Shun
Nishimura
a,
Tetsuya
Shishido
*b,
Junya
Ohyama
b,
Kentaro
Teramura
bc,
Atsushi
Takagaki
d,
Tsunehiro
Tanaka
b and
Kohki
Ebitani
a
aSchool of Materials Science, Japan Advanced Institute of Science and Technology (JAIST), 1-1 Asahidai, Nomi, Ishikawa 923-1292, Japan
bDepartment of Molecular Engineering, Graduate School of Engineering, Kyoto University, Katsura, Nishikyo-ku, Kyoto 615-8510, Japan. E-mail: shishido@moleng.kyoto-u.ac.jp; Fax: +81-75-383-2561; Tel: +81-75-383-2559
cPrecursory Research for Embryonic Science and Technology (PRESTO), Japan Science and Technology Agency (JST), 4-1-8 Honcho, Kawaguchi, Saitama 332-0012, Japan
dDepartment of Chemical System Engineering, School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-7656, Japan
First published on 20th April 2012
To compare the catalytic performances against daily start-up and shut-down (DSS) operations between co-precipitated (CP) and impregnated (IMP) Cu–Al–Ox catalysts for the water gas shift (WGS) reaction, in situ X-ray adsorption fine structure (XAFS) measurements, temperature-programmed reduction (TPR) profiles, X-ray diffraction (XRD) patterns and high resolution transition electron microscopy with an energy dispersive X-ray spectroscopy (TEM-EDS) analysis were performed. In situ XAFS studies clearly indicated that the Cu species were frequently oxidized and reduced during DSS operations with steam treatment (DSS-like operation). Based on in situ XAFS and H2-TPR profiles, the highly active and stable CP-catalyst possessed more susceptible Cu particles to oxidation/reduction (described as redoxable) than the IMP-catalyst even after the DSS-like operations. Interestingly, the XRD and TEM-EDS analysis showed small Cu particles which were covered with a card-house structure of the in situ formed boehmite in the case of the CP-catalyst after the DSS-like operations. According to these results, we concluded that the superior durability of the CP-catalyst against frequent redox changes was attributed to the nanoscale coordination with the in situ formed boehmite structure which preserves the small-redoxable Cu particles.
The daily start-up and shut-down (DSS) operation is essential for the application of the PEMFC energy system to small facilities such as homes, buildings and vehicles. Since frequent DSS operation causes the catalyst to be exposed to many different environments (e.g. temperature movement, exposure to air and/or condensing steam), deactivation of the catalyst can occur quickly. Therefore, it is necessary to consider new strategies different from the traditionally highly-functionalized catalyst systems under steady-state conditions. To this end, investigation of dynamic behavior of the structure in developing catalysts under the DSS operations will give useful information for considering the new demands of DSS operations.
Many researchers have investigated the deactivation mechanism of ceria-supported noble metal catalysts for WGS reaction such as Pt/CeO2, Au/CeO2 and Pd/CeO2. The formation of carbonates and formates on the surface of not only CeO2 but also noble metal was observed after deactivation by DSS operation in these cases.2,3 Some researchers proposed that the irreversible over-reduction of CeO2 was also related to the deactivation mechanism in the presence of excess H2.4,5 According to these reports, it was considered that reoxidation treatment with air or oxygen is necessary for regeneration in the case of ceria-supported noble metal catalysts under the DSS operations. In this aspect, addition of O2 to reaction gas6 or alkaline metals to catalysts7 was proposed for improving durability of precious metal catalysts.
In general, Cu-based catalysts have higher activity than noble metal-based catalysts, however, the former have much lower durability than the latter against DSS operations.8 Cu-based catalysts are known to be very sensitive catalysts against frequent atmosphere changes, therefore they are easily deactivated under DSS operation. The Cu/ZnO type catalysts are the most popular Cu-based catalysts for WGS reaction, and addition of Al2O3, SiO2 and various alkaline metals,9–12 investigation of effective supports13 or optimization of preparation conditions14,15 has been attempted for improvement of their catalytic performance. Formation of a spinel phase was also reported to be an effective methodology for high catalytic activity of the Cu–Al, Cu–Fe and Cu–Mn types.16–19 However, further investigation of competitive Cu-based catalysts is still necessary for mid- and long-term use. The formation of carbonates,20,21 passivation by a dense shell of hydroxide22 and sintering of Cu23,24 were proposed as the reasons for deactivation of the Cu-based catalysts under DSS operation. Nevertheless, an effective strategy towards a highly durable Cu-based catalyst has not been proposed.
Recently, the formation of the highly active and durable Cu–Al–Ox catalyst prepared by the simple co-precipitation method has been studied under both laboratory and pilot scale DSS conditions over 500 cycles.24,25 In general, the Cu/Al2O3 including ZnO (Cu/ZnO/Al2O3) catalysts were considered as a standard for Cu-based catalysts for not only the WGS reaction but also the methanol synthesis and SRM reactions because the additive ZnO had a great effect on the activity and durability owing to its spacer and promoter actions, and so on.26–28 Therefore, the catalytic feature of the Cu–Al–Ox catalyst system is an attractive finding. Further elucidation of the novel Cu–Al–Ox catalyst during DSS operations gives valuable information of progressively robust Cu-based catalysts. Herein, we investigated the catalytic dynamic behavior of Cu–Al–Ox catalysts during DSS-like operations using an in situ X-ray adsorption fine structure (XAFS) method in combination with other analytical techniques, and revealed the novel roles of the stable Cu–Al–Ox catalyst and the deactivation mechanism under the DSS operation.
These catalysts were tested for WGS reaction in a fixed bed reactor in order to determine their activity and durability under the DSS operations with steam treatment (Fig. 1). The as-prepared catalyst (200 mg) was placed between glass-wool. Prior to WGS reaction, the catalyst was reduced with a H2/N2 (5/95) gas at 523 K for 0.5 h. As the steam treatment, repetitions of temperature between 473 and 323 K under steam flow for a predetermined number of times were performed (denoted DSS-like operations hereafter). The gas composition (after moving through an ice trap to remove water) was analyzed with an on-line gas chromatograph equipped with a TCD (Molecular Sieves-5A column) and a methanizer FID (Porapak-Q column) detector. Durability of the catalyst was determined to compare its activities before and after DSS-like operations. The gas space velocity per gram catalyst (SV) for WGS reaction was 12.4 L h−1 g−1-STP.
![]() | ||
Fig. 1 Diagram of DSS-like operation in the WGS reaction. The (a)–(e) indicate the sampling points for the structural change experiment shown in Fig. 9. |
X-ray diffraction (XRD) measurement was carried out with a Rigaku RINT 1000 instrument with CuKα radiation (1.54059 Å) in the range of 2θ = 5–80° with 0.60° min−1 scanning rate. The diffraction patterns were analyzed with the database in the JCPDS (Joint Committee of Powder Diffraction Standards). The surface area of the prepared catalyst was estimated by the N2 adsorption method at 77 K using a Brunauer–Emmett–Teller (BET) model in Coulter SA3100. The morphology and elemental distribution were observed using a high resolution transmission electron microscope (HR-TEM, JEOL JEM-2100F) and a scanning TEM with an energy dispersive X-ray spectroscopy (STEM-EDS) attachment. The samples for the TEM measurements were dispersed in water, then the supernatant liquid was dropped onto a carbon coated molybdenum grid and dried in vacuo overnight.
The Cu surface area of the catalyst was determined by an N2O decomposition method at 363 K as described in the previous reports.29,30 Prior to the measurement, all samples were screened to 26–42 mesh and reduced with a H2/N2 (5/95) gas at 523 K for 0.5 h. Temperature-programmed reduction (TPR) was performed with an Ohkura BP-2 instrument interfaced with a TCD. The TCD results were normalized to the mass of the used samples, and the rate of H2 consumption was estimated based on the standard curve of pure CuO (99.999%, Koujundo Chem. Labo. Co., Ltd.) reduction. Each sample was pre-oxidized under O2 gas flow at 573 K for 1 h. Then the catalyst was purged with He and cooled to ambient temperature. The TPR profile was recorded under a H2/Ar (5/95) flow at a ramp rate of 10 K min−1.
In situ time-resolved Cu K-edge XAFS measurements were carried out at the BL01B1 of SPring-8 with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (Proposal No. 2008A1170, 2009A1662 and 2009B1690). Double Si(111) single crystals were used for division of energy. In a typical sample preparation, 15 mg of sample was diluted with 16 mg of boron nitride (Wako Pure Chem. Ind. Co., Ltd.), grained and pressed to a pellet. Thereafter, the pellet was placed in the in situ flow cell (ASPF-20-03, Kyowashinku Co., Ltd.) designed by Suzuki and Nomura.31 The Cu K-edge XAFS spectra were measured for each 60 s in the range of 8733–10362 eV at intervals of 33 s for each measurement. The obtained spectra were analyzed with Rigaku REX2000 ver. 2.5.7. software. For deconvolution analysis of the spectra,32–35 Cu metal, Cu2O and CuO were used as references for Cu0, Cu+ and Cu2+, respectively. The k3-weighted EXAFS spectra (k3χ(k)) were obtained from normalized EXAFS spectra, and Fourier-transforms (FTs) of which were performed in the range of k = 3–13 Å−1. The coordination numbers (CN) were estimated by curve-fitting analysis with the empirical parameters extracted from Cu foil for the first Cu–Cu shells (2.55 Å); the inverse FTs were performed in the range of k = 4–12 Å−1 and R = 1.78–2.61 Å.
![]() | ||
Fig. 2 Effect of DSS-like operations on CO conversion over CP Cu–Al–Ox, IMP Cu–Al–Ox and Cu/ZnO/Al2O3 (MDC-7) catalysts. Reaction temperature 473 K, CO/CO2/H2O/H2 = 7.3/7.3/27.2/58.3, SV = 12.4 L h−1 g−1. |
To compare the differences between CP- and IMP-catalysts, direct monitoring of Cu species during DSS-like operations was conducted by in situ XAFS measurements. The changes in Cu K-edge (8.98 keV) XANES spectra during 2 cycles, 5 times each, of DSS-like operations are described in Fig. 3. Both CP- and IMP-catalysts showed the same tendency as follows: the two-humped XANES spectra corresponding to Cu metal (Fig. S1(A), ESI†) were observed during the first reaction (red lines). When steam was introduced during DSS-like operations, the peak intensity around 9000 eV decreased with increase in the peak intensity around 8990 eV (green and blue lines). These changes were related to the oxidation of Cu0 to Cu+ and/or Cu2+ species (Fig. S1(A), ESI†) during the DSS-like operations. Thereafter, the XANES spectra drastically turned to the two-humped spectrum upon switching from the steam atmosphere to the reaction. It suggested that the immediate reduction of Cu+ and/or Cu2+ species to Cu0 occurred in the initial stage of the second reaction. For further investigation, |FT|s of EXAFS spectra (k3χ(k)) during 2 cycles of DSS-like operations are shown in Fig. 4. The peaks at 1.47 Å and 2.24 Å in a normal |FT| correspond to Cu–O (1.86 Å) and Cu–Cu (2.54 Å) after curve fitting, respectively. The positions of these peaks were very close to the first Cu–O shell (1.85 Å) in Cu2O and the first Cu–Cu shell (2.55 Å) in Cu metal (Fig. S1(B), ESI†). From the results of |FT| changes, it is clear that the changes in XANES spectra during DSS-like operations (Fig. 3) are associated with the oxidation of Cu metal to Cu2O, and the remarkable reduction of Cu2O to Cu metal at the beginning of the second reaction. It is also supposed that the change in XANES spectra of the CP-catalyst was much larger than those of the IMP-catalyst, indicating that a large part of Cu metal in the CP-catalyst was oxidized to copper oxides. These structural alterations during DSS-like operations were also inferable from changes in the k3-weighted EXAFS spectra (Fig. S2 compared with Fig. S1(B), ESI†).
![]() | ||
Fig. 3 Changes in XANES spectra over (A) CP Cu–Al–Ox and (B) IMP Cu–Al–Ox during 2 cycles of DSS-like operations under the WGS reaction (red lines), 1st and 2nd DSS-like operation (green lines and blue lines). |
![]() | ||
Fig. 4 Changes in |FT| of EXAFS spectra (|FT| of k3χ(k)) over (A) CP Cu–Al–Ox and (B) IMP Cu–Al–Ox during 2 cycles of DSS-like operations under the WGS reaction (red lines), 1st and 2nd DSS-like operation (green lines and blue lines). |
To elucidate the changes in Cu species during DSS-like operations, the deconvolution of the obtained XANES spectra was performed using reference XANES spectra (Fig. S1(A), ESI†) and the estimated fraction of each Cu species was plotted as a function of time. Fig. 5 and 6 show the evaluation of the fraction of Cu species during 1 and 2 cycles of DSS-like operations, respectively. Over 90% of Cu species exist as Cu0 during the first reaction in all cases, then partial oxidation to Cu+ occurred under DSS-like operations. The oxidation process of Cu0 to Cu+ was accelerated by the ramping temperature followed by the drastic reduction of Cu+ to Cu0 at the beginning of the second reaction. The partial oxidation of Cu to Cu2O by water was also proposed by some research groups.38,39 Although both CP- and IMP-catalysts showed the same redox behavior of Cu species as shown in Fig. 5 and 6, the amount of oxidized Cu species in DSS-like operations was much different. The CP-catalyst showed 36% and 76% of Cu+ species formed in 1 and 2 cycles of DSS-like operations, but the IMP-catalyst showed 13% and 23% for each. While the temperature of the DSS-like operation is different as shown in Fig. 6 and 7, the changes scarcely affected the evaluations of the amount and the fraction of Cu species during a cycle of DSS-like operation.40 Changes in the coordination number (CN) of Cu–Cu of 2.54 Å (2.24 Å in a normal |FT|) (Cu metal) indicated that the CP-catalyst has smaller Cu species than the IMP-catalyst during the first reaction before DSS-like operation, but increase in the coordination number of Cu–Cu estimated by |FT|s, approximately from 10.5 to 12, was observed after repeating 2 cycles of the DSS-like operations in the case of the CP-catalyst (Fig. S3, ESI†). The gradual growth of the crystallites of Cu metal in the CP-catalyst after the DSS-like operations was also observed in the XRD patterns (e.g. the approximate crystalline sizes of Cu(111) were 13 nm (before), 14 nm (after 10 cycles), 16 nm (after 20 cycles), and 23 nm (after 50 cycles)) (Fig. S5(A), ESI†) although it exhibited high durability.
![]() | ||
Fig. 5 Fraction of Cu species over (A) CP Cu–Al–Ox and (B) IMP Cu–Al–Ox during 1 cycle of DSS-like operation, (open circles) Cu0, (closed squares) Cu+ and (open diamonds) Cu2+. |
![]() | ||
Fig. 6 Fraction of Cu species over (A) CP Cu–Al–Ox and (B) IMP Cu–Al–Ox during 2 cycles of DSS-like operations, (open circles) Cu0, (closed squares) Cu+ and (open diamonds) Cu2+. |
![]() | ||
Fig. 7 Fraction of Cu species over (A) prereacted CP Cu–Al–Ox and (B) prereacted IMP Cu–Al–Ox during 1 cycle of DSS-like operation, (open circles) Cu0, (closed squares) Cu+ and (open diamonds) Cu2+. |
We wondered why the CP-catalyst showed good durability even though the growth of Cu metal was suggested by the |FT| and XRD patterns after DSS-like operation. In order to clarify the novel factor of the CP-catalyst, the results of evaluation of the fractions of Cu species using CP- and IMP-catalysts prereacted over 20 cycles of DSS-like operations in the laboratory before XAFS measurement were also investigated (Fig. 7). The CP-catalyst prereacted with 20 cycles of DSS-like operations possessing a high activity (42% conv.) shows a similar behavior to the fresh CP-catalyst (Fig. 5(A)), and the amount of Cu+ species (39%) during 1 cycle of DSS-like operation also had a similar value as shown in Fig. 5(A) and 7(A). Contrastively, the IMP-catalyst after 20 cycles of DSS-like operations possessing a low activity (6% conv.) had a drastically decreased amount of Cu+ species compared to the fresh IMP-catalyst (from 13% to <1%) (Fig. 5(B) and 7(B)). These results show that the Cu species in the CP-catalyst kept their original properties in terms of the oxidation after DSS-like operations, however, the IMP-catalyst experienced a big change toward tolerant to oxidation. Comparisons of correlations between CO conversion and the easily-oxidized Cu concentration (Cu+/Cu0total) estimated from the XAFS spectra in a cycle of DSS-like operation (described in Fig. 5 and 7) were examined. The fresh CP-catalyst indicated Cu+/Cu0total = 3.76 × 10−1 with 30% CO conv. and the prereacted CP-catalyst indicated 4.19 × 10−1 with 42% conv. On the other hand, in the case of the IMP-catalyst, the fresh and the prereacted ones showed 1.38 × 10−1 with 25% conv. and 5.13 × 10−3 with 6% conv., respectively. These results clearly indicate that the concentration of easily-oxidized Cu species is related to a high catalytic activity for WGS reaction.
Moreover, the fraction of Cu2+ (CuO) was flat during DSS-like operations in all cases (Fig. 5–7). Thus, the overoxidation of Cu species scarcely occurred in the DSS-like operations. It is known that water could act as an oxidant in the oxidation from Cu metal to Cu2O (Eo(Cu+/Cu0) = +0.52 V) but could not in the oxidation from Cu2O to CuO (Eo(Cu2+/Cu+) = +0.16 V) because the latter is in the range of the stability field of water.41 Furthermore, Cu+ has a possibility of transmutation to Cu2+ and Cu0 by disproportionation, however, it seemed to be rare during DSS-like operations. Presumably, the stable Cu2+ species observed in XANES analysis are attributed to CuOs surrounded by AlxOy, which is difficult to access, and/or Cu-Al-Ox composite precursors.
In the XAFS results, the DSS operations caused the redox changes of Cu species, gradual oxidation and drastic reduction between Cu metal and Cu2O, and causing Cu species to grow. The CP-catalyst suppressed the effects of the redox changes even when it had Cu species susceptible to oxidation. Considering the results shown in Fig. 2, few changes were observed over 10 cycles of DSS-like operations in cases of both CP- and IMP-catalysts, which implied that the oxidation of Cu0 to Cu2O was saturated over 10 cycles in the Cu–Al–Ox catalysts in the laboratory reactor. Briefly, the DSS-like operation described in Fig. 1 seemed to change the degree of oxidation of Cu metal to Cu2O in the catalysts, which was related to the degree of successive reduction from Cu2O to Cu metal.
To further investigate the redox behavior of CP- and IMP-catalysts, H2-TPR profiles were measured as shown in Fig. 8. All samples were pretreated in O2 flow at 573 K for 1 h, and then each profile was recorded from 323 K with 10 K min−1 under H2/Ar flow. The H2-TPR profiles supported the observation of Cu oxides on the sample after each treatment. The calcined CP-catalyst exhibited two peaks around 550 K and 620 K (Fig. 8A(a)). After reduction treatment, a single peak around 530 K was observed (Fig. 8A(b)). After that, the peak gradually shifted to the lower temperature side with increasing the amount of easily reducible CuO by repeating the DSS-like operations (Fig. 8A(c)–(f)). These results suggested that the CuOs in the CP-catalyst were gradually dispersed on the catalyst by repeating DSS-like operations.42,43 On the other hand, the calcined IMP-catalyst had a single peak around 560 K (Fig. 8B(a)). After reduction and DSS-like operations, the reduction peak was placed around 520 K with a small shift to higher temperature indicating the slight increase in the crystallinity of CuO.44,45 According to these results, the easily reducible Cu species are related to the superior activity for the WGS reaction, and the superior nature of the CP-catalyst against DSS-like operations is not only related to the easy oxidation (supported by XAFS studies) but also easy reduction (indicated by H2-TPR profiles) of Cu species. The importance of the redox behavior of Cu active species has also been investigated in various reactions.42,46,47 As the previous reports on WGS reaction, two plausible mechanisms over the Cu-based catalyst have been proposed. The redox mechanism via changing between Cu0 and Cu+ on the surface of the Cu/ZnO/Al2O3 catalyst has been reported.9,11,15,36,48 On the other hand, the associative mechanism (intermediate formation such as formates, carbonates or carboxyls) was also discussed.49,50 A more understanding of the correlation between the redox properties of Cu particles and the catalysis for WGS reaction over Cu–Al–Ox catalysts is the subject of a future investigation.
![]() | ||
Fig. 8 H2-TPR profiles of (A) CP Cu–Al–Ox and (B) IMP Cu–Al–Ox before and after DSS-like operations. After (a) calcination, (b) reduction, (c) 1st, (d) 10th (e) 20th, (f) 50th cycles of DSS-like operations. Each number indicates the amount of H2 consumption (mmol g−1) at the peak. Oxidation at 573 K under O2 flow was carried out as a pretreatment for all samples. |
According to the previous reports, the in situ formation of AlOOH during DSS-like operations is strongly associated with the high durability of the CP-catalyst.24,25 However, the formation mechanism of the boehmite and the morphology between formed boehmite and Cu species still remain open questions. To study the boehmite formation process, the structural changes during 1 cycle of DSS-like operation are investigated by XRD analysis as shown in Fig. 9. The dialog and sampling points for this experiment are described in Fig. 1 denoted (a)–(e). In the case of the CP-catalyst, small and broad peaks corresponding to CuO crystallites in the calcined catalyst (Fig. 9A(a)) and Cu metal after reduction (Fig. 9A(b)) were obtained. Thereafter, the diffractions resulting from formation of boehmite appeared during and after DSS-like operations (Fig. 9A(c)–(e)). These results suggested that the boehmite was formed by exposure of amorphous alumina in steam. Some studies reported the preparation method of boehmite under steam rich conditions.51–54 Intensity of the XRD patterns of boehmite at (130) peak at 2θ = 38.5° and BET surface area were gradually increased by repeating DSS-like operations, and their tendencies were well fitted with that of CO conversion (Fig. S6(A) and (B), ESI†) suggesting that the formation of boehmite strongly affected the increases in the catalytic activity. On the other hand, in the case of the IMP-catalyst, sharp and strong peaks corresponding to large crystalline domains of CuO and Cu metal were observed after calcination and reduction, respectively (Fig. 9B(a) and (b)). Comparing the peaks after DSS-like operations, few changes on the large Cu metal and Al2O3 were observed (further details are shown in Fig. S5(B), ESI†). It was found from XRD data that the CP-catalyst had smaller Cu metal particles than the IMP-catalyst, and the progressive changes in catalytic structure occurred only in the CP-catalyst accompanied with in situ boehmite formation by repeating DSS-like operations.
![]() | ||
Fig. 9 XRD patterns of (A) CP Cu–Al–Ox and (B) IMP Cu–Al–Ox during 1 cycle of DSS-like operation. The dialog and sampling points (a)–(e) are described in Fig. 1. (closed squares) CuO, (open squares) Cu, (closed triangles) AlOOH (boehmite), and (open triangles) γ-Al2O3. |
HRTEM and STEM-EDX analyses were carried out for the CP-catalyst after DSS-like operations to study the morphology between boehmite and Cu particles. Fig. 10 shows the HRTEM and STEM images of the CP-catalyst after DSS-like operations with elemental analysis by EDS measurement. The presence of a needle-like structure owing to boehmite formation52,54,55 was supported by the HRTEM image after DSS-like operations (Fig. 10A). Using the STEM-EDS analysis of the catalyst, small Cu particles were surrounded by the needle-like structure in the CP-catalyst after DSS-like operations (Fig. 10B and C). The needle-like boehmite structures and small Cu particles were observed in many places (e.g. see Fig. S7, ESI†). According to these results, the needle-like boehmite was formed by repeating steam treatments under the DSS-like operations, and they stabilized the small redoxable Cu particles.
![]() | ||
Fig. 10 Morphology of CP-catalyst after DSS-like operations analyzed by (A) HR-TEM, (B) STEM and (C) STEM-EDS. The green spots in (C) correspond to the presence of Cu. |
Footnote |
† Electronic supplementary information (ESI) available: Changes in EXAFS spectra, CN numbers, Cu(111) crystalline sizes, BET surface areas, and STEM-EDS mappings. See DOI: 10.1039/c2cy20133j |
This journal is © The Royal Society of Chemistry 2012 |