Vous êtes sur la page 1sur 12

International Journal of Hydrogen Energy 30 (2005) 1265 1276 www.elsevier.

com/locate/ijhydene

Low temperature water gas shift: the link between the catalysis of WGS and formic acid decomposition over Pt/ceria
Gary Jacobs, Patricia M. Patterson, Uschi M. Graham, Adam C. Crawford, Burtron H. Davis
Center for Applied Energy Research, University of Kentucky, 2540 Research Park Dr., Lexington, KY 40511, USA Received 2 September 2004; accepted 1 March 2005 Available online 25 April 2005

Abstract Partial reduction of the ceria surface leads to the formation of bridging Type II OH groups, as reported previously. These were found to react with formic acid to yield bidentate formate and water, or with CO directly to form bidentate formate. In both WGS and formic acid decomposition via dehydrogenation whereby a high water/CO or water/formic acid ratio was utilized, a normal kinetic isotope effect was observed consistent with the involvement of formate CH bond cleaving in the rate limiting step. In the current investigation, switching between a feed containing DCOOH to HCOOH led to an increase in the formic acid conversion, as well as a slight increase in the CO2 selectivity. The overall change in the CO2 yield was approximately 1.3, very close to the normal kinetic isotope effect of 1.4 reported in several studies of water gas shift on metal promoted ceria. By in situ DRIFTS, a slower D-formate decomposition rate versus H-formate in the presence of water was also observed during transient decomposition of the stabilized formate. Based on the identication of the same adsorbed reactive species and similar kinetic isotope effects, the results suggest an analogous mechanism operates for formic acid decomposition and water gas shift. 2005 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.
Keywords: Watergas shift (WGS); DRIFTS; Infrared (IR); Platinum; Ceria; Formic acid; Hydrogen

1. Introduction There is much debate over whether or not a ceria-mediated redox mechanism or a surface formate mechanism occurs for the low temperature watergas shift reaction. Understanding the mechanism and the nature of the catalytically active site is crucial for the development of improved catalysts, especially those to be used in fuel cell processors for the generation of pure CO-free hydrogen. Regarding the
paper was presented at 3rd Fuel Cell Topical Conference, 2004 AIChE Spring National Meeting, April 2529, 2004. Corresponding author. Tel.: +1 859 257 0251; fax: +1 859 257 0302. E-mail address: davis@caer.uky.edu (B.H. Davis).
This

surface formate mechanism, reduction of Pt/ceria in hydrogen leads to partial reduction of ceria in the surface layers, and the generation of bridging type OH groups. These OH groups react with CO to produce adsorbed bidentate formate. Shido and Iwasawa [1] have claimed that H2 O assists in the decomposition of surface formates to H2 and an adsorbed CO2 species (unidentate carbonate) during the rate limiting step. Indeed, they found, and we have veried, that substituting OOCH with OOCD led to a decrease in the rate by a factor of approximately 1.4, giving credibility to the mechanistic scheme. In this study, we tested another reaction that produces bidentate formate as an intermediateformic acid decomposition. Adsorption of formic acid is a dissociative process, giving rise to formate and adsorbed H together with liberation of H2 O in the process. Water-assisted

0360-3199/$30.00 2005 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2005.03.001

1266

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276

decomposition of formic acid was carried out with HCOOH and DCOOH to determine if a similar kinetic isotope effect exists for that reaction, which would lend further credibility to the possibility of a surface formate mechanism operating for the low temperature water gas shift reaction. In a broader sense, formic acid is an important reaction in its own right, since molecules such as formic acid and methanol can be steam reformed catalytically to produce hydrogen, making them important carriers for consideration from the perspective of hydrogen storage.

(3) (reproduced from [4])

1.1. Historical perspective The history behind the formic acid decomposition reaction dates back at least a century in the pioneering work of Sabatier [2], and with the advent of advanced characterization techniques, great strides have been made to provide understanding of the mechanism. Two works of note include the reviews of Mars et al. in the 1950s and 60s [3,4], who noted that . . . it is as yet impossible to give a correct estimate of the signicance the research in the eld of catalytic formic acid decomposition has for our understanding of the heterogeneous catalysis in general, but in our opinion a warning against over-estimation is justied. In the reviews of Mars, it was noted that the decomposition of formic acid on oxides takes place at temperatures above 200 C, and it proceeds mainly in one of two directions: (1) dehydration HCOOH CO + H2 O, (2) dehydrogenation HCOOH CO2 + H2 . A third reaction also is possible, the disproportionation to formaldehyde via two formic acid molecules, but as it is usually negligible, it will not be considered in this discussion. From the standpoint of catalysis, what is especially of interest is that the selectivity of the two possible pathways can be controlled by the choice of catalyst. In terms of selectivity, Mars [3] very elegantly showed that while some catalysts promoted dehydration (e.g., TiO2 , SiO2 , and Al2 O3 ), others were effective in promoting the dehydrogenation (e.g., MgO, ZnO, Fe3 O4 , ThO2 after high temperature heating, and Cr 2 O3 ). It was also of particular interest to note that these latter catalysts were also found to be effective in dehydrogenating alcohols. Upon taking a second look at those materials, it is evident that they are very often the catalysts employed for carrying out the watergas shift reaction. Based upon kinetic data and infrared observations, Mars et al. [4] concluded that there is a common intermediate involved in the decomposition of formic acid pathway, with the stability of the intermediate governing the relative rates. In 1963, they concluded that in most cases, the reaction proceeds via an adsorbed formate ion as an adsorbed intermediate, as follows:

In that work, Mars et al. [4] also reviewed the works of Tamaru [5,6] and Fahrenfort et al. [7], who concluded that while some surfaces like Au and Ag displayed a high activation energy barrier for step A, on many metals and oxides, the stability of the formate was so high that the selectivity was controlled by the relative rates of decomposition of the formate through pathways B and C. In such cases where the stability of the formate was found to be high, of particular interest to catalyst researchers studying the much-related water gas shift reaction, the dehydrogenation coordinate put forth by Mars et al. [4] was schematically represented as follows: (modied from [5])

In a number of infrared studies of formic acid adsorption on metal oxides, notably, the studies of ZnO by Hirota et al. [8] and MgO by Scholten et al. [4], surface formates were clearly identied at low temperatures (room temperature to 100 C). Upon heating the MgO samples, Scholten et al. [4] found that the decomposition of formates corresponded directly to the formation of the decomposition products, and that the rates agreed very well with the formic acid decomposition rates.

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276

1267

1.2. Extension to watergas shift on partially reducible oxides In the 1980s and early 1990s, a number of infrared studies on metal oxides for reactions involving methanol, formic acid, and water gas shift were performed in Japan. Some notable studies include those of Tamarus group [9], who investigated methanol decomposition over Cr 2 O3 ; Onishis group, which carried out the adsorption of formic acid and methanol on ceria [10]; and a series of publications by Iwasawa [1113], who studied water gas shift over ZnO, MgO, ceria, and metal promoted ceria. In each case, surface formates were found to be important intermediates for the reaction of interest, suggesting the possibility of a universal active site. In the case of water gas shift, it was found that CO can react with active surface hydroxyl groups to produce the surface formates. In the case of partially reducible oxides (e.g., ceria, thoria, and zirconia), there is increasing evidence to support the idea that the surface shell reduction process produces hydroxyl groups on the surface that are particularly reactive for watergas shift. For partially reduced ceria, these OH groups have been observed at considerably lower wavenumbers than those terminal groups (i.e., Type I) for the oxidized form of ceria (i.e., +4 oxidation state). For example, for unreduced ceria, the Type I groups appear at between 3690 and 3710 cm1 [14]. However, after reduction in hydrogen of unpromoted ceria (occurring at ca. 450 C), or at much lower temperatures with Pt or Rh addition (e.g., 200250 C), sharp bands are formed in the infrared at approximately 3650 cm1 (and with a shoulder at 3675 cm1 ) [9,1219]. There have been many labels given to describe these special OH groups. They are classied as Type II, bridging, geminal, or terminal OH groups on partially reduced ceria. The label seems to depend on the perspective of the assignee. We have observed these groups both in the presence and absence of water, and have considered two possibilities for their formation:

so that bridging OH groups are formed with a decrease in oxidation state of two Ce atoms to +3 for each pair of OH groups formed. Certainly, with the presence of a metal like Pt, the hydrogen could be dissociated to produce the bridging OH groups directly, a reductive process involving the change in oxidation state of ceria from +4 to +3. As one can see, the term geminal is appropriate if one considers the perspective of the M atom, while considering two M atoms, it is probably more appropriate to use the term bridging or Type II, since the OH group is bridging across two M atoms. Such groups have also been observed on thoria [20,21] and zirconia [22,23]. The link between the generation of the bridging OH groups and the partial reduction of ceria was established by comparing the IR band intensities at different temperatures to the surface shell reduction process as measured by temperature programmed reduction [24,25], and more directly, by in situ XANES [26]. One possibility for the observation of a shoulder is due to the possibility of resonance, which can impact the OH stretching vibration, as shown in the schematic below. After the identication of the bridging OH groups on the surface of ceria, researchers discovered that these groups react with CO to produce surface formates [12,13]. The surface formates are readily observed using IR spectroscopy. The OCO component of the formate vibration has two stretching modes, an asymmetric stretching mode which produces an intense band at ca. 1555 cm1 , and a symmetric mode which appears at a lower wavenumber of ca. 1358 cm1 . The band maxima may shift to higher wavenumbers due to surface coverage effects (e.g., dipoledipole coupling). Accompanying the OCO bands are stretching vibrations for the CH component of the formates, the more intense occurring at ca. 2845 cm1 (CD at ca. 2150 cm1 ) for bidentate formate, and a smaller band at ca. 2950 cm1 (CD at ca. 2140 cm1 ), which has been assigned to a bridged formate [12] or to another vibrational mode of bidentate formate, due to resonance [27]. The active bridging OH group densities depend on the degree of reduction of the surface shell of ceria. Much more intense formate bands are observed upon CO addition to Pt/ceria at 250 C than on unpromoted ceria, due to the ability of Pt to catalyze the surface reduction process [25,26].

On the left, one possibility is that vacancies (involving two reduced Ce+3 centers) created from the removal of surface capping oxygen atoms react with H2 O to form the bridging OH groups, or, as shown on the right, the hydrogen could be dissociated directly, aided with a promoter metal like Pt,

In our initial efforts to further understand the mechanism of the low temperature water gas shift reaction on Pt/ceria, we conducted experiments at high H2 O/CO ratios that would be typical of conditions found in a fuel processor for fuel cell applications [28]. It was found in earlier kinetic studies that the reaction orders of PCO and Pwater depend on the H2 O/CO ratio. For example, it was found that on metal/ceria catalysts, at high CO/H2 O ratios, the

1268

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276

reaction order is zero for PCO , whereas the reaction becomes rst order in PCO at high H2 O/CO ratios [27], those typically found in a fuel cell processor after secondary water injection following the high temperature shift stage. For a heterogeneous catalytic mechanism, under rst order conditions, the rate of the reaction should control the coverage of the adsorbed CO intermediate. In fact, at high H2 O/CO ratios, the kinetics have revealed that Pwater is of zeroth order. That implies the surface should be saturated by this reactant under watergas shift reaction conditions. Certainly, the bridging OH groups represent a likely possibility. Not only that, but we have found that while Pt-CO remains close to saturation when high H2 O/CO ratios are used (PCO rst order), the surface formate coverages respond in a manner consistent with them being controlled by the water gas shift reaction rate. We have therefore argued that they are likely important intermediates in the reaction pathway [18,19], as proposed by Shido and Iwasawa [12,13].

The fact that Pt remains close to saturation by CO under high watergas shift rates contradicts the recently proposed ceria-mediated redox mechanism [29,30], where surface Pt atoms are directly involved in the pathway: Pt + CO Pt-CO, Pt-CO + 2CeO2 Ce2 O3 + CO2 + Pt, Ce2 O3 + H2 O 2CeO2 + H2 . Certainly, the use of the above mechanism as the basis for design has led to the development of more active catalysts. In fact, in both the surface formate mechanism and the ceriamediated redox schemes, the partial reduction of ceria is a necessary step. As discussed previously, the formate mechanism argues that the reduction of the ceria surface shell is necessary to generate the bridging OH group active sites, while the redox mechanism argues that reduction of ceria is directly involved in the reaction mechanism. In our initial work, the redox mechanism was utilized as a means of catalyst design. However, in our own infrared studies, many of the points that served as a basis to rule

out the formate mechanism were largely discounted. For example, Hilaire et al. [29] claimed that surface formates do not form upon CO adsorption to Pd/ceria catalysts, but rather only carbonates are present. However, we observed very intense formate bands on high surface area ceria catalysts upon CO adsorption [19]. In the work of Hilaire et al. [29], switching experiments were conducted in that work between CO adsorption and O2 pulses. In that case, carbonates are the surface species, because no H2 or H2 O was introduced beforehand to provide the possibility of forming the bridging OH groups (and therefore, the surface formates upon CO adsorption). However, in a recent study, we showed that once H2 or H2 O is utilized prior to CO adsorption, surface formates are the dominant species present. Switching from CO to H2 O showed the rapid formation and decomposition of the surface formate at low temperature shift temperatures. Yet none of the above considerations is sufcient to provide certainty with regard to which mechanism is operating under low temperature shift conditions. To support the redox mechanism, it is necessary to demonstrate conclusively that H2 O can reoxidize partially reduced ceria under a hydrogen-containing atmosphere at a sufcient rate at the low temperature condition. In our initial work utilizing XANES spectroscopy [26], we did not nd evidence to support this claim, although O2 was found to rapidly reoxidize the ceria. Introducing H2 O : H2 mixtures into the XANES cell after rst reducing with H2 , the partially reduced ceria remained unchanged at low temperature shift temperatures. This suggests that water interacts with partially reduced ceria by establishing an equilibrium surface concentration of bridging OH groups via adsorption:desorption. Recently, the group of Burch [31] has adopted the approach of using switching between 12 CO2 and 13 CO2 and monitoring the responses of different adsorbed species during reverse water gas shift. By the principle of microscopic reversibility, one should expect that the forward shift mechanism should proceed via the reverse mechanism. It is interesting that the group claims that while under argon purge gas, formate decomposes more rapidly than Pt-CO, and that both decompose much more rapidly than carbonate at 225 C, but that Pt-CO and carbonate decomposition are much faster than formate decomposition on a reverse water gas shift feed. What is in contrast to most reports, is that for the reverse WGS, Burch et al. [31] reported that the formate decomposition was slower under a RWGS feed (i.e., presumably, in the presence of H2 O product) than when only argon purge gas was utilized. For the forward shift, it is well known that H2 O accelerates the forward decomposition rate of surface formates by a factor of 100, which led Shido and Iwasawa to include adsorbed water in the transition state of the formate decomposition step (i.e., the proposed rate limiting step). Shido and Iwasawa have reported that the activation energy of the formate decomposition decreases from 55.6 kJ/mol in vacuum to 33.3 kJ/mol in the presence of water. We have veried that water clearly accelerates the forward decomposition of surface formates [19], which at rst

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276

1269

glance appear to contradict the RWGS results of Burch et al. [31]. However, for a related material, ZnO, which was also proposed by Shido and Iwasawa [12] to involve a waterpromoted mechanism involving forward decomposition of formates, Shido and Iwasawa [32] concluded that the reverse WGS reaction operated by a somewhat different pathway. They reported, for example, that the decomposition rate of formate as a reaction intermediate was 1/101/30 of that in vacuum by the coexistence of CO2 (e.g., as found in a RWGS feed) [32]. As our primary focus is on exploring low temperature watergas shift catalysts for use in fuel cell processors, our studies have centered on utilizing a high H2 O/CO ratio. Thus far, we have not included CO2 in the feed, which will certainly be a component of interest in a real feed stream [28]. Under these conditions, we have also recently carried out studies utilizing isotopically labeled reactants for Pt/ceria [19,33], Pt/thoria [21], and more recently, Pt/MgO and Pt/ZrO2 [34]. In each case, we attempted to validate the normal kinetic isotope effect (NKIE) reported by Shido and Iwasawa for Rh/ceria [1] and MgO [11]. In the TPD studies of Shido and Iwasawa [1,11], where they tested four combinations of labeled formate and water, an important NKIE was identied and linked to the decomposition of adsorbed formate. These data supported the fact that virtually identical rate constants were obtained from steady state WGS reaction testing and those obtained from decomposition of the stabilized surface formates. A rather negligible NKIE was associated with water. For MgO, the ratio of formate decomposition rates was found to be (T = 307 C): rate(HCOO + H2 O)/rate(DCOO + H2 O) = 1.575. Likewise, for Rh/ceria, where strong formate bands were observed, analogous conclusions were drawn (T = 170 C): rate(HCOO + H2 O)/rate(DCOO + H2 O) = 1.33. What is also of interest in the Rh/ceria experiment, is that the authors observed the same kinetic isotope effect in switching from H2 O to D2 O in reaction testing using a closed circulating reactor system. Due to the apparent controversy regarding the WGS mechanism, we recently repeated the H2 O/D2 O switching studies, and found virtually identical changes in the rate, consistent with the normal kinetic isotope effect reported by Shido and Iwasawa. Similar NKIE are obtained for a range of materials for which a surface formate mechanism has been postulated: Pt/ceria [19,33], Pt/thoria [21], Pt/MgO [34], and Pt /ZrO2 [34]. In the case of Pt/ceria, we carried out studies where the surface formates were generated in a CO:N2 (3.75 ccm: 135 ccm) feed after hydrogen (or deuterium) reduction. 125 ccm of N2 was then replaced by H2 O (or D2 O) to maintain constant CO partial pressure, and the formate (H or D-labeled) response monitored. Utilizing the CO alone condition as a means to establish the total site density, we

calculated the formate coverages under WGS with H2 O or D2 O. It was clear that the H-formates were more reaction rate limited than the D-formates, indicating that they react faster. The ratio of the coverages exhibited a kinetic isotope effect identical to that observed in xed bed reaction testing in switching from H2 O to D2 O feed [33]. In another work, we monitored the transient decomposition of H and D-formates with either H2 O or D2 O. In that work, it was strongly suggested that the NKIE was associated primarily with CH (or CD) breaking of the formate, in agreement with the ndings of Shido and Iwasawa [1]. During decomposition, we witnessed an increase in infrared bands 1460 and 1390 cm1 that are consistent with unidentate carbonate, in agreement with the ndings of Shido and Iwasawa, who included unidentate carbonate (precursor to CO2 ) as a second intermediate of the WGS mechanism [1] (see schematic). At a time when it is evident that additional fundamental studies are required to further probe the WGS mechanism on metal/ceria, we continue to adopt the approach of utilizing isotopically labeled compounds and infrared spectroscopy. In this contribution, we take the historical perspective, and examine a related reaction to WGS, the decomposition of formic acid. In this study, we explore the likely possibility of a similar normal kinetic isotope effect for the water-assisted decomposition of formic acid, utilizing H and D labeled formic acid. More precisely, the compounds chosen were labeled only on the hydrogen associated with the C atom of the formic acid molecule. 2. Experimental 2.1. Catalyst preparation High surface area ceria were prepared via homogeneous precipitation of the nitrate in urea with aqueous ammonia in a similar manner to Li et al. [27], whereby urea decomposition is a slow process resulting in a slower, more homogeneous precipitation. On a basis of 30 g CeO2 , an appropriate amount of Ce(NO3 )3 6H2 O (Alfa Aesar, 99.5%) and 240 g urea (Alfa Aesar, 99.5%) were dissolved in 900 mL of deionized water, and to the solution about 30 mL NH4 OH (Alfa Aesar, 2830% NH3 ) was added dropwise ( 1 mL/ min). The mixture was then boiled at 100 C with constant stirring. The precipitate was ltered, washed with 600 mL of boiling deionized water, and dried in an oven (100 C) overnight. The dried precipitate was then crushed and calcined in a mufe furnace at 400 C for 4 h. Platinum was added via incipient wetness impregnation with tetraamine platinum (II) nitrate solution. The catalyst was calcined at 400 C for 4 h after metal component addition. 2.2. BET surface area BET Surface Area measurements were carried out in a Micromeritics Tristar 3000 gas adsorption analyzer. In each

1270

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276 Table 1 BET surface area and porosity results for catalysts Description Ceria Pt/ceria BET SA (m2 /g) 105.5 99.6 Pore volume (cm3 /g) 0.121 0.115 Average Pore Radius (nm) 2.29 2.32

trial, a weight of approximately 0.25 g of sample was used. The adsorptive gas was nitrogen (N2 ) and the adsorption was carried out at the boiling temperature of liquid nitrogen. 2.3. Temperature programmed reduction TPR was conducted on CeO2 supports and metal promoted catalysts in a Zeton-Altamira AMI-200 unit, which was equipped with a thermal conductivity detector (TCD). Argon was used as the reference gas, and 10%H2 (balance Ar) was owed at 30 ccm as the temperature was increased from 50 to 800 C at a ramp rate of 10 C/ min. 2.4. Diffuse reectance infrared Fourier transform spectroscopy (DRIFTS) A Nicolet Nexus 870 was used, equipped with a DTGSTEC detector. A high pressure/high temperature chamber tted with ZnSe windows was utilized as the WGS reactor for in situ reaction measurements. The gas lines leading to and from the reactor were heat traced, insulated with ceramic ber tape, and further covered with general purpose insulating wrap. Scans were taken at a resolution of 4 to give a data spacing of 1.928 cm1 . Typically, 128 scans were taken to improve the signal to noise ratio. The sample amount utilized was 33 mg of powder. A steam generator consisted of a downow tube packed with quartz beads and quartz wool heated by a ceramic oven and equipped with an internal thermocouple. The lines after the steam addition were heat traced. The steam generator and lines were run at the same temperature as that of the in situ sample holder of the DRIFTS cell. This allowed us to accurately bring the reactants to the desired reaction temperature. Water was added to the steam generator by a thin needle welded to a 1/16th in line. A precision ISCO Model 500D syringe pumps was used to feed the water. Feed gases were controlled by using Brooks 5850 series E mass ow controllers. Iron carbonyl traps consisting of lead oxide on alumina (Calsicat) were placed on the CO gas line. All gas lines were ltered with Supelco O2 /moisture traps. 2.5. Fixed bed reactor system
1 in downow tubular reactor was utilized. ApproxiA 2 mately 25 mg of Pt/ceria was diluted with inert silica to give a bed containing 0.4 g of the mixture. A bed of quartz wool upstream of the bed was added to equalize the gas ow. The water/formic acid mixture was added to the reactor by a thin needle welded to a 1/16th in line. A precision ISCO Model 500D syringe pumps was used to feed the water/formic acid mixture. A gas chromatograph (SRI 8610C) was used to monitor the products of reaction. The GC includes two columns (6 silica gel packed and 3 molecular sieve packed) and two detectors (FID and TCD). To boost the sensitivity of the CO

and CO2 signals, the GC incorporates a methanizer, such that these products can be analyzed by FID. An internal standard of N2 was also employed. A bypass line allowed the product ow to be directed to a 40 cc knockout trap, prior to sending the dry gas to the switching valve of the GC.

3. Results and discussion 3.1. Standard characterization BET surface areas and porosity parameters of ceria and Pt/ceria are included in Table 1. Addition of urea to the preparation procedure has allowed us to double the surface area of ceria over standard precipitation using ammonium hydroxide. Previous XRD measurements have indicated that the domain size of the ceria was approximately 8.4 nm. TPR proles have been reported before (e.g., see [19]), but are reproduced here for clarity. Fig. 1a reveals two important features. By comparing the reduction proles of sintered ceria and high surface area ceria, the bulk ceria is assigned to reduction at close to 750 C. Reduction of the surface shell of ceria, present as a large broad peak on the high surface area ceria support, occurs between 400 and 500 C. Pt catalyzes and therefore shifts the peak for the reduction of the surface shell of ceria to lower temperatures, but has little effect on the bulk. We have conducted XANES experiments [26] which directly measure the extent of the partial reduction of ceria, not reproduced here, which support the above conclusions. After reduction, the bridging OH groups are clearly produced on the surface of ceria, as depicted in Fig. 1b (reproduced from [19]). 3.2. Reaction tests in a xed bed reactor To determine the appropriate temperature to conduct the isotope switching studies using HCOOH and DCOOH molecules, it was rst necessary to ensure that the test was carried out away from equilibrium. The dehydrogenation reaction is slightly exothermic (7.38 kcal/mol), while that of the dehydration is slightly endothermic (+2.45 kcal/mol). In tests performed, the dehydrogenation products were preferred. Therefore, the conversion versus temperature plot was rst obtained, as shown in Fig. 2. Based on the lightoff curve using HCOOH, we decided to carry out the switching test at 210 C, far from the equilibrium conversion

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276

1271

Isotope Switching at 250C Bridging OD 1%Pt/CeO2 HSA urea TCD Signal (a.u.) Intensity (a.u.) Ceria HSA urea prep Ceria sintered at 1000C a 0 100 200 300 400 500 600 700 800 900 T (deg C)

Bridging OH c

3800 3600 3400 3200 3000 2800 2600 Wavenumbers (cm-1)

Fig. 1. TPR proles of, moving upward, bulk sintered ceria, high surface area ceria and Pt-promoted high surface area ceria (left); single beam intensity spectra of the bridging OH (OD) groups, as identied by infrared spectroscopy.

1.0

kinetic isotope effect in conversion was found to be approximately 1.2, while that of the product selectivity (CO2 /CO + CO2 ) was about 1.1, resulting in an overall NKIE for the production of CO2 to be approximately 1.3.

0.8 Formic Acid Conversion

3.3. In situ DRIFTS spectroscopy


0.6

0.4

0.2

0.0 140 160 180 200 220 240 260 280 300 320 Temperature (C)
Fig. 2. Reaction testing in a xed bed reactor utilizing a feed consisting of 135 ccm N2 , 125 ccm H2 O, and 3.75 ccm formic acid.

condition. Fig. 3a through c show the results of the switching from a feed containing DCOOH to one containing HCOOH. For an average of four points at each condition, a normal

As mentioned previously, reduction in H2 led to the production of bridging (i.e., Type II) OH groups on the surface of ceria, as shown in Figs. 1b and 4. During reduction, it is apparent that bands indicative of surface carbonates decrease during the reduction process, in agreement with earlier studies [16,19]. The reason that the bands have been assigned to carbonates is that during their decomposition Pt-CO bands evolve and then diminish. Furthermore, there are no CH stretching vibrations that would indicate that the bands are surface formates. Figs. 5 and 6 are for infrared studies for DCOOH and HCOOH formic acid, respectively. Figs. 5a and 6a show that adsorption of formic acid leads to a pronounced decrease in the bridging OH groups. Upon reaction with these sites, the bidentate surface formate bands evolve on the surface. These have been previously identied by the asymmetric (1580 cm1 ) and symmetric (1375 cm1 ) OCO stretching vibrations (Figs. 5b, 6c) accompanied by CH (or CD) stretching vibrations (Figs. 5b, 6b) at 2845 (2150 cm1 ) and 2945 cm1 (2180 cm1 ), respectively (A. Holmgren et al. [14]). The process agrees with the dissociative adsorption schematic of Mars et al. [3], as described in the Introduction.

1272

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276

0.7 D-formic acid Formic Acid Conversion 0.6 H-formic acid Carbon Dioxide Selectivity

1.00 D-formic acid 0.95 H-formic acid

0.5

0.90

0.4

0.85

0.3 (a) 0 20 40 60 80 Time (min) 20 40 60 80 Time (min) 0.7 D-formic acid Carbon Dioxide Yield 0.6 (b)

0.80 0 20 40 60 80 Time (min) 20 40 60 80 Time (min)

H-formic acid

0.5

0.4

0.3 (c) 0 20 40 60 80 Time (min) 20 40 60 80 Time (min)

Fig. 3. Reaction testing in a xed bed reactor utilizing a feed consisting of 135 ccm N2 , 125 ccm H2 O, and 3.75 ccm formic acid at 210 C and employing 25 mg catalyst: (a) upper leftchange in conversion in switching from DCOOH to HCOOH, (b) upper rightchange in dehydrogenation product selectivity in isotope switching and (c) bottomchange in dehydrogenation product yield accompanying isotope switch.

We write the process as follows:

generated from the adsorptive reaction of the bridging OH groups with CO, as shown in Fig. 7. It is clear that the bands are identical, and that CO can react with these groups to produce the same bidentate formate species:

It is instructive to compare the bands produced by reacting the bridging OH groups with formic acid with those

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276


Reduction in H :N mixture Carbonate Region Bridging OH Activation 250C 200C 150C 100C RT 1%Pt/ceria 4000 3500 3000 2500 2000 1500 1000

1273

Wavenumbers (cm )

Fig. 4. In situ DRIFTS spectra of the reduction of 1%Pt/ceria in H2 : N2 mixture (100 ccm:135 ccm).

When considering the earlier comments of Mars et al. [3,4], Tamaru [5,6] and Fahrenfort et al. [7], it is important to consider the stability of the formate species produced, as this will largely govern the type of mechanism that will occur. If the formate is rather stable, then its decomposition can be the rate controlling factor in the mechanism. In a separate study (results not included for the sake of brevity), a TPD under inert gas was performed after the bidentate formates were generated from the dissociative adsorption of formic acid. The formates, as we observed earlier with the formates generated from CO adsorption [19], were quite stable until temperatures in the neighborhood of 300 C were reached, where they were observed to decompose thermally (Fig. 8). However, as shown in Fig. 9 the formates readily decompose in the forward direction (i.e., via unidentate carbonate
OCO Stretching Surface formates

Single Beam intensity (a.u.)

0.0

C-D Stretching Surface formates

Absorbance

Absorbance

Pt-CO

-0.1

3900

3800

3700

3600

3500

3400

2200

2000

1800

1600

1400

1200

(a)

Wavenumbers (cm-1) C-D Stretching Surface Formates Pt-CO

(b)

Wavenumbers (cm-1)

OCO Stretching Surface Formates

C-D Stretching Surface Formates Pt-CO

OCO Stretching Surface Carbonate

Intensity (a.u.)

2200 2000 1800 1600 1400 1200 1000

Intensity (a.u.) 2200

2000

1800

1600

1400

1200

1000

(c)

Wavenumbers (cm-1)

(d)

Wavenumbers (cm-1)

Fig. 5. (a) Change in bridging OH groups upon adsorption of DCOOH at 140 C, (b) evolution of bidentate OCO and CD bands during adsorption of DCOOH, (c) single beam spectrum prior to formate decomposition by H2 O at 140 C, (d) single beam spectrum indicating growth of unidentate carbonate bands after formate decomposition. 10 10 l of formic acid added.

1274

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276


C-H Stretching Surface formates Absorbance Absorbance OCO Stretching Surface formates

Absorbance

Pt-CO

3900 3800 3700 3600 3500 3400

3100 3000 2900 2800

2200 2000 1800 1600 1400 1200

(a) Wavenumbers (cm-1)


Pt-CO

(b) Wavenumbers (cm-1)

(c)

Wavenumbers (cm-1) OCO Stretching Surface Carbonate Pt-CO

OCO Stretching Surface Formates Intensity (a.u.) Intensity (a.u.) 2000 1500 1000 C-H Stretching Surface Formates

C-H Stretching Surface Formates

3000

2500

3000

2500

2000

1500

1000

(d)

Wavenumbers (cm-1)

(e)

Wavenumbers (cm-1)

Fig. 6. (a) Change in bridging OH groups upon adsorption of HCOOH at 140 C, evolution of bidentate, (b) CH and (c) OCO bands during adsorption of HCOOH, (d) single beam spectrum prior to formate decomposition by H2 O at 140 C, (e) single beam spectrum indicating growth of unidentate carbonate bands after formate decomposition. 10 10 l of formic acid added.

C-H Surface Formates

OCO Surface Formates

Absorbance

Absorbance

3900 3800 3700 3600 3500

3100 3000 2900 2800 2700 2600

Wavenumbers (cm-1)

Wavenumbers (cm-1)

Absorbance
1800 1700 1600 1500 1400 1300 1200

Wavenumbers (cm-1)

Fig. 7. (Left) Change in bridging OH groups upon adsorption of CO at 250 C; evolution of bidentate (middle) CH and (right) OCO bands during adsorption of CO; dotted line shows the decrease of the CH band of formate during water gas shift, indicating that it is a likely intermediate. Here, a high H2 O : CO ratio was utilized (3.75 ccm CO: 125 ccm H2 O : 10 ccm N2 ). The catalyst shown here was not the same batch as used in the formic acid decomposition measurements.

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276

1275

0.4

1.0 0.9 Normalized Absorbance 0.8 0.7 0.6 0.5 0.4 0.3 0.2 H-formate with H2O D-formate with H2O

140C

0.3

Absorbance

0.2

0.1

0.1 0.0 0 2 4 6 8 10 12 14 Time (minutes) 16 18 20

0.0 0 2 4 6 8 10 Time (minutes) 12 14 16

Fig. 9. Water-assisted decomposition of H and D-formates at 140 C utilizing a feed containing 125 ccm H2 O and 135 ccm N2 .

Fig. 8. Thermal decomposition of surface formate at 300 C, following the CH band at 2850 cm1 in inert gas ow (135 ccm N2 ).

formation) at 140 C in the presence of water. That the formates decompose to carbonate is easily seen by comparing the spectra before and after reaction with D2 O in Fig. 5c and d, or by H2 O in Fig. 6d and e. Also shown in the gures, Pt-CO decreases only slightly during the formate decomposition, indicating that the species is not likely participating in the mechanism. The D-formates were observed to decompose more slowly than H-formate, as we have observed in a previous study, whereby the formates were generated by CO adsorption [19], pointing again to a normal kinetic isotope effect. ,

close to the normal kinetic isotope effect of 1.4 reported in several studies of water gas shift on metal promoted ceria. The slower D-formate decomposition rate versus H-formate in the presence of water during transient decomposition is consistent with the idea that CH bond breaking of formate is very likely the rate limiting step for water-assisted formic acid decomposition over partially reduced ceria. Overall, taking into account the identication of the same adsorbed reactive species and similar kinetic isotope effects, the ndings suggest an analogous mechanism operating between formic acid decomposition and water gas shift.

References
[1] Shido T, Iwasawa Y. Reactant-promoted reaction mechanism for watergas shift reaction on Rh-doped CeO2 . J Catal 1993;141:7181. [2] Mailhe A, Sabatier P. Catalytic decomposition of formic acid. Compt Rend Acad Sci 1911;152:12125. [3] Mars P. The decomposition of formic acid on oxides. Proceedings of symposium on the mechanism of heterogeneous catalysis, Amsterdam, The Netherlands, 1959. [4] Mars P, Scholten JJF, Zwietering P. The catalytic decomposition of formic acid. Adv Catal 1963;14:35113. [5] Tamaru K. Adsorption during the catalytic decomposition of formic acid on silver and nickel catalysts. Trans Faraday Soc 1959;55:82432. [6] Tamaru K. Decomposition of formic acid on metal catalysts. Adsorption measurements during surface catalysis. Bulletin of the faculty of engineering, Yokohama National University 1959;8:8192. [7] Fahrenfort J, van Reijen L, Sachtler WMH. The decomposition of formic acid on metal catalysts. Proceedings of the symposium on the mechanism of heterogeneous catalysis, Amsterdam, The Netherlands, 1960.

4. Conclusions The results are quite convincing that there is an important link between the WGS and formic acid decomposition on partially reduced ceria promoted with noble metal. Partial reduction of the surface leads to the formation of bridging Type II OH groups. These were found to react with formic acid to yield bidentate formate and water, or with CO directly to form bidentate formate. In both WGS and formic acid decomposition, we have observed an important normal kinetic isotope effect consistent with the idea that CH bond cleaving of formate is involved in the rate limiting step. It is to be pointed out that in our conditions, a high water/CO or water/formic acid ratio was maintained. In the current investigation, switching between a feed containing DCOOH to HCOOH led to an increase in the formic acid conversion, as well as a slight increase in the CO2 selectivity. The overall change in the CO2 yield was approximately 1.3, very

1276

G. Jacobs et al. / International Journal of Hydrogen Energy 30 (2005) 1265 1276 [22] Jung K, Bell AT. Role of hydrogen spillover in methanol synthesis over Cu/ZrO2 . J Catal 2000;193:20723. [23] He D, Ding Y, Luo H, Li C. Effects of zirconia phase on the synthesis of higher alcohols over zirconia and modied zirconia. J Molec Catal A: Chemical 2004;208:26771. [24] Jacobs G, Williams L, Graham UM, Sparks DE, Thomas G, Davis BH. Low temperature watergas shift: in situ DRIFTSreaction study of ceria surface area on the evolution of formates on Pt /CeO2 fuel processing catalysts for fuel cell applications. Appl Catal A 2003;252:10718. [25] Jacobs G, Chenu E, Patterson P, Williams L, Sparks D, Davis BH. Watergas shift: comparative screening of metal promoters for metal/ceria systems and role of the metal. Appl Catal A 2004;258:20314. [26] Jacobs G, Patterson PM, Williams L, Chenu E, Graham UM, Sparks DE, Thomas G, Davis BH. Watergas shift: in situ spectroscopic studies of noble metal promoted ceria catalysts for CO removal in fuel cell reformers and mechanistic implications. Appl Catal A 2004;262:17787. [27] Li Y, Fu Q, Flytzani-Stephanopolous M. Low-temperature watergas shift reaction over Cu- and Ni-loaded cerium oxide catalysts. Appl Catal B 2000;27:17991. [28] Ghenciu AF. Review of fuel processing catalysts for hydrogen production in PEM fuel cell systems. Curr Opin Solid State Mat Sci 2002;6:38999. [29] Hilaire S, Wang X, Luo T, Gorte RJ, Wagner J. A comparative study of watergas-shift reaction over ceria supported metallic catalysts. Appl Catal 2001;215:2718. [30] Bunluesin T, Gorte R, Graham G. Studies of the watergas shift reaction on ceria-supported Pt, Pd, and Rh: implications for oxygen-storage properties. Appl Catal B 1998;15:10714. [31] Tibiletti D, Goguet A, Meunier FC, Breen JP, Burch R. On the importance of steady-state isotopic techniques for the investigation of the mechanism of the reverse watergas shift reaction mechanism. Chem Commun 2004;14:16367. [32] Shido T, Iwasawa Y. The effect of coadsorbates in reverse watergas shift reaction on ZnO, in relation to reactantpromoted reaction mechanism. J Catal 1993;140:57584. [33] Jacobs G, Khalid S, Patterson PM, Sparks DE, Davis BH. Watergas shift: kinetic isotope effect identies surface formates in rate limiting step for Pt/ceria catalysts. Appl Catal A 2004;268:25566. [34] Chenu E, Jacobs G, Crawford A, Keogh RA, Patterson PM, Sparks DE, Davis BH. Watergas shift: an examination of Pt promoted MgO and tetragonal and monoclinic ZrO2 in-situ DRIFTS, submitted.

[8] Hirota K, Kuwata K, Nakai Y. Mechanism of catalytic decomposition of formic acid on silver catalysts. Bull Chem Soc Japan 1958;31:8614. [9] Yamashita K, Naito S, Tamaru K. The behavior of surface formate ions as reaction intermediates in the decomposition of methanol over Cr 2 O3 . J Catal 1985;94:353. [10] Li C, Domen K, Maruya K, Onishi T. Spectroscopic identication of adsorbed species derived from adsorption and decomposition of formic acid, methanol, and formaldehyde on cerium oxide. J Catal 1990;125:44555. [11] Shido T, Asakura K, Iwasawa Y. Reactant-promoted reaction mechanism for catalytic watergas shift reaction on MgO. J Catal 1990;122:5567. [12] Shido T, Iwasawa Y. Reactant-promoted reaction mechanism for watergas shift reaction on ZnO, as the genesis of the surface catalysis. J Catal 1991;129:34355. [13] Shido T, Iwasawa Y. Regulation of reaction intermediate by reactant in the watergas shift reaction on CeO2 , in relation to reactant-promoted mechanism. J Catal 1992;136:493503. [14] Holmgren A, Andersson B, Duprez D. Interactions of CO with Pt/ceria catalysts. Appl Catal B 1999;22:21530. [15] Li C, Sakata Y, Arai T, Domen K, Maruya K, Onishi T. Adsorption of carbon monoxide and carbon dioxide on cerium oxide studied by Fourier-transform infrared spectroscopy. 2. Formation of formate species on partially reduced cerium dioxide at room temperature. J Chem Soc Faraday Trans 1989;85:145161. [16] Laachir A, Perrichon V, Badri A, Lamotte J, Catherine E, Lavalley JC, El Fallah J, Hilaire L, Le Normand F, Quemere E, Sauvion GN, Touret O. Reduction of CeO2 by hydrogen. Magnetic susceptibility and Fourier-transformed infrared, ultraviolet, and X-ray photoelectron spectroscopy measurements. J Chem Soc Faraday Trans 1991;87:16019. [17] Binet C, Badri A, Lavalley JC. A spectroscopic characterization of the reduction of ceria from electronic transitions of intrinsic point defects. J Phys Chem 1994;98:63928. [18] Jacobs G, Williams L, Graham UM, Sparks DE, Davis BH. Low temperature watergas shift: in-situ DRIFTS-reaction study of a Pt /CeO2 catalyst for fuel cell reformer applications. J Phys Chem B 2003;107:10398404. [19] Jacobs G, Patterson P, Williams L, Graham UM, Sparks DE, Davis BH. Low temperature watergas shift: kinetic isotope effect observed for decomposition of surface formates for Pt/ceria catalysts. Appl Catal A 2004;269:6373. [20] Lamotte J, Lavalley JC, Druet E, Freund E. Infrared study of acid-base properties of thorium dioxide. J Chem Soc Faraday Trans 1983;79:221927. [21] Jacobs G, Crawford A, Williams L, Patterson PM, Davis BH. Low temperature watergas shift: comparison of thoria and ceria catalysts. Appl Catal A 2004;267:2733.

Vous aimerez peut-être aussi