A new and direct preparation method of iron-based bimodal catalyst and itsapplication in Fischer–Tropsch synthesis

Yi Zhang a, Jun Bao b, Satoshi Nagamori a, Noritatsu Tsubaki a,c,*a Department of Applied Chemistry, School of Engineering, University of Toyama, Gofuku 3190, Toyama 930-8555, Japanb National Synchrotron Radiation Laboratory, University of Science and Technology of China, Heifei 230026, PR Chinac JST, CREST, Sanbancho 5, Chiyoda-ku, Tokyo 102-0075, Japan

Applied Catalysis A: General 352 (2009) 277–281

A R T I C L E I N F O

Article history:

Received 27 May 2008

Received in revised form 8 October 2008

Accepted 11 October 2008

Available online 21 October 2008

Keywords:

Fischer–Tropsch synthesis

Iron-based catalyst

Bimodal structure

Self-organization

A B S T R A C T

A bimodal iron-based catalyst was prepared by a new one-step impregnation method. The active

components were used as the ‘‘brick’’ to directly build the small pores inside the large pores of support,

which was quite different from the previous bimodal catalysts that were prepared once more on a

bimodal support. Comparing with the unimodal catalysts and conventional co-precipitated catalyst, the

prepared bimodal catalyst exhibited excellent activity in Fischer–Tropsch synthesis due to the improved

active metal dispersion and fastened diffusion efficiency. This preparation method is much simpler than

the previous methods and can be extended to prepare various bimodal catalysts with different chemical

compositions.

� 2008 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Applied Catalysis A: General

journa l homepage: www.e lsev ier .com/ locate /apcata

1. Introduction

The reaction performance of a solid catalyst was dependent onmany factors such as support property, metal precursor identityand calcination temperature. Among these factors, the porestructure of catalyst has a significant effect on the diffusionbehavior of reactants and products, which affects not only thecatalytic activity but also product selectivity. In general, largespecific surface area favors high dispersion and creates more activesites, leading to a high catalytic conversion. However, the catalystwith large surface area usually contains small pore size, whichresults in poor intra-pellet diffusion efficiency of reactants andproducts. A bimodal structure of catalyst with both large pores andsmall pores has excellent advantages for solving this contradictionbecause the large pores lead to a high diffusion rate of reactantsand products while the small pores provide a large surface area anda high metal dispersion [1].

The reported bimodal catalysts have been prepared by a two-step process: the synthesis of bimodal pore support andsubsequent impregnation of active components. Both the large

* Corresponding author at: Department of Applied Chemistry, School of

Engineering, University of Toyama, Gofuku 3190, Toyama 930-8555, Japan.

Tel.: +81 76 445 6846; fax: +81 76 445 6846.

E-mail address: tsubaki@eng.u-toyama.ac.jp (N. Tsubaki).

0926-860X/$ – see front matter � 2008 Elsevier B.V. All rights reserved.

doi:10.1016/j.apcata.2008.10.017

and small pores of the final obtained catalysts were actuallyderived from the bimodal support and almost not affected by theintroduction of the supported active components. Until now, mostof the efforts have focused on the preparation of supports withbimodal pore structure. Previous studies of our group havedeveloped a simple method to prepare the multi-functionalbimodal support by building up nano-particles from sol or solutionof support precursor to form small pores inside the large pores ofhomo- or hetero-atom support. The cobalt catalysts derived fromthe prepared bimodal supports exhibited outstanding activity inFischer–Tropsch synthesis (FTS) [2–4]. However, the two-steppreparation has the basic disadvantage that a considerable amountof active components is difficult to enter the small pores of bimodalsupport during the impregnation, which limits the metal disper-sion on the surface of support, and finally the catalytic activity. Tofurther increase the reaction performance and simplify thepreparation of bimodal catalysts, a new method, directly usingthe catalytic active components for generating small pores insidethe large pores of support, is proposed here. For FTS using syngasfrom coal gasification, biomass gasification or CO2 reforming ofnatural gas, the iron-based catalysts are attractive because of theirhigh FTS and water–gas shift (WGS) activity, as well as excellentsulfur tolerance [5,6]. Generally, the reported iron-based FTScatalysts have been prepared by co-precipitation of the metal salts.In this study, supported iron bimodal catalyst containing copperand potassium promoters was prepared based on the proposed

Fig. 1. Pore size distributions of the prepared Fe–Cu–K/SiO2 catalysts and

corresponding supports.

Table 1Physical properties of various Fe–Cu–K/SiO2 and corresponding supports.

Catalyst Surface area (m2/g) Pore volume (ml/g) Pore size (nm)

Q-50 support 80.9 1.26 61.8

Fe–Cu–K/Q-50 94.8 0.47 6.2, 56.3

Fe–Cu–K/Q-50a 75.1 0.78 61.3

Q-3 support 785 0.51 3.7

Fe–Cu–K/Q-3 429 0.30 5.1

a Calcination at 973 K.

Y. Zhang et al. / Applied Catalysis A: General 352 (2009) 277–281278

method through a simple one-step impregnation and its perfor-mance for FTS was examined. The obtained catalyst was quitedifferent from the conventional supported catalyst as dispersedmulti-component active sites formed mesoporous structure byself-organization, ensuring much higher catalytic activity than thatof the precipitated catalyst and unimodal supported catalyst. Thisnewly developed method is a simple and tailor-made method toprepare supported catalyst with bimodal structure and can beextended to prepare various bimodal catalysts with differentchemical compositions.

2. Experimental

2.1. Catalyst preparation

The bimodal Fe–Cu–K supported catalyst was prepared byincipient wetness co-impregnation of the support with mixedaqueous solutions of Fe(NO3)3�9H2O, Cu(NO3)2�3H2O and KNO3

(Kanto Chemicals Co.). The iron loading was 20 wt%. The molarratio of Fe:Cu:K was 200:30:5. The support was commerciallyavailable silica gel Q-50 (Cariact, Fuji Silysia Co.). After impreg-nation, the precursor was dried in air at 393 K for 12 h, and thencalcined in air from room temperature to 673 K in 3 h andmaintained at 673 K for 2 h. For comparison, a catalyst with thesame composition but using silica gel Q-3 with small pore (FujiSilysia Co.) as a support was prepared by the same method.

2.2. Catalyst characterization

X-ray diffraction (XRD) patterns were measured using a RigakuRINT-2000 diffractometer with Cu Ka radiation (la = 0.154 nm).Scanning electron microscopy (SEM) images were obtained onJSM-6700F field emission scanning electron microscope (FE-SEM).BET surface area, pore size distribution and pore volume weredetermined by the adsorption method (Quantachrome AUTO-SORB-1) with N2 as the adsorbent.

2.3. Catalyst testing

The FTS reaction was conducted in a continuous flow type fixed-bed reactor under the temperature of 553 K and a pressure of1.0 MPa. The feed gas consisted of Ar (3.08%), CO (45.1%) and H2

(51.82%). 0.5 g catalyst mixed with 2.5 g quartz sand was placed inthe reactor tube and reduced in situ at 573 K for 10 h in the flow offeed gas at 0.1 MPa, followed by cooling down to 353 K in N2. Whenthe reaction temperature was reached, pressurized syngas wasintroduced and the reaction was carried on continuously for morethan 10 h. A detailed account of the experimental setup andproduct analysis has been reported elsewhere [7].

3. Results and discussion

The pore size distributions of the prepared catalysts andcorresponding supports are shown in Fig. 1. The silica gel Q-50 andQ-3, as expected, had the uniformly distributed pore with size of61.8 nm and 3.7 nm, respectively. After impregnation with mixedsolutions of iron nitrate, copper nitrate, potassium nitrate andcalcination, the pore size distribution of the obtained Fe–Cu–K/Q-3catalyst did not show a significant change, while the Fe–Cu–K/Q-50exhibited distinctly two types of pores with size of 56.3 nm and6.2 nm, respectively, reflecting the bimodal structure. The largepore was from the intrinsic pore of the Cariact Q-50 pellet, but thesize slightly decreased from 61.8 nm to 56.3 nm. The new smallpore with 6.2 nm diameter was formed by the introduced iron,copper and potassium species. The physical properties of the

supports and catalysts were shown in Table 1. Due to the formationof small pores, the BET surface area of the bimodal Fe–Cu–K/Q-50catalyst increased from 80.9 m2/g to 94.8 m2/g. These indicatedthat the introduced components were not deposited on theentrance of large pores to block the pores; otherwise the BETsurface area would be lower. The more important fact was that thepore volume of the catalyst decreased significantly from 1.26 ml/gto 0.47 ml/g, proving that these components entered the largepores of silica gel rather than deposited on the outer surfaces.Based on the above findings, it can be concluded that for the Fe–Cu–K/Q-50 catalyst, its distinct bimodal structure was formed bythe introduced active species to build the small pores on the inner

Fig. 2. XRD patterns of the Fe–Cu–K/Q-3, Fe–Cu–K/Q-50 and 973 K calcined Fe–Cu–

K/Q-50 catalysts before in situ reduction. (*) Fe2O3; (*) SiO2.

Fig. 4. Representation of the bimodal pore structure.

Y. Zhang et al. / Applied Catalysis A: General 352 (2009) 277–281 279

wall of large pores of Q-50 support, as large pore size decreasedfrom 61.8 nm to 56.3 nm. Fig. 2 shows the XRD patterns of thethree supported Fe–Cu–K catalysts before in situ reduction. For thebimodal Fe–Cu–K/Q-50 catalyst, the diffraction peaks wereindexed to Fe2O3 and SiO2 species. After calcinations at 973 K,the diffraction peaks became sharper. The Fe–Cu–K/Q-3 catalystexhibited weaker diffraction intensity than the Fe–Cu–K/Q-50catalyst and only Fe2O3 species were detected, indicating that theactive species had a higher dispersion on the large surface areasupport. No peaks assigned to copper or potassium species wereobserved on these samples, which may be due to their low contentsand high dispersion.

The microstructure of the Fe–Cu–K/Q-50 catalyst was studiedby FE-SEM. For comparison, the SEM image of Q-50 support wasalso included. As shown in Fig. 3a, the pores of the Q-50 supportwere clearly observed and the pore size was about 50 nm. After

Fig. 3. FE-SEM image of (a) Q-50 support and (b) Fe–Cu–K/Q-50 catalyst.

impregnation with iron, copper and potassium species, the innersurface of the pores of the obtained Fe–Cu–K/Q-50 had a ruggedstructure (Fig. 3b, arrow), displaying a bimodal pore structure asillustrated in Fig. 4. These results revealed that the introducedactive species indeed entered the large pores of Q-50 silica gel andacted as the ‘‘‘brick’’ to build the small pores inside the large poresof support, which was quite different from the previous bimodalcatalyst where the bimodal pore catalyst support was prepared atfirst, followed by the impregnation of active component. Aftercalcination at 973 K, the small pores of the bimodal catalystdisappeared and BET surface area decreased to 75.1 m2/g. On thecontrary its pore volume increased to 0.78 ml/g remarkably. Thisindicated that the heat treatment at 973 K resulted in the collapseof small pores and agglomeration of active components. Corre-spondingly, the large pore size of the Fe–Cu–K/Q-50 catalystcalcined at 973 K restored from 56.3 nm to 61.3 nm, very near tothe original value of 61.8 nm. For the Fe–Cu–K/Q-3 catalyst,compared with Q-3 support, its BET surface area and pore volumedecreased significantly while the pore sizes did not changeobviously, indicating that the introduced species did not enterthe small pores of Q-3 and might be deposited on the entrance ofsmall pores and block the pores.

The prepared iron-based bimodal catalyst was applied to FTSreaction to investigate the promotional role of bimodalstructure. Table 2 shows the reaction performances of iron-based catalysts prepared from Q-50 and Q-3 support. Silicasupport Q-3 is analogous to Q-50 but with average pore size ofonly 3.7 nm. The Q-3 supported catalyst with the largest surfacearea exhibited the lowest catalytic activity, indicating thediffusion-controlled regime due to its small pores. The bimodalFe–Cu–K/Q-50 catalyst showed a higher CO conversion com-pared with the one derived from Q-3, although its surface areawas much lower. After calcination at 973 K, the CO conversion ofthe obtained catalyst decreased significantly. The selectivities ofCH4 and CO2 did not change obviously. The chain growthprobability of the three catalysts was 0.65, 0.57 and 0.61,respectively. The Q-3 supported catalyst had the lowest chaingrowth probability, proving that the small pore catalyst tendedto produce lighter hydrocarbons, while the large pore wasfavorable for chain propagation [8]. The bimodal Fe–Cu–K/Q-50catalyst showed the best reaction performance, demonstratingthe promotional effect of bimodal structure. It is important tonote that unlike the previous bimodal catalysts that preparedafter the formation of bimodal support, the small pores in thepresent bimodal catalyst were tailor-made formed directly bycatalytic active components themselves, which suggested ahigher metal dispersion and active surface area. Comparing the

Table 2The reaction performance of various Fe–Cu–K/SiO2 catalystsa.

Catalyst CO conversion (%) Selectivity (%) ab

CH4 CO2

Fe–Cu–K/Q-3 32.8 9.2 36.7 0.57

Fe–Cu–K/Q-50 53.7 10.8 37.1 0.65

Fe–Cu–K/Q-50c 45.5 9.6 34.1 0.61

a Reaction conditions: 553 K, 1.0 MPa, W/F = 5 g h/mol.b Chain growth probability.c Calcination at 973 K.

Table 3The reaction performance of bimodal Fe–Cu–K/Q-50 and co-precipitated Fe–Cu–K/

SiO2 catalysts with the same chemical compositiona.

Catalyst CO conversion (%) Selectivity (%) ab

CH4 CO2

Fe–Cu–K/Q-50 89.5 10.5 37 0.60

Fe–Cu–K/SiO2 13.5 12.7 14.6 0.61

a Reaction conditions: 553 K, 1.0 MPa, W/F = 10 g h/mol.b Chain growth probability.

Y. Zhang et al. / Applied Catalysis A: General 352 (2009) 277–281280

two unimodal catalysts, Fe–Cu–K/Q-3 and high temperaturetreated Fe–Cu–K/Q-50, the latter with only the large poreshowed a higher CO conversion although its specific surface areawas much lower. This fact indicated that in this case, the poresize of catalysts played a more important role rather thansurface area on the catalytic activity. The reaction rate in slurryphase here was mainly controlled by the diffusion of reactantsand products. The variations of the activity and selectivity of thebimodal Fe–Cu–K/Q-50 catalyst are shown in Fig. 5. The catalystshowed stable high activity and low methane selectivity for 10 hcontinuously. The CO conversion increased gradually at theinitial stage and reached the steady state after about 5 h onstream. Similar results were reported in the previous studies [9–11]. Riedel et al. [10] found that time on stream had dominantinfluence on composition and structure of the catalyst. It tookseveral episodes for the reduced fresh iron catalyst to constructthe actual catalytic phase and reach the steady state. Theyconcluded that the FT activity of iron catalysts was generatedwith time, when the a-Fe reacted with carbon from COdissociation and formed the iron carbide (particularly Fe5C2),which was addressed as the true active species for FTS.

For the iron-based catalysts used for FTS, most of the previousstudies have focused on the co-precipitated catalysts and only fewreports have concerned the impregnated samples. Iron catalystsprepared by impregnation of support usually utilize strongattrition resistance property of the support material, whichmay show superior attrition resistance than a precipitated ironcatalyst and merit consideration for industrial use in FTS [12].However, due to the formation of chemical compounds betweenalkali promoter and support reducing the alkali promotionaleffect, the supported iron catalysts are less active compared withthe co-precipitated iron catalysts [13]. Anderson has reported thatiron catalysts with high support-to-metal ratios are generallyineffective FT catalysts, and highly dispersed supported Fe isdifficult to prepare [14]. To improve the dispersion and loading of

Fig. 5. Variations of the activity and selectivity of bimodal Fe–Cu–K/Q-50 catalyst

with reaction time. (&) CO conversion; (*) CH4 selectivity; (&) CO2 selectivity.

Reaction conditions: 553 K, 1.0 MPa, W/F = 5 g h/mol.

metal of the supported iron catalysts, large surface area supportshave been generally used [11,12,15]. Furthermore, multipleimpregnation steps were sometimes necessary due to the smallpore size of supports [12,16]. The presented iron catalyst wasdifferent from the conventional supported catalyst. Its prepara-tion was based on the large pore support and the dispersed multi-component active sites formed mesoporous structure by self-organization inside large SiO2 pore. The formation of bimodal porestructure realized not only high metal dispersion, but also highdiffusion efficiency of reactants and products, which ensuredmuch higher catalytic activity than that of the unimodalsupported catalyst. In this study, for comparison, an iron catalystwith the same chemical composition was prepared by theconventional co-precipitation method [17], and its performancefor FTS was also tested under the same conditions. The BET surfacearea of the co-precipitated Fe–Cu–K/SiO2 was 218 m2/g and thepore size and pore volume was 8.8 nm and 0.29 ml/g, respectively.As compared in Table 3, the precipitated iron catalyst had a quitelow CO conversion and CO2 selectivity. The supported bimodal Fe–Cu–K/Q-50 exhibited excellent FTS activity under the samereaction conditions. Its CO conversion (W/F = 10 g h/mol) wasnear 90%, about seven times higher than that of the co-precipitated sample. The high CO2 selectivity (�40%) indicatedthat the catalyst had a high WGS activity. These results suggestedthat the supported bimodal iron catalyst is a highly activecatalytic material for FTS.

4. Conclusion

An iron-based catalyst with distinct bimodal pore structure wasprepared by directly using the active components for building upsmall pores inside the large pores of support. Due to the improvedactive metal dispersion and fastened diffusion efficiency, theprepared catalyst exhibited much higher activity in FTS than thatof the unimodal catalysts and conventional co-precipitatedcatalyst. This preparation is much simpler than the previousmethods and can be extended to prepare various bimodal catalystswith different chemical compositions.

References

[1] Levenspiel (Ed.), Chemical Reaction Engineering, 2nd ed., Wiley, New York, 1972,p. 496.

[2] B.L. Xu, Y.N. Fan, Y. Zhang, N. Tsubaki, AIChE J. 51 (7) (2005) 2068.[3] Y. Zhang, M. Koike, R.Q. Yang, S. Hinchiranan, T. Vitidsant, N. Tsubaki, Appl. Catal.

A 292 (2005) 252.[4] Y. Zhang, Y. Yoneyama, N. Tsubaki, Chem. Commun. 11 (2002) 1216.[5] A. Demirbas, Prog. Energy Combust. Sci. 33 (2007) 1.[6] K.W. Jun, H.S. Roh, K.S. Kim, J.S. Ryu, K.W. Lee, Appl. Catal. A 259 (2004) 221.[7] J. He, Z.L. Liu, Y. Yoneyama, N. Nishiyama, N. Tsubaki, Chem. Eur. J. 12 (2006)

8296.[8] L. Fan, K. Yokota, K. Fujimoto, AIChE J. 38 (1992) 1639.[9] J. Yang, Y.C. Sun, Y. Tang, Y. Liu, H.L. Wang, L. Tian, H. Wang, Z.X. Zhang, H.W.

Xiang, Y.W. Li, J. Mol. Catal. A 245 (2006) 26.[10] T. Riedel, H. Schulz, G. Schaub, K.W. Jun, J.S. Hwang, K.W. Lee, Top. Catal. 26 (2003)

41.[11] J. Xu, C.H. Bartholomew, J. Sudweeks, D.L. Eggett, Top. Catal. 26 (2003) 55.

Y. Zhang et al. / Applied Catalysis A: General 352 (2009) 277–281 281

[12] R.J. O’Brien, L. Xu, S. Bao, A. Raje, B.H. Davis, Appl. Catal. A 196 (2000) 173.[13] M.E. Dry, in: J.R. Anderson, M. Boudart (Eds.), Catalysis Science and Technology,

vol. 1, Springer, New York, 1981, p. 160.[14] R.B. Anderson (Ed.), The Fischer–Tropsch Synthesis, Academic Press, New York,

1984, p. 147.

[15] L.H. Yu, X.X. Zhang, Z.J. Du, D. Wang, S.R. Wang, S.H. Wu, J. Nat. Gas Chem. 16(2007) 46.

[16] D.B. Bukur, C. Sivaraj, Appl. Catal. A 231 (2002) 201.[17] H.J. Wan, B.S. Wu, Z.C. Tao, T.Z. Li, X. An, H.W. Xiang, Y.W. Li, J. Mol. Catal. A 260 (1–

2) (2006) 255.