Activity and Stability of Cu-CeO2 Catalysts in High-Temperature Water-Gas Shift for Fuel-Cell Applications

Copper-containing cerium oxide materials are shown in this work to be suitable for the high-temperature water-gas shift (WGS) reaction integrated with hydrogen separation in a membrane reactor to generate pure hydrogen. Copper-ceria is a stable high-temperature shift catalyst, unlike iron-chrome catalysts that deactivate severely in CO2-rich gases. Such gas mixtures will prevail if a catalytic membrane reactor is used to remove hydrogen. We also found that iron oxide-ceria catalysts have much lower activities than copper-ceria catalysts. Ceria participates in the WGS reaction; its surface properties are crucial for high activity and are sensitive to the presence of dopants. The kinetics of the WGS reaction over 10 atom % Cu-Ce-(30 atom % La)Ox were measured in the temperature range 300-450 °C. A strong dependence on CO and a weak dependence on H2O were found at 450 °C, whereas inhibition by the reaction products was weak. The apparent activation energy over the catalyst stabilized in the reaction gas mixture at 450 °C is 70 kJ/mol. The catalyst lost some activity in the initial time on stream but was stabilized thereafter. A loss of catalyst surface area ( 20%) and copper enrichment of the ceria surface during the WGS reaction at 450 °C can explain the observed activity loss.

Introduction

The exothermic water-gas shift (WGS) reaction, CO + H2O ↔ CO2 + H2, is used industrially for the production of hydrogen for use in ammonia synthesis and in adjusting the CO/H2 ratio for subsequent synthesis of methanol. Recently, there has been a renewed interest in the WGS reaction because of its potential use in supplying hydrogen for fuel-cell power generation. Fuel cells are currently undergoing rapid development for both stationary and transportation applications.

Industrially, the WGS reaction is carried out in two temperature regimes: High-temperature (320-450 °C) shift reactors use an Fe2O3-Cr2O3 catalyst^1,2 that can effectively reduce CO from several percentage points to the equilibrium CO value dictated by the operating temperature and composition; further reduction of CO takes place at low temperatures over a more active catalyst based on Cu-ZnO.^1,3 Another industrially applied catalyst is Co-Mo/Al2O3 (used in the 150-400 °C temperature range), which is sulfur-tolerant.^1,4 The Cu-ZnO catalyst is very sensitive to temperature excursions, operating in the narrow temperature window of 200-250 °C; requires careful activation in H2 gas; and is readily deactivated by exposure to air or by water condensation.

Optimization of the WGS system for the production of hydrogen for fuel cells is of particular interest to the energy industry. The high-temperature shift (HTS) can be used to increase the hydrogen content of coal gas produced by gasification or of reformate gas produced by autothermal reforming of fuel oils. To this end, it is desirable to couple the WGS reaction to hydrogen separation using a semipermeable membrane, with both processes taking place at high temperatures to improve the reaction kinetics and permeation.⁵ Reduced equilibrium conversion of CO at high temperatures is overcome by product H2 removal via the membrane. There are several challenges in developing a HTS membrane reactor that is small and cost-effective. If Pd-based, the membrane should contain a minimal amount of Pd, and the catalyst should be both robust and active in CO2-rich gas. This type gas will be present over much of the catalyst, as the membrane removes the hydrogen produced from the water-gas shift reaction.

Commercial catalyst formulations containing iron oxide were recently found to deactivate in CO2-rich gases.⁶ In this work, we examine Cu-containing ceria catalysts as an alternative to the commercial iron-chrome HTS catalysts. For comparison, we also pre-pared and studied iron oxide-containing ceria catalysts.

CeO2-based catalysts have been found promising for the WGS reaction.^7-14 When treated in a reducing atmosphere (such as CO, H2, and hydrocarbons) at elevated temperatures, CeO2 forms a continuum of oxygen-deficient, nonstoichiometric oxides, CeO2-x (0 < x e 0.5),¹⁵ that can be reoxidized to CeO2 when exposed to an oxidizing environment (such as oxygen, water vapor, or NO). CeO2 retains its fluorite-type crystal structure even after the loss of considerable amounts of oxygen and the formation of a large number of oxygen vacancies. Doping of ceria with ZrO2 or La2O3 increases its reducibility, oxygen storage capacity, and resistance to sintering.^16,17 Noble metals (Rh, Pt, Pd)^7,8 and non-noble metals (Au, Cu)^9-14 on CeO2 have been reported as promising low-temperature WGS catalysts. Bunluesin et al.⁷ found that Pt, Pd, and Rh supported on ceria had specific WGS reaction rates (normalized to the catalyst area) identical for each of the metals and much higher than those for either pure ceria or each of the precious metals supported on alumina. They proposed that the water-gas shift on ceria-supported precious metals occurs through a bifunctional redox mechanism in which CO adsorbed on the precious metal is oxidized by ceria, which, in turn, is oxidized by water. Rivaling the precious metal-ceria catalysts is the Au-ceria system, which was recently reported as both very active and  stable in the WGS reaction at temperatures in the range of 175-350℃ and in realistic feed gas compositions.^9-11

Li et al.¹³ recently reported on the WGS activity of Cu- and Ni-containing ceria catalysts over the temper-ature range of 175-300 °C. The catalysts were prepared by the urea gelation coprecipitation method, which disperses the metal/metal oxide in the form of clusters on ceria, resulting in a large interfacial area and interaction with the reducible cerium oxide support. The content of Cu or Ni was in the range of 5-15 atom % (2-8 wt %), and 10 atom % La2O3 was doped into ceria as a structural stabilizer.The 5% Cu-Ce(La)Ox catalyst was  reported to retain high activity and stability at temperatures up to 600 °C.¹³ However, the gas used in that work comprised only CO and H2O. It is important to examine the stability of such catalysts in the presence of realistic reformate gas compositions. Such gases were used in a recent kinetic study of Cu-based shift catalysts, including copperceria, but only at 200 °C.¹⁴ The high-temperature stability of such catalysts needs to be evaluated.

The choice of copper-ceria for HTS applications is further rationalized by the following consideration: The formulation contains a nonprecious metal; hence, it would be a much more cost-effective catalyst for the intended application. It is also a better choice than nickel-ceria, which catalyzes the methanation reac-tion,¹² and it has better activity than gold-ceria at high temperatures.¹² Several groups have reported that the addition of a small amount of CeO2 into the Fe2O3-Cr2O3 catalyst or Cr-free Fe2O3 catalyst enhances the performance of the catalyst.^18-20 To examine whether iron oxide supported on ceria has as good activity as the copper-ceria system, Fe-Ce(La)Ox was also studied in this work.

Experimental Section

Catalyst Preparation. Cu- or Fe-containing ceria catalysts were prepared by the urea gelation coprecipitation (UGC) method,¹³ which produces more homogeneous mixed oxides with finer particle sizes than conventional coprecipitation. The UGC method consists of the following steps: (1) Dissolve urea; (NH4)2Ce-(NO3)6, La(NO3)3, or ZrO(NO3)2; and Cu(NO 3)2â3H2O or Fe(NO3)3â9H2O in required amounts in deionized water. (2) Heat the solution to boiling, (3) adding water when precipitation begins (about 30 min after boiling). (4) Age the precipitate in boiling water with constant stirring and addition of water for 8 h. (5) Filter and wash the precipitate twice in 50-70 °C water with constant stirring for 30 min each time and (6) dry the precipitate at about 100 °C for 8 h. The dried lump was then crushed into particles smaller than 150 ím in diameter and calcined in air by being heated slowly (2 °C/min) to 650 °C and kept at this temperature for 4 h.

Several catalyst formulations were prepared with La2O3 or ZrO2 used as a dopant in ceria in amounts ranging from 8 to 30 atom %. Dopants were added to improve the thermal stability of ceria.^16,17 Copper or iron oxide was added as a minor component in ceria. All compositions are expressed in atomic percentages. As an example, the notation 10% Cu-Ce(30% La)Ox means

Cu / (Cu + Ce + La) = 0.10

Ce / (Cu + Ce + La) =0.90×    0.70

and         La/ (Cu + Ce + La) =0.90×0.30

C12-4-02, a commercial catalyst containing 80-95 wt% Fe2O3, 5-10 wt % Cr2O3, <5 wt % CrO3, 1-5 wt % CuO, and 1-5 wt % graphite, was supplied by United Catalysts, Inc., in pellet form and was used after being crushed into <150-ím-size particles.

Catalyst Characterization. The elemental composition of each sample was determined by inductively coupled plasma (ICP) spectrometry. Hydrogen peroxide and nitric acid were used to dissolve the solids at room temperature.

The BET surface area of each sample was measured by single-point nitrogen adsorption/desorption cycles in a Micromeritics Pulse Chemisorb 2705 instrument, using a 30% N2/He gas mixture.

X-ray powder diffraction (XRD) analysis was per-formed on a Rigaku 300 X-ray diffractometer with rotating anode generators and a monochromatic detector. Cu KR radiation was used. Samples in fine powder form were directly pressed onto a ³/4-in. by ⁵/8-in. frosted area etched on a glass holder. A small amount of tungsten powder was added to the sample to calibrate the peak position. The crystal size of ceria was deter-mined by the Scherrer equation,²¹ and the lattice constant of ceria was determined from Bragg's law.²¹

Temperature-programmed reduction (TPR) tests with H2 were run in the same instrument according to the following procedure: A sample of about 1 g was heated to 350 °C (10 °C/min) in 50 mL/min (NTP) 20% O2/He and kept at this temperature for 30 min to fully oxidize the sample. Heating was then stopped, and when the temperature dropped to 200 °C, purge gas, 50 mL/min N2 (99.999%), was switched in. TPR began after the sample had cooled to room temperature. The reduction gas was 20% H2/N 2 (50 mL/min), and the heating rate was 5 °C/min.

Temperature-programmed oxidation (TPO) tests were run according to the following procedure: The sample was heated to 350 °C (10 °C/min) in He (99.999%, 50 mL/min) and kept at 350 °C for 30 min to remove any CO2 adsorbed. After the sample had cooled to room temperature, it was heated in a 20% O2/He (50 mL/min) gas mixture to 650 °C at a heating rate of 5 °C/min. A mass spectrometer (MKS-model RS-1) was used for detection of CO2.

X-ray photoelectron spectroscopy (XPS) was per-formed using a Perkin-Elmer 5200C instrument with a 300-W (15-kV by 20-mA) aluminum KR anode as the X-ray source. Samples used in XPS analysis were in powder form pressed on a double-sided adhesive copper tape. The tape with the sample was mounted on a sample holder and introduced into the XPS vacuum chamber. Analysis was performed after a desired low pressure in the vacuum chamber had been reached.

Activity Tests in a Packed-Bed Microreactor. CO conversion tests and WGS reaction rate measurements were performed at atmospheric pressure. Catalyst samples (<150 ím in size) were diluted with quartz salt and loaded onto a quartz frit at the center of a quartz-tube microreactor, which was heated inside an electric furnace. All samples were used without activation. The gases used were helium (grade 5.0), 10% CO/He or 50% CO/He, 50% CO2/He or pure CO2 (grade 4.0), and 50% H2/He or pure H2 (ultrahigh purity). All gas mixtures were certified calibration gases. A hydrocarbon  trap was connected  at the outlet of the CO/He gas cylinder to remove any iron carbonyl from the CO/He mixture. Water was rejected  into the flowing gas stream by  a calibrated syringe pump and was vaporized in the heated gas feed line before entering the reactor. A condenser filled with ice was installed at the reactor exit to collect water. The exit gas was analyzed with a Hewleet-Packward 5890A gas chromatograph, equipped with a thermal conductivity detector and a ¼-in. diameter  × 6-ft-long Carbosphere column for CO and CO2 separation. The CO conversion was calculated from the concentrations of CO and CO2 detected. No methane formation occurred, even when H2 and CO2 were included in the feed gas mixture.

Activation energies were calculated from  the reaction rates at CO conversions of ＜20%. The reaction rate was calculated according to the expression : 

Where Nco is the molar flow rate of CO in the feed gas in moles per second, Xco is the percentage conversion of CO to CO2, and Wcat is the catalyst weight in grams.

Results and Discussion

Table 1 lists the bulk compositions, BET surface areas, ceria crystallite sizes, and ceria lattice constants of the catalyst samples prepared and tested in this work. The compositions in the bulk are almost the same as those in the preparation. The samples containing up to 15 atom 5 Cu or Fe , after  calcinations at 650℃， have surface areas of around 90m2/g. at higher than 15 atom % content, addition fo Cu of Fe decrease the surface area of the La-doped ceria.

It has been reported by Li et al.¹³ that, in the copper-ceria system, Cu or CuO nanoparticles cannot be detected by XRD up to ~ 15 atom % copper content. Similarly, we did not see reflections of Cu compounds in all of the samples listed in Table 1. Particles of  Fe  or its oxides in our iron-ceria catalysts are also too small to be detected by XRD. In the Zr-doped samples and the 8 atom % La-doped samples, there was no reflection of lanthana or zirconia because both of these oxides go into solid solution with ceria, and the only XRD-detectable phase was ceria. In the samples with 30 atom % La, small amounts of hexagonal, tetragonal, and monoclinic crystals of lanthanum oxide carbonate (La2CO5) were identified in addition to ceria (Table 1); only hexagonal La2CO5 still existed after use of the catalyst in the WGS reaction. The lattice parameter of ceria (5.41 Å) in-creases with La³⁺ doping and decreases with Zr⁴⁺ doping, as a result of the difference in the ionic radii (La³⁺ > Ce⁴⁺ > Zr⁴⁺).

Used samples (either 32 or 20 h) lost about 20% of their surface area (Table 1). Concomitantly,  the crystal size of ceria increased by about 20% as measured by XRD.

 Cu- or Fe-Ce(8% La)Ox  catalysts (with different content of Cu or Fe ) were screened in a 2% CO-10% H2O-He gas mixture. The catalyst amount was 0.15 g, and the contact time was 0.09 g s cm^-3. The space velocity (SV) was 80 000 h^-1 (NTP), except for the commercial iron-chrome catalyst, C12-4-02 (48 000 h^-1), because of its different density. Steady-state CO conversion plots are shown in Figure 1. The Fe-Ce(8% La)Ox samples have higher activities than Ce(8% La)-Ox, indicating that addition of Fe increases the activity of ceria. The commercial catalyst C12-4-02 is more active than the Fe-Ce(8% La)Ox samples, but all of the Cu-Ce(8% La)Ox samples are much more active than C12-4-02. The CuO contained in C12-4-02 might be the reason that C12-4-02 is more active than the Fe-Ce-(8% La)Ox samples. Among the copper-ceria samples, 10% Cu-Ce(8% La)Ox and 15% Cu-Ce(8% La)Ox were the most active. The 13% Fe-Ce(8% La)Ox was the most active among the iron-ceria samples.²²

H2 TPR data for various catalyst compositions are presented in Figures 2 and 3 and Table 2. These results help us understand the differences in WGS activity among the catalysts shown in Figure1. The low-temperature peak (surface oxygen reduction) of ceria shifts to lower temperatures upon addition of Cu. As shown in Figure 2, the presence of even 1 atom % Cu shifts the peak reduction temperature by more than 100 °C. With higher copper contents, the effect is much more pronounced. Thus, Cu dramatically increases the reducibility of ceria, as has been amply documented in the literature.^13,17,23 As shown in Table 2, the normalized hydrogen consumption does not change much with addition of copper above 1 atom %. There is no need to use ceria catalysts containing more than 10 atom % copper, in agreement with the activity data of Figure 1. Figure 1 shows that the sample containing 40 atom % Cu was inferior, probably because of the lower surface area and further sintering during reaction. Addition of Fe increases the reducibility of ceria but to a lesser extent than Cu, as shown in Figure 3 and Table 2. Promotion of iron-based shift catalysts by ceria has been reported in the literature and attributed to enhancement of reducibility.^19,20 The activity tests of the iron oxide-ceria catalysts performed in the present work correlate well with the H2 TPR results of Figure 3. The best iron-ceria catalyst shown in Figure 1 was the one containing 13 atom % Fe. Interestingly, as shown in Figure 3, for the sample containing 13 atom % Fe, the iron oxide reduction begins and peaks at lower temperatures. The reduction of the bulk oxygen of ceria is not affected by the presence of either iron (Figure 3) or copper oxide (not included in Figure 2).

In many practical cases, hydrogen is obtained from coal gasification. As mentioned above, CO2 enrichment of the gas occurs when the WGS reaction is integrated with hydrogen permeation through a dense membrane. Therefore, the WGS activities of the 10% Cu-Ce(8% La)Ox and 13% Fe-Ce(8% La)Ox catalysts were tested and compared with that of the commercial catalyst C12-4-02 in a simulated coal gas mixture with a molar composition 10% CO-34% H2O-10% CO2-15% H2-He (here helium is used instead of N2) and in a CO2-rich gas containing 2% CO-10% H2O-35% CO2-He, both at a contact time of 0.09 g s cm^-3. In both gas mixtures (Figures 4 and 5), the ranking of the catalysts is the same as that in Figure 1, i.e., the 10% Cu-Ce(8% La)-Ox is more active than C12-4-02, which is superior to 13% Fe-Ce(8% La)Ox. Lund et al. ⁶ have reported a dramatic poisoning effect of CO2 on iron oxide-based catalysts, whereby the reaction rate decreased by several orders of magnitude in CO2-rich gas. Copper-ceria, on the other hand, exhibits very good activity, reaching equilibrium CO conversions above 450 °C in the simulated coal gas mixture and at high space velocity (80 000 h^-1, NTP, Figure 4). After all the measurements from  300 to 600 °C (50 °C intervals, 1.5 h at each tempera-ture), the CO conversions over both 10% Cu-Ce(8% La)-Ox and C12-4-02 were checked again at 400 °C, as shown by the single filled triangle and the single circle, respectively, in Figure 4. The CO conversions over both catalysts dropped, but both were still higher than that over the 13% Fe-Ce(8% La)Ox sample at 400 °C before exposure to high temperatures; the conversion over 10% Cu-Ce(8% La)Ox was still higher than that over C12-4-02. The activity of copper-ceria is stable in the gas mixture with the artificially high CO2 content (Figure 5). Thus, this type of catalyst has merits for the high-temperature shift reaction carried out in a membrane reactor.

 Effect of Dopant. Steady-state conversions of CO in the WGS reaction over the 10% Cu-Ce(dopant)Ox samples with dopants 8% La, 30% La, and 24% Zr were measured in the microreactor in a gas mixture of 10% CO-30% H2O-10% CO2-15% H2-He, with a catalyst amount of 0.05 g and at a contact time 0.012 g s cm^-3. The activities were in the order 30% La > 24% Zr > 8% La. Because the surface areas of these samples are all similar (Table 1), we conclude that the effect is chemical and not scalable with the surface area.

Figure 6 shows the stabilities of 10% Cu-Ce(30% La)Ox and 10% Cu-Ce(24% Zr)Ox with time on stream at 450 °C. The contact time was 0.0303 g s cm^-3. Higher conversions of CO were measured over the La-doped sample throughout the 40-h-long test period. Both catalysts lost some activity rapidly at the beginning, but their activities became stable after 20 h on stream. No carbon deposition was found on either of the two used samples, as checked by temperature-programmed oxidation (TPO). The BET surface areas of both catalysts decreased by about 20% after use for either 32 or 20 h (Table 1), and the increase in ceria crystal size of both samples (Table 1) was also around 20%. The drop of CO conversion over both catalysts (after 32 or 20 h) was more than 20%. Over both catalysts, the CO conversion after 2 h was around 1.6 times the conversion after 20 h. Therefore, the loss of surface area (together with the increase in ceria crystal size) was not the only reason for the observed activity loss. XPS analysis of the 10% Cu-Ce(30% La)Ox sample (Table 3) showed that, after reaction under similar conditions (10% CO-30% H2O-10% CO2-15% H2, 450 °C, 34 h), Cu enrichment of the surface from 13 to 21 atom % took place, whereas the surface concentration of La did not change. It is possible that copper enrichment of the ceria surface contributed to the activity loss.

Conclusions  

In this work, we have demonstrated that copper-ceria is an excellent HTS catalyst, not suffering from the severe instability of iron chrome catalysts in CO2-rich gases. Ceria participates in the WGS reaction; its surface properties are crucial for high activity. The kinetics of the WGS reaction over 10% Cu-Ce (30%La)-Ox were measured over the temperature range 300-450℃. A strong dependence on CO and a weak dependence on H2O were found at 450℃, and inhibition by the reaction products was weak. No carbon deposition took place on the catalyst under these conditions. Catalyst lost some activity in the initial time  on steam but were stabilized thereafter. Thus, cerium oxide stabilizers copper in active form under WGS conditions at temperatures as high as 450℃.