Can Carbon Dioxide Be Reduced to High Molecularweight Fischer-tropsch Products

7/27/2019 Can Carbon Dioxide Be Reduced to High Molecularweight Fischer-tropsch Products

1/7

CAN CARBON DIOXIDE BE REDUCED T O HIGH MOLECULARWEIGHT FISCHER-TROPSCH PRODUCTS?Imre PuskasResearch Services, 939 Brighton DriveWheaton, IL 60187. Te1.(630) 653-4897

Keywords: clean fuel production, carbon dioxide, Fischer-Tropschproducts, methanol, remote natural gas.INTRODUCTION.

Our interest in t he title originates from our efforts todevelop an economically viable clean fuel process from naturalgas. Natural gas, wi th its high calorific value 210.8 kcal/g-molmethane) is one of t he most preferred fuels from theenvironmental point of view because of its clean burningcharacteristics. However, natural gas found in remote areas, farfrom markets, inaccessible to pipeline transportation, cannot bereadily utilized. Currently several alternatives are practicedfor remote natural ga s utilization (1). Natural gas can be.liquified by cooling to its boiling point (-163%) and shipped inrefrigerated containers. Natural gas can also be converted tomethanol or hydrocarbon liquids (syncrude) OK ammonia at itssource, and these products shipped to market.

Both methanol and th e Fischer-Tropsch (FT) hydrocarbon liquidsare clean fuels . Their fuel uses have been evaluated (2.3).Currently they cannot compete with the less expensive crude oil-derived fuels. Methanol commands a higher price as a chemical ,but this market is relatively small (estimated 27 MM tons perannum, worldwide) compared to the huge fuel market. Increasingpercentage of the methanol production originates from remote gasusing giant plants (800-975 M ton/annum capacity), takingadvantage of the low gas costs and the economics of large sca leproduction. Historically, the methanol market can becharacterized by periods of shortages and periods ofoverproduction and low capacity utilization. Recently we proposedthe development of a methanol-syncrude coproduction technology(4) which could keep t he methanol plants running at full capacityeven in case of methanol oversupply. Th e co-production scheme ofFigure 1 would provide both economic and technological advances.In the first step, the compressed synthesis gas would bepartially converted to methanol. This reaction has equilibriumlimitations. The unconverted syngas from the methanol reactorwould be converted to hydrocarbons. This latter reaction has noequilibrium limitations. We are currently working on the detailsof a research and development plan to demonstrate the viabilityof a co-production technology. The key to success depends on thedemonstration that the effluents of the methanol reactor (amixture of Hz, CO and CO ) can be efficiently converted to highmolecular weight F T products. The percieved difficulty is causedby the presence of carbon dioxide, which is known to yieldpreferentially methane rather than high molecular weight FTproducts in reductions (5). This study was undertaken to providea stimulus for the development of a methanol-syncrudecoproduction technology. Reported cases of carbon dioxidereductions to reasonably high molecular weight FT productsalready exist. The study should be helpful to set the stage forfurther progress.HISTORICAL OVERVIEW OF CARBON DIOXIDE REDUCTIONS.

The reactions, utilization and sources of carbon dioxide hav erecently attracted considerable interest because of the possibleecological effects arising from large scale carbon di'oxideemissions into the atmosphere. An information update is providedin very recent reviews by Xiaoding and Moulijn (6) on co2reactions and usage; by Krylov and Mamedov S \ nn i t s.-, --. - -heterogeneous catalytic reactions; by Jessop, Ikariya and Noyori? ) on its homogeneous catalytic hydrogenations; by Tanaka o n itsflxation catalyzed by metal complexes 8 ) : and by Edwards on itspotential sources and utilizations (9). one of the most importantreactions of carbon dioxide is its reduction to methanol:Cu-ZnO, 200-260C, 5-10 MPa

Cot t 3H1 CHJ-OH t H 2 [E-11680


2/7

Although carbon dioxide has been reduced to methanol in the Pastin commercial operations (lO,ll), current methanol plants.usemixtures of carbon dioxide and carbon monoxide. An alternativepotential use of carbon dioxide would be its complete reductionto methane or to mixtures of Fischer-Tropsch type hydrocarbons:Ni (or Co, Fe, Ru), 25O-40O0Cco2 + 4n1 Y CH, t 2H20 [E-21

nC01 + 3nH1 Fe or Co (Ru, Ni), 200-300'C-- CHz), t 2nH20Franz Fischer and coworkers were the first to try to reducecarbon dioxide to hydrocarbon oils, after their development Ofhydrocarbon synthesis from carbon monoxide. They h ave found, thatcarbon dioxide gives preferentially methane, with so me gaseoushomologues (12). However, liquid and solid hydrocarbons were alsoobtained in some experiments (13). These early reports havenoted, that carbon monoxide was a reaction interm ediate (12) andthat liquid hydrocarbons were observed in those experiments, whenthe catalyst was alkalized or it contained a Cu component (13).In the last decades, many chemists and surface scientistshave extensively studied the reduction of carbon dioxide tohydrocarbons and the chemisorption of carbon dio xide on catalyticSUKfa,CeS. It is out of the scope of this study to review theliterature. However, a restricted number of references are cited(14-35) to sample the diversity of worldwide interests. Thecitations exclude the literature on carbon dioxide reductions tohydrocarbons which proceed via methanol intermediate.The cited studies unanimously agree with the early conclusionsthat carbon monoxide is an intermediate formed by the reverseWater Gas Shift (WGS) reaction:

catalystcol t H? o t n o [E-41

The reduction of carbon monoxide proceeds by the methanationreaction O K FT synthesis. Falconer and Zagli have proposed (34)that th e preferential formation of methane over higherhydrocarbons is caused by the high H2:CO ratio on the catalystsurface. While the major product was methane in most of thestudies, a few cases of liquid hydrocarbon formation were alsoreported. Table 1 compiles the best examples of higherhydrocarbon formations. In the Table, we have converted thereported hydrocarbon selectivity data to Anderson-Schulz-Flory(ASF) growth probability values (alphas) to provide a basis foreasy comparison of the product molecular weight distributions.ASF alpha values in the 0.6-0.7 range have been achieved, mostlyon potassium-promoted Fe catalysts. Kuester (13 ) has evaluateddifferent variations of unsupported, alkalized Co and Fecatalysts. In their work, reported in 1936,' the formation ofsolid hydrocarbons (waxes) was often observed. Unfortunately, thereported product analyses were qualitative in nature and we wereunable to deriv e chain growth probability values for productcharacterization. However, the isolation of waxes suggests thatthe chain growth probability values must have been substantiallyhigher than 0.70 and probably were the highest in Table 1. Inthe penultimate example of Table 1, the primary olefinic productswere converted to aromatic hydrocarbons over the ZSM-5 componentof the catalyst. The last example of Table 1 is a case of higherhydrocarbon formation over RhINbO,. This appears to be aninteresting case, since Rh is no{ known for FT catalysis.

In order to understand better how cOl reduction can bechanneled toward higher hydrocarbon formation, relevantfundamental knowledge on the WGS and FT reactions will bereviewed and discussed below.THE REVERSE WATER GA S SHIFT REACTION STEP.

The reduction of carbon dioxide to carbon monoxide, known asthe reverse WG S reaction LE-41, has been extensively studied (36-39) because of its industrial importance in synthe sis gasreactions and hydrogen manufacture. The most ef ficientheterogeneous catalysts for the WGS reaction ar e the Cu-based681


3/7

catalysts, particularly Cu-Zn systems, the iron oxide basedcatalysts and the alkalized, sulfided Co-Mo catalysts (39).Other metals, oxides also have some catalytic effect, but theyhave received muc h less attention. However, alkalization wasfound to increase substantially the WGS activity of manysubstances 3 9 ) . The alkalized FT catalysts have been extensiv.elystudied (13.40-43). Their WGS activity has been long known, butmost of the cited studies focussed on the effect of alkalipromotion on the changes in the rate and the products of the FTreaction. Th e alkalized FT catalysts seem to be excellentcandidates for the reduction of carbon dioxide to FT hydrocarbonsas the examples of Table 1 also suggest. Surface scientists havefound (44-45) that alkalization of FT catalysts changes therelative chemisorptions of CO and H and that alkalizationactivates the surfaces for COz chemisorption 2 4 . 4 6 ) .Carbon dioxide hydrogenation to carbon monoxide [E-4] is areversible reaction and leads to equilibrium. The equilibrium isindependent of the pressure, but is very much influenced by thetemperature. In the temperature ranges useful for th e FTreaction, the equilibrium is not favorable. Figure 2 illustratesthe equilibrium COz conversions as a function of the temperature

for 1 : 1 , 3 : l and 4:l H:CO gas compositions. Higher otconversion can be obtained if the HZ reagent is used instoichiometric excess. The equilibrium will be also favorablyshifted if th e CO i s removed from the system. This happens duringreductions to the hydrocarbon stage.THE FISCHER-TROPSCH REACTION STEP.

The F T reaction E - 3 ) has been very extensively studiedbecause of its commercial significance and because of itsscientific complexity and diversity. This brief review will berestricted to certain aspects of F T chemistry which are relevantto our objectives.In first approximation, the products of the FT synthesis aredefined by a single parameter, the chain growth probability ( a o ralpha) according to th e ASF equation:

(E-5)where C, is t he carbon selectivity (mass fraction in the idealcase when the products are olefins) of th e product with n carbonnumber and d i s th e chai n growth probability. In practice, amultiplicity of d s is produced, but an averaged 6 stillreasonably defines the products unless the range of t he Q-s isvery broad (47). Deviations from the AFS distribution have beenwidely reported. Some of the deviations are predictable and welldefined (48); others, notably t he'C1 selectivities, are not welldefined.

C, = (In' d n

For the purpose of this treatment, it is proposed, thatmethanation E-2) is an extreme case of the FT reaction (E-3)when the chain growth probability value is zero or very low. Thisunderstanding seems to be supported by t he numerous reports thatsmall amounts of ethane and propane are usually also observedduring methanation. Th e methanation catalysts are very activehydrogenation catalysts, and they hydrogenolize the metal-C1intermediates on th e catalyst surface before they could grow.Furthermore, the methanation catalysts can also hydrogenolize thehigher hydrocarbons already formed, which reactions also producemethane. Because of the se reactions, the ASF equations mayincreasingly fail to define the product distributions as thechain growth probability value decreases.

Recently we have proposed for Co/SiOl catalysts (47), that th echain growth probability is a function of the catalyst, of t hereagent and inert concentrations and of the temperature of t hecatalyst surface:

Even though the function f cannot be defined, it may bebeneficial to review our qualitative knowledge about the factorswhich together should define d. In E-6, C is the catalyst factorwhich is composed of numerous elements. Th e catalytic metal isimportant. Co, F e and Ru are known to be able to produce veryhigh &values. There are reports in the literature (49-51)682


4/7

suggesting that the dispersion of the metal can influence chaingrowth. Promoters incorporated into the catalysts can alsoinfluence chain growth. Alkali metal salts, particularly K Salts,were found to greatly increase chain growth (13,40-43). Inaddition, alkalization had a tremendous influence on th e reactioncharacteristics by changing the relative strengths of H I CO andCot chemisorptions. The hydrogenating character of the catalystwas reduced by alkalization, resulting in high olefin yields.

represent t he concentrations of t he reagentsand inerts (including products). The question is how to def inethese concentrations i n light of the knowledge, that in most FTreactions diffusion controls the rates (47). Due to complexdiffusion effects, the concentrations of the components in theimmediate vicinity of the catalyst surface might be quitedifferent, than their concentrations in the bulk gas phase. Toeliminate the need for considering diffusion effects, S rconcentrations represent the concentrations of componenvs A to 2in the immediate vicinity of the catalyst surface. Th e values ofS A . . . S ~ are related to their respective bulk gas phaseconcentrations and are dependent on the prevailing diffusionalconditions. Of course, their values can be changed by changingthe pressure of the system. Qualitative examples on the influenceof component concentration, pressure, diffusion on the chaingrowth probability are available in the literature. Thus,increasing H 'CO ratio was shown to give lower alpha values (47).Dilution of l'he feed with inert gases was also sho wn to result inlower chain growth probability (47). Diffusional changes werealso suggested for observed changes i n rate and chain growthprobability (52).

In E-6, SI . .

The influence of the reaction temperature (T) on t he value ofthe chain growth probability has been long known. Recently wehave shown, that over Co catalysts, the alpha value sharplydecreases with increasing T (47). Over alkalized F e catalysts, asreviewed by Dry (43), the effect of T appears to be much moregradual. With these catalysts, chain growth probability of about0.7 can be obtained even over 3OO0C. In Table 1 , we can s ee anexample of 0.72 chain growth probability from COl reduction at400 C over a heavily alkalized Fe catalyst.CATALYST AND PROCESS DESIGN REQUIREMENTS.

From the above review it is clear, that a combination ofappropriate catalyst design and process design is required forobtaining high molecular weight FT products i n CO reductions.The catalyst must contain a WGS component and a F\ component. TheWGS component must provide fast rates for CO formation andaccumulation. Furthermore, the surfaces must be modified forobtaining a proper balance in the chemisorptions of COl, CO andH Concerning the process design, the process parameters (T, P,Sl', feed composition) need t o be optimized for th e individualcatalyst to provide the most favorable H 'CO ratios on thecatalyst surface for high molecular weigkt FT products.Conceptually, diffu sio n control might also serve to regulate theH2:C0 ratio. I f gas phase d iffu sion controls the reagentconcentrations on the catalyst surface, the surface is expectedto be enriched in hydrogen, because of its high diffusivityarising from its small molecular s ize [ 5 2 ] . If d iffu sion throughliquids were to control the reagent concentrations on thecatalyst surface, the excessive hydrogen concentration on thecatalyst surface may be avoided, due to differences in thesolubilities of the reagents in hydrocarbon liquids [53]. We areoptimistic that utilization of knowledge in catalyst and processdesign will lead to significant increases i n the ASF growthprobability values during Cot reductions.REFERENCES.1 J.M. Fox, Cata1.Rev.-Sci.Eng., 35 (1993) 169.2 M.D. Jackson and C.B. Moyer i n Encyclopedia of ChemicalTechnology, 4th Ed., Vol 1, p.826. Wiley, New York, 1991.3 P.J.A. Tijm, ACS Fuel Division Preprints, 39 (1994) 1146.4 I. Puskas, Chemtech, December 1965, p.43.5 C.V. Krylov, A.Kh. Mamedov, Rus.Chem.Rev., 64 (1995) 877.

683


5/7

6 X. Xiaoding, J.A. Moulijn, EnergybFuels 10 ( 1 9 9 6 ) 3 0 5 .7 P.J. Jessop, T. Ikariya, R. Noyori, Chem.Rev., 9 5 ( 1 9 9 5 ) 2 5 9 .8 K. Tanaka in Advances in Inorganic Chemistry, Ed. A.G. Skyes,9 J.H. Edwards, Catal.Today, 2 3 , 5 9 ( 1 9 9 5 ) .Vol. 4 3 , p. 4 0 9 . Academic Press, San Diego, 1 9 9 5 .10 M.L. Kastens, J.F. Dudley, J Troeltzsch, Ind.Eng.Chem., 4 0

11 Anonymous, Chem.Eng., February 11 1 9 8 0 , p. 4 9 .1 2 H. Kuester, Brennst.Chem., 1 7 ( 1 9 3 6 ) 2 0 3.1 3 H. Kuester, Brennst.Chem., 1 7 ( 1 9 3 6 ) 2 2 1.1 4 P.H. Choi, K-W. Jun, S-J Lee, M-J Choi and K-W Lee, Catal.1 5 J.A. Fisher and A.T. Bell, J.Catal., 1 6 2 , ( 1 9 9 6 ) 5 4 .1 6 0 Froehlich, U. Kestel, J. Lojewska. T. Lojewski, G. Meyer,

( 1 9 4 8 ) 2 2 3 0 .

Lett., 4 0 ( 1 9 9 6 ) 1 1 5 .M. V o s s D. Borgmann, R. Dziembaj, G. Wedler, Appl.Catal.,A 1 3 4 ( 1 9 9 6 ) 1.Kobayashi, Appl,Catal., A 1 3 7 ( 1 9 9 6 ) 2 5 5 .Preprints 4 1 ( 1 9 9 5 ) 1 4 1 1 .Chem.Commun., 1 9 9 5 , p. 1 0 2 5 .Catal.Lett., 3 5 ( 1 9 9 5 ) 2 7 1 .Techn. 1 7 ( 1 9 9 4 ) 3 9 0 .

1 7 S . Mori, W-C. Xu, T. Ishidzuki, N. Ogasawara, J Imai, K.1 8 K-W Jun, S-J Lee, M-J Choi, K-W Lee, ACS Fuel Division1 9 Y. Huang, X. Meng, Z. Dang, S. Weng and C. Zhang, J.Chem.Soc.2 0 Y. Kou, Z-h. Suo, J-z. Niu, W-e. Zhang, H-1. Wang,2 1 U. Kestel, G. Froehlich, D. Bergmann, 0 Wedler, Chem.Eng.2 2 J-F. Lee, F-S. Chern, M-D. Lee and T-Y Dong, Can.J.Chem.Eng.,7 0 ( 1 9 9 2 ) 5 1 1 .

1 2 9 .2 3 T. Kai, T. Matsumura, T. Takahashi, Catal.Lett., 1 62 4 F. Solymosi, ~.Mol.Catal., 6 5 ( 1 9 9 1 ) 3 3 7 .2 5 T. Inui, T.. Takeguchi, Catal.Today, 10 ( 1 9 9 1 ) 9 5 .2 6 T. Suzuki, K. Saeki, Y. Mayama, T. Hirai, S. Hayashi2 7 C-K. Kuei, M-D. Lee, Can.J. Chem. Eng., 6 9 (1 9 9 1 ) 3 42 8 M-D Lee, J-F Lee, C-S. Chang. Bull.Chem.Soc.Jpn., 6 22 9 F. Noeaki, T. Sodesawa, S. Satoh, K. Kimura, J.Cata1

Kinet.Catal.Lett., 4 4 ( 1 9 9 1 ) 4 8 9 .2 7 5 6 .( 1 9 8 7 ) 3 3 9 .3 03 1

3 23 33 43 53 6

3 73 83 94 04 14 24 34 44 54 647

1 9 9 2 )

React.i 1 9 8 9 )

1 0 4K.R. Thampi, J. Kiwi, M. Graetzel, Nature, 3 2 7 ( 1 9 8 7 ) 5 0 6 .A.D. Thomsett, T. Hagiwara, A. Miyamoto, T.Inui, Appl.Catal., 2 6 ( 1 9 8 6 ) 3 9 1 .G.D. Weatherbee, C.H. Bartholomew, J.Catal., 8 7 , ( 1 9 8 4 ) 3 5 2.J Barrault, C. Forquy, J.C. Menezo, R. Maurel, React.Kinet.Catal. Lett., 1 7 ( 1 9 8 1 ) 9.13.J.L. Falconer, E. Zagli, J.Catal., 6 2 1980) 2 8 0 .D.J. Dwyer, G.A. Somorjai, J.Catal., 7 ( 1 9 7 8 ) 2 9 1 .C.V. Ovesen, B.S. Clausen, B.S. Hannnershoi, 0 Steffensen, T.Askgaard, I Chorkendorff, J.K. Norskov, P.B. Rasmussen, P.Stoltze, P. Taylor, J.Cata1.. 1 5 8 ( 1 9 9 6 ) 1 7 0 .C. Rhodes, G.J. Hutchings, A.M. Ward, Catal.Today, 2 3 ( 1 9 9 5 )4 3 .S-I. Fujita, M. Usui, N. Takezawa, J.Catal., 1 3 4 ( 1 9 9 5 ) 2 2 0 .D.S. Newsome, Catal.Rev.-Sci.Eng., 2 1 ( 1 9 8 9 ) 2 7 5 .D.B. Bukur, D. Mukesh, S.A. Patel, Ind.Eng.Chem.Res., 2 9( 1 9 9 0 ) 1 9 4 .L. Koenig, J. Gaube, W. Meisel, P. Guetlich, W. Gerhard, C.Plog, Ber.Bunsenges.Phys.Chem., 91 ( 1 9 8 7 ) 1 1 6 .G. Henrici-Olive, S. Olive, J.Mol.Catal., 1 6 ( 1 9 8 2 ) 1 1 6 .M.E. Dry in Catalysis Science and Technology, J.R. Andersonand M. Boudart, Eds.. Springer Verlag, Berlin, 1 9 8 1 , Vol 1,p.159.M.E. Dry, T. shingles, L. Boshoff, G.J. Oosthuieen, J.Catal.,1 5 ( 1 9 6 9 ) 1 9 0 .J Benzinger, R. Madix, Surf.Sci. 9 4 ( 1 9 8 0 ) 1 1 9 .G. Meyer, D Borgmann. 0 . Wedler, Surf.Sci., 3 2 0 ( 1 9 9 4 ) 1 2 3 .R.S. Hurlbut, 1. Puskas, D.J. Schumacher, EnergybFuels, 10( 1 9 9 6 ) 5 3 7 .48 I Puskas, R.S. Hurlbut, R.E. Pauls, J.Catal., 1 3 9 ( 1 9 9 3 ) 5 9 14 9 A.S. Lisitsyn, A.V. Golovin. V.L. Kuznetsov, Yu.1. Yermakov,

50 K. Fujimoto, T. Nobusawa, T. Fukushima, H. Tominaga, Bull.5 1 C.S. Kellner, A.T. Bell, J.Cata1.. 7 5 ( 1 9 8 2 ) 2 51 .52 I Puskas, B.L. Meyers, J.B. Hall, Catal.Today, 2 1 ( 1 9 9 4 ) 2 4353 J.S. Chou and K-C. Chao, 1nd.Eng.Chem. Res., 3 1 ( 1 9 9 2 ) 6 2 1 .

C1 Mol.Chem., 1 ( 1 9 8 4 ) 1 1 5 .Chem.Soc.Jpn., 5 8 ( 1 9 8 5 ) 3 1 64 .

684


6/7

Table 1. Reported Examples of Carbon DioxideReductions to Higher FT HydKOCaKbonS.

C.45L

4 0s 3 5

Cat Y S tFe-K/Al2O1 400 2026 69.6; 66.9 0.72 14Fe-K/A1201 300 1013 57.7; 50.4 0.66 18Fe-Mn-K 320 1013 33.8; 29.0 0.56 22Fe- K 320 1013 34.7; 28.4 0.6 5 28Fe-Cu-KCl/ 270 1520 10.0; 5.3 0.68 33Coi01 A120kFe-Cu, 150-250 101 ? se e text 13Fused Fe-ZSM-5 350 2100 38.1; 32. 6 se e text 27Rh/NbjOl 350 101 11; 10 0.21 29The Hz/CO1 feed ratios varied be tween 4:l and 1:l.bThe flrst number gives the total conversion (C O t hydrocarbons);the second number the conversion to hydrocarbons.Our best estimates of t he chain growth probabilities from thereported data, unless provided in the publication.

Figure 1. Conceptual Methanol-Syncrude Coproduction Scheme.

IIgases

:l H2:CO3:l H :CO

4:l H :CO

Tempera tu re C

685


7/7

Table 1. Reported Examples of Carbon DioxideReductions to Higher FT Hydrocarbons.'

Catal YS tFe-K/AlFe-K/A1203Fe-Mn-KFe-KFe-Cu-KC1/TiO1-A120Co,Fe-Cu,kFused Fe-ZSMRh/NblOi

(kPa)400 2026300 1013320 1013320 1013270 1520150-250 101-5 350 2100350 101

coconverslonb avalue' Reference69.6; 66.9 0.72 1457.7; 50.4 0.66 1833.8; 29.0 0.56 2234.7; 28.4 0.65 2810.0; 5.3 0.68 33? see text 1338.1; 32.6 see text 2711; 10 0.21 29

;The H1/C02 feed ratios varied between 4:l and 1:l.The first number gives t he total conversion (CO t hydrocarbons);the second number the conversion to hydrocarbons.'Our best estimates of the chain growth probabilities from thereported data, unless provided in the publication.

Figure 1. Conceptual Methanol-Syncrude Coproduction Scheme.

Irecycle loop

Figure 2 CARBON DIOXIDE CONVERSIONSIN REVERSE W G S EQUILIBRIA*VI 4 50 1

,150 2 0 0 2 5 0 300 300Temperature, C

686

1:l H2:CO:l H2:CO

4 : l H2:CO

Documents

Can Carbon Dioxide Be Reduced to High Molecularweight Fischer-tropsch Products