124 107 47 877.6 2544 +is0 0.M) 88 57 7885 2440 - 084 132 54 8135 183.7 +150 002 53 552 3178 4462 *I50 25 081 75 481 1758 3403 +150 S4 016 68 582 1815 387.3 138 87 030 62 681 3 5 3 2 5 1 2 0 a 85 r47 .75 Table 3 - Performance of SDO in asphalt cement 893 Table 4 - Marshall stability tests (MTO LS-263) Sample Stability Flow V.M.A.' Air Voids # ( KN ) (.OOl in.) ( % ) ( % ) Control I A 11.225 9.50 15.09 6.00 I B ll.M)2 9.90 14.88 6.20 II A 12.622 _-. 13.57 4.88 11 B 12.958 11.00 13.61 4.96 IlIA 13.972 10.12 14.33 5.56 IllB 12.292 10.72 15.25 6.45 5% SDO I A 14.330 9.10 14.17 5.47 I B I I .063 8.20 14.59 5.30 II B 10.995 9.40 13.29 4.72 IIlA 12.993 10.76 12.54 4.12 llIB 13.612 9.10 13.02 4. I4 I1 A 10.448 __. 13.47 5.28 2% Redicote 82s I A 6.687 2.80 18.46 2.58 I B 6.689 3.00 16.35 2.59 I1 A 12.017 9.40 14.14 5.83 II B 10.585 10.50 14.38 6.28 IllA 12.545 8.34 14.40 6.34 IllB 14.784 9.09 13.83 5.78 ' V.M.A. - Voids in mineral aggregates Calculated by the following formula: V.M.A.= 100 - [ C (lOO-AC)/G I where C = Bulk specific gravity of compacted bituminous mixture AC = Wt % asphalt cement (A/C) G = Maximum specific gravity of aggregate. assumed at 2.70 All mixes made with 5% by weight A/C - 85/100 penetration Coarse aggregate ( >#4 US Sieve) - Jamieson 140 wt %I Fine aggregate ( <#4 US Sieve) - Dibhlee I60 wt %I Table 5 - Comparison of penetration and viscosity in AAC' using different additives Penetration Kinematic viscosity 4°C 25°C 100°C 135°C (dmm) (CSt) AAC 7 30 11039 866 AAC + 2.1% SDO 7 40 6883 719 AAC + 6.0% SDO 7 50 4794 624 AAC + 12.2% SDO 11 98 4801 413 AAC + 22.0% Bow River 454°C 13 80 3079 287 Bow River 454°C > 300 da' 67 21 SDO > 400 d a 22 7 Table 6 - AAC properties before and after addition of SDO ~ Post TFOT AIC SDO Viscosity Pcnctration TFOT Viscosity Penetration Retained Q60"C Q 25°C wt loss Q60"C @ 25°C pen4 (%) (P) ( d m ) (%) (P) (dmm) (%) AAC 0 29 720 32 0.03 45 210 AAC 9 5 330 65 0.60 9 583 39 60 AAC: Aged asphalt cement n/a: n o t d y z e d Pen: Penetration 894 / / ..... ~. ....... -.- ............. ........... ..... r...... ~.-.. . . -~~ ... - I /- . . I Fig. 1 - 100 80 m 80 3 d 70 f $ " * s o 40 30 SCHEMATIC OF OFS PILOT PLANT, HAMILTON, ONTARIO ......... s.".-4 -. -,,..- .-. . - --- ....... .. -.:.= ........... - [*AJc1 +NC2 +Nc) rAlun.01 1 2 3 4 5 6 Wt% SDO In Asphalt Cement Fig. 2 - STRIPPING BY STATIC IMMERSION TEST - 40 - . - - _-. - . . _.-. 30 + Alkazn-0 *Rdcl AP +Hyl.Crk. *Harnlln +Rdct 825 1 0 1 2 3 4 5 6 X Blend In 85llOO AJC 4 Flg 3 - ANTISTRIPPING TEST ADDITIVE BLENDS VS RETAINED COATING 100 90 i B 70 z $ 60 50 10 .. .- -I 1*85/100 NC4 *150/200 NU 1 0 1 2 3 4 5 WIX SO0 b UC Fig. 4 -STRIPPING BY STATIC IMMERSION TEST 895 MANUFACTURE OF AMMONIUM SULFATE FERTILIZER FROM FGDGYPSUM M.4.M. Chou, J.A. Bruinius, Y.C Li, M. Rostam-Abadi, and J.M. Lytle Illinois State Geological Survey 615 E Peabody Drive Champaign, IL 61820 KEYWORDS: gypsum, ammonium sulfate, flue gas desulfurization ABSTRACT The goal of this study is to assess the technical and economic feasibility of producing marketable products, namely fertilizer-grade ammonium sulfate and calcium carbonate, from gypsum produced as part of lime/limestone flue gas desulfurization (FGD) processes. Millions of tons of FGD-gypsum by-product will be produced in this decade. In this study, a literature review and bench-scale experiments were conducted to obtain process data for the production of marketable products from FGD-gypsum and to help evaluate technical and economic feasibility of the process. FGD-gypsum produced at the Abbott power plant in Champaign, IL was used as a raw material. The scrubber, a Chiyoda Thoroughbred 121 FGD, produced a filter cake product contains 98.36% gypsum (CaS0,.2H20), and less than 0.01% calcium sulfite (&SO3). Conversion of FGD-gypsum to ammonium sulfate were tested at temperatures between 60 to 70°C for a duration of five to six hours. The results of a literature review and preliminary bench-scale experiments are presented in this paper. INTRODUCTION The 1990 amendments to the Clean Air Act mandate a two-stage, 10-million ton reduction in sulfur dioxide emissions in the United States'. Plants burning high sulfur coal and using FGD technologies must also bear increasingly expensive landfill disposal costs for the solid waste produced2. The FGD technologies would be less of a financial burden if successful commercial uses were developed for the gypsum-rich by-products of the wet limestone scrubbing. The degree to which FGD-gypsum is commercially used depends on its quality. Currently, high-quality FGD-gypsum with purity greater than 94% is used mainly to manufacture construction materials, i.e. stucco and gypsum-plaster, gypsum wall boards, and cement'. The amount of high quality FGD gypsum could exceed the current demand of the FGDgypsum industry. Conversion of FGD-gypsum to marketable products could be a deciding factor in the continued use of high-sulfur Illinois coals by electric utilities. One approach is to produce cost-competitive ammonium sulfate fertilizer and commercial-grade calcium carbonate from FGD-gypsum. Ammonium sulfate is a valuable source of both nitrogen and sulfur nutrients for growing plants. There is an increasing demand for sulfur in the sulfate form as a plant nutrient because of diminished deposition of atmospheric sulfur compounds from flue gas emissions and more sulfur is taken up by plants produced in high yields'. Also, the trend of using high-nitrogen content fertilizers has pressed incidental sulfur compounds out of traditional fertilizer. The current market for ammonium sulfate in the United States is about two million tons per year. It is anticipated that 5 to 10 million tons of new ammonium sulfate production may be required for fertilizer markets annually to make up for the loss of sulfur deposition from the increased restriction on acid-rain. The fertilizer industIy appears ready to accept an added source of fertilizer grade ammonium sulfate to supply sulfur in NPK fertilizer blends'. In Phase-I of this study, a literature review and a series of bench-scale experiments were conducted to obtain process data for the production of ammonium sulfate from FGDgypsum and to help evaluate technical and economic feasibilities of the process. EXPERIMENTAL PROCEDURES Sample of FGD-gypsum and methods of nnnlyses -The Abbott power plant in Champaign, Illinois operates a Chiyoda Thoroughbred 121 FGD-desulfurization system which produces one ton of gypsum for every ten tons of coal burned. The FGD-gypsum sample collected was dried in ambient air for two to four days. The particle size distributions of the sample were determined using both manual and instrumental methods. In the manual method, the sample was wet-sieved through a 100 mesh (149pm) screen and then a 200 mesh (74pm) screen. The weight % of the size-fractional samples were determined after drying. In the \ 3 896 '. I I I instwmental method, a Micro Trac I1 analyzer was used to determine the mean and standard deviation of the panicle diameter by means of laser light scattering. The amounts of free water (released at 45°C) and combined water (released at 230°c for gypsum), calcium oxide (CaO), magnesium oxide (MgO), and carbon dioxide in the sample Were determined by the ASTM method (2471. Based on these analytical results, the Compositions were calculated in terms of %CaCO, %MgCO, %CaSO,, %CaS04.2H@, and %(NH&SO,. Thermogravimetric analysis (TGA) was conducted under an air flow of 50 d h h with programmed heating from room temperature to 900°C at lO"C/min. The Weight loss profile was used for preliminary estimates of purity and composition Of gypsum Conversion of FGD-gypsum ammonium sulfate and calcium carbonate - The batch, bench-scale reactor system consisted of a 1000-mL, three-neck, round-bottomed flask fitted with a mechanical stirrer, a condenser, and a thermometer. An autotransformer and heating mantle were used to control the reaction temperature. The important reaction for producing ammonium sulfate from the FGD-gypsum is the reaction between ammonium carbonate and calcium sulfate. Two sets of experiments were conducted in this study. In the first set of experiments, the gypsum reacted with reagent-grade ammonium carbonate in a liquid medium. In the second set of experiments, ammonium carbonate, formed by the reaction of ammonia and carbon dioxide in a liquid medium reacted with suspended gypsum. The procedures for the experiments are outlined below. FGD-gypsum was added to an ammonium carbonate solution (prepared by dissolving reagent-grade ammonium carbonate in 500 mL of distilled water) in the 1000-mL reaction flask The temperature of the stirred mixture was raised from room temperature to the reaction temperature and maintained at that temperature for a range of pre-determined times. The solution which contained the ammonium sulfate product was separated from the solid byproduct, calcium carbonate, by vacuum filtration. The filtrate plus the rinsing, a total of about 600 mL of the liquid, was concentrated to a volume of about 150 mL in a constant temperature water bath. The residual concentrate was kept at room temperature to form ammonium sulfate crystals. The condensation and crystallization steps were repeated until no more crystal could be produced. The combined product was dried under ambient air before determining the total weight. In the second set of experiments, ammonium carbonate was formed by the reaction of ammonia and carbon dioxide in a liquid medium, which was then allowed to react with FGD-gypsum in suspension. After removal of the calcium carbonate, the ammonium sulfate is recovered in a similar manner by filtration, evaporation and crystallization. The ammonium sulfate produced was analyzed by melting point determination, chemical analysis and TGA analysis. The yield of the ammonium sulfate produced was obtained based on its theoretical yield from a total conversion of calcium sulfate feed. The purity of the ammonium sulfate produced was determined by chemical analysis of the nitrogen content using methods described by the Association of Agriculture Chemists (AOAC) and American Water Works Association (AWWA) procedure, and by ASTM method C-471. The calcium carbonate by-product was dried and subjected to TGA analyses to determine its purity and composition of unreacted gypsum. RESULTS AND DISCUSSIONS Characterization of the FGD-gypsum sample - The data on particle size distribution obtained by passing the gypsum sample through a series of screens and by Micro Trac I1 particle-size analyzer are shown in Table 1. Ahout 84% of the sample has particle-size smaller than 74 pm (200 mesh), and about 99% of the sample has particles of smaller than 149pm (100 mesh). The results of chemical analyses and the calculations following ASTM method C-471 are shown in Table 2 The FGD-gypsum sample has more combined water (water of hydration) than free moisture and has 98.36% gypsum (CaS04.2H20) with less than 0.01% of calcium sulfite (CaSO,). The TGA curve of the gypsum sample is shown in Figure 1. All weight loss occurred between 98°C and 207°C (peak at 158.65"C). This weight loss is related to removal of the water of hydration from gypsum. No further thermal decomposition occurred to a temperature of 900°C. Literature review - The chemistry of the process and process conditions6 were reviewed. The literature study showed that the production of ammonium sulfate from natural gypsum, ammonium, and carbon dioxide, known as the Merseburg Process, has been tried in England' and Indias in 1951 .and 1967, respectively. The process was proven to be 897 commercially feasible. The Merseburg Process for manufacturing ammonium sulfate from gypsum is based on the chemical reaction between gypsum and ammonium carbonate. Ammonium carbonate is formed by the reaction of ammonia and carbon dioxide in aqueous solution. The reaction produces insoluble calcium carbonate and an ammonium sulfate solution. The reason it is not currently used is the cost of natural gypsum and the availability of an economical source of carbon dioxide. In the early 196O's, the chemistry of the Merseburg Process was carefully studied and partially developed in the US.. At that time the Tennessee Valley Authority (TVA) studied a process in which ammonium phosphate was produced using ammonium sulfate and phosphate rock as starting materials. In the process, phosphate rock was extracted with nitric acid. The extract was allowed to react with ammonium sulfate to produce ammonium phosphate. Gypsum was produced as a by-product To minimize the costs of ammonium phosphate conversion, TVA adopted the Merseburg Process and developed a single-stage reactor both in bench scale9 and in pilot scalelo operations. The purpose was to recover the by-product gypsum and use it to regenerate ammonium sulfate for the starting material. In the regeneration, the by-product gypsum and ammonium carbonate were premixed before entering the reactor. Residence times of 0.5, 1, and 3 hours at 125°F (52°C) and 140°F (60°C) were tested, and conversions of greater than 95% were achieved9. Typical operating conditions in the pilot plant were 120°F (49"C), 2 hours residence time, and ammonium carbonate feed at or above 105% stoichiometric requirement, and the conversion was 98%*4 Bench scale testing for ammonium sulfate production - The important reaction for producing ammonium sulfate from the FGD-gypsum is the reaction between ammonium carbonate and calcium sulfate. Two sets of experiments (see experimental procedures section) were conducted in this study. The reaction conditions, amounts of reactants, and the properties of products for the two sets of experiments are listed in Table 3. The ammonium sulfate produced was confirmed both by comparing its melting point with that of a commercial standard and by examining chemical analysis data and TGA data. Based on the weight of the ammonium sulfate produced and its theoretical yield from a total conversion of calcium sulfate feed, a yield of up to 83% and a purity of up to 99% for the ammonium sulfate production was achieved. A mass balance calculation for calcium and sulfur in gypsum was conducted on experiment run No. 5 (Table 3). The results show a recovery of 98% for calcium in calcium carbonate and a recovery of 81% for sulfur in ammonium sulfate were obtained. The TGA curve for calcium carbonate produced in one of the residues is shown in Figure 2 The graph shows a weight loss occumng between 600°C and 770°C This is attributed to the evolution of carbon dioxide from decomposing calcium carbonate. A typical TGA curve (Figure 3) of the ammonium s'ulfate produced shows a total decomposition of the sample with a maximum weight loss at 418.3"C In summary, the results of these preliminary laboratory experiments suggest that high quality ammonium sulfate can be produced from the FGD-gypsum sample obtained from the Abbott power plant ACKNOWLEDGEMENT & DISCLAIMER This report was prepared by MA. M. Chou and the ISGS with support, in part, by grants made possible by U.S. department of Energy (DOE) Cooperative Agreement Number DEFC22- PC92521 and the Illinois Coal Development Board (ICDB) and the Illinois Clean Coal Institute (ICCI). Neither authors nor any of the subcontractors nor the US. DOE, ISGS, ICDB, ICCI, nor any person acting on behalf of either assumes any liabilities with respect to the use of, or for damages resulting from the use of, any information disclosed in this report. The authors would like to acknowledge D.F. Fortik of the Abbott Power Plant for suppling the FGD-gypsum sample and the project manager D.D. Banejee of the ICCI. REFERENCES 1. 2 3. E Claussen, "Acid Rain: The Strategy," Office of Communications and Public Affairs, EPA Journal, Washington DC, Jan./Feb. 1991, vol. 17, no. 1. R.E. Bolli and RC. Forsythe, "Ohio Edison Company's Clean Coal Technology and Waste Utilization Programs," EPRI 1990 SO2 Control Symposium, vol. 1, 3b. S.K. Conn, M.G. Vacek, and J.T. Moms Jr., "Conversion from disposal to commercial grade gypsum; An alternative approach for disposal of scrubber wastes." EPRI 1993 SO2 Control Symposium, voL 3,8b. 4. R.G. Hoe% J.E Sawyer, R.M. VandenHeuvel, Mk Schmitt, and G.S. Brinkman, 898 Corn response to sulfur on Illinois soils, J. Fert, 295-104, 1985. 5. N.W. Frank and S. Hirano, "Utilization of FGD Solid Waste in the Form of Byproduct Agricultural Fertilizer," EPRI 1990 SO2 Control Symposium, vol. 1, 3b. 6. Kirk-Othmer, Encyclopedia of Chemical Technology, fourth edition, vol. 2,706, John Wiley & Sons, New York, 1992 7. Higson, G.I., The Manufacture of Ammonium Sulphate from Anhydrite, Chemistry and Industry, 750-754, September 8, 1951. 8. Nitrogen, Conversion of Gypsum or Anhydrite to Ammonium Sulfate, Nitrogen, 46, MarcWApril 1967. 9. Blouin, G.M., O.W. Livingston, and J.G. Getsinger, Bench-Scale Studies of Sulfate Recycle Nitric Phosphate Process, J. Agr. Food Chem., vol. 18, 313-318, 1970. 10. Meline, R.S., H.L Faucett, CH. Davis, and A.R. Shirley Jr., Pilot-Plant Development of the Sulfate Recycle Nitric Phosphate Process, Ind. Eng. Process Des. DeveIop., vol. 10, 257-264, 1971. Table 1. Results of particle size anawis of the FGD-gypsum Siie 5% > 149pm' 0.97 149-74 pm' 15.40 < 74pm' 83.60 average diameter em), 73.88 standard deviation2 35.63 'By S I e v e analyss; 'ny Micro 'Itac 11 Table 2 Results of ASTM chemical analysis of the FGD-gypsum Composition in WL % Analytes moisture free basis combined water 20.59 GO 3292 MgO 0.01 so4 54.90 so1 co.01 co2 0.71 NH, <0.01 Free Moisture co.01 Calculated Values C~SO,.~H,O 98.36 1.60 co.01 Caco, Caso, Caso, co.01 MEOi 0.01 (Nrq),S04 co.01 / I f I Table 3. Reaction conditions and the results of fmal product and by-product analyses Caco, W4)ZW I Run Run 'Mole number Conditions ratio ' Wt% lCalculated 'purity 'Calculated 'm.p. residue yield yield ('C) 1 70'CShr 1.56 97 ND ND ND 242 2 70'C6hr 1.59 86 ND 95 82 237 3 70'C6hr 1.33 81 81 ND 83 241 NH, C02 clco, (NH4)ZSO' Run Run number conditions m o k h Lwt% in Icalculated 'purity 'Calculated h p . residue yield yield ('C) 4 60'C4h+ 1.50 1.25 ND ND 99 58 240 5 6YC6hF 1.25 1.00 94 104 9s 83 241 6 70eC6h14 1.25 1.00 ND ND 90 76 2?.7 3 ]Based on theoretical yield kom FGD-gypsum feed; 'Wet chemical analysis by ASTM c-171 and AWWA procedures; SMelting point for the standard is 240'C; 61.9S mole of gypsum used. I/ / 899 Figure 100- 95 - 90 - 85- -0.4 -0.2 0 80 - 206.73 C 79.46: 80 - 60- 40- 20- weight (:) Deriv. Weight ( : / C ) 100 0.8 90- 0.6048%/C -0.6 80- -0.4 70 - -0.2 60- -^ 0 i5a.81 c 57.83: 50 ! , -0.2 0 200 400 600 800 1000 Temperature (C) Figure 2 Typical TGA weight loss profile and first derivative for solid by-product (CaCO,) from the ammonium sulfate production. lo,% 2.5 -2.0 -1.5 -1.0 -0.5 - -0.0 o , , , , , i-0.5 \' 1' I 900 J I TECHNO-ECONOMIC EVALUATION OF WASTE LUBE OIL RE-REFINING IN SAUDI ARABIA Mohammad Farhat Ali, Abdullah J. Hamdan and Faizur Rahman DEPARTMENT OF CHEMISTRY Kin"e Fahd Universitv of Petroleum & Minerals Dhahran: Saudi Arabia Keywords: Waste Lube Oil. Re-refining, Economics INTRODUCTION Abut 80 million gallons of automotive lubricating oils are sold in Saudi Arabia. Much of this oil, after use, is actually contributing to the increased pollution of land because of indiscriminate dumping. Any scheme of secondary use of the waste lube oils would be of interest both for conservation of energy resources and for protection of environment. This paper discusses the secondary use for the used automotive lubricating oils. Process technology of Meinken, Mohawk and KTI were selected for the techno-economic feasibility study for rerefining used oil. Profitability analysis of each process is worked out and the results are compared. In many counmes the re-refining of the used oils has become an important industry. The objective of recovering high quality raffinates is attained through the use of widely differing techniques. The processes concerned can be classified according to the chemical or physical method ofused- oil pretreatment selected. Meinken process is based on chemical pretreatment whereas, both Mohawk and KTI processes employ physical methods involving distillation and eliminates the use of sulfuric acid thus providing a facility for safer operation than Meinken. The plant capacity of two existing units in Jeddah are 10.000 TPA and 80,000 TPA rerefining of waste oil. We selected a plant of 50,000 TPA waste oil re-refining for economic study of these three processes. Both Mohawk and KTI have been running full range plants in different parts of the world and appear to be efficient and viable. Meinken have successfully implemented more than 60 used oil re-refining plants world wide including Kuwait Lube Oil Company, Iran Motor Oil Company, Saudi Lube Oil Company Limited, Jeddah, and Lube Oil Co. Ltd., Jeddah. PROCESS TECHNOLOGIES Meinken Process The used oil is supplied to the re-refinery by railway tankers, road tankers or in barrels. Before the used oil flows into the waste oil storage tanks, it passes through the filters to remove solid impurities. A block flow diagram of re-refining process is shown in Figure 1. Meinken process is based on chemical pretreatment [I]. The dewatered oil is treated with sulfuric acid (96 %) and the acid refined oil is vacuum distilled to separate lube base oil from the low boiling spindle oil and gas oil. With sulfuric acid treatment it is necessary to dehydrate the feedstock completely before subjecting it to acid treatment to prevent dilution of the concentrated sulfuric acid. On the other hand, there is no need to remove crankcase "dilution" or fuel components ahead of the acid-treating step, since these could be conveniently stripped from the hot oil in the subsequent clay contacting step. Their presence during acid treatment reduces the viscosity of the oil and thereby increase the ease of separating the acid sludge. However, the sulfuric acid - treatment and clay addition produce waste streams like acid tar and spent clay resulting in a problem of waste disposal. Inspite of the disposal problem associated with Meinken process, the Meinken technology appears to be very popular. At present, there are about 60 such refiners around the world using the same system. New refineries of this w e are in various stages of consauction and planning in Kuwait, Saudi Arabia, UAE, Oman and India manifesting the technology to be well proven and widely accepted. Mohawk Process A simplified block flow diagram of the Mohawk-CEP process is shown in Figure 2. This is claimed by the licensors to be. the newest and yet proven high-efficiency re-refining technology. Mohawk technology has been licensed to Chemical Engineering Partners, a private chemical engineering design company based in California, U.S.A. The fust stage of the process removes water from the feedstock [5,6]. The second stage of the process is distillation, at this step light hydrocarbons are removed resulting in a marketable 901 fuel by-product. The third stage, evaporation, vaporizes the base oil, separating it from the additives, leaving behind a by-product called residue. This residue is used in asphalt indusq. me final processing stage is hydrotreatment which results in a high quality base oil. The Mohawk process features continuous operation, low maintenance, longer catalyst life span, reduced corrosion, and proven technology. KTI Process Kinetic Technology International (KTI) of the Netherlands, in close cooperation with Gulf Science and Technology Co. (Pittsburgh, Pa.) has developed a new re-refining process for all types of waste lubricating oils [5,101. The KTI waste lube oil re-refining process involves a series of proprietary engineering technologies that affords high economic returns without resulting in environmental loads. The main features of the KTI process include : (a) high recovery yield up to 95 7% of the contained lube oil; @) excellent product quality; (c) flexible operation with wide turndown capability; (d) no requirement for discharging chemicals or treating agents; (e) absence of non commercial byproducts; and (f) reliable, inexpensive treatment of waste water contained in the wasted lube oil. The important steps of this process are as follows. Atmospheric distillation, which removes water and gasoline. Vacuum distillation using special wiped film evaporators separates lube oil from heavy residue containing metals and asphaltenes. The next step is hydrofinishing of lube oil. Hydrogen rich gas is mixed with the oil and heated before passing through the reactor. The treated oil is then steam smpped or fractionated into cuts using a vacuum in order to obtain the right specification. ECONOMIC EVALUATION Capital Investment The total fixed capital investment to process 50,000 TPA of waste oil was obtained from Meinken [l] and Mohawk [6] in 1991. Location factor of 1.25 was used to estimate the fixed capital costs for Saudi location [2]. Table 1 lists the total fixed capital investment estimated for both the technologies. Working capital for the re-refinery was estimated by itemizing the production costs components [12]. It varies with changes in raw material prices, product selling price and so on. Economic evaluation of KTI process could not be carried out because of non-availability of complete cost data. Production costs Production costs consists of direct costs, indirect costs and general expenses. Direct cost includes expenses incurred directly from the production operation. These expenses are : raw materials (including delivery), catalysts and solvents, utilities, operating labor, operating supervision, maintenance and repairs, operating supplies, royalties and patents. Raw material prices were estimated from F.O.B. prices in Germany in September 1991 r1.31 and includes $90.0 per ton for shipping. Local price was used for sulfuric acid. Cokction Cost of waste oil in Jeddah [I], Saudi Arabia was estimated as $53.52 per ton. Byproduct( gas oil) price $1 10 per ton was taken from Petroleum Economist[9], for Caltex, Bahrain location. By-product asphalt price $130.0 per ton was taken from CMR 131, but reduced by 15% as it needs some more processing. If the asphalt residue can not be sold at international price due to low demand in this region, its price has to be further reduced. For economic analysis purposes, the price of asphalt residue was still lowered by 50 I.Th is is an approximation and the price used finally in the calculations is $55.0 per ton of asphalt residue. The raw nuterials, utilities, and manpower requirements are given in Table 2 which were obtained from Meinken [I] and Mohawk [8]. Table 3 lists raw materials, utilities and manpower costs estimated for Saudi Arabian location [2,1 I]. Natural gas price was taken as $0.5 per million Btu [21 and the benefit of low price of natural gas is reflected in utilities costs such as electricity and steam. However, process water is expensive in Saudi Arabia because it is produced from desalination plants. Operating costs which includes operating labor, supervision, maintenance and repairs and indirect costs which includes overheads, storage and insurance, and general expenses were estimated according to the standard procedures [ 7,13,141. 902 d Sumation of all direct costs, indirect costs and general expenses results in a production Cost. Table 4 illustrates production cost of re-refining waste oil resulted from the two technologies. The estimated production cost for Meinken process was $348.8 per ton and for Mohawk process it is $198.4 per ton of re-refined oil. For Meinken process the raw materials cost is about 54 % of the production COS^ Utilities is 3.0 %, operating cost is 17.2 %, total indirect costs is 20.4 % and general expenses about 7.4 % of the total product cost. The share of raw materials cost in the total product Cost is dominant. \ In case of Mohawk process the raw materials cost is about 42.7 % of the total product Cost. By-products are 12.4 %, utilities are 8.8 %,operating cost 23.0 %, total indirect costs are 25.4 % and general expenses are 12.5 % of the total product cost. So, the production cost will be sensitive to raw materials prices and sensitivity analysis was performed for different raw materials price. Profitability Analysis The profitability of an industrial opportunity is a function of major economic variables such as product selling price, raw materials prices, capital investment, energy prices and so on. Year-by-year cash flow analysis have been carried out using assumptions and financial arrangements described in Table 5. From the analysis of production costs (Table 4) components, it is obvious that the raw materials cost is the dominant item. So, sensitivity analysis were performed for 15 % lower and 15 % higherraw materials prices than prevalent in September 1991. Since the re-refined oil is not segregated into different neutral oils and bright stock, following typical composition was assumed: 10 % 300 SN, 80 % 500 SN and 10 % bright stock. Based on LUBREF, Jeddah [4] base oil prices of various grades an estimated selling price of $415.60 per ton is used in the financial analysis. The year-by-year cash flow analysis for international raw materials prices (base case) in September 1991 and for 15 % lower and 15 % higher raw materials prices have been carried out. The results of cash flow analysis are summarized in Table 6. Figure 3 shows the effect of raw materials prices on internal rate of return (IRR). The total fixed capital investment is very high for Meinken process (28.750 million $) as compared to Mohawk process (17.7 13 million $). The working capital amounts to a high value of 4.998 million US. Dollars for Meinken as compared to relatively low value of 3.050 million Dollars for Mohawk. The payback period(PBP) and break-even-point (BEP) for Meinken Process are high as expected compared to Mohawk process, which are 8.16 years and 53.8 % of the full production. The PBP for Mohawk is 1.40 years, and BEP is 28.65 %. The IRR for Meinken and Mohawk are estimated to be 11.24 I and 45.36 %. Thus, the total positive annual cash flow for Mohawk process appears to be. more attractive than that for Meinken. The high profitabilities of Mohawk process are due to lower capital costs as a result of (i) excluding hydrogen plant and (ii) possibly due to relatively not well established technology as compared to Meinken process. The main disadvantage of Mohawk process is that, the plant has to be located near a refinery or petrochemical plant (because of hydrogen supply) to be able to realize such high prufitabilities. If the facilities are to be provided with an independent hydrogen plant, then the capital costs may go up significantly and subsequently profitabilities will be dropped. ACKNOWLEDGEMENT The investigators wish to acknowledge King Abdul Aziz City for Science and Technology (KACST) for funding this Research Project (AR-10-60). The facilities and suppon provided by the Department of Chemisuy and the Research Institute of King Fahd University of Petroleum and Minerals (KFUPM) is also gratefully acknowledged. REFERENCES 1. 2. B. Meinken, Private Communication, B. Meinken Project and Construction Management Consultants, Haltern, Germany, 1991. SRI, PEP Yearbook International, Volume 1, SRI International Menlo Park, California, U.S.A., 1989. 903 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Chemical Marketing Reporter, Schnell Publishing Company, Inc., New York, U.S.A., September 1991. LUBEREF, Private Communication, Petromin Lubricating Oil Refining Company, Jeddah, Saudi Arabia, 1991. Ali, M. F. and Hamdan, A. J. Studies on Used Lubricating Oil Recovery and Rerefining, Third Progress Report, KACST AR-10-60, KFUPM, Dhahran, 1990. Magnabosco, L.H., M. Falconer and K. Padmanbhan. The Mohawk-CEP Re-refining Process, The Proceedings of Sixth International Conference on used oil Recovery and Reuse, San Francisco, California, May 28-31, 1991. Garrett, D.E., Chemical Engineering Economics, Van Nostrand Reinhold, New York, U.S.A.. 1989. Mohawk, Private Communication, Mohawk Lubricants, A Ltd., Bumaby, B.C., Canada V5G 4G2, 1991. Petroleum Economist, p.31, June 1991. Division of Mohawk Oil co. KTI, Waste Lube Oil Re-refining for King Fahd University of Pemleum and Minerals, Dhahran, Document No. 10091, Kinetic Technology International Corp., California, U.S.A., 1989. TECNON, List of Heavy Petrochemical Indusmes for Royal Commission for Jubail and Yanbu, Madinat Yanbu Al-Sanaiyah, K.S.A., TECNON (UK) LTD., Peuochemicals Marketing and Planning Consulting Services, London, U.K. 1988. Bechtel, L.R., Estimate Working Capital Needs, Chemical Engineering, 67 (4): 127 1960. Axtell, O., Economic Evaluation in the Chemical Process Industries, John Wiley and Sons, New York, U.S.A., 1986. Ulrich, G.D., A Guide to Chemical Engineering Process Design and Economics, John Wiley and sons, New York, U.S.A., 1984. 904 / i f Table 1. Capital investment of 50,OOO TPA waste oil re-refining plant in Saudihbia. Roc+ss Technology Meinken, Germany 28.750 Mohawk, Canada 17.713 Total Fixedcapital in 1991 (Million US $ ) Table 2. Raw materials utilities and manpower requirements per ton of product. -- Waste oil. ton - Sulfuric Acid, ton - Activared clay, ton - Lime. ton - Ammonia water(2395). ton Catalyst. kg * Gas oil. ton * Asphalt, ton Lulk.x - - Cooling water, ton Pmcess water, ton * Hydmgen, ton -Steam, ton Manoower: Fuel oil, ton * Total men for 3 shifs -ve) Sign indicates by-prcduci Meinken pmcess 1.266 0.095 0.049 0.214 0.008 - - 0.060 - 0.075 75.000 - - - 33 Mohawk pmcess 1.343 - - - - 3.76 0.135 0.176 0.116 2.003 97.020 0.003 0.667 31 , Table 3. Raw materials, utilities and manpower costs in Saudi Arabia. Item *- Waste oil, ton - Sulfuric Acid, ton Activated sludge, ton *- Lime, ton - Ammonia water (23%),ton Catalyst, kg Ihditia * Fuel oil, ton Cooling water, ton * Process water. ton 9 Electricity, Kwh Hydrogen, ton * steam. ton Manwwer: * One man year ($/year) Cost ($/unit) 53.52 160.00 673.00 316.00 387.00 3.41 110.00 0.019 0.803 0.015 65.000 4.630 18.000 905 Table 4. Production cost data. Total fixed capital Working capital SIDF loan Annual vanable expenses Annual fixed expenses Annual sales Payback period (years) Bd-even-point (% capacity) IRR (96/year) h e m .- BR ayw-p m-cadtuecri~asls -operating Cost Indirect cost General Expenses TOlal production cost _ _ _ _ _ ~ Meinlren Mohawk RocesS RocesS 28.750 17.713 4,999 3.111 16,875 11,356 7,587 2.873 4,746 3.852 16.410 15,377 8.2 1.4 53.8 28.7 11.2 45.4 Meinken mess 188.21 -6.64 70.63 71.03 25.56 348.79 Table 5. Basis of financial calculations. I ltem Mohawk Recess 84.70 -24.52 63.08 50.34 24.82 198.41 Project life Consuuction period Depreciation method Salvage value Equity/SIDF loan SIDF Loan fee Loan payment Tax rile Inflation C.a pital expenditure: 1st year *- 2nd year 3rd year Capacity utilization: * Istyear * 2nd and subsequent years ~ _ _ _ _ _ Calculaied Basis 20 years 3 w Straight line zao 50% each 3% 7 equal installments stanin8 2 years afler plant start-up 2.5 % 0.0 ?6 20% of fixed capital 45% of fixed capital 35% of rued capital plus working capital 6046 100% Table 6. Profitability of rerefining 50,OOO TF’A waste oil in Saudi Arabia (IO00 $) 906 r I 90- 80- -R 70.- z I3T ' I I 1 (I rc' d VACWY usnunow Figure 1. Block flow diagram of Re-refining of used oil by Meinken process Figure 2. Block flow diagram of Re-refining of used oil by Mohawk-CEP process. 04 85 90 95 100 105 110 PERCENT OF RAW MATERIAL PRICES Figure 3. Effect of raw material prices on IRR. 5 907 THE EFFECT OF PROPYLENE PRESSURE ON SHAPE-SELECTIVE ISOPROPYLATION OF BIPHENn OVER H-MORDENITE Y. Suai, X. Tu'), T. Matsuzaki, T. Hanaoka Y. Kubota, J.-H. Kimb), and M. Matsumoto" National Institute of Materials and Chemical Research, AIST, Tsukuba, Ibaraki 305, Japan ') esearch and Development Center, Osaka Gas Co., Ltd., Osaka 554, Japan "Department of Materials Science, Tottori University, Tottori 680, Japan Keywords: shape-selectivity, acid site, isomerization INTRODUCTION Catalytic alkylation of aromatics using zeolites has been the subject of much research, because it is essential to match the dimensions between reactants, products, and zeolite pore to achieve highly shape-selective catalysis [1,2]. H-Mordenites have been found as potential catalysts for the shape-selective alkylation of polynuclear aromatics such as biphenyl [3-91, naphthalene [lo-141, and terphenyl [151. We previously described the activity and the selectivity of the slimmest DIPB isomers, 4,4'-DIPB, were enhanced by the dealumination because of reactions inside pores, and discussed the participation of external acid sites for the H-mordenite with the low Sio?/d203 ratio because the pores were choked by coke-deposition [3-7]. However, mechanisms of the catalysis by acid sites and of the steric requirement of substrate and products in mordenite pores have not been fully understood. In this paper, we describe the effect of propylene pressure on the isopropylation of biphenyl over a highly dealuminated H-mordenite, and discuss the role of acid sites at intracrystalline and external surfaces. EXPERIMENTAL Catalyata and reagents. H-Mordenite (HM(Z2Q), SiWAlfi = 220) was obtained from Tosoh Corporation, Tokyo, Japan, and calcined at 500 'C just before use for the reactions. Biphenyl and propylene were purchased from Tokyo Kasei Co. Ltd., Tokyo, Japan, and used without further purifications. The alkylation was carried out without solvent using a 100 or 200 ml autoclave. Oxygen in the autoclave containing biphenyl and HM(220) was purged out with flashing N2 before heating. After reaching reaction temperature, propylene was supplied to the autoclave and kept at a constant pressure throughout the reaction. A standard set of the reaction included: 200 mmol of biphenyl, 2 g of HM(220), 0.8 MPa of propylene pressure, and 250 "C of temperature. Propylene pressure was expressed by the difference between before and after the introduction of propylene. Isomerization of 4,4'-DIPB. The isomerization of 4,4'-DIPB was examined under the condition as follows: 100 mmol of 4,4'-DIPB, 1 g of HM(220), 0-0.8 MPa of propylene pressure, 250 "C of temperature, and 4 h of period. Product analysis. The products were analyzed with a HP-5890 GC equipped with a 25 ml Ultra-1 capillary column, and identified with a HP-5978 GC-MS. The yield of every product was calculated on the basis of biphenyl used for the reaction, and the selectivities of each IPBP and DIPB isomers are expressed as: Alkylation. Each DIPB (IPBP) isomer (mmol) Total DIPB (IPBP) isomers (mmol) Selectivity of DIPB (IPBP) isomers = RESULTS and DISCUSSION Effects of propylene pressure on the isopropylation The isopropylation over HM(220) reached to above 80 % of conversion within 800 min at 250 "C under every propylene pressure among 0.1-0.8 MPa as shown in Fig. 1. No significant effect of propylene pressure was observed in the initial rate of the isopropylation although there were some differences in the late stage. These results showed that primarily principal reaction was the alkylation over KM(220) under every pressure. Figure 2 shows the profile of the formation of isopropylbiphenyl (IPBP), diisopropylbiphenyl,@IPB), and triisopropylbiphenyl CnIPB) isomers expressed on the basis of the converslon of biphenyl under various propylene pressures. The yield of products under every pressure was on the same plot. These profiles show that the alkylation proceeds by the same reaction paths under every propylene pressure. 908 / I' f I &?me 3 shows the effect of pressure on the yield of 4- and 3-IPEiP. No swmt effects of propylene pressure on the yields were observed. The yield of 4-IPI3P reached the maximum at 50-60 96 conversion, and decreased as further alkylation proceeded. On the other hand, the yield of 3-IPBP increased monotonously with the conversion of biphenyl. The predominant formation of 4-IPBP in the isopropylation of biphenyl IPBp isomers occurs through the shape selective catalysis inside the pores at any Propylene pressures. The isomerization of IPBP isomers was not observed under these Conditions. The difference of the formation of 4- and 3-IPBP isomers can be understood on the difference of their participation to form DIPB isomers. The profiles of 3- and 4-IPBP shows that 4-IPBP is a sole precursor to produce DIPB isomers, and that 3-IPBP does not participate to +e formation of 3,4'-DIPB, at least, except the late stage of the reaction. Highly selective formation of 4,4'-DIPB is also ascribed to the shape selective catalysis inside H-mordenite pores. The higher selectivity of 4,4'-DIPB compared to that of 4-IPBP shows that the isopropyl group of 4-IPBP gives more severe restriction than the hydrogen group of biphenyl at the transition states between propylene, biphenyl, and acid sites in the pores. The formation of 4,4'-DIPB over HM(220) was much influenced by propylene pressure as shown in Fig. 4. The selectivities were as high as 80 % under all pressure conditions at early stages, and kept almost constant during the reaction under higher pressures than 0.3 MPa' However, the decrease of the selectivity was observed under lower pressures than 0.2 MPa, although no significant effects of pressure on the selectivities were observed in the formation of IPBP isomers. The decrease of the selectivity of 4,4'-DIPB corresponded to the increase of that of 3,4'-DIPB. The yields of 4,4'- and 3,4'-DIPB were in linear relations to the yield of combined DIPB isomers under higher pressure than 0.3 MPa. These results show that the alkylation occurs in steady state under high propylene pressures. The yield of 4,4'-DIPB under such a low pressure as 0.1 MPa was deviated downwards from the linear plot under higher pressures, and the upward deviation occurred for the yield of 3,4'-DIPB. The amount of 4,4'-DIPB decreased after reaching the maximum at higher conversion. The inq-ease of the yield of 3,4'-DIPB corresponded to the decrease of the case of 4,4'-DIPB. These results show that the decrease of the selectivity of 4,4'-DIPB is not due to the change of shape-selectivity of the pore, but to its isomerization to 3,4'-DIPB because 3,4'-DIPB is more thermodynamically stable isomer than 4,4'-DIPB. The analysis of the products encapsulated inside the pores after the reaction showed that the distribution of 4,4'-DIF'B was as high as 90 % under every propylene pressure [16]. Further isopropylation of 4,4'-DIPB occurred only in a small amount even under high propylene pressure. This means that alkylation of 4,4'-DIPB is prevented in the pore. H-mordenite pore allows the formation of 4- and 3-IPBP, and 4,4'- and 3,4'-DIPB, while the formation of TrIPB is forbidden in the pore. Takahata and his co-workers found also the increase of the selectivity of 4,4'-DIPB with raising propylene pressure over H-mordenite with the low siOy'M203 ratio [17]. Fellmann proposed that the increase of the selectivity was ascribed to the steric crowding with accumulation of propylene at the active sites [IS]. However, our results show that the change of the selectivity of 4,4'-DIPB with propylene pressure is due to the isomerization of 4,4'-DIPB by preferential adsorption of propylene at the external surface as discussed above. The behavior of 4,4'-DIPB under propylene pressure The behavior of 4,4'-DIPB during the reaction is one of the essential factors for product distributions. Figure 5 shows the stability of 4,4'-DIPB in the presence of propylene pressure over HM(220). Without propylene or under low propylene pressure, 4,4'-DIPB was isomerized significantly to 3,4'-DIPB accompanying IPBP isomers formed by the dealkylation. 'However, the formation of 3,4'-DIPB decreased with the increase of propylene pressure. These tendencies correspond well with the influences of propylene pressure in the alkylation. 4,4'-DIPB was found as exclusive isomer encapsulated in H-mordenites under these conditions (16). These results lead us to conclude that active sites for the isomerization of 4,4'-DIPB are not inside the pores, but at external surface. The inhibition of the isomerization under high propylene pressure show that propylene adsorbs more preferentially than 4,4'-DIPB does. Adsorbed propylene prevents the adsorption of 4,4'-DIPB, and retards its isomerization. However, the adsorption of 4,4'-DIPB predominates over that of propylene under low propylene pressure to result in the enhancement of its isomerization. Further isopropylation of 4,4'-DIPB under the conditions was observed only in small amounts even under high propylene pressure; ie., the alkylation of 4,4'-DIPB is prevented inside the pores. This is one of the reasons why shape-selective isopropylation occurs in the catalysis of H-mordenite. H-mordenite pore allows the formation of 3- and 4-IPBP, and 4,4'- and 3,4'-DIPB, while the formation of TrIPB is forbidden in the pores. The higher steric restriction of 3,4'-DIPB compared with that of 4,4'-DIPB results in the lowly formation of 3,4'-DIPB because molecular diameter of the former isomer is bigger than that of the latter. The relatively high amount. of nIPB was observed in the isomerization of 4,4'-DIPB under 0.1 MPa. TrIPB is formed by the alkylation of DIPB isomers at the external surface because of low propylene pressure. . 909 Mechanistic aspects on the catalysis The isopropylation of biphenyl over H-mordenite occurred inside the pores by successive addition of propylene to biphenyl. Principal factor controlling the catalysis is ascribed to steric restriction at the transition state composing biphenyl, propylene, and acid sites. Predominant formation of 4-IPBP in the isopropylation of biphenyl to IPBP isomers proceeds through the shape selective catalysis inside the pores. The difference of 4- and 3-IPBP isomers for their formation and for their alkylation to DIPB isomers is ascribed to the difference of their accommodations inside the pores. 4-IPBP is a sole precursor to form DIPB isomers, and that 3-IPBP does not participate in the formation of 3,4-DIPB. Higher selectivity of 4,4'-DIPB compared with that of 4-IPBP is due to the bulkiness of 4'-isopropyl group in 4-IPBP. Propylene pressure changed significantly the selectivity of DIPB isomers in the isopropylation of biphenyl over H-mordenite. The isomerization of 4,4'-DIPB occurred under low pressure of propylene. However, total formations of IPBP and DIPB isomers were on the same profiles under any propylene pressure. This means that the isomerization occurs after the formation of 4,4'-DIPB. The distribution of 4,4'-DIPB encapsulated in the pores was highly selective under any pressures 1161. These results support that the isomerization of 4,4'-DIPB occurs at the external acid sites. The isomerization under high propylene pressures is prevented by the preferential adsorption of propylene on the acid sites, whereas under low propylene pressure, the adsorption of 4,4'-DIPB predominates over that of propylene, and thus, the isomerization of 4,4'- DIPB occurs at external acid sites. The high selectivity of 4,4'-DIPB during the isopropylation suggests that the difference of the rate among acid sites at intracrysatlline and external surface is not so significant. It also reflects the amount of acid sites at the surfaces. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. S. Csicsery, Zeolites, 4 (1984) 203. Y. Sugi and M. Toba, Catal. Today, 19 (1994) 187. T. Matsuzaki, Y. Sugi, T. Hanaoka, K. Takeuchi, T. Tokoro, and G. Takeuchi, Chem. Express, 4 (1989) 413. Y. Sugi, T. Matsuzaki, T. Hanaoka, K. Takeuchi, T. Tokoro, and G. Takeuchi, Chemistry of Microporous Crystals (Stud. Surf. Sci. Catal., vol. 60), T. Inui, S. Namba, T. Tatsumi (eds.), Kodansha, Tokyo, and Elsevier, Amsterdam, 1991, pp. 303-330. X. Tu, M. Matsumoto, T. Matsuzaki, T. Hanaoka, Y. Kubota, J.-H. Kim, and Y. Sugi, Catal. Lett., 21 (1993) 71. Y. Sugi, T. Matsuzaki, T. Hanaoka, Y. Kubota, J.-H. Kim, X. Tu, and M. Matsumoto, Catal. Lett., 27 (1994) 315. Y. Sugi, T. Matsuzaki, T. Takeuchi, T. Hanaoka, T. Tokoro, X. Tu, and G. Takeuchi, Sekiyu Gakkaishi, 37 (1994) 376. G.S. Lee, J.J. Maj, S.C. Rocke, and J.M. Garces, Catal. Lett., 2 (1989) 243. T. Matsuda and E. Kikuchi, Zeolites and Microporous Crystals (Stud. Surf. Sci. Catal., vol. 83), T. Hattori and T. Yashima (eds.), Kodansha, Tokyo, and Elsevier, Amsterdam, 1994, pp. 295-302. A. Katayama, M. Toba, G. Takeuchi, F. Mizukami, S. Niwa, and S. Mitamura, J. Chem. SOC., Chem. Commun., (1991) 31. P. Moreau, A. Finiels, P. Geneste, and J. Solofo, J. Catal., 136 (1992) 487. Y. Sugi, J.-H. Kim, T. Matsuzaki, T. Hanaoka, Y. Kubta, X. Tu, and M. Matsumoto, Zeolite and Related Microporous Materials: State of the Art 1994 (Stud. Surf. Sci. Catal., vol. 84), J. Weitkamp, H.G. Karge, and W. Holderlich (eds.), Elsevier, Amsterdam, 1994, pp. 1837-1844. C. Song and S. Kirby, Microporous Materials, 2 (1994) 467. J.-H. Kim, Y. Sugi, T. Matsuzaki, T. Hanaoka, Y. Kubota, M. Matsumoto, and X. Tu, Microporous Materials, in press. G. Takeuchi, H. Okazaki, M. Yamae, and K. Kito, ADD~C.a tal.. 76 (1991) 49. , . I . _. unpublished results. K. Takahata, M. Yasuda, and H. Miki, Jpn. Tokkyo Kokai, 88-122635. J. Fellmann, Catalytica Highlights, 17 (1991) 1. 910 i 100 80 20 0 0 0.1 0 0.2 W 0.3 0 0.4 A 0.8 0 200 400 600 600 Reaction period (min) Fig. 1 Effect of propylene pressure on catalytic activity of HM(220) in the isopropylation of biphenyl. Reaction conditions: biphenyl 400 m o l , HM(220) 2 g, propylene pressure 0.1-0.8 MPa, temperature 250 'C. 0 0 20 40 60 80 100 Conversion (%) Fig. 2 Product Profile of the isopropylation of biphenyl. Reaction conditions were the same as Fig. 1. 0 20 40 60 80 100 Conversion (?h) Fig. 3 Effect of propylene pressure on the yield of 4- and 3-IPBP over HM(220). Reaction conditions were the same as Fig. 1. 911 0 Yield of DIP6 (“70) Fig. 4 Reaction conditions were the same as Fig. 1. Effect of propylene pressure on the yield of 4,4‘- and 3,4’-DIPB over HM(220). 40 I - 30 E. ,” 20 E .0-) CI 0 U a 2 n i o 0 0 0.1 0.2 0.4 0.8 Propylene pressure (MPa) Fig. 5 Effect of propylene pressure on the isomerization of 4,4’-DIPB over HM(220). Reaction conditions: 4,4’-DIPB 100 m o l , HM(220) 1 g, temperature 250 ‘C, period -1 h. 912 \ S E L E ~ V E DIALKYLATION OF NAPHTHALENE WITH HINDERED ALKYLATING AGENTS OVER HM AND HY ZEOLITES UNDER LIQUID PHASE CONDITIONS. Patrice Morearr, Afrrrie Firtiels, Patrick Geneste, Frkdhic Morearr and ]orris Sorofo. brhoratoire de Matiriurrx Cntalytiqrres et Catalyse err Cliirrrie Orgnriiqrie, URA CNR S 478, Ecole Nntiorinle Sigiirierire de Chimie , 8 rile de I’Ecole Norrimle, 34053 MONTPELLIERC edex I, FRANCE. Keywords: zeolites, shape-selective alkylation, naphthalene, heterogeneous catalysis. INTRODUCTION Whereas the use of zeolite catalysts has been widely investigated for the shapeselective conversions of mononuclear aromatic hydrocarbons, such as alkylation of toluene or isomerization of xylenesl-3, in contrast relatively few reports are available on the conversion of polynuclear aromatics, such as naphthalene derivatives. Among the latter, 2,6-dialkylnaphthalenes are the most valuable compounds, since, as precursors of 2,6-naphthalene dicarboxylic acid, they are potentially useful raw materials for production of high quality polyester fibers and plastics4 and of thermotropic liquid crystal polymerss. The interest of such derivatives was shown, in the recent years, by the increasing number of patents relevant to their preparation and separation. Following the pionneering work of Fraenkel et al? on the alkylation of naphthalene with methanol over various zeolites, much attention has been paid, in the early nineties, to studies on the activity and selectivity of zeolites in the isopropylation of naphthalene’-9. The interest of the use of bulky substituents in such reactions over zeolites has been clearly demonstrated by recent papers10-*2. The present communication is concerned with the results obtained in the alkylation of naphthalene over a series of HM and HY zeolites, using isopropyl bromide, cyclohexyl bromide and cyclohexene as alkylating agents, under liquidphase conditions, leading to a better regulation of the reaction pathway. EXPERIMENTAL Materials. Analytical grade cyclohexane, isopropyl bromide, cyclohexyl bromide, cyclohexene and naphthalene (Aldrich Chemical) were used as supplied. Catalysts. H mordenite (Zeolon 100-H, Si/Al = 6.9 from Norton) and three dealuminated mordenites were used for the isopropylation reactions. The dealuminated mordenites were prepared, according the published procedurel3, from Zeolon 100-H (HM) by treatment in 1M HCI solution at 100°C for 3 h or refluxing in 3 M HCI solution for 6 h or in 6 M HCI solution for 12 h. The resulting powders, washed and oven dried at llO°C, had Si/Al atomic ratios of 9 (HMI), 13.1 (IIMz), and 20.6 (1-hf3). The HY catalyst was derived from the thermal decomposition of NH4Y (Linde SK 41, Si/AI = 2.5 from Union Carbide). The US-HY was supplied by Chemisch Fabrik Uetikon, Zurich (26-05-01, Si/AI = 2.5). The CVDmodified zeolites were obtained according the silanation procedureg, and fully characterized by various techNcs14. For the cyclohexylation reactions, the US-HY used was the same as above. The dealuminated HY (Si/Al = 19.5) and HM (%/AI = 10.8) were from Zeocat, Montoir de Bretagne (ZF 520 and ZM 510). Catalytic runs. The isopropylation of naphthalene was carried out in a 0.1 liter stirred autoclave reactor (Sotelem). In a typical Nn, the autoclave was charged with 1 of zeolite freshly calcined in air at SOOT, a mixture of 5 mmol naphthalene and 10 mmol isopropyl bromide in 50 ml of cyclohexane and heated to 200°C. Samples were withdrawn periodically and analyzed by GLC (Altech OW capillary column, 10 m or 25 m x 0.25mm). reactions with cyclohexyl bromide. When cyclohexene was used, the procedure was the following: the autoclave was charged with naphthalene (5 mmol), cyclohexane (50 ml) and the catalyst (1 g), and heating was started; cyclohexene (10 mmol) was then added, drop by drop, by means of a stainless steel pressurized funnel, and the // I The same procedure and analysis technic were used for cyclohexylation / 1 913 , niixture was stirred in tlie same conditions as above. For the isolalion and purificalion of 2,6-dicycloliexyInaplithalene, the procedure was llie following: afler cooling, tlie calalysl was fillered and cyclohexane evaporaled; tlie crude product solidified at room temperature, [lie solid was then fillered and recrystallized from elhanol (my 152°C afler two recryslallizalioiis). The structure was confirmed by GC-MS, If1 and 1% NMK speclroscopy together willi Xray crystallography~5. RESULTS AND DISCUSSION lsopropylation reaction over HM and IIY zeolites. The isopropylation of naplillialene with isopropyl bromide over a series of niordeiiiles and Y zeolites show that bolli conversion and seleclivity depend largely upon tlie nature and the slruclure of llie calalysl. Tlie main resulls, obtained in lliis study and reporled in Table 1, can be summarized as follows: - morcleiiites are less active than Y zeolites; - in bolli cases, a high p-selectivity is found, leading to the selective formation of 2- iso~~ro~~yliiap1il1ialaennde a mixture of 2,6- + 2,7-diiso~~ropylnaplillialenes; - llie formation of Irialkylnaplitlialenes cannot be avoided over untrealed zeolites. The origin of llie high pselectivily observed is differeiil depending on tlie zeolite. Over H-mordenites, such a selectivily is explained as Ilie result of transition-stale shape seleclivity, due lo Ihe constrained environment in tlie channels of tlie mordenile; the steric hindrance of the I-isopropyliiaplillialene (uisomer) does not allow its formation inside tlie tight one-dimensional tunnels of tlie zeolite. Over Y zeoliles, tlie pseleclivity has been shown to be due to a thermodynamic equilibrium favorable to the 2-isopropyliiaplitlialene; [lie 1- isoproi~yliiaplillialene (kinetic product) is iiiilially formed inside llie tlireedimensional large-pore slructure of Y zeolites, and it is then rearranged into tlie 2- isonier (Iliermodynamic product) at high temperatures. Over I l ie two kinds of zeolites, 2,6- and 2,7-diisopropylna~~litlialenesa re tlie main disubstituted derivatives. Such a result is expected taking into account tlie p seleclivity observed in the monoisopropylation step, wliatever [lie origin of this selecl ivity. I n both cases, in our experimental conditions, the same distribulion belween llie 2,6- and 2,7- isomers (2,6-/2,7- ratio = 1) is observed; such a result seems not surprising, i f we consider Ihat these two isomers have tlie same kinetic diameter (6.5A), and that llieir production and subsequent diffusion in tlie pores or cavities of zeolites occur in the same way. Nevertheless, a higher 2,6-/2,7- ratio is generally observed over niordeiiites by some a u t l ~ o r s ~va~ri~ou~s~ a~ss;u mplions, such as differences in diffusion rates or in the ease of transition-slate formation for the two isomers, might explain these higher ratios, obtained in different experinienlal conditions. These results are very encouraging regnrding the efficiency of Ilie calalytic nclivity of zeolites in isopropylalion of naplillialene and heir shape seleclivily properties. Nevertheless, the selective formation of the 2,6-isomer is not possible in any case. Tlie possibilily of an improvemenl of such a seleclivity has then been considered by tlie use of eillier modified zeoliles or more hindered alkylaliiig agenls. lsopropylation over CVDmodified zeolites. Shape-seleclivily of zeolites may be improved by reducing the number of active siles of the external surface. II is known Ilia1 silanation of zeolites leads not only lo such a deactivalion of the outer active sites bul also to a uniform control of the pore-opening size of the ze0lile1~,~~. The “chemical vapor deposition” silanation metliodlR leads, iii parlicular, to a remarkable enliancemenl of the reaclanl and product sliape-selectivity~9~~~. Over such CVD-modified zeoliles, the formation of tlie bulky tria~kyltia~~lit~ialeniess lotally suppressed i n l l ie case of niordenites, and considerably reduced with tlie t i Y zeolites (2% compired with 14% over llie untreated zeolite) (Table 1). Such a result clearly demoiilrales that tlie trialkylation reaction occurs on llie acidic siles located on tlie external surface of mordeniles. The acidic outer siles of the IIY zeolites are also largely involved in tlie formation of tlie triisopropyl derivatives, which can be, neverlheless, produced also inside the cavities of these three-dimensional large-pore Y zeolites. The results obtained in the isopropylation of the 2-isopropyliiaplillialene over CVD-modified HY zeolite, i.e. a drastic decrease in the amount of triisopropyliiaplitlialenes (4% instead of 18% over tlie untreated f4Y), leading lo an 914 \‘ / \ Y J enliancetiiell~ of [lie pseleclivily over the CVD-modified 1 IY for IIiis reaclioti (85x1, codirlii such an involvement. Wilh CVD-modified 1 IM, llie 2-isopropyltiaplillialetie is llie major product (90% a1 10% coilversion); despile (lie low conversion, Iliis resull inusl be lakeii into accounl because 2-isopropylnaplillialene can be easily separated from llie reaclion iiiixlure by siinple dislillation, and then isopropylaled as slarling material for diisopropylnalvlllllalelie production. Willi CVD-modified I-IY zeolites, a high psdeclivily (>90X) is oblained, corresponding lo 63% ot 2-isopropylnaplillialene and 30% of 2.6- + 2,7-diisopropylnaphllialenes at 70% conversion. The overall resulls show that, from a synllielic poiti1 ot view, the CVDmodified zeolites appear IO be the best calalysts witti wliicli it is possible to selectively obtain p- and ~ p ' i-s omers. Cycluliexylatiun reaction over IIY zeolites. As already said above, no seleclivily in 2,6-diiso~~ro~~ylnaplillialecnoeu ld be found despite (lie liindrance of llte isopropyl group; moreover, the separalion of 2.6- and 2,7-diisopropyl isomers is very difficult. 11 was reported that, in the cycloliexylalion reaclioti of naplillialene under conventional conditions (i.e. over Friedel-Crafts calalysls, such as aluminum chloride), llie 2,6-dicycloliexyInaplilhalene could be isolaled, in a very low yield, from the reaclion mixture by crystallizalion21.22. Taking inlo account such a property and the sleric hindrance of llie cycloliexyl group, we studied the catalylic aclivilies of a sample of It-mordenile and IWO samples of I IY zeolites in the cycloliexylation reaclion ot naplillialene with cycluliexyl bromide and cyclohexene respeclively (Table 2). The I h x m l e n i t e presents a weak activity in llie reaclion willt the cyclohexyl broiiiide, as shown by the low conversion of naplillialene (676) a1 200"C, whereas the I I Y zeoliles appear lo be very efficient even at lower temperalures. The ullraslable zeolile I-IY (SiIAI = 2.5) arid lhe dealutninaled sample (Si/Al = 20) exliibil siiiiilar aclivily and selectivities a1 Ihe same temperalures, as shown on Table 2. M e n cycloliexene is used as the alkylating agenl instead of cycloliexyl broiiiide, a sliglil difference is observed i f the reaclion is carried out under the same conditions (naplillialene and alkylaling agent put together in the auloclave), due lo diinerisalion of cyclohexene. When cyclohexene is added drop by drop to tlie stirred mixlure, llie same resulls are illen obtained, bolh in conversion and seleclivity. The compirison of cyclohexylalion with isopropylalion of naplitlialene over llie ullraslable zeolile US-CIY under lhe same condilions is given on Table 3. In bolh reaclions, a high conversion of naphthalene is obtained after short reaction times (10 iiiiii with cyclohexyl bromide, 1 Ii w i l l i isopropyl bromide). The cyclohexylalion reaclion yields an increasing amount of 2,6- and 2,7- dicycloliexyliiaplillialenes, logellier wilh a significant decreasing amourit of lrialkyl derivatives due to the steric hindrance of (he cycloliexyl group, leading IO an iiiiprovernenl of llie 0-p' seleclivily (82% compared will1 71%). Neverllieless, [ l i e relative distribution of the 2,6- and 2,7- isomers does not dramatically change (2,6- /2,7- = 0.95 lor isopropyl arid 1.1 for cycloliexyl. Such a resull confirms lliel Y zeolites increase l l i e 141' selectivity, but do not lead to llie predoiiiitiaiil foritlation of otie given dialkyl isomer. The advantage of [lie cycloliexylation in comparison witlt the isolvropylation is direclly relaled lo Ilie physical properlies of llie 2,6-dicycloliexyInapliIIialene, wliicli is sqiaraled from the mixture by crystallization. Pure 2,6-dicycluliexyIiia~lilIialeiie (while cryslals, nip = 152°C) is isolaled in tiiotlernle yields (19 lo 277%) depending on (lie zeolilesa. 915 As shown above, this crystalline 2,6-dicyclohexylnaphthalene contains a crystallographic symmetry centre; the cyclohexyl subsliluents adopt a chair conformation, and the presence of the two bulky substituents involves only a slight deviation from flatness for the naphthalene ring's. CONCLUSION The liquid-phase alkylation of naphthalene with hindered alkylating agents such as isopropyl and cyclohexyl derivatives can be carried out efficiently over HM and HY zeolites. High conversions and efficient 6-v selectivities are obtained after very short reaction times at 200°C. In the case of isopropylation with isopropyl bromide, the use of zeolites modified by silanation of the external surface leads to an improvement of such a selectivity by suppressing or reducing the formation of the triisopropyl derivatives. The use of cyclohexyl derivatives, cyclohexyl bromide or cyclohexene, as alkylating agents, yields, over HY zeolites, an increasing amount of 2,6- and 2,7-dicyclohexylnaphthalenes, together with a significant decreasing of the trialkyl derivatives. Moreover, the 2,6-di~clohexyInaphtlialene, a crystalline compound, is easily separated from the reaction mixture by crystallization, which is, to our knowledge, the first exemple of the production of a pure 2,6- dialkylnaphtlialene. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Venuto, P.B. Micropororis Mnfrriuls 1994,2, 297 and references therein. Chen, N.Y.; Kaeding, W.W.; Dwyer, F.G. /. Alii. Clrrrri. Soc. 1979,707,6783. Kaeding, W.W.; Chu, C.; Young, L.B.; Weintein, B.; Butter, S.A. 1. Cnlnl. 1981, 67,159. Caydos, R.M. in "Kirk Othmer Encyclopaedia of Chemical Technology" (Kirk, R.E.; Othmer, D.F., Eds.); Wiley: New York, 1981, vol 15, p 698. Song, C.; Schobert, I4.H. Firel Proc. Tech. 1993,34,157. Fraenkel, D.; Cherniavsky, M.; Ittah, B.; Levy, M. /. C,rlnl. 1986, 207,273. Katayama, A,; Tobe, M.; Takeuchi, G.; Mizukami, F.; Niwa, S.; Mitamura, S. /. C h i . Soc., CJrmi. Cotrrin., 1991, 39. Fellmann, J.D.; Saxton, R.J.; Weatrock, P.R.; Derouane, E.C.; Massiani, P. U.S.Patent no 5,026,942,1991. Moreau, P.; Finiels, A,; Geneste, P.; Solofo,J./. Cnlrrl. 1992,736,487. Song, C.; Kirby, S. Micropororis Msferinls 1994,2,467. Sugi, Y.; Kim, J.H.; Matsuzaki, T.; Hanaoka, T.; Kubota, Y.; Tu, X.; Matsumoto, M. Sfrid. Sirif. Sri. Gzfnl. 1994,84, 1837. Chu, S.J.; Chen, Y.W. Appl. Cnfnl. A 1995, 723,51. Fajula, F.; Ibarra, R.; Figueras, F.; Gueguen, C. /. Grlrrl. 1984,89,60. Chamoumi, M.; Brunel, D.; Fajula, F.; Ceneste, P.; Moreau, P.; Solofo, 1. Z ~ o l i t e s19 94,74, 282. Moreau, P.; Solofo, J.; Geneste,P.; Finiels, A.; Rambaud, J.; Declercq, J.P. Acfn Cryslnllogr. 1992, C48,397. Niwa, M.; Kato, S.; Hattori, T.; Murakami, Y. /. Cliern. SOC., Frirndny Trnm I 1984,80,3135. Niwa, M.; Kawashima, Y.; Murakami, Y. /. Clretri. Soc., Frimdoy Trtnis I1985, 87,2757. Niwa; M.; ltoh, H.; Kato, S.; Hattori, T.; Murakami, Y. 1. Ckerrr. Soc., Cketrr. Corrrrri. 1982, 819. Niwa, M.; Kato, S.; Hattori, T.; Murakami, Y. /. Plrys. Clierri. 1986,90,6233. Bein, T.; Carver, R.F.; Farlee, R.D.; Stucky, G.D. /. Ani. Clr~rrr. Soc. 1988,170, 4546. Bodroux, E. Airri. Cliiiii. Pliys. 1929, 77,535. Pokrovskaya, E.S.; Stepantscva, T.C. j. Cen. Clrriti. 1939,9,1953. Solofo, I.M; oreau, P.; Geneste, P.; Finiels, A. P a In t. Appl. WO 91 0159, 1991. 916 Table 1. Isopropylation of Naphlalene over a Spries of Untreated and CM) Modified Zeolites at 200°C with Isopropyl bromide J Catalyst lime. naphlh. product distribulion. D selectivity, 5% h 0Wl.ll COIIY., s 2bAiU Dim Erlect MLEci DlPN Tu'N 2-1PN 2.5 27- oltm HM 24 18 81 4 4 3 8 72 09 HMt 24 60 74 7 6 5 10 70 87 HM2 24 20 61 13 11 9 9 74 a5 HM3 24 16 55 14 13 7 9 80 82 HY 1 e6 42 16 17 12 14 73 74 US-HY 1 97 28 19 20 16 17 71 67 CVD-HM I IO 90 3 4 3 0 70 97 CVD-HY 1 70 63 15 15 5 2 w 93 Table 2. Cyclohexylation of Naphtalene over Zeolites at ?OU' C with Cyclohe.lyl bromide (CB) and Cyclohexene (CH) Calalyrt Alkylal. lime, naphth. pmduct distribution, p. vlectivity 2% agent min cmv., D MCN CCN TCN 2b.-2.7- DCN MCN IXN DCN Hhl CB 70 6 5 6 4 4 46 16 27 H W CB 10 % 31 67 2 6 43 82 w 2 0 CB 10 94 53 46 1 0 43 79 CHa 10 90 85 15 53 2o 36 CHb 2S 98 4 4 5 4 2 7 41 n a CH charged together with naphthalene in the autoclave before heating. b CH added drop by drop. C time corresponding to the end of addltion of naphthalene. Table 3. lsopropylalion and Cyclohexylation of Naphtalene over US-HY Zeolite at 2OWC with Isopropyl bromide (IB) and Cyclohexyl bromide (CB) Alkyl. time naphlh. product distribution, s *lect,"Oty. 7. agenl min c a w 3 DAN^ - MANa 26 27- L6+2.7- o l k n TANC DAN ~~ Is 60 97 28 19 20 39 16 17 71 CB 10 96 31 29 26 55 12 2 82 1 MAN : manoalkylnaphthalenes. DAN : dialkylnaphthalmes. =TAN : bialkylnaphthalenes. 911 REGIOSELECTIVE ISOPROPYLATION OF DINUCLEAR AROMATICS OVER DEALUMINATED MORDENITE CATALYSTS a r e w D. scheuta and Chunshan Song Department of Materials Science and Engineering Fuel Science Program. 209 Academic Projects Building Pennsylvania State University, University Park, PA 16802 Keywords: napht halene. biphenyl, shape-selective alkylation Summarv Selenivc addition of propylene tn naphthalene or biphenyl over dealuminated H.mordcnite (HM)r ~r a l y s t si s being used to produce 2.6-diisnpropylnaphth;1len(e2 .6.DIPN) and 4.4'- diisopropylbiphenyl (4,4'-DIPB), respecrively. When oxidized, these selectively suhstituted dinuclcar ammatics become monomer!+ for liquid cryslalline polymers and engineering plastirs "' Wo. and others. have shown that IIM deolumination increases alkylation regiosclectivity for isopropylation of binuclear aromatics.6-'C In this paper. we more closely examine the effects of IIM dealurninat ion on catalyst anivity and regoselectivity, as well as effects on c:;italyst physical properties Two different mordcnitcs were dealuminated by mineral arid 1e;iching. HM 14 and I-IM38, having SiO~N20, (mol ratio) of 14 and 38, respectively. For naphthalene isopropylation. dealu. mination of IIM14 increases the 2.6-DIPN isomer selectivity from 30 to 60%. Dealumination of HM.38 gives similar results but with lower rcgioselectivity. For comparison. 4,4'-DIPU regiosclcctivity was examined in biphenyl isopropylation over a series of mordenites with SiO~U,O, 14-230. Selectivity for the 4.4'-isomcr increased from 66 to 87%. Therefore, increased selectivity for the slimmest diisopmpyl-isomer with deahnination is a general property: it occurs with differrnt mordcmtr starting materials and different, but similar in srm and shape, reactant moleculcs. Selectivity for p-substitution of naphthalene seems tu correlate with changes in IIM mesopore volume brought about by deulumination. An increase in the mesopore volume is mirrored by an increase in 2.G-DIPN isomer selectivity. IiM micropore voliimos do not chnngc appmciably It has been shown that the two PP-disubstituted naphthalenes have nearly identical critical diameters. biit 2.6-DIPN has a somewhat more linear structure than 2.7.DIPN."6 Consequently. 2,G-DII'N has a lower activation energy for diffusion in lIM.G This explains why HM catalysts typically give 2.61'2.7 DII'N isomer ratios greater than unity We have used X-ray powder diffraction to measure the dwrease in HM UNI cell volumes caused by dcalumination The 2.61'2.7 DlPN ratio shows an approximate inverse relationship with the unit. cell volumes. A probable cuplanaition is that the unit cell contractions caused by dea1umin;ition decrease the channel diameter. slightly, resulting in more snug fit for the Pp-disubstituted isomers in the channels As if consequence, the difference in diffusion rates for 2.6- und 2.7-DIPN IS magnified. Catalyst Preparation. The procedures used in this work have been described earlier.' Mordenite catalysts CBV 10A (NaM14), CBV 20A (HM21) and CBV 30.4 (HM38) were supplied as 10 pm average particle size powders P Q Corporation). HM14 was generated from NaM14 by sodium-exchanged with 1 M NH,Cl. Dealumination was accomplished by stirring HM in aqueous hydrochloric or nitric acid at reflux temperature. Time and acid concentration were varied to control the extent of aluminum removal. All catalysts were calcined 5.5 h at 465 "C except HM230. HM230 was prepared according to a procedure described by Lee et al. for extensive aluminum removal.'6 Accordingly, NaM14 was first treated at reflux with 1 M HCI to generate the HM54 sample. In the second step, HM54 was calcined at 700 "C and treated with 6 M HNO,, followed by final calcination at 700 'C. Samples were dissolved using lithium metaborate fusion and analyzed for silicon, aluminum and sodium by ICP-AES. Sorption data and residual sodium content for the catalysts are listed in Table 1. Catalyst Eualuotion. AS described previously, catalyst testing was done in a tubingbomb batch reactor charged with 0.10 6 catalyst, 1.0 g (7.8 mmol) naphthalene, and 0.66 g (15.fi mmol) propylene.' Naphthalene and biphenyl (Aldrich, 99% grades) and propylene (Matheson, 99.5 % minimum, polymer purity) were used bs supplied. Solution products were analyzed by GC-MS and GC-FID for qualitative and quantitative analyses. respectively, using a 30m x 0.25mm DB-17 (J&W Scientific) column. X-ray powder diffraction GRD) was done on a Scintag 3100 diffractometer using nickelfiltered Cu Ka radiation. The Cu Ka2 component was stripped from the patterns using the standard Scintag algorithm, so the wavelength used in the calculations is 1.540598 A. Samples were mixed with ca. 10 wt% -325 mesh silicon internal standard for 28 corrections. The scan rate was 0.5" 201min with 0.02" steps. XRD pattern indexing and determination of lattice constants for HM (CMC~I space group) was done using the JCPDS-NEWLSQ82 unit cell refinement computer p r ~ g r am.T' ~hi s program minimizes the sum S defined in equation 1, where Ocorr are I J I I the observed Bragg angles, corrected for instrumental and physical peak shifts, and OcaIC are calculated h m th e m n t u nit cell parameters and the weighting factors Whkk Starting values for Owlc are determind from the input unit cell parameters which are then adjusted, using a nonlinear least-squares method, to minimize S. This approach can lead to cell parameters accurate to a few parts in 1 0 , 0 0 0 . ~ ~ Sorption analyses were done on either of two automated instruments: a Coulter SA 3100, or Quantachrome Autosorb. Samples were outgassed at 400 “C. Multipoint surface areas were calculated by the BET method. Micropore volumes were calculated using the T-plot method. Zsopmpylation ofNaphthalene. Catalyst test data for the two serigs of dealuminated HM catalysts is presented in Table 2. Greater than 98% of the products are isopropylnaphthalenes UPN’s). The predominant side reactions result in small amounts of alkylnaphthalenes that are not solely isopropyl-substituted. Mass balances are greater than 96% in all cases, with material losses being primarily attributed to carhonaceous deposits on the catalyst. The effects of dealumination on catalyst performance are shown in graphically in Figures 1-3. HM54 has abnormally low activity so its data for were omitted from these graphs because. HM230 has higher activity than expected, apparently due the higher activation temperature used in its preparation. Both HM14- and HM38 derived catalysts show similar activity patterns: naphthalene conversion first increases, then decreases as aluminum is removed from the lattice (volcano plots, Figure 1). This type of activity trend is quite common for HM and is due to the decreasing acid site densit and increasing acid site strength that occurs with dealumination as SiOdA120, ratios, probably due to gross depletion of acid sites. HM38-derived catalysts retain reasonable activity to higher ratios, indicative of higher acid site concentrations. HM93 (from HM38) shows a moisture loss on ignition 2.5 times the amount desorbed from HM90 (from HM14). This also suggests that HM38-derived catalysts have higher acid site concentrations than HM14-derived catlysts at the same Si0,/.N2O3 ratio. As shown in Figure 2, DIPN yield and selectivity are increased hy dealurnination up to the same maxima defined in Figure 1. Beyond these maxima, monoalkylation dominates and TrIPN+ production falls to near zero. However, p- substitution selectivity (%P-MIPN and %2,6-DIPN) continues to increase with more extensive aluminum removal (Figure 3). This means that a larger fraction of the naphthalene and 2-MIPN molecules are reacting within the confines of the HM channels where a-substitution is sterically blocked. Less reaction occurs on the external surface acid sites which are non-selective. The first alkylation step is much more rapid than the second. Furthermore, since ortho-substitution of naphthalene by propylene is sterically prevented, formation of TrIPN+ products must involve or-substitution. Consequently, TrIPN+ product concentrations also decrease considerably at higher levels of dealumination. Sugi et al. made similar observations on this reaction.’ They also showed that the external-surface acid sites of HM I28 could be preferentially deactivated to improve P- substitution selectivity, while still maintaining the activity for selective substitution inside the channels. Figure 3 also shows that the ultimate attainable p- Substitution selectivity depends on the choice of HM starting material. The HM catalysts have, on average, 38% of the their total pore volume in the mesopore region (20-600 A diameter). Using XRD line-broadening, we determined the mean crystallite dimensions for HM14, HM74 and HMllO to be 0.23 f 0.02 pm, and for HM230, 0.14f 0.01 pm5 Laserscattering measurements reported by the manufacturer show that the mordenite starting materials have average particle sizes of about 10 pm. It is likely that most of the mesopore volume is in the intersticies between crystals in the catalyst particles, and dealumination increases the interstitial (mesopore) volume. The constraints of the microporous channels not only gives rise to the desired regioselective alkylations, but also impede diffusion of the desired products. If formation of the p-substituted products is diffusion limited, an increase in the mesopore volume should increase the rate of their production. Figure 4 shows that the p-substitution selectivity does, in fact, closely arallel HM mesopore volume. Lee et al. showed a similar trend for isopropylation of biphenyl.”To account for concurrent, but less pronounced, increase in micropore volumes, it has been proposed that lattice-bound aluminum is removed from the 4-membered rings that separate the 12-membered rings of the main channels h m t he neighboring 8-membered ring side-pockets.I8 Lee e t al. have suggested that propylene may preferentially diffuse through these 8-membered ring channels that run perpendicular to the main channels, and are inaccessible to naphthalene.I6 Pore volume changes do not seem to explain why the 2,6/2,7 DIPN isomer ratio increases with dealumination, considering these two isomers have nearly identical critical diameter^.^.^ Horsley et ai. used molecular graphics screening and molecular mechanics calculations to provide convincing evidence that 2,6-DIPN, with its slightly more linear structure, has a lower activation energV for dfision than 2,7-DIPN in the minochannels of HM.6 With its isopropyl groups on the same side of the molecule, steric repulsions are maximum when 2.7-DIPN diffuses into the pore windocus; whereas, diffusion of 2,6-DIPN is significantly less hindered.6 Since the channel diameter is obviously a critical parameter in determining the 2.6/2,7 DIPN ratio, we used X-ray powder diffraction to measure the changes in HM unit cell volumes caused by dealumination. These data are listed in Table 3. where the cited errors limits are the standard errors calculated by the LsQ82 program. Mordenite was the only phase detected in the patterns. HM79 seems to be an anomaly because its cell parameters are much lower than expected. The XRD pattern for catalyst is of low intensity suggesting that some structural collapse may have occurred during discussed elsewhere.6,18. 7 ’ Both series of catalysts show a severe loss in activity at high 919 its dealumination. Still, there is a general trend of unit cell contraction with dealumination, with the largest change being in the b-direction. HM71 shows the following magnitudes of contraction relative to HM14: a, 0.59%; b, 0.77%; c. 0.65%; and volume, 2.0%. As shown in Figure 5 for the six samples, excluding the anomalous HM79 data, the 2,6/2,7 DIPN ratio shows an approximate reciprocal relation to the changes in unit cell volume. A probable explanation is that HM dealumination causes a slight shrinkage in the channel diameter that results in more snug fit Of the 2,6- and 2,7.DIPN isomers in the channels. Diffusion of 2.7-DIPN becomes even more hindered than in the non-dealuminated HM case. Zsopropylotion of Biphenyl. General applicability of the dealumination procedure for improvement of regioselectivity was evaluated by examining biphenyl isopropylation over selected dealuminated mordenites. The experimental results are showdin Table 4 and in Figures 6-7. Greater than 99% of the products are isopropylbiphenyls with a small percentage of other alkylbiphenyls which are not solely isopropyl-substituted. Tetrambstitution of biphenyl is not observed. The trends in alkylation regioselectivity are remarkably similar to those observed for the naphthalene reaction. Conversion and DIPB yield increase with initial dealumination, then both decrease at higher SiO,/AI,O, ratios. Formation of 3-MIPB is not effected much by dealurnination; whereas, the concentration of 2-MIPB in the product goes to near zero. Neglecting isomers with isopropyl groups ortho to each other, there are 10 possible DlPB isomers. We observe 9 peaks in the GC-MS having m/z of 238. At present, only the three DIPB isomers listed in Table 4 can be identified with certainty. Of the remaining 7 isomers, there is the 3,5-isomer, and 6 isomers involving substitutionat the 2-position(s). Consequently, it is not so surprising that the concentrations of compounds labeled other-DIPB closely parallel the 2-MIPB concentrations. Overall, isopmpylation of biphenyl is a more eficient process than isopropylation of naphthalene. Separation of mono-, di- and polysubstituted products from each other is easy in comparison to separation of positional isomers. Dealuminated HM can give 4,4'-DIPB isomer selectivity over 80%, and an isomer ratio with the next most concentrated isomer, 3,4'-DIPB, of about 8. On the other hand, selectivity for the target 2,6-DIPN isomer is only about 60% with a 2.6/2,7 DIPN not exceeding 2.3 in this work. It should be noted that 2.6-DIPN isomer selectivity of over 65%, with 2,6/2,7 DIPN exceeding 2.6, can he achieved by adding a small amount of water to the reactor or by increasing the reaction t emp e r a t ~ r eo, r~ b y using isopropanol as the alkylating agent.3 Conclusions Dealumination of HM14 increases the 2,6-DIPN isomer selectivity from 30 to 60%. While a similar tend is observed for dealumination of HM38, lower regioselectivity is obtained. However, HM38 retains reasonable activity to higher Si0,/A1,03 ratios than HM14 does. Both factors demonstrate that performance of the dealuminated catalysts is dependent upon the choice of starting material. In comparison biphenyl isopropylation experiments, it was found that 4,4'-DIPB regioselectivity can be increased from 66 to 87% by HM dealumination. Therefore, increased selectivity for the slimmest diisopmpyl-isomer with dealumination is a general property: it occurs with different mordenite starting materials and different, but similar in size and shape, reactant molecules. Comparing regioselectivity and sorption data, we found that a higher percentage of reaction occurs in the confnes of the mordenite channels when the density of non-selective external surface acid sites is diminished by dealumination. Relative diffusion rates seem to be a major controlling factor in determining selectivity. Reducing diffusion resistance by increasing the mesopomus volume in the catalyst particle intersticies results in an increase in DIPN yield and 2,6-DI PN isomer selectivity. As reported elsewhere, 2,6-DIPN has a slightly smaller critical diameter and lower activation energy for diffusion in HM than 2,7-DIPN.3,6 We used a careful analysis of the unit cell parameters to show that the 2,6/2,7 DIPN ratio increases as the unit cell volumes decrease with aluminum removal. A prnhable explanation is that HM dealumination causes aslight shrinkage of the channel diameter, increasing the difference in diffusion rates for 2,6- and 2,7-DIPN. AcknowledeenxuLa We would like to acknowledge the encouragement and support of Prof. Harold Schohert at the Pennsylvania State University. We would also like to thank the PQ Corporation, Inc. for graciously providing the mordenite starting materials with detailed analytical data, and Prof. Deane K. Smith a t the Pennsylvania State University for his assistance in the XRD work. I. 2. 3. 4. 5. Song, C.; Schobert, H. H. Specialty Chemicals and Advanced Materials from Coals: Research Needs and Opportunities Am. Chem. SOC.Di u. Fuel Chem. frepr. 1992, 37(2), 524-532. Song, C.; Schobert, H. H. Opportunities for Developing Specialty Chemicals and Advanced Materials from Coals Fuel Process. Technol. 1993, 34, 157-196. Song, C.; Kirby. S. Shape-Selective Alkylation of Naphthalene with Isopropanol Over Mordenite Catalysts Microporous Materials 1994, 2, 467-476. Song, C.; Schobert, H. H. Non-Fuel Uses of Coals and Synthesis of Chemicals and Materials Am. Chem. SOCD.i u. Fuel Cheni. f repr . 1995, 40(2), 249-259. Schmitz. A. D.; Song, C. Shape-Selective lsopropylation of Naphthalene Over Dealuminated Mordenites Am. Chem. SOC.Di u. Fuel Chem. frepr. 1994, 39(4), 986-991. 920 ' I ' I s 1 i / I I 6. 7. 8. 9.. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. Horsley, J. A.; Fellmann, J. D.; Derouane, E. G.; Freeman, C. M. Computer-Assisted Smening of Zeolite Catalysts for the Selective lsopmpylation of Naphthalene J. Catal. 1994, hktev, A. S.; Chekriy, P. S. Alkylation of Binuclear Aromatics with Zeolite Catalysts in "&lites and Related Microporous Materials: State of the Art 1994,"S tud. surf. SC~c.a lol. Chu, S.-J.; Chen, Y.-W. lsopmpylation of Naphthalene Over p Zeolite Ind. Eng. Chem. Res. 1994, 33, 3112-3117. sugi, Y.; Kim, J.-H.; Matsuzaki, T.; Hanaoka, T.; Kubota, Y.; Tu, X.; Matsumoto, M. The IWmpylation of Naphthalene Over Cerium-Modified H-Mordenite in "Zeolites and Related Microporous Materials: State of the Art 1994," Stud. Sur/. Sci. Catal. 1994, 84, 1837-1844. sub, Y.; Matsuzaki, T.; Hanaoka, T.; Kuhota, Y.; Kim, J.-H. Shape-Selective Alkylation of Biphenyl Over Mordenites: Effects of Dealumination on Shape-Selectivity and Coke Deposition &tal. Lett. 1994, 26, 181-187. sugi, Y.; Toba, M. Shape-Selective Alkylation of Polynuclear Aromatics Catalysis Today 1994, 19, 187-212. 'h, X.; Matsumoto, M.; Matsuzaki, T.; Hanaoka, T.; Kubota, Y.; Kim, J.-H.; Sugi, Y. Calal. k t t . 1993, 21, 71-75. Moreau, P.; Finiels, A; Geneste, P.; Solofo, J. Selective lsopropylation of Naphthalene Over Zeolites J. Calal. 1992, 487-492. Katayama, A,; Toba, M.; Takeuchi, G.; Mizukami, F.; Niwa, S.-i.; Mitamura, S. Shape- Selective Synthesis of 2,6-DiisopropylnaphthaleneO ver H-Mordenite Catalyst J. Chem. SOC., Cheni. Commun. 1991, 39-40. Fellmann, J. D.; Saxton, J.; Wentrcek, P. R.; Derouane, E. G.; Massioni, P. Process for Selective Diisopmpylation of Naphthyl Compounds Using Shape Selective Acidic Crystalline Molecular Sieve Catalysts U. S. Patent No. 5,026,942,1 991. Lee, G. S.; Maj, J. J.; Rmke, S. C.; Cards, J. M. Catal. Lett. 1989, 2, 243-248. Hubbard, C. R.; Lederman, S. M.; Pyrros, N. P., "A Least Squares Unit Cell Refinement Program;" National Bureau of Standards: Washington, D.C., and JCPDS-International Centre for Diffraction Data: Swarthmore, PA, July 1983. Mishin, I. V.; Bmner, H.; Wendlandt, K.-P. Synthesis and Properties of High-Silica Zeolites with Mordenite Structure in "Catalysis on Zeolites," (Ka116, D.; Minachev, Kh. M., Eds.); H. Stillman Publishers, Inc.: Boca Raton, FL, 1988, pp 231-275. Seddon, D. The Conversion of Aromatics Over Dealuminised Mordenites Appl. Catal. 1983, 7, 327.336. 147, 231-240. 1994, 84, 1845-1851. Table 1. Sorption Data and Residual Sodium Content for Mordenite Catalysts surface area, i / g pore vo~umec, m'/g catalyst NazO,wt% total micro meso total micro meso NaMl4' 6.24 466 457 10 0.312 0.174 0.138 ~ ~ 2 1 ' 0.02 606 536 70 0.317 0.207 0.110 HM38' 0.07 512 429 82 0.293 0.167 0.126 HM14 0.19 nab'' na na na na na HM54 0.15 no na na na na na HM62 <0.01 504 413 91 0.250 0.163 0.087 HM70 <0.01 556 395 161 0.280 0.180 0.099 HM71 CO.01 572 497 75 0.419 0.191 0.125 HM74 <0.01 583 509 74 0.385 0.1% 0.148 HM79 0.14 na no na na no na HM90 <0.01 540 471 69 0.313 0.188 0.125 HMB~ 0.01 no no no no no na HMllO <0.01 539 480 59 0.362 0.184 0.138 ~ ~ 1 4 00.02~ no na no na na no HMUO <0.01 498 437 60 0.342 0.168 0.136 'Dataas reported by supplier. bNot avalble.eHM14andNaM14areassumdto have very similar sorption properties.' Produced by dealuminaton of HM38. 921 Table 2. Isopropylation of Naphthalene Test Data for HM14- and HM38-Derived Dealuminated Mordenites ""it cell paramdm q A b.d c.A volume.A' product distnbution, mol% catalyst conv.,% MIPN DIPN TrIPN+' HM14 76 63 , 32 3.6 HM54 43 75 22 1.2 HM62 78 61 34 3.9 HM70 82 52 40 5.6 HM71 74 59 37 2.3 HM74 47 75 23 0.6 HM79 69 65 29 3.6 HM90 36 79 19 0.9 HMllO 15 84 14 0.4 HMv0d 41 72 25 1.1 HM14 HM54 HM71 HM74 HM79 HMllO HM230 18.163*0.005 18.05Ii0.004 18.056*0.w6 18.091+0.009 17.947i0.016 18.075iO.006 18.047i0.008 20.314*0.w4 20.162i0.005 20.157+0.oW 20.217i0.009 19.866i0.012 20.233*0.006 20.197+0.009 7.490+0002 7.447i0.003 7.441i0.003 7.460a0.003 7.383a0.004 7.463*0.002 7.452in.003 2764*1 2710*1 2708i1 2730il 2632+2 2730il 2717i2 isomer distribution, mol% 2-MIPNb 2,6DIP$ 2,7-DIP$ 2,6'L7 60 33 19 1.76 68 50 24 2.11 54 29 16 1.78 58 44 20 2.17 64 51 22 2.29 71 55 25 2.24 59 39 21 1.86 70 53 24 2.21 83 61 30 2.05 74 58 25 2.32 HM38 73 60 34 4.4 58 39 19 1.99 HM93 84 48 43 6.5 62 48 22 2.21 HM140 38 77 21 0.4 73 56 25 2.20 'Tri-and tetraisopropylnaphthalcnes. Mole percent in MIPN products. E Mole percent in DPN products.d Calcined at 700OC. Table 4. Isopropylation of Biphenyl Test Data for Selected Dealuminated H-Mordenites product MIPB isomer DIPB isomer distribution, ml% distribution. nul% distribution. ml% catalyst conv.,% MIPB DIPB TrIPB 2- 3- 4- 3.3'- 3,4'- 4,4'- other HM14 49 74 25 0.5 9.2 24 66 3.9 17 66 13.8 HM21 60 64 33 1.5 8.1 26 66 3.0 I5 72 10.8 HM38 71 54 42 3.0 10.2 28 62 2.7 13 72 12.2 HM71 46 62 37 1.1 3.1 24 73 1.3 I1 83 4.9 HM230 23 67 32 0.5 2.0 I8 80 0.9 10 87 2.5 101 . , , , , , , 0 50 100 150 200 250 SiOdAlrh (molar) Figure 1. Naphthalene conversions for HM14-derived (solid h e ) and HM38-derived (broken line) catalysts as a function nfSiO,/N20, ratio. 922 ,f I I I f .-- 0 0 50 LOO IS0 200 2 0 SiOJAlpO, (molar) Figure 2. Naphthalene isopropylation product distributions for HM 14-derived (solid lines) and HM3B-derived (bra-, ken Lines) catalysts as a function of SiOz/Al,O, ratio. I 0 1 . , , , , , , . . 0 so 100 I50 200 210 Si%/Al& (molar) Figure 3. Naphthalene isopmpylation isomer distributions for HM 14-derived (solid lines) and HM38-derived (broken lines) catalysts as a function of SiO,/AI,O, ratio. Miaopors Volrmc = I IO0 80 - -s P v 60 n 9 40 .E Le! Ma- V o b s 0.08 IW IS0 zoo 210 SiQ/Al,O~ (molar) Figure 4. Comparison of pore volumes (solid lines) and P-substitution selectivities (broken lines) for dealuminated HM14 catalysts as a function of SiOz/AI,O, ratio. 923 I , 2.4 2700 I . , . I . , . , . I 1.4 0 so 100 150 200 250 SIWAIO (molar) Figure 5. Comparison of unit cell volumes (solid line, left axis) with 2,6/2,7 DIPN ratios (broken line, right axis) for dealuminated HM14 catalysts as a function of SiO.JAl,O, ratio. Error bars represent four-times the standard error in unit cell volumes from Table 3. 0 JO 100 1 IO 2w SiOdAllO, (molar) Figure 6. Isopropylation of biphenyl conversion and product distribution as a function of SiO,/AI,O, ratio. 2-MIPB 0 so 100 I so 200 Sio2/AI2Q (molar) Figure 7. Biphenyl isopropylation isomer distributions as a function of SiO,/AI,O, ratio. 924 i I i SYNTHESIS OF POLYESTERS WITH RIGID BIPHENYL SKELETON BY CARBONYLATION-POLYCONDENSATIOWNI TH PALLADIUM-PHOSPHINE CATALYSTS Y.Kubota K.Takeuchi, T.Hanaoka, and LSugi National Institute of Materials &d Chemical Research, AIST, Tsukuba, lbaraki 305, Japan Keywords: polyester, carbonylation-polycondensation,D BU INTRODUCTION Stiff macromolecules are expected, when properly processed, to produce materials with high dFg%q of molecular orientation and order which should result in superior mechanical strength [1,2]. BI henyl derivatives are promising components for advanced materials such as heat-resistant porymers and liquid crystallme polymers. &polyesters based on terephthalic acid, isophthalic acid, and bisphenol A have already been used because they are heat-resistant and transparent. Wholly aromatic polyesters containing biphenyl-4,4'-dicarboxylate moiety in place of terephthalate and isophthalate moiety must be highly potential for beat-resistance. For this reason, we tried to synthesize biphenyl-containing polyesters. Among several synthetic methods, carbonylationgob'condensation method originally developed by Imai and his co-workers [3], and subsequently y Perry and his co-workers [4] on the basis of Heck reaction [5] seemed to be the most straightforward way to get target polyesters. We report herein the successful synthesis of polyesters which contain rigid biphenyl skeleton by palladium-catalyzed carbonylation-polycondensation. Especially, introduction of 9,lOdihydrophenanthrene moieties was found to afford highly soluble polyesters in organic solvent. This is advantageous for polyester formation by carbonylation-polycondensationin solution and for molding resulting polyesters. EXPERIMENTAL 4,4'-Dibromobiphenyl (DBBP) and 4,4'-diiodobiphenyl (DIBP) were obtained from Aldrich Japan, Tokyo, Japan, and purified by recrystallization from toluene. 2,7-Dibromo- 9,lO-dihydrophenanthrene (DBDHP) and 2,7-diiodo-9,10-dihydro henanthrene (DIDHP) were prepared from 9,lO-dihydrophenanthrene by known methods [6,7!. All other materials were obtained commercially, and used with appropriate purifications. The wei ht average molecular weight (Mwa)n d the number average molecular weight (M") were determinetby means of gel permeation chromatography on the basis of a polystyrene calibration on a Yokogawa HPLC Model LClOO System (column, Tosoh TSK-Gel G4000HHR; eluent, chloroform or chloroform/l,l,l,3,3,3-hexafluoro-2-propanol (HFIP) = 3/1 (v/v); detection, UV (wavelength: 254 nm)). Thermal characteristics were studied with a Mettler FP800 Thermal Analysis System and a MAC Science TG-DTA 2000 apparatus. In a 50 ml stainless steel autoclave equipped with a magnetic stirrer was placed 845.1 mg (2.5 mmol) of DBDHP, 570.7 mg (2.5 mmol) of bisphenol A, 17.7 mg (0.1 mmol) of PdCIz, 104.9 mg (0.4 mmol) of PPh3, 10 ml of chlorobenzene, and 0.82 ml (5.5 mmol) of 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU). Carbon monoxide was introduced at 1.1 MPa of an initial pressure and then heated with vigorous stirring at 130 "C in an oil bath for 3 h. After excess carbon monoxide was purged, reaction mixture was poured into 100 ml of methanol. Precipitated polymer was separated from methanol by decantation, dissolved in 50 ml of chloroform, and then poured into 100 ml of methanol again with stirring. Polymer was filtered, washed with 100 ml of methanol, and dried in vacuo to afford poly[oxy-1,4- phenylene(l-methylethylidene)-1,4-phenyleneoxycarbonyl(9,lO-dihydro-2,7-phenanthrenediyl-) carbonyl] (1) as white or pale green yellow solid. The yield was 1.09 g (95%). Mw and Mw/Mn determined by GPC were 102,600 and 2.5, respectively. The 10 % weight loss temperature (Tio) was 414 "C and the 5 % weight loss temperature (Ts) was 394 "C in ai[ Found: C, 79.93; H, 5.09; Br, 1.35 %. Calcd. for (C31Ha)n: C, 80.85; H, 5.25 %. IR and 'H and 'C-NMR spectral data were satisfactory for polyester 1. Materials. Molecular weight measurements. Typical procedure for carbonylation-polycondensation. The other polyesters were obtained by analogous procedures. RESULTS AND DISCUSSION Effect of reaction parameters on the carbonylation-polycondensation The synthesis of polyesters by the carbonylation-polycondensation is shown in Eq. 1. To make clear factors controlling the synthesis, parameters such as temperature, CO pressure, solvent, and base were studied for polyester 1 from DBDHP and bispbenol A. Figure 1 summarizes the effect of temperature on the molecular weight of 1 by the use of pph:, as a ligand and DBU as a base. The carbonyldtion-pol~condensatiown as dependent on temperdturc, and the molecular weight was the highest at 120-130 C. Low molecular weight is due to low reaction rate at low temperatures, and to side reactions at high temperatures. Effects of solvent on the molecular weight of 1 by the use of 1,3-bis(diphenylpbosphino) propane (dppp) as a ligand and DBU as a base are shown in Table 2. Although the alcoholysis of acyl-palladium jntermediate is expected to be favorable in polar solvent, the solubilit of product is more important factor than the polarity of solvent in our case. Polyester 1 was founito be easily dissolved in chlorobenzene, nitrobenzene, dichloromethane, and chloroform. Among them, only chlorobenzene mediated effectively the carbonylation-polycondensation. The efficiency of may be due to good solubility of polyesters in it. Table 2 summarizes the effect of base on the molecular weight of 1. The highest molecular weight was obtained for DBU. DBU was under wide variety of conditions because it is a strong base (px, = 11.5 ). Because the salt from DBU and HBr liberated by the carbonylation was highly soluble in organic solvents, it is easier to remove HBr than the salt of other bases such as EtsN and i-Pr2NEt. Under optimum 925 co Base + HO-R-OH PdCIZ- Phosphlnz DBDHP condition, 1 was obtained in 95% yield with high molecular weight (Mw =l.0x105) as described in experimental part. Catalytic activity of palladium-phosphine complex on the carbonylation-polycondensation is important for the molecular weight of polyesters. The effect of phosphine on the molecular weight of 1 is summarized in Table 3. The use of four moles of PPh per mole of palladium was necessary to prevent catalyst decomposition, probably by a cluster formation [8]. However, large excess of PPh3 (PPhpd = 10) inhibited polymer formation because ex= ligands coordinated to metal center reduce the coordination of substrate and carbon monoxide. This is a different feature from the case of the ethoxycarbonylation of DBBP, where catalytic activity kept high even at high PPh3lPd ratio [9]. Bidentate phosphine such as dppp has been described to be more effective ligand for the carbonylation than monodentate hosphine such as PPh3 [9-121. In our previous work 91, high catalytic activities were observecf for hidentate phosphines, a, &bis(diphenyIphosphinojalkanes ( P ~ ~ P ( C H Z ) ~(Pn=P2~-5Z) ), cspecially for dppp (n=3) and dppb (n=4) in palladium catalyzed ethoxycarbonylation of DBBP and DBDHP. The catalysts with these li ands have been much more active than those with PPhs. The effectiveness of bidentate ligan$ is due to the formation of chelated complex with palladium, such as six-membered chelate ring for dppp. In the carbonylationpolycondensation, dppp gave the highest molecular wcight of 1 among them. Two fold excess of dppp for palladium was also required for the carbonylation and for high molecular weight, however, large excess of dppp prevented them. Figure 2 shows the effect of CO pressure on the molecular weight of 1 for PPhs and dppp. The molecular weight was the highest under CO pressure around 1 MPa, and then gradually decreased with further incrcase of CO pressure. Both PPh3 and dppp were effective ligands in the carbonylation-polycondensation, especially under relatively low pressure of 0.6-1.5 MPa. However, PPh3 was more excellent than dppp for the formation of high molecular weight polyester 1 under these conditions. The effect of CO pressure on the molecular weight of 1 for DIDHP was different from the case for DBDHP as shown in Fig. 2. The molecular weight of 1 for DIDHP was low under every pressure. The structure of dihalide and bisphenol affected the carbonylation-polycondensation. Table 4 shows the effect of dihalide on the molecular weight of 1 by the use of dppp and DBU. Polyester 1 from DBDHP or DIDHP has hi er solubility than poly[l,4-phenylene(l-meth I ethy1idene)-1,4- phenylenecarbonyl(4,4'-bipheny~ne)carbonyl] (8) from DBBP or DIBP anB accompanied the enhancement of its molecular weight. Such an increase of solubility is owing to the effect of bulkiess of 9,lOdihydrophenanthrene moiety. The molecular weight of 1 from DBDHP was higher than that from DIDHP. The difference between bromide and iodide should be explained by the difference of rate determining steps in successive reactions. Significant increase of the molecular weight of 8 was observed by the use of DIBP instead of DBBP. This was due to the enhanced reactivity of DIBP, which was compensated for negative factor of insolubility. When pdibrornobenzene was used as a dihalide, polyester 9 was obtained with low molecular weight due to the low solubility of the resulting polyester. Table 5 summarizes the effect of bisphenol on the molecular weight of polyesters. Bisphenols having sufficiently bulky alkyl spacers resulted moderately high molecular wcight. Especially, bisphenol A gave the best results among them. Bisphenols with phenyl groups gave relatively poor results. Polyester 6 from bis(4-h ydroxypheny1)sulfone had low molecular weight. Polyester 7 from 4,4'-thiobisphenol was insoluble In the solvent. These low molecular weight is due to their low solubility. Mechanistic aspects of the carbonylation-polycondensation According to previous paper by Moser and his co-workers [8 active species for the carbonylation of aromatic halide is expected to be Pd(0)Ln complex (L: pkosphine moiety; n: 1-4), which is formed in situ from Pd@I)CIz and phosphine or from Pd(II)LzC12 complex with phosphine. Oxidative addition of aryl halide to Pd(0)Lo species, followed by CO insertion, and base mediated the alcoholysis to yield ester with regeneration of Pd(0)Ln. This mechanism is plausible and applicable to the carbonylation-polycondensation. We found several characteristic features in our carbonylation-polycondensation. The molecular weight of 1 depended on the type of base. Strong organic bases such as DBU, TMEDA, and DABCO preferred to typical amines such as Et3N and i-PnNEt. These results suggest the alcoholysis step is a key step for the increase of molecular weight. Similar effects of basc wcre obsewed in the carbonylation of 4-bromobiphenyl [12]. The molecular weight of 1 for DIDHP was inferior to that for DBDHP. The effects of CO pressure on the molecular weight of 1 for DIDHP were quite different from the case for DBDHP. These results mean that the oxidative addition of halide to Pd(0)Ln is not so important for the increase of the molecular weight of resulting polyesters. Figure 3 shows time dependence curves on CO consumption hy the use of d pp aid PPh3, respectively, under atmospheric pressure. The periods, for which half amount of carion monoxide was consumed for the reaction, were 16 min and 32 min in the case of dppp and of PPh3, 926 ' I respectively. This means that catalytic activity for d pp is apparently higher than that for PPh3. Figure 4 shows h e de pendence on the Mwb y use of $pp and PPh, respectively, under the same condition as in Fig. 3. Initial rate of the increase of Mw for dppp is obviously larger than that for PPh3, whereas final Mw for PPhs is higher than that for dppp. These results mean that catalytic activity is not parallel to final molecular weight of resulting polyesters, and that some inhibiting reactions n q occur during the growth of the polyester in the case of dppp. We should note the side reactions because the rate of polyester formatlon should not affect the final molecular weight of polyester. To know whether side reactions take place or not, the phenoxycarbonylation of DBBP using the catalyst with dppp or PPh3 was examined. Diphenyl biphenyl-4,4'-dicarboxylate was observed as a sole product from DBBP in high yield in both cases. Only a detectable product except the esters h.om DBBP was phenyl benzoate judging from GC analysis. This product should form via the carbonylation of chlorobenzene used as a reaction solvent. Amount of phenyl benzoate was 0.115 mmol with dppp and 0.044 mmol with PPh3, respectively, from phenol (6.0 mmol) and chlorobenzene (10 ml) in the presence of DBBP (2.5 mmol). The amount of phenyl benzoate arising during the &onylation-polycondensation with dppp was about three times larger than that with PPhs. It is evident that the benzoyl complex as an intermediate of the carbonylation of chlorobenzene can act as a terminator of the carbonylation-polycondensation. This is a possible reason why the molecular weight of 1 with dppp is lower than that with PPhs in spite of higher catalytic activity of palladium-dppp complex. These results show that in order to obtain the polyesters with high molecular weight, the selectivity of the catalyst for the carbonylation is more important factor than catalytic activity. Thermal properties of polyesters Wholly aromatic polyesters 1- 9 did not have melting point and glass transition temperature up to 400 "C. The 10 %weight loss temperature (no) values of them were above 380 "C in air. On the basis of TG profiles, polyesters 1 and 8 were more stable than 9 at 300-390 "C, although they lost their weights faster than 9 above 400 "C. Polyesters 1 and 5 were soluble in chloroform and dichloromethane. Soluble polyesters with MW larger than 10,000 easily formed transparent casting films. REFERENCES 1 2 3 J . 4 . Jin, S. Antoun, C. Ober, and R. W. Lmz, Brit. Polym. J., 12 (1990) 132. M. Ballauff, Angew. Chem. Int. Ed. Engl., 28 (1989) 253. M. Yoneyama, M. Kakimoto, and Y. Imai, Macromolecules, 21 (1988) 1908; 22 (1989) 2593.4152. I 4 R.J. PeiyyS.R. Turner, R.W. Blevins, Macromolecules, 26 (1993) 1509. 5 R.F. Heck, Adv. Catal., 26 (1977) 323. 6 D.E. Pearson, US. Patent, 3988369 (1976). 7 Von H.O. Wirth, K.H. Gonner, and W. Kern, Makrornol. Chem., 63 (1963) 53. Gakkaishi, 37 (1994) 70. 8) W.R. Moser, A.W. Wang, N.K. Kildahl, I. Am. Chem. Soc., 110 (1988) 2816. 9) Y. Sugi, K. Takeuchi, T. Hanaoka, T. Matsuzaki, S. Takagi, and Y. Doi, Sekiyu 10 Y. Ben-David, M. Portnoy, D. Milstein, J. Am. Chem. SOC., 111 (1989) 8742. 11 R.E. Dolle, S.J. Schmidt, L.1. Kruse, J. Chem. SOC., Chem. Commun., (1987) 904. 12 I Y.Kubota, T.Hanaoka, K.Takeuchi, and Y. Sugi, Synlett., (1994) 515. i Table 1. Effects of solvent on the synthesis of 1') Yield Mw Run Solvent (%) (x103) (M~IM") 1 Chlorobenzene 99 2 Anisole 99 3 Flurobenzene 98 4 Benezene 94 5 Nitrobenzene 0 a) Reaction conditions: DBDHP 2.5 mmol, bisphenol A 2.5 mmol, PdClz 0.1 mmol, dp p 0 2 mmol, DBU 5.5mmol, solvent 10 ml, CO pressure 2.1 MPa, temperture 12OoC, periocf3 h.' Table 2. Effects of base on the synthesis of 1') Yield Mw Run Base (%) (x103) ( ~ ~ l ~ n ) 92 16 8 DABCO") 96b) 12 9 EON 92b) 6 DBU 95 b' 7 TMEDA~) 99 10 'i-P&t 22 a) Reaction conditions: DBDHP 2.5 mmol, bisphenol A 2.5 mmol, PdCh 0.1 mmol, PPh3 0.4 mol. DBU 5.5 mmol, chlorobenzene 10 ml. CO Dresswe 2.1 MPa. temoerture 120T neriod .. ., -1 r - - - - - 3 h. b) dppp 0.2 mmol, &XN;N'-tetramethylethylene diamine. e) 1,4-Diazabicyclo[2.2.2]octane. was used as a ligan'd. c) Palladium preci$tate was-observed. d) I 927 Table 3. Effects of catalyst on the synthesis of la) Yield Mw Run Catalyst (%) (x103) (M~IM.) 11 PdC12nPPh3 39 12 PdCld4PPe 92 13 Pd(PPh3)4 96 19 14 PdC12ndppeC) 94 15 PdClddpppd) 97 16 PdCld2dppp 99 17 PdCWdppg) 39 16 PdClfldppb 95 19 PdClSdpppe‘) 91 a) Reaction conditions: DBDHP 2.5 mmol, bisphenol A 2.5 mmol, PdClz 0.1 mmol, phosphine 0.1-0.5 mmol, DBU 5.5 mmol, chlorobenzene 10 ml, CO pressure 2.1 MPa, temperture 120 “C, period 3 h. b) Pd(PPh3)4,0.1 m o l . c) 1,2-bis(diphenylphosphino)ethane. d) 1,3-bis(diphcnylphosphino) propme. e) 1,4-bis(diphenylphosphino)hutane. f ) 1,5-bis(di-phenylphosphino)hexane. Table 4. Polyesters of dihalides with bisphenol A’) Yield MW Run Dihalobiphenyl Polyester (%) (x103) (MW/Mn) 99 35 98 24 20 DBDHP 1 {% 21 DIDHP 1 22 DBBP 8 94 5.6 (6.3j 2.6 [;:;{ 23 DIBP 8 24 p-Dibromobenzene 9 96 94 12 a) Reaction conditions: dihalide 2.5 mmol, bisphenol A 2.5 mmol, PdCIz 0.1 mmol, dppp 0.2 mmol, DBU 5.5 m o l , chlorobenzene 10 ml, CO pressure 2.1 MPa, temperture 120 “C, period 3 h. Table 5. Polyesters of bisphenols with DBDHPa) Yield Mw Run Bisphenol Polyester (%) (x103) (MwIMn) Remarks 26 Bisphenol A 1 99 35 27 4,4’-Cyclohexylidencbisphe I 2 93 20 26 4,4’-(~-Butylidene)bispheno;8j 3 97 13 29 4,4’-Ethylidenebisphenol 4 96 12 30 4,4’-(l-Phenylethylidene)bis henol 5 99 25 31 4,4’-SulfonylbisphenoP 6 32 4,4’-Tiobisphenol I 99 insoluble tt a) Reaction conditions: DBHDP 2.5 mmol, bisphenol 2.5 mmol, PdCh 0.1 mmol, dppp 0.4 mol, DBU 5.5 mmol, chlorobenezene 10 ml, CO pressure 2.1 MPa, temperture 120 “C, period 3-5 h. b) CO pressure 1.1 MPa. c) Ourtlook of reation mixture just after the reaction. -: homogensous. t: slightly suspended. tt: suspended. 12 l!5- 3 0 X F v 4 0 90 110 130 150 Reaction temperature (“C) Fig. 1. Effect of reaction temperature on the molecular weight of 1. Reaction conditions: DBDHP 2.5 m o l , bisphenol A 2.5 mmol, PdClz 0.1 mmol, PPh3 0.2mmo1, DBU 5.5 mmol, chlorohenzene 10 ml, CO pressure 1 .I MPa. period 3 h. 928 i 10 DBDHP-PPh3 0 DBDHP-dppp DIDHP-PPh3 2 - ' 0 0 1 2 3 CO pressure (MPa) Fig. 2. Effect of CO pressure on the molecular weight of 1. Reaction conditions: dihalide 2.5 m o l , bisphenol A 2.5 mmol, PdCIz 0.1 mmol, PPh3 02mmol or dppp 0.1 mmol, DBU 5.5 mmol, chlorobenzene 10 ml, temperature 120 "C, period 3 h. 0 30 60 90 120 Reaction period (min) Fig. 3. CO consumption during the carbonylation-polycondensation.R eaction conditions: DBDHP 2.5 mmol, bisphenol A 2.5 mmol, PdCIz 0.1 mmol, PPh3 0.4 mmol or dppp 0.2 mmol, DBU 5.5 mmol, chlorobenzene 10 ml, CO pressure 0.1 MPa. 10 1 a 0 X -c 3 PPh3 - 1 n - 0 dPPP 0 5 10 15 20 25 Reaction period (h) Fig. 4. Time dependence of the molecular weight on the s nthesis of 1 Reaction conditions: DBDHP 2.5 mmol, bisphenol A 2.5 mmol, PdClz 0.1 mmoi: PPh3 0.4 mmol or dppp 0.2 mmol, DBU 5.5 mmol, chlorobenzene 10 ml, CO pressure O.1MPa. 929 SHAPESELECTIVE HYDROGENATION OF NAPHTHALENE OVER ZEOLITESUPPORTED Pt AND Pd CATALYSTS Andrew D. S c u G rainne Bowers and Chunshan Song Department of Materials Science and Engineering Fuel Science Program, 209 Academic Projects Building Pennsylvania State University, University Park, PA 16802 Keywords: hydrogenation, bifunctional catalysts, shape-selectivity Per-hydrogenation of naphthalene produces both cis-decalin (c-DeHN) and trans-decalin (t- DeHN). Huang and Kang reported the rate data shown in Scheme 1 for this reaction catalyzed by Pt/AI,O,.' Isomerization of c-DeHN was treated as irreversible, and it was assumed that dehydrogenation of DeHN could be neglected. We have found it possible to achieve high selectivity 1 I Scheme 1. Naphthalene hydrogenation pathways over Pt/AI,O, from ref. 1. Rate constants are for reaction at 200 "C in units h-'. for one DeHN isomer by appropriate catalyst selection. For example, PtlHY gives 100% naphthalene conversion to decalins with 80% selectivity for c-DeHN. There numerous potential industrial applications for c-DeHN; such as, the production of sebacic acid which can be used in the manufacture of Nylon 6,10 and softeners. Conversely, catalysts that promote the thermody- ~mi c a l l yfa vored c-DeHN to t-DeHN isomerization can be made to give nearly 95% t-DeHN. This reaction can he used in fuel upgrading applications, to increase the thermal stability of the fuel. Considerable effort has been invested in this laboratory to develop jet fuels with improved thermal stability, particularly for high-performance jet aircraft (see ref. 2, for example). In this application, the fuel is also used as the primary heat sink for cooling. As the fuel temperature is raised, fuel degradation leads to the formation of solid particulates in the fuel.' Over time, the particulates agglomerate and are deposited, plugging filters, fuel lines and fuel injectors. A jet fuel's overall carbon-forming propensity can be reduced by limiting its aromatic content as in the oomplete hydrogenation of naphthalene. In thermal stressing of jet fuels, Song et al. have shown that cyclic alkanes have higher thermal stability than normal alkanes.'-3 Cycloalkane conformation also effects high temperature stability, and it has been shown that t-DeHN is more stable than c-D~HN.'.~T ests on the thermal stressing of petroleum-derived fuels and model compounds have shown that addition of t-DeHN can significantly retard the rate of carbon dep~sition.~ Two types of zeolites were used in this work: H-mordenite (HW and HY. Mordenite has a two-dimensional channel structure with elliptically shaped channels of diameter 6.7 x 7.0 A. Since naphthalene's critical diameter is very close to the HM channel dimension, transition state selectivity is induced on reactions of naphthalene occurring in the channels. Relative diffusion rates of the products can also effect selectivity. HY has large cavities in its interior and narrow channel openings. A reactant that has entered the channel structure may reorient itself and react at catalyst sites on the walls of the cavities. However, because of restricted difision at the channel openings, only molecules ofappmpriate diameter will be produced at an appreciable rate. The final t-DeHhVc-DeHN (t/c) ratio in the product is governed by several factors related to the bifunctionality of the catalysts. While the initial tlc ratio may be governed by several factors, zeolite acid character significantly influences c-DeHN isomerization. It has been found that catalysts based on dealuminated HM give the highest t-DeHN selectivity. Choice of the noble metal, Pt or Pd, is also important. In a w r d with previously reported data for naphthalene hydrogenation using noble metal catalysts on non-zeolite ~ u p p o r t so,u~r data show that Pdlzeolitc has higher initial selectivity for t-DeHN, and also isomerize c-DeHN at higher rates than platinum on the same zeolite. Metzl particle sizes determined from X-ray powder diffraction (XRD) linewidths show large variations on the different zeolites, Cotolyst Prepmotion. The zeolites were supplied in NH4-form and used as received. Table 1 lists their properties. Two portions of each zeolite were loaded with metal, one with Pt, the other Pd, to generate a total of eight catalysts. Incipient wetness impregnation of either aqueous H,PtCI,+H,O (Aldrich 99.995% Pt, metal basis) or PdCI, (Aldrich, 99.999% Pd, metal basis) dissolved in dilute hydrochloric acid (sufficient to form soluble PdCI,'.) was used to achieve a nominal metal conoentralion of6 wt%. Following drying in vacuo, the catalysts were calcined in 930 air at 460 "C for 2 h. Noble metal reduction occurs during the catalyst test, in the hydrogen Pressurized reactors. Catalyst Evaluation. A 30 mL, stainless steel tubingbomb batch reactor was used for catalyst tests. A tee-shaped design was used where most of the reactor internal volume is in the horizontal member that contains the catalyst and reactants. The horizontal member is connected by a 10" length of 114" 0.d. tubing to a pressure gauge and valve. The reactor was charged with 0.4 g catalyst, 1.0 g (7.8 mmol) naphthalene (Aldrich, 99%), 4.0 g n-tridecane reaction solvent, and 0.35 g n-nonane internal standard. The charged reactor was flushed with H, then sealed and leak-tested with H,. Finally, the hydrogen pressure was adjusted to 1500 psig cold ( ca. 0.2 g) to Start the test, Naphthalene begins to react immediately, even at room temperature, so a consistent procedure was established to minimize the time between reactor pressurization and the start of the run. The reactor was affixed to a holder and placed in a fluidized sand-bath heater so that approximately two-thirds of the total length was immersed. Vertical agitation at 240 cyclehin was used to provide mixing. The reactions were done at 200 "C for 6-60 min. At the end of each test, the reaction was quenched in cold water. After cooling, the gas headspace was collected for analysis and the reactor was opened, Acetone was used to wash the reactor contents onto a filter and the filtrate was analyzed by GC/GC-MS (30m x 0.25mm DB-17 column, J&W Scientific), while the solid was dried for XRD examination. X-ray powder diffraction analyses (XRD) were done on a Scintag 3100 diffractometer using Cu Ka radiation and a scan rate of 1" 20 /min with 0.02" steps. Diffraction line widths were measured using a profile-fitting program which assumes a peak shape intermediate between Gaussian and Cauchy. Manual measurements were always used to check the calculated results, especially for very diffise lines where the computer routine fails. Mean metal crystallite size was calculated by application of the Scherrer equation (wavelength 1.54056 A, Scherrer constant 0.89).6SiIicon powder (-325 mesh) was used as an external standard for measuring instrumental and spectral broadening. Ka,-doublet broadening corrections, and pure line profile determination for low-angle reflections were done as described elsewhere.6 For most of the catalysts, only the Pt or Pd (111) diffraction line was suitable for profile analysis. The lower intensity lines of the metals were interfered with by zeolite patterns. Results and Discussion XRD Observations. XRD for the Pt and Pd catalysts removed from the reactors following the 60 min runs are shown in Figures 1-2. In each case, only diffraction lines corresponding to the zem-valency state of the metals are observed. Several catalysts from the 30 min runs were also examined, and only metallic phase is observed. Therefore, in situ hydrogen treatment is adequate for complete metal reduction. Differences between the samples are striking, especially for the Pt catalysts. WHY shows very sharp and intense metal diffraction lines (large Pt particles), but F'tfHM38 and Pt/HM17 show very broad, diffuse lines indicative of small Pt particles. PtlM21 is intermediate. Closer examination of PtlHM38 shows that the Pt-phase is bidisperse. A sharp line appears superimposed over a very broad band, both arising from reduced platinum. Pd catalysts (Figure 1) all show significant line-broadening. The trend in line width increase on going from metaUHY to m e t a m 1 7 is also observed for Pd, but is less pronounced. Average metal particle sizes for the Pt and Pd catalysts are compared in Table 2 and Table 3, respectively, Acnvacy of the XRD particle size technique is generally accepted to be i 10.20%. When XRD lines become very broad and diffuse, accurate line-width measurements are difficult, however, This was more of a problem for Pd which has a lower scattering power (z = 46) than Pt (z = 78). Baseline zeolite signals from HM38 and HM17 cloud the Pd (11 1) reflection enough that accurate measurements are not possible. Conservatively narrow best-guesses at the line widths were used to determine the values cited in the tables. Effects of Catalyst Composition. Test data for 60 min runs at 200 "C for the eight Pt- and Pd-loaded zeolite catalysts are compared in Table 2 and Table 3, respectively. The products of Nap hydrogenation are almost exclusively isomeric DeHNs. In some cases, a small amount of tetrahydmnaphthalene CTeHN) is observed. Gas headspace analyses show 5-50 ppm levels of C,. C, hydrocarbons. Every catalyst gives ca. 100% Nap conversion in 1 h, so it is not possible to rank catalyst activity based on these data. Yet, the translcis DeHN ratio is highly dependent on both the zeolite and the metal. Palladium gives higher t-DeHN selectivity than platinum on the same zeolite. Catalysts based on HM38 gave the highest t-DeHN selectivity. There is a definite upward trend in t/c ratio wlth HM S10dAl2O3 ratio. PtfHY shows amazingly high selectivity for c-DeHN. Comparing catalyst with the same metal, there does not seem to be any correlation between metal crystallite size and DeHN isomer selectivity. Neither naphthalene hydrogenation nor c- DeHN isomerization involve C-C 0-bond breaking. Consequently, the overall reaction should be structure insensitive and independent of metal particle size. Effecfs ofRun Duration. In order to determine the practical equilibrium tfc ratio, four tests were done with P M 2 I at extended reaction times (Table 4). An approximately constant t/c of 13.6 is obtained within 6 h reaction time. This value is somewhat lower than the calculated equilibrium constant for c-DeHN to t-DeHN isomerization of 20.5 at 200 "C6 However, ca. 14 is the practical limit as conCmed using other Wzeolite and Pdheolite catalyst^.^ Some decalin may reside in the portion of the reactor that extends above the sand level in the fluidized sand bath, the cold zone. and may not react. According to calculations, if even 5% of the decalin doesn't react to form an equilibrium amount oft-DeHN, the equilibrium constant value falls to 13.0.6 . . . 931 I t was already known that complete naphthalene conversion to a mixture of decalins occurred within the first hour of each test. We wanted to find out if the initial product distribution was signiiicantly different than what we had observed in 60 min runs. Additional runs were done with m 3 8 , Pd/HM38, PtHY and Pd/HY at 15 and 30 min. The metal/HY catalysts were also tested at 6 min. It should be noted that 6 min is the approximate time required for the interior of the reactor to equilibrate to the reaction (sand bath) temperature. Greater than 99% naphthalene conversion occurred within the shortest run period for each test. Decalins were the only hydrogenation products with the exception of the 6 min run with Pt/HY, where 48% TeHN was also observed. However, less than 3 % TeHN was observed in the longer runs with PtlHY. Plots of log(c-DeHN) vs time are linear (Figure 3) showing that the isomerization of c-DeHN is firstorder. Considering Scheme 1, when all of the TeHN has been consumed, the rate expression for disappearance of c-DeHN simplifies to a simple first-order expression in c-DeHN concentration. This simplification does not hold when the TeHN concentration is non-zero, so the 6 min data point for PdHY is not included in the determination of k,. Values determined for the rate constant in this work are compared with literature values in Table 5. It can be seen that not only does Pd have a higher initial selectivity for t-DeHN, but it also isomerizes t-DeHN faster than Pt. The values of k, determined in this work are considerably higher than those reported by Huang and Kang,' and h i and Song for direct isomeriaation of c-DeHN.' Huang and Kang did not report the mass ofcatalyst used, so it is not possible to compare values on a per-gram catalyst basis. h i and Song used the same PtIHM38 catalyst that was used here. We are still not certain what causes the discrepancy between these data and the results of hi and Son Other detailed kinetic studies on hydrogenation of aromatics have recently been The PtlHY catalyst is highly selective for c-DeHN and does not promote the isomerization. We are unable to explain this result at present but suspect that it may be due, in some way, to the large Pt particle size (1700 A). It is possible that a unique type of shape-selectivity may arise from partial blockage of the channel openings by the large metal particles. Further understanding may be gained by electron microscopy and H, chemisorption to determine metal dispersions. It has been established that the naphthalene hydrogenation process can be tailored to produce either c-DeHN or t-DeHN by appropriate choice of the zeolite and metal species. Selectivity for t-DeHN increases with SiO.$U,O, ratio in the HM catalysts, so that catalysts bascd on HM38 gave the highest t-DeHN selectivity. Compared to Pt on a given zeolite, Pd shows a higher initial selectivity for t.DeHN, and a higher rate for c-DeHN to t-DeHN isomerization. The practical equilibrium dc ratio is 14 at 200 "C. Metal crystallite sizes are highly dependent on the zeolite. Pd generally had a higher dispersion than did platinum on a given zeolite. Naphthalene hydrogenation and c-DeHN isomerization are structure insensitive reactions. Therefore, DeHN isomer selectivity does not show a correlation with particle size. Uniquely high selectivity for c-DeHN has been obtain with P W . It has been proposed that partial channel blockage by the large metal particles of this catalysts give rise to a unique type of shape.selectivity. We wish to thank the following persons at the Pennsylvania State University: Prof. Harold Schobert for his encouragement and support, and W.-C. h i for his thoughtful comments on this work. This work was jointly supported by the US. Dept. of Energy, Pittsburgh Energy Technology Center, and the Air Force Wright Propulsion Laboratory. We would also like to thank Mr. W. E. Harrison 111 of USAF and Dr. S. Rogers of DOE for their support. 1. Huang, T.-C.; Kang, B.-C. Kinetic Study of Naphthalene Hydrogenation Over Pt/AI,O, Catalyst Ind. Eng. Chern. Res. 1995, 34, 1140-1148. The authors neglected to divide the slopes from activation energy plots by the gas constant (1.987 caUmol'Q for reporting activation energies. The data cited in Scheme 1 are the correct values. 2. Song, C.; Eser, S.; Schobert, H. H.; Hatcher, P. G. Pyrolytic Degradation Studies of a Coal- Derived and a Petroleum-Derived Aviation Fuel Energy & Fuels, 1993, 7, 234-243. 3. Song, C.; h i , W:C.; Schobert, H. H. Hydrogen-Transferring Pyrolysis of Cyclic and Straight- Chain Hydrocarbons. Enhancing High Temperature Thermal Stability of Aviation Jet Fucls by H-Donors Am. Chern. SOCD.i u. Fuel Chern. Prepr. 1992, 37, 1655. Weitkamp. A. W. Stcreocheinistry and Mechanism of Hydrogenation of Naphthalenes on Transition Metal Catalysts and Conformational Analysis of the Products Adu. Calol. 1968, Klug, H. P.; Alexander, L. E., "X-ray Diffraction Procedures for Polyc~s tal l inean d Amorphous Materials;" Wiley: New York, 1974. See example calculations on p 699. Lai, W.-C.; Song, C. Zeolite Catalyzed Conformational Isomerization of cis-Decahydronaphthalene. Reaction Pathways and Kinetics Am. Chern. SOC. Diu. Fuel Chem. Prepr. 1995, 40, in press. Schmitz, A. D.; Song, C. unpublished results. Kom, S. C.: Klein M. T.;Q uann, R 2. Poiynuclear Aromatic Hydrocarbons Hydrogenation. 1. Experimental Reaction Pathways and Kinetics Ind. Eng. Chm. Res, 1995, 34, 101.117. 4. 18, 1-110. 6. 6. 7. 8. 932 I 9. Stanislaus, A,; Cooper, B. H. Aromatic Hydrogenation Catalysis: A Review Catal. Reu.-Sci. Eng. 1994, 36, 75-123. Table 1. Properties of the Zeolite Starting Materials SiO,/Al20,, NapO, surface area, zeolite id. zeolite type supplier molar wt % m2/g HY zeolite Y Linde LZ-Y62 5.0 2.5 948 17.0 0.05 480 , Linde HM17 mordenite LZ-M-8 HM21 mordenite PQC CBoVw 2.0,A In c. 21,1 0.02 606 HM38 mordenite pQC cBOVr p30'' AIn' 37.5 0.07 512 Table 2.Naphthalene Hydrogenation Data for Platinum Catalysts in 60 min Runs at 200 "C Product Distribution (mole %) trans. cis- total trandcis average metal catalyst conv., % TeHN DeHN DeHN DeHN DeHN particlesize,A PtlHM17 100 0.0 37 63 100 0.58 120 PUHM21 100 0.0 45 55 100 0.83 590 PUHM38 100 0.0 70 30 100 2.34 50 (750)' PUHY 100 2.6 15 82 97 0.18 1700 a Bidisperse metal. Particle sizes for the broad (and sharp) components of the Pt (1 11) are indicated. Table 3. Naphthalene Hydrogenation Data for Palladium Catalysts in 60 min Runs at 200 'C Product Distribution (mole %) trans- cis- total trandcis average metal catalyst conv., % TeHN DeHN DeHN DeHN DeHN particle size, .k PdIHM2I 100 0.0 75 25 100 3.00 310 Pd/HM38 100 0.0 82 18 100 4.42 < 60" PdmY loo 0.0 73 27 100 2.69 310 Pd/HM17 100 0.0 65 35 100 1.84 < 60" a Diffuse diffraction line makes accurate width measurement difficult. Metal particles are no larger than the indicated size. Table 4. Equilibrium DeHN Isomer Distribution from Naphthalene Hydrogenation Over PdlHM21 Catalyst at 200 "C Product Distribution (mole %) trans- cis- total translcis time(h) cow,% TeHN DeHN DeHN DeHN DeHN 1 100 0.0 76.5 23.5 100 3.26 6 100 0.0 93.1 6.92 100 13.46 10 100 0.0 93.2 6.78 100 13.74 24 100 0.0 93.2 6.85 100 13.69 933 Table 5. Comparison of c-DeHN Isomerization Rate Constants from This Work and the Literature PtlHM38 Pd/HM38 WHY Pd/HY PtMM38' Pt/Al,O,b k , h-' 0.78 0.84 0 0.35 0.30 0.11 k4, 15.2 16.4 0 6.9 ' 10.9 - mmol/gcat'h 'Isomerization of c-DeHN from ref. 6. bNaphthalene hydrogenation from ref. 1. PdlHhll7 Pd/HM38 Pd/HMZl PdlHY n - " 2s Figure 1. XRD patterns for palladium catalysts in the region of the Pd (111) and (200) lines. Figure 2. XRD patterns for platinum catalysts in the region of the Pt (111) and (200) lines. WIG438 -1.4 - P O N - 1 8 . . . , . , . , , , , PdRmDK 5 15 25 35 45 5s 65 Time (Tin) Figure 3. First-order rate constant plots for c-DeHN to t- DeHN isomerization at 200 "C. 934 CATALYTIC HYDROGENATION OF POLYAROMATIC COMPOUNDS USING COKE-OVEN GAS INSTEAD OF PURE HYDROGEN. C.E.Braekman-Danheux, A.H.Fontana. Ph.M.Laurent and Phhlivier Service de Chimie Gknkrale et Carbochimie, Facultd des Sciences Appliqukes, Universitk Libre de Bruxelles, CP 165, Av. ED. Roosevelt, 8-1050. Bruxelles, Belgium. Keywords: catalytic hydrogenation, pol yaromatic, coke-oven gas. ABSTRACT In order to improve the economy of the conversion process of polyaromatic molecules to their hydroclromatics analogs, camlytic hydrogenation of phenanthrene has been canid out under pressure of different simulated coke-oven gases instead of pure hydrogen. The influence Of reaction time, temperature and pressure on the hydrogenation yields and on the nature of the obtained products has been studied. Comparisons have been made with reaction with pure hydrogen in the same conditions. The influence of the different components of a real coke-oven gas has also been pointed out. The results indicate that coke-oven gas wn be used if the goal is not to obtain perhydroaromatics compounds for a thermal cracking, but to give partly hydrogenated compounds to be used as hydrogen donor solvent in a coal liquefaction process. The results have been applied to coal-tar highly aromatic fractions. INTRODUCTION Coal derived heavy oils conlain principally polyaromatics hydrocarbons (P.A.H.). Previously published studies1,2 showed that these molecules must be hydrogenated to their perhydrogenated analogs in order to obtain high yields of light aromatics (B.T.X.) and ethylene by thermal cracking. Moreover, these perhydrogenated molecules could provide one of the best solution to the requirement of modern jet plane fuels 3-5. If the hydrogenation is not complete, the partly hydrogenated oils could also be used as hydrogen donor solvent in a coal liquefaction process. An important economic factor in hydrogenation processes is the hydrogen cost To reduce this cost will be beneficial. In this perspective, Doughty and aL6.' studied the hydrocracking of coal derived liquids using bimetallic catalyst and a gas mixtureof 93% I-12 I 10% CO instead or pure hydrogen. They showed that the conversion to low boiling point materials was lower in presence of 10% CO, probably because the CO molecules occupy the H2 sites at the catalyst surface. Fu and aL8 performed the hydrogenation of model compounds in presence of petroleum solvent using syngas (H2 : CO = 1: 1). The experiments were carried out in a microreactor during 45 minutes. The results showed that the hydrogenation of anthracene at 350°C under H2 pressure gave 90% conversion. The reaction with syngas at the same temperature gave only 65.9% anthmcene conversion. The most important product was dihydroanthncene. The use of WCO mixtures may introduce other competing reactions which lead -to undesirable products. For example, if the experimental conditions are similar to those used in the methanation reaction, hydrogen and carbon monoxyde will be consumed, producing unwanted methane and water. Even with low levels of CO, catalyst poisoning may be increased. The loss in &pic activity would lead to reduced conversions, which have to be compared to the gain in cast due to the use of a coke-oven gas. Moreover, industrial gases used directly may contain HzS which could poison the catalyst. The aim of this work is to hydrogenate polyaromatics hydrocarbons using the coke-oven gas instead of pure hydrogen. Coke oven-gas contains approximatively 55% Hz, 30% CH4, the Mance being made by COZ, CO, CnHm, N2. Phenanthrene has been chosen as a model compound for the P.A.H. in the heavy oils. Many works over catalytic hydrogenation of P.A.H. were published, but anyonestudy the possibility of using coke-oven gas instead of pure hydrogen. EXPERIMENTAL Procedure Experiments were performed in a 250 ml stainless steel PARR 4570 autoclave. 10 g of phenanthrene and 2 g of catalyst are introduced in the reactor. without solvent. The inskdlation is purged with the reactive gas and pressurised at ambiant temperature. Temperature is then raised as quickly as possible to the desired one. The reaction system is continuously stirred at 250 rpm. After reaction, the autoclave is cooled at ambient temperature and depressurised. The gases and the liquids are collected and analyzed. Catalysts A commercial nickel-molybdene catalyst ( 3% NiO, 15% Moo3 on alumina, surface area : 300 m*/g) was used in all experiments. For comparison, some experiments were carried out with a cobalt-molybdene catalyst and also with a palladium catalyst. 935 Analyses Gases and liquids products are analyzed by gas chromatography. The g.c. conditions are described elsewhere'. The products, first identified by gdms, are dihydrophenanthrene (DHP), tetrahydrophenanthrene (THP), sym-octahydrophenanthrene(sOHP), asym-octahydrophenanthrene (asOHP). and perhydrophenanthrene (PHP). Mass balances are controlled after each experiment. The compositions of the different simulated coke-oven gases used in this work are given in table 1. RESULTS AND DISCUSSION Hydrogenation of phenanthrene with pure hydrogen (gas 1) The hydrogenation of phenanthrene under H2 pressure was studied as a fonction of reaction time. Temperature and pressure were kept constant (370'C-21 MPa).The results are summarized in table 2 and confirms these of Colglough 10 at the same temperature and pressure. It can be seen also that the yield of cracking products remains low. These results will be used for the comparison with the hydrogenation performed with the other gases mixtures. Hydrogenation of phenanthrene with simulated coke-oven gases. Influence of reaction time. The hydrogenation of phenanhrene was studied as a function of time, at 370°C and under 2 1 MPa of a gas containing the two main components of a coke-oven gas, the Mance being made by nitrogen (gas 3). The results are presented in table 2. As in pure hydrogen, the conversion of phenanthrene is high. more than 90% after 2 hours.The PHP yield increases regularly but cannot reach the value obtained under pure hydrogen, even after 16 hours of reaction. Influence of temwrature The hydrogenation of phenanthrene with gas 3 was studied between 300 "C and 450°C. Pressure and reaction time were kept constant (21MPa- 16 hours). The results are shown in table 2. It can be seen that a rise in temperature favour the hydrogenation reactions. The maximum yield of PHP is obtained at 370°C. At higher temperature, the yield of PHP decreases while the yields of cracking products and phenanthrene increase. Higher temperatures introduce ring opening reactions, lcading to the formation of lower molecular weight products. The formation of aromatics compounds by deshydrogenation reactions is also favoured by increasing tcmperature and explain the high yield,of phenanthrene. lnfluence of pressure. The hydrogenation of phenanthrene was studied between 11 and 25 MPa of gas 3. Temperature and reaction time were kept constant (370°C-16 hours). Table 2 gives the yieldof the products as a function of pressure. At the lower pressure, the conversion of phenanthrene is low and the hydroaromatics species are the major products. The hydrogenation of phenanthrene is favoured at higher pressure. These results are also in agreement with these of previous works. Influence of the comwnents of the coke-oven gas. In order to lam more about the influence of each components. catdytic hydrogenation of phenanthrene was studied under pressure of different simulated coke-oven gases. Temperature, pressure and reaction time were kept constant (37OoC-21MPa-16 hours). Results are shown in table 2. The results are also compared to the one obtained under 11.5 MPa of hydrogen, this pressure corresponding to the M partial pressure in the coke-oven gas. - Hydrogen influence Under pure H2 pressure, the conversion of phenanthrene is 99%. The yield of PHP is more than 80% and the yield of cncking products is small. The conversion and the yield of PHP decrease in presence of nitrogen (gas 2). However. the yield of cracking products become important.The comparison between the results obtained under 11.5 MPaof hydrogen and under 21 MPa of gas 2 permitted to distinguish between the influence of total pressure and hydrogen partial pressure. It can be seen that for the same hydrogen pressure, the yield of cracking products is less important if the total pressure is higher. - Methane influence The comparison of the results of reaction with gas 2 and gas 3 showed that the yields of the different hydrogenated products are not significately influenced by the presence of methane. In our experimental conditions. methanc does not handicap the atalytic hydrogemtion of phcnanthrene. - Influence of the other components The presence of ethane, ethylene and carbon dioxide (gas 5) modifies slightly the composition of the hydrogenation products. On the other hand, the presence of CO lead to an important drop of the yields of hydrogenated compounds and specially the PHP. The cracking bccomes also important: it raises from 10% to308. The influence of the carbon monoxide can be explained by the following hypotheses: - formation of alkanes by Fisher-Tropsch type reactions, consuming hydrogen, for example: CO + 3HZ -. CH4 + H20 - nanial de-tivation of :he calalysls, lhe carbon monoxide occupying preferentially some active sites c'. 936 i Y In our experimental conditions, these reactions occur as shown by the following experimenb. The composition of the hydrogenating gas (gas 4) has been compared before and afler reaction in 3 cases: Expl : the autoclave contains coke-oven gas (gas 4) alone; Exp2 : the autoclavecontains coke-oven gas (gas4) and the NiMo catalyst; Exp3: the autoclave contains coke-oven gas (gas 4). the NiMo catalyst and the phenanthrene. The results are shown in table 3. The experiments 2 and 3 confirm the reactivity of carbon monoxyde: the yield of methane increases while the yield of hydrogen decreases when coke-oven gas is tmted in presence of the catalyst. The formation of propane could also be explained by Fischer-Tropsch reaction. Partial hydrogenation of ethylene occurs also, even in the absence of the Thermal cracking of the hydrogenated compounds. In order to verify the thermal behaviour of the hydrogenated compounds, thermal cracking experiments have been performed on the mixtures obtained after hydrogenation of the phenanthrene with gas 4 and with pure hydrogen. The cnckings are made at atmospheric pressure undernitrogen at 800°C and with a residence time of 1 s. As shown in table 4, the BTX yields obtained by thermal cracking are directly related to the amount of perhydrogenated compounds present in the hydrogenation products, as explained previously 'I. Hydrogenation of heavy oils. Two different industial oils have been hydrogenated : an heavy naphtha fraction of petroleum (HLN) and a chrysenic fraction of a coal tar (HC). The experimental conditions are: ?": 370°C, P : 21 MPa, t : 16 h, cat.: sulfided Ni-Mo. The hydrogenating gas used is gas 4 and the results are compared with hydrogenation under pure hydrogen. Due to the complexity of the oils, the thermal cracking of the mixtures obtained after hydrogenation has been directly performed, mainly to compare the BTX yields. It has to be pointed out that the untreated heavy oils do not contain BTX. The results, summarized in table 5. indicate clearly that the amounts of perhydrogenated compounds are lower when the simulated coke-oven gas is used for the hydrogenation, confirming the results obained on the model substance. calalyst. CONCLUSIONS The results obtained during this research have shown, the possibility to use a coke-oven gas to perform the hydrogenation of PAH with a commercial catalyst. But it wm not possible, even by increasing the reaction time, to obtain with a gas containing 55% H2 the Same yield of PHP as the one obtained with pure hydrogen. The influence of the various components of the coke-oven gas on the hydrogenation yields has been investigated: CH4 does not handicap the cahlytic hydrogenation of the phenanthrene; the presence of C2H4, C2H6, and CO2 modifies slightly the composition of the hydrogenation products; the presence of CO leads to an important drop of the yields of hydrogenated compounds and specially the perhydrogenated. It can be concluded that, if the goal of the hydrogenation is not to obtain perhydrogenated compounds for a chemid upgrading by thermal cracking, but to give partly hydrogenated compounds to be used as hydrogen donor solvent in cml liquefaction processes, then the hydrogen can be economically replaced by a coke-oven gas. AKNOWLEDGEMENTS. The authors wish to thank the Commission of the European Communities. Coal Directorate, for their financial support for the project (ECSC Project 7220-EC/208). REFERENCES. I. Cyprhs R. and Bredael P., Fuel Process.Technol., 1980. 3,297. 2. Bernhardt R.S., Ladner W.R., Newman J.O. and Page P.W., Fuel, 1981, 60,139. 3. Perry M.B, Pukanic G.W.and Ruether J.A., Preprints Amer.Chem.Soc.Div.Fue1 Chem.,1989, 4. Sullivan R.F., Preprints Amer.Chem.Soc.Div.Fue1 Chem., 1986,31,280. 5. Greene M.,Huang S., Strangio V., ReillyJ., Preprints Amer.Chem.Soc.Div.Fue1 Chem.,l989, 6. Doughty P.W., Harrison G. and Lawson G.J., Fuel, 1989, 68, 298. 7. Doughty P.W., Harrison G. and Lawson G.J., Fuel, 1989, 68. 1257. 8. Fu Y .C., Akiyoshi M., Tanaka F.and Fiyika K., Preprints Amer.Chem.Soc.Div.Fuel Chem., 9.Braekman-Danheux C., Cyprks R.. Fontana A, Lauren1 Ph. and Van Hoegaerden M., Fuel, 34,1206. 34.1 197. 1991.36.1887. 1992. 71.251. 10. Colclough P.: Proceedings of ECSC Round-Table on Coal Valorization , 1982.36. 937 Table 1. Composition ( vol.% ) of the gases used for hydrogenation. Gas 1 : 10096H2 Gas 2 : 55% HZ, 45% N2. Gas 3 : 55% H2,15% N2,30% CH4 Gas 4 : 55% HZ, 1% N2,30% CH4,6% CO, 3% CZH4,3% C2H6,2% C02 Gas 5 : 55% HZ, 7% N2,30% CH4,0% CO, 3% CZH4,3% CZH6,2% COZ Table 2. Hydrogenation of phenanthrene (catalyst : sulfided Ni-Mo) Gas T t P Phen. DHP THP OHP PHP Others (“C) (h) (MPa) (Wt%) (Wt%) (Wt%) (Wt%) (Wt%) (Wt%) 1 370 1 21 5.8 8.3 5.9 26.8 48.0 0.9 2 1.2 7.3 1.9 24.8 60.2 1.2 4 0.7 4.9 1.6 23.6 67.9 1.1 16 0.7 1.9 0.6 20.7 81.2 1.1 16 11.5 3.9 5.8 6.0 16.5 55.6 2.4 3 370 2 21 6.8 11.1 13.1 33.7 26.6 0.5 4 6.2 10.0 10.5 25.9 38.9 1.3 8 5.4 7.5 8.2 18.7 47.9 2.3 12 5.2 6.0 6.8 17.9 54.3 7.1 16 2.8 5.4 5.2 14.6 61.0 8.9 3 300 16 21 3.4 0.5 1.9 72.5 12.7 2.6 350 2.7 5.7 4.8 24.1 58.0 1.3 400 5.3 8.3 7.9 20.1 40.5 7.3 450 25.0 5.4 5.2 3.5 33.9 18.0 3 370 16 11 25.3 7.3 13.6 13.5 21.4 3.5 15 12.2 6.1 11.4 19.1 37.8 5.3 25 0.9 3.2 1.9 9.7 72.8 7.2 2 370 16 21 3.7 5.6 5.8 13.9 59.8 0.8 4 7.4 5.2 8.0 16.4 26.6 29.2 5 3.1 5.2 5.0 14.6 57.8 9.0 Table 3. Behaviour of the hydrogenating gas (gas 4) during the reactions (vel.%) H2 CH4 CO CO2 CzH4. CzH6 C3H8 Nz Gas4 55.0 30.0 6.0 2.0 3.0 3.0 0.0 1.0 Exp.l 54.3 30.9 5.8 2.4 1.3 4.3 0.0 1.0 Exp.2 52.4 36.7 1.1 2.1 0.3 5.2 1.3 1.0 Exp.3 39.4 46.0 1.7 2.1 0.3 8.1 1.4 1 .o 938 Table 4. Thermal cracking of the phenanthrene hydrogenation products. Cracking conditions : To: 800°C. residence time : 1 s, P : 0.1 MPa NZ. Hydrogenation conditions : To : 370'C, t : 16 h, P : 21 Mh wt% of the main compounds. Product 1 : hydrogenated with gas 1. Product 2 : hydrogenated with gas 4. Product 1 Product 2 Before cracking After cracking Before cracking After cracking Benzene 10.6 2.6 Toluene 4.8 1.7 Xylenes 1.7 0.8 Total BTX 17.1 5.1 Phenanthrene 0.7 3.9 7.4 15.4 DHP 1.9 5.2 0.4 THP 0.6 8.0 0.8 OHP 20.7 0.8 16.4 1.2 PHP 81.2 1.9 26.6 0.5 Table 5 . Thermal cracking of the heavy oils hydrogenation products. Hydrogenation and thermal cracking conditions are the same as in table 4. BTX yields (in wt%) HLN HC gas 1 gas 4 gas 1 gas 4 Benzene 6.7 3.6 5.1 Toluene 5.5 2.9 2.8 Xylenes 2.3' 1.9 1.5 Total BTX 14.8 8.4 9.4 2.1 0.7 0.5 3.3 1 I I 939 Role of Iron Catalyst on Hydroconversion of Aromatic Hydrocarbons Eisuke Qata, Xain-Yong Wei', Akio Nishijima'. Tom0 Hojo' and Kazuyuki Horie w e n t of Chemistry and Biotechnology. Graduate School of Engineering, The University of Tokyo; 7-3-1, Hongo. Bunkyo-ku. Tokyo 113, Japan 1. Introduction A symposium on iron-based catalysts for coal liquefaction was held at the 205th ACS National Meeting[l]. and some of the p a p have been published in Energy & fietels[2]. Reviews of the development of catalysts for coal liquefaction were also published in Journal of the Japvln Insrimre of Energy[3], and Ozaki reviewed the results of the studies of upgrading residual oils by means of thermal cracking and c d t i g under reduced pressures, catalytic cracking over nickel ores and iron oxides, and hydrodesulfurization. as well as hydrodemetallization[4]. We reported that catalysis of metallic iron and iron-sulfide catalysts were affected by the S/Fe ratio; the activity inmsed with pyrrhotite formation and the activity was accelerated by the presence of excess sulfur[5-8]. Activity of pyrite FeS, for phenanthrene hydrogenation[9] and activity of natural ground pyrites for coal liquefaction[lO] decreased with storage under air. On the other hand, the NEDoLprocess for a coal liquefaction pilot plant of 150 t/d which is one of the national projects in Japan, will use pyrites as one of the catalysts for the first-stage because FeS, has high activity and is low in price. In this paper, we describe in detail the role of iron catalysts in hydroconversion of aromatic hydrocarbons such as diphenyl (DPh), dinaphthyl (DNp) and diarylalkanes (DAAs) constructed with monocyclic aromatic-units and/or bicyclic aromatic-units and both monocyclic and bicyclic aromatic units and linked with from one to three methylene-groups. 2. Experimental Materbls: (I-Naphthy1)phenylmethane (NPM) and di(1-naphthyhethane (DNM) were synthesized by heating naphthalene (NpH) with benzyl chloride and 1-chloromethylnaphthalene in the p e n c e of metallic zinc powder catalyst, respectively[ll], and 1,2-di(l-naphthyl)ethane (DNE) was synthesized by the reaction of I-bromomethylnaphthalene with metallic iron powder catalyst in boiling water [12]. 1.3-Di(l-naphthyl)ppane (DNP) was synthesized by a coupling reaction of 1-naphthylmagnesium bromide with 1.3-dibromopropane in the presence of copper(1) bromide catalyst in hexamethylphosphric biamide (HMPA) solvent [13]. These diarylalkane (DAAs) were purified using conventional methods such as vacuum distillation, separation with silica and alumina column chromatography and rrcrystallization from the solutions. The other substrates such as 1-methylnaphthalene (1-MN), diphenyl (DPh). 1,l'-dinaphthyl (DNp). diphenylmethane (DPM), 12diphenylethane (DPE), 1,3-diphenylpropane (DPP), triphenylmethane (TF'M), 1-[4-(2-phenylethyl)benzyIlnaphthalRle (PEBN); hydrogen-donors such as telralin (THN), 9.10-dihydrophenanthrene (DHP) and 9,lO-dihydroanthracene (DHA); and the solvent decalin (DHN). were purchased commercially and further purified, if necessary, by conventional methods. Pyrite FeS, and metallic iron ultra f i e powder Fe were synthesized by Asahi Chemical Industry Co. Ltd. and Vacuum Metallurgical Co. Ltd., respectively. procedure: In typical reactions, 1.0 g of I-MN or PEBN or 7.7 mmol of D h . the prescribed amount of FeS, or Fe catalysts and 30 ml of DHN, as well as the prescribed amount sulfur (SFe ratio = 2.0) if necessary. were placed in 90 ml or 150 ml stainless steel, magnetically stirred autoclaves. After pressurization with 10 MPa of hydrogen. nitrogen or argon. the Catalysts: ^" Present Address: 'Department of Coal Reparration snd Utilization. China Univewrsiiy of Mining snd Technology. Xuzhou (221008). Jiangsu, China: hrface Characterization Laboratory. National Institute of Malerids and chemical Research. 1-1, !gashi. Tsukuba-shi, Ibaraki 305. Japan; 'Depament of Industrial Chemistry, College of Science ard Technology. Nihon Universlly, 1-8. Surugadai. Kanda, Chiyoda-ku. Tokyo 101. Japan. 940 I autoclave was heated to the desired reaction temperature from 300°C to 400°C within 20 min and maintained for 1 hr. It was then immediately cooled in an ice-water bath. Analyses: The reaction Products were identified by GC-MS (Shimadzu GCMS QP-1000, equipped with a 0.24 nun (I.D.) x 50 m (I,.) glass capillary column chemically bonded with OV-1 )and quantified by GC (Shimadzu GC-lSA, equipped with the same capillary column ). Oxidation of FeS, CarOlySl: FeS, was oxidized at room temperature. 80°C. 150°C and 200°C for the desired time under atmospheric air. The bulk structure of iron catalysts oxidized and recovered after the reaction was analyzed by using XRD ( Rigaku tknki Model RlNT 2400 ) and XPS (Perkin-Elmer Model PHI 5500). Substrates and the notations areshown in follo.wing-: PI Qc c c PHN 6 6 6 -$ DPP TPM DNp PNM DNM DPM DPE DNE DNP pEBN 3. Results and Discussion Thermolysis of aromatic hydrocarbons was strongly affected by the bridged-methylene length, aromatic-ring size in the arylalkane structure. and the presence of molecular hydrogen (H,) in the reaction system as shown in Table 1. Under treaction conditions of 4oooC for 1 hr. DPM was very stable in both the absence and presence of H,. The reactivity of diarylalkanes (DAAs) constituted of two phenyl-rings increased in the following order: DPM << DPE << DPP. These conversions were slightly increased by the presence of H,. Generally, the reactivities of DAAS constituted of two naphthyl-rings were higher than those of diphenylalkanes (DPAs). It is particularly remarkable Uiat many radical-adducts between species formed from C,,,k-Co,k bond cleavage and solvent DHN molecule was recovered. and many phenylethyldecalins and naphthylethyldecalins were produced especially in the case of diarylpropane thermolyses and the formation was slightly depressed by the presence of gas phase H, molecule. From other experiments it was shown that the reaction between styrene and DHN under the same reaction conditions formed many phenylethyldecalins as solvent adducts. suggesting that these solvent adducts were produced when arylethylenes were formed in the reaction system. These results ultimately indicate that C-C bond scission was proceeded by a radical chain reaction and the reactivity of DAAs was governed by dissociation energy of C,,k-C,,k and C,,&,, bonds in the absence of catalyst. Tables 2 and 3 show the effect of bridged methylene length in the diarylalkane structure on the hydmonversion with pyrite FeS, catalyst at 300°C for 1 hr, and the effect of aromatic ring-size and number in the arylalkane structure. DPE and DPP were not converted even after IO hrs. The reaction of DPh yielded only cyclohexylbenzene via the hydrogenation of one benzene-ring. DPM hydrocracking also proceeded via the C,,-Co,t bond scission, but DPM was much less reactive than triphenylmethane CTPM). DNp hydrocracking resulted in the corresponding tetra-hydrogenated 1 ,l'-dinaplithyls as main products. Reactivity of DNM was the highest in this hydroconversion series. DNM hydrocracking mainly produced naphthalene (NpH) and I-MN, via hydrogen addition to the ipso-carbon of DNM. Only a small amount of hydrogenated di(l-naphthyl)methanes (H-DNMs) was produced. DNM hydrocracking was much easier than that of DPM and TPM. Drastically different from DNM, the reactions of DNE and DNP mainly yielded the hydrogenated 1,2-di(l-naphthyl)ethanes (H-DNEs) and the hydrogenated 1,3-di(l-naphthyl)propanes (H-DNPs) rather than decomposed products, respectively. The result shows that the cleavage of the C,,-C,,t linkage in DNE and DNP is much more difficult than that in DNM. The total selectivity of decomposed products in the case of DNp was higher than that in DNE hydroconversion. Reaction of DNP mainly produced hydrogenated 1,3-di(l-naphthyl)- propanes (H-DNPs), and a small amount of hydrocracked products such as NpH and (I-naphthyl) 941 propane (1-NP). Table 4 shows the effects of Fe and FeS, catalysts and reaction temperature on the hydroconversion of DPM. FeS, catalyst has more hydrocracking sites than hydrogenation sites. while Fe catalyst has highly active sites and mainly produced hydrogenated diphenylmethanes (H-DPMs). Table 5 shows the additive effects of hydrogendonors (H-donors) on DPP thermolysis and the additive effects of metallic Fe and FeS, catalysts on DPP hydroconversion at 4oo°C for 1 hr. DPP conversion decreased with H-donor addition in the order: none > THN >> DHP > DHA. These results are easily undexstood because the resulting PhCH,' abstracts hydrogen atoms from the Hdonors readily more than it does from DPP. In other words, the H-donors inhibited the radical chain reaction in DPP thermolysis by donating their benzylic hydrogen to PhCH,'. Table 5 also demonstrates the catalytic effects of FeS, and Fe on the DPP thermolysis when compared with the non-catalytic reaction of DPP under H, of 10 Mpa at 400°C. FeS, greatly promoted C,,-C,,k bond scission as DPP hydrocracking. Under H,, the rate for DPP hydrocracking in the presence of FeS, was ca. 2-fold faster than that in the absence of the catalyst and dramatically decreased the formation of solvent adducts. Fe catalyst promoted DPP hydrogenation, but DPP conversion was low, about the same as that under N,. 'fhis result suggests that Fe catalyst promoted DPP hydrogenation and inhibited thermolysis. The inhibiting effect of Fe on DPP thermolysis remains to be investigated. It appears remarkably that the formation of solvent adducts such as phenylethyldecalins (PEDS) was drastically inhibited by FeS, and metallic Fe catalysts. FeS, was oxidized to ferrous sulfate FeSO,'H,O even at w m temperature under atmospheric air, and the catalytic activities of oxidized FeS, on I-MN hydrogenation were decreased with increases in the storage time. Recently, Linehan and co-workers [14,15] reported the C-C bond scission activity of PEBN on well-characterized eleven synthesized iron-oxygen compound catalysts in the presence of elemental sulfur and a hydrogen-donating solvent in detail. As shown in Figs. 1 and 2, thermolysis of PEBN under Ar was stable, but the conversion of PEBN was effectively accelerated in the presence of 10 MPa H, and the main readon was changed from C,,k-C,, bond scission (Route [A] )t o C,,-C,, bond scission (Route [a).Me tallic Fe catalyst mainly accelerated the hydrogenation (Roufe [B]), and FeS, catalyst promoted C-C bond scission of Roufe [a. FeS, catalyst activity was decreased with the oxidation of FeS, by oxygen in the air. However, the deactivated FeS, catalysts were mactivated by the presence of excess elemental sulfur in the system, and the reaction proceeded along Roufe [C]. 4. Concluding Remarks Hydroconversion of aromatics and arylalkanes as a model reaction of coal liquefaction and heavy petroleum residue degradation was carried out in the absence or presence of metallic iron Fe and pyrite FeS,. Thermolysis of some diarylalkanes proceeded slowly by the radical chain reaction. The reaction rate was reduced by the addition of hydrogen-donating solvents and was slightly accelerated in the presence of hydrogen molecules. Metallic Fe catalyst accelerated the hydrogenation of aromatic-rings, espially bicyclic-rings, more than that of monocyclic-rings. FeS, catalysts, which is converted to pyrrhotite Fe,,S under reaction conditions, promoted C,,-C,, bond cleavage of diarylmethanes only, and also promoted the hydrogenation of diarylethanes and diarylpropanes. C-C bond cleavage of arylalkanes was related to the hydrogen-accepting ability. C-C bond dissociation energy and resonance energy of the species after C-C bond scission. Oxidation of pyrites and its catalysis were also investigated. It was found that the catalytic activity of pyrites in the hydrogenation of 1-methylnaphthalene and 1-[4(2-phenylethyl)benzyllnaphthalened ecreased with oxidation under air, and deactivated pyrites was reactivated by addition of sulfur to the reaction system. ' ' Acknowledgment The authors wish to thank to the New Sunshine Rogram Promotion Headquarters. Agency 942 J J Of Industrial Science and Technology, Ministry of International Trade and Industry Of Japan for financial support. References 1. hprt."Syrnposium on Iron-Based Catalysts for Coal Liquefaction", Div. Fuel Chem.. 2. Energy &Fuels, &(l), (Special issue), p-2-123 (1994). 3. J. Jpn Instit. Energy, a (1). (Special issue). p-2-49 (1994). 4. H. m i , Sekiyu Gakkaishi, 3-6, (3), 169 (1993). 5. E. Ogata, E. Niki, hoc. 27th Conf. Coal Sci. Jpn. Soc. Fuel (Tokyo). pp-115 (1990). 6. X.-Y. Wei, E. Ogata, E. Niki, Chem. Lett., 2199 (1991). 7. X.-Y. Wei, E. Ogata, Z.-M. Zong. E. Niki, Energy&Fuels, 6. 868 (1992). 8. E. Ogata, K. Ishiwata, X.-M. Wei, E. Niki, Proc. 7th Inter. Conf. Coal Sci.(Banff). Vol.' 9. E. Ogata, T. Suzuki, K. Kawamura. Y. Kamiya, 26th Conf. Coal Sci. Jpn. Soc. Fuel 205th ACS National Meeting (Denver), Vol. 38 (No.l),pp-1-238 (1993). 11, pp-349 (1993). (Sapporo), pp121 (1989). 10. K. Hirano, T. Hayashi, K. Hayakawa. 30th Conf. Coal Sci. Jpn. Soc. Fuel (Tokyo), 11. S. Futamura, S. Koyanagi, Y. Kamiya.FueL 63. 1660 (1984). 12. N. P. Buu-Hoi and N. Hoan, J. Org. Chem.. u, 1023 (1949). 13. J. Nishijima. N. Yamada, Y. Horiuchi. E. Ueda. A. Ohbayashi, and A. Oku, Bull. Chem. 14. J. C. Linehan, D. W. Matson. and J. G. Darab, Energy & Fuels, 8, 56 (1994). 15. D. W. Matson, J. C. Linehan. J. G. Darab. and M. F. Duehler, Energy & Fuels. 8, pp-161 (1993). Soc. Jpn.. Se, 2035 (1986). 10 (1994). Table 1 Effect of Chain-length and Rlng-sizefn the Diarylalkanc Structure and 10 MPa of H y d a t 4 M l Cf or 1 hr. Substrate 0 selectivity (mol % ) Benzene o o 0 98.2 o 3.4 o o o n Naphthalene 0 0 0 0 0 0 61.7 1.5 12.6 0 Arylmethane o o 20(r 101.a' 100' 96.6' 61.7b 174.4~1 3 2 . ~1~05 .0b Arylethane 0 0 0 98.2' 2.9' 16.5' 0 9.7d 6.8 Arylethvlene 0 0 0 0 9.8" 6.2' 0 0 0 14.9' Table 2 Elfects of Chain-length in the Diqlalkane S~NC~UonI XH ydmconversionw ith b i l e F e m 0 0 " C fo r 1 hr. 6.2) (36.1) (0) (0) --- --- ___ ___ Selectivity( mol 9% ) Beraene 0 1 0 0 0 0 0 0 0 0 Naphthalene 0 0 0 0 17.3 95.7 1.8 10.3 ArylmelhanC 0 loob 0 0 0 89.0' 2.3' tr' H-DAAs 100 0 0 4x3' 2.9 868 88.1 Arylelhane ' 0 0 0 0 0 ' 0 o.Ad tr" Arylpropnc 0, 0 0 0 0 0 0 7.0 a: ReactionTme 10 hrs: b Toluene: c: 1-Melhylnaphlhdene: d Elhylnaphlhalene: e: HydmgnrsM Diarylalkanes: I: Cyclohenylbwzme: 8: HydmgRlsted Diaphlhyls. 943 Table 3 EIlects of Ring+ vd Number in the Diarylmee Slr~cture 0- C for 1 hr. DPM WM NPM DNM 3.1 36.1' 21.4 86 92 Selectivity ( mol 96 ) Benzene 100 99 102.6 tr 0 Naphthalene 0 0 96 mlmethane 100' 99' g.6' 76:' 89' Telralins 0 0 0 9.3 7.9 Diarylmethane --- --- 97.k 12O -_- H-DAAs' 0 1.5 W 15 2.9 a: Reaction Time IO lus: b: Tolumc: c. I-Melhylnaphhlcne: d: Diphenyhelhane: E: (Z-NsphlhylJ-phmylme~~1c. ;H yckqenared diqlalkanes. Table 4 Effects of Iron Catalysts and Reaclion Tempxature on Hydroconversion of Diohnvlmdhanc for 1 hr. Tcm , Caialysi (g) COW. &&$itv ( mol % ) ("c o) CHx' PhH MCP To1 DCHM BCH 300 Fe (0.23 ) 58.4 0 0 0 0 13.4 86.6 300 FeS, (0.50)' 3.1 0 100 0 100 0 0 400 Fe (0.02)' 79.3 3.2 0 9.8 0 42.1 57 400 FeS. (0.50) 59.1 0 98 0 97.2 2.4 0 a: Cyclohexk: b w, ~,Methylcydohexanc: d: Toluene: e: Dicyc~ohexyhcthane: I: &nzykyclohexam: 8: Addi lc~o fS ulfur 0.05 8. Tabic 5 Additive Fllect of Hydrogcn-Donating Solvcnts and Iron Catalysts on Conversion R c a c l ~ o n _ o L D _ i p h c ~ l ~ ~ ~ ~ c - a I ~ C l o r l h f Additive (g) Gas React. Conv. SleCtivity ( mo10/0) Phase Timethi1 (Yo) PhH' Tal' Elmd S& CHPP' BCHP P E D L 1 39.9 0 100 2.9 9.8 0 0 87.4 71)s 3 1 28.6 0 100 3.8 19.5 0 0 76.7 DHP 7.5' N: 1 7.4 0 100 6.4 39.8 0 0 53.8 None 0 1 4615 3.4 96.6 16.5 6.2 0 0 73.4 None 0 $ 2 7:.1 :.I E 9 ",.I :.I 0 68.9 Fe 0.02 2 65.9 10.7 53.948.8 0 26.9 8.5 15.8 FeS 0.5d 2 1 69.7 8.7 87.6872 2.6 3.7 0 0 ~ e S ~ O . S d - H : _ _ 2 ~ 6 _ _ 9 , 8 6 . 8 8 6 . ~ 1 . 9 _ 3 5 0 0 a: Rcacuon Condilion DPP 7.5 mol. DHN 30 ml. PH2 IO MPa: b Benzene: c: Toluenc: d: Elhylbcmcnc: e: Slyrene: I: C~c lohe~lphcnyl~ropan8e: :B icyclohcxyllropane: h Phenylelhyldceslins: i: mo l : j: Additionof Sulfur 0.05 8. - D H A L N 1 1 0 . 9 2 . 0- Fe 0.02 H, 1 40.0 6.6 63.4 56.2 0 24.7 5.3 13.8 100 :f-J 40 20 0 o z 4 B a i o 1 2 1 Rescllon Time ( hr I :ig. 2 Reaction 01 1-[4(2-Phenylethyl)benzyl]naph~halene Fig. 1 EHecl of Reaction Time on the Conversion Reactlon 01 1 -[4-~2-Phenylethyl)benryllnaphthalene. 0 HI. 350°C. 0 Hz .380"C. 0. HZ . 400'C. 0 ' Ar .380"C. A. H2. 380'C. Fe Cal(0 05 g) ReaCrtonCondilwlns PEEN 1 Og. DHN 30ml PH2 1OMPa. ' Ar. 400°C. A: H2 - 380°C. DHA Adalllon(0HNPEBN = 3 0). 944 \ i I / I i CATALYSIS OF ALKENE AND ARENE HYDROGENATION BY THERMALLY ACTIVATED SILICA Venkatasubramanian K. Rajagopal Robert D. Guthrie Department of Chemistry University of Kentucky Lexington, KY 40506 Burtron H. Davis Kentucky Center for Applied Energy Research 3572 Iron Works Pike, Lexington, KY 4051 1 Keywords: Deuterium Incorporation, Diphenylacetylene, Hydrogenation Stereochemistry INTRODUCTION In a recent study we described the hydrogenation of stilbene, a-methylstyrene and anthracene at 410 OC under 14 MPa of hydrogen or deuterium gas in the absence of added catalysts to give diphenylethane, cumene and 9,lO-dihydroanthracene respectively.' As we were simultaneously working on the hydroliquefaction of coalmodel compounds attached to silica: we became aware of work by Bittner, Bockrath and Sola? which demonstrated catalytic effects of thermally activated silica in reactions involving H, and D,. Using a.pulse-flow microreactor, Bittner showed that after fumed silica is heated at 330 OC for 16 h in an argon stream it catalyzes the reaction H, + D, --f 2 HD at temperatures as low as 120 OC. Moreover, this material catalyzes the hydrogenation of ethene to ethane at 150 OC and produces ethane-d, when D, is used as a flow gas. We became curious to see whether silica activated in this way would Sewe as a hydrogenation catalyst in a static reactor. Our results and additional information about the catalytic behavior of this material are described below. EXPERIMENTAL Hydrogenation Procedure. Approximately 300 mg of fumed silica (Cab-0-Si1 M-5, Cabot Corporation) was placed in a length of (16 mm 0. d.) glass tubing with a section of 1 - 2 mm capillary ( ca. 16 cm long) attached to one end. Plugs of glass wool were placed between the silica powder and the tube exits. This assembly was heated in a tube furnace either at 330 OC or at 430 OC for 16 or more hours with argon flowing in the capillary tube and out through a ca. 2 mm hole in a stopper placed in the wide end. After this activation time, the tube was cooled to room temperature and a substrate was introduced maintaining the argon atmosphere during the addition process. The argon flow was then discontinued, the assembly evacuated, and the noncapillary end of the tube sealed under vacuum. Argon was readmitted and the reaction vessel was placed in a steel tube reactor under H, or D, pressure. The assembly was heated for the desired time period in a fluidized sand bath as described earlier! When the heating period was complete, the apparatus was cooled, the pressure released, the glass reactor section cracked open and the silica hydrolysed with ca. 30 mL of 1 M aqueous NaOH for ca. 15 h. At this point a measured aliquot of an external standard (biphenyl in CH,CI,) was added. The aqueous solution was acidified and extracted three times with CH,CI? The glass reaction vessel was washed with CH,CI, and the washings combined wlth the extracts. The solvent was removed by rotary evaporation and the products analyzed by GC and GCIMS. Exchange Experiments. Experiments to assess surface exchange were carried out in the same way as for the D, experiments described above except that instead of cracking the tube open after the D, treatment, a sample of phenol in benzene was added to the tube through the capillary opening using small diameter polyethylene tubing. Most of the, benzene was pumped out and the capillary section sealed. The entire assembly was then placed in a tube furnace at 400 OC for 60 min. We have shown in other experiments that phenol-d, undergoes replacement of three of its ring deuterium atoms by hydrogen atoms atoms on heating with Cab-0-Si1 at temperatures above 140 OC.' After heating, the sealed tube was opened and the contents hydrolyzed in NaOH solution as described in the previous section. Control experiments showed that phenol could be recovered quantitatively from the hydrolysis workup and that the workup did not remove ring deuterium atoms. . 945 RESULTS AND DISCUSSION The results from hydrogenation of several alkenes in the presence of thermallyactivated silica are given in Table 1. The compound examined most extensively was stilbene (1,2-diphenylethene), STB. Table 1. Reaction of Unsaturated Compounds with D, or H, in the Presence and Absence of Thermally-Activated Silica. 300 90 D, A I29 + 6gh I <1 a A = Activated, NA = not activated. cidtfans = 0.085. cidfrans = 0.087. Product appears to be mixture of dihydronaphthalene-d, and tetralin-d,. e Remainder appears to be alkene isomers. DPA = Diphenylacetylene. Product contains 3.3% diphenylethane, DPE, and 38.8% mixed cis- and trans-STB, cidffans = 0.26. This run was carried out with a different procedure, however, and it is not certain that all of the DPA was available for reaction. Product is a mixture of DPE (29%) and STB with no significant amount of residual DPA. The STB shows cidfrans = 0.1 8. It is clear that for STB there is no reaction with D, at 350 O C in the absence of silica and nearly complete hydrogenation to DPE after 90 min in the presence of activated silica. For the three runs carried out in the presence of silica, the deuterium distribution in the DPE produced was d, = 85.4%, d3 = 6.0% d4 = 3.5% in the first run, d, = 83.7%, 'L, = 10.6% d4 = 2.0% in the second run, and 4 = 63.3%, d3 = 25.4% d4 = 7.2% in the third run. The balance was a small amount of do DPE present as an impurity in the STB. Thus, the product was mainly d, material which underwent additional deuteration at longer times and higher temperatures. 'H NMR showed that neither the DPE nor the STB remaining contained any significant amount of aromatic D whereas there was a prominent ' signal for the aliphatic D in the DPE produced. It will be noted that unactivated silica does catalyze the reaction to a lesser extent, but this might be expected in view of the fact that reaction temperatures are similar to activation temperatures. With naphthalene, there appears to be a small amount of dihydronaphthalene and tetralin being formed and these contain mainly 2 and 4 atoms of D respectively. However, we have not been able to increase the yield much above 1% by increases in time or temperature. Possible reasons for this situation are discussed below. Nonene was picked as a prototypical nonaromatic alkene and it shows hydrogenation to nonane. The saturated material produced under these conditions is a mixture of ca. 64% nonane-d, and 36% nonane-d,. The unreacted 1 -nonene is mainly undeuterated, but the precision required to determine small amounts of D was unavailable due to extensive mass spectral fragmentation. Both of the nonene isomers, presumably cis- and trans-2-nonene are mainly d, material but contain about 30 % do material. It thus seems likely that 1-nonene can isomerize under these circumstances and that at least part of the process does not involve the intermediacy of nonane-d,. The reaction is not limited to alkenes as anthracene can be reduced to dihydroanthracene, mainly with two atoms of D. Diphenylacetylene (DPA) is also 946 J I hydrogenated. At 350 OC, the STB produced is largely converted to DPE-d,. At 300 O C , substantial amounts of intermediate STBs are observed, mainly 4. The STB from these experiments seemed to be slightly enriched in the cis-isomer but the analysis was inherently imprecise due to the GC overlap between cis-STB and DPE. In order to provide more convincing evidence that the initial product of hydrogenation of DPA is cis- STB, the reaction.temperature was lowered to 250 OC. Data for hydrogenation of DPA (with H,) under these conditions'is presented in Table II. Table 11. Yields of Products from Reaction of Diphenylacetylene with H, at 250 "C Over Thermally-Activated Cab-0-Sil. I' I a STB = Stilbene. DPA = Diphenylacetylene. Thermal equilibration gives a ratio of 0.1 at 350 'C. OPE = Diphenylethane. e Reasons for the dramatic decrease in DPE formation in the 60 min run are uncertain. But, for this run the silica was activated at 430 "C rather than the 330 OC used for all of the other runs. This result is being checked. It is clear from the data in Table II that the predominant product of hydrogenation of DPA at 250 OC is cis-STB by a ratio of at least 3 to 1. The equilibrium ratio of cis- /trans-STB has not been established at 250 OC, but the 5th and 6th runs of Table I show that at 350 "C the ratio is 0.086 and a lower value might be expected at 250 "C. Thus there seems little doubt that cis-STB is the kinetic product of DPA hydrogenation. This would seem to clearly rule out radical processes in which the two hydrogen atoms are transferred independently. The fact that we see none of the products expected from 1,2- diphenylethyl radicals: tetraphenylbutane or 1,l -diphenylethane also supports this ~ conclusion. It appears that, whatever the active site for silica-catalyzed hydrogenation might be, it is similar to a metal-surface type catalyst in that H atoms are transferred in pairs and transferred in a stereoselectively syn fashion. It seems likely that the trans-STB produced in this reaction results from the thermal isomerization of the cis-isomer. Comparison of the STB runs in Table I carried out at 300 and 350 OC to the run in the last column of Table II shows that a temperature change of 100 O C does not have the effect on the conversion of STB to DPE that would be expected for a thermally activated reaction. For the thermal hydrogenation of STB studied earlier, reaction proceeds at a kinetically convenient rate at 410 "C but is not measurable after comparable times at 350 "C. It would thus seem a reasonable hypothesis that the rate-limiting step in the catalytic process has a low enthalpy of activation and may be controlled by geometric restrictions for access to the site. It seemed logical that if D, is combining with silica in order to be activated for addition to unsaturated molecules, that the formation of OD bonds must be involved and, this being the case, that the process would provide a mechanism for exchange of the Si- OH groups on the silica surface with the D, atmosphere. In their microreactor process, Bittner did not obselve formation of HD when D, was passed over the thermally-activated silica? Nevertheless, it seemed possible that this would happen under the higher pressures used in our experiments. To this end we carried out a series of experiments in which silica was activated then heated with D, followed by removal of the D, under vacuum and heating with phenol. Earlier work had shown that the ortho and parapositions of phenol undergo exchange with OH groups on the silica surface at temperatures above 140 OC.' After recovery from an aqueous workup the deuterium content of the phenol was determined by GC/MS analysis. Data are presented in Table 111. Although there is some scatter in the data, it is clear that heating with D, at temperatures above 200 OC introduces SiOD grou s on the silica surface. Calculations based on the expected5 4.5 SiOH groups per nm P of surface area suggest that roughly 947 75% of the surface SiOH groups are replaced by SiOD groups in the high temperature runs. The threshold temperature for the exchange reaction with D, appears to be between 200 and 250 'C. In the four runs carried out at 250 'C, the exchange with D, seemed about twice as great with silica which had undergone prior activation, however, the effect of activation on exchange was not as great as that on the STB reduction. Table 111. Deuterium Content of Phenol Exchangeda with Cab-0-Si1 (300 mg) Previously Treated with D, at 14 MPa at Various Temperatures. a See Experimental section for details of the exchange experiments It remains uncertain at this point whether the mechanism for deuterium exchange of the silica surface hydroxyl groups and the mechanism for hydrogenation of alkenes are linked. At least for the alkene reaction, we can roughly estimate the concentration of active sites by the following experiment. Activated silica is heated with D,, a process which Table 111 demonstrates will convert most of the surface SiOH groups to SiOD groups. The resultant deuterated silica (300 mg) which then has at least 0.33 mmole of D on the surface (after pumping off excess D,) is heated with excess STB. The STB then contains a small amount of DPE-d,. This amount is less than 0.015 mmole or roughly 10 % of the equivalents of D, present on the surface. We find in some experiments that the thermally activated silica also is capable of catalyzing the hydrogenation of aromatic rings. The circumstances of this occurrence and the type of compound which is susceptible are under continuing investigation. As shown in Table I, a small amount of naphthalene is hydrogenated under the conditions described. To our astonishment, DPE is more extensively hydrogenated than naphthalene provided that the silica is thermally activated either for several days at 330 "C or at 430 "C for 16 h. We have found that the latter procedure gives reasonably reproducible results. Representative experiments are listed in Table IV. The reaction gives two products with gas chromatographic retention times which are very similar to DPE. With H, these products have mass spectra which match l-cyclohexyl-2- phenylethane (CPE) and 1,2-dicyclohexylethane (DCE). Each is extensively deuterated and the 'H NMR spectrum of the mixture shows an envelope of purely aliphatic D atoms in the range of 1 to 1.5 ppm from TMS. The D content of the two reduction products suggests that much of the material arises from the replacement of the four benzylic H atoms in DPE as well as the addition of three D, molecules per reduced benzene ring. However, evidence for exchange at the aliphatic sites of the saturated rings is provided by the presence of up to CPE-d,, and up to DCE-4,. The DPE which is recovered is mainly DPE-d, but there is evidence for some exchange in the aromatic rings as well. Astonishingly, when naphthalene is mixed with DPE and subjected to the conditions of Table IV, neither compound is hydrogenated and, moreover, the presence of naphthalene prevents even the exchange of the benzylic hydrogens as the DPE recovered contains very little deuterium. From a thermodynamic standpoint, it should be easier to hydrogenate one ring in naphthalene than the isolated phenyl rings of DPE. It . 948 I I Run 1 Run 2 Run 3 Activation Time 42 h .16 h 16 h Activation Temp. 330 OC 430 OC 430 OC Run Time l h l h 7 h / Run Temp. 1 350 OC I 35OoC 1. 350 OC Yield CPEa I 30.4% I 22.6%e I 37O/oe I Recovery DPEC 11 Yield DCEb I 5% I 5%: 1 13%: II 64% I 72Yog 50 %g ACKNOWLEDGEMENTS The authors gratefully acknowledge a grant from the United States Department of Energy, Pittsburgh, DE-FG22-91-PC91291, supporting this work. REFERENCES 1. Rajagopal, V.; Guthrie, R. D.; Shi, B.; Davis, 6. H. Prepr., Div. Fuel Chem., Am. Chem. SOC. 1994, 39, 673. 2. Guthrie. R. D.; Ramakrishnan, S.; Britt, P. F.; Buchannan, Ill., A. C.; Davis, 6. H. Prepr., Div. fuel Chem., Am. Chem. SOC. 1994, 34, 668. 3. Bittner, E. W.: Bockrath, B. C.; Solar, J. M. J. Catal 1994, 749, 206 4. Guthrie, R. D.; Shi. B.; Sharipov, R.; Davis, 6. H. Prepr. Div. fuel Chem., Am. Chem. 5. Information provided in product bulletin by Cabot Corporation for Cab-0-Si1 M-5. 6. Dent, A. L.; Kokes, R. J. J. Am. Chem. SOC. 1970,92, 6709-6718. SOC. 1993, 38, 526-533. 949 HYDROGENATION/DEHYDROGENATION OF MULTICYCLIC COMPOUNDS USING ATTM AS CATALYST PRECURSOR Richard P. Dutta and Harold H. Schobert Fuel Science Program Pennsylvania State University University Park, Pa 16802. Keywords: Hydrogenatioddehydrogenation; kinetics; thermodynamics. Introduction Coal liquefaction can be considered a viable technical alternative for production of advanced fuels if the coal macromolecule can be broken up into low molecular weight fragments and hydrogenated to decrease the concentration of aromatics in the final product. Previous studies have shown that the initial breakdown of coal can be achieved using various catalysts and various conditions. However, if the final product is to be a very high quality distillate, the coal liquids still need funher hydrotreatment if they are to be satisfactory. One way to improve the quality is to add another step to the liquefaction process. This would employ a very active catalyst to hydrogenate the products from the first liquefaction stage. However if operating costs are to be kept to a minimum, it would be advantageous to hydrogenate the coal fragments as they are being released during the first stage of liquefaction. Burgess has shown that ammonium tetrathiomolybdate (A'ITM) can be used as a catalyst precursor ina process for conversion of coal to a "proto-jet fuel" [I]. Coal conversion up to 95% were observed but the products were aromatic and contained some phenols. Various temperature strategies have been formulated for coal liquefaction. The majority of these strategies are concerned with the depolymerisation of coal and the avoiding of retrogressive reactions. Another important aspect of temperature strategies is the thermodynamic behavior of released coal fragments, With careful 'fine tuning' of the reaction conditions, it could be possible to have advantageous thermodynamics in the system along with reasonably fast kinetics of depolymerisation of the coal macromolecule. Basically, a trade-off between kinetics and thermodynamics is possible. Model compound studies can, be used to understand the fundamental behavior of coal fragments during coal liquefaction and coal liquids upgrading. The literature on hydrogenation of model compounds is vast and has been recently reviewed by Girgis [Z]. For the past several years we and our colleagues have been investigating the hydrogenation and dehydrogenation chemistry of a variety of polycyclic compounds. This work has aimed at investigating some of the fundamental chemical processes involved in various aspects of fuel utilization. The compounds investigated have included decalin and tetralin [3], anthracene [4], phenanthrene (5.61, pyrene [6,7] and chrysene [6]. This paper will discuss the hydrogenation and dehydrogenation behavior of naphthalene and pyrene. Kinetic and thermodynamic parameters will be calculated from product distribution trends. From these parameters it should be possible to outline a possible reaction strategy that allows all these compounds to remain in their hydroaromatic states during a coal liquefaction operation. Experimental All reactions were carried out in 25ml microautoclave reactors (made of type 3 16 stainless steel). In all runs, 3+0.01g naphthalene or pyrene (Aldrich, 99%. used as received) and 0.075M.005g ammonium tetrathiomolybdate (Aldrich, used as received) were weighed into the reactor. The reactor was then evacuated and pressurized with hydrogen to 7MPa. Heating was accomplished by lowering the reactor into a fluidized sand bath preheated to the desired temperature. After a measured reaction time, the reactor was quenched to room temperature by immersing it in a cold water bath. The products from the reaction were removed from the reactor using THF. The THF was removed by rotary evaporation and the product was weighed. It was found that in all cases the weight of the products equaled the weight of the original compound before reaction. The products were dissolved in acetone and analysed using a Perkin-Elmer 8500GC. In order to determine the dehydrogenation behavior of the hydrogenated pyrenes, the products from pyrene hydrogenated at 350°C and 60 minutes, 4WC and 80 minutes, and 450°C and 40 minutes, were catalytically dehydrogenated under Nz. This was accomplished using the same reactors as in the hydrogenation step. The products from the three hydrogenations listed above were weighed into the reactor along with a 1 wt% (metal) loading of A'ITM. The reactor was pressurised with approx. 3MPa N2 and immersed in a sand bath at the desired temperature and for the desired reaction time. After this time, the reactor was quenched as before, and the products were removed using THF. The products were analysed using GC as before. The dehydrogenation behavior of tetralin was investigated in a similar way to the hydrogenated pyrenes. Results and discussion 1. Naphthalene Thermodvnmcs. Figure 1 shows the product distributions of naphthalene hydrogenation at 350, 400 and 450'C for various reaction times up to 3hrs. In all cases only tetralin was detected as a 950 \ \ \ I \ hydrogenation product of naphthalene. No decalin was observed. Cracking/isomerisation products of tetralin were observed at 450'C. but total concentration did not exceed 5wt%. At 450'C. conversion of naphthalene to tetralin reaches a maximum at 51%. At 4WC, conversion is 62% and at 350'C the reaction does not reach equilibrium, but after 3hrs conversion is 72% (to calculate Kp for 350'C reaction, extrapolation used to assume 95% conversion). Figure 2 shows how the product distribution at equilibrium varies with temperature. From these equilibrium compositions, Kp values were calculated as below: Kp = Itetralinleq. [naphthalene]eq.[H~p ressure12 Table 1 reports the Kp values for naphthalene hydrogenation. As expected, Kp decreases with increasing temperature. This is because thermodynamics controls the extent of the reaction as the temperature increases. The variation in Kp with temperature can be used to find the enthalpy of the reaction. This is done using the van't Hoff isochore equation: d In Kp/dT = AH/RT* A van? Hoff plot gives a value of -32 kcalhol. This is in agreement with values given in the literature (-29-32 kcal/mol) [8,9]. 1. Reversible reaction kinetics of the forward and reverse rate constants: l€&ic% This can be modeled as a first order process with an effective rate constant equal to the sum - dCA/dt = kfCAPH2" - k&N where CA = concentration aromatic at time t, CN = concentration hydroaromatic at time t, kf = forward rate constant and kr = reverse rate constant. On integration, the following expression is derived: In CA - CAe 1 CAo - CAe = -(kf+kr) t where CAO is the initial aromatic concentration and CAe is the equilibrium aromatics concentration. 2. Irreversible reaction kinetics. to obtain the forward rate constant: If equilibrium effects are negligible, a simple pseudo-first order kinetic model can be used If values of the above equation are plotted upto the time where equilibrium effects the reaction, a good approximation of kfcan be obtained. Table 1 shows the calculated values of the rate constants calculated from plots of the above kinetic equations. Figure 7 shows an Arrhenius plot of the calculated rate constants. An activation energy of 14.7 kcaVmol for the forward reaction, and 25.5 kcaYmol for the reverse reaction were calculated. From this plot, dehydrogenation would be favored over hydrogenation at a temperature of 416°C. !hW If hydroaromatics are to be produced from aromatics. two factors have to be considered. I. Conversion. 2. Length of time to get to the desired conversion level. As can be seen from the data, conversion decreases with increasing temperature, but the kinetics of the reaction are slower at lower temperatures. From the data, it can be concluded that high temperatures are desirable for the first 40 minutes of reaction, but after this time thermodynamics limit the conversion. At this point, it is then advisable to drop the temperature to below 400°C. and continue to convert naphthalene to tetralin as seen in the 350'C reaction. To hydrogenate only at 350'C would take too long to achieve respectable conversions, i.e. conversion at 350°C and 120 minutes is the same as 450°C and 60 minutes. Therefore in hydrogenating naphthalene to tetralin a reverse temperature stage reaction is proposed. Stage I . 4OO'C and 40 minutes reaction time. Stage 2.350'C and 60 minutes reaction time. Dehydrogenation reactions of tetralin Figure. 3 shows the product distribution of tetralin dehydrogenation at 350,400. and 450'C for reaction times up to 30 minutes. As temperature increases, the rate of dehydrogenation increases. and the conversion of tetralin to naphthalene also increases. At 350 and 400°C. conversion to naphthalene does not exceed 13%. but at 450'C conversion is 42%. This explains , . ICS vs T h e m o d v n m 951 the rapid approach to equilibrium seen in the hydrogenating reactions and the relatively low conversions seen at the high temperature of 450°C. 2. Pvrene I ~- -stribution of pyrene. dihydropyrene, tetrahydropyrene and hexahydropyrene are shown in figures 4 and 4b. From these product distributions it can be seen that temperature is affecting the conversion of pyrene to hydrogenated pyrenes. At 450°C. equilibrium is reached after 20 minutes. with 28% conversion of Dvrene. At 4WC. eauilibrium is reached after 80 minutes. ~~~~ ~. with 45% conversion of pyrene. At 33&C, equilibrium is iot observed, even after 120 minutes of 'reaction. Conversion at this point is 55% pyrene to hydrogenated pyrenes. These product distribution trends are similar to that observed for naphthalene hydrogenation in that as temperature increases. conversion decreases but the rate of reaction to equilibrium increases. Figure 5 shows the equilibrium composition of pyrene and total hydrogenated pyrenes. Kp values are reported in table 1. Kp decreases with increasing temperature. These values can be used to determine the enthalpy of reaction as described earlier. A value of -6.4 kcaVmol is obtained from a van't Hoff plot. This value is a reasonable comparison to the value obtained by Johnston (-10 kcal/mol)[lO]. Kinetics. A similar model is used for evaluation of pyrene kinetic data as was used for naphthalene. Rate constants ar reported in table I . Figure 8 shows an Arrhenius plot for the calculated rate constants. An activation energy of 6.83 kcal/mol for the forward reaction, and 21.5 kcaVmol for the reverse reaction were calculated. From this plot, dehydrogenation would be favored over hydrogenation at temperatures above 350°C. Kmetics vs thermodvnarmcs. stage reaction is proposed for pyrene hydrogenation: Stage 1.400'C and 20 minutes reaction time. Stage 2. 350'C and reaction time set for the desired conversion. Dehydrogenation of hydropyrenes. Figure 6 show the dehydrogenation product distributions of dihydropyrene, tetrahydropyrene and hexahydropyrene. It can be seen that dehydrogenation is rapid and complete at 450'C. At the lower temperatures, dehydrogenation is slower and complete dehydrogenation to pyrene is not seen in the 30 minutes reaction used in this study. Comparisons between naphthalene ind pyrene as ring size increases, enthalpy of reaction increases and activation energy decreases. Future Work The work will be expanded to include 3-ring systems and other 4-ring compounds. When the parameters are calculated for these compounds and plotted vs ring size, molecular weight etc., it should be possible to make predictions as to how other compounds behave under hydrogenating/dehydrogenating conditions. Ideal temperature strategies will be estimated from the product distribution curves and compared for the different compounds. References 1. Burgess, C. PhD Thesis 1994, Pennsylvania State University. 2. Girgis, M. J. and Gates, B. C. Ind. Eng. Chem. Res. 1991, 30, 2021. 3. Song, C. and Hatcher, P. G. Prepr. Pap.- Am. Chem. SOC., Div. Pet. Chem. 1992.37, 529. 4. Song, C., Ono, T. and Nomura, M. Bull. Chem. SOC. Jpn. 1989,62, 680. 5. Song, C., Schobert, H. H..Matsui. H., Prep. Pap.- ACS, Div. Fuel. Chem.. 1991, 36(4), 1892. 6. Dutta, R., BSc Thesis. 1991, Nottingham Polytechnic. 7. Tomic, J., PhD Thesis. 1993, Pennsylvania State University. 8. Frye, C. G. J. Chem. Eng. Data. 1962, 7, 592. 9. Frye, C. G. and Weitkamp. A. W. J. Chem. Eng. Data. 1969,14, 372. 10. Johnston, K. P. Fuel.1984, 63, 463. Table 1. Kinetic and thermodynamic parameters for naphthalene and pyrene hydrogenation . . The same arguments apply for pyrene as they did for naphthalene. A reverse temperature Table 1 shows a comparison of the parameters for the two compounds. It can be seen that IK~ ~p ~p Umin-1 k/min-l Wmin-1 AH z.; 350C 400C 45OC 35OC 4OOC 45OC kcdmol kcdmol Naphthalene 0.0045 0.00037 0.00027 kf 0.0059 kf 0.0150 kf 0.0301 -32 kr 0.0027 kr 0.0125 kr 0.0194 kr 0.0100 kr 0.0342 kr 0,1103 Ppne 0.0078 0.0058 0.0038 kf 0.0088 kf 0.0139 kf0.0187 -6.4 6.8 952 80 70 60 50 @' 40 30 20 IO 0 0 50 100 150 200 timelmin Figure I . Naphthalene-tetralin product distribution vs time 100 60 5 1: 0 --8- naphthalene ! 99 %) and it was used without further purification. Model compound reactions A reactor with a capacity of 33 mL was loaded with ca. 0.25 g NMBB, 1 wt % catalyst Precursor (1 wt % Mo based on NMBB) and 0.14 g solvent (tridecane). When water was added, the molar ratio of H20 to NMBB was 10, corresponding to a wt ratio of HzOMMBB of 0.56. The reactor was purged three times with H2 and then pressurized with 6.9,MPa H2 at room temperature for all experiments. A preheated fluidized sand bath was used as the heating source and the horizontal tubing bomb reactor was vertically agitated to provide mixing (about 240 strokedmin). After the reaction the hot tubing bomb was quenched in cold water. The liquid contents were washed with 15 ml CHCIj through a low speed filter paper for qualitative and quantitative GC analysis of the filtrate. All runs were carried out at least twice to confirm reproducibility. When sulfur was added, the atomic ratio of S:Mo was 4:l. The products were identified by GC-MS using a Hewlett-Packard 5890 IJ GC coupled with a HP 5971A mass-selective detector operating at electron impact mode (EI, 70 eV). The column used for GC-MS was a J&W DB-17 column; 30-m X 0.25-mm. coated with 50 % phenyl 50 % rnethylpolysiloxane with a coating film thickness of 0.25 fim. For quantification, a HP 5890 I1 GC with flame ionization detector and the same type of column (DB-17) was used. Both GC and GC-MS were temperature programmed from 40 to 280 "C at a heating rate of 4 "Chin and a final holding time of 15 min. The response factors for 10 of the products were determined using pure compounds. More experimental details may be found elsewhere (8). RESULTS AND DISCUSSION NMBB Reaction at 350 "C Effect of Precursor Type and S Addition Table 1 presents the results of non-catalytic and catalytic runs of NMBB with dispersed catalysts at 350 "C. NMBB is essentially inert at 350 "C under H2 pressure in a non-catalytic run. A'ITM showed remarkable catalytic effect on NMBB conversion at 1 wt % Mo loading. The main products are 4-methylbibenzyl (4-MBB) and naphthalene, which were formed from cleavage of bond a in NMBB. It is clear that the molybdenum sulfide in situ generated from AlTM at 350 "C is catalytically active, and can promote the cleavage of C-C bond a in NMBB. A?TM decomposition also generates extra sulfur. However, our results in Table I shows that adding sulfur alone, or HzO alone, had little effect on NMBB conversion. The material in situ generated from Mo(C0)6 at 350 'C acted as a hydrogenation catalyst. The dominant product with Mo(CO)~is tetrahydro-NMBB (TH-NMBB). Sulfur addition to Mo(CO)6 increased NMBB conversion significantly, from 50.8 to 94.3 %. Adding sulfur also changed the product distribution pattern. The major products with Mo(C0)6 + S are 4-MBB and naphthalene arising from cleavage of bond a. The mn with Mo(CO)6 + S also produced considerable amounts of bibenzyl and methylnaphthalene, probably via cleavage of bond b in NMBB. Figure 1 compares the product distribution for runs at 350 "C. An interesting result was found in the run with Mo(CO)6 at 350 "C. Most of the total conversion of 50.8 % can be attributed to the formation of TH-NMBB derivatives (45.5 mol %). This finding suggests that under low severity reaction conditions the initial step in hydrocracking of NMBB is the addition of hydrogen. Several TH-NMBB derivatives (MW 326) can be detected in the GC-MS analysis, indicating hydrogenation of different aromatic moieties in the model compound. At elevated temperatures activated Mo(CO)6 cleaves NMBB completely; no more TH-NMBB derivatives can be detected, as described later. Effect of HzO Addition The addition of H20 to AmM enhanced NMBB conversion and increased the yields of 4- MBB and naphthalene. Therefore, the co-use of ATTM and water appears to be beneficial for NMBB hydrocracking at 350 "C. However, adding H20 to Mo(CO)6 decreased NMBB conversion to the level close to a non-catalytic run. This indicates that added H20 either inhibited the formation of a catalytically active phase or passivated the active sites on the surface of the active phase or reacted to form some kind of catalytically inactive material. However, adding HzO to Mo(CO)~k s system did not have significant effects on NMBB conversion or product distribution. It is interesting to note that H20 addition to the catalytic runs with either ATTM or Mo(C0)6 + S system did not alter the product distribution pattern, suggesting that the added water did not alter the reaction pathways in these cases. NMBB Reaction at 400 "C Table 2 shows the results for non-catalytic and catalytic runs of NMBB at 400 "C. NMBB is not very reactive in a non-catalytic run at 400 "C under H2 pressure, as its conversion is below 4 %. sulfur, however, began to show catalytic effect when the temperature is increased from 350 to 400 "C. Both ATTM and Mo(C0)6 afforded higher conversion of NMBB at 400 OC than the corresponding runs at 350 "C. ATTM alone is a more effective catalyst precursor than MO(CO)~ done, in terms of higher NMBB conversion (93.0 vs. 79.6 %). Addition of water to A n M in the mn at 400 "C, however, had negative impact on NMBB conversion. These results are consistent . 969 with those for catalytic hydroliquefaction of coal, where H20 addition had a strong promoting effect for mns at 350 "C, but inhibiting effect for runs at 400 "C (6). It appears from our results that water has two opposing effects on NMBB conversion at 350 and 4M) "C. Possibility exists that the ratio of water to catalyst is also influential. Farcasiu et al. (9) reported that NMBB cleavage at 420-430 "C with various dried iron oxide precursors were different from rehumidified catalysts. Addition of small quantities of water increases, to some extent, the catalytic activity. Completely rehumidified iron oxides showed very low catalytic activity compared to partially hydrated iron oxide. The activity of the system as an acidic catalyst is destroyed by larger amounts of water (longer rehumidification time). Figure 2 further compares the product distribution for runs at 400 "C. For the runs with ARM and ATTM+H20,4-MBB and naphthalene are the major products. In the case of Mo(CO)6, the yield of tetralin is higher than that of naphthalene. Apparently, the activity and selectivity of a dispersed Mo catalyst for NMBB hydrocracking depends on the catalyst precursor type and reaction conditions. Since it is the precursor that was charged into the reactor, an activation into catalyst is involved during the heat up and the subsequent reaction. It is known (3.5.10) that the S-free catalyst precursors like metal carbonyls require the addition of sulfur for sufficient activity in coal liquefaction; activation of A'ITM into the catalytically active species (close in composition to MoS2) occurs at a temperature of 2325-350 "C. This temperature range was used in our model reactions. Sulfur addition to Mo(CO)6 generates MoS2 after high temperature activation. The resulting product distribution at 350 OC is very similar to runs with A'ITM (Table 1). We assume that the active catalytic species is similar to that from ATTM. Unlike ATTM, the organometallic complex Mo(CO)6 decomposes at much lower temperatures. The active catalyst particles will be readily available under the conditions employed. This may rationalize why the NMBB conversion is higher with Mo(CO)6 than with ATTM at 350 "C. However, for runs at 400 "C, the NMBB conversions with A'ITM and Mo(CO)6 are similar to each other (8). With respect to the effect of the catalyst loading level, we have reported some results on NMBB hydrocracking over dispersed catalysts at 2.11 wt % metal loading (8). Decreasing Mo loading level from 2.11 wt % to 1 wt % (this work) did not have negative impacts on NMBB conversion with ATTM, but caused some changes in product distribution from NMBB with Mo(C0)6. CONCLUSIONS Dispersed fine particles in situ generated from either water-soluble precursors such as A'ITM . or oil-soluble precursors such as Mo(CO)6, can he effective Mo catalysts for promoting the cleavage of certain C-C bonds such as bond p in NMBB at 350-400 "C. When the sulfur-free precursor is used, adding sulfur helps to improve catalytic activity, particularly hydrocracking activity. When ATTM is used at low temperature (350 "C), adding water seems to be beneficial in improving NMBB conversion. ACKNOWLEDGMENTS We wish to thank Dr. H. Schobert for his encouragement and support. This project was supported by the U.S. Department of Energy, Pittsburgh Energy Technology Center under contract DE-AC22-92PC92122. We are grateful to Dr. U. Rao of PETC for his support. We also thank Mr. R. Copenhaver for the fabrication of reactors. REFERENCES 1. Bergius, F. and Billiviller, J. German Patent No. 301.231, Coal Liquefaction hocess, 1919. 2. Mochida, I. and Sakanishi, K. Advances in Catalysis, 40, 1994, 39-85. 3. a) Artok, L.; Davis, A.; Mitchell, G. D.and Schobert, H. H. Energy & Fuels, 1993,7, 67- 17. b) Garg, D. and Givens, E. N. Fuel Process. Technol., 8,1984. 123-34. 4. a) Hirschon, A. S.; Wilson Jr., R. B. Cod Science II , ACS Sym. Ser., 1991, 273-83. b) Hirschon, A. S.; Wilson Jr., R. B. Fuel, 71,1992, 1025-31. 5. a) 0. Ymada, T. Suzuki, J. Then, T. Ando and Y. Watanabe, Fuel Process. Technol., 11, 1985, 297- 311. b) T. Suzuki, T. Ando and Y. Watanabe, Energy & Fuels, 1,1987, 299-300. c) S. Weller, Energy Fuels, 8 , 1994, 415-420. 6. Song, C. and Saini, A. K. Energy & Fuels, 9,1995, 188-9. 7. a) Bockrath, B. C.; Finseth, D. H. and Illig, E. G. Fuel Process. Technol., 12, 1986, 175. b) Ruether, J. A.; Mima, J. A.; Koronsky, R. M. and Ha, B. C. Energy & Fuels, 1,1987, 198. C) KhYa. y.; Nobusawa. T. and Futamura, S. Fuel Process. Technol., 18, 1988, 1. 8. a) Schmidt, E. and Song, C. Prepr. Pap. - Am. Chem. Soc.. Div. Fuel Chem., 35,1994 733-737. 970 b) Song, C.; Schmidt, E. and Schobert, H. H. DOE Coal Liquefaction and Gas Conversion, Contractors' Review Meeting in Pittsburgh, (September 1-8, 1994), 593-604. 9. Farcasiu, M.; Smith, C.; Pradhan, V. R. and Wender, I. Fuel Processing Technology, 29, 10. a) Song, C., Parfitt, D.S.; and Schobert, H.H., Energy & Fuels, 8,1994, 313-9. a) Song, C.; Nomura, M. and Miyake Fuel, 65,1986, 922-6. C) Song, C.; Nomura, M. and Ono, T. Prepr. Pap. - Am. Chem. SOC. Diu. Fuel Chem., 1991, 199-208. 36(2), 1991, 586-96. Table 1: Effect of S and H2O on hydrocracking of NMBB at 350 "C. aMethyltetrahydronaphthalene. *when S was added, the atomic ratio S:Mo was 4: 1. Table 2: Effect of Mo-based catalyst precursors on hydrocracking reactions of NMBB at 400 "C. aMethylteWahydronaphthalene, *when S was added, the atomic ratio S:Mo was 4 1. 971 Runs at 350 "C with added H20 Figure 1. Effect of S and water on hydrocracking of NMBB Runs at 400 'C m y . 100 .E ATTY.HIO. 400 .c Y ~ I E O ~4.0 0 .c Figure 2: Effect of Mo-based catalysts on hydrocracking of NMBB. 972 CATALYTIC CONVERSION OF POLYCYCLIC AROMA'ITC HYDROCARBONS: BASED CATALYST TO ALKYLARENES. Tom Autrey, John C Linehan, Donald M Camaioni, Tess R Powers, Eric F McMillan and James A Franz Pacific Northwest Laboratory, P.O. Box 999, RichIan4 WA 99352 USA MECHANISTIC INVESTIGATIONS OF HYJlROGEN TRANSFER FROM Ah' IRONKey Words. Catalysis, mechanism, hydrogen transfer Mnduction. model compounds has been demonstrated to increase the efficiency of liquefaction during the early stages of catalytic coal hydrokatment.' Despite numerous model compound studies, the mechanism of "liquefaction" remains controvemid. Wei and c o w ~ r k m ~ ~ ~ pmposed a hydrogen atom - ipso displacement pathway, however, this pathway alone cannot explain the observed selectivity! Farcasiu and co-w0rkers7 p p s e d a mechanism in which the alkyl-arene moiety is activated to undergo bond scission by electron transfer to the catalyst. However, a key step in this pathway, unimolecular scission of the radical cation, has been argued to be kinetically and thmcdymmcally unfavorable? We recently reprted a beneficial charactaistic of the FdS cataly* generated in situ by the reaction of sulfur with iron oxyhydroxides produced by the RTDS process: scission of strong carbon-carbon bo& withotd addition of hydi.ogen g a and with minimum fonnarion of light gaws?9 We investigated a series of mom, di-, and trimethyldiphenylmethanes and found tha! in all cases the benzyl group is prefmtially displaced. We proposed a variant of the Farcasiu mechanism' in which a hydrogen atom is transferred to the ipso position of the radical cation, which then scissions a benzyl cation. Back electron transfer h m the reduced catalyst surface to the benzyl cation would give the benzyl radical! We refer to this radical ion mechanism as "ET/H"'. The mechanism is consistent with (1) the pmpsed redox properties of the cataly*' (2) the obmed "dealkylation" selectivity, benzyl >> methyl, and (3) the low yield of tmndkylation products. Recent examination of the structure - reactivity relationships for the methylated diphenylmethanes shows the rate of catalytioinduced bond scission correlates not only with the ease of m e ox idation but also comlates with the stability of the ipso radical adduct.'!' To accOmmOdate the observed selectivity for benzyl scission >> methyl scission by a radical pathway, we propsed a reversible hydrogen atom transfer between the catalyst surface and the arenes, in which case, the rate of back hydrogen transfer h m th e ipso and nonipso adducts must be fast co@ to scission of methyl radical. between the ETm radical ion pathway and the reversible hydrogen atom transfer pathway. Experimental. catalyst p m o r . " 9,lO- dihydrophenanthme (DHP), xanthene, and ohyhydiphenylrnethane were purchased hm Aldrich. The DHP was distilled and recrystallized h m methanolldichlomethane. 1,2ditolylethanol was available h m a previous study.12 o- and p benzyldiphenyl ether were prepared by the same methd used for our synthesis of the alkyldiphenylrnethanesmethanes.'T)h e isomers were separated on a chromotron@ eluting with pentane. the sulfur (3 mg), and the DHP solvent (100 mg) were loaded into the glass tubes and sealed under vacuum. The thermolysis was canid out in sealed 5-nun 0.d. borosilicate glass tubes immersed in a fluid& sand bath regulated at 400°C for 1 h. The Gc and Gc/us analysis were carried out as described previously." Results and Discussion. Given the parallel shucture - reactivity trends for ion, radical, and radical-ion intermediates in the alkylarene series previously examined, we prepared a new series of model compunds to discriminate between the ion, radical, and radical ion hydrogen transfer pathwayj. A comparison of diphenyl ether analogs with our diphenylmethane model compounds was suggested as an approach to obtain insight into the proposed multi-step ETJHA mfer step or a free radical @way.14 The cation formed by ipso addition of Hatom to the radical cation of diphenyl ether (DPE) was suggested to be more resistant to bond scission than the analogous cation obtained h m diphenylmethane @PM) beoluse of the differences in the stability of the leaving pup, PhO(+) << PhCH2(+), while the The utility of iron-based nanophase catalysts in the liquefaction of coal and coal In the present study, we designed and prepared model compounds to differentiate Matedals. All catalytic experiments used the RTDS-prepared, &line fenihydrite ~ S t d k sThe. m odel wmropound (1 5 mg), the &line fmihydrite (3 rng), 973 opposite selectivity was predicted for a radical @way, PhO-, > PhCH2-. U n f ~ w l y , while the pscission of a phenoxy cation h m DPE is expected to be significantly slower than the pscission of a benzyl cation hin DPM the first step, oxidation of the DPE, is faster than oxidation of DPMI5 Thus, an "external" comparison of DPM and DPE derivatives could complicate direct kinetic comparisons. To alleviate this conam, we used model compounds that had both the PhG and PhCH2- substituents in a single molecule - xanthone @A), pbenzyldiphenyl ether (PBDPE), and 0-benzyldiphenyl ether (oBDPE). This approach avoids complications caused by diffktcnm in oxidation potential of diphenylmethane and diphenyl ether and takes advantage of the selectivity differences between scission of PhO* and PhCHz*.16 Thus, the appearance of PhCH,Ph would be consistent with a &n-oxygen (PhCH2Ph-OPh) ike radical bond scission pathway, and the appearance of PhOPh would be consistent with an appanmt carbon-carbon (PhCHz- PhOPh) cationic bond scission pathway. starting material in 60 minutes with the formation of DPM, DPE, toluene and phenol. The ratio of DPmPE (Le. &n-oxygd&n-wrbon bond scission) is 8:l. This mult offers significant insight into the mechanism of catalytioinduced bond scission. The selectivity of the 8: 1 ratio for scission of benzyl over phenoxy radical h m pBDPE is significantly less than expected for pscission of PhCH2(+) over PhO(+)." Therefore, we have considerably less confidence with the involvement of a cationic intermdate formed either by acidic proton transfer or multi-step ET/HA pathways for promoting bond scission. What is interesting about these results is that the selectivity is the oppsite of the relative stabilities of the benzyl and phenoxy radicals. TherefOK bond cleavage must not be the rate-limiting step for reaction of this molecule. Ihe selectivity is consistent with ratelimiting formation of the ipso adducts. The stability of the ipso adduct, (a), leading to formation of DPE and the benzyl radical is 3 Q kcallmolls more stable than the ipso adduct, (b), leading to formation of DPM and the phenoxy radical (Scheme 1). Thermolysis of oBDPE under the same d o n co nditions again yields DPM, DPE, toluene and phenol, however, the ratio of DPIWDPE is 1:l. The apparent lower selectivity observed h m the catalytic thermolysis of oBDPE could be due to a competing neophyl-lie phenyl migration, 1,5 addition, in Scheme 2. These Arl-5 radical reanrtngements are horn to occur,19e specially whcn heteroatom termini are involved.m21 Tautormerization of the phenol, followed by unimolecular scission,n can yield diphenylmethane by an alternative pathway. phenylxanthene - formed fiom addition of the diphenylmethyl radical to the ortbposition, 1,6 addition, followed by disproportionation-is detected in the thermolysis of oBDPE. This provides further evidence for the presence of the precursor to the neophyl rearrangement pathway under the reaction conditions." 'Ihermolysis of xanthene under the same catalytic reaction conditions yields little bond scission &er 60 minutes, no detectable 2-methyldiphenyl ether, and only traces of 2- bond scission products, toluene and phenol. Here, the absence of significant quantities of scission products in the xanthene thermolysis is probably due to competing reversible reactions that regenerate the starting material. Because the leaving group is "attached," little bond scission is observed. ynimolecular S s i s i ~ ~ QEafrly ~mec~han.isti c studies demonstrated the preference for catalytic-induced scission of diqhethane l i g e s over the thermally labile bibenzylic linkages in 4(l-naphthylrnethyl)bibenzyl (NMBB). Farcasiu and coworkers invoked single electron oxidation of the m e fo llowed by a unimolecular scission to yield naphthalene and Cmethylbibenzyl (referred to as -A- bond cleavage).' However, the observed selectivity is the opposite of that expected based on reactions in solution or reactions in the gas phase based upon Sagmentation reactions of the NMBB radical cation in a mass spectr0meer.S Almost as surprising as radical cation cleavage at the diarylmethane -A- bond is the suggestion that the positive charge is carried on the naphthyl not the benzyl group, given the difference between stabilization of benzylic cation and a benzylic radical. To favor -A- bond scission over -D bond scission, the catalyst must uniquely stabilize the naphthyl cation and/or destabilize the benzyl cation. This novel unirnolecular scission pathway is reported to be suppotted by atom superposition and electron delocalization molecular orbital (ASEDMO) methcds, however, AM1 and MNW theoretical methods indicate cleavage of the -D bond is favored25 cation dissociation mechanism. Although the bond dissociation energy @DE) for bibenzyl radical cation, at about 30 kdrnol, is substantially weaker than the bibenzyl bond we expect that -D bond scission in the NMBB radical cation will be a minimum of 40 kdmol, given that oxidation of I-methylnaphthalene is ca 10 kcaVmol more favorable than oxidation of pxylene, and comatively assuming no barrier for the ET process that Thermolysis of pBDPE in DHP- FdS at 400 "C leads to ca 70% consumption of the The expected side product h m this radical reanangement pathway, 9- NMBB probably is not the best model compound to test the unimolecular radical 974 I genaates the radical cation. A bania of this magnitude probably cannot compete with bn~lecularr eactions of the radical cation or alternative fke radical pathways, scheme 3. A more judicious choice of model cornpmds, for example one with a much lower radical cation BDE, could provide support for the proposed electrun transfer pathway if the bond were hken more rapidly than purely thermal pathways allow. ~phenylethanol radical cation has a BDE of 15 kdmol and therefore is expeckd to dissociate ca. 8 orders of magnitude faster than the NMBB radical cation at 400°C, and will be more likely to compete with other pathways. Thennolysis of D E in DHP for 60 minutes at 400°C mlts in ca 5Wh conversion to yield 4,4'-pdimethylbibenql, pxylene, 4-methylbenzylalcohol, 4-methylbenzaldehyde and a tlace of toluene. The mtio of toluene topxylene is 1:25. The 4,4'-pdiiethylbibenzyl is formed by a reduction pathway that competes with unimolecular scission. Since the 4,4'-pdmethylbibenzyI is thermally stable under the reacton conditions, the xylene is predominately formed h m unimolecular thermal scission of the starting material. . Thermolysis of D E in DHP containing the FdS catalyst for 60 &Utes at 400°C results in complete conversion to yield 4,4'-pdimethylbibenzyI, pxylene, and toluene. The pxylene is most likely formed fiom the thermal background and reduction of the alcohol and aldehyde since no oxygen-wntaining products are detected. The most significant finding is the ratio of toluene to p q l e n e has i n d to 1: 1. The presence of the FdS catalyst pathway apparently increases the yield of toluene! Toluene is not a product expected hm single electron oxidation of the diphenylethan~l.'~ The formation of toluene under the catalytic conditions is more rationally explained by an ipso hydrogen displacement pathway. Hydrogen atom addition to the phenyl ring a- to the hydroxy group leads to the pscission of a stabilized ketyl radid. As we have previously dosaved efficiency of bond scission is strongly dependent on the stability leaving group." If the radical cation of DTE was formed under the reaction conditions, instantanenus unimolecular scission of the bibenzylic bond would have occurred to yield pxylene as the major product. While we are convinced DTE is an improved probe molecule for investigaiing the radical ion pathway, we would l i e to investigate betta models, e.g. the methyl ether of DE, that are not expected to be reduced under the reaction conditions. We are confident that electron transfer f?om the akylarenc to the catalytic surface does not occur, otherwise we would have observed much higher yields of xylene. Admittedly, oxidation of this xylene derivative will be more endergonic than oxidation of the naphthyl moiety in NMBB, however, the selective catalytic pathways seem to operate even for Single ring model compounds.69 We cannot guess what the ASEDMO methods would lind for single electron txansfer cleavage pathways of DE, but experiment^'^ and AM1 calculations26 suggest highly efficient bibmzylic bond scission. Summary and Conclusions. lhe results of our model compund studies suggest that fiee radical hydrogen transfer pathways h m the catalyst to the akylarene are responsible for the scission of strong carbonsarbon bonds. mere are two requisites for the abed selective bond scission. First, and most importantly is the stability of the ipso adduct precursor leading to displacement, the more stable the adduct the more probable bond scission. 'Ihis explains why benzyl radical displacement > phenoxy radical displacement in benzyldiphenyl ether and explains why PhCH2CH2PhCH2ra dical > naphthylmethyl radical h m N MBB. Second, given "equal" ipso adduct precursor stabilities, e.g. methyldiphenylmethane, the stability of the departing radical determines the selectivity. This explains benzyl radical > methyl radical in the methylated diphenylmethanes and explains why a-hydroxyphenethyl radical > methyl radical in 1,2ditolylethanol. We have assumed little physical interaction between the molecules and the catalytic surface and have been able to satisfactorily explain most of the observed selectivity. However, for IWBB we expect a higher selectivity for -A- bond scission relative to -5 bond scission, given the ca 6 kdmol Merence between the radical adduct formd by the hydrogen atom addition to 1-methylnaphthalene and pxylenc. It is possible that physical properties play a role in lowering the selectivity in NMBB bond scission. Also, we realize that catalysts p r e p a r e d by other methods may contain different activity sites and operate by different mechanisms. We used 1,2-pditolylethanol (DE)as a probe for the electron transfer mechanism I 915 Acknowledgment This work was supported by the US. Department of Energy, office of Basic hergy Research Chemical Sciences Division, Prccm and Techniques Branch. The work was conducted at Pacific Northwest LaboratoIy, which is operated by Battelle Memorial InstiMe for the U. S. Department of Energv under Contract DEACO6-76Ru) 1830. We thank Dean Matson for OUT supply of the worlds best catalytic precu~sors. Support for TRP and EFM was provided h u & AWU-NW under grant DE-FGO6-89ER-75522 with the U.S. Department of herpy. Glossary of Acronyms. BDE bond dissociation enera RTDS DHP 9,l O-diiydmphenanhe pBDPE pbenzyldiphenyl ether oBDPE ebenzyldiphenyl ether DPM diphenylmethane DPE diphenyl ether DTE 1,2ditolylethanol NMBB 4-( 1 -naphthylmethy 1)bi~enzyl References and Notes. Rapid T h d Decomposition of precursOrs in Solution 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. See the first 21 papas in Energv & Fuels, 1994,1, and references thmin. We, X-Y, Ogata, E., Niki, E., BUN Chem &e. Jm 1992, 65, 1 114. Wei, X-Y, Ogata, E., Nib, E., Chemistry Letters, 1991,2199. Wei, X-Y, Ogata, E., Zong, EM, Niki, E. Energy Fuels 19!?2,6, 868. Wei, X-Y, Zong, SM, Energy Fuels 1992, 6,236. F r a q J.A.;C amaioni, D. M; Alnajjar, M S.; Autrey, T.; Lmehm, J. C. Am Chem Soc, Dv Fuel Chem Preprints, 1995,40(2) 203. Anaheim, Ck Farcasiu, M; Smith, C.; pradhan, V. R; Wader, I. FuelPrm. Tech. 1991,29, 199. L m e h J. C.; Matson, D. W.; Darab, J. G.; Autrey, S. T.; Franz, J. A; Camaioni D. M Am Chem Soc, Div Fuel Chem Preprints, 1994,39(3) 720. Washington D. C. Autrey, T.; Camaioni, D. M; Lmehm, J. C.; Wartob, H. W.; Franz, J. A. in preparation. To be submitted to Enersy Fuels. Matson, D. W.; Linehan, J. C.; lhab, J. G.; Budder, M F. Energv Fuels 1994, 8, 10. Camaioni, D. M; Franz, J. A. 1 Org Chem. 1984,49, 1607. See reference 9 for general preparation. A few drops of sulfiuic acid was added to a solution of diphenylether and benzyl alcohol in methylene chloride. The reaction was stirred at room temperature for 24 h. Distillation under vacuum yielded both the ortho and para isomers. McMillen, D. M; Malhotra, R Am Chem Soc, Dlv Fuel Chem Preprints, 1995, 4q2) 221. Anaheim, CA. IP of DPE (8.09eV) DPM(8.55eV) Lias, S. G.; Bartmass, J. E.; Liebman, J. L.; Holmes,J.L.;Levin,RD.;Mallard,W.G.JChemP~s.Ref:Data,1988,17, suppl. * designates either radical or radical ion. Personal communication with Ripu Malho@ the difference in appearence potentid for oxidation of PhO* -> PhO(+) and PhCHp -> PhCH,(+) is ca. 34 kcal/mol. Assuming a ABDE PhGH and PhCH,-H of 3-4 kcaVmol this predicts a selectivity of several ordm of magnitude even at 400°C. McMillen, D. F.; Golden, D. M. Ann Rev. Phys. Cheq 1982, 33, 493. Wmtein, S.; Heck, R; Lappork, S.; Baird, R Eprientiu, 1956, 12, 138. Kochi, J. K;Gilliom., R D. 1 Am Chem Soc. 1964,86, 5251. DeTar, D. F.; Hlynsky, A. 1 Am Chem Soc. 1955, 77,4411. McMillen, D. F.; Ogier, W. C.; Ross, D. S. 1 @. Chem 1981,46, 3322. Thermolysis of o-bismethyldiphenyl ether appearj to give *xylene under the catalytic conditions. This product can only come &om a Ar,-CS neophyl rearrangement. Frgatnentation of NMBB by 70 eV electron impact in a mass spectrometer leads to a Sagmentation pattern that has the charge on the -Db ond scission pmdua.~n, o charge is detectable on the naphthalene hgment: MS m/z 322 (kft, 50); 231(100); 215 (40); 127 (0); 91 (15). Ada, H. F., C~mPanion, L. Subbaswamy, K. R Energv Fuels 1994, 8, 71. Camaioni, D. M1 Am Chem Soc. 1990, 112,9475. P w J. H; J-h. Energv Fuels 1994, 8, 421. 976 I \ 4 I J f Scheme 1 Ph d 0 \ P h - PhnPh + PhO' @) Scheme 2 QfJJ-m \ \ Ph Ph t 1.6 addition Ph bh I bh Ph I Do DH .,,Ph L Ph Ph Scheme 3 NapAPh - * L P h ~ J ~ ~ ~ ~ L : T \ c'a Nt a aly p sC A P L p h -D- bond scission *CHfPh 1 ? j -A- bond scission competing 'I pathway C WPh NapnTh cn,t Naphthalene + I *CH2\ph Naphthalene 977 EFFECT OF MODIFIER Pd METAL ON HYDROCRACKING OF POLYAROMATIC COMPOUNDS OVER Ni-LOADED Y-TYPE ZEOLITE AND ITS APPLICATION AS HYDRODESULFURIZATION CATALYSTS Takema Wada, Satoru Murata, and Masakatsu Nomura, Department of Applied Chemistry, Faculty of Engineering, Osaka University, 2-1 Yamada-aka, Suita, Osaka 565, Japan Keywords: hydrocracking, hydrodesulfurization, metal-supported zeolite catalyst INTRODUCTION Coal tar obtained from coal carbonization is a treasure of polyaromatic hydrocarbans, where more than 400 kinds of aromatic compounds are found to be contained. Naphthalene’s content in coal tar is about 9.0 %, being used as starting materials for phthalic anhydride, dye staff, pharmaceutical products and synthetic resins. On the other hands, phenanthrene and pyrene are contained in the yield of 5.0 % and 2.1 %, respectively, being used only for production of carbon black and antiseptics of timbers. Application of these three or four ring aromatic compounds for stating materials of fine chemicals is not yet developed so extensively. The development of new catalysts being able to convert these aromatics into mono or diaromatic compounds is one of objectives for utilization of polyaromatics. Hydrocracking of polyaromatic compounds is believed to proceedvia formation of terminal-naphthenic ring of starting aromatic compounds, followed by cleavage of the naphthenic ring to produce alkylated aromatic compounds which has less numbers of ring than starting aromatics. Accordingly, hydrogenation of aromatic rings and cracking of resulting naphthenic rings are key steps of hydrocracking reaction, so that dual functional catalysts such as metal-supported acid catalysts are considered to be one of the best catalysts!-2 Zeolite has controlled pore structures and strong acidity enough to crack naphthenic rings, being characteristics in exchanging metal species with ease. We have been studying the hydrocracking of polyaromatic compounds over Ni-loaded zeolite catalysts (ZSM-5, mordenite, and Y-type) and found the fact that pore size of zeolite exerts an interesting effect on product distribution? We also conducted computer-simulation for diffusion phenomena of the polyaromatic hydrocarbons in the pore of these zeolites and found that diffusion ability of the substrate affects strongly the product distribution! Recently we found that modifying of Ni-loaded Y-type zeolite by Pd-loading enhanced hydrocracking ability of the catalyst. In this report, we would like to refer to the results of both hydrocracking reaction of pyrene and hydrodesulfurization of dibenzothiophene using Pd-modified Ni-loaded Y-type zeolite. Experimental Section heoaration of metal-SUDDOrted zeolites The NH,-substituted Y-type zeolite (50 g) was stirred in 1000 ml of aqueous solution of Ni(NOJ, (0.25 M) at 90°C for 96 h, then being filtered and dried at 1 10°C to obtain the nickel cation substituted zeolites. As to the Pd supported one, NH,-substituted Y-type zeolite (15 g) was treated in aqueous solution (200 ml) of [Pd(NHJ,](NO,), (0.025 M) at 40 “C for 24 h. As to Ni-Pd-Y catalysts, NH,- substituted Y-type zeolite was treated with aqueous [Pd(NHJ,](NO,), solution, followed by aqueous Ni(NO,), solution. The resulting cation exchanged zeolites were calcined in a stream of air at 500°C for4 h, being submitted to the reduction with \atmosphere at 450 “C for 1 h. The content of nickel and palladium in each zeolite was determined by using a Rigaku Denki System 3270 type fluorescence X-ray analyzer, this being summarized in Table 1. Hydrotreatment of pvrene or dibenzothiophene The substrate (I g) and the catalyst (0.5 g) were placed in a 70 ml SUS 316 autoclave, which were pressurized to 70 kg/cm2 with hydrogen, being followed by heating up to 350°C with the rate of 8 “C / min. Reaction time is the duration being kept at 350°C. After the gaseous product was collected, its aliquot was submitted to GC analysis with a Shimadzu GC-3BT (active carbon column. 2 m) and a Shimadzu GC-8AIT (silica gel column, 60/80 mesh, 3 m). The liquid product was recovered by washing the inside of the autoclave with CYCI,. According to the analysis by a JEOL JMS-DX- 303HF type GC-MS, components of liquid products were assigned, their quantitative analyses being conducted by a Shimadzu GC-14APFSC (CBP-I capirary column, f 0.5 mm x 25 m). The carbon deposited on the catalyst was calculated based on the microanalysis of the recovered catalyst. Results and Discussion Hvdrocrdcking reaction of ovrene In a previous study, we conducted hydrocracking reaction of phenanthrene and pyrene over three 978 \ / J I different Ni-supported zeolite catalysts such as Ni-loaded ZSM-5, mordenite, and Y-type zeolite at 350°C for 1 h at 70 k&m2 of hydrogen and found that pore size of zeolites is controlling product distribution. Conversion of the substrates depended also on the size of these zeolites. By the use of Ni-loaded Y-type zeolite (Ni-Y) catalyst, phenanthrene was completely converted to gases and oneor two-ring compounds, while in the case of pyrene both hydrogenated pyrene and unreacted one Still remained in the product. This results could be interpreted by the fact that the pore size of Y-type zeolite is larger than the molecular size of phenanthrene, however, it is somewhat smaller than that Of pyrene. In order to improve the activity of Ni-Y catalyst, we examined modifying of Ni-Y catalyst by loading second metal species. We prepared Pd-modified Ni-Y (Ni-Pd-Y) catalyst by ion exchange of Nq-substituted Y-type zeolite with aqueous [Pd(NH,),I2' solution followed by NiZ+s olution. Concentration of metal species on the resulting zeolite is summarized in Table I . To examine catalytic activity of Pd metal itself, we also prepared Pd-loaded Y-type (Pd-Y) zeolite. Using these three catalysts, hydrocracking of pyrene was conducted in a 70 mL autoclave at 350°C for 1 h at 70 kg/cm2 of hydrogen, the results being shown in Figure 1. In the presence of Pd-Y or Ni-Pd-Y catalyst, pyrene was completely converted to gases (from methane to butane) and derivatives of benzene or cyclohexane, especially in the case using Ni-Pd-Y catalyst, yield of gases reached to 70%. Amounts of carbon deposited on the catalyst were 9.1% for Ni-Y, 9.6% for Pd-Y, and 7.0% for Ni-Pd-Y catalyst. Figure 2 summarizes the distribution of mono-ring compounds produced from hydrocracking of pyrene over three catalysts. In the case using Ni-Y catalyst, the selective formation of each compound was not observed, while, the yield of cyclohexanes was higher than that of bezenes in the case using Pd-Y and Ni-Pd-Y catalysts. No qand C,-benzenes and C,-cyclohexanes was observed in products obtained from the reaction using Ni-Pd-Y catalyst. These results indicated that activity of Ni-Pd-Y catalyst toward hydrogenation and cracking reaction is the highest among three catalysts. This finding agrees well with the fact that the ratio ofi-butane to n-butane (2.1) of gaseous products in the reaction using Ni-Pd-Y catalyst was slightly higher than that (1.8) from the reaction using Ni-Y catalyst. This higher activity of Ni-Pd-Y catalyst might be partly due to the decrease of carbon deposited on the catalyst, this often leading to deactivation of the catalyst. Effects of reaction temperature and duration on the extent of hydrocracking were also investigated, the results being shown in Figure 3. In the reaction at 350°C for 0 min, 25% of pyrene was still remained and main products were hydrogenated pyrenes (60%). With the reaction time being longer than 30 min, no pyrene was recovered and gases and mono-ring compounds became main products. In the reaction at lower temperature such as 325"c, conversion of pyrene was still higher than that in the reaction using Ni-Y catalyst at 350°C. These results suggested that modifying of Ni-Y catalyst by Pd can reduce the reaction temperature of hydrocracking of pyrene compared with Ni-Y catalyst. -In a previous chapter, we found that modifying of Ni-Y catalyst by Pd-loading resulted in very high activity for hydrogenation or hydrocracking of polyaromatic hydrocarbons. So, we tried to apply this modified catalyst for hydrodesulfurization (HDS) of dibenzothiophene (DBT). HDS reaction of DBT was conducted at 300°C for 1 h under 70 kglcm2 of H,. Figure 4 shows the product distribution of HDS reaction by using Ni-Y and Ni-Pd-Y catalysts. In the case using Ni-Y catalyst, 15% of DBT was recovered along with 12% yield of sulfur-containing compounds, while, using Ni-Pd-Y catalyst, DBT was almost converted to gases and monoring compounds and no sulfur-containing compound was observed in the liquid products. These results suggest that high activity of Ni-Pd-Y catalyst is much more effective for HDS reaction. Now, we are conducting characterization of this active catalyst. This work was supported by Grant-in-Aid for Scientific Research No. 07242249 from the Ministry of Education, Science and Culture, Japan. 4. REFERENCES 1) 2) 3) 4) Hayes Jr., H. W.; Parcher, J. F.; Halmer, N. E. Ind. Eng. Chem. Process Des. Dev., 1983.22, 401. Lapinas, A. T.; Klein, M. T.; Gates, B. C.; Macris. A.; Lyons, J. E./&. Eng. Chem. Res., 1987, 26, 1026. Matsui, H.; Akagi, K.; Murata, S.; Nomura, M.J. Jpn. Petrol. Insr., in confribution. Matsui, H.; Akagi, K.; Murata. S.; Nomura, M. Energy Fuels, in press. 919 Table 1. The catalyst employed in this study Contents (wt%) Ni Pd Ni-Y 5.5 - Ni-Pd-Y 3.3 3.6 Pd-Y - 3.7 naphthalenes and tetralins hydrogenated Ni-Y Pd-Y Ni-Pd-Y 0 20 40 60 80 100 Yield (wt%) Figure 1. Hydrocracking of pyrene over metal-suppotted Y-type zeolites at 350 OC for 1 h Ni- Pd-Y Ni-Pd-Y Figure 2. Distribution of mono-ring compounds in hydrocracking of pyrene over three metal-supported Y-type zeolite catalysts at 350 OC for 1 h under 70 kglcm' of H, 980 naphthalenes and tetralins hydrogenated pyrenes unreacted gases \ phenarenes coke pyrene J 325 OC, 15 350 OC, o 325 OC, 60 350 OC, 15 350 OC, 30 350 OC, 60 0 20 40 60 80 100 Yield (wt%) Figure 3. Hydrocracking of pyrene over Ni-Pd-Ycatalyst ies biphenyls and phenylcyclohexanes b z e n e s \\ cyclohexanes thiophenols and DBT derivatives gases coke S in coke Ni-Pd-Y k i . 1 0 20 40 60 80 100 Yield (wt0/.) Figure 4. Hydrodesulfurization of DBT over metal-supported Y-type zeolite at 350 OC for 1 h 981 SELECTIVE HYDRODESULFURIZATION OF 4,L-DIMETHYL DIBENZOTHIOPHENE IN THE MAJOR PRESENCE OF NA_P_H__T_H_A LENE OVER MOLYBDENUM BASED BINARY AND TERTIARY SULFIDES CATALYSTS Takaaki ISODAS',h inichi NAGAOX, iaoliang MA, Yozo KORAI and lsao MOCHIDA * Institute of Advanced Material Study, Kyushu University, Kasugakouen 6-1. Kasuga, Fukuoka 816, Japan Keywords : Selective HDS, Ru-CoMo / A1203, 4,6-dimethyldibenzothiophene INTRODUCTION It has been clarified in previous papers (I) that the sufficient desulfurization of the refractory sulfur species in the diesel fuel such m 4-methyldibenzothiophene and 4,6- dimethyldibenzothiophene (4,6-DMDBT) should be achieved in its sulfur level of 0.05 70 which is currently regulated (2). Such refractory species have been proved lo be desulfurized through the hydrogenation of one or both phenyl rings in the substrate to moderate the steric hindrance of inethyl groups located in the neighbours of the sulfur atom (3). Hydrogenation of neighboring phenyl rings should competitive with the aromatic partners in the hydrogenation step. being severely hindered (4). The diesel fuel tends to be more aromatic due to crude of more aromaticity and more blending of the craked oil. In the present study. the catalytic activities of CoMo and NiMo of different contents of Co or Ni and Ru-CoMo I A1203 were examined for the desulfurization of 4,6-DMDBT in decane and decane with naphthalene lo find selective catalysts which desulfurize 4.6-DMDBT in the presence of naphthalene through preferential hydrogenation through its phenyl ring. The key step is assumed lo be the hydrogenation step of4.6-DMDBT in the competition with naphthalene. Ru was the selected as the third component of metal promoter which was added to CoMo / Al2O3, since sulfur atom in 4.6-DMDBT can be a prefereable anchor to the nobel metal sulfide in competition with the aromatic hydrocarbon (5-6). Characterization of Ru-CoMo / A1203 using XPS, XRD and HREM to find the origin of the selectivity for the desulfurization in the aromatic hydrocarbons. EXPERIMENTALS Chemicals and Catalysts : 4.6-DMDBT was synthesized according to the reference (7). Commercially available (NH&Mo04. Co(N03)26H2O, Ni(N03)26H20 and RuC1.~3H20w ere used as catalyst precusor salts. Commercially available A1203 was selected 3s the catalyst support. The precusor salt was impregnated onto AI203 according to an incipinent wetness impregnation procedure. Impregnation additives such as HCI, H3PO4. malic acid and citric acid are used in the impregnation solution. Contents of metal oxides supported were 3s follows; Co(0 - 3 wt%)-Mo(lS wt%) / A1203. Ni(0 - 5 wt%)-Mo(lS wt%) I A1203. Ru(0.75 wt %)-Co(O.ZSwt%)-Mo(lS wt%) / A1203, respectively. After the impregnation, the catalyst was dried at 160"C, calcined at 420'C under air now. and presulfurized at 360°C for 2h by flowing H2S (5vol %) in H2 under atomospheric pressure just before its use. Reaction ; HDS of 4.6-DMDBT in decane with naphthalene was performed in a SO 1111 batch-autoclave at 300°C under 2.SMPa H2 pressure for 0.5 - 23h, using 1.5g catalyst and log substrate including solvent. The concentrations of 4,6-DMDBT and iiaphthalene were 0.1 and10 wt %, respectively. After the reaction, products were qualitatively and quantitatively analyzed by GC-MS, GC-FID (Yanaco G-3800 and G- 100) and GC-FPD(Yanaco G-3800 and S0inl OV-101). XPS : X-ray photoelectoron spectra (XPS) was taken on a ESCA 1000 (Shirnazu co.) with Mg Ka radiation energy of 1253.6 eV. The binding energy was identified according to the references (8- I I ). 982 i XRD ; X-ray diffraction (XRD) was taken on a X-ray diffraction meter (Rigaku co.) with Cu bget electrode at 40 KV voltage. X-ray diffraction was performed according to the. procedure described by International Center for Diffration Data (12). HREM ; High resolution electron micrographs (HREM) of catalysts were taken on JEM -2000 EX (Jeol co.) at 200 KV accelation voltage at magnification of 1 SO to 500K. RESULTS Inhibition with Naphthalene of €IDS Reaction over NiMo and CoMo catalysts Fig.l(A) illustrates HDS activity of NiMo catalysts for 4,6-DMDBT in decane and decane with IO wt% naphthalene at 300°C under 2.5MPa H2 for 2h. The catalysts gave 95 - 10%co nversion of 4.6-DMDBT in decane regardless of Ni content which increased very much the desulfurization compared to that over Mo / A1203 in decane. Naphthalene of IO.wt% reduced the conversion to 45% on Mo / A1203 and Ni( I)-Mo I A1203. and 77% on Ni(S)-Mo / A1203. Significant retardation by naphthalene was observed over Ni-Mo / A1203 especially when Ni content was low. Fig2 shows the HDS products from 4.6-DMDBT over Mo and NiMo / AI203 NiMo / AI2O3 produced B4,6 as the major product and A4.6 as the second major while Mo / A1203 did A4.6 as the major and 84.6 as the second major in decane. Overall desulfurization was certainly enhanced by addition of Ni through the hydrogenation. Naphthalene of IO wt% reduced very much the desulfurization over Mo / AI203 and Ni( I)-Mo /AI2O3. Mo / A1203 and Ni-Mo / A1203 with less content of Ni suffered large reduction of both products by IO wt% naphthalene. Large amount of Ni increased the 84.6 overcoming the inhibition of naphthalene while A4.6 suffered very much the retardation regardless of Ni content. Fig.l(B) shows conversion of naphthalene to tetralin and decalin. Ni of I wt% or more addition accelerate very much the hydrogenation of naphthalene, giving 100% tetralin and 25% decalin. while Mo / A1203 provided SO% conversion to tetralin without decalin. Table I summaries the yield of minor products, hydrogenated one (H) and desulfurized one without hydrogenation (C4.6). Addition of Ni increased the yield of C4.6 and reduced that of H. Fig.3(A) illustrates the HDS of 4.6-DMDBT over CoMo / AI203 in decane and decane with IO wt% naphthalene. The CoMo catalysts allowed 100% conversion of 4.6-DMDBT regardless of Co content under the present conditions in decane. Naphthalene of IO wt% reduced the conversion to 7.5% in Co(0.25)Mo / A1203. 87% over Co( I )Mo /A1203. and 68% over Co(3)Mo / AlzO3. It should be noted that Co( I)Mo / A1203 surfered the smallest retardation by naphthalene. Fig.4 shows yields of products from 4,6-DMDBT over CoMo / AI203 of different Co contents. B4.6 was the major product, of which yield increased markedly by addition of Co to Mo, reaches the maximum of SO% by Co of I wt% in decane and 60% in decane with IO wt% naphthalene. Addition of Co increased very sharply the yeild of B4.6 the major product. I wt% of Co giving the maximum yield. A4.6 the second major product, was produced most over Mo / Al203. Addition of Co reduced it sharply in decane. Yield of A4.6 decreased markedly in decane with 10% naphthalene over Mo I A1203, while addition of 0.25 and I wt% of Co slightly increased it. 90 naphthalene over Ru(0.75 wt%) wt%)-Mo / Al203, respectively, by 2h. High HDS conversions of 9S - 100 % were obtained over all three catalysts in decane. Naphthalene of IO wt% in decane reduced the HDS conversion over the catalysts, however the extent reduction depended in the amounts of Co or Ni. Co content of 0.7.5 wt% exhibited 22% reduction while Co and Ni of 3 and 5 wt% severe suffered 30 and 39 % reduction, respectively. Hydrogenation conversion of coexistent naphthalene is shown in Fig.5 (B). Ru- NiMo / A1203 provide the highest conversions of naphthalene to tetralin and decalin of 100% and 20%. respectively, while the conversions were 90% and IO%, respectively, over Ru-Co(0.25 wt%)Mo / A1203. Hence, Ru-CoMo / A1203 with 0.2.5 wt% Co exhibited the highest HDS activity of 4,6-DMDBT by minimum hydrogenation of coexistence naphthalene. 983 Hydrodesulfurization and Hydrogenation Selectivities Fig.6 compares the conversions of 4.6-UMUBT hydrodesulfurization and naphthalene hydrogenation over NiMo, CoMo and additive with Ru(0.75 wt%)-C0(0.25 ~ 1 % ) - Mo( IS wt%) / AI203 at 300'C for 0.5h and 2h where 0.1 wt% 4.6-DMDBT and IO wt% naphthalene were present in decane. There were found two kinds of the catalysts : one exhibited a larger conversion of naphthalene with much less conversion of 4,6-DMDBT, and the others were large conversions of 4.6-DMDBT. The first group of cayalyst contained NiMo, the second one contained Ru-CoMo and Ru-CoMo with additive H3P04 and HCI. In order to compare the reaction selectivity between the HDS for sulfur compound and hydrogenation (HGN) for aromatic hydrocarbon, relative selectivity was introduced according to the equation (I). Selectivity ratio is shown as follows ; Selectivity ratio of 4.6-UMDBT = (Reaction mole ratio of 4.6-DMDBT over NiMo / A1203 or Ru-CoMo / A1203 ) / ( 1 ) Fig.7 shows the selectivity ratio of 4.6-DMDBT calculated from data of Fig.6 versus conversion of naphthalene. It clarified hydrogenation activity for naphthalene decreased with increasing the selectivity ratio of 4,6-DMDBT over these catalysts. Particular Ru- CoMo-HCI / A1203 showed the highest hydrodesulfurization selectivity for 4,6-DMDBT, giving ratio of I .41, while giving ratio of I .28, I . I2 and I ,O over Ru-CoMo / A1203. Ru-CoMo-P / A1203 and CoMo / AlzO3, respectively. NiMo / A1203 was inferior to CoMo / A1203 for hydrodesulfurization selectivity of 4,6-DMDBT, giving ratio of 0.38. Especially, Ru-CoMo-HCI / A1203 showed as 1.7 times higer selectivity for hydrodesulfurization of 4.6-DMDBT as that of NiMo / A1203, while 0.4 times lower hydrogenation activity of naphthalene. XPS Analysis Fig.8 shows XPS of Mo 3d in the Ru(x)-Co(y)-Mo( ISwt%) I A1203 (0 S x. y 5 I wt %) catalysts before and after presulliding. Before sulfiding, two Mo 3d 3/2 and 3d 5 ~ 2pe aks were found at 235 and 232 eV of binding energies (8-1 I), indicating Moo3 species in all catalysts regardless of them compositions. Sulfiding sharply the peaks to 22.5 and 222 eV, respectively in Mo / A1203 and CoMo / A1203. indicating the Mo(ll) species (8- I I ). Ru-CoMo / A1203 and Ru-Mo / AI203 exhibited two peaks at 222 and 218 eV with a very small peak at 22.5 eV. indicating major presence of Mo 3d 5/? of Mo(0) at 218 eV after the sulfiding. These resuts indecated two kinds of Mo species, such as MoS2 and metal Mo existed on sulfided Ru-CoMo / A1203 catalyst. XRD Analysis Fig.Y shows XRD spectra of a series of Ru(x)-Co(y)-Ma( IS wt%) / AI203 (0 _< x. y 5 I wt %) before presulfiding. There were two large peaks ascribed to alumina of 45.7' and 663". respectively , with all catalysts. Three sharppeaks were identified with MOO, of 23.4'. 2.5.6" and 27.3". respectively, over the Mo based on catalysts. The intensity of these three peaks increased with increasing content of Ru on CoMo / AI203 indicating the large crystals of Moo3 in the presence of Ru. The peaks of 33.7' and S3.9' were identified to Ru02, while no peak was ascribed to Co oxide. No definite peaks related to Mo, Ru and Co species were found after the sulfiding. HREM Fig.10 shows HREM micrographs of sulfided Ru(0.75 wt%)-Co (0.25 wt%)-Mo (IS wt%) / Al2O3. Fig. IO (a) and (b) shows MoS2 layers and RuS2 crystals as dotted spots, respectively, under 20K magnification which were typically observed on Ru- CoMo / A1203. Large magnification of (a) under 50K clarified large length and thickness of MoS2 layers. DISCUSSION (Reaction mole ratio of 4.6-DMIIBT over CoMo / A1203) Fig.1 I illustrates the reaction pathway of 4.6-UMUBT in decane ovre Mo sullidc based on cataysts. There were two desulfurization routes, one is the desulfurization through the hydrogenation of one phenyl group; i.e.. hydrodesulfurization route. and the other is desulfurization without apparent hydrogenation; i.e., direct-desulfurization route. The former reaction route is strongly hinderd by the dolninamt presence of naphthalene. \ \ 004 I In earlier works, it has been proposed desulfurization active site of sulfided CoMo and NlMo catalysts were anion-vacancy at edge plane of MoS2 ( 13). Voorhoeve and Stuiver Proposed "Intercalation model" which Ni and Co located edge plane of MoS2 (14), Delmon proposed "Contact synergy model" which high activity brought contact between liny Co& and MoSz crystal (15). and Tops0e insisted the mechanizum of high activity by "Co-Mo-S phase model" which located on edge plane of MoS2 (16). Additive co-catalysts. such as Ni or Co brought the high activity of MoS2. The role of Ru addition to Mo sulfide based on caytalysts are classified two categories, reduction of hydrogenation activity for aromatic hydrocarbon and promotion of hydrogenation selectivity for sulfur compound. In addition XPS spectra indicated additive Ru in CoMo catalyst was easily reduced MoS2 to metal Mo. Hence hydrogenation activity of aromatic hydrocarbon was controled by additive Ru, and it suggests compeatitive reaction of 4.6-DMDBT with naphthalene on hydrogenation active site was relieved. Crystals of RuSz and (Co)-MoS2 were found existing separately by HREM, it m y be suggest 4,6-DMDBT takes precedence over the hydrogenation of naphthalene on the RuS2* and hydrogenated 4.6-DMDBT was completely desulfurized over (Co)- MoS2. In order to design of the higher selective hydrodesulfurization catalyst, it will be necessary to high dispersion of Ru on the surface of support, and optimization of the amount of CoMo and Ru on catalyst. Other side of the aspect. it will be worthwhile to try the hybrization of Ru/ A1203 and CoMo / A1203. LITERATURE CITED (1) Isoda. T.. Ma. X.. Mochida. 1.. J.Jpii. Per. ln.\r..37. 368 (1994). (2)Takatuka. T.. Wada. Y.. Suzuki. H.. Komatu. S.. Morimura. Y.. J.Jpii. Per. Insr.. 3.7. 197 (1992). (3) Isodn. T.. Ma. X., Mochida. I.. J . J p Prr. Insr., 37. SO6 (1994). (4) Isoda. T.. Ma. X., Nagno. S.. Mochida. I., J . J p . Per. Inst., 38. 25 (1995). ( 5 ) Isoda. T.. Kisamori. M.. Ma, X.. Mochida. I., Abstract of Symposium on Jpn.Pet. Inst.. p.42. 17 - (6) Isoda. T.. Ma. X., Nagao. S.. Mochida. I.. Abstract of Symposium on Jpn. Pet. Insl.. p.716. 26 - (7) Gerdil. R.. Lucken. E.. J.Am. Chem Sor.. 87. 213 (196.5). (8) Chung.P.L.. David, M.H.. J. P/iys. C/ieni..B. 4.56 (1984). (9) Gajardo. P., Mathieux, A.. Grange. P.. Delmon. B.. Appf. C[im/.. 1, 347 (1987). (IO) Walton. R.A.. 1. Cflro1..4$ 488 (1976). ( I I ) Ledoux. M.J.. Hantzer. S.. Guille. 1.. Bull. Soc. Cbem. Bclge.. 26. 8SS (1987). (I21 International cenue for diffiaction data. "Inorganic Phases", (1989). (13) Prins,R.. De Beer, V.H.J.. Somurjai. G.. Girol.Rev. Sci.Eng..& I (1989). (14) Voorhoeve. R.J.H.. S1uiver.J.C.M.. J.C~irfl/.,2.l2,2 8 (1971). (15) Grange. P. , Delmon. B.. 1. Le.\.\ Common Met.. &, 353 (1974). (16) Topsw. N.Y.,Topsoe. H.. 1. C[ird., 84, 386 (1983). 18 May 1994. Japan. 27 Ocl. 1904. Japan. 985 .- 0 0 0 1 2 3 4 5 0 1 2 3 4 5 NiO (wt"/.) (A) 4,6-DMDBT (B) Naphthalene Fig.1 Inhibition with naphthalene of HDS reaction over NiMo catalysts. (300eC-2.5MPa, 4,6-DMDBT 0. I wt% + Nap IOwt% in decane, Catalyst content; 15 wt% ) to tetralin 1 A s? Y C 0 0s) c 0 0 .- E IO0 50 n naphthalene - 0 1 2 3 0 1 2 3 coo (Wh) (A) 4,6-DMDBT (6) Naphthalene Fig.3 Inhibition with naphthalene of HDS reaction over to tetralin CoMo catalysts. I 1 In decane in decane with 1Owt% naphthalene fw I in decane with 1Owt% 0 1 2 3 0 1 2 3 coo (WtVO) (A) 8 4 6 (6) A4.6 Fig.4 Major products from 4,6-DMDBT over CoMo catalvsts. 986 I I -1 00 -100 C C 0 v) 50 2 50 Q) > C > C 0 0 g E .0- .- ii " 0 " 0 (A) 4,B-DMDBT (e) Naphthalene to tetralin -: in decane Kl , conversion of ndphthdhe : n + 5wl% Naphthalene = to tetralin : conversion of produced retralin to decalin ; n' + lOwt% * Fig.5 Inhibition with naphthalene for the HDS reaction of ' 4,6-DMDBT over Ru(0.75)-Co(0.25)-Mo-P, Ru(0.75) -Co(3)-Mo and Ru(0.75)-Ni(5)-Mo / AlzOs. (300°C- 2.5MPa-2h, 4,6-DMDBT O.lwt% +Nap lOwt% in decane, Catalyst content; 15 wt% ) 0 2 1: Ni-Mo 2: CO-MO 3 RU-COMO-H~PO~ 4 Ru-COMO 5: Ru-COMO-HCI 0.5 40 60 80 100 Conversion of naphthalene (Yo) Fig.7 Effect of the additive on the HDS selectivity ratio of 4.6-DMDBT r (a) (b) under 20K magnification (c) I d I, 4nK II Fig.10 HREM micrographs of sulfide Ru(0.75)-Co(0.25)-Mo(I5 ) / AIaO, __ 1 Fig. I I Reaction pathway of 4,6-Dimethyldibenzothiophene over Mo sulfide based on catalyst. ( 300C-2.5MPa) 988 SELECTIVE HYDRODESULFURIZATION OF 4,B-DIMETHYLDIBENZOTHIOPHENE IN THE DOMINANT PRESENCE OF NAPHTHALENE OVER HYBRID CoMo I AI203 AND Ru I A1203 CATALYSTS Takaaki ISODA*, Shinichi NAGAO, Xiaoliang MA, Yozo KORAI and lsao MOCHIDA * Institute of Advanced Material Study, Kyushu University, Kasugakouen 6- I , Keywords : selective HDS, Ru / A1203 catalyst, 4.6-dimethyldibenzothiophene Kasuga. Fukuoka 8 16, Japan INTRODUCTION It has been revealed that significant desulfurization of refractory 4-methyldibenzothiophene and 4.6-dimethyldibenzothiophene (4.6-DMDBT) is very essential to achive the low sulfur level of gas oil requested by the current regulation ( I ) . Their direct desulfurization through the interaction of their sulfur atom with the catalyst surface is sterically hindered by its neighbouring methyl groups. The substrate is found kinetically to be hydrogenated at one of its phenyl rings prior to the desulfurization in order to reduce the steric hindrance through non-planaring configuration (2-4). NiMo / A1203 was reported to be superior to CoMo / AI203 in the deep desulufurization. because of its higher hydrogenation aciivity (2). However, such a hydrogenation route suffers severe inhibiiion by aromatic species in their dominant presence (3). because 4.6-DMDBT must compete with the aromatic species to the hydrogenation sites on the catalysts. The aromatic species up to 30 wt % in the gas oil was that completely stop the desulfurization of ihe particular substrate (3). The catalyst for the selective hydrogenation of 4.6- DMDBT in ihe dominant aromatic partners is most wanted to achive its extensive desulfurization in the pas oil, although there have been reported activitics of various transition metal sulfides for HDS of dibenzothiophene (5). and hydrogenation of aromatic hydrocarbons (6). The present authors have reproted that different hydrogenation selectivity for 4.6- DMDBT and naphthalene over mixed sulfides of molybdemum and other transition metals (7). Ru-CoMo / A1203 catalyst which was impregnated from aquation HCI of Co, Mo and Ru salts showed four times higher hydrogenation selectivity for 4,6-DMDBT than NiMo I AI203 catalyst(8). In the present study, HDS of 4,6-DMDBT in decane containing a significant amount of naphthalene was examined over a hybrid of CoMo / A1203 and Ru / A1203 to design the selective hydrogenation and succsesive desulfurization of 4.6-DMDBT in an aromatic moiety. Its activity was compared to those of CoMo / A1203. NiMo / A1203 and Ru / AI203 in their single use. I I lp I (' I' t EXPERIMENTALS Chemicals and Catalysts ; 4.6-DMDBT was synthesized according to the reference (9). Commercially available (NH&MoOd, Co(N03)26H20, and RuC13 3H20 were used as catalyst precusor salts. A1203 as the catalyst support was commercially available. The precusor salt was impregnated onto A1203 according to an incipinent wetness impregnation procedure. Content weight of metal oxide on each catalyst as follows ; Co(0.25 wt%)-Mo(l5 wt%) IA120.1. Ni(lwt%)-Mo(lS wt%) / AI203 and Ru(6 wt%) / A1203. respectively. After the impregnation. the catalyst was dried at 160'C. calcined at 420'C under air flow, and presulfurized at 360°C for 2h by flowing H2S (5 vol %) in HZ under atomospheric pressure just before its use. Reaction ; HDS of 4.6-DMDBT in decane with naphthalene was performed in a 50ml batch-autoclave at 300°C under 2.5MPa H2 pressure for I .O - 2.Sh. using I .Sg catalyst and log substrate including solvent. The concentrations of 4.6-UMDBT and naphthalene were 0.1 and10 wt %. respectively. After the reaction, products were qualitatively and quantitatively analyzed by GC-MS, GC-FID (Yanaco G-3800 and G-100) and GCFPD( Yanaco G-3800 and 50ml OV-101). k 989 RESULTS HDS Activity of 4,6-DMDBT Fig. I (A) and (B) illustrates the conversion of 4.6-DMDBT and naphthalene versus reaction time. respectively, over CoMo / A1203. NiMo I Al203, Ru / A1203 and a hybrid of CoMo I AI203 and Ru I A1203 at 300°C. CoMo I A1203 exhibited an excellent activity for HDS of 4,6-DMDBT. giving conversions of 46% by I h and 74% by 2h ils shown in Fig. I(A). The particular NiMo I AI203 was inferior to CoMo I A1203, giving conversions of 24% by I h and 47% by 2h. Ru I A1203 was very inactive for HDS. giving conversions of 6% by Ih and 8% by 2h. The hybrid showed the highest activity for HDS of 4.6-DMDBT among the catalyst examined, giving conversions of 71% by Ih, 87% by 2h and 90% by 2.Sh. when 20 wt% of Ru I A1203 and IS wt% CoMo / AI203 were used. NiMo I A1203 showed high activity for the hydrogenation of naphthalene. giving conversion of 90% by I h. Tetralin and decalin were the products. their yields of the latter produced being 6% by I h and 18% by 2h, respectively. CoMo / AI203 and its hybrid with Ru / A1203 exhibited similar activities, being much inferior to NiMo / AI203 to give conversion of 61 and 77% by 1 h, respectively. Decalin of 80% produced by I h over CoMo / Al203,5% by I h over the hybrids. Ru / A1203 was very inactive, giving a conversion of 10% by I h and 23% by 2h. Products frum 4,6-I)MDBT Fig.2(A) and (B) illustrates the product yields from 4,6-DMDBT over the CoMo / A1203, NiMo I A1203. Ru / A1203 and a hybrid of CoMo I A1203 and Ru I AI203 at 300 'C. The major products were hydrodesulfurization products B4.6 and A4,gespectively. B J , a~n d A4.6 were produced through the hydrogenation of one or both phenyl ring in 46DMDBT. In addition, hydrogenation products of H and desulfurized product C4,6 were also found in minor yields, being produced through the hydrogenation of one phenyl ring and successive direct sulfur elimination, respectively. by the yields over CoMo I A1203. NiMo / A1203. Ku / AI203 and hybrids with 10 and 20 wt% Ru I AI2O3, respectively. as suininnrized in Table I . CoMo I A1203 and the hybrid provided H4 by the yields of 24 and 3590, respectively, by Ih, while the yields of H were 3 and 10% by I h, respectively. The yield of H decreased beyond 1 h over CoMo / AI203 and hybrids, indicating its consecutive reaction pathway. Large yield of A4.6 over the hybrid was noted, being produced of A4.6 43% and 45% by 2h over hybrid with Ru / AI203 IO wt% and 20 wt%, respectively, while giving conversions of 29% and I I % by 2h over CoMo I A1203 and NiMo / Al203, respectively. Yield of C4,h were 0 to 2%. respectively over these catalysts, indicating very minor contribution of direct elimination of sulfur from 4.6-DMDBT as reported previously (2). The palticular Ru I AI203 was inferior in the desulfuri7ation to CoMo / AI203 and the hybrid, however it produced more H, giving its yield of 6% by Ih, and 8% by 2h. Longer reaction time beyond I h increased the yield, although no definite product of desulfurization was found. DISCUSSION Fig3 illustrates the hydrodeslfurization scheme of 4.6-DMDBT carries two methyl groups on 4 and 6 carbons neighbouring sulfur atom. Because two methyl groups on the sulfur htoin. sterically hinder the interaction of sulfur through its Pz orbital with the sulfur vacancy of sulfide catalyst, the direct elimination of sulfur at0111 is strongly hindered. The hydrogenation of one of two phenyl rings hreakes the co-planiuity of the dibenzothiophene skelton, moderating the steric hindrance of the methyl groups in the neighbors of the sulfur atom. Furthennore the hydrogenation of the neighboring phenyl ring increascs electron density of the sulfur atom enhancy its elimination through electron density interaction with the active site. Thus, it is very essential to hydrogenate the phenyl ring of 4.6-DMDBT for the accelerate of its desulfurization. The hydrogenation of 4.6-DMDBT at one of its phenyl ring certainly compeates the hydrogenation active site with aromatic partners of dominant presence in the deisel oil as observed in the present study. The selective hydrogenation of 4.6-DMDHT is very essential to accelerate its desulfurization. NiMo I A1203 exhibited preferable hydrogenation of naphthalene on the conversion-base to that of 4,6-DMDBT in the dominant presence of the former substrate. While CoMo I A1203 and the hybrid of CoMo / AI203 and Ru / AI203 did similar or slightly preferable selectivity to 4.6-DMDBT. respectively. Thus, the latter catalyst promoted the largest desulfurization activity of 4.6-DMDBT with the smallest hydrogenation of naphthalene. 990 The products for 4.6-DMDBT arc classified into three calegories. hydrogenation, direcf-desulrurizatioii and desullurization through the hydrogenation. RU / AI203 produce more hydrogenation product of 4.6-DMDBT than CoMo I AI203 and the hybnd catalyst. because of its insecutive desulfurization reactivity. while the hybrid catalyst gave the largest yield of bicyclohexyl which is the desulfurized product of both rings hydrogenated. Based on the above discussion, the hybrid catalyst performed the selective desulfurization of 4.6-DMDBT in the dominant presence of naphthalene through the selective hydrogenation of substrate over Ru / A1203 and the desulfurization of hydrogenated products over CoMo / A1203. High activity of NiMo / AI203 for the non-selective hydrogenation rules out the efficiency of its hybrid with Ru I A1203. The origin of selectivity for 4.6-DMDBT may be worthwhile for speculation. although no sufficient evidence is available at moment. n orbital localized on the sulfur atom in 4,6-DMDBT may interact prefereable to d orbital of the sulfide catalyst to that of the napthalene ring, being free from the steric hindrance of its methyl groups. Such a Sr - Md interaction may be expected more strongly with Ru than Ni. Co or Mo because of the higher polarizibillty of noble Ru, allowing the higher selectivity of Ru / A1203 to 4.6-DMDBT than naphthalene. LITERATURE CITED (.11. T akatuka., T.~. Wa.da. Y.. Suzuki. H.. Komatu. S.. Morimura. Y.. J.Jpn. Per. Insr. ~~ . . . . , 35, 197(1992). (2) Isoda, T., Ma, X., Mochida. I., J.Jpn. Per. /nsl.,37, 368 (1994). (3) Isoda, T., Ma, X., Mochida, I., J.Jpn. Per. Insr., 31, 506 (1994). (4) Isoda, T., Ma. X., Nagao, S., Mochida, I., J.Jpn. Per. Insr., 38, 25 (1995). (5) Lacroix. M., Boutarfa, N., Guillard, C., Vrinat, M., Breysse, M..J. Crrfd., 120. (6) Des Los Reyes, J.A.. Vrinat, M., Geantet, C., Breysse, M., Grimblot, J., J. (7) Isoda. T.. Kisamori, M., Ma, X., Mochida, I., Abstract of Symposium on Jpn (8) Isoda. I., Ma, X., Nagao, S., Mochida, I.. Abstract of Syinposium on Jpn. Pet. (9) Gerdil, R., Lucken, E., J.Am. Chem. SOC., 87.213 (1965). 473 (1989). Cuful., 142,455 ( 1993). Pet. Inst.. p.42, 17 - 18 May 1994, Japan. Inst.. p.316, 26 - 27 Oct. 1994, Japan. 100 h 2 C 0 Q> C .- E 50 s 0 a) Conversion of produced lelralin lo declin I NLMo n CUMO CoMcqlSulC)+ K d A b O ~ ( l O w l l ) CoMNlSMlr)+ 0 1 2 3 0 1 2 3 K d A 1 ~ l I r l 2 0 ~ 1 ~ ) Reaction time (hour) (A) 4,6-DMDBT (B) Naphthalene Fig. I Cocversions of 4,6-DMDBT and naphthalene over a hybrids of CoMo / , 4 1 2 0 3 and Ru /Al203. (30O0C-2.5MPa, 4,6-DMDBT 0.1 wt% + Nap IOwt% in decane, Catalyst content; 15 wt%, (CoMo / A1203) / (Ru /Ah03) = 1 .O and 2.0) to tetralin 991 60 #! p 40 20 h v .- s c '0 P n 0 15 h #! Y q10 2 5 .- r CI 'El P n 0 0 1 2 3 0 1 2 3 Reaction time (hour) (A) 646 (6) H Fig.2 Products from 4,6-DMDBT over a hybrids of CoMo / A I 2 0 3 and Ru /AlzOx Product distribution of 4,6-Dirnethyldibenzothiopheneo ver NiMo. CoMo, Ru I Alz01 and a hyhrid of CoMo I All01 and Ru I AIzOt. Table I Co-Mo I A1203 29 Ni-Mo I AIzOi 12 Ru I AI203 0 Co-Ma I Ah0 1 + 43 Ru I Al2Ol(towt%) Co-Mo I Ah01 + 45 Ru I A1203(XIwts) a) icilCtiiin condition: 31llK-2.SMPa-2h. 4.6-DMDBT O.lwt% and ~ a i pn w m Fig3 CH3 CH, B Reaction Pathway of 4,6-Dirnethyldibenzothiophene over Mo Sulfide Based on Catalyst. ( 300"C-2.SMPa) 992 I / HYDRODEOXYGENATION OF 0-CONTAINING POLYCYCLIC MODEL COMPOUNDS USING NOVEL ORGANOMETALLIC CATALYST PRECURSORS Stephen R. Kirby, Chunshan Song and Harold H. Schoben Fuel Science Program, 209 Academic hojects Bldg. Penn State University, PA 16802 Keywords: Deoxygenation, catalyst, Liquefaction. Oxygenated compounds are present in virtually all coals [I]. Phenols (and related hydroxyl compounds) have been identified as components of coal-derived distillates (2.31. Ethers and related compounds, connecting structural units within the coal matrix, have been proposed as sites for the depolymerization of the coal [4] and also ethers, together with carboxyls and phenolics, have been implicated in the facilitation of retrogressive, crosslinking, repolymerization reactions (5.61. Low-rank coals ( i.e. lignites and subbituminous coals ) include significantly more oxygen-containing groups than coals of higher rank [7]. With the increase in the extraction of lower rank coals in the U.S. and research into their use as liquefaction feedstocks [5,8,91, the imwrtance of ox_vee-n functionalitv removal from coal and coal-derived liquids is all the more apparent. The removal of these functionalities from the distillate products of coal liquefaction can be both complicated and expensive, and often leads to substantial reductions in distillate vields 131. Therefore. deoxwenation durine the liouefaction urocess would be beneficial. ?his god may be. attinablLhh the use of gulphideb bimetalli'c catalysts dispersed onto the coal using an organometallic precursor (10.1 I]. Model compound studies using multi-ring systems, or those of comparable molecular weight, were performed to investigate the capabilities of these catalysts. The model compounds selected represent a variety of oxygen functionalities, possibly present in coals of differing rank [12-141, contained within polycyclic systems. They include: anthrone (carbonyl); dinaphthyl ether (aryl-aryl ether); xanthene (heterocyclic ether); and 2,6-di-t -butyl-4-methylphenoI (hydroxyl). EXPERIMENTAL All experiments were performed in a 22ml capacity microreactor. A 0.5g sample of model compound was loaded into the reactor. Solvent was added in a 1:2 weight ratio to model compound and catalyst precursors were added at 2.46mol% concentration (unless otherwise stated). The catalyst precursors used were (NH4)2MoS4 (ATTM), [Ph4P]@i(MoS4)2] (Ni-Mol) and C ~ ~ CO~MOZ ( C(OCo)MZ So-~T 2). Air was removed by flushing the reactor three times with H2 to 1OOOpsi. The reactor was then repressurized to IoOOpsi H2. Reactions were performed at 300°C. 350°C and 400°C for 30 minutes. All reactions were carried out in a fluidized sand bath equipped with a vertical oscillator driving at a setting of 55 (-250 strokes per minute). At the end of the reaction the microreactor was quenched in cold water. Tridecane (0.25g) was added to the microreactor as an internal standard. The' microreactor contents were then extracted with acetone and diluted for analysis. Capillary gas chromatography (GC) connected to a flame ionization detector (Perkin Elmer-8500) and gas chromatography / mass spectrometry (Hewlett Packard-5890) were used for the quantitative and qualitative analysis of the product distribution, respectively. RESULTS AND DISCUSSION Product distributions have been grouped as oxygen-containing and deoxygenated for the purposes of this article. The conversions of anthrone, dinaphthyl ether, xanthene and 2,6-dir-butyl4-methylphenola re shown in Figures 14, and the product distribution of dinaphthyl ether is given in Figure 5. Generally, the addition of any catalyst to a system under the conditions studied increases the total conversion. For example, at 400°C dinaphthyl ether undergoes 26% thermal conversion; this yield is increased to 72% in the presence of AITM, 88.5% with Ni-Mol, and 100% using CoMo-T2. However, any improvement in the product quality, especially deoxygenation and ring reduction, in the presence of these catalysts is also important, and the variation of these factors for the different oxygen functional groups will be the main focus of this discussion. Anthrone Under non-catalytic conditions anthrone converts to anthracene through thermal reaction of the carbonyl oxygen. Anthracene then reacts further to form a variety of hydrogenated ring species, such as di- and tetrahydroanthracene. In the presence of AmM, the formation of oxygen-containing compounds in the products at 350T and 400°C (substituted naphthols and phenols) suggest hydrogenation of 993 the carbonyl oxygen to a hydroxyl group before extensive conversion to anthracene. Reduction in the yield of these oxygen functionalities in the ATTM reaction at 400°C may indicate the possibility of an increase in the conversion of these Species tO non-oxygenated products. Conversion of anthrone to oxygen-free products is increased considerably using the C ~ M ca~talyTst p~recu rsor. This implies that CoMo-T2 has the capability to increase the conversion of carbonyls without additional phenol or naphthol production. This may be achieved by either rapid C=O cleavage prior to ring hydrogenation, rapid phenol conversion to oxygen-free products, or by the prevention of initial hydroxyl group formation. From the reactions of 2,6-di-r-butyl-4-methylphenowl ith CoMo-T2, it can be Seen that this catalyst, although removing some hydroxyl functionality, does not promote the ready conversion of phenols to non-oxygen containing species. Variations in the oxygen-free products of anthrone conversion are also apparent for the different catalyst precursors. Ni-Mol appears to promote the formation of 1,2,3,4- tetrahydroanthracene (THA), whereas CoMo-T2 demonstrates the facilitation of 9,10- dihydroanthracene (DHA) production. ATTM seems to have equal affinity for the formation of both products. Ni-Mol and ATTM both exhibit an increase in the formation of 1,2,3,4,5,6,7,8-octahydroanthracene (OHA) at 400°C (0% under catalyst-free conditions to 11.8% and 11.3% respectively), which only appears in very low yields with CoMo-T2 (1.5%). This reduction in OHA yield for the CoMo-T2 precursor is comparable to increases in anthracene and DHA production, suggesting selective hydrogenation of the 9- and 10- positions (i.e. the carbonyl carbon). Dinaphthyl Ether Under non-catalytic conditions naphthalene is the major product of dinaphthyl ether (DNE) hydrogenation, with low yields of 2-naphthol, although total conversion is very small (26%). Oxygen functionality removal is increased in the presence of all the catalyst precursors, although to a lesser extent than for anthrone. ATTM increases DNE conversion to oxygen-free products (63.6% at 400°C) with the balance of the products being phenols, naphthols (1.8%) and ring-reduced derivatives of the starting material. Phenol and naphthol yields decrease from 350°C to 400°C. again implying that ATTM facilitates hydroxyl group removal. High conversions to tetralin and naphthalene are achieved in the presence of CoMo- T2 (51.6% and 40.2% respectively at 400OC). Phenols and naphthols are present in larger yields than for anthrone, suggesting the cleavage of a single C-0 bond followed by hydrogenation of the phenoxy (or naphthoxy) group. Ring-reduced derivatives of DNE produced at 350°C are absent at 400°C and naphthol yields decrease across the same temperature range. These reductions in oxygen compound yields are accompanied by increases in tetralin, naphthalene and alkylbenzene formation. The product distributions (0 : non-0) of reactions of A7TM, Ni-Mo and CoMo-TZ with DNE (Figures 2 and 5) distinctly show the latter precursor to be the most favourable for C-0-C bond cleavage to oxygen-free products. Xanthene In the absence of a catalyst xanthene is totally unreactive. Addition of ATTM or CoMo-T2 produces noticeable reaction at 350°C and 400°C. At 350°C the products from both precursors are phenols, cycloalkyl- and long-chain alkylbenzenes formed by C-0 and C-C bond cleavage. However, at 400°C ATTM produces an increase in oxygen-free products with no increase in phenols, although conversion to non-oxygen containing species is low (24.9%). Increases in oxygen-free product yields are also achieved with CoMo-T2 at 400°C. but with accompanying increases in phenol formation. This gain in phenols may be attributed to the formation of short-chain (Cl-C2) alkylphenols from longer chain alkylphenols, implying that CoMo-T2 favours C-C cleavage over C-OH. The comparably large conversion to oxygen-free products and phenols reinforces the ability of CoMo-T2 to cleave ether linkages, and inability to remove hydroxyl groups. However, the low conversions of xanthene illustrate the unreactive nature of the starting material. 2,6-Di-r-butyl4-methylpheno(lD BMP) Under non-catalytic reaction conditions the conversion of DBMP involves the cleavage of one, or both, of the r-butyl groups to produce 2-butyl-4-methylphenol (BMP) and ultimately 4-methylphenol (100% at 400OC). No reaction occurs at 300°C in the absence of a catalyst. When a catalyst is present the removal of the butyl groups becomes more favourable and formation of the above products takes place. At 350°C with ATIU, almost all the starting material has reacted and only a small poaion remains as BMP (13.5%). The major product, 4-methylpheno1, then undergoes catalytic hydrogenation and hydroxyl removal to form toluene and methylcyclohexane. At 400°C these reactions proceed to a greater extent, resulting in greater yields of both Products (46.5% and 20.2% respectively). In the Presence of CoMo-T2, DBMP appears to lose both butyl groups SO rapidly that no 2-r-butyl-4-methylphenol is isolated, so 4-methylphenol is the only product at 994 I 300°C. At 350°C it exhibits some further conversion to methylcyclohexane (1.6%) and at 4W"' toluene and methylcyclohexane are produced. DBMP is a reactive compound through loss of its butyl groups. However, the hydroxyl group C-OH bond is very resistant to reaction and is only cleaved, to a substantial degree, in the presence of the ATTM precursor. CoMo-T2 removes the OH-group, but only to a small extent. Investigations using the Ni-Mol precursor are not as advanced as those for ATTM and CoMo-T2. Presentation of these results is planned for future articles. CONCLUSIONS From the non-catalytic data shown there is a clear order of starting material reactivity : 2,6-di-t-butyl-4-methylphenol> anthrone > dinaphthyl ether > xanthene. However, the reactivity order of the oxygen functionalities in the presence of the various catalysts is different. For non-catalytic conditions the order appears to be : carbonyl >arylaryl ether )> substituted phenol = heterocyclic ether. In the presence of A'lTh4 this sequence changes slightly to : carbonyl >substituted phenol = aryl-aryl ether B heterocyclic ether and for reactions involving CoMo-T2 the reactivity order appears to be : carbonyl aryl-aryl ether > heterocyclic ether > substituted phenol. These differences in reactivity order emphasize the effect of the nature of the oxygen functionality on the deoxygenating capabilities of the catalysts and that different catalysts can have different roles in promoting hydrodeoxygenation and reduction, depending on the nature of the starting material. They also highlight the undesirability of phenolic and heterocyclic ether structures in liquefaction systems. Both these structures types are quite unreactive under liquefaction conditions and any reaction has a tendency to form high yields of single-ring phenols. When applied to coals, these findings suggest that coals differing from each other in the form of which oxygen functional groups are dominant, may show quite different kinds of liquefaction products, depending on which catalyst precursor was chosen. ACKNOWLEDGEMENTS The authors wish to express their appreciation to the U.S. Department of Energy, Pittsburgh Energy Technology Centre for supporting this work, Dr. E. Schmidt for synthesizing the catalyst precursors and Mr. R.M. Copenhaver for the fabrication of the microreactors. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. C. Song, L. Hou, A. K. Saini, P. G. Hatcher and H. H. Schobert, (1993). Fuel Processing Technology, 34 249-276. R. E. Pauls, M. E. Bambacht, C. Bradley, S. E. Scheppele and D. C. Cronauer, (1990). Energy & Fuels, 4 236-242. C. Burgess, (1994). "Direct Coal Liquefaction: A Potential Route to Thermally Stable Jet Fuel'', pp. 167. L. Artok, 0. Erbatur and H. H. Schobert, (in press). C. Song, H. H. Schobert and P. G. Hatcher, (1992). Energy & Fuels, 6 326-328. M. A. Serio, E. Kroo, S. Charpenay, R. Bassilakis, P. R. Solomon, D. F. McMillen, A. Satyam, J. Manion and R. Malhotra, (1993). Proceedings of Contractors' Review Conference: "Coal Liquefaction and Gas Conversion": "The Dual Role of Oxygen Functions in Coal Pretreatment and Liquefaction: Crosslinking and Cleavage Reactions", Pittsburgh, PA, 15-44. S. M. Solum, R. J. Pugmire and D. M. Grant, (1989). Energy & Fuels, 3 187-193. C. Song and H. H. Schohert, (1992). Am. Chem. Soc. Div. Fuel Chem. Prepr., 37 42. L. Huang, C. Song and H. H. Schobert, (1992). Am. Chem. Soc. Div. Fuel Chem. Prepr., 31 223. C. Song and H. H. Schobert, (1993). "Novel Bimetallic Dispersed Catalysts for Temperature-Programmed Coal Liquefaction", PeM State University DE-AC22- C. Song, D. S. Parfitt and H. H. Schobert, (1993). Catalysis Letters, 21 27-34. L. M. Stock, (1989). Accounts of Chemical Research, 22 421-433. R. Hayatsu, R. E. Winans, R. G. Scott, L. P. Moore and M. H. Studier, (1978). Fuel, 57 (2), 541. J. H. Shinn, (1984). Fuel, 63 (3), 1187 92PC92122-TPR- 1. 995 Figure 1. Yield of oxygenated and deoxygenated products of anthrone as a function of temperature and catalyst precursor. 100 90 80 Q 70 5 60 gc 50 9 40 " 30 20 10 0 B . . None None None AlTM ATIU AlTM Co Co Co Ni- 300°C 350T 400°C 300T 350°C 403°C Mo-T2 Mo-T2 MeT2 Mol 300°C 35wc 400°C w c Figure 2. Yield of oxygenated and deoxygenated products from dinaphthyl ether as a function of temperature and catalyst. . . . . . None None None ATTM ATTM ATI'M Co Co Co Ni- 300T 350T 400°C 300°C 350°C 4oooC Mo-X? Mo-n Mo-n Mol 300°C 350°C W0C 4oo°C 996 / React.ternp.('C) Cat. Precursors ROdUCU(wt%) Tetralin Naphthalene THDNE OHDNE 2-Naphthol Methylphenol Alkylbenzenes Conv. (wt%) THnaphthol I i 300 350 400 300 350 400 400 300 350 400 None None None A?TM A'ITM A?TM Ni-Mo CoMo- CoMo- CoMo- TZ TZ "2 1.2 1.3 30.5 24.4 24.6 4.4 47.2 51.6 1.9 4.1 22.9 1.4 26.1 38.3 39.1 1.9 28.6 40.2 7.4 6.3 12.0 0.3 7.5 0.3 1.7 3.2 2.6 2.3 1.0 4.1 5.9 3.5 3.2 0.6 0.8 2.1 0.3 1.5 0.5 0.3 0.4 0.6 0.7 1.9 0.9 2.0 0.5 3.1 1.9 5.3 26.1 2.7 70.7 71.9 87.6 6.9 94.3 100 I' Figure 3. Yield of oxygenated and deoxygenated products of xanthene as a function of temperature and catalyst. 60 50-/ 5 Oxygenated - 3 40-/ .: 30-/ e -5 6 20 IO 0 None None None ATIU A m A m Co Co Co 300°C 350°C 400°C 300OC 350°C 400°C Mo-T2 MwT2 M o l 2 300°C 350°C 400°C Figure 4. Yield of oxygenated and deoxygenated products of 2,6-di-1-butyl-4-methylphenol as a function of temperature and catalyst. None None None AlTM AlTM AlTM Co Co Co 300°C 350°C 40°C 300T 350°C 400°C Mo-T2 Mo-n Mo-n 300T 350°C 40°C 997 SYNTHESIS AND REACTIVITY OF NEW BIMETALLIC OXYNITRIDES S. Ramanathan, C. C. Yu and S. T. Oyama Department of Chemical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061 Keywords: Bimetallic oxynitrides, Hydrodenitrogenation, Hydrodesulfurization ABSTRACT A new series of catalysts, transition metal bimetallic oxynitrides of the form M1M20,Ny (M1 = V, Nb and Cr, M2 = Mo), was prepared. The catalysts were synthesized by nitriding the bimetallic oxide precursors in an ammonia gas stream at 1000 cm3/min ( 6 . 8 ~ 1 0p~m ols-l) using a heating rate of 5 Klmin (8.3~ 10-2 Ks-1). The catalysts were characterized by x-ray diffraction, CO chemisorption and surface area measurements. The activity of these catalysts for hydroprocessing was studied in a three-phase trickle bed reactor operated at 3.1 MPa and 643 K. The liquid feed consisted of 3000 ppm sulfur (dibenzothiophene), 2000 ppm nitrogen (quinoline), 500 ppm oxygen (benzofuran), 20 wt% aromatics (15 wt% tetralin and 5 wt% amylbenzene) and balance aliphatics (tetradecane). The activities of the bimetallic oxynitrides were compared to a commercial Ni-Mo/Al2O3 (Shell 324) catalyst tested at the same conditions. The bimetallic oxynitrides were found to be active for the hydrodenitrogenation (HDN) of quinoline. In particular, V-Mo-0-N exhibited higher HDN activity than the commercial Ni-Mo/A1203 catalyst. hydrodesulfurization (HDS) activity of the bimetallic oxynitrides ranged from 9-25% with V-Mo-0-N showing the highest HDS activity among the oxynitrides tested. The INTRODUCTION Monometallic nitrides have been investigated extensively since the 50's and 60's [1,2]. However, in order to take advantage of the catalytic properties of the carbides and nitrides, it is important to prepare these materials in high surface area form. Conventional powder metallurgy methods such as direct nitridation or carburization of metal or metal oxide powders resulted in compounds of typically low surface area (c 10 m2g-l). Significant progress has been made in the preparation of these materials in high surface area form in the last decade and a half [3-61. One of the techniques developed during this period was the temperature programmed reaction [7] method of preparing high surface area compounds from oxide precursors. The technique offers the advantage of lower synthesis temperatures than the conventional methods. In addition, the transformation of the oxide to the carbidelnitride phase is direct, bypassing the metal phase, which is the most prone to sintering. However, most of the work on the temperature programmed reaction is focused on the synthesis of monometallic carbides and nitrides. There is little work reported on the preparation of high surface area mixed transition metal nitridesloxynitrides in the literature. Transition metal carbides and nitrides were found to be active for a number of hydrocarbon reactions [a]. One of the major applications of transition metal carbides and nitrides has been in hydroprocessing. Petroleum feedstocks is gaining importance with the need to process heavier resources. Nitrogen removal is always accompanied by the consumption of excess hydrogen due to the difficulty involved in the C-N bond scission. The development of catalysts that are selective to C-N cleavage is an important goal, and this paper reports an investigation on a new class of catalysts which is different in structure and properties from the conventional Ni-MojAl2O3 and Co-Mo/A1203 hydrotreating catalysts. Removal of nitrogen and sulfur from 998 1 'i After the initial results by Schlatter, et al., [ 9 ] on quinoline HDN, most of the hydroprocessing work has been concentrated on molybdenum nitride catalysts, both supported and unsupported [lo- 141. nltrogen is partially exchanged by oxygen and a second transition metal is also introduced in the interstitial compound. SYNTHESIS AND CHARACTERIZATION The bimetallic oxynitrides were prepared by nitriding the bimetallic oxide precursors using the temperature programmed reaction technique [15]. The oxide precursor was loaded in a quartz reactor placed in a furnace (Hoskins S O O W ) . Ammonia reactant gas was passed over the oxide bed at a flow rate of 1000 cm3/min (68Ox1O2 pmols-1). was ramped linearly at 5 Kjmin ( 8 . 3 ~ 1 0 -K~s -l) to the final synthesis temperature (TmaX) and held at that temperature for a period of time (thold). The effluent gases from the reactor were analyzed by an on-line mass spectrometer (AmetekIDycor, MAlOO). Once the reaction was completed, the gas flow was switched to helium and the reactor was quickly cooled down to room temperature by removing the furnace. The catalysts were passivated at room temperature in a 0.5% 02/He gas mixture before exposure to the atmosphere. conditions used in the preparation of these materials is presented in Table 1. The bulk phase purity of the samples was identified by x-ray diffraction (XRD) (Siemens Model D500 with a CuKa monochromatized radiation source). Figure 1 presents the XRD patterns of the passivated bimetallic oxynitrides. The patterns did not show any features of the starting oxide material and moreover, all the patterns indicate that the oxynitrides have a face centered cubic arrangement. In addition, the linebroadening of the peaks indicates the presence of small crystallites. Elemental analysis indicated that the actual composition of the catalysts was V2.~Mo1.001.7N2.4, Nb2.0M02.603.0N4.2 and Cr1.0M01.302.3N1.4. N2 physisorption and CO chemisorption measurements were carried out to obtain the specific surface area and the number of exposed surface metal atoms. Prior to surface adsorption measurements, the catalysts were activated in a flow of 10% H2/He gas mixture at 738 K for 2 h. The surface areas, CO uptakes and the number densities are summarized in Table 2. The number densities indicated in Table 2 reveal that only a maximum of 14% of the total metal atoms are available for the chemisorbing molecule. These values are typical of the interstitial compounds due to the prior occupation of the sites by N and 0, which were not removed during the activation process. REACTIVITY Experimental runs consisted of testing a series of oxynitride catalysts for their activity in hydrodenitrogenation (HDN), hydrodesulfurization (HDS) and hydrodeoxygenation (HDO). The reactions were carried out in a three-phase trickle-bed reactor at 3.1 MPa and 643 X. was loaded, corresponding to a total surface area of 30 m2. prior to Catalytic testing, the oxynitrides were activated in flowing hydrogen at 723 K for 3 hours. The commercial Ni- ~o/Al2O3 Catalyst was sulfided in a flow of 10% H2S/H2 gas mixture. After the activation process, the reactors were cooled down to 643 K and hydrogen was pressurized to 3.1 MPa. Hydrogen flow to the reactor was maintained at 150 cm3(NTP)/min (100 pmo1S-l) using mass flow controllers. Liquid feed rate was set at 5 cm3h-l using high-pressure liquid pumps. passed over the catalyst bed in a cocurrent upflow mode and out This paper reports a new family of nitrides, where the The temperature of the reactor bed A summary of the synthesis Typically about 0.2-1 g of the catalyst The gas and liquid 999 to the liquid sampling valve. The liquid feed composition used in all the experiments was 3000 ppm S (dibenzothiophene), 2000 ppm N (quinoline), 500 ppm 0 (benzofuran), 20 wt% aromatics (15 wt% tetralin and 5 wt% amylbenzene) and balance aliphatics (tetradecane). The reactions were carried out for a period of 60 hours. The liquid samples were analyzed off-line by gas chromatography. The activity of the catalysts was compared on the basis of equal surface areas of 30 m* loaded in the reactor. Figure 2 shows a comparison of the activities of the catalysts for HDN, HDS and HDO at 3.1 MPa and 643 K. Clearly, the oxynitrides show considerable activity for the HDN of quinoline. In fact, V-Mo-ON exhibited higher activity than the commercial sulfided Ni- MofAl,O, catalyst. All the catalysts showed similar product distribution and the major hydrodenitrogenated product was propylcyclohexane. The HDN activity of the catalysts was stable even after 60 hours on-stream. The HDS activity of the oxynitrides ranged from 9-25%, with V-Mo-0-N displaying the highest HDS activity among the oxynitrides tested. The oxynitrides showed high initial HDS activities, but they deactivated after about 25 h on-stream. The major product from the HDS of dibenzothiophene was biphenyl. The oxynitrides were also active for the removal of oxygen from benzofuran. The HDO activity ranged from 12-32% and the major deoxygenated product was ethylcyclohexane. In fact, the V-Mo-0-N showed higher overall activity than the corresponding monometallic nitrides 1161. for the removal of sulfur and oxygen from the liquid feed. X-ray diffraction patterns of the spent catalysts indicated that the bulk phase purity of the samples was preserved. did not show extraneous oxide or sulfide peaks indicating that the oxynitrides were stable towards heteroatoms even after prolonged exposure at elevated temperatures. The commercial Ni-Mo/Al2O3 catalyst showed high activities The patterns CONCLUSIONS A new series of catalysts, bimetallic oxynitrides of transition metals, was prepared in high surface area form. They were found to be active for the hydrodenitrogenation of quinoline. Interestingly, V-Mo-0-N displayed higher HDN activity than the commercial Ni-MO/Al2Oj catalyst. The new catalysts were found to be sulfur resistant under the reaction conditions. The bimetallic oxynitrides displayed better activity and stability than the monometallic nitrides. ACKNOWLEDGMENT Support for this work by Akzo Nobel and the Department of Energy, Office of Basic Energy Sciences is appreciated. REFERENCES 1. Toth, L. E., "Transition Metal Carbides and Nitrides", Academic Press, New York, 1971. 2. Juza, R., in Advances in Inorganic Chemistry and Radiochemistry", (H. J. EmelGus and A. G. Sharpe, Eds.), Vol. 9, p. 81, Academic Press, New York, 1966. 3. Volpe, L., Oyama, S. T., and Boudart, M., "Preparation of Catalysts 111". p. 147, 1983. 4. Volpe, L., and Boudart, M., J. Solid State Chem. 59, 332 (1985) . 5. Volpe, L., and Boudart, M., J. solid State Chem. 59, 348 (1985) . 6 . Ledoux, M. J., Guille, J., Hanzter, S., Marin, S., and Pham- Huu, C . , Extended Abstracts, Proceedings MRS Symposium, Boston, Nov 26-Dec 1, p. 135, 1990. I . Oyama, S. T., Ph.D. Dissertation, Stanford University, 1981. 1000 a . 9. 10. 11. 12. 1 3 . 1 4 . 15. 16. Oyama, S. T., Catal. Today, 15, 179 ( 1 9 9 2 ) . Schlatter, J. C., Oyama, S. T., Metcalfe, J. M., III., and Lambert, J. M., Jr., Ind. Eng. Chem. Res., 27, 1648 ( 1 9 8 8 ) . Lee, K. S . , Abe, H., Reimer, J. A., and Bell, A. T., J. Catal., 139, 34 ( 1 9 9 3 ) . Abe, H . , and Bell, A. T., Catal. Lett., 18, 1 ( 1 9 9 3 ) . Colling, W. C., and Thompson, L. T., J. Catal., 146, 193 ( 1 9 9 4 ) . Nagai, M., and Miyao, T., Catal. Lett., 15, 105 ( 1 9 9 2 ) . Nagai, M., Miyao, T., and Tuboi, T., Catal. Lett., 18, 9 ( 1 9 9 3 ) . Yu, C. C., and Oyama, S. T., J. Sol. St. Chem., 116, 205 (1995) . Yu, C. C., Ramanathan, S., Sherif, F., and Oyama, S. T., J. Phy. Chem., 98, 13038 ( 1 9 9 4 ) . T a b l e 1. Summary of the Synthesis Conditions Catalyst Final Temperature Soak Period Tmax I K thold 1 V-Mo-0-N 1037 0.5 Nb-No-0-N 1063 0 . 3 Cr-Mo-0-N 1013 0 . 3 Table 2 . Results of Surface Adsorption Measurements Catalyst CO Uptake Surface Area Number Density pmolg-1 m29-1 x 1015 cm-2 V-MO-0-N 1 6 7 7 4 0 . 1 4 Nb-MO-0-N 1 1 . 2 1 2 1 0.0056 Cr-Mo-0-N 163 90 0 . 1 1 1001 F i g u r e 1. ~~~ 20 30 40 50 €Q 70 80 90 20 I degrees X-ray diffraction patterns of the fresh catalysts. 90 80 70 60 'C 50 0 40 z 30 20 10 0 = HDN a HDS 0 HDO r F i g u r e 2 . catalysts at 3.1 MPa and 643 K. Comparison of the HDN, HDS and HDO activities of the i I t ll 1002 I SYNTHESIS OF MESOPOROUS MOLECULAR SIEVES AND THEIR APPLICATION FOR CATALYTIC CONVERSION OF POLYCYCLIC AROMATIC HYDROCARBONS Kondam Madhusudan Reddy and Chunshan Song Fuel Science Program, Department of Materials Science and Engineering The Pennsylvania State University, University Park, PA 16802, USA Mesoporous molecular sieves, AI-MCM-41, hydrogenation, isopropylation, hydrocracking. polycyclic aromatic hydrocarbons Keywords: INTRODUCTION Molecular sieves such as Y and ZSM-5 are widely used catalysts in acid-catalyzed reactions for the production of fuels, petrochemicals, and fine chemicals 11-31, Despite their enormous use as environmentally safe catalysts, they are limited to convert relatively small molecules as their pore size is retuicted to nucropore size range (usually 1.4 nm). However, with the growing demand of technologies for treating heavier feeds, as well as for synthesizing large molecules for producing commodities and fine chemicals, it is necessary to develop catalysts with wider pores. Recently, Mobil workers have reported a new series of mesoporous molecular sieves (4.51: MCM-41 is one of the members of this extensive family of mesoporous series possessing a hexagonal array of uniform mesopores. Many reports have since appeared on synthesis and characterization of these new materials (6-101.H owever, information on their catalytic activity is still very limlted. The pore dimensions of these materials can be tailored (in the range of 1.5-10.0 nm or more) through the choice of surfactant and auxiliary chenucals as templates and the crystallization conditions in the synthesis procedure. The BET surface area of these materials IS more than 1000 m2/g with high sorption capacities of 0.7 cclg and greater. Moreover, these matenals can be synthesized in a large range of framework SUA1 ratios and therefore c3n develop acid sites of different strength. Hence, these new mesoporous aluminosilicate molecular sieves, AI-MCM-41, might open new possihilities in developing catalysts for processing large molecules. As part of our ongoing project on liquefaction of coal and upgrading of coal liquids, we intend to use these mesoporous aluminosilicates molexlar sieves as catalysts to upgrade the coal derived oils to transportation fuels, particularly thermally stable jet fuels. We have studied the synthesis and characterization of these materials [ I I]. In this paper. we repon some of the results on the synthesis and their application for the catalytic conversion of model polycyclic aromatic hydrocarbon compounds. EXPERIMENTAL The mesoporous aluminosilicate molecular sieves, AI-MCM-4 I , were synthesized hydrothermally in 100 ml Teflon lined autoclavcs from a mixture of reactants with the following composition: 50Si02-xA120~-2.19(TMA)~0-15.62(CTMA)Br-316w5hHe2re0 ;- 0.5, 1.0 and 2.0. The details of synthesis are given elsewhere [I I ] . Three series of smples with varying Si/AI ratios. and using three different aluminum sources (aluminum isopropoxide. pseudo boehmite and aluminum sulfate) were synthesizcd. Some synthesis parameters and their physical characteristics are shown in Table 1. The AI-MCM-41 samples were characterized by chemical analysis, X-ray diffraction, nitrogen adsorption, thcrmogravimetric analysis, and solid state NMR Prior to catalytic runs, the organic template from the as-synthcsized solids was removed by calcining the samples in a tubular furnace at 550 OC for one hour in nitrogen and 6 hours in air flow. The calcined samples were exchanged with ammonium nitrate. The protonated form was then obtained by calcining these ammonium exchanged samples at 550 UC for 3 hours. Finally, 3wt% Pt was loaded by wet impregnation, with a required amount of hexachloro platonic acid (AldnchJ solution and the sample in a beaker and evaporating the water at room temperature while stirring it overnight. The Pt loaded samples were then calcined in air at 450 OC for 3 hours. Mesoporous molecular sieve catalysts were tested for the following reactions: I ) hydrogenation of naphthalene and phenanthrene, 2) isopropylation of naphthalene and 3) hydrocracking of I J.5- triisopropyl benzene. A 30 cc stainless-steel tubing bomb batch reactor was used for all the experiments. During the reaction, reactors were heated in a fluidized sand-bath under vertical shaking (240 cycles/min.). All the chemicals were used as supplied. The standard reactor charge w3s 0.10 g of catalyst and 1.0 g of reactant and other reaction conditions are given in appropriate Tables. At the end of the reaction, the reactor was quenched in cold water. After collecting the reaction products in acetone solution, they were analyzed by GC (Perhn-Elmer 8500) using DB-17 fused silica capillary column. The products were identified by GC-MS (HP). RESULTS AND DISCUSSIONS Three series of AI-MCM-4 I samples using three different aluminum sources. aluminum isopropoxide, pseudo boehmite (Catapnl B), and aluminum sulfate. with SUA1 ratios 50, 25, and 12.5 were synthesized. Details are shown in Table 1. The crystallinity, the incorporation of aluminum in framework and the acidity were studied by XRD, nitrogen sorption. thermal analysis of n-butylmine on samples, 27Al MAS NMR. The results on the synthesis and charactenzation were reponed in our earlier paper [I I]. X-ray diffraction patterns showed that all the samples are well crystallized and phase pure with a very strong peak and three weak peaks 11.21. A typical XRD Pattern of AI-MCM-41 is shown in Figure I . It was observed from nitrogen sorption and XRD studies that the samples prepared with aluminum sulfate are less crystalline compared to the 1003 other two series of samples prepared with different aluminum sources. However, the incorporation of aluminum framework was found to be efficient with aluminum isopropoxide and aluminum sulfate, compared to pseudo beohmite. The aluminum incorporation was characterized by increase in the interplanar spacings from XRD and 27Al MAS NMR. The acidity due to the presence of aluminum in the framework was determined by a thermal analysis of n-butylamine on samples. As the aluminum incorporation is higher in samples prepared with aluminum sulfate and aluminum isopropoxide, they have shown better acidity compared to other samples prepared'with pseudo boehmite [I I]. Catalytic test results in the reactions of hydrogenation of naphthalene and phenanthrene, isopropylation of naphthalene, and hydrocracking of 1,3,S-triisopropylbenzene are presented in Tables 2-5. The initial observation is that they are active in these reactions with good conversions. However, reactions occurred non-selectively as expected, because of the no shape-selective nature of mesoporous materials with wide pores. In the case of hydrogenation of naphthalene. conversion was almost hundred percent with all the catalysts (see Table 2). There was a large amount of unconverted tevalin observed for the MCM- 41 catalysts prepared with pseudo beohmite, whereas for other MCM-41 catalysts the conversion of tetralin to decalin was almost complete. The r-decalinlc-decalin ratios for all the MCM-41 catalysts are low and in case of the MCM-41 catalyst prepared with pseudo boehmite the ratio is lower compared to the earlier results reported on mordenite [ 12,131. These results indicate that in this reaction. the isomerization of c-decalin to r-decalin probably takes place on acid sites. If that is the case, MCM-41 samples are less acidic as compared to mordenite. hence the r-decalinkdecalin ratio is low for these materials. Moreover this ratio is lower for the MCM-41 catalyst prepared with pseudo beohmite because of the poor incorpowtion of aluminum in the framework. leading to poor acidity. Table 3 shows the product analyses in hydrogenation of phenanthrene over three different MCM- 41 catalysts. They were all active but product selectivities were different compared to earlier results reported [ 141; especially sym-octahydroanthracene was formed less, which is an isomerized product from sym-octahydrophenanthrene. This isomerization was believed to be occurring on acid sites [ 141. Hence MCM-41 catalysts are not as acidic as other zeolites. especially the MCM- 41 catalyst prepared with pseudo beohmite. Similar observations were reported by earlier authors [6,81. The product analyses of isopropylation of naphthalene using propylene are presented in Table 4. The alkylation over zeolites is known to occur on acid sites. The MCM-4 I catalyst prepared with pseudo beohmite was not as active ar the other two MCM-41 catalysts. It indicates that the catalyst prepared with pseudo beohmite is less acidic. which again confirms the poor incorporation of aluminum in the framework compared to the other two catalysts. From the product analyses it is also clear that tri and tetra isopropylnaphthalene are formed in large quantities which is a clear indication of non selective nature of these mesoporous materials compared to other zeolites [ 12, 131. Non selective nature of these mesoporous materials can also be verified from the a and p substituted product selectivities, which are different from the results obtained on mordenite and Y zeolites [12,13]. Table 3 also shows the effect of h loading in the reaction of isopropylation of naphthalene. Both Pt and non h containing catalyst showed more or less similar activity, however, selectivities were different. The Pt loaded catalyst yielded more tri and tetra substituted isopropylnaphthalenes, which indicates that the alkylation processes seems to be more efficient with Pr loaded catalysts. This may be due to the bifunctional nature of the catalyst. In zeolite catalysis, it is a known fact that the bifunctional catalysts are more susceptible to the coke formation. In these experiments, because of less reaction time and with limited availability of propylene, deactivation due to coke has not been noticed. However, with continuous supply of propylene for longer reaction times there might be a noticeable difference in catalyst stabilities with and without h loading, In alkylation reactions over zeolites, the type of alkylating agent is known to have an effect. For example, the alkylation with alchols was found to be less efficlent compared with respective alkenes [ 131. This could be due to the water formation in the reactions with alcohol and water, that may be suppressing the activity of acid sites. Similar results were observed with these mesoporous molecular sieves. The conversion of naphthalene to isopropylnaphthalenes was found to be less when isopropanol was used as an alkylating agent compared to propylene. Apart from the hydrogenation and alkylation. hydrocracking is an important reaction in the process of upgrading the heavy oils. Zeolites are known to be good hydrocracking catalysts with good activity, selectivity, and stability. Table 5 shows the results of hydrocracking of bulky molecule, 1,3,5 triisopropylbenzene over mesoporous molecular sieves. Both PUAI-MCM-41 and H/AIMCM- 41 were found to be active in this reaction. The main products on WAI-MCM-41 were mono and di substituird isopropylbenzenes. However, the conversion of Pi loaded catalyst is 100% and products were of lower molecular weight, mainly C& hydrocarbons. This may be due to the hydrogenation reaction thus leading to further cracking. The conversion of 1,3.5 oiisopropylbenzene was negligible without catalyst at similar conditions. CONCLUSIONS We observed that mesoporous molecular sieve catalysts are capable of converting bulky polycyclic aromatic hydrocarbons. Due 10 the large size of mesopores relative to the substrate molecules, however, reactions are found to be occurring non-selectively. The efficiency of the mesoporous 1004 I I i I molecular sieves can be significantly different depending on their synthesis method, especially with the source of alumjnum used. The catalysts prepared with aluminum isopropoxide and alumintun sulfate were found to be more active compared to the ones prepared with pseudo boehmite. n e Pt loading and type of alkylating agent are influential in the conversion of polycyclic aromatic compounds, as reflected in the product selectivities. ACKNOWLEDGMENTS We ?e grateful to Dr. H. H. Schobert for his encouragement and support. Financial support was provided by US Department of Energy and US Air Force. We wish to thank Mr. W. E. Harrison 111 of USAF and Dr. S. Rogers of DOE for their support, and Dr. A. Schmitz for assistance in perfonrung out catalytic mns and GC analysis. REFERENCES 1. J. E. Naber. K. P. deJone, W. H. J. Stork, H. P. C. E. Kuiuers, and M. F. Post, Stud. in Surf 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Sci. and Cd., 84 (1994) 2197. W. F. Hoelderich, M. Hesse, and F. Naumann, Angew. Chem. Int. Ed. Eng., 27 '( 1988) 226. W. 0. Haae. Stud. in Surf: Sci. and Catal.. 84 (1994) 1375. C. T. Cesie, M. E. LeoGowicz, W. J. Roih, J.'C. V&uli, and J. S. Beck, Nature, 359 (1992) 710. J. S. Beck, J. C. Vartuli, W. J. Roth, M. E. Leonowicz, C. T. Kresge, K. D. Schmjtt, C. T. W. Chu, D. H. Olson, E. W. Sheppard, S . B. McCullen, J. B. Higgins, and J. C. Schlenker, J. Am. Chem. Soc , 114 (1992) 10834. C.-Y. Chen, S. L. Burkett, H.-X Li and M. Davis, Microporous Mater. 2 (1993) 17. C.-Y. Chen, H.-X Li, and M. Davis, Microporous Mater., 2 (1993) 27. A. Coma, V. Fornes, M. T. Navarro and J. Perez-Pariente, J. Catal. 148 (1994) 569. M. Janicke, D. Kumar, G. D. Stucky, and B. F. Chmelka, Stud. in Surf: Sci. and Catal., 84 (1994) 243. R. B. Borade, and A. Clearfield, Catalysis Letters, 31 (1995) 267. K. M. Reddy, and C. Song, submitted for publication. C. Song, and S. Kirby, Microporous Mater., 2 (1994) 467. A. D. Schmitz, and C. Song, Fuel Chemistry Division Preprints of 208th ACS Symp., Washington D. C., 1994. p.986. C. Song, and K. Moffatt, Microporous Mater., 1994,2(5), 459. Figure 1. A typical X-ray powder diffraction pattern of AI-MCM-41 molecular sieves 1.5 2.5 35 4.5 5.5 6.5 7.5 8.5 9.5 Two Thela Table 1. Synthesis and physical characteristics of mesoporous molecular sieves SiO2/A1203 (mole ratio) BET surface Pore Size from Sample Source of AI Input Output k e a (m2/g) sorption (A) MRK9a AI isopropoxide 100 88.4 1147 27.67 MRK9b AI isopropoxide 50 53.8 1206 28.02 MRKlOa Catapal B 100 95.5 1010 21.92 MRKlOb Catapal B 50 44.3 ____ MRKl la Al sulfate 100 164.6 834 25.38 MRKl lb Al sulfate 50 87.4 ---- -____ --___ I005 Table 2. Naphthalene hydrogenation over Pt/MCM-41 catalysts Reaction conditions: 0.1 g catalyst, 1.0 g naphthalene, 1000 psi H2 pressure 200 OC temperature and 1 hour reaction time naphthalene Product distribution (wt%) t-lc- Catalyst conv. (%) tetralin t-decalin c-decalin total decalins decalins MRK9b 100.0 0.18 33.15 66.67 99.82 0.497 MRKlOb 99.7 25.50 18.01 56.48 74.49 0.319 MRKllb 100.0 0.00 32.25 67.75 100.00 0.476 Table 3. Hydrogenation of phenanthrene over Pt/MCM-41 catalysts Reaction conditions: 0.1 g catalyst, 1.0 g phenanthrene, 1500 psi H2 pressure 300 -C temperature and 2 hours reaction time Product distribution (wt%) Phenanthrene Conv. (%) 79.61 88.00 66.63 1,2,3,4-tetrahydrophenanthren(eT HP) 7.03 6.84 14.69 9,lO-dihydrophenanthrene (DHP) 41.59 45.47 54.29 sym-octahydrophenanthrene (sym-OHP) 14.82 31.25 13.25 unsym-octahydrophenanthrene (unsym-OHP) 15.27 12.99 9.51 sym-OHNsym-OHP 1.41 0.07 0.09 MRK9h MRKlOb MRKlIb sym-octahydroanthracene (sym-OHA) 16.70 1.40 7.7 1 tetradecahydrophenanthrenes (TDHP) 3.94 2.03 0.00 Table 4. Isopropylation of naphthalene over MCM-41 catalysts Reaction conditions: 0.1 g catalyst, 1.0 g naphthalene, 150 psi propylene 200 0C temperature and 2 hours reaction time Product distribution (wt%) Catalyst WMRK9b Pt/MRK9b PtlMRKlOb WMRKllb naphthalene Conv. (%) 92.48 96.62 37.41 90.25 2-isopropylnaphthalene 1 1.82 9.13 24.59 11.57 1-isopropylnaphthalene 16.27 9.88 56.13 20.24 diisopropylnaphthalenes 42.1 1 39.25 16.4 I 42.19 triisopropylnaphthalenes 25.37 34.26 2.51 22.35 tetraisopropylnaphthalens 4.41 7.46 0.29 3.63 2,6-diisopropylnaphthalenes 3.18 4.87 0.80 2.09 2.7 di isopropyl naphthalenes 3.96 3.83 0.77 2.09 Table 5. Hydrocracking of 133-triisopropylbenzene over MCM-41 catalysts Reaction conditions: 0.1 g catalyst, 1.0 g triisopropylbenzene, 1500 psi H1 pressure 350 0C temperature and 2 hours reaction time Product distribution (wt%) Catalyst no catalyst WMRK9b WMRK9b 1,3,5-triisopropylbenzene Conv. (%) 1.02 67.69 100.0 isopropy lbenzene 0.00 19.56 0.0 1,3-diisopropylbenzene 94.20 63.09 0.0 1,4-dUsopropylbenzene 3.30 7.71 0.0 others, mainly C,-C, hydrocarbons 4.10 9.63 100.0 1006 L ? ZEOLITE-CATALYZED RING-SHFT ISOMERIZATION OF sym-OCTAHYDROPHENANTHRENE INTO sym-OCTAHYDROANTHRACENE. EXPERMENTAL RESULTS AND CALCULATED EQUILIBRIUM COMPOSITIONS Wei-Chuan Lai, Chunshan Song*, Adri van D u d and J. W. de Leeuw! Fuel Science Program, 209 Academic Projects Building, The Pennsylvania State University, University Park, Pennsylvania 16802, U3.A *Division of Marine Biogeochemistry, Netherlands Institute of Sea Research (NIOZ), P.O. Box 59. 1790 AB Den Burg, The Netherlands Keywords: Zeolites. isomerization, octahydrophenanthrene, octahydroanthracene INTRODUCTION Phenanthrene and its derivatives are abundant in coal-derived liquids from coal carbonization. pyrolysis. and liquefaction; however, they are used in industries only to a limited extent despite of considerable efforts [Song and Schobert. 19931. On the other hand, their isomers, anthracene and !tS derivatives such as sym-octahydroanthracene (sym-OHA), are more useful materials for industrial applications. Anthracene and its derivatives may be used as the starting materials for the manufacturing of anthraquinone (an effective pulping accelerator), pyromellitic dianhydride (PMDA, the monomer for polyimides such as Du Pont's Kapton), and dyestuffs [Song and Schoben. 1993, 1995). Thus, it is desirable to conven phenanthrenes to anthracenes. There has been much research on the catalytic hydroprocessing of phenanthrene under high H2 pressures and at temperatures generally in excess of 623 K [Nakatsuji et al.. 1978; Haynes et al., 1983; Salim and Bell, 1954: Lee and Salterfield, 1993; Girgis and Gates, 1994; Landau et al., 1994; Korre et al.. 19951. However, relatively little information about the ring-shift isomerization of phenanthrenes into anthracenes at lower temperatures is available. It has been shown in our earlier exploratory work that some chemically modified mordenites and Y-zeolites may selectively promote the transformation of sym-octahydrophznanthrene( sym-OHP) into sym-OHA [Song and Moffatt. 1993, 39941. Cook and Colgrove (1994) reponed the acid catalyzed isomeriration of phenanthrene, anthracene, and sym-OHA under FCC conditions (755 K). The objective of this work is to clarify the effects of reaction conditions and catalyst properties on the catalytic isomerization of sym-OHP into sym-OHA. In addition, the Molecular Mechanics (MM3) [Allinger et al., 19901 calculations were performed lo find the equilibrium coriipositions of sym-OI1P and qm-OHA, and to establish the upper limit of the catalytic conversion. We wish to establish activity and selectivity data that could point to an inexpensive way of making anthracene derivatives from phenanthrene denvatives. EXPERIMENTAL The sym-OHA and Aym-OHP chenucals were obtained from Aldrich Chemical Company and TCI America, respectively, and were used as received. Their purities were analyzed in our laboratory using gas chromatography (GC) and gas chromatography-mass spectrometry (GC-MS). The catalysts used in the catalytic isomerization reactlons include: two hydrogen mordenites (HML8 and HM30A) and two noble metal loaded mordenitcs (WHM3OA and Pd/HM30A). PmM30A and Pd/HM30A were prepared by incipient wetness impregnation method; Le., the salt of platinum and palladium were dispersed into the mordenites by incipient wetness impregnation of corresponding aqueous H2PtC16 or H2PdC14 dissolved in hydrochloric acid. The noble metal loading on the support was kept at nominally 6 wtR. The metal-loaded catalysts were calcined in air at 723 K for 2 h after being dried in vacuum oven. The details of the preparation and properties of the catalysts are described elsewhere [Song and Moffatt. 1994; Schmitz et al., 19941. Catalytic isomerization reactions were carried out in 28-mL horizontal type sbiinless steel tubing bomb reactors, which were charged with 0.6 mmole of sym-OHP or sym-OHA (0.1 12 9). 1 ml of 1.3.5-trimethylbenzene solvent, and 0.2 g of catalyst, at 473-573 K for 0.15-12 h under an initial pressure of0.79 MPa UHP N2 or H2. The unifomuty of concentration and temperature inside thc reactor was obtained by agitating the reactor vertically at 240 cycledmin. After the reaction, the gas products were collected in a gas bag, and the liquid products were recovered by washing with acetone. The recovered catalyst was stored in a vial for thermogravimeuic analysis performed later. The gaseous products were quantitatively analyzed using a Perkin-Elmer Autosystem GC equipped with two detectors, a thermal conductivity detector (TCD) and a flame ionization detector (FID). The liquid products were analyzed on an HP 589011 GC coupled with an HP 5971A Mass Selective Detector (MSD) and quantified by a Perhn-Elmer GC 8500 equipped with an FID. More details for the analytical procedures may be found elsewhere [Song et al., 1994). RESULTS AND DISCUSSION ~ a l c u l a t e de quilibrium compositions from MM3. The equilibrium compositions of sym- OHP and sytn-OHA at three temperatures were calculated to establish the theoretical upper linut of the catalytic conversion. The procedures are as follows. Calculations on the various conformers present in an equilibrium mixture containing only sym-OHA and sym-OHP were first performed. In order to obtain the raw geometries for the molecules and to find the different conformers present, the DELPHI-molecular mechanics program was used with the MM3 force field. The obtained minimized geometries were used as starting points for the MM3 program. using the MM3 force field (both the 1992 versions). Only slight differences were found between the optimized DELPHIand MM3-geomeuies. Five and six different conformers were found for sym-OHA and sym-OHP, 1007 J I i respectively, as shown in Table 1. Table 1 also shows the calculated heats of formation and entropies for the different conformations of sym-OHA and sym-OHP at 298 K. The literature. value of the heat of formation of sym-OHA is -8.89 kcallmol [Pedley et al., 19861. The good agreement between the literature value (-8.89 kcal/mol) and the calculated one (-9.17 kcal/mol, in Table 1) indicates that the MM3 force field should be applicable to problems like these. From these data, the Gibbs free energy Ofsym-OHA and sym-OHP at 298 K could be obtained. In similar fashion, the Gibbs free energies of these compounds at 473, 523, and 573 K may be calculated by correcting the heats of formation and entropies in Table 1 for temperatures. From these free energies, the equilibrium composition of the sym-OHNsym-OHP mixture was calculated. It should be noticed that one may get different MM3 calculation results from using different estimated thermodynamic parameters of sym-OHA and sym-OHP. It was found that the error in the MM3 calculations using current parameters for such kind of mixture compositions is about 10%. The calculated mixture compositions will be presented later and compared with experimental data following the presentation of experimental results. Effectiveness of HMLS. HML8 was studied more extensively than other catalysts in this work, because it was found to be the best among several zeolitic catalysts in our previous work at 523 K for 2 h [Song and Moffatt, 1993, 19941. Table 2 presents the results of sym-OHP isomerization over HML8 with sym-OHP as the starting material at three different temperatures (473, 523, and 573 K) under N2 environment. The purity analysis results of the starting material was also presented in Table 2. It can be seen that the as-received reactant contains about 91% of sym-OHP, 2.9% of sym-OHA, 5.5% in total of various hydrogenated phenanthrenes (asym-OHP, 9,10-DHP, 1,2,3,4,4a,IOa-HHP, and 1,2,3,4-THP), and 0.6% of other impurities. The purity analysis is important, because without this information it is possible that impurities are mistaken as products. The reactivities of 9.10-DHP and 1,2,3,4,4a,IOa-HHP will be briefly discussed first. 9.10-DHP does not react to a great extent except under severe conditions, e.g., 573 K for 1 h; on the other hand, 1,2,3,4,4a,lOa-HHP reacts quickly even at a lower temperature, 473 K. The main reaction of 9,10-DHP, which accounts for about 2.7% of the starting reactant, was believed to be dehydrogenation to phenanthrene under current reaction conditions. The supporting evidence is from the fact that the sum of 9,IO-DHP and phenanthrene was relatively constant (about 2.8) for most of the reaction conditions. Other researchers also suggested the hydrogenation of 9,lO-DHP to 1,2,3,4-THP [Lemberton and Guisnet, 19841 and asym-OHP [Nakatsuji et al., 19781; however, these hydrogenation reactions are less likely under the current reaction conditions judging from the deficiency of hydrogen and the overall product distribution of 1,2,3,4-THP and asym-OHP. To verify this, experiments need to be performed with 9.10-DHP as the sole reactant. For 1,2,3,4,4a, loa-HHP, the main reactions were believed to include dehydrogenation to 1,2,3,4- THP, and ring-contraction isomerization and ring-opening cracking. The major reaction of interest in this work is the isomerization of sym-OHP to sym-OHA; Table 2 shows that HML8 was an effective catalyst for this reaction because it could afford over 90% selectivity with reasonable conversion (about 50%). In order to approach equilibrium conditions, relatively long residence times have been employed, e.g., 12 h at 473 K, 4 h at 523 K, and 1 h at 573 K. From Table 2, the reactions seemed to have reached the asymptotic equilibrium after 12 h at 473 K, 2 h at 523 K, and 0.25 h at 573 K. In other words, the yield of sym-OHA reached its maximum at those reaction conditions, and longer reaction times at 523 K and 573 K only served to decrease the sym-OHA yield because of enhanced competitive side-reactions or secondary reactions such as ring-opening cracking and subsequent dealkylation (to form mainly alkyltetralins), conventional ring-contraction isomerization, and dehydrogenation. The selectivity towards sym- OHA suffered severely when the reaction temperature was increased to 573 K, when side-reactions became significant. Table 3 compares the pseudo-equilibrium yields from MM3 calculations and experimental results at three temperatures. There exists discrepancy between the calculated data and the experimental results. The MM3 calculations indicate that the reaction temperature has a moderate effect on the equilibrium ratio of sym-OHA to sym-OHP. For example, the ratio changes from 2.56 to 1.95 when the temperature is increased from 473 K to 573 K. However, the experiments show that although the reaction temperature affects the reaction rate and selectivity, the equilibrium ratio is not sensitive to the reaction temperature. For example, the equilibrium ratios are all close to 1.3 for the temperature range studied. Table 3 shows that the pseudo-equilibrium yields of sym-OHA from experiments are considerably lower than the calculated equilibrium yield from MM3 method. There might be some explanations for the discrepancy. The calculation assumes that sym-OHP and sym- OHA are the only reacting species in the reaction system. However in the real experimental system since there are other reactants as well as by-products, the actual equilibrium state now depends on more simultaneous chemical reactions. For the current reaction system, at higher temperatures, e.g., 523 K and 573 K, there are significant side reactions as can be seen from the smaller sym- OHA selectivity or the decreasing sum of sym-OHP and sym-OHA. Other simultaneous reactions might include the reactions due to the solvent used (1.3.5-trimethylbenzene) as well. Since many reactions are involved, the composition is a resultant of all the operative reactions and is determined by the thermodynamic equilibrium for the total system. In short, the presence of such simultaneous reactions have shifted the equilibrium state of sym-ow and sym-o~. Some experiments using purc syni-OHA (rather than sym-OHP) as starting material were also performed to check the reversibility of the ring-shift isomerization and to confirm the experimentally obtained pseudo-equilibrium composition (Table 3). The results of sym-OHA isomerization over HML8 with sym-OHA as the starting material at three temperatures under N2 environment are given 1008 i f I in Table 4. The results demonstrate the reversibility of the ring-shift isomerization. Besides, it Was shown, again that HML8 was quite selective for the ring-shift isomerization except under severe condmns. From Table 4, the reactions seemed to have reached pseudo-equilibrium after 4 h at 473 K, 0.5 h at 523 K, and 0.15 h at 573 K judging from the fact that the yield of sym-OHP reached its maximum at those reaction conditions. The pseudo-equilibrium compositions of sym- OHA and sym-OHP and their ratios reported in Table 4 are very close to those reported earlier (see Tables 2 and 3); for example, the equilibrium ratio is not sensitive to the reaction temperature and is close to 1.3. From the data in Tables 2-4, on the basis of reaction rate alone, it appears desirable to operate the reaction at a high temperature level such as 573 K. However, an examination of the reaction selectivity reveals that the selectivity towards sym-OHA drops rapidly at 573 K due to significant side-reactions. Taking both rate and selectivity into consideration, it seems that 523 K for 0.5 h might be the desirable condition for the sym-OHP isomerization over HML8. Effectiveness of HMJOA. HM30A was used to examine the possible effect of dealumination on the ring-shift isomerization in comparison to HML8. The Si02/A1203 molar ratios of HML8 and HM30A are 17 and 35, respectively. Table 5 presents the results of sym-OHP isomerization over HM30A with sym-OHP as the starting material at two temperatures (473 and 523 K) under N2. Tables 2 and 5 suggest that both catalysts reacted comparably near the pseudo-equilibrium condition at 473 K for 12 h; this could be seen from the similar conversion level, selectivity, and sym-OHA to sym-OHP ratio. However, at 523 K, they behaved slightly different. The Hmordenites with relatively larger SiO2/A12O3 ratio, i.e., HM30A. seem to exhibit higher activity for sym-OHP conversion but lower selectivity towards sym-OHA plus THA. The comparison of selectivity could be seen at similar conversion level, 51.9%. where the selectivity is 91.9% and 83.5% for HML8 and HM30A, respectively. In short, HM30A shows promising results at 473 K in terms of both activity and selectivity, but is not as selective as HML8 at 523 K. More experiments with additional catalysts will be needed to clarify the effect of dealumination on the ring-shift isomerization. Effectiveness of PtIHMJOA and Pd/HM30A. The motivation of using these two noble metal loaded mordenites on the ring-shift isomerization comes from the promising activity and selectivity results of HM30A at 473 K as well as our study on conformational isomerization of cisdecalin using the same two catalysts [Lai and Song, 19951. It was found that Pt- and Pd-loaded mordenites are very effective catalysts under H2 atmosphere for the conformational isomerization of cis-decalin even at 473 K. Because of their high activity and selectivity on the isomerization of naphthalene derivatives, W 3 O A and PdlHM3OA were also used in this study. Table 6 presents the results of isomerization over these two catalysts with sym-OHP or sym-OHA as the starting material at 473 K under 0.79 MPa of H2 or N2. The effect of noble metals can be seen by comparing the results of HM30A (PS42, in Table 5) and those of Pt- and Pd-loaded HM30A (PS47 and PS48, respectively, see Table 6). Mordenites loaded with noble metal exhibits much higher activity but lower selectivity towards hydrogenated anthracenes. The activity for sym-OHP conversion is: Pd/HM30A > Pt/HM30A > HM30A. The lower selectivity towards sym-OHA for both PdMM30A and Pt/HM30A is due to the significant dehydrogenation reactions to form THA and THP or even phenanthrene under N2 environments. The results are not beyond expectation because both Pt and Pd are well known active metals for dehydrogenation and hydrogenation. Hydrogenation was not apparent because of the H2 deficient environment, since reactions were performed under N2. The dehydrogenation activity drops in the order Pd > Pt. As can be seen from Table 6, changing the gas environment from N2 to HZ significantly affects the final product distribution because the dehydrogenation under N2 was replaced with hydrogenation under Hz. Figure 1 presents the temporal plots of major products from sym-OHP isomerization using Pt/HM30A and Pd/HM30A at 473 K under excess Hz (8 mmole). Figure 1 provides information about the reaction pathways. For example, for the catalyst Pt/HM30A, it can be seen that sym-OHA was the primary product; its yield reached a maximum at 0.3 h and then decreased due to enhanced hydrogenation. On the other hand, PHA and PHP appeared to be the secondary products resulting from hydrogenation of OHA or OHP. Table 6 shows that for W 3 0 A after 0.3-h reaction time, the ratio of sym-OHA to sym-OHP approached a constant value of 1.28, which was very close to the pseudo-equilibrium mole ratio determined from the study of HML8 and HM30A. All these results suggest that sym-OHA formation be the major primary reaction and hydrogenation of OHP and OHA did not occur until sym-OHA and sym-OHP reached pseudoequilibrium. In addition, it was demonstrated that Pt/HM30A shows promising activity and selectivity at 473 K under H2. Optimal products may be reached in a short period of time, e.g., 0.15-0.3 h. Longer reaction time under H2 environment beyond the pseudo-equilibrium stage is not beneficial to the production of sym-OHA, due to pronounced hydrogenation to form deeper hydrogenation products such as perhydrophenanthrene and perhydroanthracene . The other noble metal catalyst, Pd/HM30A, showed slightly different results. Pd again shows higher activity but lower selectivity towards hydrogenated anthracenes. Its superior hydrogenation ability might have changed the reaction pathways. For example, instead of being a secondary product only, PHP might as well be a primary product too. The hydrogenation and isomerization reactions might have proceeded in parallel instead of in series for Pd/HM30A under H2 environment. The supporting evidence is that the ratio of sym-OHA to sym-OHp did not approach the pseudo-equilibrium mole ratio until very late in the reaction when sym-OHP was almost completely consumed. The results seem to suggest that such strong ability of dehydrogenation and 1009 hydrogenation might make Pd a less favorable catalyst for the ring-shift isomerization of sym-OW. The results of sym-OHA isomerization over F't/HM30A and Pd/HM30A with sym-OHA as the starting material under Hz environment, are also given in Table 6. The results demonstrate the reversibility of the ring-shift isomerization and the important role of deeper hydrogenation. SUMMARY The MM3 calculations were performed to find the equilibrium compositions of sym-OHP and sym- OHA and were compared to experimental results using four catalysts including HML8, HM30A, PtMM30A. and PdMM3OA. The MM3 calculations showed a moderate effect of the reaction temperature on the equilibrium ratio of sym-OHA to sym-OHP. However, the experiments showed that although the reaction temperature affected the reaction rate and selectivity, the equilibrium ratio was not sensitive to the reaction temperature. The presence of simultaneous side reactions was believed to have shifted the equilibrium state. The rate and selectivity of the isomerization reactions depended on both the metal and support type of the catalysts, but the equilibrium ratio was not sensitive to the catalysts used and was close to a constant value of 1.3. The activity for sym-OHP conversion is: PdMM30A > PaM30A > HML8 = HM38. Longer reaction time beyond the pseudo-equilibrium stage was not beneficial to the production of sym-OHA, especially for PdRfM30A and Pt/HM30A due to the pronounced hydrogenation or dehydrogenation. Pt/HM30A showed promising activity and selectivity at 473 K under H2 with optimal reaction time of 0.15-0.3 h. The desirable condition for HML8 was 523 K for 0.5 h. HM30A showed promising results at 473 K, but was not as selective as HML8 at 523 K. Too strong an ability of dehydrogenation and hydrogenation might make Pd/HM30A a less favorable catalyst for the ring-shift isomerization of sym-OHP. ACKNOWLEDGMENTS We are grateful to Dr. T. Golden and Dr. V. Schillinger for providing the mordenites samples, and to Dr. A. Schmitz of PSU for preparing the noble metal catalysts. C.S. would like to thank Prof, H. H. Schobert and Prof. P. B. Weisz for their encouragement. REFERENCES Allinger, N. L.; Li, F.; Yan, L.; Tai, J. C. J. Computational Chem. 1990,11, 868-895. Cook, B. R.; Colgrove, S. G. Prepr.-Am. Chem. Soc., Div. Pet. Chem. 1994, 39 (3), 372-378. Girgis, M. J.; Gates, B. C. Ind. Eng. Chem. Res. 1994,33, 1098-1106, Haynes, H. W. Jr.; Parcher, J. F.; Helmer, N. E. Ind. Eng. Chem. Process Des. Dev. 1983.22, Korre, S. C.; Klein, M. T.; Quann, R. J. Ind. Eng. Chem. Res. 1995.34, 101-117. Lai, W.-C.; Song, C. Prepr.-Am. Chem. Soc.. Div. Fuel Chem. 1995.40, in press. Landau, R. N.; Korri, S. C.; Neurock, M.; Klein, M. T. In Catalytic Hydroprocessing of Petroleum and Distillates. M. C. Oballa and S. S. Shih, Eds., Marcel Dekker, Inc.: New York, 1994, pp421-432. Lee, C. M.; Satteriield, C. N. Energy & Fuel 1993, 7, 978-980. Lemberton, J.-L.; Guisnet, M. Appl. Catal. 1984, 13, 181-192. Nakatsuji, Y.; Kubo, T.; Nomura, M.; Kikkawa, S. Bull. Chem. Soc. Jpn. 1978,51, 618-624. Pedley, J. B.; Naylor, R. D.; Kirby, S. B. Thermochemical Data of Organic Compounds. 1986, Chapman and Hall: London. Salim, S. S.; Bell, A. T. Fuel 1984.63, 469-476. Schmitz, A. D.; Bowers, G.; Song, C. In Advanced Thermally Stable Jet Fuels. Technical Report, U.S. Department of Energy, DE-FG22-92PC92104-TPR-9O,c tober 1994, H. H. Schobert et al., Eds., pp37-42. Song, C.; Moffatt, K. Prepr.-Am. Chem. Soc., Div. Pet. Chem. 1993.38 (4). 779-783. Song, C.; Schobert, H. H. Fuel Processing Technology 1993.34, 157-196. Song, C.; Lai, W.-C.; Schobert, H. H. Ind. Eng. Chem. Res. 1994,33, 534-547. Song, C.; Moffatt, K. Microporous Materials 1994,2,459-466. Song, C.; Schobert, H. H. Prepr.-Am. Chem. Soc.. Div. Fuel Chem. 1995,40 (2), 249-259. 401-409. ' Table 1. The calculated heats of formation and entropies at 298 K Conformation AHf (kcahole) S (caVmol-K) sym-octahydroanthracene double chair (symmetric) -9.17 105.54 double chair (asymmetric) -9.17 105.54 boatchair -6.08 ins 21 double boat (symmetric) -3.00 107.48 double boat (asymmetric) -3.03 107.47 sym-cctahydrophenanthrene double chair (symmetnc) -7.98 106.33 double chair (asymmetric) -7.57 106.89 boatchair I -5.05 108.78 boat-chair U -4.83 108.76 double boat (symmetric) -1.72 108.82 double h a t (asymmetric) -1.62 108.71 1010 Table 2. sym-OHP isomerization over Hh4L8 with sym-OHP as the starting material Expt. Temp. Time Roducts (W% of feed) x c ~ ~ 1 . dR atio no. (0 (h) sym- m a sy m- q m - 9.10. HHPa THPa Phen.a Othersb (46) (%) sym-OHA OHA OHP OHPa DHPa sym-OHP 0.w 2.90 91.04 0.32 2.67 1.74 0.77 0.07 0.50 0.03 PS6 473 0.50 14.70 0.11 77.99 0.35 2.78 0.43 1.40 0.15 2.10 13.1 91.2 0.19 PS8 473 1.00 28.10 0.21 64.38 0.34 2.59 0.52 1.20 0.20 2.45 26.7 95.3 0.44 ps9 473 2.00 42.30 0.26 50.35 0.36 2.57 0.22 1.16 0.20 2.58 40.7 97.5 0.84 PSI4 473 4.00 49.64 0.33 42.25 0.35 2.60 0.04 1.21 0.17 3.41 48.8 96.5 1.17 PSI5 473 6.00 51.44 0.45 40.20 0.33 2.54 0.04 1.29 0.16 3.55 50.8 96.4 1.28 PS41 473 12.0 52.15 0.42 40.07 0.35 2.62 0.04 1.24 0.17 2.94 51.0 97.4 1.30 PSI0 523 0.25 35.90 0.48 55.62 0.36 2.53 0.51 1.21 0.20 3.18 35.4 94.5 0.65 PSI 523 0.50 48.76 1.0040.70 0.30 2.53 0.12 1.67 0.33 4.59 50.3 93.1 1.20 PS3 523 2.0049.32 1.3039.11 0.22 2.23 0.03 1.65 0.49 5.65 51.9 91.9 1.26 PSI8 523 4.00 47.74 1.56 38.42 0.13 1.95 1.88 0.75 7.57 52.6 88.2 1.24 PSI3 573 0.15 44.96 1.21 41.20 0.20 2.20 0.12 2.03 0.60 7.48 49.8 86.8 1.09 PS7 573 0.25 45.50 1.95 36.58 0.10 1.50 0.27 2.26 0.80 11.04 54.5 81.8 1.24 PS4 573 0.50 42.67 2.13 35.64 0.15 1.75 0.12 3.70 1.12 12.72 55.4 75.6 1.20 PS5 573 1.00 33.62 3.20 27.19 0.10 0.80 0.12 3.98 1.53 29.46 63.9 53.1 1.24 a THA = 1.2,3.4-tetrahydroantene; asym-OHP = 1,2,3.4,4a,9,10,10a-octahydrophenan~ne; 9,lO-DHP = 9, IOdihydrophenanthrene; HHP = 1,2,3,4,4a, loa-hexahydmphenanlhrehrene: THP = 1.2,3.4-teuahydrophenanthrene; Phen. = phenanthrene. others include products of ring-contraction isomerization and ring-opening cracking and subsequent dealkylation. X = conversion of sym-OHP (weight % of feed). This row presents the purity of as-received sym-OHP. * Selectivity to sym-OHA plus THA, defined as the percentage of sym-OHP conversion. e Table 3. Pseudo-equilibrium composition of the sym-OHAlsym-OHP mixture Temperature MM3 calculation Experimental dataa (K) OHA ; OHP Ratio OHA : OHPb Ratio 473 71.9% : 28.1% 2.56 52.2% : 40.1% 1.30 523 68.7% : 31.3% 2.19 49.3% : 39.1% 1.26 573 66.1% : 33.9% 1.95 45.5% : 36.6% 1.24 a Data from Table 2 12 hat 473 K, 2 hat 523 K. and 0.25 h at 573 K. Other products make up the remainder Table 4. sym-OHA isomerization over HMLA with sym-OHA as the starting material Expt. Temp. Time Products (wt% of feed) Xa Se1.b Se1.c Ratio no. (K) 01) sym- THA sym- HHP THP Others (%) (95) (9%) sym-OHA OHA OHP sym-OHP PS27 473 1.00 61.44 0.23 35.52 0.18 0.08 2.54 38.6 92.1 92.8 1.73 PS28 473 4.00 55.07 0.40 41.22 0.14 0.10 3.07 44.9 91.7 92.3 1.34 PS33 473 8.00 55.42 0.39 40.56 0.16 0.09 3.38 44:6 91.0 91.5 1.37 PS25 523 0.25 56.18 0.67 37.95 0.25 0.20 4.75 43.8 86.6 87.6 1.48 PS26 523 0.50 53.36 0.76 40.69 0.27 0.30 4.62 46.6 87.2 88.5 1.31 PS32 523 2.00 51.38 1.04 39.91 0.19 0.57 6.91 48.6 82.1 83.6 1.29 PS29 573 0.15 50.91 1.26 39.15 0.23 0.92 7.53 49.1 79.8 82.1 1.30 PS30 573 0.25 45.97 1.67 36.39 0.18 1.57 14.22 54.0 67.4 70.6 1.26 PS31 573 0.50 40.65 2.28 32.36 0.16 2.52 22.03 59.4 54.5 59.0 1.26 a X = convenjon of sym-OHA (weight % of feed, sym-OHA). b Selectivity to sym-OHP. defined as the percentage of sym-OHA conversion. C Selectivity to sym-OHP plus HHP and THP. defined as the percentage of sym-OHA conversion. Table 5. sym-OHP isomerization over HM30A with sym-OHP as the staning material products (W% of feed) Xa Se1.b Ratio . Expt. Catalyst Temp. Time no. type 0 (h) sym- THA sym- asym- 9JO- HHP THP Phen. Others (46) (5%) sym-OHA OHA OHP OHP DHP sym-OHP 0.03 ps42 HM30A 473 12.00 52.57 0.64 39.03 0.30 2.39 0.14 1.40 3.53 52.0 96.7 1.35 Psi6 HM3OA 523 0.50 44.47 1.72 39.19 0.21 1.55 0.14 2.34 1.05 9.33 51.9 83.5 1.13 psi7 HM30A 523 1.00 42.87 2.06 35.90 0.12 1.20 0.11 2.54 1.29 13.91 55.1 76.2 1.19 0.00C 2.90 91.04 0.32 2.67 1.74 0.77 0.07 0.50 1011 Table 6. Ring-shift isomerization over noble metal loaded mordenites with sym-OHF' or sym-OHA as the starting material at 473 K under 0.79 MF'a of H2 or N2 Catalyst wHM30A Pd/HM30A Time (h) O.Wf 0.30 0.15 0.30 1.00 2.00 2.00 0.30 0.15 0.30 2.00 2.00 Staning reactant OHP OHP ow ow OHP OHA OHP OHP OHP OHP OHA 0.79-MPa N2 or H2 N2 H2 H2 H2 Hz HZ N2 H2 H2 H2 H2 Expt. no. Roducls (wt% of feed) sym-OHA ?nA PHAa sym-OHP asym-OHP 9.10-DHP IMP THP Phenanthrene PHPa Others x ( W b SelectivityC( So) Selectivity" (%) Selectivitve (So) PS47 PS45 PS43 PS36 PS39 PS38 PS48 PS46 PS44 PS37 PS40 2.90 42.52 47.82 48.07 40.62 34.76 37.21 38.10 30.17 26.53 8.01 4.03 2.57 3.70 9.94 12.63 12.06 10.00 16.72 33.77 40.1 I 91.04 33.07 40.16 38.28 31.72 27.13 29.44 29.99 36.17 24.80 6.47 3.21 0.32 1.14 2.62 3.53 3.67 3.65 3.61 1.95 4.23 3.47 1.55 0.82 2.67 0.71 0.01 0.03 0.13 0.14 1.47 0.01 1.74 0.02 0.02 0.06 0.06 0.03 0.02 0.02 0.08 0.77 10.45 0.12 0.91 0.36 12.66 0.06 3.35 5.20 0.08 5.25 5.06 11.53 15.12 13.24 0.07 16.94 25.34 45.85 47.40 0.50 0.89 1.55 1.30 2.31 5.44 3.89 1.54 2.46 3.06 4.35 4.43 58.0 50.9 52.8 59.3 63.9 62.8 61.1 54.9 66.2 84.6 96.0 81.7 93.3 92.6 80.3 69.9 72.4 67.9 60.9 46.0 46.9 3.3 74.5 53.6 7.77 0.20 0.05 9.00 sym-OHA/sym:OHP 0.03 1.29 1.19 1.26 1.28 1.28 1.26 1.27 0.83 1.07 1.24 1.26 a PHA = pxhydroanthracene; PHP = perhydrophenanthrene. X = conversion of sym-OHP or sym-OHA (weight % of feed). Selectivity to sym'-OHA plus "HA and PHA. defined as the percentage of sym-OHP conversion. Selectivity to sym-OHP. defined as the percentage of qm-OHA conversion. e Selectivity to hydrogenated phenanthrenes, defined as the percentage of sym-OHA conversion. This column presents the purity of as-received sym-OHP. (a) PtlHM30A 50 - 40 - 30 - 20 - I sym-OHA 11 sym-OHP 0.0 0.5 1 .o 1.5 2.0 Time (h) 50 (b) PdlHM30A - 0.0 0.5 1.0 1.5 2.0 Time (h) Figure 1. Temporal plots of major products from sym-OHP isomerization at 473 K under H2. 1012 i I ADAMANTANES FROM PETROLEUM, WITH ZEOLITES L. Deane Rollmann, Larry A. Green, Robert A. Bradway, and Hye Kyung C. Timken J i Mobil Research and Development Corporation, Paulsboro Research Laboratory, PO Box 480, Paulsboro, NJ 08066 Keywords: Adamantane, zeolite, hydrocracking ABSTRACT Circumstances have been found under which adamantanes are significantly concentrated and, it is believed, formed in a petroleum refinery, and catalysts have been identified which are effective in recovering these compounds from a complex mixture of similarly boiling hydrocarbons. In an example detailed below, nearly 10% adamantanes, largely methyl-substituted derivatives, were found in and isolated from a refinery stream by selectively removing the non-adamantanes with a Pt-containing zeolite Beta catalyst. INTRODUCTION ' Despite their discovery in the early 1930s in the heavy Hodonin crude of eastern Europe (1 ), adamantanes occasioned relatively little interest until a facile chemical synthesis was reported, in 1957 (2). Although notable as part of the diamondoids found in certain natural gas condensates (3, 4), adamantanes appear never to exceed about 0.02-0.04% in crudes (5), a concentration too low for economic recovery. Unsubstituted adamantane was first prepared by the AICI3-catalyzed isomerization of hydrogenated cyclopentadiene dimer, tetrahydrodicyclopentadiene (THDCP), an approach which was quickly expanded to include a number of methyl adamantanes (6). Solid acid catalysts such as silicaalumina (7) and HY zeolite (8) were also able to effect the THDCP-adamantane transformation, but none was apparently competitive in yield and stability with AIC13 and/or AIBr3. of adamantanes was a change in feedstock, from THDCP to a variety of tricyclic perhydroaromatics (9). The effective catalyst was an AIX3-HX-hydrocarbon mixture, where X = chloride or bromide. The products were methyl or polymethyl adamantanes, each having the same molecular weight as the feed tricycloalkane, often in a yield of some 60% or more. An example is shown in Figure 1. Subsequent work showed that numerous acid catalysts would convert tricyclic naphthenes (tricycloalkanes) into methyl adamantanes, namely, chlorinated PVAI2O3 (1 I), silicaalumina (12). silica-alumina with Group Vlll metal (13), and REWREY, usually with Group Vlll metal (1 4). exist in a crude in the form of high boiling polycyclic naphthenes and aromatics. In a modern refinery, these precursors, boiling above about 3OO0C, commonly encounter acid catalysts in both a fluid catalytic cracking (FCC) unit and in a hydrocracker (HDC). Thus, the present experiments focused on HDC recycle streams. EXPERIMENTAL (16). Framework SiO~/A1203ra tios were approximately 50 for Beta and 200 for 'ultrastable" Y (USY). For better comparison with the USY, a sample of the Beta catalyst was dealuminated to a similar framework SiO2/ AI203 ratio and designated "low activity" Beta (LoAct-Beta). In the experiments, all catalysts contained alumina binder, all were 24/60 mesh, all contained 0.5% Pt or Pd, and all were brought to initial operating conditions (232% and 2.5 mPa) in flowing hydrogen. The experiments were conducted in a downflow tubular reactor, at 2.5 mPa, with a Hn/hydrocarbon (H21HCJ mole ratio of 3 - 4, at temperatures of 230" 7 330°C, and at 1 - 4 WHSV (weight hourly space velocity). Day-to-day Catalyst aging was not significant in these experiments. In the present context, the most significant post-1960 advance in the synthesis Based on the chemical synthesis work, potential precursors to adamantanes . Two zeolites were used in the experiments, Beta (15) and "ultrastable" Y (USY) 1013 Gas chromatography (gc) results were obtained with a 60m DB-1 capillary column (J&W Scientific, 0.25 mm id, 0 . 2 5 ~fil m). The gc-mass spec analyses were performed on a Kratos Model MSBORFA, with a Hewlett Packard Series II 5890 gc and a 30m DB-5HT column (0.32 mm id, 0.11 film). Ionization was by electron impact. RESULTS AND DISCUSSION Refinery streams selected for testing are shown in Table 1. Since most methyl and ethyl adamantanes boil between 180°C and 240°C. streams were selected to bracket that range, namely, a 135O - 21OOC HDC heavy naphtha, a 175' - 375% HDC recycle stream, a 175' - 26OOC portion of the HDC recycle stream, and a 120' - 245°C hydrotreated kerosene, all from a refinery sourced largely with heavy crude. In addition, a 175O - 290°C analog of the above HDC recycle stream was obtained from a refinery sourced with light, conventional crude. Crude type was a consideration since, in general, heavy crudes are enriched in polycyclic alkanes relative to light, "conventicfinal" crudes (1 7). Isolation of adamantanes. The adamantanes shown in Figure 2a were obtained when the heavy crude HDC recycle stream described in Table 1 was passed over the WLoAct-Beta catalyst at 325OC and 1.2 WHSV. Approximately 90% of the feed, which boiled above 175°C was converted to lower boiling, mostly gasolinerange hydrocarbon. The remaining high-boiling material contained over 70% adamantane and methyl adamantanes. Gas chromatography (gc) and gc-mass spec showed the presence of diamantanes as well. Comparison of Figures 2a and 2b showed the striking similarity between this potential refinery product and a mixture of naturally occurring adamantanes recovered from a deep gas condensate (4). Despite the understandable difference in carbonnumber and isomer distribution, every major peak in the product from Pt/LoAct-Beta corresponded to a peak in the condensate adamantanes. Gc-mass spec confirmed the molecular weights indicated in Figure 2. Adamantanes free from diamantanes were obtained by using a lower boiling portion of the HDC recycle stream. When the above experiments were repeated with a 175' - 260°C fraction of the HDC stream, this time using the high-activity Pt/Beta (50 Si02lA1203 ratio) at 260°C and 2.0 WHSV, conversion to lighter hydrocarbon was 86%. The product "mixed methyl adamantanes" (MMAs) were virtually indistinguishable from those obtained with full-range HDC recycle, and material boiling higher than the MMA's (e.g., diamantanes) was c 0.1% of the product. MMA yield was substantial. With the 175' - 260°C feed, 9.1 g of MMA was obtained from 100 g of feed, representing 32% of the 3RN's. A second experiment, under slightly milder conditions, yielded 9.3 g of MMA, or 33%. The product MMA, separated from lower boiling hydrocarbons by distillation, was a colorless liquid with a density of 0.89 glcc. That the MMAs were associated with the recycle stream was further aflirmed by a "blank" experiment with Pt/Beta and the 135" - 210°C HDC heavy naphtha, at 255°C and 1.9 WHSV. The product containing only 0.6% MMA's, essentially all of which were bridgehead-methyl isomers boiling between 180° and 2OO0C. Yield of adamantanes was much lower in a refinery operating on light, conventional crude. When a 1 :1 blend of the light crude HDC recycle and the HDC heavy naphtha in Table 1 was processed over high-activity WBeta, the product contained only 0.9 % MMA's. A final experiment was conducted to probe for adamantanes in the crude supply to the heavy crude refinery. The feed was the hydrotreated kerosene, a stream which had never contacted a zeolite catalyst but which, given its 120° - 245°C boiling range, should contain any MMA's in the crude. As shown in Table 1, it analyzed 0.6% 3RNs. When processed over both PdlBeta and PUBeta, the products contained 0.4% MMA's. This result strongly suggests that, while some portion of adamantanes did enter the refinery with the crude, the bulk was being formed, either in the HDC (and possibly FCC) unit or in these noble metaVzeolite experiments. Formation of adarnantanes. A model compound was used to probe possible formation of adamantanes over Pt and PdlBeta catalysts under the conditions Of these experiments. Based on the adamantane literature and on commercial 1014 f f availability, perhydrofluorene (PHF) was the selected for most of the experiments. Boiling at 253OC, it should convert to 1,3,5-TriMA, as depicted in Figure 1. The PHF to 1.3.5-TriMA conversion process was largely absent over zeolite Beta, for reasons which will be discussed below. Dissolved at 10% in HDC heavy naphtha and processed over high-activity WBeta at 265% and 1.6 WHSV, the yield of MMA's based on PHF was less than 5%. (PHF was 100% converted.) The very small amount of new MMA's in the product had methyl or ethyl groups on non-bridgehead Cabons, and little or none of the "end" product, 1,3,5-TriMA, was formed. A similarly low MMA yield was obtained with phenanthrene, a molecule which might be expected to hydrogenate and isomerize to 1,3,5,7-TetMA over a noble metaVzeolite catalyst. Choke of zeolite. Beta was selected for the first experiments because it is a member of a class called "large-pore' zeolites (18). a class which includes zeolite Y. The Beta pores, like those of Y, are formed from 12-membered rings of linked silicon and aluminum oxide tetrahedra. The opening of those pores in Beta is an elliptical 6.4 x 7.6 A, while those of Y are a circular 7.4 A (19). Unsubstituted adamantane, a spherical molecule with a Van der Waals diameter of 7.4 A, is known to penetrate the Pores of zeolite Y (8), but should have difficulty penetrating Beta, whose critical pore dimension is only 6.4 A. Thus Beta was the zeolite of choice for isolating adamantanes. It is noteworthy, with respect to this approach to adamantanes, that MMA's had been isolated earlier by hydrocracking narrow-boiling, laboratory crude extracts over 10% Pt-on-diatomite, at temperatures of about 430°C (20). Zeolite Y, while also effective in isolating adamantanes, was much more effective than Beta in generating them, as shown by experiments with PHF. When the PHF experiment described earlier was repeated over Pdp/ at 590°C and 1.7 WHSV, the product contained 3.5% MMA's which, when corrected for the heavy naphtha contribution. represented an approximately 27% yield based on PHF. Gc-mass spec showed four new non-bridgehead products, all C13 MMA's, and 1,3,5-TriMA was enhanced in concentration relative to the other bridgehead isomers. PHF. a C13 molecule, was 100% converted. The PHF experiments demonstrated size-selective differentiation between zeolites Y and Beta, and they strongly suggested that some portion of the adamantanes isolated from these refinery streams were formed in the HDC unit and were only being concentrated in the experiments with Pt and Pd/Beta. Higher severe experiments with Pd/Beta further demonstrated the sizediscriminating ability of this zeolite. When the contact time between HDC heavy naphtha and Pd/Beta was increased, the smallest of the MMA's were converted, namely, unsubstituted adamantane and 1 -MA. The larger MMA's presumably could not enter the pores of zeolite Beta and were essentially unconverted. With PcUY, at the same HDC heavy naphtha conversion levels, the relative reactivity relationships were reversed. The larger MMA's were preferentially converted. CONCLUSIONS These results show that adamantanes, while present in crudes, can both be formed and concentrated in certain refinery operations, most notably in an HDC unit, and that their amount depends on crude source, catalyst, refinery configuration, and operating conditions. A level of some 10% adamantanes is not unexpected in the 175O - 26OOC portion of the HDC recycle stream in a refinery sourced by heavy crude. These adamantanes can be isolated very effectively from such streams by mild hydrocracking Over large-pore zeolite catalysts, such as zeolite Beta. ACKNOWLEDGEMENTS assistance and to M. Granchi, B. Hagee. M. Nicholas, K. Peters, R. Quann, and W. Rogers for helpful input and advice. Samples of deep gas diamondoids were provided by C. Chen and D. Whitehurst. REFERENCES Special thanks go to W. Weimar and F. Daugherty for excellent laboratory I . Landa, S., and Machacek, V., Coll. Czech. Chern. Comrn., p, 1 (1933). 2. Schleyer. P. von R.. J. Am. Chem. Soc., a,32 92 (1957). 3. Sokolova. I. M., Berman, S. S., Abryutina. N. N.. and Petrov. A. A.. Khim. Technol. Topl. Masel, 5.7 (1989). Chem. Abst., m, 81021. 1015 4. 5. 6. 7. 8. 9. Wingert, W. S., Fuel, a,37 (1 992). Petrov, A. A,, "Petroleum Hydrocarbons,' (Springer-Verlag. Berlin, 1984) p. 93. Fort, R. C., Jr., and Schleyer, P. von R., Chem. Rev., M, 277 (1964). Plate, A. F., Nikitina, 2. K., and Burtseva. T. A., Neftekhimiya.1. 599 (.1961.). Ch em. Abst., z, 4938a. Lau, G. C., and Maier. W. F., Langmuir, 9, 164 (1987). Schneider, A., Warren, R. W., and Janoski, E. J., J. Am. Chem. SOC., M, 5365 (1964). 10. &hneider, A,, Warren, R. W., and Janoski, E. J., in "Proceedings, 7th World Pet. 11. Johnston, D. E., McKervey, M. A., and Rooney, J. J., J. Am. Chem. SOC., u, 2798 12. Bagrii, E. I., and Sanin, P. I., US Patent 3,637,876 (January 25, 1972). 13. Landa, S., and Podrouzhkova. V., Petrol. Chem. USSR, 14,135 (1974). 14. Honna, K., Shimizu, N., and Kurisaki. K., US Patent 3,944,626 (March 16, 1976). 15. Wadlinger, R. L., Kerr, G. T., and Rosinski, E. J., US Patent 3,308,069 (1967). 16. Breck, D. W., "Zeolite Molecular Sieves,' (R. F. Krieger, Malabar, FL.1989) p. 507. 17. Tissot, B. P.. and Welte, D. H., 'Peroleum Formation and Occurrence," (Springer- 18. Higgins. J. B., Lapierre, R. B., Schienker, J. L., Rohnan, A. C., Wood, J. D., Kerr. 19. Meier. W. M.. and Olson. D. H.. Zeolites. 12. 1 H 992). Cong., Mexico City," 5, 427 (1967). (1 971). Verlag, New York. ed. 2, 1984) p. 390. G. T., and Rohrbaugh, W. J.. Zeolites, 8, 446 (1988). 20. Yakubson, Z: V., Aref'ev; 0. A.; and PetrorA. A,, Netekhimiya, u, 345 (1973). Chem. Abstr., Bt 11 6756. Table 1. Boiling ranges and three-ring naphthene (3RN) contents of refinery streams tested for adamantanes. ciu!%Ka . . m' Heavy crude HDC recycle stream 175 - 375 24 Fraction boiling below 260' (46 %) 175 - 260 28 Heavy crude HDC heavy naphtha 135- 210 1.2 Light crude HDC recycle stream 175 - 290 3.4 Heavy crude hydrotreated kerosene 120 - 245 0.6 By mass spectrometry 1016 Overall: perhydrofluorene 1,3,5-trimethyladamantane + C13H22 ',,HZ Figure 1. A simplified reaction scheme for the preparation of 1.3,5-trimethyl. adamantane from perhydrofluorene (10). 1-MA 1,3,CTflMA Et4.1 4 CI 4 I ,3,5,7-TetMA a Et- C15 2a n 2b Figure 2. (a) Gc trace snowing adamantanes isolated from 175" - 375°C heavy crude HDC recycle by processing over'Pt/Beta. (b) Adamantanes recovered from a deep gas condensate. Asterisks in 2a indicate peaks which match those in 2b and which have MMA molecular weights by gc-mass spec. 1017 EOL~T&CATALYZED CONFORMATIONAL ISOMERIZATION OF Cis- DECAHYDRONAPHTHALENE. REACTION PATHWAYS AND ~ T I C S . Wei-Chuan Lai and Chunshan Song* Fuel Science Program, 209 Academic Projects Building, The Pennsylvania State University, University Park, Pennsylvania 16802, USA. Keywords: Zeolites, isomerization, decahydronaphthalene, kinetics INTRODUCTION It was shown in literame that decalin (DHN) may be one of the potential endothermic jet fuels that can serve as the primary heat sink to cool the hot surfaces and system components [Donath and Hess, 1960; Lander and Nixon, 1987; Taylor and Rubey, 19871. Commercial decalin solvents from industrial hydrogenation processes consist of almost equimolar mixtures of cis- and trans- DHN. Although the physical properties of these two isomers are similar, their chemical properties are different. One example is their difference in thermal stability at high temperatures. We have previously shown that as jet fuel components, trans-DHN is superior to cis-DHN for high temperature application because the former is much more stable at high temperatures [Song et al., 19921. The excellent thermal stability at high temperatures is desirable for future high Mach aircraft. Besides, trans-DHN has the desirable ability of inhibiting the solid deposit formation from jet fuels and their components at high temperatures [Song et al., 1994al. For example, adding 50 vol% trans-DHN to a JP-8P fuel, n-tetradecane, and n-butylbenzene thermally stressed at 723 K for 4 h significantly reduced the deposit formation from 3.1 to 0.1 wt%, from 3.0 to 0.1 wt%, and from 5.6 to 0.0 wt%, respectively. Although cis-DHN also has some potential industrial applications, it is desirable to convert cis-DHN to trans-DHN for fuel stability consideration at high temperatures. There has been much research on the catalytic hydrocracking or dehydrogenation of decalin under high pressures and at temperatures generally in excess of 673 K [Ritchie and Nixon, 1967; Shabtai et al., 1979; Constant et al., 1986; Mostad et al., 1990a,b; Nimz, 1990; Sousa-Aguiaret al., 19941. However, relatively little information about the conformational isomerization of cis-DHN into trans- DHN at lower temperatures is available. Petrov et al. (1977) reported the isomerization of cis- and trans-DHN on a nickel catalyst in the temperature range of 393-453 K. They claimed that the isomerization took place only in the presence of hydrogen. Our earlier exploratory work has shown that some mordenites and chemically modified zeolites may promote the isomerization of cis-DHN into trans-DHN at 523 K for 2 h under N2 environment [Song and Moffatt, 1993, 19941. This work extended previous exploratory studies on the catalytic isomerization of cis-DHN to trans- DHN. The objective of this work is to examine the effects of reaction conditions as well as catalyst properties on the catalytic reaction. An overall kinetic model for the catalytic reaction was proposed and empirical equations were presented to predict the selectivity. EXPERIMENTAL The chemicals, cis-DHN, trans-DHN, and DHN (an almost equimolar mixture of cis- and trans- DHN) were obtained from Aldrich Chemical Company and were used as received. Their purities (>99%) were analyzed in our laboratory using gas chromatography (GC) and gas chromatographymass spectrometry (GC-MS). The six catalysts used in the catalytic isomerization reactions include: a hydrogen Y zeolite (HY), a metal ion-exchanged Y zeolite (LaHY), a hydrogen mordenite (HM30A). and three noble metal loaded mordenites (Pt/HM30A, Pd/HM30A, and WHMZOA). The noble metal loaded mordenites were prepared by dispersing the salt of platinum or palladium into the mordenites by incipient wetness impregnation method. The noble metal loading on the support was kept at nominally 6 wt%. The details of the preparation and properties of the catalysts are described elsewhere [Song and Moffatt, 1994; Schmitz et al., 19941. Catalytic isomerization reactions were carried out in 28-mL horizontal type stainless steel tubing bomb reactors, which were charged with 1 g of cis-DHN, trans-DHN, or DHN (7.23 m o l ) and 0.2 g of catalyst, at 473-548 K for 0.15-8 h under an initial pressure of 0.79 MPa UHP N2 or Hz. The reactor was agitated vertically at 240 cycles/min to ensure the uniformity of concentration and temperature inside the reactor. After the reaction, the gas products were collected in a gas bag and were quantitatively analyzed using a Perkin-Elmer Autosystem GC equipped with two detectors, a thermal conductivity detector (TCD) and a flame ionization detector (FID). The liquid products were recovered by washing with acetone and were analyzed on an HP 589011 GC coupled with an HP 5971A Mass Selective Detector (MSD) and quantified by a Perkin-Elmer GC 8500 equipped with an FID. The catalyst was recovered and stored in a vial for thermogravimetric analysis performed later. More analytical details may be found elsewhere [Song et al., 1994bl. RESULTS AND DISCUSSION Calculated composition of equilibria. The equilibrium compositions of trans-DHN and cis- DHN at several temperatures were calculated to establish the theoretical upper limit of the catalytic conversion. The equilibrium constant (K) is related to the Gibbs energy change (AGO) by Q. (1) InK=- - AGO RT Using the data in Reid et al. (1987) for a binary mixture system of cis-DHN and trans-DHN, we have determined the general expression for the equilibrium constant as a function of temperature as shown in Q. (2) 1018 R In K = + 14.83 In T - 0.0365 T + 1.885x10-’ T2 - 4.49~10.T~3 - 85.2 (2) where R is the gas constant (8.3 14 J/mol-K) and T is the temperature in K. The computed heat of reaction, equilibrium constant, and composihon are shown in Table 1. It should be noticed that because of the exponential nature of Equation 1, the calculated results highly depend on the thermodynamic parameters used. For example, if the Gibbs energy was off by just 5-lo%, the estimation error for the equilibrium constant at 473 K may be as large as 16-3570. The calculated mixture compositions will be compared with experimental data following the presentation of experimental results. Effectiveness of zeolitic catalysts. Table 2 shows the products distribution from catalyzed isomerization of cis-Dm using I-g commercial DHN (an almost equimolar mixture of cis- and trans-isomers) as starting reactant at 473 or 523 K under an initial pressure of 0.79 MPa UHP H2 Or N2. Pt- and Pd-loaded mordenites, i.e., F’UHM30A, Pt/HM20A, and Pd/HM30A, are very effective catalysts under H2 atmospherz for the conformational isomerization of cis-DHN to trans- DHN even at low temperature, 473 K. The conversion selectivity towards trans-DHN reached nearly 100%; in other words, there were almost no side-products accompanying the isomerization. Take -30A as an example. The experimental final product composition of 92.3% trans-DHN and 7.3% cis-DHN at 473 K as shown in Table 2 is very close to the calculated equilibrium composition (95.3% trans-DHN, 4.7% cis-DHN), which is shown in Table 1. Although Pt- and Pd-loaded mordenites were effective catalysts under H2 atmosphere, they became less effective under N2 atmosphere (see Table 2). W 3 0 A i s a better catalyst than Pd/HM30A at 473 K under N2 atmosphere in terms of conversion and trans-DHN selectivity although they are almost equally effective under H2. It is interesting to look at the yield change of tetralin, which initially existed as an impurity (0.7 wt%), under different gas environment. Tetralin was completely hydrogenated into decalin under H2 environment because of the hydrogenation ability of Pt and Pd. On the other hand, the noble metals under N2 served to dehydrogenate the decalin to tetralin and thus increased the yield of tetralin. For the other three catalysts studied, HY, LaHY, and HM30A. they are much less effective than hand Pd-loaded mordenites, and do not react at all at 473 K. It is also interesting to note that although the effectiveness of F‘t- and Pd-loaded mordenites depends on the gas environment (H2 or N2). H2 has no impact on the performance of LaHY. The data in Table 3 seems to show that the hydrogen Y zeolitc (HY) performed about as well as the metal ion-exchanged Y zeolite (LaHY), and HM30A is the least effective one among the catalysts studied. Reaction pathways and kinetic data. We further investigated the performance of LaHY and HY intending to get the kinetic data, These isomerization reactions were carried out using 1 g of cis-DHh’ instead of DHN mixture, and 0.2 g of catalyst, at 508-548 K for 0.15-8 h under an initial pressure of 0.79 MPa UHP N2, The experimental results are shown in Table 3. Isomerization is the dominant reaction under the conditions employed. The dominant products are two types of isomers: trans-DHN (from conformational isomerization) with as high as 81% selectivity (defined as the ratio of molar yield of the product to the conversion), and other decalin isomers from ringopening or ring-contraction isomerization. Although cracking products are not shown in Table 3, they are in general small (less than 4% selectivity) except under severe conditions such as at 538- 548 K for 1 h. There was no apparent dehydrogenation reaction from decalin to tetralin observed judging from the gradually decreasing yield of tetralin, which initially existed as an impurity in cis- DHN (0.27 wt%). Figure 1 presents the trans-DHN selectivity vs cis-DHN conversion plots for LaHY and HY catalysts at four different temperatures. There are a few features that may be pointed out from examining Table 3 and Figure 1. First, the more complete data in Table 3 seem to indicate that HY performs slightly better than LaHY in terms of criteria such as activity and selectivity. This observation is somewhat different from what we said earlier that they perform equally well judging from the data in Table 2 where commercial decalin was used as the starting material. Second, selectivity towards ?runs-DHN decreases with increasing temperature. This is not unexpected since the isomerization from cis-DHN to trans-DHN is exothermic, 13212 Jlmol (95 Jig or 3.16 , kcal/mol) at 523 K (our calculation in Table I). Third, the product (trans-DHN) selectivity decreases with increasing conversion level under isothermal condition, and displays a concave downward behavior, which could be empirically fitted by a second degree polynomial as demonstrated later. The trend of selectivity vs conversion in Figure 1 provided useful information about the reaction pathways. It implied that the reactions proceeded through a parallel-consecutive network [Bond, 19871. Based on the previous observations, a simple reaction pathway model for the catalytic reaction of cis-DHN to products was proposed. It should be noted that readers interested in more detailed mechanisms from cis to trans isomers may refer to the review by Weitkmp (1968). The overall reaction is modeled as the parallel-consecutive kinetic scheme shown in Figure 2. The isomerization between cis- and trans-DHN was known to be a reversible process; thus the interconversion between them was also included in Figure 2. However, our experimental data using both cis- and trans-DHN have shown that the forward reaction from cis- to trans-DHN is much faster than the backward reaction, i.e., kl >> kl. In addition, for the reaction condltions studied, the reactions were taken to be approximately first order. With these assumptions, the rate equations may be written as the following: 1019 &% = - (kl + k2) A dt dt dt = kl A - k3 B &= k;? A + k3 B Equations 3-5 may be solved to give A/& = exp [-(kl+k2) t] = exp [- k tl C = A0 - A - B (6) (7) (8) B/Ao = [kl/(kl+k;?-k3)1 texp (-k3 t) - exp (- k 01 Because we are mainly interested in the yield of trans-DHN and only limited data are available in the current work, we did not intend to find all the kinetic parameters. Instead, we used the experimental data to find the lumped rate constant k (equal to kl+k;? in Eq. 6) and then developed empirical equations to predict the product yield of trans-DHN. The procedures are described as follows. First, the rate constant k was determined by using Eq. (6) for all the experiments shown in Table 3. Then, the rate constant (k) was correlated by the Anhenius law as shown in Eq. (9) k = A* e-Ea I RT (9) where A' (h-I) is the frequency (or preexponential) factor, Ea is the apparent activation energy (kcalhol), and R is the gas constant (kcal mol-' K-l). The Ea and A values determined from Arrhenius plots are as follows: for HY Ea = 49.9 kcdmol and A' = 6.03 x 10' h-' (10) for LaHY Ea = 54.6 kcdmol and A* = 3.36 x h-' (11) Third, empirical equations were developed to predict the product yield of rrans-DHk Based on the results in Figure I, we proposed that the selectivity of trans-DHN could be presented by a second degree polynomial as shown in Eq. (12) Selectivity of tr~ns-DHN= a1 + a;?X + a3 X2 + a4 T + a5 X T (12) where X is the cis-DHN conversion, T is the temperature in K, and ai (i=l, ..., 5) are the empirical parameters to be found. Using the data in Table 3, we have determined the parameters ai as follows: for HY Sel. = 1.938 + 0,779 X - 0.285 X2 - 0.00228 T - 0.0012 X T (13) for LaHY Sel. = 1.534 + 2.745 X - 0.082 X2 - 0.00146 T - 0.0054 X T (14) In order to check the reliability of these equations in predicting the conversion and selectivity, predictions based on equations 6, 10, 11, 13, and 14 are compared with experimental results. Figures 3 and 4, respectively compare the experimental selectivity and yield of trans-DHN to the values predicted from the empirical equations for both catalysts. The line corresponding to exact agreement is drawn as a diagonal. It is clear from Figures 3 and 4 that the predicted values are generally in good agreement with experimental values over a wide range of conversion. SUMMARY This work presented some exploratory studies on the effects of reaction conditions as well as catalyst properties on the catalytic isomerization of cis-DHN to trans-DHN. The theoretical equilibrium compositions of trans-DHN and cis-DHN at several temperatures were calculated and compared with experimental data. The catalytic reactions were studied under N;? or H;? environment using HY, LaHY, HM30A. Pt/HM30A, Pd/HM30A, and PtmM20A. Pt- and Pd-loaded mordenites are very effective catalysts under Hz atmosphere for the conformational isomerization of cis-DHN to trans-DHN at 473 K; however, they are less effective under N;? atmosphere. PUHM30A is a better catalyst than PdMM30A at 473 K under N2 atmosphere in terms of conversion and trans-DHN selectivity although they are almost equally effective under H;?. HY, LaHY, and HM30A are much less effective than Pt- and Pd-loaded mordenites, and their performance was not affected by H;?. Besides, they do not react at all at 473 K. HY performs slightly better than LaHY, and HM30A is the least effective one among the catalysts studied. Selectivity towards fruns-DHN decreases with both increasing temperature and increasing conversion level (under isothermal condition). A simple reaction pathways model containing parallel-consecutive kinetic scheme was proposed. The Ea and A* values for the cis-DHN conversion were determined from Arrhenius plots. Empirical equations capable of predicting product yields were also developed. In short, equations 6, 10, 11, 13, and 14 may be used to predict reaction conversion and major products. ACKNOWLEDGMENTS We are grateful to Prof. H. H. Schobert and Prof. P. B. Weisz for their encouragement and support, and to Ms. Cindy Chan of PSU for carrying out many of the catalytic experiments. Funding was provided by U.S. Department of Energy and U.S. Air Force. We wish to thank Mr. W. E. Harrison III of USAF and Dr. S. Rogers of DOE for their support. 1020 REFERENCES Bond, G. C. Heterogeneous Catalysis: Principles and Applications. Second Edition. 1987, Oxford University Press: Oxford. OD 52-53. Constant, W.*D.; Price, G. L..McLaughlin, E. Fuel 1986.65, 8-16. Donath, E. E.; Hess, M. Chemical Engineering Progress. April 1960,56 (4), 68-71. Lander. H. R.: Nixon. A. C. Preor.-Am. Chem. Soc.. Div. Pet. Chem. 1987.32 (2). 504-5 1 1. Mostad, H. B.: Gis, T. U.; Ellestad, 0. H. Applied catalysis 1990a, 58, 105:117.. Mostad, H. B.; Riis, T. U.; Ellestad, 0. H. Applied Catalysis 1990b, 63, 345-364. Nim, M. Zeolites 1990.10. 297-300. PetrOV, L.; Angelova, L.; Shopov, D. Bokl. Bole. Akad. Nauk 1977,30 (I), 85-88. Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The Properties of Gases & Liquids. Fourth Edition. Ritchie, A. W.; Nixon, A. C. Prep.-Am. Chem. Soc.. Div. Pet. Chem. 1967, 12 (3). 117. Rober,ts, R. M.; Madison, J. J. J. Am. Chem. SOC. 1959.81, 5839-5839. Schmitz, A. D.; Bowers, G.; Song, C. In Advanced Thermally Stable Jet Fuels. Technical Repon. U S . DeDarIment of Enerav. DE-FG22-92PC92104-TPR-9O,c tober 1994, H. H. 1987, McGraw-Hill Book Company: New York, N.Y. -. Schobert et al., Eds., pp37-42. Shabtai, J.; Ramakrishnan, R.; Oblad, A. G. In Thermal Hydrocarbon Chemistry; Obald, A. G.; Davis. H. G.: Eddineer. R. T.. Eds.: Advances in Chemism Series 183: American Chemical Society Washingtln, D. C.,.19791 pp 297-328. 1655- 1663. Song, C.; Lai, W.-C.; Schobert, H. H. Prepr.-Am. Chem. Soc., Div. Fuel Chem. 1992,37 (4). Song, C.; Moffatt, K. Prepr.-Am. Chem. Soc., Div. Pet. Chem. 1993.38 (4), 779-783. Song, C.; Lai, W.-C.; Schobert, H. H. Ind. Eng. Chem. Res. 1994a, 33, 548-551. Song, C.; Lai, W.-C.; Schobert, H. H. Ind. Eng. Chem. Res. 1994b, 33, 534-547. Song, C.; Moffatt, K. Microporous Materials 1994,2, 459-466. Sousa-Aguiar, E. F.; Pinhel da Silva, M.; Murta Valle, M. L.; Forte da Silva, D. Prepr.-Am. Taylor, P. H.; Rubey, W. A. Prep.-Am. Chem. Soc., Div. Pet. Chem. 1987,32 (2). 521-525. Weitkamp, A. W. In Advances in catalysis and Related Subjects. D. D. Eley, H. Pines, and P. B. Chem. Soc., Div. Pet. Chem. 1994,39 (3). 356-359. Weisz, Eds. Academic Press: New York and London. 1968, Volume 18, ppl-110. Table 1. Calculated heat of reaction, equilibrium constant and composition for a binary mixture system of cis-decalin and trans-decalin Temperam Heat of reaction Equilibrium Composition (wt %) 0 AH (J/mol) conStanla cis-DHN rmnrDHN 473 - 13205.8 20.5 4.65% 95.35% 508 -132 IO. 1 16.3 5.79% 94.21% 523 ~ 13212.0 14.9 6.30% 93.70% 538 -13213.7 13.7 6.82%. 93.18% 548 - I32 14.6 12.9 7.17% 92.83% 573 -13215.7 11.4 8.06% 91.94% 623 , -13210.6 9.1 9.87% 90.13% 673 -13192.9 7.6 1 I .69% 88.31% 723 . I3 160.2 6.4 13.48% 86.52% a) Equilibrium constant for the reaction: cis-DHN c1 trans-DHN K= [trans-DHN] [cis-DHN] Table 2. Catalyzed isomerization of cis-decalin (starting reactant is 1-g commercial decalin) under an initial pressure of 0.79 MPa UHP Hz or Nz Catalyst Temp. Time Gas Product (wt % of fed) tnmdcir xc Se1.d Ype (K) fi) tm-DHN cis-DHN Tetralin Othersb ratio (%) n na 48.34 50.62 0.70 0.34 0.41 12.7 0.46 12.8 0.55 11.5 1.88 1.9 1.23 - 1.3 0.30 0.95 14.44 3.5 15.42 3.5 LaHY 523 2.0 H2 65.79 16.73 0.59 16.89 3.9 HM30A 523 2.0 N2 53.02 31.60 0.38 15.00 1.7 a) This row presents the purity of as-received commercial decalin including 0.34% n-decane. b) Unreacted n-decane plus products of ring-contaction and ring-opening reactions. C) Conversion of cis-decalin (weight % of feed). a) Selectivity to rranr-decalin, defined as a fraction of cisdecalin conversion. 43.4 1.00 43.4 1.00 42.7 1.00 17.2 0.82 8.2 0.75 0 - 31.6 0.56 31.8 0.53 33.9 0.52 19.0 0.25 \ f WM30A 473 2.0 H2 92.34 Pd/HM30A 473 2.0 H2 92.31 PUHMZOA 473 8.0 H2 91.50 PmM30A 473 2.0 N2 62.40 pd/HM3OA 473 2.0 N2 54.50 LaHY 473 2.0 N2 48.29 HY 523 2.0 N2 65.92 LaHY 523 2.0 N2 65.15 7.25 0 7.23 0 7.95 0 33.46 2.26 42.43 1.84 ' 50.73 0.68 19.04 0.60 18.82 0.61 1021 Table 3. Catalyzed isomerization of cis-decalin (starting reactant is 1-g cis-decalin) under an initial pressure’of 0.79 MPa UHP N2 Catalvst: 0.2 e of LaHY temperature (K) residence timea (min) reaction timeb (min) product yieldC (wt 96) nanC-DHN cis-DHN rrm-/cis-DHN ratio cis-DHN conversionC rate constant, k (h-l) trm-DHN selectivity temperature (K) residence timea (min) reaction timeb (min) product yieldC (wi %) rrMs-DHN cis-DHN trans-lcis-DHN ratio cis-DHN conversionc . - 508 508 508 508 523 523 523 523 538 538 538 548 548 548 60 120 240 480 30 60 120 120 9 18 60 9 18 60 54 114 234 474 25 55 115 115 4 13 55 5 14 56 7.4 11.5 22.7 43.0 20.4 29.0 47.7 46.3 11.7 32.7 50.4 24.2 37.9 42.5 90.5 85.2 70.8 42.8 72.0 60.1 29.4 32.5 83.5 51.4 10.2 61.2 31.2 4.7 0.08 0.13 0.32 1.00 0.28 0.48 1.62 1.42 0.14 0.64 4.94 0.40 1.21 9.08 9.1 14.4 28.8 56.8 27.6 39.5 70.2 67.1 16.1 48.2 89.4 38.4 68.4 94.9 0.11 0.08 0.09 0.11 0.78 0.55 0.63 0.58 2.62 3.02 2.45 5.83 4.92 3.19 0.80 0.79 0.79 0.76 0.73 0.73 0.68 0.69 0.72 0.68 0.56 0.63 0.55 0.45 Catalyst: 0.2 g of HY 508 508 508 523 523 523 538 538 538 538 548 548 548 120 240 480 30 60 120 9 I8 30 60 9 I8 30 I14 234 474 25 55 115 4 13 25 55 5 14 26 21.1 47.6 57.0 23.1 46.1 52.7 20.7 34.6 51.9 54.5 35.8 47.8 50.3 73.9 37.8 23.0 69.3 36.8 25.6 71.6 51.1 19.4 7.4 45.9 23.1 10.2 0.28 1.26 2.48 0.33 1.25 2.06 0.29 0.68 2.67 7.35 0.78 2.07 4.93 25.8 61.8 76.6 30.3 62.9 74.0 28.0 48.5 80.3 92.2 53.7 76.5 89.4 rate constant. k (h-l) 0.16 0.25 0.18 0.87 1.08 0.70 4.91 3.06 3.89 2.78 9.16 6.19 5.17 rrm-DHN selectivity 0.81 0.77 0.74 0.76 0.73 0.71 0.73 0.71 0.64 0.59 0.67 0.62 0.56 a) b) c) Reactor residence time in sand bath preheated to reaction temperature. Corrected reaction time (reactor residence time minus heat-up time). Based on the initial amount of cis-DHN. 0.7 - 0.5 1 + 0 508K(LaHY) e 508K(HY) A 523K(LaHY) A 523 K (HY) 0. 5 38K(LaHY) 538 K (HY) -t 548 K (LaHY) ” 548K(HY) 0.4 0.0 0.2 0.4 0.6 0.8 1.0 cis-Decalin conversion Figure 1. trans-Decalin selectivity vs cis-decalin conversion plots for LaHY and HY catalysts at four different temperatures. 1022 I I I / Figure 2. Proposed overall reaction pathways for the catalytic reaction of cis-decalin. 0.9 - 0.8 - 0.4 0.5 0.6 0.7 0.8 0.9 trans-DHN selectivity (Experimental) Figure 3. Predicted versus measured values of trans-decalin selectivity for the catalytic isomerization of cis-decalin using HY and LaHY at 508-548 K for 9-480 min. . 9* a 2 b .Ih 0.6 - 0.5 - 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 trans-DHN yield (Experimental) Figure 4. Predicted versus measured values of trans-decalin yield for the catalytic isomerization of cis-decalin using HY and LaHY at 508-548 K for 9-480 min. 1023 SOXAL" PILOT PLANT DEMONSTRATION AT NIAGARA MOHAWK'S DUNKIRK STATION Peter K Strangway Research and Development Deparhnent Niagara Mohawk Power Corporation Syracuse, New York 13202 Keywords: SOXAL" Process, SO, Emissions Control, Regenerable FGD System INTRODUCTION ' The Clean Air Act Amendments of 1990 made it necessary to accelerate the development of scrubber systems for use by some utilities burning sulfur-containing fuels, primarily coal. While many types of Flue Gas Desulfunzatlon (FGD) systems operate based on lime and limestone scrubbing, these systems have drawbacks when considered for incorporation into long-term emissions control plans. Although the costs associated with disposal of large amounts of scrubber sludge may be manageable today, the trend is toward increased disposal costs. Many new S0,conml technologies are being pursued in the hope of developing an economical regenerable FGD system that recovers the SO, as a saleable commercial product, thus d m i z i n g the formation of disposal waste, Some new technologies include the use of exotic chemical absorbents which are alien to the utility industry and utilities' waste treatment facilities. These systems present utilities with new environmental issues. The SOXALTMp rocess has been developed so as to eliminate such issues. The objective of the nominal 3 Mw SOXAL pilot plant at Niagara Mohawk's Dunkiuk Power Station was to demonstrate the technical and economic feasibility of this regenerative FGD process to remove SO, from the flue gas of a coal-fired boiler. The key demonstration component was the integration of a bipolar membrane system with proven sodium scrubbing and steam shipping technologies. Previously, bipolar membrane systems had been commercially proven in applications unrelated to flue gas desulfurization. Sodium alkali scrubbing of the type used in the SOUL process is an accepted and proven method for removing SO, from gaseous streams. It is the system of choice in many industrial applications due to its lower capital requirements, higher S0,removal efficiencies, and low maintenance costs. A large number of sodium scrubbers have been operated successfully at industrial and utility sites. The main drawback of such systems is the higher cost of the sodium scrubbing solution versus the reagents required for calcium-based systems. The SOXAL process minimizes the cost of sodium scrubbing hy regenerating the scrubbing solution for reuse while simultaneously recovering the sulfur as a saleable product. PROCESS DESCRIPTION The SOXAL FGD process has four major unit operations which are illustrated schematically in Figure I : I . 2. 3. 4. The primaryreactions in the sodium sulfite (NqSO,) scrubber are as follows: S0,removal is accomplished by the reaction ofthe S0,with NqSO, in the scrubbing solution to form sodium bisulfite (NaHSO,). In addition, aportion of the NqSO, in the scrubbing solution is oxidized to sodium sulfate (NqSO, ) by reaction with oxygen (0,)in the flue gas stream. The NqSO, can be recovered in a saleable crystalline form. Regenerationo f the spent scrubbing solution is achieved in the primary bipolar regeneration unit which is shown schematically in Figure 2. Each cell has a bipolar membrane and a cation selective membrane. The bipolar membranes separate the water molecules into hydrogen (H') and hydroxyl (OH) ions, and NaHSO, is converted to NqSO, h addition, sodium ions ma") migrate across the cation selective membrane into the base compartment. These beeome associated with OH-ions and form NaOH. Most ofthis NaOH reacts with NaHSO, to form NqSO, for recycle to the scrubber. The HSO; anions that remain in the acid compartment associate with H+ ions from the . . The prescrubber removes chlorides, fluorides and residual particulates by water scrubbing. The sodium sulfite scrubber removes the S0,from the flue gas. The bipolar membrane cell stack regenerates the spent sodium bisulfite solution. The steam stripper removes the SO, from the sulfurous acid. bipolar membrane to form sulfurous acid (H2S03). The pdally sahkited H$03 stream is continuously withdrawn from the cell stack and is subsequently decomposed into S0,gas and water molecules in the steam stripper. PILOT PLANT FACILITIES The 3 MW SOUL pilot plant demonstration f d t y was installed at Niagara Mohawk's Dunkirk Power Station on Lake Erie near Buffalo, NY. This station has two (2) 100 MW and two (2) 200 MW tangential coal-fued boilers. The slip-stream of the flue gas for the 3 MW pilot plant was extracted after the induced draft fan of Unit No. 4. During the demomation, this boiler was fd with bituminous coal from Pennsylvania and West Virginia which had an average sulfur cantent of 2. I %, ash content of 7. I%, and heating value of 13, IO0 Btu per pound. The 3 MW SOXAL pilot plant was located adjacent to Unit No. 4 to minimize the amount of ductwork required to provide the slip-stream of flue gas to the scrubber and to return the processed flue gas back to the station stack. In addition, most of the utilities required for the pilot plant were available with minimal interconnect distances between the station and the pilot plant. Both the prescmbber and scrubber were designed and supplied by Advanced Air Technology. The water-based presaubber measured 4.5 feet in diameter by 25 feet in height. It had a Hastelloy quench section, an FRP shell, 1024 \ I a six-fwt bed of polypropolyene packing, and an FRP mist eliminator. The sodium-sulfite-based scrubber measured 4.5 feet in diameter by 40 feet in height. It had two, six-foot high polypropylene-packed stages. %ahon of these units was easy and reliable. No fouling was observed. Considerable particulate matter was removed by the prescrubber, and there was no significant carryover from the prescrubber to the scrubber. The bipolar membrane cell stack used during the demonstration had 44 two-membrane cells. Each cell included a single bipolar membrane and a cation membrane. Only 176 square feet of cell area was required at the pilot plant, and standard commercial bipolar membranes were used. A DC rectifier provided the energy required to regenerate the absorbent solution. The initial cell stack was operated for over three months with no hardware problems. While some individual membranes occasionally had to be replaced, the overall performance of the stack was well within expectations. At no time was testing delayed due to membrane failures. The pilot plant steam stripper column measured 16 inches in diameter by 2 1.5 feet in height. It had a stainless steel shell and contained a six-foot bed or random Kynar packing. A small bed of this same packing material was used as the mist eliminator. Early in the test program, instrumentation failures gave the erroneous impression of low stripping eiliciency. However, once these problems were identified and corrected, the steam stripper column operated as designed. Although some problems were encountered relative to pumps, flow meters, weld leaks, etc., most of these were corrected prior to the main demonstration program. As a result of failure of the continuous SO, gas analyzer to operate properly, it was necessary to use an outside testing service during the last four months of the test period in order to obtain accurate and continuous SO, measurements. In general, most of the instrumentation installed was reliable and pedormed up to expectations after initial start-up. The operators found the pilot plant facility to be easy to operate with minimal staffmg. Two engineers and four operators manned all shifts, including the 7 days per week, 24 hours per day periods of continuous testing. TEST DESCRIPTION The 3 MW S O W pilot plant test program took place over a seven-month period and included both continuous operation and paramelic tests. Failure of the continuous SO, analym severely limited the amount of quantitative SO, data collected during the first two months of testing. During this time, it was possible to demonstrate continuous operation of the bipolar membrane cell stack in integrated operation with the scrubber and stream stripper systems. The overall process was kept in balance while producing regenerated scrubbing solution and concentrated SO,. Immediately after a previously scheduled one-month boiler outage, parametric testing was initiated in accordance with the following test plan: m si!&a!G * Initial Baseline Studies' Establish Baseline - .Absorber Parametric Studies - Lower Stage pH - Recycle Rate -Number of Beds ~ SO, Concentration - Base pH - Recycle Rate - Cell Stack Current - Conversion Rate - Cell Stack Temperature * Cell Stack Parametric Studies Maximize Absorption Maximize Absorption Minimize Oxidization Minimize Oxidizationhlaximize Absorption Reduce Flush Cycle Reduce Cost and Flush Cycle Reduce Power Consumption Reduce Power Consumption Reduce Power Consumption - Stripper Temperature Studies Optimize Efficiency - Overall Optimized Operation Maximize AbsorptiodSO, Removal During the four months of parametric testing, the Dunkirk boiler was operated at a reduced load overnight and was shut down on weekends due to a lack of power demand. As a result, parametric testing was carried out on a "decoupled" basis, live days per week. In other words, when studies were conducted on the absorber, the cell stack unit was shut down, and vice versa. The pilot plant was operated from full storage tanks of either spent or regenerated absorbent. The portion of the process not undergoing testlng at any given time was operated overnight to replenish the spent or regenerated absorbent inventory for the next day's testing. The parametric studies were not felt to have been significantly affected by the unanticipated boiler cycling and shutdowns. TESTRESULTS The data collected during the demonstration period is summarized in Table 1. During the fust two months of testing, the pilot plant was operated continuously. During the last four months of parametric testing after the boiler outage, the pilot plant consistently demonstrated over 98 percent SO, absorption as is shown in Figure 3. During this same period the SO, concentration in the flue gas ranged between 1000 and 1500 ppm as is shown in Figure 4. It appears that the same high level of SO, absorption was probably achieved during the initial twomonth continuous run when the SO, analyzer was not operational. The test results also show that when higher inlet SO, levels were obtained by recycling some of the recovered SO, to the scrubber, SO, absorption was 1025 enhanced and oxidation of the snubber solution was reduced Oxidation of the scrubber solution is an important parameter in the economics of the SOXAL process. Test results showed that total oxidation during SO, absorption was well within the design range even without the use of additives or any other attempt to minimize oxidation. The major parameters associated with the operation of tbe bipolar membrane cell stack are its power consumption and durability. During the demonstration period, power consumption by the cell stack was consistently in the range of 1100 to 1300 kWton of SO, removed as is shown in Figure 5. This is consistent with anticipated power consumption and the value that was used in EPRI’s 1990 economic evaluation of the process. During over 2,500 hours of pilot plant operation, the bipolar membranes proved to be extremely durable. An acid wash process was used to minimize fouling of the membranes during the demonstration period, and the optimum operating conditions ncedcd to minimize membrane washing were determined. CONCLUSIONS 1. 2. 3. 4. 5. ACKNOWLEDGMENTS This project was sponsored by the U.S. Department of Energy’s Pittsburgh Energy Technology Center based on Conbxt No. DE-AC22-91PC91347. AlliedSignal’s Aquatech Systems Division was the prime contractor with responsibility for design, fabrication, and operation of the pilot plant. Additional funding and the host site were provided by Niagara Mohawk Power Corporation, with co-funding from the Empire State Electric Energy Research Corporation and the New York State Energy Research and Development Authority, Continuous integrated operation of the absorption and regeneration portions of the pilot plant was demonstrated. The ease of independently operating these porhons of the SOXAL system was also demonstrated. Over 98 percent SO, removal was consistently achieved. Stable bipolar membrane performance was proven. Cell stack power consumption and scrubber oxidation were consistent with plant design expectations. Table 1 Summary of Test Data’ SO, FlueGa SO, Power Test Concen- Flow Rate Absorption Consumed GEk lr81ion (%) , 1026 I Table 1 (Contmued) Notes 1. 2. 3. 4. Each data point typically represents lhe average of four mensurements taken during an eight-hour test. Simultaneous indicates continuous operation of both absorption and regeneration processes. All other tests were conducted in a "decoupled" mode. These flow rates we in ACFM (actual cubic feet per minute). All others are in DSCFM (standard cubic feet per minute - dry basis). Baseline regeneration tests were performd with 400 amps of operating cwent. Since.the membranes had four square feet of cross-sectional area, this is equivalent to 100 ASF (amps per square foot). Figure I SOXAL Pmess Flow Sheet Figure 2 Schematic of Bipolar Membrane Regeneration Unit 1027 100 98 2 96 92 : 94 2 00 g 90 N 06 04 02 80 I I . I Test Figure 3 SO, Absorption Efficiency I 1400 I 600 mm Test I Figure 5 Cell Stack Power Consumption i 1028 MILLIKEN CLEAN COAL PROJECT-UPDATE G. S. Janik, S. C. Chang, and P. A. Szalach New York State Electric (L Gas Corporation Corporate Drive - Kirkwood Industrial Park Binghamton, NY 13902-5224 J. B. Mereb, J. A. Withum, and M. R. Stouffer CONSOL Inc. Research and Development 4000 Brownsville Road Library, PA 15129 Keywords: Milliken, Clean Coal Project, SO, and NO, Control INTRODUCTION The Milliken Clean Coal Demonstration Project was selected for funding in Round 4 of the U.S. DOE’S Clean Coal Demonstration Program. The project’s sponsor is New York State Electric and Gas Corporation (NYSEG). Project team members include CONSOL Inc., Saarberg-Holter-Umwelttechnik (S-H-U), NALCO/FuelTech, Stebbins Engineering and Manufacturing Co., DHR Technologies, and ABB/CE Air Preheater. The project will provide full-scale demonstration of a combination of innovative emission-reducing technologies and plant upgrades for the control of sulfur dioxide (SO ) and nitrogen oxides (NO,) emissions from a coal-fired steam generator w h o u t a significant loss of station efficiency. The demonstration project is being conducted at NYSEG’s Milliken Station, located in Lansing, New York. Milliken Station has two Combustion Engineering 150 MWe pulverized coal-fired units built in the 1950s. The S-H-U FGD process and the LNCFS-Level I11 low-NO, burner are being installed on both units. I. S-H-U Process A. Background The Saarberg-Holter Umwelttechni k GmbH (S-H-U) flue gas desulfurization (FGD) process commenced operation at the NYSEG Milliken Station in mid-January 1995; Unit 1 operation is scheduled to begin in late June. The S-H-U SO control technology is based on a forced oxidation, formic acid-enhanced wet timestone scrubber. Project goals include: Demonstration of up to 98 percent SO, removal efficiency while burning high-sulfur coal; Production of commercial grade gypsum and calcium chloride by-products to minimize waste disposal; 0 0 0 Zero wastewater discharge; 0 Space-saving design; 0 A low-power-consumption scrubber system. Parametric testing of the S-H-U process is scheduled to begin September 1, 1995. The test program will provide operation and performance data to confirm that the S-H-U FGO process can meet regulatory requirements for new and existing utility boilers. The data also will provide a basis for process optimization and for economic evaluation. The physical and chemical data required for by-product sales or disposal of gypsum, FGD blowdown sludge, and calcium chloride will be developed. B. Description of the S-H-U Contactor As shown in Figure 1, the absorber has a cocurrent section followed by a countercurrent section. There are four slurry spray headers on the cocurrent side and three on the countercurrent side. The two-stage design helps maintain the slurry pH in the optimum range. Also, cocurrent operation reduces pressure drop. The two-stage absorber is designed to be compact, allowing easier retrofit. The absorber is constructed of concrete and is lined with corrosionand abrasion-resistant ceramic tiles. This design is expected to reduce maintenance. I d ? The FGD system is designed for zero waste water discharge. A blowdown stream is removed and treated to control the scrubber chloride concentration and produces a saleable concentrated calcium chloride solution. 1029 C. Start-up Results The scrubber is operating using four or five spray headers which provides an L/G .of 119 to 157 gal/kacfm. The dewatering system produces gypsum containing less than 10% moisture by weight. TO achieve the design slurry chloride concentration, the brine concentrator system started up until June 1995. The following table shows preliminary SO, removal results, 0. Parametric Test Plan To define the performance limits of the S-H-U FGD system, Unit 1 will operate at design conditions, provide long-term data, and evaluate the FGD load-following capability. The steady-state chloride level is expected to be about 40,000 ppm C1 by ut. For each test, the scrubber pressure drop and SO, removal will be measured. The effect of process variables on gypsum crystal morphology will be studied during tests using the design sulfur coal. The project will use coals which contain 1.6, 3.2 (design coal), and 4 weight percent sulfur. The following is a discussion of the parameters to be varied. The plant design is based on a coal sulfur content of 3.2 weight percent. The coal sulfur content will be varied over a range of 1.6 to 4.0 weight percent using at least three different coals. The purpose i s to demonstrate the S-H-U technology with low-sulfur coal, design coal, and high-sulfur coal. Parametric tests will not be performed using the high-sulfur coal; instead, the process will be operated at optimum conditions based on the results of parametric tests using the design coal and computer modeling results. The design scrubber slurry formic acid concentration is 800 ppm. Formic acid concentrations of 0, 400, 800, and 1600 ppm will be tested. The purpose is to demonstrate the effect of formic acid concentration on SO, removal and scrubber operability. Various combinations of spray headers in the cocurrent and countercurrent sections will be tested. The purpose is to generate data to optimize SO, removal performance and scrubber energy consumption. The mass transfer coefficients will be determined individually for the cocurrent and countercurrent sections using the results from these tests. By changing the number of spray headers in operation at constant flue gas flow, the scrubber L/G ratio will be varied. The design gas velocity is 20 ft/sec in the cocurrent scrubber section and 12 ft/sec in the countercurrent section. Tests at higher velocity (15 to 20 ft/sec in the countercurrent section) will be performed on the Unit 2 scrubber by shunting gas flow from Unit 1 to the Unit 2 scrubber. The purpose is to provide data on high gas velocity scrubbers. Recent literature (e.g., Ref. 2) suggests that FGD capital cost can be reduced significantly by increasing the design velocity in the absorber. These tests will be performed using the design formic acid concentration (800 ppm). The design limestone grind is 90% -170 mesh when using formic acid and 9o"x -325 mesh with no formic acid. For comparison purposes, tests will be performed using 90% -170 mesh without formic acid and using 90% -325 mesh at 800 ppm formic acid concentration in the scrubber. The test coal sequence is low-sulfur coal (1.6%) followed by the design coal (3.2%), and lastly the high-sulfur coal (4%). The test plan includes 103 six-hour tests using low-sulfur coal, 61 seven-day tests using design sulfur coal, and one two-month test using high-sulfur coal. The tests are statistically designed to study parametrically the effect of formic acid concentration, L/G ratio, and mass transfer on scrubber performance. 11. Low-NO, Concentric Firing System-Level 111 (LNCFS-111) Limestone utilization will be held constant at the design level. A. Background Both Milliken units were retrofitted with the LNCFS-111 burners. The objective was to reduce NO,, emissions to comply with the 1990 Clean Air Act Amendments 1030 A 1 i (cm), while continuing t o produce marketable fly ash. The Unit 1 burner r e t r o f i t was in 1993 and the Unit 2 r e t r o f i t in 1994. New coal m i l l s were installed during the burner outage. The effectiveness o f LNCFS-I11 burner r e t r o f i t t o reduce NO emissions was evaluated i n short-term tests (2-4 hours each) and long-term Zests (60 days) while burning a h i g h - v o l a t i l e eastern bituminous coal. The short-term tests were s t a t i s t i c a l l y designed t o evaluate the impact of burner operating parameters on NO, emissions and loss-on-ignition (LOI). The long-term t e s t consisting o f 60 measurement days was used t o estimate the annual NO emissions and was consistent with the U t i l i t y A i r Regulatory Group (UARG) recommhdations. The baseline tests were conducted on Unit 2 and the p o s t - r e t r o f i t tests were conducted on Unit 1, since Unit 1 was not available f o r baseline t e s t i n g p r i o r t o i t s r e t r o f i t . Conducting baseline testing on one u n i t and p o s t - r e t r o f i t t e s t i n g on the other u n i t was an acceptable option because p r e - r e t r o f i t NO emissions from the two units d i f f e r e d by less than 0.03 lb/MM Btu. Long-term h emissions from the two M i l l i k e n u n i t s were 0.64-0.68 lb/MM Btu at 3.5%-4.5% O2 a t the economizer outlet. E. Parametric Test Program Results The short-term parametric tests evaluated the impact o f b o i l e r load, excess O,, and burner tilt on NO emissions and LOI. P o s t - r e t r o f i t t e s t i n g included as additional parameters ;ill c l a s s i f i e r speed, SOFA tilt, and SOFA yaw. Variation o f CO was not a consideration in t h i s study because CO measurements were less than 13 ppm f o r the baseline tests and less than 23 ppm f o r the p o s t - r e t r o f i t tests. Figure 2 shows f u l l b o i l e r load (140-150 MWe) variations of NO emissions and LO1 with economizer 0, f o r the baseline and the p o s t - r e t r o f i t tests. Only postr e t r o f i t tests i n which over-fire a i r (SOFA and CCOFA) flows and m i l l c l a s s i f i e r speeds did not vary were included i n Figure 2. A t the same 0 level, the scatter o f the data was p a r t l y due to experimental variation and l o the variation o f other parameters, such as burner tilt. Under both baseline and p o s t - r e t r o f i t conditions, higher 0, levels increased NO, emissions and reduced LOI. A simple inverse relationship was observed between baseline NO emissions and LOI. The p o s t - r e t r o f i t relationship between NO emissions and LO! was more complex because o f the larger number o f the LNCFS-111 parameters. The LNCFS-I11 configuration t y p i c a l l y had 0.17-0.19 lb/MM Btu lower NO, emissions and 2.4%-2.9% (absolute) higher LO1 r e l a t i v e to the baseline. In general, NO, reductions were about 35% and p o s t - r e t r o f i t LO1 levels were about 4%. The e f f e c t o f m i l l c l a s s i f i e r setting on NO emissions and LO1 a t 120 MWe f o r d i f f e r e n t economizer 0 levels (3.05, 3.4%: and 4.5% nominal) are shown i n Figure 3. Increasing tbe c l a s s i f i e r speed corresponds t o f i n e r pulverized coal (increasing c l a s s i f i e r speed 40 rpm i s estimated t o increase coal fineness from 75% t o 90% through 200 mesh) which dramatically reduced LOI. Furthermore, NO emissions could be reduced by as much as 0.05 lb/MM Btu by increasing th; c l a s s i f i e r speed 40 rpm. Similar trends were observed a t f u l l b o i l e r loads. Baseline changes i n burner tilt had a s i g n i f i c a n t e f f e c t on NO emissions and a minor effect on LOI, whereas, p o s t - r e t r o f i t changes i n burner tilt had s i g n i f i c a n t e f f e c t s on both NO emissions and LOI. Increasing the LNCFS-I11 burner tilt below the horizontar (negative t i l t ) was e f f e c t i v e i n reducing both NO emissions and LOI, but was l i m i t e d by i t s impact on the main steam teiperature. Following the burner r e t r o f i t , a control algorithm provided automatic v a r i a t i o n i n burner tilt t o maintain the main steam temperature. Changes in SOFA tilt had minor e f f e c t s on NO, emissions, LOI, and steam temperatures. Furthermore, changes i n SOFA yaw had minor e f f e c t s on NO, emissions, but increased LO1 i f the SOFA yaw was d i f f e r e n t from the fuel f i r i n g angle, However, SOFA yaw changes were accompanied by automatic changes i n burner tilt to maintain steam temperatures, and the effects o f the two parameters on LO1 could not be isolated. C. Long-Term Test Results Long-term measurements (60 days) were used t o estimate the achievable annual NO emissions, and t o evaluate the effectiveness o f the LNCFS-I11 burner r e t r o f i t : Figure 4 compares long-term NO emissions from the two M i l l i k e n units (baseline and LNCFS-111) at f u l l b o i l e r joad (145-150 MWe). A t 3.3%-3.6% economizer 0 , NO emissions dropped from baseline levels o f 0.64 lb/MFI Btu t o p o s t - r e t r o f i t l e t e l s of 0.39 lb/MM Btu, corresponding t o a reduction o f about 3996. At a b o i l e r i' Y 1031 load of 80-90 MWe and a t 4.5%-5.0% economizer 0 , NO,, emissions dropped from baseline levels o f 0.57 lb/MM Btu t o p o s t - r e t r o f i t levels o f 0.41 lb/MM Btu, corresponding to a reduction o f about 28%. In summary, NYSEG believes LNCFS-111 burner r e t r o f i t i s a cost-effective technology t o comply with T i t l e I V o f the 1990 CA4A. NO emissions below 0.4 lb/MM Btu could be achieved, while maintaining salable fly'ash. To date, burner operations are acceptable. REFERENCES 1. Glamser, J.; Elkmeir. M.; Petzel, H-K. "Advance! Concepts In FGD Technology: The S-H-U Process With Cool ing Tower Discharge, Journal o f the A i r Po l l u t i o n Control Association, Vol. 39, No. 9, September 1989. 2. Carey, T.R., Skarupa, R.C., Hargrove, O.W., and Moser, R.E. "EPRI ECTC Tes! Results: Effect o f High Velocity on Wet Limestone Scrubber Performance, Presented at the 1995 SO, Control Symposium, Miami, FL, March 28-31, 1995. k Figure 1. (One of Two Absorbers Shown) SCHEMATIC OF S-H-U FGD SYSTEM AT THE NYSEG MILLIKEN STATION Flue Gas To Slack Slurry From Recirculslion Flue Gar From Boiler- Slurry From Recirculation Pumps 1032 I 1 I I I ' I 1033 SCR AND HYBRID SYSTEMS FORUTILITY BOILERS: A REVIEW OF CURRENT Kent D. Zammit Electric Power Research Institute 3412 Hillview Avenue, P.O. Box 10412 Palo Alto, California 94303 EPRI-SPONSORED RESULTS Keywords: Selective Catalytic Reduction (SCR), NOx Reduction, Hybrid Systems Selective Catalytic Reduction (SCR) has been widely demonstrated in Europe and Japan as a postcombustion NOx control technology. However, most of this experience has been gained using relatively low-sulfur fuels, typically less than 1.5 percent. By comparison, the application of SCR in the United States has been much more limited, and to date the experience base is virtually non-existent for coal- and oil-fired boilers. Higher fuel sulfur content corresponds to higher concentrations of SO2 and SO3 which can lead to potential poisoning and more rapid deactivation of the catalyst. In addition, SCR catalysts have the potential to oxidize SO2 to so3, which can lead to serious problems with ammonium sulfate and/or bisulfate deposition in the air preheater, marketability of fly ash, and potential increases in plume opacity. A number of elements present in fly ash, such as arsenic and alkaline metals, may poison the active sites of an SCR catalyst. Regulatory forces stemming from the 1990 Clean Air Act Amendments have the potential to require the use of SCR in the US. for both new and existing units. In response to uncertainties in the cost and feasibility of SCR for the U.S. utility industry, EPRI has sponsored a multi-pilot plant test program to evaluate the feasibility and cost of SCR as a function of fuel type and SCR/host boiler configuration. This paper discusses three of those pilots: the high sulfur/high dust unit at the National Center for Emissions Research, TVA; the post-FGD unit at EPRI's Environmental Control Technology Center, NYSEG; and the residual oil-fired unit at the NiMo Oswego Steam Station. Each pilot represents a 1 MW(e) equivalent SCR reactor divided into two parallel sections to allow for simultaneous testing of two catalyst types. Operating conditions for each pilot are listed in Table 1. The pilot SCR catalysts were designed to maintain certain performance criteria over guaranteed and design lifetimes of 2 and 4 years, respectively. Performance goals call for 80% NOx conversion with residual ammonia (slip) levels of less than 5 and 2 ppm at the exits of the second-to-bottom and bottom catalyst layers, respectively. At TVA (5 catalyst layers), the 5 and 2 ppm slip limits apply to the outlets of the fourth and fifth catalyst beds, and at NYSEG and NiMo (3 layers), the limits apply to the outlets of the second and third catalyst beds. TVA HOT SIDE, HIGH SULFUIUHIGH DUST SCR PILOT The TVA pilot was eventually equipped with features to counter the effects of fly ash on the SCR catalysts, including relatively large cell openings, a non-catalytic "dummy" catalyst layer, screens above each catalyst layer, streamlined reactor and sootblowers above the first and fourth layers. The TVA pilot was operated between May 1990 and May 1994 for a total of approximately 22,000 hours. Catalyst Activity. Both initial TVA test catalysts exhibited significant deactivation which was exacerbated by frequent boiler/pilot shutdowns early in the test program. Analysis of ash samples from the reactor identified a mechanism in which the ash deposits become enriched with sulfur via interaction with ambient moisture during shutdowns. As the moist acidic deposits reacted with alkaline ash constituents, hard deposits were formed that permanently plugged a number of catalyst channels. This mechanism may have also occurred on a smaller scale on the catalyst surface and within catalyst pores, and contributed to formation of a masking layer and consequential loss of catalyst activity. The original V/Ti catalyst was tested for the entire pilot operating duration. Results of catalyst sample activity measurements by the manufacturer are shown in Figure 1. The measurements were made on small sections of the catalyst sample that were free of plugged channels; therefore, results were directly comparable with data shown in the 1034 figure from selected European experience. In all cases, the samples from bed 1 exhibited a higher activity than those from bed 3, which may indicate the positive influence of sootblowers located above the first catalyst layer, but not above the third layer. Figure 1 also shows the activity curve for replacement V/Ti test elements installed in the center of catalyst beds 1 and 3 for approximately 8,000 hours exposure. The replacement elements featured different hardness values than the original catalyst charge. Although the replacement elements exhibited a lesser rate of deactivation, the positive effects of altering catalyst hardness are not entirely conclusive because a lower sulfur coal was being fired while the replacement elements were in place. The original zeolite catalyst failed to meets its performance criteria after 5,000 hours of operation and was replaced after approximately 12,000 hours with a reformulated zeolite design from the same vendor. The reformulated catalyst showed improvement in its baseline activity and in the rate of deactivation compared to the origmal catalyst. Bulk and surface chemical measurements were also made by both catalyst vendors to monitor changes in the composition of the catalyst and the accumulation of potential catalyst poisons. Bulk analysis results indicate increases in the concentrations of arsenic, sodium, and potassium with increasing exposure time. Catalyst Plugging Countermeasures. The TVA pilot represented a severe environment with respect to potential catalyst plugging due to the high ash loading and the alkalinity of the ash. Periodic reactor inspections revealed considerable buildup of solids on the outer catalyst blocks, which resulted from "wall effects" in the relatively small reactor. Over the course of the test program, the flue gas pressure loss across the V/Ti catalyst increased from below 4 inches to over 10 inches of water. Manual counting of the plugged channels showed that nearly 55% of all V/Ti catalyst channels had become permanently plugged when testing ended. A number of strategies were implemented or considered during the test program to limit increases in catalyst channel plugging. These include screens, sootblowers, vacuuming, and moisture avoidance. All strategies employed at the pilot were successful to a certain degree, but their need and practicality for full-scale SCR application will vary. Other Operating Issues. Several operational issues were encountered during the TVA test program that provided pilot experience with full-scale SCR design implications. These include: ammonia injection nozzle pluggage in the high sulfur/high dust environment, artifact reactions over sampling probe materials during NOx and ammonia sampling, process control issues associated with zeolite catalyst ammonia adsorption/desorption times, and CEM system maintenance and sample preconditioning issues specific to high sulfur/high dust SCR systems. NYSEG POST-FGD SCR PILOT The NYSEG SCR Pilot begin testing in December 1991 and is currently operating, with test catalysts exposed to flue gas for approximately 21,000 hours. Key pilot results include catalyst performance in the relatively clean post-FGD environment and cost issues associated with flue gas reheat. A recuperative heat-pipe heat exchanger (HPHE) recovers heat from the gas exiting the reactor and an additional 185°F of reheat input is required to maintain a reactor temperature of 650°F. Heat Exchanger Fouling Effects. Because the cold end of the NYSEG HPHE operates below the maximum condensation temperatures of ammonium sulfate/bisulfate and sulfuric acid, the test program was focused on evaluating exchanger fouling effects (ie., heat transfer loss, increase in gas pressure drop, and corrosion). Figure 2 shows the relative decline in heat transfer and increase in flue gas pressure drop across the return side of the heat exchanger during operating periods with distinct animonia slip levels. In the figure, heat transfer is expressed as a fraction of the design rate to normalize exchanger efficiency for changes in gas flow and reactor temperature. The figure also shows the effects of internal water-washing between operating periods. Water-washg was very effective in dissolving and removing deposits, and 1035 consequently in restoring heat exchanger performance and pressure drop to original conditions. Figure 2 also illustrates the importance of minimizing ammonia slip from SCR catalysts in the post-FGD configuration. At the pilot unit, severe heat exchanger performance . degradation was avoided when the average ammonia slip was held to below 2-3 ppm. An extensive corrosion testing program was undertaken to examine the potential 1 problem of cold-end corrosion of the HPHE tubes. The surface temperature of these tubes is typically 60°F colder than the bulk flue gas temperature. The program included on-lie corrosion monitoring, test samples, and test heat pipes of various metals. The results of this program are beyond the scope of this paper. Catalyst Activity. A post-FGD configuration requires less catalyst and problems associated with high flue gas sulfur and fly ash content are avoided. No screens, "dummy" catalyst layers, or reactor sootblowers are required. The overall catalyst volume is lower than its high dust counterpart because of the higher surface area-tovolume ratio inherent to a smaller pitch catalyst. The favorable reactor environment also lessens the rate of catalyst deactivation, and further reduces catalyst volume requirements to achieve a given catalyst life. The original composite V/Ti catalyst exhibited severe deactivation and was replaced after only 5,300 operating hours. Activity changes occurred exclusively during pilot shutdowns, which suggested that ambient moisture had aided in the mobilization and penetration of catalyst poisons throughout the active catalyst surface layer. Contaminants that penetrated the catalyst included silicon, sodium, potassium, phosphorus, and sulfur, while calcium and iron were concentrated at the surface. The source of the contaminants is the fine ash and FGD carryover solids that are lightly deposited on the catalyst surfaces during operation. The original extruded V/Ti catalyst and the replacement composite V/Ti catalyst showed no measurable activity change in pilot tests over 13,100 and 12,700 hours, respectively. Therefore, given the same performance goals, a post-FGD catalyst would be expected to exhibit a substantially longer catalyst life than its high sulfur, high dust counterpart. A short testing period was dedicated to catalyst evaluation at temperatures below the typical lower limit of 600°F to determine the potential for reducing the operating temperature in the post-FGD configuration. To accomplish this, SCR catalysts would need to overcome performance effects from kinetic limitations at low temperature, and from possible fouling due to condensation of ammonium-sulfur compounds on the catalyst surface and within the catalyst pores. At reactor temperature of 550"F, the extruded V/Ti catalyst exhibited marginally lower, but steady performance over 1,400 operating hours even though catalyst fquling was detected. The composite catalyst exhibited a more severe performance decline that varied with changes in the inlet NOx concentration from the host boiler. After each test period, both catalyst's performance was restored to original levels when the temperature was raised to baseline (650°F) conditions. NiMo RESIDUAL OIL SCR PILOT The NiMO pilot unit was operated between October 1991 and October 1993, with flue gas flowing through the unit for approximately 4,800 hours. The host boiler is used for load following, typically cycling down to 20% MCR overnight, and with hourly load changes of up to 60% MCR. Catalyst Activity. Activity changes were measured by both catalyst vendors on samples taken after 2,400 and 4,100 operating hours. The relative activity of the corrugated plate catalyst increased during both test intervals, and exceeded, the original activity by 24% by the end of the test program. The effect was attributed to deposition of vanadium from the flue gas on the catalyst surface, since SO;! oxidation rates also increased with time over both catalysts. The measured fuel oil vanadium content varied between 55 and 170 ppm during the test program. The activity of the top layer of composite V/Ti catalyst decreased somewhat during the test program, but no overall performance change was detected via pilot NOx conversion i \ 1 1036 and ammonia slip measurements at the reactor exit (after 3 layers). After 4,100 hours, the activity of the top layer declined by roughly 20% based on the average of values from samples taken from the tops of the first and second layers, and essentially no activity change was seen in the second and third beds during the course of the test program. Deactivation in the first layer was attributed to masking by a thin layer of solids found on the catalyst surface. The catalyst vendor concluded that solids deposition and consequential activity loss at the top of the reactor was exacerbated by the aggressive catalyst pitch (3.6 mm). A more conservative catalyst design and additional measures to prevent solids deposition (Le., sootblowers above every catalyst level) are advisable for full-scale SCR systems in similar heavy oil service. Catalyst Plugging and Deposition. Although the particulate content of the flue gas from the NiMo host boiler is considerably less than that at TVA, problems with catalyst deposition and pluggage were encountered throughout the test program. Reactor deposits were found to consist of oil ash, magnesium oxide (MgO) and magnesium sulfate, the magnesium source being fuel oil additives. Plugging countermeasures for the pilot were limited to sootblowers above the first catalyst layer and routine catalyst cleaning during system shutdowns. Although not proven at the pilot scale, more strict control of MgO usage may reduce solids deposition and catalyst pluggage effects in full-scale SCR systems for residual oil boilers. In addition, sootblowers were found to be highly effective in preventing catalyst pluggage in this service in a detailed evaluation at another EPRI-sponsored pilot. Other Operating Issues. Operational lessons from the NiMo pilot study include the demonstration of direct liquid ammonia injection, and process control issues associated with inconsistent aqueous ammonia concentrations and deep cycling of the host boiler. SCR Design and Operational Recommendations Report Results from all EPRI-sponsored pilots are currently being incorporated into a guidance document entitled SCR Design and Operational Recommendations: RbD Lessons Learned (EPRI Report TR-105103). The report will be released later in 1995, and will include results and design implications from the three pilot studies described in this paper, in addition to the results from the advanced SCR pilot system at the Pacific Gas & Electric Company's Morro Bay Station, and the multi-pilot SCR system at Southern Company Service's Plant Crist sponsored under the DOE'S Clean Coal Technology Program. HYBRID SCR Hybrid selective catalytic reduction (SCR) systems consist of either a combination of SCR techniques (i.e., in-duct SCR combined with air heater SCR) or selective non-catalytic reduction (SNCR) in combination with SCR. Depending on unit-specific parameters, a hybrid can offer advantages that include: reduced capital cost, higher NOx reduction without extensive unit modifications; lower system pressure drop; safer and less expensive chemical storage; lower ammonia slip; and operational flexibility. However, a hybrid system can present some drawbacks that may make them less beneficial. These include: system complexity, higher chemical costs, and potentially higher capital costs. EPRI commissioned a study to document the current experience and develop a tool by whichutilities can determine the applicability of Hybrid SCR to meet their NOx reduction goals, a guideline for selecting the best configuration, and a reference for developing the design parameters necessary to implement the technology. There are a number of technical and commercial considerations which must be resolved prior to designing or procuring a Hybrid SCR system. The boiler operating, temperature, and emissions data necessary for the final design are presented along with the process desi@ variables which must be specified. Procurement suggestions are included to assist the user addressing some of the more pertinent commercial issues. 1037 Table 1 Typical SCR Pilot Operating Characteristics . Heat Tmnsfer Host Boiler Fuel Type Pilot Configuration Total Flue Gas Flow, scfm Reactor Temperature, OF Inlet NO,, ppm Inlet SO2, ppm Inlet SO3, ppm Particulate, gr/dscf High (2.5-5.0%) S Coal Hot Side/High Dust 2100 700 450 2000 20 3.0 Med. (1.5-2.5%) S Coal Post FGD 2wO 650 300 150 5 0.0012 1.5% S Residual Oil Hot Side 2000 700 2M)-1000 BM) 23 0.091 0.9 - 0.8 - 0.7 - 0.6 - 0.5 - 0.4 - First Bed "Replacement" Third Bed "Replacement" European Experience _. Range of Selected 0.3 ' 0 4000 8000 12000 16000 Zoo00 24 Exposure Time (hrs) 00 Figure 1 Relative Changes in TVA Vmi Catalyst Activity vs. Exposure Time 84 0 . 8 4 I2.5 4 18.5 I Average SCR Reactor Ammonia Slip (ppm) Figure 2 NYSEG Pilot Heat Exchanger Performance and Return Side Pressure Drop vs. Time (Heat Exchanger Was Water-Washed Between Operating Periods) 1038 I i REMOVAL OF MULTIPLE AIR POLLUTANTS BY GAS-PHASE REACTIONS OF HYDROGEN PEROXIDE Vladimir M. Zamansky, Lac Ho, Peter M. Maly, and William R. Seeker Energy and Environmental Research Corporation 18 Mason, Irvine, CA 92718 Keywords: Air Pollution, Hydrogen Peroxide, NO, INTRODUCTION Hydrogen peroxide is a large-volume chemical with a wide range of applications in different industries. If properly stored, hydrogen peroxide solutions in water are stable, with no loss of the effective substance. Environmental applications have become a major area of use for hydrogen peroxide because it is not itself a source of pollution, and water and oxygen are the only reaction byproducts. There is a variety of developed or developing environmental technologies which use H,O, as an active reagent: detoxification and deodorization of industrial and municipal effluents; low temperature removal of nitrogen oxides, sulfur dioxide, cyanides, chlorine, hydrogen sulfide, organic compounds; low temperature treatment for catalytic NO-to-NO, conversion, etc. This study develops a novel concept of high-temperature H,O, injection into combustion gases or other off-gases followed by gas-phase reactions of H202 with NO, SO,, CO, and organic compounds. Experimental and modeling data show that a water solution of hydrogen peroxide injected into postcombustion gases converts NO to NO,, SO, to SO,, and improves the removal of CO and organic compounds due to chain reactions involving OH and HO, radicals. The existence of the chemical reaction between NO and hydrogen peroxide has been proven earlier experimentally by Azuhata et al.' at long residence times of approximately 12 sec which are not applicable to air pollution control. In this study effective NO-to-NO, and SO,-to-SO, conversion, as well as CO and CH, oxidation, was predicted by kinetic modeling and measured experimentally in the temperature range 600-1 100 K in a practical range of reaction times (6) from 0.2 to 2.0 s. EXPERIMENTAL In the current work, the bulk of experiments were carried out in a flow system which consists of four parts, a gas blending system, a liquid injection system, a reactor, and an analytical train. The gas blending system is a set of rotameters capable of preparing a flowing mixture of 0, with addition of NO, CO or CH, in N, as a carrier gas. The liquid injection system includes a burette containing 3% H,Op,O solution and a precision metering pump for delivery of the solution through a capillary tube to the heated reaction zone. For the study of the SOJH20, reaction, dilute sulfuric acid was added into 3% H,O, solution. At a temperature of 500-600 K H,SO, is converted into H,O and SO, and this is a convenient means of producing a gas mixture containing known amounts of H,O and SO,. The rates of pumping the water solutions of H20, and H,SO, were chosen so as to provide the desired concentrations of H,O, and SO, in the gas mixture. The prepared gas and liquid mixtures go to the reactor which was located in a 1 m three zone electrically heated furnace. The first and the third zones (25 cm each) were heated to 450-600 K to evaporate the liquid, to preheat the gas mixture and to avoid condensation of the reaction products in the reactor. In the second heating zone (50 cm long) which was the reaction zone, the temperature was varied from 450 to 1300 K. All tests with air pollutants were performed with a 2.7 cm ID quartz reactor. The experimental gas mixture could be passed through the reactor and then sent to analysis, or it could be sent directly to analysis. The analytical train included a Thermoelectron Chemiluminescent NOiNO, analyzer, a Thermoelectron Gas Filter Correlation' CO analyzer, a Thermoelectron Pulsed Fluorescence SO, analyzer, Flame Ionization Total Hydrocarbon analyzer, and permanganate titration of H,O,. In addition to the laboratory-scale experiments done with synthetic gas mixtures, a set of experiments on NO-to-NO, conversion was also carried out at pilot scale in a 1 MBtu/hr Boiler Simulator Facility (BSF) burning natural gas with stoichiomemc ratio of 1.2. The furnace has two sections: a vertically down-fired tower (56 cm in diameter and 6.7 m height) and a horizontal convective pass (20 x 20 cm cross section, 14.2 m long) simulating typical temperature profiles of full-scale utility boilers. Solutions of hydrogen peroxide. methanol or their mixtures (15% in water) were injected by a fluid nozzle into the convective pass at different temperatures. Flue gas was sampled downstream in the convective pass at different temperatures with residence times from 0.2 to 2.0 s. Most experiments were conducted with sampling at about 500 K and analysis by NO, and CO meters. The Chemkin-I1 kinetic prograd and a reaction mechanism based on Miller and Bowman review paper) were used for kinetic modeling. 1039 LABORATORY-SCALE RESULTS Gas phase reactions of H202 are complicated by heterogeneous processes, and therefore, a preliminary set of experiments was done for a mixture without air pollutants (1100 ppm H,O, - 7.3% H,O - balance air) to define the degree of H,O, heterogeneous decomposition under different experimental conditions and to estimate how much H202 will be available for the useful homogeneous reactions with air pollutants. A water cooled impinger with a known amount of KMnO, solution was installed at the exit of the reactor. The concentrations of hydrogen peroxide leaving the reactor were defined by "on-line titration", e.g. by measuring time for which the gas passes through the KMnO, solution until decoloration. Results show that about 75% H,O, decomposes at temperatures which are lower than the threshold temperature for homogeneous decomposition. The measured heterogeneous rate constant was 5.5exp(-1,25Om s.', and it was included in modeling. A substantial excess of H,O, was used for the tests with air pollutants. It is worth noting, however, that in the scope of scaling up the process the surface chemism becomes less important in large size industrial installations. All concentrations of hydrogen peroxide, shown in this Section, are calculated values after substraction of the heterogeneously decomposed HzO, bebore the reaction zone from initial H20z concentrations. ,Conversion. Two set of tests were performed to demonstrate the NO to NOz conversion in the presence of HzO; variation of temperature and variation of reaction time. Two initial gas mixture compositions were used for the tests with different temperatures: (I) 100 ppm NO - (160- 220) ppm H,O, - 4.2% 0, - 4.8% H,O - balance N, and (2) 100 ppm NO - (90-120) ppm H,O, - 4.4% 0, - 1.7% H,O - balance N,. The flow rates of the H,O, solution were 0.05 and 0.14 mumin correspondingly and reaction time 6=1-2 s. Experimental and modeling results for the same conditions are presented in Figure 1. One can see that experimental and modeling results agree at least qualitatively and that at H20f10 ratios equal to 1.6-2.2 and 0.9-1.2, the achievable NO-to-NO, conversions are 95 and 80%. In the next set, the experiments were conducted at 820 K. and air was ' added to the mixture (1) in order to increase the gas flow rate and decrease the reaction time. The concentrations of NO and H,Oz were adjusted to the same levels as in previous tests. Concennations of H,O and 0, were 4.7-7.0% and 4.2-15%. Five various air flow rates from 2.5 to 10.0 Vmin were checked, and there were no visible difference in the final NO concentration: it was in the range of 10- 12 ppm at ~=0.4-1.4s . sn@- . Average gas mixture composition for these experiments was 100 ppm SO, - (160-220) ppm H,O, - 4.2% 0, - 4.8% H,O - balance N, and the reaction times were between 1.0 and 1.6 s. Under certain conditions SO, reacts with H202 to form SO,. No sulfur dioxide was formed at T=600-1100 K when hydrogen peroxide was absent in the mixture. SO, measurements and modeling for different H,O, concentrations are shown in Figure 2. One can conclude that SO, is converted to SO, in a temperature range of 800-1100 K with up to 7545% efficiency. _ _ ion of CO Puunated by H2Q2. The goal of this set of experiments was to show the improvement of the CO oxidation in the presence of H202. In other words, it was expected according to kinetic calculations that HzOZ will make it possible to reduce CO concentrations at lower temperatures than that without HzO,. The results of experiments and calculations are compared in Figure 3 at t,=1.0-1.5 s for three mixtures: (1) 90 ppm CO -4.2% 0, - (160-220) ppm H20z - 4.8% H,O - balance Nz. (2) the same mixture but without H,O,, and (3) the same mixture but without H,O, and H,O. Modeling for the mixture (3) was done at 10 ppm H,O in the mixture because in experiments it was prepared without special drying. It is clear that experiments and modeling well agree and that H,O, promotes CO oxidation but at rather low extent, about 20% at 860-960 K. WQIn the presence of H,,O, the te.mperatu re limit of CH, removal is substantially shifted to lower temperatures. This is shown in Figure 4 at t,=1.0-1.8 s for three mixturns: (I) 90ppm CH, -4.2% Oz - (160-220) ppm H,O, - 4.8% H,O - balance N,, (2) 90 ppm CH, - 4.4% 0, - (90-120) ppm H202- 1.7% H20- balance N,, and (3) the same mixture as (I) but without H,O, and H,O. In the temperature range from 790 to 1060 K, the addition of H,O, can provide from 20 to 90% CH, removal. Maximum performance is observed at T = 900 - 1040 K. . . PILOT-SCALE RESULTS An attractive method of NO-to-NO, and SO,-to-SO, conversion by injection of methanol into the flue gas was described by Lyon et al.' In pilot-scale experiments recently performed by Evans et al.', 87% NO-to-NO, conversion was achieved. Unfortunately, a problem with using methanol is the formation of CO as a by-product. Each molecule ofNO or SO, converted into NO, and SO, produces a molecule of CO, and CO is not oxidized to CO, at methanol injection temperatures. Results of NO and CO measurements after injection of H,O, and CH,OH are shown in Figure 5 for two initial NO levels of 400 (Figure 5a) and 200 ppm (Figure 5b). For all tests the molar ratio of [Agentl/[NO] was 1.5, and O,concen!mtion in flue gas was 3.8%. Maximum NO-to-NO, conversion 1040 I 1 I' / i Was in the range of 80437% for H20, injection and 87-92% for CH,OH injection. For comparison, in the PRvious tests', 87% NO-to-NO, conversion was achieved by CH,OH injection. The minimum Of the temperature window is shifted to lower temperatures in the case of H202 injection as predicted by kinetic calculations. The mechanisms of NO-to-NO, conversion by H,O, and CH,OH injection are similar, and therefore the slight decrease in performance of H,O, can be explained by the heterogeneous decomposition which might be still noticeable in the 20 x 20 cm duct. AS for CO measurements, the H,O, injection almost does not affect 24 ppm CO exiting the furnace tower, which is consistent with the laboratory-scale tests for low H20z levels. Methanol injection generates high CO emissions of about 600 and 300 ppm as shown in Figure 5. Methanol is less expensive than H20,. Therefore, if CO emissions are considered to be the primary drawback of CH,OH injection, one strategy might be to add as much CH,OH as possible within CO limits, and then add enough H,O, to obtain target NO conversion. In light of this, several tests were performed in which the agent consisted of various combinations of CH,OH and H,O,. In Figure 6 measured NO and CO concentrations are shown at different agent injection temperatures for various [H,OJ/ICH,OH] mixtures. Initial NO and CO levels for these tests were 70 and 30 ppm respectively, and total CH,OH+H,O, concentration was always 105 ppm. At H,OJCH,OH=l:l (52.5 ppm H203 NO-to-NO, conversion was approximately the same as for pure CH,OH injection, and then NO conversion decreases incrementally as [H,O,]/[CH,OH] ratio increases. The CO emissions increase also incrementally as CH,OH concentration grows (Figure 6b). The temperature window for NO-to- NO, conversion had about the same minimum for all H,OJCH,OH mixtures and incremental temperature shift was not observed. This is explained by appearance of OH radicals at lower temperatures in the presence of H,O,. The NO and CO concentrations shown in Figure 66 were measured at the minimum point of the H,OJCH,OH temperature window, 800 K, except for the NO and CO concenuations after injection of pure methanol ([H202]=0). These concentrations were measured at 866 K, the minimum point of the CH,OH temperature window. DISCUSSION It is known that NO, is much more soluble in water than NO. Kobayashi et al.6 demonstrated that NO, can be removed by aqueous solutions of various inorganic and organic reagents. Senjo et al.? reported several methods of NO, removal by sodium salts. It was also proven by Zamansky et al.' that NO, can be removed efficiently in modified calcium-based SO, scrubbers. Since flue gas desulfurization systems are increasingly required for SO, removal after combustion of sulfur containing fuels, the conversion of relatively inert NO into much more reactive NO2 and conosive SO, into SO, becomes promising for combined NO, and SO, removal. Hydrogen peroxide injection is a "green" process. It is not dangerous for the atmosphere. there is no additional soot, CO or nitrogen compounds formation as may be expected from urea, cyanuric acid or methanol injection. H,O, can be injected as a water solution at various concentrations. The products of the H,O,decornposition at high temperatures are H,O and 0, which are environmentally acceptable. Therefore, hydrogen peroxide can be applied in any reasonable excess to air pollutants for their complete or partial removal depending on current needs without risk of ammonia, CO or other dangerous compound breakthrough. In the homogeneous H,O, decomposition the total amount of OH radicals increases due to dissociation: H202 + M - 2 OH + M. The hydroxyl radicals formed have several reaction routes, including (1) the reaction with H,O, molecules to form HO, radicals: OH + H,O, - H,O + HO,; (2) chain termination steps, such as OH + HO, - H,O + 0,; and (3) interaction with carbon-containing compounds, such as CO, CH, and other organics: OH + CO - CO, + H, OH + CH, H,O + CH,, etc. The total CH,-O, reaction, CH, + 20, = CO, + 2H,O. is promoted in the presence of OH radicals. As known from the literature', H,O, enhances oxidation of some other organic compounds due to the chain processes involving OH and other active species. Cooper et aL9 found that injection of H,O, in dilute air mixtures of heptane and isopropanol increases the rate of their destruction at T = 910-1073 K and f = 0.26-0.94 s. The HO, radicals, formed in the reaction of OH radicals with H,O,, play an important role in pollutants removal. The interaction of HO, radicals with NO, HO, + NO - NO, + OH, is the only rapid NO reaction at low and moderate temperatures, and this is the principal route of NO-to-NO, conversion. The HO, species react also with SO, followed by HSO, thermal decomposition: HO, + SO, - HSO, + 0, and HSO, + M - SO, +OH + M. Both modeling and experimental results show that NO is not converted to NO, in the absence of H,O,, but SO,, CO, and CH, are converted to SO, and CO, at higher temperatures even without H,O, addition. However, in non-ideal practical combustion systems all these pollutants, SO, and carboncontaining compounds, are present in flue gas, and H,02 injection will reduce their concentrations. 1041 The position of the H,O, temperature window is defined by chemical nature of H,O, reactions. At temperatures lower than 600 K the homogeneous H20, decomposition is very slow and OH and HO, radicals are not formed. At temperatures higher than 1100 K, concentrations of all radicals in the system become very high, and the rate of recombination reactions which are quadratic on radical concentration prevails over the rate of their reactions with molecules. An important factor is also the decomposition of HO, radicals at temperatures higher than 1000 K. Thus, H,O, is active only in the temperature range of 600-1 100 K. It is believed that four chain reactions are involved in removal of air pollutants: NO: SQ&tmyal: OH + H,O, - H,O + HO, HO, + NO - NO, + OH OH + H,O, - H,O + HO, HO, + SO, - HSO, + 0, HSO,+ M - SO,+OH + M H + 0, - OH + 0 -reduction: CH, + 202 = CO, + 2H,O (promoted in the presence of OH radicals) - chain reaction - chain reaction -chain reaction - chain reaction CO: OH+CO-CO,+H Thus, the single reagent can remove multiple air pollutants. One can use H,O, injection in combination with other NO, control technologies, such as reburning, ammonia or urea injection, etc. to reduce NO to a very low level. In this case rather low NO concentrations (100-200 ppm) will react with H,O,, which reduces the cost for the additive and reduces the residual (after scrubbing) NO, concentration, preventing the NO, brown plume. For example, in the COMBINOX process which includes reburning, urea injection, methanol injection and SO,/NO, scrubbing, H,O, could either completely or partially replace methanol to meet CO regulatory limits. Assuming 90% NO-to-NO, conversion by H,O, injection and taking into account the pilot-scale results in other COMBINOX steps the total process will reduce NO, emissions by 96%. CONCLUSIONS This paper demonstrates the feasibility of multiple pollutants removal (NO, SO,, CH,, and CO) by hydrogen peroxide injection within reaction times (0.2-2.0 s), temperatures (600-1 100 K), and other conditions which are in the practical range for ils application in boilers, furnaces, engines and other combustion installations. In the presence of H,O,, maximum NO-to-NO, conversion was 95% in the flow system and 87% in pilot-scale at H,OflO = 1.5. SO, was effectively converted to SO, with up to 85% efficiency. CO-to-CO, conversion was slightly enhanced by about 20% at temperatures of about 900 K. Formation of carbon monoxide is incrementally increases when methanol is added to H,O,. Mixtures of methanol and hydrogen peroxide can be injected to remove NO and to meet CO regulations at reduced cost for the additive. In the presence of H,O,, CH, is effectively (70-90%) removed from flue gas at 1000 K and at H,O#JO= 0.9-2.2. Kinetic modeling describes quantitatively or at least qualitatively all substantial features of NO, SO,, CO and CH, reactions with H,Oz. ACKNOWLEDGMENT This work was supported by the U.S. Department of Energy under a grant No. DE-FG05-93ER81538, Project Officer - Dr. Robert S . Marianelli. REFERENCES I. Azuhata, S , Akimoto, H. and Hishimum, Y . AIChE Jourrial, v. 28, pp. 7-11 (1982). 2. Kee, R.J., Rupley, F.M. and Miller, J.A. Sadia Nat.L.ab.Rcport No. SAND89-8009 (1989). 3. Miller, J.A. and Bowman, C.T. Progr. Energy Combust. Sci., v. 15, pp. 287-338 (1989). 4. Lyon, R.K., Cole, J.A., Kramlich, J.C. and Chen, S.L. Comb. Flame, v. 81, pp. 30-39 (1990). 5. Evans, A.B.. Pont, J.N., and Seeker, W.R. (1993). Development of advanced NO, control concepts for coal-fired utility boilers. EER Rcporf, DOE coiitract DE-AC22-90PC90363. 6. Kobayashi, H. Emir. Sci. Techno/, v. 11, No. 2, pp. 190-192 (1977). 7. Senjo. T. and Kobayashi, M. U.S. Patenr, 4,029,739 (1977). 8. Zamansky, V.M., Lyon, R.K., Evans, A.B., Pont, J.N., Seeker, W.R. and Schmidt, C.E. (1993). Development of process to simultaneously scrub NO, and SO, from coal-fired flue gas, 1993 SO, Control Spy., EPRIEPADOE, Boston, Val. 3, Session 7. 9. Cooper. C.D., Clausen. C.A., Tomlin, D., Hewjett. M. and Martinez, A. J. Hazard. Mat., v. 27, pp. 273-285 (1991). 1042 I O L 500 EE" 40 d 020 0 500 a 1300 1043 100 , I a 500 700 900 1100 1300 Temperature, K -53 ppm H202 120 - -& - 63 ppm H202 b 80 -- E n - 60.- -C 40 -- 2 0 20.- 500 700 900 1100 Temperature, K 1300 Figure 4. CH, oxidation by H202 injection. (a) - experimental and (b) - modeling data for t,= 1.0-1.8 s; mixture (1): 90 ppm CH, -(160-220) ppm H202 - 4.2% 0, - 4.8% H,O - balance N,; mixture (2): 90 ppm CH, - (90-120) ppm H202- 4.4%0, - 1.7% H,O -balance N,. 600 E n 400 8 9 200 U (II I 0 600 700 800 900 1000 1100 Temperature, K 600 700 800 900 1000 1100 Temperature, K Figure 5. NO and CO concentrations after pilot-scale injection of H20, (solid curves) and CH,OH (dash curves). [Agentl/[NO]=lS. (a)-[NOl,=400 ppm,(b)-[NOl,=200 ppm. 80 - . & - - 8 4 ppm H202 B -.c9.5p pm H202 n 105 ppm H202 6 60 E" 40 L .. . --C 20 2 0 650 700 750 800 850 900 950 Temperature, K 100 E :: 80 20 0 I.. ...[.C.O..I.. ................... NO> - 120 Figure 6. NO and CO concentrations after pilot-scale injection of H,O,/CH,OH mixtures. ([H,OJ+[CH,OHI)/[NOl,=l.[5N,O ],=70 ppm. (a) - temperature windows for various H20JCH,0H mixtures, (b) - NO and CO concentrations at 800 K (at [H,OJ=O data are shown for 866 K). I 1044 r" EPRICON: Agentless Flue Gas Conditioning For Electrostatic Precipitators Peter Paul Bibbo V.P. & G.M. of APCD Research-Cottrell, Inc. Division of Air & Water Technologies Branchburg, New Jersey Keywords: Electrostatic Precipitator, SO3 Gas Conditioning, Oxidation Catalyst INTRODUCTION Achieving efficient particulate control in coal burning electric utility plants is becoming an increasingly difficult proposition, given the variety of regulatory, technical, operating and environmental pressures that exist in the U.S. For most powerplants, particulate control is achieved by an electrostatic precipitator (ESP). Under optimal conditions, modern ESPs are capable of achieving particulate removal.efficien- CieS of 99.7% and higher ... well within the regulatory levels Prescribed by the Clean Air Act. Unfortunately, optimal conditions are not always present. ESPs are sensitive to flue gas . conditions, and those conditions may change dramatically after a fuel switch or the installation of some types of emissions control technology upstream of the ESP. Gas conditioning has been shown to be an effective means of returning flue gas to the 'optimal" conditions required for efficient ESP operation following a fuel switch to a low, or at least, lower sulfur coal. Borrowing technology common in conventional soap-making plants around the turn of the century, sulfurburning SO3 gas conditioning has been the solution to may difficult fuels in electrostatic precipitators. Although it has contributed most to improved ESP performance after a fuel switch, conventional gas conditioning has significant drawbacks, including the need for maintaining a little chemical plant, and otherwise storing or handling toxic materials. In an effort to develop an alternative to conventional SO3,gas conditioning, the Electric Power Research Institute (EPRI) initiated a research and development project that has produced an alternative and modern technology for flue gas conditioning, now called EPRICON, and licensed it to Research-Cottrell. FLUE GAS CONDITIONING Changing Flue Gas Conditions The majority of ESPS now operated by U.S. electric utilities are more than 20 years old, and were designed to operate primarily on high sulfur fuels. When designed, these devices were capable of meeting opacity standards of 20 per cent and emissions levels in the range of 0.1 lb/MMBtu. Those earlier emissions control standards have been replaced by a host of subsequent regulations, most recently the Clean Air Act Amendments of 1990, many of which directly or indirectly affect particulate collection. Switching from high sulfur to a lower sulfur coal is currently the favored means of attaining compliance under Title IV of the CAAA, which regulates acid gas emissions. Different coals have different chemical and physical characteristics, however, and can be expected to change flue gas conditions and particulate properties substantially. Some low sulfur coals have high ash contents, for example, and will increase particulate loading, which may strain the ash handling system. For coals with a very low sulfur content, typically one per cent or below,. the resulting flyash exhibits high electrical resistivity, which may significantly reduce ESP perfcrmanze. Addressing High Resistivity is converted to SO3 (typically less than 2%). When temperature and humidity conditions are favorable, the SO3 thus generated is absorbed on the surface of the flyash particles and is suffficient to reduce ash electrical resistivity. under acceptable resistivity levels and other good operating conditions, ESPs can achieve collection efficiency over 99.9%. High particle resistivity (typically above 5E10 ohm.cm) will decrease the ESP's overall collection efficiency, however, because dust begins to limit current flow and sparking voltage in the ESP. AS an alternative to enlarging the ESP, gas conditioning can restore the required resistivity conditions to ideal performance levels. Early applications of gas-conditioning used liquid SO3 which was vaporized and diluted with dry air, or concentrated sulfuric acid, which was vaporized with hot air. A second generation of 1045 A small fraction of the SO2 produced by the combustion of coal gas-conditioning technology using SO2 as feed material was developed. More recently, burning molten elemental sulfur to produce SO2 prior to the catalyst bed was proven, and this technology emerged in the 1970's as the dominant choice. The EPRICON Process The EPRICON process provides required gas conditioning without the need for external agents, such as liquid SO2 or vaporized molten sulfur. In addition, it eliminates the need to filter the gas of particulates prior to its entry into the gas-conditioning chamber, and eliminates the need for an additional fan to move the conditioned gas into the electrostatic precipitator. The process (Figure 1) operates by withdrawing a small fraction of the flue gas from a location in the boiler where the operating temperature is in the range of 800°F to 90O0F. This fraction of flue gas, or slipstream, is then passed over a catalyst heated by the gas, where between 30-70 percent of the SO2 in the flue gas is converted to SO3. The slipstream, now SO3- rich, is re-injected after the air preheater but ahead of the ESP to provide the required SO3 for the reduction of resistivity. The feasibility of the technology is dependent on case-by-case conditions. If, for example, 5ppm of SO3 can treat the ash adequately and the flue gas contains 500 ppm, from 1 to 2 percent of the gas must be treated. Conversely, if 15 ppm of SO is needed, a little over 3 percent of the gas containing 580 ppm of SO2 would have to be treated. Three percent is considered to be the upper limit of a range for continuous operation that has been identified, as economically and technically desirable, although operation above this range to deal with difficult but temporary coal supplies is feasible. PILOT PLANT ed by EPRI to determine the operability of the catalyst in a slip-stream flue gas system over a period of time. The pilot system was constructed at Alabama Power Company's plant Miller and identified a number of design parameters for the EPRICON process. This pilot is still in operation. FULL SCALE DEMONSTRATION installed a full-scale turnkey EPRICON system on a 250MW public utility boiler in the Northeastern U . S . This boiler is about 25 years old, and was originally designed to fire a high sulfur coal. The new compliance coal is to cover a wide variety oE sources all of which will contain much lower sulfur than the original design. The boiler is equipped with its original precipitator, which cannot meet emissions regulations while the boiler is firing compliance sulfur coal. This full scale demonstration system (Figure 2) incorporated the fundamental premises of the EPRICON technology, such as avoidance of pre-cleaning the gas (the catalyst operates in "dirty" raw flue gas) and the absence of an air mover to push the slipstream through the catalyst chamber (gas flow is induced through the catalyst by the differential pressure across the air preheater). The full scale system also borrowed some of the design parameters of the pilot program, mainly the catalyst itself and its arrangement, but after that, the differences from the pilot were many. Inlet Duct which provided the convenient design choice to provide two parallel catalyst chambers, each with its own gas take-off. The boiler gas remaim split all tho way through ::le grecipitators, which is ideal for side-by-side diagnostic and characterization tests. Also. there was no need to mix gas from two different temperature sources. The twin inlet ducts are fabricated from 1/4" ASTM-A242 plate and insulated with 5" of mineral wool covered with a flat aluminum lagging. The ducts are simply supported at the boiler casing penetration and the top of the catalyst vessels. AII expansion joint, a guillotine isolation damper, and motorized flow control damper are installed right at the boiler off-take. Catalyst Chamber Although there is a variety of catalyst formulations and substrates that can perform the necessary conversion, it was decided to Stay with the same catalyst that was selected for the pilot. (Figure 3 ) The chamber is a rectangular cross-section 6 ' - 6 x 10'- 4, fabricated from 1/4" A242 plate and has the catalyst blocks arranged in six ( 6 ) layers (two (2) layers have purposely been left empty for future catalyst addition, if necessary). The cata- A pilot program on a pulverized coal-fired boiler was conduct- In the spring of 1994, Research-Cottrell designed and The boiler is physically split in the convective section, 1046 L l / lYSt is supported in the chambers by means of fabricated tee sections. The gas flow through the chamber is vertically downward. A generous gap was left between catalyst layers for fitting with 'puff' blowers to knock of ash deposits that can form on the flat tops of the catalyst blocks, but acoustic devices were also installed as a alternative to air blowing. Outlet Duct And Distribution System This outlet duct is fitted with a guillotine shut-off damper Provided to isolate the chamber for maintenance. Penetration of Converted flue gas into the main gas duct is by means of a unique "expansion box" from which the distribution header is hung. The header answered one of the questions from the pilot study: simple injection pipes and full height air foils have proven excellent performance in terms of treated gas injection and distribution upstream of a precipitator that is very close Coupled to the air preheater. System Control Modulation of the system is simple. A flow transmitter in the inlet duct modulates a double lovered flow control damper in the inlet duct directly down stream of the inlet isolation guillotine . PERFORMANCE variety of extraction and instrumented test procedures. pilot tube and thermocouple. Good agreement was achieved on the North (designated side 11) chamber between the measured flow rate and the flow rate indicated by the installed electronic flow meter. Flow rates were measured at full boiler load and at a reduced boiler load. At full load, gas volumetric flow rate ranged from 23,500 to 28,200 ACFM at approximately 850°F per side. Lower boiler load tests were run between 13,400 and 15,300 ACFM per side. SO3 Conversion ber during 16 characterization tests using both an analyzer installed on the boiler and by standard wet chemical procedure. Again, agreement between these methods was good, so eventually, most reliance was placed on the instrument reading which, besides being faster, tends to be more accurate. SO3 measurements by analyzer are not possible, so the Goksoyr-Ross controlled condensation method was used. SO3 conversion can be approximated by the difference in SO2 concentration at the inlet and outlet of the EPRICON chamber, and by direct measurement in SO3 at the inlet and outlet, the difference being the apparent conversion from SO2 to SO3 by the action of the catalyst. Direct SO3 measurement indicated a conversion from about 10 ppm at the inlet to about 200 ppm at the outlet, for an average conversion of over 70% at full load expressed in standard units. (Figure 4) At low load, conversion increased, as expected, to about 85%. Compared to SO2 measurements, the SO3 levels at the outlet of the chamber appear to be understated. However, the Goksoyr-Ross method is a non-isokintetic technique which would tend to under-collect fly ash at the EPRICON outlet. If any SO3 were to become attached to flyash particles, perhaps by adsorption above the condensation temperature, this fraction of the converted SO2 could easily be missed by the test method. Conditions At The Precipitator Inlet SG3 concentrations at the BSP inlet ranged between 12 and 23 ppm at high and low boiler loads, respectively. So3 and temperature uniformity were of great interest in the design stage, so gas sampling at several locations in a grid across the ESP face was done to measure both SO3 and gas temperature. The results showed acceptable uniformity for both parameters, and prove the adequacy of the injection apparatus for this technology. Temperatures were also measured with EPRICON dampered off. Average flue gas temperature rise across the face of the ESP was uniformly above 10°F. a little lower than expected, which is most likely attributable to the somewhat lower than expected gas outlet temperature from the chambers. SO3 concentration again is probably slightly understated due to the non-isokinetic nature of the direct measurement procedure. Flyash Resistivity And Precipitator Current Density Fly ash resistivity was not measured directly during these first characterization tests, but ESP power levels were recorded with and without EPRICON valved in. Power levels were, monitored 1041 Characterization tests were run in June and July 1994, using a Flow rates were established using EPA approved methods with a SO was measured at the inlet and outlet of the EPRICON chamwith one EPRICON chder on line and the other chamber cut Off with its outlet isolation damper. The on-line chamber was then shut off and the other chamber was brought on line. In each case, the change in ESP power was significant and rapid, showing a strong correlation between EPRICON chamber SO2 content and ESP corona power. (Figure 5) The fact that each EPRICON chamber serves a separate precipitator reinforces this conclusion. Total ESP power was increased about 200% on Side 11 28 kw to 68 kw and a little less on Side 12 (35 kw to 65 kw). Overall ESP was increased from 0.25jWattsjFt2 to 0.53 WattsjFt2. Second Full Scale Unit near-identical 250MW boiler at the same plant site. Since a complete battery of characterization and performance tests were not completed prior to the decision to install this second system, the catalyst chambers are virtually identical except that the second unit has a simpler access system. This unit was completed in December, 1994. THE BOTTOM LINE Compared to conventional gas conditioning, the EPRICON gas conditioning system minimizes the need for external chemicals or apparatus to achieve a reduction of resistivity. The system is applicable to power stations with high resistivity ash, often produced by the use of low-sulfur coals, that can be treated adequately with SO3. That reduction of electrical resistivity will enhance the performance of the ESP particulate-collection device. Capital Cost Based on these two, 250 MW installations, the EPRICON technology is expected to cost under $4.50/kw on a completely installed turnkey basis. These two boilers are big enough to scale well to most other utility sizes except perhaps units over 600 MW or so. Between 100 and 600 MW, the use of dual chambers should be a preferred choice when separate or unitized precipitators are installed, and this is typically the case. Installation labor and auxiliaries such as dampers, expansion joints, and access systems comprise over 50% of the total system cost. Operating Costs The operating costs of EPRICON are noted in two areas: thermal penalty due to the 3 percent of flue gas unavailable for heat exchange through the air preheater, and maintenance of the catalyst bed. Thermal penalties are estimated to be insignificant for slipstreams of 3 percent or below however, this assumption will be vigorously tested in full scale tests. Catalyst rejuvenation costs are anticipated every two years to restore SO2 conversion efficiency at a minimum of 50 percent. This translates to less than 7 cents per kw per year. as a result of breakage. Catalyst replacement costs are estimated at approximately $1,000 annually. Present Status BIBLIOGRAPHY Brown, Robert F., Quantitative Determination of Sulfar Doxide, Sulfur Trioxide and Moisture Cont of Flue Gases. Dahlin, R.S., et al, SRI and Others. A Field Study of a Combined NH3 Conditioning System on a Cold-S Fly-Ash Precipitator at a Coal-Fired Power Plant. Dismukes, Edward 9. A Review of Flue Gas Conditioning 1983 with Ammonia & Organic Amines. Paper presented at the 76th Annual Meeting of the Air Pollution Control Association, Atlanta, GA., June 19-24, 198:. Linsberg, Mark, Ferrigan, James, Krigmont, Henry. Evaluation of an SO3 Flue Gas Conditioning Program for Precipitator Enhancement at the J.M. Stuart Station. Presented at the 1987 Joint Power Generation Conference, Miami, FL, October 4-8. Eskra, Bryan, Kinney, Bill G., One Year's Operating Experience with SO3 Condition on a Large Coal-Fired Unit's Electrostatic Precipitator. Presented at the Air Pollution Control Ass. Annual (75th) Meeting., New Orleans, LA June 20-25 1982. La Rue, J.M.. Latham, B.F., SO3 Conditioning Agent System for Fly-Ash Precipitators. Singhvi, R., Sulfur Dioxide to Sulfur Trioxide Conversion Using Vanadium Pentoxide as a Catalyst Determination of Sulfur Dioxide Concentration at the Catalytic Converter Outlet. WhalCO, Cumings, W.E., Reamy, W.H., Baltimore Gas & Electric Experience with Combined SO3/?i~i3 Injection for Precipitator Performance Improvement. - In October, 1994, work began on a second EPRICON system on a A second maintenance cost is incurred for catalyst replacement 1048 I Figure 1. EPRICON Process I Figure 2. Full Scale System (one of two sides) Figure 3. Catalyst I I ,/ 1049 IO is 0 ac .u O.fL.U.1 Figure 4. Apparent SO, Enrichment I Figure 5. Precbilator Power Enhancement I 1050 CONTINUOUS REMOVAL OF SULFUR OXIDES AT AMBIENT TEMPERATURE, USING ACI?VATED CARBON FIBERS AND PARTICULATES Y. Fei, Y.N. Sun, E. Givens and F. Derbyshire Center for Applied Energy Research, University of Kentucky. 3572 Iron Works Pike, Lexington, KY 4051 1-8433 Keywords: Activated carbon fibers, flue gas, clean-up, SO2 removal INTRODU~ION The control of sulfur dioxide emissions from fossil fuel.combustion and other industrial processes has ken recognized as one of the major environmental issues, in both developed and developing countries. In the US, energy-intensive and space-consuming sorbent scrubbing processes that are widely used to remove SO2 from flue gases also produce huge amounts of process wastes. The management and disposal of the by-product wastes by landlill not only represent poor resource utilization, but can cause furrher environmental and land use problems. Activated carbons have offered alternative technologies for the clean-up of flue gas streams. A dry process for the simultaneous removal of sulfur and nitrogen oxides has been commercialized by Mitsui - Bergbau Forschung, using granular activated carbons[l]. Carbon is lost in this pmess by chemical reaction and by athition, and to supplement this loss accounts for about half of the process operating cost. In addition, high capital costs are associated with the large reactor volumes and the systems to transport ganular carbons in moving bed operations, and have provided obstacles to the wide-scale development and use of the process. In the early 1970s. studies were made of the continuous oxidation and hydration of sulfur dioxide over granular activated carbons in a mckle bed, with the desorption of sulfuric acid by flowing water in the same reactor [2]. Similar concepts of water desorption have been also proposed for the regeneration of activated carbons [3]. An important feature of these methods is that sulfur species are converted to useful chemicals in the form of sulfuric acids. However, the wet desulfurization process is limited by slow rates of oxidation and mass transfer through liquid phase. Improvements have been recently made to increase process effectiveness and to obtain high concentration sulfuric acid, including cyclic operation of trickle beds, higher reaction temperature (at 80 OC rather than ambient temperature) and the loading of platinum on activated carbons [4,5]. A combined process that also removes NOx has been proposed through selective catalytic reduction (SCR) with ammonia in a separate unit, possibly using a different activated carbon catalyst [5]. On the other hand, Mochida and his coworkers at Kyushu University, Japan, have found that activated carbon fibers (ACF) produced commercially from polyacrylonitrile (PAN) are very effective catalysts for the continuous removal of SO2 from humidified model flue gases [6,7]. These and other commercial activated carbon fibers have also exhibited activity for NO oxidation into NO2 at ambient temperature [E]. For the past few years, we have been investigating the synthesis of general purpose carbon fibers and activated carbon fibers from different isotropic pitch precursors [9,10.11]. In collaboration with the Japanese researchers, we have found that certain fibers that we have synthesized in the laboratory are very active for the oxidation of SO2 and NO [12]. The results were so encouraging that we constructed a reaction system to make further investigations. In this paper, we describe the performance of activated carbon fibers and particulate activated carbons for the continuous removal of SO2. The effects of heat treatment, particle size, and several basic engineering parameters of the catalyst bed were also examined. EXPERIMENTAL. Comparisons of catalytic activities for SO2 conversion were made using three different types of activated carbon fibers and a commercial granular activated carbon in two particle size ranges. The activated carbon fibers were produced from coal-tar pitch (a commercial product from Osaka Gas Co.) and synthesized in this laboratory from shale oils and coal liquids (The details of the preparation procedure have been described elsewhere [9,101). The granular activated carbon (BPL type, Calgon Corp.) was produced from bituminous coals and was selected for the study because this material has been already tested in model flue gas-water systems for SO2 oxidation [4,5,13,141. Different particle size ranges were obtained by grinding and sieving. The properties of the activated carbon fibers and particles are summarized in Table 1. Their BET surface areas are varied from 980 to 1060 m?g. The activated carbon samples were either used directly or after heat treatment h nitrogen at 8CCI OC for 1 hour. 1051 Figure 1 shows a schematic of the reaction system. The flow of dry gases from cyliiders were metered by mass flow controllers (MFC) into a mixing chamber. Water was added to the gas mixture exiting the mixing chamber, by passing a stream of air through a water bubbler that is maintained at constant temperature. The combined gas mixture. was fed to reactor at a flow rate that can be varied from 100 to 3000 ml/min. A tubular glass reactor (typically, 08 x 110 m) was equipped with a insulating jacket for liquid media to be circulated to maintain a stable reaction temperature. The catalyst bed dimensions can be altered through exchanging different size tubular reactors. The SO2 concentration in the gas stream was monitored continuously with an infrared analyzer. . The reactor exit gas was passed through a liquid collector and an ice trap before entering the SO2 analyzer in order to reduce the water vapor pressure to a low and steady level. The liquid products from the reactor were drained into the liquid collector. RESULTS AND DISCUSSIONS Both fiber and particulate activated carbons in their as-received forms exhibited measurable activity for the oxidative removal of SOz from the simulated flue gas, Figure 2. In each case, after a short time on stream, SOz was detected in the emuent gases and increased in concenuation to a steady value. The steady-state removal (SSR) of SO2 is dependent upon the type of activated carbon. The shale oil-derived fibers showed the highest activity, with 60% SO2 removal at steady state. This result is consistent with our earlier findings [12]. The Osaka Gas fibers had much lower activity as observed by the Kyushu University group [7], and were comparable to the performance of some granular carbons. The much higher activity of the shale oil fibers is believed to be related to their high nitrogen content (coal-tar pitch fibers -0.5 wt% versus >2.5 wt% for shale oil products). although the specific role and form of the nitrogens is not understood. It is to be noted that the activity of the granular carbons is significantly increased upon reducing the particle size. This indicates the importance of mass transfer limitations in the reaction process and that these can be reduced by using smaller particle sizes. In practical terms, a catalyst bed consisting of h e panicle activated carbons would give a high pressure drop, especially in the two-phase flow regime where sulfuric acid is draining through the bed. By u carbon fibers in some suitable arrangement (other than loosely packed), the advantages of reducing mass transfer effects could be realized without the attendant penalty in pressure drop. The open pore structure of fiber beds would facilitate fast contact with the reaction surfaces contained in 10 - 20 microns filaments and assist liquid drainage. Figure. 3 shows the effects of prior heat treatment on the catalytic activities of both particulate and fiber activated carbons. Heat treatment has been found to be. effective for improving the catalytic activity of commercial PAN and coal tar pitch-based activated carbon fibers [6,7,8]. At equivalent loading, the activity of the fibers decreased in the order, shale oil >> coal liquids > coal tar pitch (Osaka Gas fiber). As in Figure. 2, the small particle granular carbon is somewhat more active than the Osaka Gas fibers, although at double the loading. A comparison of Figures 2 and 3 shows that the pretreatment procedure greatly increased the steady state activity of the shale oil fibers, from 60 % to about 90 % SO2 removal. It can be seen that at high loading, 100% steady state removal was achieved with heat-treated shale oil fibers and this activity was maintained for at least 72 hours. In mnnast, the extent of activity improvement is smaller for Osaka Gas fibers and the small particle BPL carbons. Table 2 summarizes the typical reaction conditions and parameters of the catalyst beds for the two forms of activated carbons: shale oil fibers and small particle granules. Because of the low density of the fibers, only half the weight of the particulate activated carbons can be packed in a similar volume. With areactant gas flow rate at 200 drnin, space velocities of 10380 and 9180 h.1 were obtained for the activated carbon fiber and particle beds, respectively. Under these conditions, about 90 % SO2 removal was achieved, using heat-treated shale oil fibers with a bed depth of only 23 mm (Figure 3). Complete removal of SQ was obtained by the activated fiber bed 46 m deep, with a corresponding space velocity of 5180 hl. In contrast, using a trickle bed reactor with granular activated carbons, a few meters depth would be needed to achieve 95% SQ removal at velocity of from loo0 to 2850 h-I [151. The high rates of mass transfer and reaction over activated carbon fibers would permit the treatment of high SO2 content flue gases and production of more and highconcentration sulfuric acids as by-product SUMMARY The catalytic performance of fibrous and particulate activated carbons obtained from different precursors was investigated for SO, removal at ambient temperature, using a humidified model flue gas. Despite their similar BET surface areas, activated carbon fibers prepared in the laboratory from shale oil and coal liquids were found to exhibit much higher activity than a 1052 COnlmercial activated carbon fiber pToduced from coal tar pitch. This confirmed our early findings. It iS considered that the high nitrogen content of the shale oil fibers is an important connibutor to their high activity. However, the form of nitrogen species, and the nature and the role of the SWace groups are not yet understood. Comparisons between activated carbon fibers and parricdates indicate that the small dimensions (a couple of tens of micron diameters) of the fibers is a key factor to realidng the full catalytic potential for this application, because of high mass msfer resistance in the gas-liquid-solid system. Carbon fiber beds can provide an open pore SUucture through which reactants and products in both gas and liquid phases can flow to reach and to interact with the surfaces of carbon catalysts. REFERENCES 1. Y. Komatsubara, I. Shuaishi, M. Yano and S . Ida, Nenryo Kyokaishi (I. Fuel Society, Jpn), 64, 255 (1985). 2. M. Hamnan and R. Coughlin, "Oxidation of SO, in a trickle bed reactor packed with carbon", Chemical Engineering Science, 27,867(1972). 3. K. Yamamoto. K. Kaneko and M. Seiki. Kogyu Kogaku Kaishi. 74.84(1971). 4. P. M. Haure, R. R. Hudgins and P. L. Silveston, "Periodic operation of a trickle bed reactor". AIChiE Journal, 35(9), 21437(1989). 5. S. K. Gangwal, G. B. Howe, J. J. Spivey, P. L. Silveston, R. R. Hudgins, J. G. Metzinger. "Low-temperature carbon-based process for flue-gas cleanup", Environmental Progress, 12(2), 28(1993). 6. S. Kisamori, S. Kawano and I. Mochida, "Continuous removal of SO, in the model flue gas over PAN-ACF with recovering aqueous H,SO,", Chemistry Letter, 1893(1993). 7. S. Kisamori, K. Kuroda, S . Kawano, I. Mochida, Y. Matsumura and M. Yoshikawa. "Oxidative removal of SO, and recovery of H,SO, over poly(acryloniai1e)-based activated carbon fibers". Energy & Fuel, 8,1337(1994). 8. I. Mochida, S . Kisamori, M. Hironaka, S. Kawano, Y. Matsumura and M. Yoshikawa, "Oxidation of NO into NO, over activated carbon fibers", Energy &Fuel, 8,1341(1994). ' 9. Y.Q. Fei, F. Derbyshire. M. Jagtoyen and G. Kimber. "Synthesis of carbon fibers and activated carbon fibers from coal liquids", Proceedings, Eleven Annual Conference, Pittsburgh Coal Conference, Pittsburgh, PA, September 12-16, p.174, 1994. 10.Y.Q. Fei, F. Derbyshire, M. Jagtoyen and I. Mochida, "Advantages of producing carbon fibers and activated carbon fibers from shale oils", Proc., Eastern Oil Shale Symposium, Lexington, KY, USA, Nov.16-19, 1993, p38. 11. Y.Q. Fei, M. Jagtoyen, F. Derbyshk and I. Mochida, "Activated carbon fibers from petroleum, shale oil and cod liquids", Ext. Abstracts and Program, International Conference on Carbon, Granada, Spain, June 3-8, p.666, 1994. 12. F. Derbyshire, Y.Q. Fei. M. Jagtoyen and I. Mochida, "Activated carbon fibers for gas clean-up", Abstracts, International Workshop: Novel Technology for DeSOx and DeNOx, Fukuoka, Japan, Jan. 13-14,1994. 13. A.R. Mata and J. M. Smith, "Oxidanon of S0,in aickle bed reactor", Chem.Eng. Journal, 22, 229(1981) 14. H. Komiyama and J. M. Smith, "S0,oxidation in slurry of activated carbon", AIChiE Journal, 21, 664(1975) 15. P. L. Silveston and S. K. Gangwal, "SO2 removal in a periodic operated aickle bed", Proceedings, Eleven Annual Conference, Pittsburgh Coal Conference, Pittsburgh, PA, September 12-16. p.797, 1994. 1053 Table 1 Properties of Activated Carbon Catalysts Sample ID AF-SK25 AF-CE AF-010 BC2060 BC6012 Type Size'(pm) B ~ Pre~cursor $ ~ ~ ~ fiber 6 - 16 986 shale oil fiber 8 - 18 101 3 coal liquid fiber 8 - 20 1057 coal tar particle 200 - 600 1048 coal Darticle 600 - 1200 1020 coal ' Diameter for fibers or particles AF-SK25 0.25 8 1 23 1.16 Table 2 Comparison of reaction conditions and SO, removal for activated carbon fibers and particles AC twe I fiber I Darticte 1 Temperature (OC) Space velocity b) (hl) Catalyst bed: AC ID a) Weight (9) Diameter (mm) Depth (mrn) Volume (cm3) 30 30 10380 91 80 BC2060 0.50 8 26 1.31 ISteady state removalb)(%)) 89 1 19 a) Samples heat-treated at 800 OC for 1 h in nitrogen. b) Reactant gas at 200 mVmin: SO, 1000 ppm, 0,5 vol % H,O 10 vol %. N, balance. CsO,/N, o*N2al Nz MFC 3 r 4 i Analyzer i Figure 1 Schematic of reaction system for evaluation of SO, continuous removal at ambient temperature: 1, Mass flow controller; 2, Mixing chamber; 3, Water bubbler; 4, Reactor; 5, Liquid product collector; 6, Ice trap 1054 Removal (“A) Shale 011 fiber (0.25g. AFSK25) 40 n Partlcle (OSg, BC2060) Osaka Gas fiber (0.259, AF-010) 0 6 12 18 24 30 36 Reaction Time (h) Figure 2 Activity of as-received activated carbons for SO, removal at 30 OC; Reactant gas: 200 mllmin, lOOOppm SO,, 5 vol% 02,lO vol% H,O in N, Removal (%) 100 Partlcle (OSg, BC2060) 0 6 12 18 24 30 66 72 Reaction Time (h) Figure 3 Activity of heat-treated activated carbons for SO, removal at 30 OC; Reactant gas: 200 mllrnln, lOOOpprn so,,5 vol% 0,, 10 vol% H,O in N, Heat treatment: 800 OC, 1 h, in nitrogen / 1055 THE EFFECT OF H20 ON THE ACTIVITY OF Cu/ZSMS-BASED CATALYSTS FOR LEAN-NO, REDUCTION Hung-Wen Jen, Cliff Montreuil, and Haren Gandhi Chemical Engineering Department Ford Research Laboratories, Ford Motor Company Mail Drop 3179, 20000 Rotunda Drive Dearborn, MI 48121 Keywords: CU/ZSM~; NO, Reduction; Steam Deactivation. INTRODUCTION The reports on the high activity of Cu/ZSMS catalysts for the reduction of'NO in excess Oz [1,21 have generated great interest in automotive industry. The successful development of catalysts capable of catalyzing the NO,-reduction under lean conditions is a requisite for the application of lean-burn engine technology to production vehicles. The technology offers the potential of enhancing fuel economy and lowering engine-out pollutants [ 3 1 . A practical automotive catalyst has to have sufficient activity and long-term durability over the entire range of operating conditions. In the process of evaluating Cu/ZSMS-based catalysts for lean-NO, reduction in our laboratory, it was found that the activity decreased as the time on-stream increased. Later, the main cause of the deactivation was determined to be H20 (steam). The deactivation has been shown to be accompanied by de-alumination of the zeolite structure using "Al-nmr spectroscopy [41. The deactivation of Cu/ZSMS under conditions of typical vehicle exhaust is well known now, but. there is no report with detailed data representing the process of steam deactivation and comparing the reactivities of fresh and deactivated catalysts under a broad range of temperatures. In this report, the results from our study concerning the effects of steam on the activities of Cu/ZSMS catalysts are presented. The detailed data for the experiments leading to the finding of steam deactivation are included. Also, the activities for fresh and steam-deactivated Cu/ZSM5 catalysts are compared between 300 and 600 OC. The temporary effect of steam poisoning on .the . activities for lean-NO, reduction depended on the catalysts. The variation may be related to the nature of Cu-sites on the Cu/ZSMS catalysts. EXPERIMENTAL The catalysts used in this report were either powder samples or cordierite monoliths washcoated with Cu/ZSMS. Cu/ZSMS materials were prepared by a conventional exchange method using HZSMS or NaZSMS and Cu-acetate. The activity of a catalyst in a flow reactor system was determined by the difference between the inlet and outlet concentrations of a reaction gas. Gas concentrations were monitored using commercial gas analyzers for NO,, HC (total hydrocarbon), CO, and 02. RESULTS AND DISCUSSION In Figure 1, the activity of a monolith catalyst containing Cu/ZSMS was measured versus the time in the exhaust generated from a pulsed flame combustor [SI. In the combustor, isooctane vapor mixed in a flow of air was thermally combusted. Extra oxygen was added into the exhaust to simulate lean-burn engine exhaust. The NO,-conversion decreased with the on-stream time. One hour on stream was comparable to 30 miles of vehicle operation. The durability of the Cu/ZSMS-based catalyst. in the exhaust of combusted isooctane was not good. There are several possible sources that can deactivate a catalyst in the automobile exhaust. The results in Figur,e 2 were obtained to determine the effect of SOZ. The NO,-conversions for two identical monolith catalysts were measured. One catalyst was exposed to a synthetic gas mixture with 20 ppm SO2, while the other one was exposed to the same gas mixture but without S02. The NO,-conversion for either catalyst decreased with time. The 1056 two curves of NO,-conversion versus time were superimposable. The comparison in Figure 2 clearly shows that SO2 is not the cause Of the observed deactivation. Figure 3 shows the activities of three identical catalysts versus aging time. The aging process was simply the heating of a catalyst in a flow of air. After a certain period of aging, the catalyst was moved to a flow reactor system and the activity was measured using a dry mixture of reaction gases. Two catalysts were aged at 480 OC, one in dry air and the other in wet air Containing 10% H20. The conversion of NO,, HC or CO remained constant for 300 hours over the catalyst aged in d,'y air. The Conversion for the catalyst aged in wet air at 480 C decreased with aging time. The sole difference between the constant and decreasing activities was the existence of 10% H20 in the aging media (air). Clearly, the heating in the presence of H20-steam caused the deactivation of Cu/ZSM5 catalysts. The decrease in the activity for the catalyst aged at 380 'C in wet air was also detectable, even though the rate of decrease was smaller than that aged at 480 OC. In order to compare the activities in a broad range of temperatures, the NO,-conversion for a fresh CuNaZSM5 catalyst was measured in a temperature-programed-cooling process from 600 to 300 "C at 12 'C/min(Figure 4). The addition of 9% H20 into the reaction mixture caused a significant decrease in the NO,- conversion. The activity generally could be regained when the 9% H20 was turned off, if the exposure to the steam was not long and the temperature was not very high. The same experiment was done for the same catalyst which had been aged in 20% H20 (figure 5). The NO,-conversion for the aged catalyst was lower than that for the fresh catalyst as expected. However, the addition of 9% H20 had little effect on the NO,-conversion of the aged catalyst. The result indicated that the part of the activity vulnerable to the temporary poisoning of the steam present in the reaction mixture was the first lost to the long term steam-deactivation. The phenomenon may be related to the existence of different Cu-sites on Cu/ZSM5 catalysts. CONCLUSION Cu/ZSM5-based catalysts for lean-NO, reduction deactivated after long term exposure to the simulated exhaust gas mixture. The cause of deactivation is the exposure at high temperature to steam that is always present in vehicle exhaust. For the fresh CuNaZSM5 catalyst, HzO had a temporary poisoning effect on the NO, conversion. For the steam-aged CuNaZSM5 catalyst, the poisoning effect of H2O on the NO,-conversion was not noticeable. ACKNOWLEDGMENT The CuNaZSM5 samples were kindly provided by Carolyn Hubbard and Mordecai Shelef. REFERENCES (1) Held, W,Konig, A., Richter, T. and Puppe, L., SAE Paper (2) Sato, S., Yu-U, Y., Yahiro, H., Mizuno, N. and Iwamoto, M., ( 3 ) "Automotive Fuel Economy: How far Should We Go?", National 900496 (1990). Appl. Catal. 70, L1 (1991). Research Council, National Academic Press, Washinton, D. C., 1992, pp 217-226. (4) Grinsted, R. A., Jen, H. W., Montreuil, C. N., Rokosz, M. 3. and Shelef, M., Zeolites, 13, 602 (1993). (5) Otto, K., Dalla Betta, R. A. and Yao, H. C., J. Air Pollution Control Association, 2, 596 (1974). 1057 ~ MILES (Thousands) Figure 1. Activity of CdZSM5-Containing Monolith aged in Pusled Flanie Combustor SiOdAI101=32, 2.41~1%C u on CdZSM5, SV=30,000 1:r.I. T=482 "C HOURS Figure 2. ENect of Aging in SO1 for CdZSM5-coi:taining Monolith in Flow Reactor SiOl/Al1O1 = 32, 2.41~~1%Cu on CdZSM5, SV = 50,000 hi', T = 482 "C 3.45%02, 1517ppniC1H6, 756ppn1CIHB.490ppn:N0,0.3%C0,0.1%H,, 12% CO1, 10% H20,N 2b alance 100 BO - s 60 - c 0 7 40 c ._ 2 6 20 0 Rx. Condilion: 50.000 hr-1 T-482T 500 ppm NO 1600 PP C3Hg 800 Ppm c3Hg 0 25% CO Aging Time (Hour) Figure 3. Erect of Aging in H1O for CdZSMS-conlaining Monolith in Flow Reacfor SiO21AI20, = 32. 2.41~1% Cu on CdZSM5 1058 80 ... s '. - 60 0 0" 40 .p 20 0 1 I 600 550 500 450 400 350 300 Temperature. C Figure 4. EWcl ofH20 in Rcacrion Mislure on Nosonversion for Fresh CuNaZSM5 Si02/A120, = 46. 2.8 $VI% Cu, 0.15 g sample 5% 02,11 20 ppni C3 (C,HdC,He = 2). 55Oppm NO, 0.5 lliiiin Flow GOO 550 500 450 400 350 300 Temperature. C Figure 5. ElTcct of H20 in Reaction Mislurc on NOsonvcrsion for Steam-aged CuNaZSMS SiO2/AI2O, = 46, 2.8 \VIVO Cu, 0.15 g s3mplc 5% 021,1 20 ppm C3 (C3HJC,Hs = 2). S50ppin NO, 0.5 Vmin Flow 1059 ON THE MECHANISM OF NO DECOMPOSITION AND SELECTNE CATALYTIC REDUCTION BY HYDROCARBONS OVER CU-ZSM-5 In our s t u d y o f d i r e c t NO decomposition[4,5], we observed that the ls+4p electronic transition of CuQ) in Cu- ZSM-5 appears as a narrow, intense peak which is an effective measure of changes in the population of copper oxidation states. This transition is quite intense after Cu- ZSM-5 is activated in inert gas flow. The number of oxygen atoms surrounding the copper ions also drops from 4 to 2 during the auto-reduction. After the admission of transition intensity decreases but by no a NOM, gas mixture, the CuQ) Is+4p Di-Jia Liu AlliedSignal Research and Technology 50 E. Algonquin Road, Des Plaines. EL 60017-5016 1 I 1- a IP-wsrr, .-a- 1 Keywords: Mechanism, NO decomposition, NO selective catalytic reduction by hydrocarbons The initial reports on the catalytic activity of CU-ZSM-5 during NO decomposition and selective catalytic reduction (SCR) by hydrocarbons[ 1-31 have generated a lot of excitement and the followup research on this catalyst in recent years. Although the lack of hydrothermal aging stability may prohibit its practical application, Cu-ZSM-5 provides an excellent system for studying the mechanism and the structure-function relationship of the zeolite based NOx reduction catalysts. Reported here are our most recent analysis of the data obtained from the investigations of the catalytic mechanisms of NO removal over a Cu-ZSM-5 catalyst using the in situ X-ray absorption spectroscopy method. Two mechanism were studied and compared, they are a) direct NO catalytic decomposition and b) NO SCR by hydrocarbons in M oxygen-rich gas mixture. The difference and similarity between the two mechanisms were found through the analysis of cuprous and cupric ion transition energy shifts, the changes of local coordination structure, the influence of cuprous ion formation/catalytic activities by Cu exchange level and the type ofhydrocarbon used in the catalytic reactions. The results are summarized in Table I. I the normalized cuprous ion concentration and found it correlates well with the NO decomposition rate from 300 to 500 "C. Shown in Fig. 1. This finding supports the conjecture that CuQ) participates in a redox mechanism during catalyzed NO decomposition in Cu-ZSM-5 at elevated temperature. The active site is a two-oxygen coordinated cuprous ion. In our study of the SCR of NO by hydrocarbons[6], we observed that, even under fraction of copper ion in ZSM-5 Can be reduced in different gas mixmcs with NO conversion level. to Cu(1) at elevated temperature. The rate of 1060 I I formation of Cup) is less sensitive to the exchange level than to the type of hydrocarbons used. The Power of reducing Cu(II) to Cup) follows the sequence, C,Hp2,H,>CH4, with methane practically equals to zero. The similar Cu Is-Mp transition was observed although the peak energy shifted at different reaction temperature, indicatingthe formation of the Cup)-organic ligands possibly allylic species duringthe catalysis. XANES spectra showthat the Cum Is+4ptransition intensity changes with the reaction temperature in a similar pattern as the NO conversion activity (solid line in Fig. 2, obtained from Ref. 7) in the NO/C,HJO, mixture. Shown in Fig. 2. For comparison purposes, we also studied the Cum concentration change in a similar gas mixture where propene is stoichiometrically replaced by methane or propane. Unlike propene, methane shows no selectivity for NO reduction over CU-ZSM-5. We did not observe any window of Cup) enhancement. Propane is a selective reducing agent and we indeed observed awindow of Cup) enhancement although the intensity is much weaker than that observed with propene. Our study indicates that, even in a strongly oxidizing environment, cupric ion can be partially reduced by propene or propane to form a Cu(1) which is a crucial step for effective NO conversion through a redox mechanism. Table I. The difference and similarity in oxidation state, coordination structure and reaction mechanism between NO decomDosition and NO SCR bv hydrocarbons over CU-ZSM-5 NO Decomposition The cuprous ion. Cu(I), is observed during the direct NO catalytic decomposition, suggesting a redox mechanism in which the catalyst's active site is Cu(I). Cup) is formed through the auto-reduction at elevated temperature which involves a dicopper process. Cu(I) formation is sensitive to the Cu exchange level and the "excessively" exchanged Cu-ZSM-5 maintains higher concentration of Cup) than that of "underexchanged under the reaction conditions. The ls+4p electronic transition of Cup) does not shift its energy at different reaction temperature, indicating that no significant variation occurs to Cu(1) coordination environment The cuprous ion formed through autoreduction is coordinated by two oxygen atoms. No clear higher shell structure is observed. Under direct NO decomposition, copper ions consist of the mixture of Cu(I) and Cu(II), Cum) is coordinated by four oxygen atoms. Cup) concentration increases with the reaction temperature, and is correlated with the NO decomposition rate from 300 to 500OC. Discrepancy is observed at 600 "C. Cup) concentration decreases sensitively with the increase of the oxygen concentration i-n the gas phase. NO SCR by Hydrocarbons The cuprous ion is also observed during the NO selective catalytic reduction by hydrocarbons, suggesting a redox mechanism which involves the conversion between Cu@) and Cu(1). Cu(1) is formed through the reduction by hydrocarbons. The rate of formation of Cu(1) is not sensitive to exchange level, rather it is very sensitive to the type of hydrocarbons used. The reducing power is C&> C,H,>CH, with methane practically equals to zero. The ls-Mp transition energy is shifted at different reaction temperature in the NO/C,HJO, mixture, indicating that the local coordination of Cu(1) varies with the reaction conditions. No energy shift is observed in propane or methane mixture. Cu(I) formed by olefine (propene) reduction is also likely to be coordinated by two oxygen atoms, with a possible Cu allylic bond which has not been identified unambiguously. No evidence of the allylic compound formation was observed for propane and methane mixtures. ~~ Cup) concentration as the function of the reaction temperature depends on the gas compositions. For SCR by propene, normalized Cup) intensity at various temperature appears to overlap with the normalized reaction rate versus the temperature. Cue) concentration also decreases with the increase of the oxygen concentration, but with less degree of sensitivity. 1061 References: 1. M. Iwamoto , H. Yahiro, Y. Mine, S. Kagawa, Chem. Lett., (1989) 213. 2. M. Iwamoto, H. Yahiro, S. Shundo, Y. Yu-u, N. Mizuno, Appl. Catal. 69 (1991) L15-Ll9. 3. W. Held, A. KBnig, T. Richter, L. Puppe, SAE paper 900496. 4. Di-Jia Liu and Heinz J. Robota, ACS Symposium Book Series. No. 587, Reduction of Nitrogen 5. Di-Jia Liu and Heinz J. Robota, Catal. Lett. 1291 (1993). 6. Di-Jia Liu and Heinz J. Robota, Applied Catalysis B: (Environmental) 4, 155, (1994). 7. M. Iwamoto, N. Mizuno, and H. Yahiro, Sekiyu Gakkaishi, 34,375, (1991). Oxide Emissions (U.S. Ozkan, S.K. Aganval. and G. Marcelin, eds.), Chapter 12, p 147. 1062 HYDROCARBON SPECIFICITY OVER CU/ZSM-5 AND CO/ZSM-5 CATALYSTS IN THE SCR OF NO T. Beutel, B. Adelman, G.-D. Lei and W.M.H. Sachtler V.N. Ipatieff Laboratory Northwestern University Evanston, IL 60208-3000 Keywords: CdZSM-5, Co/ZSM-5, NO, reduction, Adsorbed NO,, H-Abstraction 1. Introduction A large variety of catalysts has been proven to be active in the selective catalytic reduction of NO by hydrocarbons. Although 02,gaactss as a nonselective competitor for the direct combustion of hydrocarbons, the addition of O2 enhances the rate of NO reduction'. This enhancement has been attributed to the oxidation of NO which leads not to NO,, gas but rather to adsorbed nitrogen oxide complexes (NO, groups). Although the reactivity of these NO, groups has not been fully investigated, there are literature data to suggest that the hydrocarbon must first be activated. Cant and coworkers2 observed a first order isotope effect when CH, and CD, were used as reductants. The authors concluded that H-abstraction was the rate limiting step for both N2 and C02 formation. In general, the chemistry for the selective reduction of NO by hydrocarbons may be comparable to the chemistry of a cold flame3. For these reactions, H-abstraction is the first step in hydrocarbon activation. It is therefore plausible that the NO, groups are the sites responsible for the H-abstraction reaction4. The role of NO, groups on CdZSM-5 and Co/ZSM-5 has been investigated by FTIR spectroscopy to determine their thermal stability and reactivity towards C3Hs and CH,. The nature of the evolved gases has been analyzed in separate experiments by mass spectroscopy. 2. Experimental 2.1. Catalyst preparation CdZSM-5 and Co/ZSM-5 catalysts were prepared via ion exchange at room temperature (r.t.) using a Cu(OAc), or CO(NO~s)o~lu tion with NdZSM-5 (UOP lot #13023-60). Elemental analysis via inductively coupled plasma spectroscopy gave the following data: CdAI = 0.56, SUA1 = 18, NdAl = 0.0; Co/A1= 0.48, SUA1 = 18, Na/A1=0.34. Prior to IR or MS experiments the samples were calcined for 2 hrs at 500°C in an UHP 0 2 flow. 2.2. FTIR spectroscopy Spectra were collected on a Nicolet 60SX FTIR spectrometer equipped with a liquid N2 cooled detector. The samples were pressed into self-supporting wafers and mounted into a Pyrex glass cell sealed with NaCl windows. Spectra were taken at r.t. accumulating 50 scans at a spectral resolution of lcm-'. The samples could be pretreated in situ in a gas 1063 flow at temperatures up to 500°C in a heating zone attached to the glass cell. After in situ calcination in UHP 02, as described previously, the sample was purged at r.t. for 1 hr with 25 ml min-l UHP He then saturated in a stream of NO (0.45%) and 0 2 (75%) with a He balance. For the reduction studies the samples were heated to the reaction temperatures at 6"/min in flowing C3H8 or CH4 (0.25% hydrocarbon in He) at a total flow rate of 30 ml min-I. Before cooling to r.t. the sample was purged for 10 min with He. Spectra were taken at r.t. 2.2. IMS analysis For the analysis of released gases, 400 mg of sample were calcined ex situ to 500°C in UHP O2 and then saturated with NO2 (OS%, balance He) at r.t. The reactor was transferred to a glass, recirculating manifold equipped with a Dycor Quadrupole Gas Analyzer. Prior to the reduction experiments the sample was heated in vacuo to 225°C for CdZSM-5 and 150°C for Co/ZSM-5. A sample loop was then filled with a known amount of hydrocarbon; evolved gases were allowed to recirculate over the sample. The signal intensities were normalized by an Ar standard. A secondary loop to the manifold was charged with 3 g of 5 wt.% Ni/Si02 pre-reduced at 400°C. This loop was sealed from the reactor and manifold during the experiment and was used to remove CO from the post-reaction analysis of the evolved gases. 3. Results 3.1. FTIR spectroscopy Fig. 1A shows the FTIR spectra of CdZSM-5 after the exposure to NO + O2 at r.t. and subsequent purge at 200°C in He. There are three distinct bands at 1628, 1594 and 1572 cm" which are attributed to Cu" bonded nitro and nitrate groups. These NO, groups are stable in He at 200°C for over 14 hrs. However in C3H8 all band intensities decrease. A plot of the band intensities, measured as peak heights and normalized by their initial intensities, is presented in FiglA'. The rates of reaction of the three NO, groups are different. One of the nitrate groups (1594 cm-I) reacts fast, whereas the other nitrate group (1572 cm-') reacts sluggishly. The reactivity of the nitro group (1628 cm-l) exhibits an induction period of 20 min after which it is consumed at a comparable rate to the nitrate group at 1594 cm-l. In CH,, the CumNO, groups are not depleted at temperatures below the thermal decomposition. In the case of Co/ZSM-5 the main feature after NO + O2 saturation is shown in Fig 1B. It consists of two broad bands at 1526 and 1310 cm-I. The former band is ascribed to a Co2+*ON0 complex. The Co*NO, adsorption complex is less stable than the CueNO,. Approximately 60% of the Co 0 NO, adsorbates are desorbed after thermal treatment at 150°C for 14 hrs. The reactivity of the remaining NO, groups with C3H8 at 150°C is shown in Fig.lB'. The normalized intensities of the adsorption band at 1526 cm-' are plotted in Fig .1B' for propane and methane. Unlike Cu*NO,, CoeNO, reacts with CH4. 3.2. MS spectroscopy Fig.2 shows the evolution of N2 when CdZMS-5 or Co/ZSM-5 samples, pre-saturated with NO2, are exposed to C3H8 or CH, at reaction temperatures of 225°C for CdZSM-5 1064 I and of 150°C for Co/ZSM-5. When C3H8 is used as the reductant, N2 evolution from CdZSM-5 is rapid but terminates after 30 min exposure to hydrocarbon. N2 evolution from Co/ZSM-5 proceeds at a slower rate; an increase in N2 is still detected after 90 min exposure to hydrocarbon. When CH, is used as the reductant no reaction occurs over CdZSM-5, but over Co/ZSM-5 N2 evolution is detected. Co *NO, reaction with CH4 is slower than Co NO, reaction with C3H8. 1 4. Discussion I NO, complexes are formed on CdZSM-5 and Co/ZSM-5 after saturation with NO2. The IR spectroscopic signature, thermal stability and chemical reactivity of Cu- and Cobonded NO are found to be different. CdZSM-5 contains not only Cu2+ ions, but also [CU-O-CU]~o' xocations and CuO oxides. Upon interaction with NO2 Cu2' ions form nitro complexes while oxocations and oxide react to nitrate complexes. On the other hand, Co/ZSM-5, which contains only Co2+ ions, can only form NO2 complexes. Unlike Cu2+ NO2, these are most likely Co2+* ON0 nitrito complexes. Although deNO, catalysis over both Co/ZSM-5 and CdZSM-5 may be initiated in the same manner, H-abstraction, the two display a different hydrocarbon specificity; CdZSM-5 requires C2+ olefins or C,+ paraffins, whereas Co/ZSM-5 is active with CH, and higher hydrocarbons. The type of the NO, groups differs which may explain the differences in hydrocarbon specificity . Assuming that the activation of hydrocarbon occurs via an H-abstraction as stated by others3,,, this reaction is affected by NO, groups. While exposure to C3Hs leads to N2 formation from both samples, only Co/ZSM-5 formed N2 upon CH, exposure. It appears that H-abstraction from CH, is difficult with Cu*NO, but facile with Co*NO,. The influence of the metal ion on the selectivity in NO reduction may be indirect by furnishing different types of NO,, The fate of the hydrocarbon radical is not yet clear. It has been proposed that a reactive intermediate containing at least one carbon, nitrogen and oxygen atom is formed on the catalyst surface which reacts further with NO to from N,. The role of NO,, and the nature of the reactive intermediate are currently under investigation. 5. Acknowledgments We would like to thank the following for grant in aid: V.N. Ipatieff Fund, Ford Motor Corporation and Engelhard Corporation. T. Beutel thanks for a stipend from the Deutsche Forschungsgemeinschaft. M. Iwamoto, Proc. of the Meeting of Catalysis Technology for the Removal of Nitrogen A.D. Cowan, R. Dumpelmann and N. W. Cant, J. Catal., 151 (1995) 356. F. Witzel, G.A. Sill and W.K. Hall, J. Catal., 149 (1994) 229. y. Li, T.L. Slager and J.N. Armor, J. Catal., 150 (1994) 388. Monoxide, Tokyo, Japan (1 990) 17. J 1065 1 .2 CdZSM-5-113 A' 1628 A I B 15 94 I L 1526 I A 13 10 I Wavenumber. an-1 0 20 40 60 80 100 Time, min 1.2 €3' i d 1 . 7 0.2 0 100 200 Time, min Fig. : FTIR spectra of CdZSM-5 (A) and Co/ZSM-5 (B) after calcination, exposure to NO + O2 at r.t. and He purge at 200°C (A) and 150 "C (B). Graph A', the relative intensities of NO, bands at 1628 cm-' (a), 1594 cm" (b) and 1572 cm-' (c) in 0.25 % propane at 200 "C vs. time. Graph B'. the relative intensities of NO, band at 1526 cm-' in 0.25 YO methane (d) and 0.25 % propane (e) at 150 "C vs. time. 1066 0.9 A 0.8 - 0.7 zcn 0.6 3 0.5 2E 0.4 Q, 0.3 u) 0.2 -2 m C CI , u Q tn u) : 0.1 a 0 0 500 1000 1500 2000 2500 3000 3500 Time in seconds Fig.2: N2 evolution from Co/ZSM-5 at 15OOC (b, c) and CdZSM-5 at 225°C (a, d) upon interaction with CH, (c. d) and with propane (a, b) vs. time. Samples have been calcined, saturated with NO2 at r. t. and outgassed at the respective reaction temperature prior to reaction. 1067 DEACTIVATION OF PT-ZSM-5 FOR SELECTIVE REDUCTION OF No K. C. C. Kharas', H. J. Robota', D.4. Liub, and A. K. Datyec 'AlliedSignal Environmental Catalysts, P.O. Box 580970, Tulsa, OK 741585-0970, USA bAlliedSignal Research and Technology, 50 E. Algonquin Road, Des Plaines, IL, 60017-5016, USA Qepartment of Chemical and Nuclear Engineering, The University of New Mexico, Albuquerque, NM, 87131-1341, USA Keywords: NOx reduction, Pt catalysts, catalyst deactivation INTRODUCTION Recent reports suggest the use of Pt-ZSM-5 with hydrocarbons to reduce NO selectively under oxidizing conditions (1,2) and our laboratories, among others (3.4) are investigating the use of Ptzeolites to reduce NOx in the emission of Diesel or gasoline lean-bum vehicles. Here we consider the activity ofPt-ZSM-5, both as a fresh catalyst and der deactivation, and characterize a troubling aspect of catalyst deactivation. In severely deactivated materials, TEM reveals a film to have formed over Pt metal; we suggest this film is siliceous material derived from the zeolite and that an important mode of catalyst deactivation is due to geometric site blockage by this film. EXPERWENTAL R-ZSMJ catalysts, containing 0.50 wt%, 1.31 wt%, 2.52 wt%, 4.4 wi%, and 4.67 W?? F't, were prepared using H-ZSM-5 supplied by PQ Corporation containing a SdAl ratio of 30.5, tested using two test gases, and aged in the model gas mixture for one to filly hours at 700 "C or 800 "C. One gas mixture included 700 ppmv NO, 3300 ppmv propene, 1000 ppmv CO, 330 ppmv H2, 7.5% 02, 20 ppmv S a , 10% H20, 10% COZ. The other gas mixture was similar, consisting of 700 ppmv NO, 700 ppmv propene, 300 ppmv CO, no Hz, 20 ppmv SOz, 7.5% 0,1oo/O C a , and 10% HzO. The automated gas delivery and data acquisition system used chemiluminescent NOx, NDIR CO and N20. FID hydrocarbon, and paramagnetic 0 2 detecton to monitor catalyst activity. Catalyst performance is conventionally reported by % converted except for N20 formation by NO reductio& which is more informative to express as % NO reduced to N20. One gram of 20-40 mesh granules were typically tested with GHSV of 110,000 hi'. Inclusion or omission of 20 ppm S a did not dect performance. EXAFS and XANES were obtained using an in sihr reactor described elsewhere (5) at Beamline X-18B of the National Synchrotron Light Source at Brookhaven National Laboratory. X-ray diffraction intensity data was obtained by standard procedures. TEh4 images were obtained using a JEOL 4000EX microscope with 1.8 A resolution. RESULTS AND DISCUSSION Pt-ZSM-5 deactivates rapidly. We examined the catalyst with highest initial performance most extensively. Catalysts were subjected to catalysis at 700 "C and 800 "C using either synthetic gas blend for up to 50 hr with temperature ramps interspered to allow the monitoring of dactivation. Figure 1 shows NO reduction performance under three conditions. Curves labeled 700°C and 800°C show performance after aging the catalyst at those temperatures in the gas blend that included 3300 ppmv propene while the cuwe labeled 700 ppmv involved aging the catalyst at 700°C using 700 ppmv propene. When fresh or mildly aged, the catalyst reduces NO over a greater temperature range when 700 ppmv propene are present compared with 3300 ppmv. When 3300 ppmv propene is present, an exotherm of over 200°C occurs after propene lightof. The magnitude ofthis exotherm is su5cient to close the temperature window when hydrocarbon levels are high (6). Figures 1 and 2 show deactivation is more rapid at 800°C than at 700 "C. Deactivation proceeds more quickly when 3300 ppmv is used in the synthetic gas compared with 700 ppmv propene. For aging times greater than 20 hr, essentially no NO was reduced when 3300 ppmv propene was used. Figure 2 shows deactivation for propene oxidation for experiments utilizing 3300 ppmv propene. Progressive deactivation occurs for both aging temperatures. The higher temperature aging is clearly much more severe. For example, thirty hr aging at 800 "C is more severe than 50 hr aging at 700 "C. Deactivation is most rapid initially. Judging on the basis of Tm increases, Figure 2 , 1068 t 8 8OO0C 0 x ° C - ~O_Opp_m_v Inlet Temperature. "C 100 200 300 400 M O 600 100 200 300 400 500 600 l l o l I , I I , t 9 0 I so 10 -in -100 200 360 400 500 600 100 200 300 400 500 600 Inlet Temperature. O C Figure 1. NO reduction, measured as conversion ofNOx, deteriorates as aging proceeds. Higher temperature or higher propene concentrations accelerate deactivation. mesh 0 20 hr 50hr- . L O h r - 1-A -3ohr_ ... x 40 hr Inlet Temperature, OC 100 200 300 400 500 600 110 90 70 50 30 10 -10 I 110 90 70 50 30 10 -10 100 200 300 400 200 600 Inlet Temperature, C Figure 2. HC conversion progressively deactivates; 800 "C aging is much worse than 700 "C aping. Tests used 3300 ppmv propene. 1069 shows performance losses during the first hour to be comparable to the next nine. The rate of HC deactivation slows to a nearly constant amount during the 700 "C aging. During 700 "C aging, the first 20 hr and the next 30 hr of aging resulted in Ts increases of about 100 "C for propene oxidation. At first glance, rates of HC deactivation do not appear to decline during the more severe 800 "C aging, However, an anomalously large incremental deactivation may have occurred between 30 and 40 hr during the high temperature aging. Comparable incremental deactivations occur between 10 and 20 hr, 20 and 30 hr, and 40 and 50 hr. These incremental deactivations at 800 "C are considerably larger than those observed at 700 "C. consistent with more severe, progressive deactivation at the higher temperature. We now proceed to physical characterization to gain insight into deactivation prior to returning to the catalytic results. TEM examination of 6esh 4.7 wt% Pt-ZSM-5 reveals "stringy" regions, sometimes over 100 A in length, of Pt. EXAFS and XANES analysis shows the Pt in 6esh 4.7 wt% F't-ZSM-5 to be oxidized. Metallic Pt is not detected by X-ray diffraction in the fresh material but is detected by XRD, EXAFS, and XANES after aging. TEM analysis of catalysts aged at 800 "C for 1 hr or 50 hr also reveals Pt metal. After 1 hr, 800°C catalysis, faceted Pt particles occur, although many smaller Pt particles have poorly defined surfaces. Some particles are as large as 500 A in diameter while particles about 100 A appear most common. For example, one sector of a typical TEM micrograph contained 23 particles with a median diameter of 120 & mean diameter of 180 4 with a range of diameters of 70 - 535 k Catalysts aged for 50 hr do not appear to be substantially more highly sintered. For example, one sector of a typical TEM micrograph of a sample aged for 50 hr contained 55 particles with a mean diameter of 100 4 a median diameter of 70 4 with a range of sizes 60m 40 - 400 k While Pt Sintering may be the cause of the initial, rapid deactivation, the data do not support the hypothesis that the progressive deactivation we observe is due to continued sintering of Pt. While observable particles do not appear to be increasing in size, the TEM results do not exclude the pos particles are increasing in mass. Small, unobservable, catalytically relevant Pt particles could persist as the aging proceeds, and their gradual loss may not cause noticeable increases in observable Pt particle sizes. Nevertheless, their gradual loss may be a cause of progressive deactivation if these postulated, unobservable entities are indeed catalytically relevant. CO dispersion measurements fail to provide evidence for the existance of the postulated small particles; CO isotherms of the aged materials are consistent with large particles whose surfaces, to a substantial degree, are inaccessible to the probe molecule. Faceted Pt particles observed after 50 hr, 800°C catalysis frequently exhibit features consistent with a superficial film on their surfaces while these features are absent in the 1 hr samples. We suggest this film is due to silica or silica-alumina derived from the zeolite itself. This suggestion will be buttressed by a catalytic experiment discussed below. A surrendipitous E M experiment performed on a Pt particle on the edge of a zeolite particle in a fresh sample provide corroborating evidence. This initially featureless Pt particle was is situated over a hole in the holey carbon TEM grid and, once noticed. was subjected to close-up examination using a more focused beam. Figure 3a and b show this particle as initially observed @ut the image is enlarged photographically) and during high magnification E M im aging (without photographic enlargement). Fringes consistent with (1 11) Pt domains are observed in Figure 3b while fringes due to crystalline zeolite (not observable in the small regions shown) vanish during focused beam exposure. A film of zeolite can be observed in the "northern" region of Pt in Figure 3b. The focused electron beam probably caused local heating that drove that region of the sample toward equilibrium: thermal reduction of oxidized Pt to Pt metal proceeded together with local amorphization ofthe zeolite. The zeolitic material spreading onto the Pt surface suggests that, as the catalyst approaches equilibrium at high temperature, siliceous films derived 60m the zeolite may cover otherwise active Pt metal. Films in severely deactivated materials manifest themselves as a white line, followed by a dark lie, then followed by another white line at the edge of a Pt particle. Each solid-vacuum interface, under conditions of slight underfocus, should result in one white line. Pairs of parallel white lines are strong evidence of film formation. Such pairs of white lines are common at the Pt/vacuwn surfaces in samples aged 50 hr at 800°C (for example, Fig. 3d) but only a single white lines are observed after 1 hr, 800°C aging (for example, Fig. 3c). Powell and Whittington used SEM to demonstrate Pt encapsulation by silica at temperatures of 1070 50 nm Fig 3. (a) and @) Pt particle and nearby zeolite in the fresh catalyst prior to (a) and after @) beam damage. (c) F’t metal crystallite observed in sample aged 1 hr. (d) Pt metal crystallite, covered by superficial film, observed in sample aged SO hr. 702°C-11020C (7,8). Theu model describing driving forces toward encapsulation appears to account for some of the phenomena we observe. Their model predicts encapsulated Pt particles will not tend to increase in size, as is observed here. Progressive catalytic deactivation should accompany progressive encapsulation. Figure 2 shows an increasing AT between the onset of propene conversion and the attainment of 25% conversion as deactivation p r d s . This steady increase in AT appears consistent with continually decreasing numbers of sites, just as an encapsulation model would predict. Successll development of durable Pt-moleah sieve catalysts requires additional understanding of this mode of deactivation that leads to effeaive countermeasures. An alternate explanation ofthe films observed in the TEM invokes carbonaceous deposits formed by gradual coking of the catalyst. oxidizing treatment might be expected to remove any such materid. We aged the catalyst 50 hr in the gas blend containing only. 700 ppmv propene. The rate of deactivation did decline, but since less propene is oxidized, temperatures at and near pt might be c 1071 considerably lower than in experiments utilizing 3300 ppmv propene. After the @I& the was treated in 15.00A 02/85.0% N2 for one hour at 550°C. Ifthe films were due to coking ofF't, this treatment should result in at least partial combustion of this film, and a concomittant increase. in available F't. A subsequent temperature ramp in the synthetic exhaust gas (700 ppmv propene) did not result in an increase in NO reduction. This is consistent with films arising from the parent zeolitic material, and not from the catalytic process. Efforts at further characterization of these films, such as TEM halography, are ongoing. CONCLUSIONS Fresh Pt-ZSM-5 shows reproducibly high degrees of NOx reduction in a narrow temperaturr window at very low temperatures. Initial performance is a function of Pt content: the more the better. NO selectivity toward N20 is about 10%. Initially, Pt in the zeolite is oxidized but, apparently, not crystalline. A fraction of the F't may occur as cations at exchange sites but relatively large (100 A) F't-containing particles occur. After one hour of lean-bum catalysis, the F't is unambiguously metallic. The EM-observable fraction ofPt may increase after reduction. F't-ZSM- 5 is not durable. After about 30 hr catalysis at AFR 22 and 800°C. no NOx reduction pafonnance remains. NOx reduction is barely detectable &. 2% ma.) after 40 hr catalytic aging at AFR 22 and 700 "C. Pt sinters; faceted 40-700 A particles can be observed by EM after 50 hr at 800 "C. al deactivation may be associated with moderate sintering of F't, the unusual progressive deactivation is accompanied by formation of films on the F't crystallites. We suggest these films are siliceous material derived from the zeolite itself and that the progressive deactivation is due to simple geometric site blockage. Acknowledgements High resolution E M was performed at the High Temperature Materials Laboratory (HTh4L.) at Oak Ridge National Laboratory. AD acknowledges receipt of a faculty fellowship at HTML during which this TEM work was performed. Mike Reddig performed the CO chemisorption measurements. Brad Hall provided painstaking photographic services. References 1. H. Muraki, T. Inoue, K. Oishi and K. Katoh, European patent application number 91 120322.2, November 27, 1991. 2. K. Ishibashi, N. Matsumoto, K. Sekizawa and S. Kasahara, Japanese patent application, Kokai Patent PublicationNo. 187244 - 1992, July 3, 1992. 3. B.H. Engler, J. Leyrer, E.S. Lox, K. Ostgathe, SAE 930735. 4. M. Iwamoto, H. Yahiro. H. K. Shin, M. Watanabe, J. Guo, M. KOMO, T. Chikahisa, T. Murayama, ADDI. Catal. B: Environmental, 5, (1994), Ll-L5. 5. D.-J. Liu and H.J. Robota, Catalvsis Letters, 21, (1993) 291-301. 6. K.C.C. Kharas, J.R. Theis, "Performance Demonstration of a Precious Metal Lean NOx Catalyst inNativeDiesel Exhaust", SAE 950751, 1995. 7. A New Mechanism of Catalyst Deactivation", J. Molec. Catal. 20 (1983) 297-298. 8. A New Mechanism of Cdya D ea,ct.ivJation", 81 (1983) 382-393. J B.R Powell and S.E. Whittington, "Encapsulation: B.R Powell and S.E. Whittington, "Encapsulation: 1012 LEAN NO, REDUCTION OVER Auly-Al,O, M.C. Kung, J.-H. Lee, 1. Brooks and H.H. Kung Center for Catalysis and Surface Science Northwestern University, Evanston, 11. 60208. Key words:Au/y-Al,O,, deposition-precipitation, lean NO, Reduction Introduction The 1990 Clean Air Act Amendment has set a schedule for compliance of new, more saingent standards for automohiles over the next ten years. In the mean time, the strong push to increase fuel economy of vehicles has led to the exploration of the use of lean-hum, gasoline engines. Unlike conventional engines, these engines operate with a large excess of air. The major ohstacle in the development of such engines is the lack of a practical exhaust catalyst for the reduction of NO, emission, since the current three-way catalysts are ineffective for NO, removal in the oxidizing atmosphere (i.e. under lean conditions) of the exhaust of such engines. I The discovery of Iwamoto (1) and Held (2) et al., showing that CU-ZSM-S catalyzes the selective reduction of NO by hydrwarhons in an oxidizing atmosphere, promoted extensive research in this area. Although this catalyst is active and selective, it has hydrothermal stahility prohlems, due to the degradation of the zeolite framework (3). Since zeolites are metastahle structures, the prohlem of hydrothermal stahility may he circumvented hy the use of metal or metal oxide supported on thermally stahle, large surface area oxides, such as y-Al,O,. The use of hase metals such a copper are not suitahle hecause they form compounds with alumina at high temperatures. Extensive work on the supported Pt group metal catalysts (4.5) suggests that, although thew catalys~$m ay he potential practical catalysts for diesel engines, their optimum range of operation temperatures(200-300 "C) is tcw low for the lean-hum engines A briefreport hy Haruta et al. (6), showing that a 1 wt. % Au/y-Al,O, catalyst, prepared hy the precipitation-deposition method, has an NO conversion of 40% in the presence of 1.8% H,O and 5 % 0, at 300°C. suggests that Au could he a potential component of a practical lean NOx catalyst. Furthermore, it has heen demonstrated hy Haruta et al. (7) and Parravano et al. (8) that the Au particle size is strongly dependent on the preparation method, and that the particle size of the Au catalyst has significant influence on h)th the catalytic and chemical properties of Au. Thus, an invtstigation of Ady-Al,O, (the dependence of its caplytic properties on the preparative methods and its hydrothermal stahility) as potential lean NO, catalysts may he fruithl. Experimental Procedures . y-Al,O, support was prepared hy hydrolyzing aluminum isopropoxide (99.99+ % Aldrich) dissolved in 2-methylpentane-2.4-diol (99+% Aldrich) using the methcwl of Masuda et a1.(9) . It was dried in air at 100°C and calcined in flowing dry air to 460°C, and then in 2.4% H,O to 700°C at a ramping speed of I"C/min. Then the y-Al,O, was sintered in 7% H,O for an additional 2 hrs at 700" C. The average surface area of such preparations ranged from 215-240 mY/g. Ctrprecipitated Ady-AI#, was prepared from a solution of HAuCI, (99.999% Aldrich) and AI(NO& (99.997% Aldrich) using 1 M Na,CO, as the precipitating agent (IO). The catalyst was suction filtered, washed and calcined at 350°C for 4 hrs. Deposition-precipitation of Au/y-Al,O, was conducted in a manner similar to that of Haruta et al. (10). This involved the reaction of Au with the support in the presence of Mg citrate. 50 mL of 5.32 mM solution of HAuCI, was mixed with 2.5 g of y-AI,O, powder. The initial pH on mixing y-Al,O, and HAuCI, was 4.01 and the pH 4.4 sample was prepared without adjusting the pH (the value 4.4 heing that recorded ,ius1 hefore the addition of Mg citrate). For the r s t of the samples, the solution was adjusted to the desired pH with Na,CO, or HNO,. After 1073 the desired pH was achieved, the solution was stirred for half an hour hefore the addition of Mg citrate. The molar ratio of Mg/Au wa. 2.5. The reaction was allowed to proceed in the dark with continuous sZning for 2 h after the addition of y-Al,03 to the Au solution. Then the suspension was suction filtered, redispersed in room temperature douhly distilled H,O, stirred hrietly and suction filtered. This washing procedure was repeated two more times with cold H,O and once with hot H,O (ahout 80°C). The filtered paste was placed in a 100°C drying oven for ahout 2 h, gently crushed and placed in a 350°C oven for 4 hrs., and finally activated in a reaction mixture of NO, C,H, and 0, at 450°C. The last procedure was used hecause sometimes activity increases were observed with time on stream at high temperatures. The lean NO, reaction was conducted in a feed of loo0 ppm NO, IO00 ppm C,H,, 4.8% 0, and 1.6% H,O with the halance He. The weight of the catalyst was 0.5 g and the total flow rate was 104 cc/min. The catalysa were evaluated with respect to three parameters:the maximum NO conversion, the temperature of maximum NO conversion and the NO competitiveness factor at the maximum NO conversion. The NO competitiveness factor is a measure of the efficiency of the catalyst to use NO instead of oxygen in the oxidation of propene and is defined as N0,,,,d*100/(9*C,H6 where 9 is the numher of oxygen at6ms needed to convert C,H6 completely to CO, and H,O. The Au and AI contents were determined hy ICP. It has been reported ( I I ) that the dissolution of Au required a solution containing a good ligand for Au as well as an oxidizing agent. Thus HCI .was added to provide the chloride ligand. and HNO, was added as the oxidizing agent. However, y-Al,O, would only dissolve with the further addition of concentrated HF. Thus all three acids were needed. The CI- concentration was determined using Quantah titrators (Fisher Scientific). The accurdcy of the titrators were verified using different concentrations of NaCl solutions. Results and Discussion The deposition-precipitaticm method is a multistep process. it involves (a) hydrolysis of AuCIi anion to a mixture of IAuCl,OHr, IAuCl,(OH),- and IAuCI(OH),I-. (h) ?dsorption of the negatively charged IAu CI,OH,J- species onto the positively charged sites on the oxide surface, and subsequent formation of AI-0-AU hond hy the condensation ofthe OH groups on the y-Al,O, and OH ligand ofthe Au complex, (c) polymerization of the surface Au complex through further reaction of the OH groups of the adsomed Au complex with other Au .species in solution, and (d) addition of Mg citrate lo physically hltrk the adsorhed Au polymeric clusters from coagulating. Each of these steps could affect the final Au particle size through their influence on the relative rates of condensation and plymerization. Of all the preparation variables, the pH of the solution appears to he of primary importance as it controls hoth the number of adsorption sites on the y- AI@,, and the distrihution of the various Au complexes in solution. The iselectric point (IEPS) of y-AI#, ranged from 6.5 to 9.4 (121. As the pH of the sdution deviates from the IEPS towards the more acidic regime, the positive charge density on the alumina surface increases. This translates to more condensation sites for the anionic Au complexes, and thus higher uptake of Au. The nature of IAuCI,,OH]' in solution as a function of pH was determined by measuring the CI- concentration in solution. Surprisingly, the hydrolysis of the AuCI,' complex was very rapid in the range of pH 4-8, resulting in an average replacement of 2.6 CI' ligands hy OH ligands per Au complex within half an hour of reaction. Longer reaction time did not increase the CI' concentration in solution. Thus it appears that the effect of pH is primarily in the determination of the numher of condensation sites for Au on y-A1,Os. Since the lower pH preparations have more nucleation sites, it is possible that the average Au particle sizes are smaller on such preparations. 1074 G Au loading. wt.% Temp. "C at max NO conv. Tahle 1 shows the % Au dep)sited on y-Al,O, as a function of the pH of preparation Solution (if all of the Au in solution was deposited onto the y-Al,O,. a Au hading of 2.5% was expected). The pH 4.4 sample (one with no adjustment of pH) had the highest Au loading and the Au loading decreased with increasing pH. This is in accordance with the fact that the density of positively charged sites on the support surface decreases with increasing pH. The pH 4.0 sample. for which the pH was maintained hy continuous addition of HN03, had a lower Au loading. possihly hecause of the competitive adsorption of the anionic NO; ions and the dissolution of alumina. 1.3 I .8 1.7 I .4 1 .o 385 385 365 365 365 Tahle I: Effect of pH during Au Precipitation on NO Reduction Activity NO competitiveness factor. % DH I 4.0 ' I 4.4 I 5.5 I 7.0 I 8.2 I 3.8 5. I 4.4 3.6 2.8 ~~ I Max. NOconv. % I 33.3 I 42.4 I 45.1 I 33.5 I 28 I The temperature of maximum NO conversion is Usually a reflection of how active a lean NO, catalyst is. At this temperature, the hydrtkarhon conversion is close to or at loa%. The Au samples with the highest Au loading (pH 4.4) had the highest temperature of maximum NO conversion. Assuming that this sample also had the smdllest particle size, then the activity of the catalyst is prOhdhly dependent on the particle size of Au. Interestingly. the NO competitiveness factor of the various samples also decreased with increasing pH of the preparation solution. This suggested that the effectiveness of the catalyst in the reduction of NO, might he related tu the particle size of Au also. These samples, prepared hy d~)sition-preparationm ethod, were compared with a sample prepared hy co-precipitaticm. The Au content of the co-precipitated sample was 0.33%, although it waq prepared with a solution which would result in a 2% Au loading if complete precipitation chieved. This sample reduced NO exclusively to N,O; showing a 24% conversion of NO to N,O and 70% conversion of C,H, at 375°C. Although the Au loading of this catalyst is low, a 0.21 % Au sample prepared hy the deposition-precipition method converted NO exclusively to N,. Thus, different preparation method resulted in catalysts with different lean NOx hehavior. Besides the possihle structural difference that may contrihute to this difference in catalytic hehavior, the presence of Na+ may also he a factor. The co-precipitated catalysts has suhstantially more Na' left over left in the sample after washing. An essential properly of a practical lean NO, catalyst is high hydrothermal stahility. The stahility of a I g sample ofa 1 .O% Ady-Al,O,, prepared by the deposition-precipitation method way tested in a lean NO, reaction mixture with I .5% H,O for 22 h hetween 400 and 5OO0C, and then at 500°C for 7 more hours with the water in the feed increased to 6%. No deactivation was observed. However, in a more stringent test using only 0.5 g of the 1 % Au sample and 9% water in the feed, a 19% decrease in activity was ohserved after 8 h of reaction at 500°C. and another 20% decrease after another 8 h. Conclusions The activity and selectivity of supported Au catalysts varies with the pH of the solution during preparation in the deposition-precipitation method. These catalysts are superior to that prepared hy the co-precipitation method. The Auly-AI,OS catalysts showed unexpectedly high stahility 1015 under reaction conditions at high temperatures and high water concentrations. Thus, Au supported on y-Al,O, has the potential of king an important component of a practical lean NOx catalyst. Acknowledgement This work was supported hy the U.S. Department of Energy, Basic Energy Sciences, and General Motors Corporation. References (I) M. Iwamolo and H. Hamada, Catal. Today, IO, 57 (1991). (2) W. Held, A. Konig, T. Richter and L. Ruppe,.SAE Paper No. 900496 (1990). (3)G.A. Grinsted, H.-W. Jen, C.N. Monbeuil, M. J. Roskosz, and M. Shelef. Zeolites. 13, 602 (1993). (4) R. Burch, P.J. Millington and A.P. Walker. Appl. Catal. B, 4. (1994) 65. (5) A. Ohuchi, A. Ohi, M. Nakamurd, A. Ogata, K. Mizuno and H. Ohuchi, Appl. Catal.,, (1993)71., (6) S. Tsuhota, A. Ueda, H. Sakurai, T. Kohayashi and M. Haruta, ACS Symposium &)ok Series, No. 552 (Ed. J. Armor) Chapter 34 (1994). (7) M. Haruta, N. Yamada, T. Kohayashi and S. Iijima, J. Catal., 115, 301 (1989).(11)M. (8) S. Galvagno and G. Parravano. J. Catal. 55, 178 (1978). (9)K. Meda, P. Mizukani, S.4. Niwa. M. Toha, M. Watanahe and K. Masuda, J. Chem. Soc. Faraday Trans. , 88(1), 97-104. (10)M. Haruta. S. Tsuhota, T. Kohayashi, H. Kageyma, M. Genet and B. Delmon. J. Catal., 144, 175 (1993). ( 1 1)R.J. Puddephdtt, The Chemistry of Gold, Elsevier, Amsterdam (1978). (12)G.A. Parks, Chem. Rev.. 65, 117 (1965). 1076 1 THE EFFECT OF FUEL SULFUR LEVEL ON THE HC, CO AND NOX CONVERSION EFFICIENCIES OF PDIRH, PT/RH, PD-ONLY AND TRI-METAL CATALYSTS / / / /I D. M. DiCicco, A. A. Adamczyk and K. S. Patel Chemical Engineering Department Ford Research Laboratoly KEY WORDS: CATALYSTS; SULFUR POISONING; SI ENGINE EVALUATION INTRODUCTION: Due to additional requirements imposed by the 1990 amendments to the Clean Air Act. automotive emissions systems niust perform at high efficiencies for 100,000 miles"'. Howcvcr, fuels containing sulfur, can reduce the efficiency of inany modern catalyst formulations'*~". Additionally. the Northeast Ozone Transport Commission (OTC) has petitioned the U.S. Eiivironmentill Protcction Agency (EPA) to require region-wide adaptation of the California Low-Emission Vcliiclc staiidards without tlie application of California's reformulated gasoline program'" which is ncccssary to keep the level of fuel sulfur low. As will be seen, this will result in reduced catalyst iiclivity in tlie OTC, siiicc typicill gasolines contain sulfur levels which vary considerably. Gasolines cn~~liiiiiii5ig0p pinS and 5OOppmS only represent the IOth and 75Ih percentile o f US.c ommercial hiiiiiiiiel' fuels'-'. As will be shown, tlicsc high levels of fuel sulfur will lower the performance of high :~clivityc ;itnlyst Ibriiiulations and niay in;ike compliancc with LEVIULEV eiiiissions lcvcls exlreincly difficult i f not inipossible without the adaptation of low-sulfur fuels. ISI'ICRIMENTAL: Dynainometcr-based catalyst durability testing and evaluiitions were used to ~lcterinineth e effects of fuel sulfur levels on HC, CO and NOx conversioii efficiencies of fully for-iiiiil;~ceiIl' d-only, Tri-Metal (PI/Pd/Rh). PVRh and Pd/Rh catalysts. These four catalyst technologies wcrc evaluated at two degrees of catalyst aging (4K and lO0K miles) using three fuel sulfur levels (34. 266 mid 587 ppniS), For all testin0 ditertiarybutlydisulfide was used as the fuel-sulfur dopant. Tcst procedures included a series of eqLEiibrium lightoff. transient lightoff and dynamic AiriFuel ratio s ~ exp~criinpeiits. ' These experiments were designed to reflect the most common operating conditions of a vehicle's emission system during typical driving. The lightoff experiments were clcsigiied to niiiiiic the cold start process of the vehicle as the catalyst warms up. The dynamic A/F ~r:ilio cxpcriiiicnt was designed to mirror the conditions which occur during feedback control of the eiiginc at cruise. The slightly rich pcrforinance of the emissions system which occurs during mild tl.iiiisiciits ciiii hc ;isscssed froin the A/F ratio sweeps resented. The effects of fuel sulfur on all cciii<~itinnisir c presented. TO expose t ~ i cca talysts to suPfur, jomin. of engine opcration using a fuel with :I pr-esci-ihcd sulfur level at ;in AIF ratio of 15.3 and a catalyst inlet gas teniperature o f 4 W C BREAK AGING AN0 OUT EVALUATION 8 BOX * * CONTROLLER 4.6L-ZV.VB '''1 rcn ENGINE - 101.1 out no- "08.. *".".' TO EXHAUS (111.1) TO EXHAUS To,-w..l B.*h.. TO EXHAUS Figitre I: Sclreniatic of errgirie lest facility. Ex )eriniental Hardware Exhaust eases from a 1993 Ford4.GL 2V engine were routed through h h cxchaiiger into a lest catalyst b r i d (see figure I) for evaluation. The brick was located 2m ~I~~w~ist roefi itliiici exhaust manifold flange. The inlet gas temper;lture to the catalyst was regulated by ;~d~ustintgh e load on the engine or by adjusting the amount o f water flow through the heat cxcli;iiigcr. Correspondingly. the flow rate to the test catalyst was controlled by adjusting either the c~iminc load or by diverting a fraction of the exhaust flow through a second flow path in parallel with ili:c:itiilyst sample. The amount of diverted exliaust was measured by a laminar flow element (see 1:igtu;c I; LFE) in tlic secoiidary stream. To allow for transient lightoff experimcnts on each catalyst mpid switching valve was placed in the exhaust streani to initially divert the exhaust flow around tllc test catalyst so the initial state of the catalyst could be set to ambient conditions. Continuous gas s;~~nples(o ne pre- and one oit catdlyst) were withdrawn into two Horiba emissions benches and :I~~:~Iyzecdiic li second for C8,';;taI ;ICs and NOx. A UEGO sensor and Air/Fuel Ratio Controller Iproviclcd tlic necessary hardware to control tlic engine AIF ratio in a prescribed way. DC DYNAMOMETER C;ltalgst forniulatioiis, description and acin Eight catalyst bricks of four different ~ol-~nulatiow~eisre evaluated. One Iormulation'was PI/Pd/Rh ( l / l4/ l ) ; one was Pd-only (0/l/0); one as PVRh (S/O/I); and the other was PdlRh (0/9/l). Their respective precious metal loadings were IOS, I 10: 60 and 4OgFt3. They were all fully formulated containing stabilizers, scavengers and base llietiil oxldes. The Tri-inetal and the Pd-only catalysts were of a two-layer washcoat design. In each layer, Ihe particle sizes were optimized to promote higher catalyst efficlency when sulfur is added to llic feedgas. They all contained 400 cell/iii' and a cell wall thickness of 0.068in. They were all of c 1077 I ' .(b). . # , . I . . . 0 - , , , , , , , , . . . , . . I Figure 2: a) Sweep tesf; b) Irassierrl lightnjf lest; c) eqiiilihrirrnt Iiglrloff test. Pd-only 4K. the same dimension (3.15"x4.75"~6.O0) and total volume (76in'). Preceding experimentation. four catalysts were dynamometer aged to the equivalent of 4K miles and four to IOOK miles of vehicle use. During this procedure, a commercial unleaded gasoline which contained 16OppmS was used. Am Sweep Test Description The A/F sweep test was conducted by operating the engine at a steady state air-flow of 3U.3ds wGle ramping the h c l flow rate from a lean-to rich A/F ratio. This ramp consisted of the superposition of linear and sinusoidal components. The linear component ranged from c1.0 to -I.OA/F ratios about stoichiometry and occurred over 360s. The sinusoidal component had an aniplitude of 0.5 A/F; its frcquciicy was I Hz. It was used lo evoke all active kinetics over the catalyst including tlie 0, storage inechanisiii. For all experiments, the AIF ratio sweep started at an A/F ratio of 15.2 (A/F,,,,= 14.2 for California Reforinulated Fuel) and proceeded to an A/F ratio of 13.2. The inlet gas temperature at the catalyst was 450k5"C and the space velocity (at STP) into the 76in' catalytic iiionolilh was 85,000 Hi'. Figure 2a shows a typicd rcsult of a sweep test. The abscissa represents AAIF ratio (i.c., AIF,, ,,,,,I -A/F,,,,,,,). The ordinate shows the CO. HC and NOx conversion efliciencics, the CO-NOx crossover efficiency, aiid the A/F ratio operational wiiidow. Values are also marked at a slightly rich A/F ratio. since these values are used later to show the effects of sulfur level on fuel-rich catalyst perforniaiice. These results are critical in determining the "best" catalyst performance for a vehicle operating under warmed up conditions and mild accelerations. Equilibrium Light-Off Test Description The equilibrium light-off test was performed to assess how the-low-temperature chemistry over the catalyst evolves without the complications associated with transient substrate warmup. It was ponducted by "slowly" (12.3"CImin) in; creasing the inlet gas temperature to the catalyst, tliiis allowing thc datalyst substrate to the~mally cquilibrate during expcrinientation. This was accoin lished by passing tlic engine exhaust tlirough ii water controlled heat exchanger, which regulated tfe temperature of gases entering tlie catalyst. As above, the engine was operated at steady state; its air flow rate was 30.3gIs; and its mean A/F ratio was 14.2. About tliis mean A/F ratio, the IHz, i3.5 A/F ratio modulation was applied. During the experiment two gas samples were withdrawn continuously and analyzed every second for CO, total I-ICs and NOx: Corresponding catalyst conversion efliciencics (([I -[I )/[I x 100%) were determined as a function of inlct gas temperature into the catalyst. Figiie y%ho;! :I typical catalyst equilibrium light-off trace for tlie 4K Pd-only catalyst. Clcarly marked are the temperatures coirespoiiding to 50% conversion of CO, I-IC and NOx. These values are used latcr lo :mess the carly liglitoff potential of catalyst formulations aiid the effects of sulfur poisoning on catalyst lightoff. Transient Lirht-Off Test Description To assess how a combination of substrate thermal inertia and the low-temperature catalyst chemistry affects the lightoff performance of the catalyst. the Transient Light-Off Test was conducted after cooling the catalyst brick to 3 8 9 C to define the initial sl:ile of the catalyst. These conditions are typical of those which occur during the cold start of a vehicle. Here, the engine was operated at the same conditions used for tlie equilibrium lightoff experiments. Initially, gases from the engine by-passed-the catalyst through a diverter valve while the engiiie was stabiliied for the experiment. At the start of the transient lightoff experiment, the engine exliaust gas flow was suddenly switched into the flow pith which contained the cold catalyst brick. Two gas samples were withdrawn continuously and analyzed every second for CO. total HCs and NO!, :!!id the corresponding conversioii efficiencies were determined as a function of tinie from the hcgiiiiliiig of the warmup period of the catalyst. Figure 2c shows a typical transient light-off trace iis conversion efficiency versus lime, :ind marks the time necessary to attain 50% conversioii of the iiilct CO,,,HC iiiid NOx. Prior to this time, mostly raw emissions pass the catalyst into the atmos here and t l i i s lightoff" lime must be !:iiiii!nizedleliininat~t~o attain LE\.' or ULEV ernissions levers. RFSULTS and DISCUSSION: Typical vehicle operation includes cold start activation, warmed-up stoichiometric cruise, and sliyhtly-rich accelerations with a11 modes present in the FTP-75@)d riving schedule used to assess vehic e emissions performance. Over this cycle, a vehicle typically produces ;in eiigiiie-out emissions level of I-3gImi THC, IO-12gImi CO. and 1.5-3.Og/mi NOx. These eiiiissions are then converted at hi41 efficiency over the catalyst system to more ertvironinentally acceptable chemical species. 6 attain IOOK ULEV emissions levels (0.055/2.1/0.3g/ml; 1078 I 100 - - - - \ \ . \ ,' , : 70. , , ; '\\.'.*.;:--.-;:. . : . . . . . ~ 0 - - '- - - .\ , . >. - - _: . . . ----:--+ . . x,: : .. - _ .. . -x 0 100 200 300 400 500 60t Fuel Sullur Level (ppmS) 60 Fuel Sulfur Level (pprnS) * PdlRh (4K) + PtlRn (4K) + c=d-onmy ( 4 ~ ) - mi-metal ( 4 ~ ) Y- PalRn (-00%) + PtlRh (1OOK) jY. PCI-only (-0OK) -b Trl-metal (100K) FiKrrre 3: n) CO nrrd NOx conversiori eSJiieiicy nt the A/F ratio corrcsporrdiirg /o COINOX crossover ~ I W ; , I ~sw eep /csfirrg; b) HC conversion efficie;icy o/ CO/NOs crossover. Solid ciirves rcpreseiit 4K ngcd cnfolysls; doshed ciirves IOOK cafn[y~t~. I-ICICOINOx), avern~ee mission system efficiencies of greater than 97%, 8 I% and 86% are necessary. However. these averages assunie that tlie emissions system is operational and functioning at high efliciency from key-on of the vehicle. Generally, the vehicle and emissions system start cold and the c;it;rlyst requircs time to warin to its lightoff temperature, hence passing unconverted emissions to the ;itiiiospheic. Since CO and HC emissions are abundant during cold start, the average CO and HC efficiencies over the remainder of tlie drive cycle must be significantly hi ,her than the averages specified above. As seen in figure 3, when sulfur level is low, these high ekficiencics are obtained fnr Pd-oiily and Tri-metal catiilysts and would also be obtained for WRh and PdlRh with more c:it;ilyst voluiiie iii the emissions system. However, at higher levels of fuel sullur and at IOOK aging. iill efficiencies drop well below the levels needed to attain LEV aiid ULEV. As seen later. catalyst lightoff is also negatively inipactetl by fuel sulfur, thus further exacerbating the problem. Wariiictl Ut, Catalyst Operation. Figure 3 presents the catalyst efficiencies ilt the A/F ratio corresponding to the COINOx crossover point (see Fig 2) and figure 4 shows them at an A/F ratio of 14.0. These NF ratios are chosen since they reflect many of tlie typical opcrating points of a w;iriiied up vehicle that occur during cruise and mild accelciatioiis. Since three way catalysts must siiiiult;ineoiisly convert HC. CO and NOx at high efficiency, the COlNOx cross over point is normally nciir the AIF ratio corresponding lo optimum catalyst operation. As seen i n figure .la, the COINOX efficieiicics of all catalyst formulations are greater than 96% efficient at low sulfur levels and at low niilmge. tvlorcover, when aged to the equivalent of IOOK miles, these formulations have conversion cflicicncics in excess of 92% when low sulfur levels are present in the fuel. Here, the efficiencies of ilie Tri-metal and tlie Pd-only are in excess of 96.5% after IOOK aging, :ind the efficiencies of the PtfRIi and the PdlRli formulations are 91% and 92%. respectively. However, for IOOK aged catalysts, wlieii sulfur is added to the fuel during evaluation, the COlNOx efficiencies of these catalysts drop. For the Tri-nictal (thc most resistmt to sulfur poisoning due to its multi-layer structure and advanced st;ibili7.crs), tlie efficiency falls from 98% to 86% when the fuel sulfur level goes froin 34 to 587 ppiiiS; Ptl-only from 96% to 69%; PURh from 92% to 65%; and the PdRh from 92% to 65%. At 4K. the ordering of sensitivity to sulfur poisoning is similar to the above at IOOK aging with tlie itmount of lost pel-formancc being less. In terms of the change in emissions throughput (( 1.0- %Eff/l~)O]l,,wsI[ I .O-%Eff/IOO),, hS). the effect of changing fuel sulfur level from 34 to 587ppmS w d d iiicrease the amount of C6 aiid NOx delivered to the atmosphere by approximately 4-9 times thc i1iiinunt dclivcrcd when the fuel sulfur level is low. As seen for thc IUOK catalysts. much of this lost perforiiiance occurs when the fuel sulfur level increased from 34 to 267ppmS with the catalysts becoming less sensitive to the addition of sulfur above these levels. The cffect of sulfur on HC conversion efficiencies is shown in figure 3b. Trends are similar to those discussed above. However, since HC conversion efficiency must be extremely high to meet LEV or ULEV emissions regulations, the level of efficiency loss due to the addition of sulfur to the fuel will In;ike it extremely difficult or potentially impossible to reach these low emissions levels with the most ;Itlv;iiicetl catalyst forinulation developed lo date. As seen iii figure 3, near stoichiomctry tlie Pd-only cstalyst raiiks first behid the Tri-metal in efficiency throughout the range of sulfur application. Even tliough it is susceptible to sulfur'*'. its higher initial activity at low sulfur is retained throughout the r;ingc of typical sulfur application when operated near stoichiometry. Its performance is higher than t l i i i t of the PtfRh or PdlRh catalysts studied. As mentioned, this is in part due to its higher initial ;ictivity and in part due Io the combination of materials which comprise its washcoat to reduce its sclisitivity lo sulfur. Here, the catalyst is of a multi-layer design containing an abundance of ceria :llid I:ii!thana plus scavengers to inhibit the detrimental effects of sulfur. Furthermore, the particle siziiig in ciicli Inycr has bcen optimized to enlinnce rcac:ion at high sulfur level. Rcsi~ltso f catalyst perforinance at ai i average fuel-rich AIF ratio of 14.0 (0.2 rich of stoichiometry i ~ i o~s(ci~lla ting at IHz) are shown in figure 4. Here, HC, CO and NOx efficiencies are presented for 4K slid IOOK aged catalysts as fuel sulfur level is increased from 34 lo 567 pmS As seen. the l,crforln;ince of iill catalysts is substantially rcduced when sulrur is added to the Rei. As ail example, \vIicIi the sulfur level is low. the NOx conversion efficiency for a l l catalyst formulations is greater i11:111 95% for hoth 4K and IOOK aged catalysts. Here, the Tri-metal foriiiulatioii shows the least 1079 1 100, 1 - =- - w 0 Fuel Sullur Level (ppmS1 801 : I I O 100 200 300 400 500 60 CO' : - _- _. .....i.- -.- 0 100 200 300 400 500 60 Fuel Sulfur Level (ppmS) 20 0 100 200 300 400 500 60 Fuel Sullur Level IpPmS) 1:igirre 4: IIC, CO arid NOx efficiency versus srrlJlrr level. Solid crimes iridicate 4K aged cntnl~slr;d aslied curves iridicate IOOK colnl~sts. Note: See fisrrrc 3 for legend. sensitivity to sulfur having its NOx efficiency drop from 98% to 95% for both 4K and IOOK of aging. The order of NOx efficiency loss under rich operating conditions among all catalyst formulations goes from Tri-metal to PURh. to Pd/Rh and to Pd-only. The Tri-metal being the least sensitive and the Pd-only being the most sensitive as sulfur is added to the fuel. Gencrall , to meet LEV and ULEV emissions levels, Ndx conversion efficiencies around 90% are necessary at IOOK miles. As seeti in figure 4, when sulfur level is low, all advanced catalyst formulations have an efficiency well above this value. However, when sulfur is added, the NOx conversion efficiency of both the PdRh and the Pd-only drop below the levels tieeded to attain LEV or ULEV emissions levels. Moreover, with the possible addition of a high-speed, high-acceleration driving cycle to the test procedures, meeting the NOx standard with a high sulfur level in the fuel becomes even more difficult. Catalyst Lialitoff Experiments. 111 :iddilion to the emissions generated during continuous operation, more than 80% of the CO and HC emission occurs during cold start of the vehicle before the catalyst becomes active. Any increase in "lightoff' temperature or "lightoff' time due to sulfur addition will present major problems in meeting LEV and ULEV emissions levels, since the exiting flux of HCs and CO are high during this period. Lightoff temperature corresponds to the temperature of the substrate at which the conversion efficiency of CO, I4C or NOx reaches 50%. Lightoff time refers to the time during the transient test rocedure at which tlie conversion efficiency oYC0. HC or NOx reaches 50% conversion. Figure 5 shows the cffect of added fuel sulfur on catalyst lightoff temperatiire of CO for all formulations studied. Lightoff temperature of IIC and NOx will follow the same trends as of CO. since they are strongly dependent on the heat generated by the exotherin during CO lightoff. \ In figure 5, the lightoff temperature is plotted as n functioti of sulfur level for each catalyst formulation. The tri-metal and the Pd-only catalysts have the lowest lightoff tempcraturcs of all formulations at 4K and IOOK aging. At low sulfur level, the lightoff temperature for the Pd-only and tri-metal catalysts are about 35°F lower than for the PURh and PdRh catalysts. This is due to the excellent low temperature CO and 14C kinetic properties of Pd. Since the triimetnl catalyst has :I multi-layered washcoat, tlie Pd-containing laye; inthis structure promotes low tcinper;iture lightoff. At higher sulfur levels, the lightoff temperature for the Pd-only and the tri-metal forniulations continue to be lower than the PURh and the Pd/Rh catalysts due to their higher initial activities and the incorporation of stabiliziers and scavengers into their formulations to resist sulfur Ipoisoning. At IOOK :uid high sulfur levels, both the tri-metal and the Pd-only formulations have the lowest lightoff temperature. Upon reproducible vchicle cold s(art, a direct rclationship should exist between catalyst lightoff temperature, lightoff time and cold start emissions, assuming the catalysts have identical substrate llierninl inertia. and heat and iiiass transfer characteristics. Here, transient lightoff experiments were coiiilucted to assess thc lightoff time of each formulation. at all sulfur levels and at 4K and IOOK. Figure 6 shows the results of these transient lightoff experiments. The curve shows the relationship betwceii lightoff time for our experimental geometry and catalyst lightoff temperature. I t should be iotcd that the exact values of lightoff time are unique to these experimental conditions. Both mass flow rille and inlet gas temperature profile ore critlcal to the absolute values of lightoff time. As cxpcctcd, as tlie lightoff lciiiperaturc iiicreases. the liglitofl' time increases. Since all catalyst bricks were of the same geometry, containing the same therm! incrtin and geometric surface area the only inajor difference between formulations arises through their differences in critical lightoff temperature. As scen, there is ii direct linear correspondeoce between lightoff temperature and lightoff time for a l l catalysts studied. This sug5ests that the catalyst which retain the lowest lightoff temperatures during aging and poisoning will lightoff sooner during vehicle cold start, thus producing fewer cold start emissions. As seeii in the figures 5 and 6, increased sulfur concentration increases lightoff temperature for a l l cafalysts, suggesting that the corresponding vehicle emissions will be impacted in n negative manner. 1080 . . 0 100 200 300 400 500 60 Fuel Sullur Level loom51 Liglrfoff fenrpcrafrrref . r CO: Soli8 curve 4K; dashed curve IOOK. I:irrrre 5: I 7igrrre 6: Trarrsierif lightoff lime versus lislrfoff feniperaturc for all cofalysts. Catalyst cleaninp, To assess the regeneration of the catalyst after exposure to sulfur, catalyst performance for all formulations was evaluated at several stages of cleansing. These stages included: I) exposure at 260ppmS; evaluation at 260ppmS; 2) evaluation at 34ppmS; 3) evaluation at 34ppmS after high-temperature (660°C), rich (A/F=13.6) cleaning for 30tnin; 4) exposure a! 587ppmS. evaluation at 587ppmS; 5) eval!iation at 34ppmS; 6) evaluation at 34ppmS atter high tern erature, rich cleaning for 30 min. As seen in &ure 6 for a Pd-only catalyst at the CO-NOx crossover point, more than half of the efficiency loss due to sulfut poisoning is regained wheti evaluation procccded using a lowsulfur fuel. However, to regain nearly all efficiency loss, a rich high temperature cleansing of tlie catalyst was iiecessar and is in agreement with the work of BeckYz’ et al. In addition, trends for lightoff temperature are 1 similar to these in that the application of sulfur in fuel raises the lightoff temperature and time of rhc catalyst and a high tcmperaturc cleansing is necessary to return it to its pre-exposure levels. COorNoxatCmsover HCalCrossover l?grrre 7: Currversiorr efjiciericy at CO/NOx p , ~ ~ o lcraf~olyy sl. Dofa “1: , I ) ex;,=- 26n, evol=260; 2) e.rp=260, eval=34; 3) exp=34; cvol=34; 4) cxp=SS7, evaI=SS7; 5) cxp=587, ettal=34; 6) cxp=34, evak34 ppmS. CONCLUDING REMARKS: In evaluating fully formulated Tri-metal, Pd-only, Pt/Rh and PdIRIi c;italysts ;it 4K or IOOK miles of aging during tlie application of 34, 260 or 557 ppmS to the f d stock. resuIIs indicate that the application of sulfur reduces catalyst efficiency (the Tri-metal being the lciist affectcd) near stoichiometry and rich of stoichiomctry. Moreover, it increases the liglitoff tcinperature and the lightoff titile of a l l forinulations evaluated. The consequence of these results is the suggcstion that. when o erdted on fuel containing elevated sulfur levels. overall vehicle cliiissioiis syslem pcrforinance will & & J e due to the increased sulfur level. Fortunately. when fuel sulfiir is removed much of the lost efficiency is regained. but to fully regain lost efficiency, a high teinpcratitre. rich cleaning process must be applied. As seen, conversion efficiencies for CO, HC and NOx iiccessary to achieve LEV or ULEV emission levels will be significantly lowered due to fuel sdrur atid cui impede attainment of these levels. ACKNOWLEDGEMENTS: The authors thank Mr. David Osborn and Mr. Arlhur Kolaskn for coitducring the sweep and lightoff experiments. I) C;rlvctt,J.Ci., I-lcywood,J.B.. Sawycr,R.F. and Seinfe1d.J.H.. Science 261 17-45 (1993). 2) Bcck, D.D..S omniers, J.W. and DiMaggio, C.L.. Applied catalysis’ B:Eb;ironnhal. 3, 3) Kock1.W.J.. Bens0n.J.D.. Burns,V.R., Gorse,R.A.,Jr, Hockhauser A.M., Knepperj.C., Leppird.N.R.. Painler,L.J., Rapp,L.A.. Reu1er.R.M. and Rutherford,J.A’.. SAE paper 932727, (Soc of Auto..Engr.) (1993). 4) California Emission Standards iii the Northeast States, Environmental and Safety Engiiieering. Internal Report, March 7. 1994. 5) I990 Motor Vehicle Manufacturers’ Association Summer Gasoline Survey (American Automobile Manufacturers Association, Detroit. 1990). 6) U.S. Federal Register, Vol 37. No. 221, Part If. November 15 (1972). 205-227 (1994). 1081 THE EFFECT OF SO2 ON THE CATALYTIC PERFORMANCE OF CO-ZEOLITES FOR THE SELECTIVE REDUCTION OF NOX BY METHANE Yueiin Li and John N. Armor Air Products and Chemicals, Inc. 7201 Hamilton Boulevard, Allentown, PA 18195-1501, USA Keywords: NOx reduction, So2 effect, Co-zeolites INTRODUCTION Selective catalytic removal of Nlox from stationary emission sources is an important and challenging task. Beyond the SCR (selective catalytic reduction of NOx) technology with ammonia, a variety of alternative approaches have been explored in the past few years, such as direct NO decomposition [.l,.and references .therein] and NO reduction by hydrocarbons [2- 141. Most of the current studies involve CI to C3 hydrocarbons as selective reducing agents for NOx over metal zeolite catalysts. Among many performance factors, e.g., activity, selectivity and catalyst stability, the inhibition or poison of catalyst by exhaust by-products, such as Hz0, S a and other compounds are also important issues. The effect of Hz0 on catalyst performance was tested for many of these systems. However, the effect of So2 on NO conversion for these systems has not been sufficiently addressed. Low levels of sulphur compounds exist in most of the fuel sources we use today and is known to poison many catalysts. Building upon our earlier work 12, 15-20], we extended our study to the effect of So2 on the catalyst performance. We describe here our studies on the effect of So2 and/or H20 on catalytic performance over Co-ZSM-5 and Co-ferrierite. EXPERIMENTAL The preparations of Co-ZSM-5 and ferrierite were described previously 1161, and they have the following compositions: Si/AI=I 1 and Co/AI=0.49 for Co-ZSM-5, and Si/AI=8.5 and Co/AI=0.39 for Co-ferrierite. The catalytic activities were measured using a microcatalytic reactor in a steady-state plug flow mode. Normally a 0. IO g of sample was used for activity measurement. The feed mixture typically consisted of 850 ppm NO, lo00 ppm CH4 and 2.5% 02, and the total flow rate was 100 cc/min. (The space velocity was 30,000 h-I based on the apparent bulk density of the zeolite catalyst, - 0.5 g/cc). Water vapor was added to the feed tising a H20 saturator comprised of a sealed glass bubbler with a medium-pore frit immersed in de-ionized H20. Helium (25 cclmin) flowed through the bubbler, carrying the H20 vapor to the feed. For reactions involving SOz, a special, two-inlet reactor was used to minimize the contamination of the system by S a exposure. A .$@/He mixture (212 ppm) was added to the reactor via a separate inlet, and this So2 stream was mixed with other gases (NO/He, @/He and CHdHe) in the quartz reactor within the furnace. The final concentration of SO:! in the mixture was 53 ppm. TPD measurements were conducted in the same reactor system. For a typical TPD measurement, a 0. lg sample was used. A sample was pretreated in situ at 500°C in flowing He for lh. Alternatively, a catalyst was allowed to undergo a steady-state NO/CH4/@ reaction in the presence of SO2 at 550°C for 2 h then flushed with He at the same temperature for lh. In both cases, temperature was decreased to 25°C in flowing He. The NO adsorption was carried out at 25°C by flowing a NO/Ar/He mixture (1700 ppm NO, 5500 ppm Ar) through the sample at 100 cc/min, and the effluent of the reactor was continuously monitored by a mass spectrometer (UTI 100C). Typically, a period of 30 minutes is sufficient to achieve a saturation for NO adsorption with 0.1 g catalyst. After the NO adsorption, the sample was then flushed with a stream of He (100 cc/min.) at 25°C to eliminate gaseous NO and weakly adsorbed NO. As the gaseous NO level returned to near the background level of the mass spectrometer, the sample was heated to 500°C at a ramp rate of 8"Clmin in flowing He (100 cc/min.), and the desorbed species were monitored continuously by the mass spectrometer as a function of timeltemperature. RESULTS AND DISCUSSION The effect of SOz addition on the NO conversion over a Co-ZSM-5 catalyst was tested first with a dry feed. In the absence of SOz, 39% NO was converted to Nz at 500°C. Upon addition of 53 ppm of SOz, the NO conversion quickly increased to >50%, then gradually decreased with time and reached to a stable level (- 32%) in - 2.5 h. The dramatic change of NO conversion in the initial period upon So;! addition reflects the accumulation process of So2 on the catalyst. Obviously, the first portion of SO:! deposited on the catalyst has most impact on the NO conversion, and the steady-state NO conversion obtained after 2h in the So2 containing stream indicates an achievement of an equilibrium condition for adsorption and desorption of So;?. Interestingly, increasing the reaction temperature to 550°C in the presence 1082 of raised the NO conversion to a new steady-state level at 55%, which is even higher than that with a Sm-free feed at the same temperature (27%). Further increasing the temperature to 600°C decreased the NO conversion to 42%. which is still twice of that in the absence of SOZ (Table 1). Note, in the absence of So2, the NO conversion has a maximum level at - 450°C on Co-ZSM-5. The addition of SOZ shifts the optimum temperature to - 550°C. Therefore, much higher NO conversions can be obtained at T t 550°C in a SOZ containing Stream. Table I summarizes the impact of SOZ and/or Hz0 on the catalytic performance of Co- ZSM-5. At 6oo"C, the addition of HzO (2%) + SOZ (53 ppm).does not have a significant impact on the NO conversion. However, at T 5 550°C. the presence of both So2 and Hz0 significantly reduces the stabilized NO conversion. The positive effect of SOZ with a dry feed at 550"C.diminishes when HzO is added. Note, at 550°C 2% HzO.alone has no impact on the conversion. Addition of S o 2 also decreases the CH4 conversion (see Table I) in a way consistent with the change in NO conversion. When SO2 was added to the feed at 5OO0C, a continuous decrease in CH4 conversion was observed. For steady-state runs, substantially lower CH4 conversions resulted from SO2 addition. This decrease is more pronounced at lower temperatures and with the co-presence of HzO vapor. The selectivity of CH4 is gieatly enhanced as the result of S a addition. At 500°C, CH4 is consumed exclusively for the reduction of NO. .To determine the fraction of Co covered by SO2 during a'steady-state NOICH4IOz reaction, NO adsorption at room temperature and TPD of NO were carried out on a fresh and SO2 exposed Co-ZSM-5 catalysts. A fresh Co-ZSM-5 was pretreated in situ at 500°C for 1 h in flowing He (100 cchin). A separate sample of Co-ZSM-5 was exposed to a feed containing 53 ppm So2, 850 ppm NO, IO00 ppm CH4, 2.5% 02 at 550°C for 2 h. The sample then was flushed with He at the same temperature for lh to flush out the gaseous SOz and subsequently cooled down to room temperature in Rowing He. The TPD measurements with the SO2 exposed Co-ZSM-5 indicate a complete disappearance of the NO desorption peak at 360°C and decreased intensities for the desorption peaks at 290 and 220°C. The low temperature desorption peaks are unaffected. The quantification of the TPD measurements gives 0.88 mmollg (1.35 NO/Co) and 0.65 mmollg (1.0 NOICo) for fresh and SOz exposed Co-ZSM-5. respectively. The SO2 coverage is 26% of the total Co sites. We reported earlier that Co-ferrierite is more active and selective than Co-ZSM-5 [ 16, 191. However, Co-ferrierite is more sensitive to SO2. At 500°C (in the absence of HzO), upon addition of 53 ppm SO;?, the NO conversion, initially 61 %, decreased with time and stabilized at 16% after - 2 h. Increasing temperature to 550°C in the presence of So2 raised the NO conversion to 23% initially, and the conversion increased only slightly in 1.5 h. After eliminating the SU2 from the feed the conversion increased with time (from 25 to 32% in 1.5 h). [As shown in Table 2, in the absence of SOz the NO conversion is 50% at 550°C] Note, a small further decrease resulted due to the addition of 2% Hz0 vapor in a So2 containing feed. In the presence of SOz and Hz0, the optimum operating temperature is shifted to 600'C. With 2500 ppm CH4 [Our normal [CHI] is loo0 ppm.], NO conversions of 65 and 45% were obtained in the presence of 53 ppm SOz under dry and wet conditions, respectively, which are comparable to those in a SOz free feed (56% in dry feed and 51 % in wet feed). Similar to Co- ZSMJ, a dramatic decrease of CH4 consumption was found due to So2 addition. The TPD profiles performed on Co-ferrierite show a substantial reduction in intensity for the NO desorption peaks at - 160 and 220°C on the SOz exposed Co-ferrierite, while other desorption peaks remain same. The amounts of NO desorption integrated from the TPD measurements are 0.91 mmol/g (1.35 NO/Co) and 0.68 mmol/g (1.0 NO/Co) for fresh and SOz exposed Co-ferrierite, respectively (26% reduction of NO desorption on the So2 exposed Co-ferrierite). Obviously, the change in topology of zeolite has a strong impact on the effect of SOz. With a dry, SO2 free feed, Co-ferrierite is more active than Co-ZSM-5 for the NO/CH4/02 reaction, but in the presence of SO2, Co-ZSM-5 is more active. Under certain conditions, SOz even doubles the NO conversion on Co-ZSM-5. It is possible that SOZ preferentially adsorbs on the sites on the outer surface of the zeolite or at the entrance of the 10-member rings. Based on our earlier studies of Co-zeolites with various exchange levels of Co2+, we believe these sites are less selective for the NO reduction but more active for the combustion of CH4. Exposure of SOZ at high temperatures may selectively poison these sites. The TPD profile of Co-ZSM-5 indicates a wide distribution of sites. While on Co-ferrierite, the NO desorption is 1083 dominated by the peak at 160 oC, and this peak was significantly reduced by SOz exposure. On Co-ferrierite Sa reduces the NO conversion at all temperatures. In contrast to Sa poisoning, HzO molecules adsorb on Coz+ sites uniformly, and consequently the CH4 selectivity does not change significantly (see Tables 1 & 2). CONCLUSIONS Co-ZSM-5 and Co-ferrierite behave differently in response to SOz addition. Over Co- ZSMJ, SOz significantly enhances the NO conversion at T > 500°C in a dry feed; while over Co-ferrierite, Sa greatly reduces the NO conversion. However, Co-ZSM-5 suffers significant activity loss when both Sa and H20 are added to the feed. On Co-ferrierite, the presence of both S a and H2O only caused a modest decrease in NO conversion compared to Sa alone. On the other hand, on both catalysts SOz inhibits the CH4 combustion activity more than NO reduction. As a result, the CH4 selectivity improved dramatically. SOZ poisons the catalyst by strongly adsorbing on the Co2+ sites. The degree of the reduction of the number of sites over both catalysts was measured by TPD and revealed that about 30% of the Co2+ sites are blocked under steady-state reaction conditions at 550°C. Interestingly, the preference of SOZ adsorption on COZ+ sites is not the same on these two catalysts (due to their different structural characteristics), which may be the reason why they response to So;? addition differently. REFERENCES 1 2 3 4 5 6 7 8 9 IO 1 I 12 13 14 15 16 17 18 19 20 M. Iwamoto, in M. Misono, Y. Moro-oka and S. Kimura (Editors), Future Opportunities in Catalytic and Separation Technologies, Elsvier, Amsterdam, 1990, p. 121. Y. Li and J.N. Armor, Appl. Catal. B, 2 (1993) 239. S. Sato, Y. Yu-u, H. Yahiro, N. Mizuno, and M. Iwamoto, Appl. Catal., 70 (1991) LI. H. Harnada, Y. Kintaichi, M. Sasaki, and T. Ito, M. Tabata, Appl. Catal. 64 (1990) LI. M. Misono and K. Kondo, Chem. Letter, (1991) 1001. M. Iwamoto, N. Mizuno and H. Yahiro, in in Proc. 10th Intl. Cong. Catal., 19-24 Jul., 92, Budapest, Hungary, pp 1285. C. I. Bennett, P.S. Bennett, S.E. Golunski, J.W. Hayes, and A.P. Walker, Appl. Catal. A 86, (1992) LI. Y. Ukisu, S. Sato, G. Muramatsu, and K. Yoshida, Catal. Lett. 16 (1992) 11. Y. Teraoka, T. Harada, T. Iwasaki, T. Ikeda, and S. Kagawa, Chem. Lett. (1993) 773. C. Yokoyama and M. Misono, Bull. Chem. Soc. Jpn., 67 (1994) 557. K. Yogo, S . Tanaka, T. Ono, T. Mikami, and E. Kikuchi, Microporous Materials, 3 (1994) 39. K.A. Bethke, D. Ah, and M. Kung, Catal. Lett. 25 (1994) 37. E. Sakamoto, T. Ohnishi, and T. Arakawa, in Stud. Surf. Sci. Catal. Vol 84, J. Weitkamp, H.G. Karge, H. Pfeifer and W. Holderich (Eds.), Elsevier, 1994, ~1537. F. Witzel, G.A. Sill and W.K. Hall, J. Catal, 149 (1994) 229. Y. Li and J.N. Armor, Appl. Catal. B, I (1992) L31. Y. Li and J.N. Armor, Appl. Catal. B, 3 (1993) LI. Y. Li and J.N Armor, I. Catal. 145 (1994) 1. Y. Li, B.J. Battavio and J.N. Armor, J. Catal., 142 (1993) 561. Y. Li and J.N. Armor, J. Catal. 150 (1994) 376. Y. Li, T.L. Slager and J.N. Armor, I. Catal. 150 (1994) 388. Table I Effect of so? on conversions/selectivity over Co-ZSM-Sa w 2 1 500 oc 550 oc 600 OC (ppm) dry wetb dry wetb dry ivetb NO conv . 0 39 30 27 28 21 22 (%) 53 32d 15 55 25, 18C 42 24 (%) 53 13d 6 47 25, 19c 93 78 CH4 conv. 0 91 38 100 86 loo 100 Selectivity 0 18 33 11 14 9 9 selectivity. Table 2 Effect of SO7 on conversions/selectivity over Co-ferrieritea [SO21 500 oc 550 oc 600 OC (ppm) dry wetb dry wetb dry wetb NO conv . 0 61 28 50 40 40,56d 32,5Id (%) 53 1 6C 13 25 18 30, 65d 24, 45d (%) 53 6C 5 IO 9 53,56d 31,55d CH4 selec. 0 43 52 22 23 17, Id 14, sd (%) 53 -100c -100 loo 85 24, 20d 33, 14d a Feed: 850 ppm NO, loo0 ppm CH4,2.5% 02; 2% H20 added; stabilized conversion (CHq] = 2500ppm. CH4conv. 0 60 23 93 75 100, I d loo, l d 1084