Methane Decomposition: Production of Hydrogen and Carbon

of hydrogen production include autothermal reforming and partial oxidation. However, all these processes involve the formation of large amount of CO...
3 downloads 745 Views 2MB Size
Download PDF
Love PNG Images
Methane Decomposition: Production of Hydrogen and Carbon Filaments BY T.V. CHOUDHARYa AND D.W. GOODMANb a ConocoPhillips Company, Bartlesville Technology Centre, Bartlesville 74004, USA b Department of Chemistry, Texas A&M University, College Station, TX77843, USA

1

Introduction

Hydrogen, presently, finds application as a chemical rather than a fuel in commercial operations. However, being a non-polluting source of energy, hydrogen is predicted to be the ‘‘fuel of the future’’.1 One of the most potential applications for hydrogen is to power fuel cells. Major automobile manufacturers are currently working towards developing fuel cell vehicles; such vehicles are expected to significantly curtail the pollution from the transportation sector. Fuel cells, because of their modular nature, can be utilized to provide heat and electricity not only to single homes but also to provide a large amount of electricity to a large grid network. Fuel cells can be broadly classified into two types; high temperature fuel cells such as molten carbonate fuel cells (MCFCs) and solid oxide polymer fuel cells (SOFCs), which operate at temperatures above 923 K and low temperature fuel cells such as proton exchange membrane fuel cells (PEMs), alkaline fuel cells (AFCs) and phosphoric acid fuel cells (PAFCs), which operate at temperatures lower than 523 K. Because of their higher operating temperatures, MCFCs and SOFCs have a high tolerance for commonly encountered impurities such as CO and CO2 (COx). However, the high temperatures also impose problems in their maintenance and operation and thus, increase the difficulty in their effective utilization in vehicular and small-scale applications. Hence, a major part of the research has been directed towards low temperature fuel cells. The low temperature fuel cells unfortunately, have a very low tolerance for impurities such as COx; PAFCs can tolerate up to 2% CO, PEMs only a few ppm, whereas the AFCs have a stringent (ppm level) CO2 tolerance. Methane, due to its abundance and high H/C ratio (highest among all hydrocarbons) is an obvious source for hydrogen. Steam reforming of methane represents the current trend for hydrogen production. Other popular methods Catalysis, Volume 19 r The Royal Society of Chemistry, 2006

164

Catalysis, 2006, 19, 164–183

165

of hydrogen production include autothermal reforming and partial oxidation. However, all these processes involve the formation of large amount of COx as by-product.2–4 Hydrogen generated by these conventional methods can be utilized by the low temperature fuel cells only if COx (CO for PEM and CO2 for AFC) are completely eliminated (to ppm levels) from the stream prior to its introduction into the fuel cell. The process required to eliminate CO from the hydrogen produced in the steam reformer is briefly described below. The steam reformer products containing B10% CO (depending on the feedstock and conditions employed) are passed into water gas shift reactors (WGSs) where CO is reacted with water to form CO2 and hydrogen.5 Generally two WGS reactors are used in series (high temperature and low temperature) to minimize the amount of water. The WGS shift reactors are extremely bulky. Finally, the CO content is reduced to a few ppm in the preferential oxidation reactor (PROX). The hydrogen can be introduced in the fuel cell only after this circuitous procedure of removing CO. AFCs would additionally require removal of CO2 to ppm levels. Also, it is known that high levels of CO2 in the hydrogen stream can be detrimental for the performance of PEM fuel cells.6 Other conventional process of hydrogen production such as partial oxidation and auto-thermal reforming also entail similar procedures for COx removal. Removal of COx to ppm levels from the hydrogen stream makes the process extremely complex and bulky and thereby prohibits the use of the existing hydrogen production technology for use in vehicular and small-scale stationary fuel cell applications. Hydrogen production routes, which do not require complex COx removal procedures, are therefore desired for fuelling low temperature fuel cells. Recently, there has been a great deal of interest in investigating the catalytic decomposition of natural gas (whose major constituent is methane) for production of hydrogen. Since only hydrogen and carbon are formed in the decomposition process, separation of products is not an issue.7 The other main advantage is the simplicity of the methane decomposition process as compared to conventional methods. For example, the high- and low-temperature water-gas shift reactions and CO2 removal step (involved in the conventional methods) are completely eliminated. This review will address the following topics related to the methane decomposition process: (a) fundamentals of methane decomposition, (b) effect of support and promoters on the methane decomposition process (c) alternate reactor design for improving the process yields and (d) catalyst regeneration. Catalyst regeneration is extremely important for the practical application of the clean hydrogen production process; issues related to catalyst regeneration by steam, air and CO2 will be summarized separately. Since hydrogen production via methane decomposition is a relatively new field there are several unresolved issues. This review will attempt to bring forth these issues. Under certain process conditions, high yields of carbon filaments can be obtained on the catalyst during the catalytic decomposition of methane. Currently, there is a great interest in these carbon filaments, as the unique properties exhibited by these materials can be exploited in a number of applications such as catalyst support, energy storage devices, selective adsorption agents and

166

Catalysis, 2006, 19, 164–183

reinforcement materials. In this review we will focus on factors that influence carbon filament formation rates/yields and properties. This review will not address the (relatively low yield) synthesis of specialty single walled carbon nanotubes; a recent review on this topic can be found elsewhere.8 2

Hydrogen Production

The theoretical hydrogen formation reaction via decomposition of methane can be represented as: CH4 - 2H2 þ C

DH1073 ¼ 90.1 kJ/mol(CH4)

This moderately endothermic process results in the formation of 2 moles of hydrogen per mole of methane consumed above a certain threshold reaction temperature. A gradual catalyst deactivation is expected due to the accumulation of carbon on the catalyst. The catalyst can be regenerated by removing the carbon on the catalyst in a separate step. Thus, hydrogen production by this approach involves two distinct steps: (a) catalytic decomposition of methane and (b) regeneration of catalyst. ðaÞ CH4 !2H2 þ C ðbÞ C þ H2 O=O2 =CO2 !COx þ H2 and clean catalyst surface At the outset, this section will address studies related to the catalytic methane decomposition step and then subsequently describe the work undertaken on the combined step-wise reforming (two step) process. 2.1 Catalytic Decomposition of Methane for Hydrogen Production. – The methane decomposition reaction for hydrogen production has garnered considerable interest in the past 4–5 years. The recent interest in this approach for producing hydrogen stems from the stringent requirement of CO-free hydrogen for the proton exchange membrane fuel cells. For vehicular and small scale stationary applications, it is necessary that the fuel reformer be compact; this is difficult for the conventional processes since high CO conversion efficiencies require large water gas shift reactors. Recent studies have addressed different issues such as methane decomposition fundamentals, support/promoter effects and reactor design. The ensuing discussion will show that while some interesting issues about catalytic methane decomposition (as a method for generating pure hydrogen) have been uncovered, a significant amount of work still needs to be undertaken for better understanding this process. 2.1.1 Methane Decomposition Fundamentals. The fundamentals of methane decomposition have been extensively investigated on model-single crystal catalysts;9,10 an exhaustive review on this subject can be found elsewhere.11 Herein, only the studies undertaken on the fundamentals of methane

Catalysis, 2006, 19, 164–183

167

decomposition reaction on high surface area catalysts will be described. Otsuka and co-workers used hydrogen-deuterium exchange studies to investigate the methane decomposition reaction mechanism over Ni/SiO2.12 An isotopic effect was observed for the CH4/CD4 decomposition on the catalyst. Furthermore, substituted H–D methanes were not observed when a CH4–CD4 mixture was decomposed over the Ni/SiO2 catalyst. Based on this the authors suggested that the first C–H bond cleavage was the rate determining step for the decomposition of methane to carbon and hydrogen. In line with this theory, the authors also observed a reverse isotopic effect between hydrogen and deuterium when the carbon deposited on the catalyst was hydrogenated back to methane. The methane decomposition mechanism was further studied by performing the following set of experiments sequentially: (i) decomposition of 12CH4 (ii) decomposition of 13CH4 (iii) hydrogenation of deposited carbon. The studies showed that the carbon that was deposited last was hydrogenated first; thus indicating that there was no significant scrambling between the carbon atoms. Using an array of catalyst characterization techniques, the same group further investigated the structural changes of the Ni species in the Ni/SiO2 catalyst during the methane decomposition reaction.13 Prior to the reaction, the Ni species on the catalyst were in the metallic state. The Ni metal particles were found to aggregate [X-Ray Diffraction (XRD) studies] as soon the catalyst was contacted with methane. Following this initial aggregation at the onset of the methane decomposition reaction, no significant change in the structure of the Ni species was observed until towards the end of the reaction. During the rapid catalyst deactivation stage, Ni K-edge X-Ray Absorption Near Edge Structure (XANES) studies indicated the formation of Ni carbide species. As will be discussed later, along with Ni/SiO2 the Ni/TiO2 catalyst is also a promising catalyst for the methane decomposition reaction. Zein et al.14 have very recently investigated the kinetics of methane decomposition on a Ni/TiO2 catalyst. Their studies suggested a first order rate law for the decomposition reaction and activation energy of 60 kJ/mol. Interestingly, their studies indicated that the methane adsorption step on the catalyst surface was the rate determining step. This is in contradiction to studies on the Ni/SiO2 catalyst wherein,12 the scission of the first C–H bond was proposed as the rate determining step for the methane decomposition reaction. Carbon-based catalysts have also been considered for the methane decomposition reaction.15 Yoon and co-workers have recently investigated the kinetics of methane decomposition on activated carbons as well as on carbon blacks.16,17 In case of activated carbons the authors observed mass transport effects in the catalyst particles and also significant pore mouth plugging. The reaction order was found to be 0.5 and the activation energy was found to be B200 kJ/mol for the different activated carbon samples. On the other hand, for

168

Catalysis, 2006, 19, 164–183

the carbon black catalysts a reaction order of near unity was observed and the activation energy was lower than that observed for the activated carbons. Recent studies have shown that unreduced Ni catalysts,18 depending on the synthesis procedure, are also efficient for the hydrogen production reaction. Since most fundamental studies have been undertaken on reduced Ni catalysts (Ni0), it will be interesting to investigate methane decomposition fundamentals on unreduced Ni catalysts. 2.1.2 Effect of Support. Methane decomposition on various Ni-supported catalysts has been extensively investigated by the Goodman group.19,20 These studies were directed towards understanding the role played by the support in determining the nature of surface carbon and CO content in the hydrogen stream. Time on stream methane activity studies at a reaction temperature of 823 K revealed comparable initial methane decomposition activities for the Ni/ HY, Ni/HZSM-5, Ni/SiO2 and the Ni/SiO2/Al2O3 catalysts.20 However, unlike the Ni/SiO2, Ni/HY and Ni/SiO2/Al2O3 catalysts, which showed methane conversion activity for several hours, a rapid deactivation (in ca. 1 h.) was observed in case of Ni/HZSM-5. Transmission Electron Microscopy (TEM) images of Ni/HZSM-5 catalyst after the reaction showed an encapsulating type of carbon (Figure 1), which explained the rapid deactivation of the catalyst. On the other hand, carbon filaments (Figure 2) were observed in case of Ni/SiO2, Ni/HY and Ni/SiO2/Al2O3 catalysts.19 The presence of the Ni particle at the apex of the carbon filaments elongates the life-time of the catalyst. It is

Figure 1 TEM image of Ni/HZSM-5 after methane decomposition at 823 K19

Catalysis, 2006, 19, 164–183

169

Figure 2 TEM image of carbon filaments formed after methane decomposition at 823 K on Ni/HY19

noteworthy that while there was rapid deactivation of the Ni/HZSM-5 catalyst at 823 K, the catalyst had a much greater stability at 723 K. TEM images revealed the presence of carbon filaments at lower temperatures (723 K), which resulted in greater catalyst stability for methane conversion. X-Ray photoelectron Spectroscopy (XPS) of the spent samples showed the presence of carbidic and graphitic carbon at low reaction temperatures (r723 K), whereas only the graphitic species were observed at higher temperatures. This is in excellent agreement with studies on model Ni catalysts (single crystal).21 Previously it has been noted that methane decomposition may lead to CO formation via reaction of the carbonaceous residue with the oxygen of the oxide support.22 Since the CO content in the hydrogen stream is a critical parameter for the PEM fuel cells, it is necessary to achieve an accurate quantification of CO (ppm levels). Although this aspect has been neglected in most studies, in our studies particular attention was devoted towards the CO quantification issue.19,20 Quantitative estimation of CO (to ppm levels) was achieved by utilizing the analysis system showed in Figure 3. The effluents from the reactor were first introduced in the thermal conductivity detector (TCD) for detection of hydrogen and methane; Ar was employed as a carrier gas. Analysis of CO was carried out by converting it into methane in a methanizer prior to its introduction in a flame

170

Catalysis, 2006, 19, 164–183

Figure 3 Schematic of the experimental set-up employed to study the methane decomposition reaction; methanizer coupled with the FID was used to quantify ppm levels of CO in the hydrogen stream20

ionization detector (FID). Essentially 100% conversion efficiency (CO to methane) was achieved by operating the methanizer at 673 K with large amounts of hydrogen. This was accomplished by routing all of the hydrogen flow to the FID through the methanizer. An auxillary flow of carrier gas was employed to obtain the optimum carrier/hydrogen ratio for maximum detector sensitivity. The CO formation rates showed a common trend for all the catalysts; high initial rates that rapidly decreased with time and finally stabilized.20 The rate of CO formation was found to increase with increasing temperatures and decrease with increasing gas space velocities (decrease in contact time). The space velocity effect was especially pronounced in the initial period of the methane decomposition reaction. The CO content in the hydrogen stream at a reaction temperature of 823 K was ca. 50, 100 and 250 ppm for Ni/SiO2, Ni/SiO2/ Al2O3 and Ni/HY respectively after the CO-formation rates had stabilized. Diffuse Reflectance Infra-Red Spectroscopy (DRIFTS) studies showed the presence of approximately 55 m-moles of hydroxyl species at 823 K on 0.1 g of the Ni/SiO2 catalyst. The hydroxyl groups on the support were held responsible for CO formation during the methane decomposition reaction. The authors proposed that supports with a greater content of reactive hydroxyl groups at a given methane decomposition temperature would show higher CO formation. It should be noted that some of the support hydroxyl groups may also be involved in CO2 formation, further complicating the issue. Since CO2 is relatively benign to the PEM fuel cell, the CO2 content was not measured in our study.19 However, the knowledge of CO2 content is important to estimate the effect of support hydroxyl groups on the CO formation. Extensive studies involving different catalyst supports (effect of temperature on hydroxyl groups) coupled with methane decomposition studies (quantitative detection of ppm levels of CO and CO2) under different process conditions will be needed to obtain a satisfactory understanding about the influence of support on the CO formation.

Catalysis, 2006, 19, 164–183

171

Takenaka et al. have investigated the effect of several supports (Ni-based) on the activity and stability for the methane decomposition reaction.23 While the catalysts with SiO2, TiO2 and graphite supports showed high activities and long life-times, the catalysts with Al2O3, MgO and SiO2–MgO2 supports were found to be inactive. Characterization of the catalysts by XRD and XANES revealed that in case of the active catalysts Ni was present in the metallic state, whereas in case of the inactive catalysts Ni formed an oxide compound with the support. Studies on catalysts with different silica supports showed that the catalyst activity/stability was also dependent on the pore structure of the support; the silica support devoid of pore structure was found to enhance the catalyst activity/stability. While a large number of studies have been undertaken related to the effect of support on hydrogen production activity/stability, the influence of Ni particle size has not been studied in detail. Otsuka and co-workers have recently reported that the Ni metal particles within a specific size range (60–100 nm) for a Ni/SiO2 showed longest catalytic life for the methane decomposition reaction.24 However, more information is desirable on this topic; for example it will be interesting to systematically investigate the effect of particle size on the hydrogen production rate. Since the hydrogen production rate and the catalyst stability are both important to the practical application of this process, it is important that future studies address both these aspects simultaneously. 2.1.3 Bimetallic Catalysts and Promoters. Shah and co-workers compared the methane decomposition activities and stabilities for monometallic (Pd, Mo or Ni) and bimetallic M–Fe (M ¼ Pd, Mo or Ni) catalyst above 673 K.25 Their studies showed that the bimetallic M–Fe catalysts produced hydrogen at significantly higher rates than the monometallic (M) catalysts. The Pd– Fe catalyst was found to be the most active methane decomposition catalyst at 973 K. Chen et al. have investigated the effect of Cu content on the methane decomposition activity and stability of bimetallic Ni–Cu/Al2O3 catalysts.26 The 2Ni–1Cu–Al catalyst was found to be superior to the 15Ni–3Cu–2Al, 3Ni– 3Cu–2Al and 1Ni–1Cu–1Al catalysts; high activity required optimized levels (not too high and not too low) of Cu in the catalyst. The authors believed that the introduction of Cu (especially at high levels) transformed the catalyst into a quasi liquid state between 973 and 1013 K thus making them less stable. Similar to this study, Li and co-workers (who investigated a series of Ni–Cu–Nb2O5 catalysts) also observed that optimized levels of Cu were required to maximize hydrogen yields. The best catalyst (65Ni–25Cu–5Nb2O5) gave a yield of 7274 mol H2/mol Ni.27,28 The methane decomposition reaction is severely constrained by equilibrium. A few studies have also been undertaken to circumvent the equilibrium constraints.29,30 Otsuka and coworkers used the addition of CaNi5 to Ni/SiO2 for cheating equilibrium.29 The physical mixture of CaNi5 and Ni/SiO2 showed greater than equilibrium methane (decomposition) conversion due to the hydrogen absorption property of CaNi5.

172

Catalysis, 2006, 19, 164–183

2.1.4 Reactor Design. The studies discussed until this point were all undertaken in conventional fixed bed reactors; this segment focuses on methane decomposition studies in membrane and fluidized bed reactors. Application of membrane reactors for hydrocarbon reforming is known to increase the yield of hydrogen.31 In line with this, Ishihara et al. observed greater than equilibrium methane conversions using a 90%Pd10%Ag hydrogen permeable membrane reactor during the decomposition of methane over Ni/SiO2 catalyst.30 The permeated hydrogen was swept by argon (Ar) gas and an increase in the Ar flow-rate was found to enhance the hydrogen production. The positive effect in the hydrogen yields was more pronounced at higher temperatures (4700 K) due to higher hydrogen permeability at these temperatures. A constant conversion of 70% (10% CH4 in N2) was observed at 773 K over a time period of 60 h. Contact times larger than 50 g-cat  h  mol1 and sweep Ar flow rates higher than 200 ml  min1 were found to be favorable for the process. Utilization of sweep gas on the permeate side results in dilution of the permeated hydrogen; this can be a serious limitation for producing pure hydrogen. Also, membrane reactors have a tendency to get fouled; hence when considering a membrane reactor it is important to address the fouling issue. Recently Weizhong and co-workers have used a two stage fluidized bed reactor to study the methane decomposition reaction over a Ni–Cu/ Al2O3 catalyst.32 The temperature in the lower stage of the reactor was held constant at 773 K, while the temperature in the upper stage was controlled between 773 K and 1123 K. Operation at higher temperatures is desired as it increases the hydrogen production rates. Unlike the fixed bed reactor/single stage fluidized bed reactor studies, wherein a rapid catalyst deactivation was observed at 1123 K, the catalyst showed significantly lower deactivation rate in the two stage fluidized bed reactor (upper stage at 1123 K and lower stage at 773 K). The authors believed that the two stage temperature operation decreased the disparity in carbon production and diffusion rates (which was responsible for rapid catalyst deactivation in fixed bed/single stage fluidized bed reactors operating at high reaction temperatures). While this is an interesting concept, the suggested reactor design may be too complex for practical operation. 2.2 Step-wise Methane Reforming: Regeneration Issues. – The catalyst is gradually expected to deactivate due to accumulation of carbon on the catalyst surface during the methane decomposition reaction. This means that after a certain reaction time period, the catalyst has to be regenerated or replaced with a new catalyst (expensive approach). The latter approach could be used for synthesizing carbon filaments with high yields. However, frequent replacement of catalyst is not a practical approach for hydrogen generation. It is therefore essential to employ the regeneration strategy for hydrogen production. The hydrogen production process therefore consists of two steps (a) methane decomposition (Step I) and (b) catalyst regeneration (Step II). This segment, which will focus on the combined hydrogen production process (stepwise reforming), has been sub-divided by the type of regeneration gas used

Catalysis, 2006, 19, 164–183

173

(steam/air/CO2). Issues related to regeneration efficiency and energy efficiency will be addressed in this section. 2.2.1 Step-wise Reforming with Steam Regeneration. The step wise methane steam reforming process, which involves the catalytic decomposition of methane in step I and steam regeneration in step II, has been investigated by our group at relatively low reaction temperatures.33,34 The studies were performed in a pulse mode so as to ensure accurate quantitative analysis of the carbon removed in the regeneration step. No catalyst deactivation was observed in this study; in contrast consequent pulsing of methane without intermittent regeneration (i.e. without Step II) showed an exponential deactivation of the catalyst. The amount of surface carbon removed varied from 92% to 100% (of the amount deposited in Step I) in the various cycles; 95% of the carbon was removed on an average. The CO content in the hydrogen produced in step I was less than 20 ppm. The average amount of hydrogen produced per mole of methane consumed in Step I was 1.1, thus indicating the presence of hydrocarbonaceous residue on the catalyst surface. Recent neutron vibrational studies have revealed the presence of methylidyne (CH), vinylidene (CCH2) and ethylidyne (CCH3) species on Ni–based surfaces after methane dissociation at low temperatures (o673 K).35 The ethylidyne species were found to be less stable than the vinylidene and methylidyne species with increasing methane decomposition temperatures. Amiridis and co-workers employed a continuous flow reactor to study the step-wise steam reforming process.36 In the first step, methane was decomposed over 15% Ni/SiO2 catalyst at 923 K and space velocity of 30000 h1 for 3 h. In the second step, the catalyst was regenerated with steam until no hydrogen was observed in the product stream. Ten reaction cycles performed as described above showed no significant decrease in catalytic activity.36 XRD patterns collected after individual cycles suggested that a large fraction of the carbon deposited in Step I was removed in the regeneration step. It is noteworthy that there was no significant change in the crystallite size of Ni during the reactionregeneration cycles. In agreement our recent studies on a pulse mass analyzer balance have indicated that ca. 75% of the surface carbon (deposited in Step I on Ni/Al2O3/SiO2 at 823) can be removed during the steam regeneration step at 823 K in the continuous flow mode.20 Choudhary and co-workers have investigated the step-wise steam reforming process in two parallel reactors;37,38 methane decomposition and carbon gasification were carried out simultaneously by switching a methane containing feed and steam containing feed between the two reactors at pre-determined time intervals. Amongst the various Ni supported catalysts (ZrO2, MgO, ThO2, CeO2, UO3, B2O3, MoO3, HZSM-5, Hb, NaY, Ce(72)NaY and Si–MCM–41) screened for this cyclic reaction, Ni/ZrO2 and Ni/Ce(72)NaY were found to be the most suitable catalysts. The degree of carbon removal by steam increased significantly on increasing the regeneration temperature from 773 K to 873 K.39 Since issues related to pressure drop are extremely important for practical operation, the pressure drop across the reactor was

174

Catalysis, 2006, 19, 164–183

also monitored in this study. An exponential increase in pressure drop was observed across the reactor during the methane decomposition reaction after a certain threshold of carbon deposition was exceeded. This study highlights the importance of optimizing the switch time between the methane decomposition reaction and regeneration reaction (optimizing run lengths for the two processes). Both methane decomposition and regeneration by steam are endothermic processes, and hence the step-wise steam reforming (like conventional methane steam reforming) is expected to be an energy intensive process. The hydrogen generation process is expected to be more energy efficient for air/oxygen based regeneration; studies related to step-wise air/oxygen methane reforming are summarized in the next section. 2.2.2 Step-wise Reforming with Air/Oxygen Regeneration. The step-wise reforming with air has been employed in the past by Universal oil products (Hypro Process).40 The process utilized a 7% Ni/Al2O3 catalyst in a fluidized bed reactor-regenerator. Catalytic decomposition of methane occurred in the fluidized bed reactor at B1150 K followed by regeneration with air at B1475 K in the fluidized bed regenerator. The product stream consisted of 93–95% hydrogen and unreacted methane. We have also recently investigated the reaction/regeneration (by air) cycles on Ni/HZSM-5 at 723 K in a fixed bed reactor.20 In this case, the methane decomposition step was performed for 1 h following which the catalyst was regenerated using an oxidation–reduction cycle. There was no apparent decrease in catalytic activity throughout the 12 cycles studied at 723 K. Similarly, Zein and Mohamed observed stable catalytic activity for six methane decomposition-regeneration cycles on a 15MnOx–20NiO/TiO2 catalyst.41 Monnerat et al. have investigated the methane decomposition and air regeneration process over a Ni gauze catalyst (Ni-grid with Raney type outer layer).42 Their studies revealed an optimal reaction performance when the cycle consisted of 4 min of reaction period followed by 4 min of regeneration period. In a second study, Mirodatos and co-workers investigated the same process on a Pt/CeO2 catalyst at 673 K under forced unsteady-state conditions.43 No CO was detected in the products under these conditions in either the cracking or the oxidative regeneration steps. Utilization of air can effectively increase the energy efficiency of the process as the exothermic regeneration step can be employed to drive the endothermic hydrocarbon decomposition step. However on the flip side, air regeneration may lead to sintering of the catalyst especially in fixed bed reactors. Villacampa et al. investigated several reaction-regeneration cycles on a co-precipitated Ni/ Al2O3 catalyst.44 Although, the initial activity for hydrogen was recovered after each regeneration step, the regenerated catalyst had a significantly higher deactivation rate. This effect was most prominent after the first catalyst regeneration. The increase in deactivation rate was attributed to the sintering of Ni. Otsuka and co-workers, on the other hand, observed an excellent stability for Ni/Al2O3 and Ni/TiO2 and catalysts for the step wise reforming

Catalysis, 2006, 19, 164–183

175

process.45 This may be related to the following differences in the catalyst reactivation procedures between the two studies: (i) The regeneration temperature was significantly lower in the latter study (smaller exotherm expected). (ii) Also no reduction step was employed in between the regeneration and the reaction step in the latter study (smaller exotherm expected). Based on this, it is apparent that the exothermic catalyst reactivation reactions need to be appropriately controlled to avoid Ni sintering/catalyst deactivation. While steam regeneration appears to be more benign for the catalyst life (several reaction-regeneration cycles), regeneration by air is more energy efficient. It may therefore be interesting to perform the regeneration using a combination of steam and air (oxy-steam regeneration). This topic deserves attention in future investigations. 2.2.3 Step-wise Reforming with CO2 Regeneration. While most of the work to date has focused on regeneration by steam/air/oxygen, few studies involving regeneration by CO2 have also been undertaken. Takenaka et al. investigated the step-wise reforming reaction with CO2 on Ni/SiO2, Ni/Al2O3 and Ni/TiO2 catalysts; the methane decomposition reaction was carried out at 823 K, while the carbon gasification by CO2 was performed at 923 K.46,47 The supports played a crucial role in determining the hydrogen production stability for the process. A gradual decrease in the hydrogen yield (total hydrogen produced in each cycle) was observed for the Ni/SiO2 catalysts during consecutive reactionregeneration cycles. However, there was no decrease in the hydrogen yield for the Ni/Al2O3 and Ni/TiO2 catalysts for several consecutive reaction regeneration cycles. The author claimed a 495% conversion of the carbon to CO in the regeneration step. The structural changes of Ni species for the different catalysts occurring during the consecutive reaction-regeneration cycles were monitored by XANES, XRD and Scanning Electron Microscopy (SEM) to enhance the understanding of the role played by the different supports.48 Based on these studies the authors arrived at the following conclusions: (i) Ni particles in the 60–100 nm range are most effective for the methane decomposition reaction. (ii) While the fresh Ni/SiO2 catalyst had Ni particles in the 40–100 nm range, consecutive reaction-regeneration reactions led to sintering/agglomeration of Ni particles (4200 nm were observed). The Ni sintering was responsible for the inferior performance of the Ni/SiO2 catalyst. (iii) The Ni/Al2O3 and Ni/TiO2 catalyst on the other hand maintained the optimized size distribution of Ni particles throughout the consecutive reaction-regeneration cycles and hence were superior catalysts. Due to its highly endothermic nature, regeneration by CO2 is unfortunately a very energy intensive process.

176

3

Catalysis, 2006, 19, 164–183

Production of Carbon Filaments by Catalytic Methane Decomposition

Carbon filaments (CF), due to their unique properties, have the potential to be used in several applications such as selective adsorption agents, in energy storage devices, catalyst supports and reinforcement materials.49–56 In the last few years significant efforts have been directed towards optimization of the process condition for CF formation. Researchers have investigated CF formation on different metal-based catalysts. In this section, the studies related to production of CF (mainly issues related to rates/yields and quality) have been categorized based on the metals used for catalyzing the methane decomposition reaction. This section will not address issues related to the CF growth mechanism; detailed information about the CF growth mechanism may be found in reference.53 3.1 Ni-based Catalysts. – Ni-based catalysts have been by far the most investigated catalysts for the CF formation via methane decomposition. This may be attributed to the high CF yield obtained on Ni-based catalysts. CF yield is defined as the total amount of CF formed per gram of catalyst at complete deactivation. Since the catalyst has to be essentially replaced for subsequent CF formation, from an economics point of view it is desirable to achieve extremely high CF yields. Shaikhutdinov and co-workers have exhaustively investigated CF formation yields on co-precipitated Ni-alumina and Ni-Cu-alumina catalysts.57,58 The amount of CF formed per gram of the catalyst was found to increase with increasing Ni content in the Ni-alumina catalyst. However, the CF yield was found to be radically small for pure Ni powder.57 In good agreement, studies by Toebes et al. also showed negligible CF formation from methane decomposition on unsupported Ni catalysts.59 Low CF yields from methane decomposition on unsupported Ni catalyst have been attributed to the presence of large Ni particles (50–1000 nm) with low index planes, since low index planes are incapable of dissociating the unreactive methane molecules. Li et al. employed a Ni-alumina catalyst prepared from Feitknecht compound for maximizing CF yields from methane.60 Similar to the work by Shaikhutdinov and co-workers,57 the total amount of CF formed was found to increase with increasing Ni content of the catalyst. The total amount of CF formed was dependent on the reduction temperature as well as the reaction temperature. Although the rate of CF formation increased at higher temperature there was a decrease in the total yield of CF due to rapid deactivation of the catalyst. Ermakova and co-workers manipulated the Ni particle size to achieve large CF yields from methane decomposition.61,62 The Ni-based catalysts employed for the process were synthesized by impregnation of nickel oxide with a solution of the precursor of a textural promoter (silica, alumina, titanium dioxide, zirconium oxide and magnesia). The optimum particle size (10–40 nm) was obtained by varying the calcination temperature of NiO. The 90% Ni–10% silica catalyst was found to be the most effective catalyst with a total CF yield of 375 gCF/gcat. XRD studies by the same group on high loaded Ni-silica

Catalysis, 2006, 19, 164–183

177

showed that the nickel particles seemed to ‘‘self organize’’ to the optimum size (30–40 nm) during the course of the methane decomposition reaction, i.e. smaller particles underwent sintering to form larger particles whereas larger particles were found to undergo dispersion.63 The authors proposed that deactivation occurred when the distance separating the metal particles (present on filament ends) increased to an extent such that reversible merging and dispersion of the Ni particles was prevented. Several studies have considered the addition of Cu to Ni-based catalysts for enhancing the methane decomposition CF yields. Studies by Shaikhutdinov and co-workers showed that the addition of Cu decreased the rate of CF formation but greatly increased the stability of the catalyst.57 The maximum CF yield (240 gCF/gcat) was obtained for a 3% Cu-87% Ni catalyst. Li et al. also studied the doping effects of Cu on the CF formation.64 In this work, addition of small amounts of Cu not only increased the total amount of CF formed, but also increased the growth rate at 873 K. This was unlike the work by Shaikhutdinov et al. where addition of small amount of Cu had decreased the growth rate but increased the overall CF yield by significantly increasing the life time of the catalyst.57 Addition of large amounts of Cu had a detrimental effect on the performance of the catalyst at relatively low methane decomposition temperatures, however the high Cu content catalyst was found to be the most effective catalyst for CF formation at higher temperature when methane was co-fed with hydrogen. Reshetenko et al. also used the Feitknecht compound as a precursor for preparing copper(8–45%) promoted Ni catalysts and carried out a detailed investigation of the methane decomposition reaction.65 The highest CF yield (525 gCF/gcat) was obtained on the 75Ni-15Cu/ Al2O3 catalyst at 898 K. In general, Ni-based catalysts in their reduced (Ni0) forms are used for CF generation from methane. However, some recent studies have shown that it may not be necessary to pre-reduce the Ni catalysts. Qian and co-workers observed methane conversions approaching equilibrium on an unreduced Ni– Cu/Al2O3 catalyst in a fluidized bed reactor.66 The corresponding methane conversion for the reduced catalyst was significantly lower from the onset of the reaction. The CF yields were also considerably higher for the unreduced catalysts. The authors believed that in situ reduction of the lattice oxygen (in case of the unreduced catalyst) provided energy for the endothermic methane decomposition process. Also, since some the hydrogen produced in the reaction was consumed in situ, this assisted in shifting the equilibrium in the direction of CF formation. This is in contrast to recent fixed bed reactor studies by Suib and co-workers, wherein a significantly higher initial methane conversion for the reduced Ni catalyst (40% Ni/SiO2 catalyst prepared from nitrate salts) was observed as compared to the unreduced version.18 However, these studies do suggest that unreduced catalysts depending on the synthesis procedure may provide large CF yields. Since the pre-reduction treatment is an important process parameter for CF formation, it would be worthwhile to obtain a detailed understanding apropos the effect of catalyst reduction on methane conversions/CF yields.

178

Catalysis, 2006, 19, 164–183

The large surface area of CF makes it an attractive candidate for catalyst support material. Avdeeva and co-workers used CF with different textural properties as supports for Ni to study the methane decomposition reaction.67,68 Ni supported on the CF, which was obtained from methane decomposition on a Ni–Cu/Al2O3 at 898 K, showed the highest yield for secondary carbon (224 gCF/gNi). Highly porous CF supports were found to be most effective for the secondary generation of CF. Compared to studies related to CF optimization fewer studies have been undertaken related to the nature/quality of the CF formed from the methane decomposition process. Studies by Baker and co-workers on Ni and Ni–Cu catalysts revealed that the structural characteristics of CF were strongly influenced by the nature of the catalyst particles.69,70 While particles rich in Ni resulted in the formation of smooth filaments, Cu-rich alloy particles gave rise to filaments having a spiral conformation. The filament size (25–100 nm) was found to be strongly dependent on particle size of the catalyst. The morphology and surface structure of CF (F1) produced on Ni-alumina catalysts and carbon (F2) produced in case of Ni-Cu-alumina catalysts were studied by scanning tunneling microscopy (STM) and High-Resolution Transmission Electron Microscopy (HRTEM) by Shaikhutdinov et al.58 The carbon surface of the filaments was found to be rough and was formed by misoriented edge planes of graphite crystallites. In case of the F1 the basal graphite planes lay inclined to the fiber axis, whereas the basal planes were perpendicular to the filament axis for F2. HRTEM micrographs indicated a closed layer structure on the edges for F2, which was contrary to the open structure observed for graphite crystallites. Kuvshinov and co-workers observed that the CF texture could be modified by changing the CH4:H2 feed ratio.71 Their work also suggested that the surface area of carbon growth centers was an important parameter for determining the maximum CF yield on Ni catalysts using pure methane. From a practical view point it is essential to have an excellent understanding of the CF yields in relation to desired CF properties (surface area, structural/ mechanical properties etc). Unfortunately this aspect of CF production has been seriously neglected. In a couple of studies, the surface area of the CF formed during methane decomposition process has been related to the CF yields.65,72 These studies clearly show that the surface area of the CF and CF yields are both strongly dependent (however in a different way) on the catalyst and process conditions. The BET surface area of the CF obtained on the catalyst, which showed highest CF yield (catalyst: 75Ni–15Cu/Al2O3 and yield: 525 gCF/gcat), was 233 m2/g. On the other hand, the CF with highest surface area (286 m2/g) was obtained with a 45Ni–45Cu/Al2O3 catalyst, which had a corresponding CF yield of only 118 gCF/gcat. While the CF BET surface area was found to decrease with increasing methane decomposition temperatures, the CF yield was found to pass through a maximum with increasing reaction temperature.72 The above study clearly demonstrates the importance of optimizing the CF yields and CF quality simultaneously.

Catalysis, 2006, 19, 164–183

179

3.2 Fe and Co-based Catalysts. – Catalysts which are not Ni-based have low CF formation rates/yields. However, it is interesting to consider metals such as Fe and Co since the properties of the CF depend on the metal employed for catalyzing the methane decomposition reaction. Ermakova and co-workers have investigated the methane decomposition reaction on un-supported Fe powder as well as supported Fe catalysts.73 With the exception of SiO2 support, the other supports (Fe/Al2O3, Fe/Al2O3, Fe/ TiO2) gave similar CF yields as unsupported Fe powder (17 gCF/gcat). The interesting behavior of Fe/SiO2 system motivated them to investigate it in greater details.74 Their studies showed that the silicates depending on their concentration in the catalyst could have either a promoting effect or an inhibiting effect on CF formation. The Fe/SiO2 catalyst with optimal silicate content showed a yield of 45 gCF/gFe. It should be noted that although the yield of CF on Fe-based catalysts is small, Fe-based catalysts produce predominantly thin walled CF (considered to be more valuable than other CF). Bennissad et al. investigated CF formation on Fe-based catalysts using CH4– H2 mixtures at temperatures to 1423 K.75,76 Under these conditions thicker fibers (ca. 1 m) were obtained, but when heating was stopped at 1323 K, the normal structure of CF was observed. Shah et al. investigated CF formation on bimetallic Fe–M (M ¼ Pd, Mo or Ni) catalysts.25 The bimetallic catalysts were found to be more active for the CF formation than the corresponding monometallic catalysts. While only CF formation was observed at the methane decomposition reaction temperature range of 973–1073 K, amorphous carbon and carbon flakes were observed concomitant with CF at reaction temperatures above 1173 K. Otsuka and co-workers have recently investigated the structural changes of Fe species and the nature of the CF formed during the methane decomposition reaction on Fe2O3/Al2O3 and Fe2O3/SiO2 catalysts.77 XANES studies showed that during the methane decomposition reaction the smaller sized Fe particles were transformed into Fe3C while the larger particles were converted into carbon atom saturated g-Fe species. The supports had a profound effect in determining the nature of the CF formed in the process; multi-walled CF and chain like carbon fibers were formed on Fe/Al2O3, while CF composed of spherical carbon units were formed along with chain like carbon fibers on Fe2O3/SiO2 catalysts. Avdeeva and co-workers studied the methane decomposition reaction on Coalumina catalysts with varying concentration of Co.78 The CF formation was found to be maximized at 60–75% content of Co. No induction period was observed for the Co-alumina catalysts, which was contrary to their previous experience involving Ni-alumina catalysts.57 Also in this case a different variety of filaments (not observed on Ni-based catalysts) with hollow-like core morphology were observed. Takenaka et al. have recently investigated the effect of supports on CF formation for Co-based catalysts.79 The Co/Al2O3 and Co/ MgO catalysts were found to be superior to the Co/SiO2 and the Co/TiO2 catalysts for CF formation. Based on catalyst characterization studies the

180

Catalysis, 2006, 19, 164–183

authors claimed that the 10–30 nm range for Co particles was preferred for CF formation. The authors also found that the temperature had a significant influence in determining the nature of the CF. Multi-walled CF were formed in the temperature range of 873–973 K, whereas helically coiled and bamboo-like CF were preferentially formed at 1073 K. Smith and co-workers investigated the effect of metal support interaction on the CF formation on a series of Co-silica catalysts.80 The metal support interaction was manipulated by addition of either BaO, La2O3 or ZrO2 to silica. The rate of catalyst deactivation was found to increase with the increase in the metal support interaction. Competition between CF formation and encapsulating carbon formation controlled the catalyst deactivation rate. In case of the catalysts with high metal support interaction, the encapsulating carbon formation was dominant and hence led to a rapid deactivation of the catalyst. It is unfortunate that very few studies have been undertaken which relate the methane decomposition process conditions and CF yields to the CF quality (surface area, structure/texture, mechanical strength etc). From a process application view point it is extremely important to comprehensively investigate this aspect of CF formation. 4

Concluding Remarks

Catalytic methane decomposition has received considerable attention in recent years. The reaction has been investigated for two main applications (a) production of hydrogen and (b) synthesis of carbon filaments. The important conditions necessary for clean hydrogen production are as follows: (i) High conversion of methane (to COx-free hydrogen) to avoid costly product separation. (ii) Absence of pressure drop issues across reactor (iii) Effective regeneration of catalyst (iv) Stable life of catalyst over several cycles While the methane decomposition step has been extensively investigated, unfortunately less attention has been devoted to other aspects. To avoid a pressure drop it is important to optimize the run lengths for the methane decomposition step and the regeneration step. In order to assess the commercial viability of the process, it is important to study the process over several cycles (4100). If the metal-based catalysts are not reduced in between the decomposition and deactivation stage, significant amount of CO may be formed due to reaction of the metal-oxide with methane during the initial stages of the methane decomposition reaction. Most studies have employed argon gas with a thermal conductivity detector to analyze product gases. Such an analysis procedure is not astute for accurate quantification of ppm levels of CO/CO2. Special attention should be paid towards this analysis as the main advantage claimed in this process is the production of clean hydrogen (without further need for purification).

Catalysis, 2006, 19, 164–183

181

Following are the important conditions necessary for carbon nanotube synthesis (i) High yield of CF (ii) CF with desired properties While several studies have been undertaken to optimize the CF yield, only a few studies have addressed the quality factor for CF. It is important to tune the synthesis procedures such that CF yields/properties are consistent with the desired applications. References 1. Web-site: http://www.eere.energy.gov/RE/hydrogen.html. 2. J.N. Armor, Appl. Catal., 1999, 176, 159. 3. V.R. Choudhary, A.S. Mamman and S.D. Sansare, Angew Chem. Int. Ed. Engl., 1992, 31, 1189. 4. J.R. Rostrup-Nielson, Catal. Today, 1993, 18, 305. 5. J.M. Thomas and W.J. Thomas, ‘‘Principles and Practice of Heterogeneous catalysis’’, VCH, Weinheim, 1997, p. 545. 6. N. Vanderborgh, Private Communication, 2000. 7. T.V. Choudhary, C. Sivadinarayana and D.W. Goodman, Chem. Eng. J., 2003, 93, 69. 8. M. Terrones, Ann. Rev. Mater. Res., 2003, 33, 419. 9. Y.-N. Wang, R.G. Herman and K. Klier, Surf. Sci., 1992, 279, 33. 10. T.V. Choudhary and D.W. Goodman, J. Mol. Catal., 2000, 163, 9. 11. T.V. Choudhary, E. Aksoylu and D.W. Goodman, Catal. Rev. Sci. Eng., 2003, 45, 151. 12. K. Otsuka, S. Kobayashi and S. Takenaka, J. Catal., 2001, 200, 4. 13. S. Takenaka, H. Ogihara and K. Otsuka, J. Catal., 2002, 208, 54. 14. S.H. Zein, A.R. Mohamed and P.S. Sai, I&EC, 2004, 43, 4864. 15. N.Z. Muradov, Energy and Fuels, 1998, 12, 41. 16. M.H. Kim, E.K. Lee, J.H. Jun, S.J. Kong, G.Y. Han, B.K. Lee, T.-J. Lee and K.J. Yoon, Int. J. Hydrogen Energy, 2004, 29, 187. 17. E.K. Lee, S.Y. Lee, G.Y. Han, B.K. Lee, T.-J. Lee, J.H. Jun and K.J. Yoon, Carbon, 2004, 42, 2641. 18. R.A. Couttenye, M.H. De Villa and S.L. Suib, J. Catal., 2005, 233, 317. 19. T.V. Choudhary, C. Sivadinarayana, C. Chusuei, A. Klinghoffer and D.W. Goodman, J. Catal., 2001, 199, 9. 20. T.V. Choudhary, C. Sivadinarayana, A. Klingerhoffer and D.W. Goodman, Stud. Surf. Sci. Catal., 2001, 136, 197. 21. D.W. Goodman, R.D. Kelley, T.E. Madey and J.M. White, J. Catal., 1980, 64, 479. 22. P. Ferreira-Aparicio, I. Rodriguez-Ramos and A. Guerrero-Ruiz, Appl. Catal. A: Gen., 1997, 148, 343. 23. S. Takenaka, H. Ogihara, I. Yamanaka and K. Otsuka, Appl. Catal. A: Gen., 2001, 217, 101. 24. S. Takenaka, S. Kobayashi, H. Ogihara and K. Otsuka, J. Catal., 2003, 217, 79. 25. N. Shah, D. Panjala and G.P. Huffman, Energy and Fuels, 2001, 15, 1528. 26. J. Chen, Y. Li, Z. Li and X. Zhang, Appl Catal. A: Gen., 2004, 269, 179.

182

Catalysis, 2006, 19, 164–183

27. 28. 29. 30. 31. 32.

J. Li, G. Lu, K. Li and W. Wang, J. Mol. Catal. A, 2004, 221, 105. J. Li, G. Lu and K. Li, Chem. Lett., 2004, 33, 652. K. Otsuka, S. Kobayashi and S. Takenaka, Appl. Catal. A: Gen., 2000, 190, 261. T. Ishihara, Y. Miyashita, H. Iseda and Y. Takita, Chem. Lett., 1995, 2, 93. E. Kikuchi, Cattech, 1997, 1, 67. Q. Weizhong, L. Tang, W. Zhanwen, W. Fei, L. Zhifei, L. Guohua and L. Yongdan, Appl. Catal. A: Gen., 2004, 260, 223. T.V. Choudhary and D.W. Goodman, Catal. Lett., 1999, 59, 93. T.V. Choudhary and D.W. Goodman, J. Catal., 2000, 192, 316. C. Sivadinarayana, T.V. Choudhary, L. Daemon, J. Eckert and D.W. Goodman, Angew Chem. Int. Edit., 2002, 41, 144. R. Aiello, J.E. Fiscus, H.-C.Z. Loye and M.D. Amiridis, Appl. Catal. A: Gen., 2000, 192, 227. V.R. Choudhary, S. Banerjee and A. M. Rajput, J. Catal., 2001, 198, 136. V.R. Choudhary and A.M. Rajput, US Patent 2004118047, 2004. V.R. Choudhary, S. Banerjee and A.M. Rajput, Appl. Catal. A: Gen., 2002, 234, 259. K. Cox and K. Williamson, Hydrogen: Its technology and implications, CRC Press, Boca Raton, FL, 1977, Vol. 1. S.H.S. Zein and A.R. Mohamed, Energy and Fuels, 2004, 18, 1336. B. Monnerat, L. Kiwi-Minsker and A. Renken, Chem. Eng. Sci., 2001, 56, 633. C. Marquez-Alvarez, E. Odier, L. Pinaeva, Y. Schuurman, C. Millet and C. Mirodatos, Stud. Surf. Sci. Catal., 2001, 136, 483. J.I. Villacampa, C. Royo, E. Romeo, J.A. Montoya, P. Del Angel and A. Monzon, Appl. Catal. A, 2003, 252, 363. K. Otsuka, S. Takenaka and H. Ohtsuki, Appl. Catal. A: Gen., 2004, 273, 113. S. Takenaka and K. Otsuka, Chem. Lett., 2001, 3, 218. S. Takenaka, Y. Tomikubo, E. Kato and K. Otsuka, Fuel, 2004, 83, 47. S. Takenaka, E. Kato, Y. Tomikubo and K. Otsuka, J. Catal., 2003, 219, 176. R.T.K. Baker, Carbon, 1989, 27, 315. G.E. Gadd, M. Blackford, S. Moricca, N. Webb, P.J. Evans, A.N. Smith, G. Jacobsen, S. Leung, A. Day and Q. Hua, Science, 1997, 277, 933. P. Chen, X. Wu, J. Lin and K.L. Tan, Science, 1999, 285, 91. N.M. Rodriguez, M.S. Kim and R.T.K. Baker, J. Phys. Chem., 1994, 98, 13108. N.M. Rodriguez, J. Mater. Res., 1993, 8, 3233. V.A. Likholobov, V.B. Fenelonov, L.G. Okkel, O.V. Goncharova, L. Avdeeva, V.I. Zaikovskii, G.G. Kuvshinov, V.A. Semikolenov, V.K. Duplyakin, O.N. Baklanova and G.V. Plaksin, React. Kinet. Catal. Lett., 1995, 54, 381. K.P. De Jong and J.W. Geus, Catal. Rev.-Sci.-Eng., 2000, 42, 481. V.B. Fenelonov, L.B. Avdeeva, V.I. Zheivot, O.V. Goncharova and L.G. Pimneva, Kinetics and Catalysis, 1993, 34, 483. L.B. Avdeeva, O.V. Goncharova, D.I. Kochubey, V.I. Zaikovskii, L. Plyasova, B.N. Novgorodov and S.K. Shaikhutdinov, Appl. Catal. A: Gen., 1996, 141, 117. S.K. Shaikhutdinov, V.I. Zaikovskii and L.B. Avdeeva, Appl. Catal. A: Gen., 1996, 148, 123. M.L. Toebes, J.H. Bitter, A.J. van Dillen and K.P. Je Dong, Catal. Today, 2002, 76, 33. Y.D. Li, J.L. Chen and L. Chang, Appl. Catal. A: Gen., 1999, 163, 45. M.A. Ermakova, D.Y. Ermakov, G.G. Kuvshinov and L.M. Plyasova, J. Catal., 1999, 187, 77.

33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54.

55. 56. 57. 58. 59. 60. 61.

Catalysis, 2006, 19, 164–183

183

62. M.A. Ermakova, D.Y. Ermakov and G.G. Kuvshinov, Appl. Catal. A: Gen., 2000, 201, 61. 63. M.A. Ermakova, D.Y. Ermakov, L.M. Plyasova and G.G. Kuvshinov, Catal. Lett., 1999, 62, 93. 64. Y.D. Li, J.L. Chen and L. Chang, J. Catal., 1998, 178, 76. 65. T.V. Reshetenko, L.B. Avdeeva, Z.R. Ismagilov, A.L. Chuvilin and V.A. Ushakov, Appl. Catal. A: Gen., 2003, 247, 51. 66. W. Qian, T. Liu, F. Wei, Z. Wang and Y. Li, Appl. Catal. A: Gen., 2004, 258, 121. 67. S.K. Shaikhutdinov, L.B. Avdeeva, B.N. Novgorodov, V.I. Zaikovskii and D.I. Kochubey, Catal. Lett., 1997, 47, 35. 68. T.V. Reshetenko, L.B. Avdeeva, Z.R. Ismagilov and A.L. Chuvilin, Carbon, 2004, 42, 1143. 69. M.S. Kim, N.M. Rodriguez and R.T.K. Baker, J. Catal., 1991, 131, 60. 70. M.S. Kim, N.M. Rodriguez and R.T.K. Baker, J. Catal., 1992, 134, 253. 71. G.G. Kuvshinov, Y.I. Mogilnykh, D.G. Kuvshinov, V.I. Zaikovskii and L.B. Avdeeva, Carbon, 1998, 36, 87. 72. T.V. Reshetenko, L.B. Avdeeva, Z.R. Ismagilov, V.V. Pushkarev, S.V. Cherepanova, A.L. Chuvilin and V.A. Likholobov, Carbon, 2003, 41, 1605. 73. M.A. Ermakova, D.Y. Ermakov, A.L. Chuvilin and G.G. Kuvshinov, J. Catal., 2001, 201, 183. 74. M.A. Ermakova and D.Y. Ermakov, Catal. Today, 2002, 77, 225. 75. F. Benissad, P. Gadelle, M. Coulon and L. Bonnetain, Carbon, 1988, 26, 61. 76. F. Benissad, P. Gadelle, M. Coulon and L. Bonnetain, Carbon, 1989, 27, 585. 77. S. Takenaka, M. Serizawa and K. Otsuka, J. Catal., 2004, 222, 520. 78. L.B. Avdeeva, D.I. Kochubey and S.K. Shaikhutdinov, Appl. Catal. A: Gen., 1999, 177, 43. 79. S. Takenaka, M. Ishida, M. Serizawa, E. Tanabe and K. Otsuka, J. Phys. Chem. B, 2004, 108, 11464. 80. X. Li, Y. Zhang and K.J. Smith, Appl. Catal. A: Gen., 2004, 264, 81.