The Nature of Active Copper Species in Cu-HMS Catalyst for

Jun 2, 2009 - The Nature of Active Copper Species in Cu-HMS Catalyst for Hydrogenation of ... Cu content higher than 5 wt % resulted in the collapse o...
0 downloads 0 Views 2MB Size
J. Phys. Chem. C 2009, 113, 11003–11013

11003

The Nature of Active Copper Species in Cu-HMS Catalyst for Hydrogenation of Dimethyl Oxalate to Ethylene Glycol: New Insights on the Synergetic Effect between Cu0 and Cu+ Anyuan Yin, Xiuying Guo, Wei-Lin Dai,* and Kangnian Fan Department of Chemistry and Shanghai Key Laboratory of Molecular Catalysis and InnoVatiVe Materials, Fudan UniVersity, Shanghai 200433, People’s Republic of China

Downloaded via UNIV OF EDINBURGH on January 29, 2019 at 07:25:33 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ReceiVed: March 25, 2009; ReVised Manuscript ReceiVed: April 29, 2009

Copper-containing mesoporous HMS catalysts prepared via a one-pot synthesis method based on sol-gel chemistry have been systematically characterized focusing on the effect of copper loading. Structural characterization of a series of different copper loading samples was performed by means of N2 adsorption, X-ray diffraction, Fourier transform infrared spectroscopy, transmission electron microscopy, temperature programmed reduction, N2O titration, and X-ray photoelectron spectroscopy. It is concluded that the copper loading has a great influence on the pore structure of the catalyst. On the basis of the characterizations, the copper species on calcined CuO/HMS samples and reduced Cu/HMS samples were assigned. The synergetic effect between the Cu0 and Cu+ is considered to be responsible for the enhanced catalytic performance in the hydrogenation of dimethyl oxalate (DMO) to ethylene glycol (EG). The maximum ratio of Cu0/Cu+ obtained via tuning the copper loading can result in the highest DMO hydrogenation activity and EG selectivity. 1. Introduction Creation of nanocomposite materials by introducing catalytic active species into a matrix is an area of growing interest in catalysis. The functionality of nanocomposites has been attributed to their surface properties, including surface-to-bulk atomic ratio, polyhedral surface morphology, concentrations of surface defects (coordinatively unsaturated ions due to planes, edges, corners, anion/cation vacancies, and electron excess centers), surface acid-base strength, and shape selectivity.1 Immobilization of nanoscopic materials in high-surface area supports with improved catalytic activity and product selectivity is a task of great economic and environmental importance in the chemical industries.1,2 Since the discovery of the M41S family of silica-based mesoporous materials in 1992,3 increasing attention has been paid to the novel materials,4-6 due to their great applications as catalysts, absorbents and host materials based on their large internal surface area and narrow pore size distributions. Among those mesoporous materials, HMS silicas, particularly those with sponge-like textural structure,7 have found promising applications as heterogeneous catalysts8 and supports for the immobilization of reagents.9 According to the results of a previous study,10 pure silica did not show any catalytic activities due to the lack of active sites necessary for catalysis. Hence, it is necessary to incorporate guests such as metals or metal oxides into the silicate framework to create active sites for catalytic interaction. Metals such as Al, Ti, Fe, W, Zr, Co, and Mn have been successfully incorporated into the mesoporous structures;11-17 however, there are few reports describing the use of HMS as a support for immobilizing copper species.18 Copper-based catalyst is widely used in many industrial reactions, and copper ions have been introduced to mesoporous silica by means of various methods, including wet-chemical processes, such as ion exchange (IE), aqueous impregnation (IM), deposition precipitation (DP), grafting, microemulsion techniques (MT), supercritical * To whom correspondence should be addressed. E-mail: wldai@ fudan.edu.cn. Fax: (+86-21) 55665572.

techniques, sono-chemical reduction, as well as various drychemical techniques, such as radiation methods, laser ablation, and chemical vapor deposition.19-24 It has also been reported that copper-containing mesoporous silica can be directly synthesized by mixing copper complexes with a suitable organofunctional silicon alkoxide.25,26 But it is difficult to introduce the divalent metallic ions into the framework of tetravalent silicon without destroying the mesoporous structures, and indeed, the amount of copper that could be retained in a mesoporous silica framework is usually small. Wang et al. reported that copper content exceeding 3 wt % would result in the collapse of the ordered mesoporous framework of MCM41.10 Karakassides et al. reported a synthesis approach of mesoporous Cu-SiO2 composite with Si(OCH3)4 and Cu(NO3)2 used as Si and Cu sources, respectively, and they found that a Cu content higher than 5 wt % resulted in the collapse of the ordered mesoporous structure.26 Recently, Derrien et al. reported copper-containing monodisperse mesoporous silica nanospheres by a smart one-step approach. However, this synthesis pathway does not allow adequate control of the material morphology under higher copper mass concentration.27 In our previous study,28 Cu-HMS catalysts prepared by the traditional incipient wet impregnation method exhibited good catalytic performance; however, the catalytic activity would decline under higher liquid hour space velocity (LHSV) (higher than 0.2 h-1) mainly due to the limited copper loading (5 wt %). So higher copper loading is necessary to further enhance the catalytic performance of the hydrogenation reaction under higher LHSV. In addition, our previous results29 have shown that the synergic effect of Cu0 and Cu+ might be proposed to be responsible for the promising hydrogenation activity, but the intrinsic nature of the active site and the effect of pore structure of the support still remain open to discussion. Herein we report a method of incorporating higher copper oxide species into the mesoporous HMS silica based on sol-gel chemistry via a one-pot synthesis (OPS) approach for the effective preparation of a novel Cu-HMS catalyst that is highly efficient in the hydrogenation of dimethyl oxalate (DMO) to

10.1021/jp902688b CCC: $40.75  2009 American Chemical Society Published on Web 06/02/2009

11004

J. Phys. Chem. C, Vol. 113, No. 25, 2009

ethylene glycol (EG). Special attention is paid to the effects of the copper mass concentration on their structural properties and catalytic behavior in the hydrogenation reaction. To better understand the role of the copper component, the catalytic behavior of different copper loading supported HMS catalysts has been investigated. In contrast with the previous studies, the 20 wt % Cu/HMS impregnated catalyst was also synthesized for comparison. To gain further insight into these effects, the relationship between the structure of the catalyst and the catalytic activity of different loadings of copper-supported HMS catalysts is evaluated in light of a systematic characterization of physicochemical properties of the catalysts by N2-physisorption, X-ray diffraction (XRD), transmission electron microscopy (TEM), Fourier transform infrared spectroscopy (FT-IR), H2-temperature programmed reduction (TPR), N2O titration, and X-ray photoelectron spectroscopy (XPS). The catalytic performance of the reduced Cu/HMS catalysts was evaluated by using the selective hydrogenation of DMO to EG as the probe reaction. 2. Experimental Section 2.1. Catalyst Preparation. Copper-containing HMS catalyst was prepared by the OPS method. A typical procedure is as follows: A 7.81 g sample of dodecylamine (DDA) was dissolved in 54 mL of ethanol and 82 mL of deionized water under vigorous stirring. After that, a certain amount of aqueous solution of Cu(NO3)2 · 3H2O (0.5 mol/L) was added. Then a stoichiometric amount of tetraethylorthosilicate (TEOS) dissolved in ethanol and isopropyl alcohol (IPA) was added dropwisely; the molar composition of this solution was 1.0:7:1 (SiO2:ethanol:IPA). The stoichiometric amount of Si is relative to the amount of Cu. The mixed solution was then stirred at 323 K for 4 h and the resulting gel was aged for 24 h at 313 K. The as-received gel was collected by centrifugation, washed three times with deionized water and once with absolute ethanol, then dried in air overnight at 373 K, followed by calcining in air for 4 h at 723 K. The final calcined samples were designated as wCu-HMS, where w denotes copper loading. For comparison, the reference Cu-HMS catalyst with 20 wt % copper loading denoted as 20Cu-HMS-im was prepared by the standard incipient wet impregnation of HMS with Cu(NO3)2 aqueous solution. Mesoporous siliceous HMS was prepared according to a well-established procedure delineated by Tanev and Pinnavaia30 using TEOS as silica source and DDA as template agent. Typically, the pure HMS material was prepared by dissolving 5.04 g of DDA in 53.33 g of deionized H2O and 39.42 g of ethanol under vigorous stirring before the addition of 21.39 g of TEOS dropwise. The solution mixture was then stirred at 313 K for 0.5 h. The resulting gel was aged for 18 h at ambient temperature to afford the crystalline templated product. After that, the resulting solid was recovered by filtration, washed with deionized water, and dried at 373 K, followed by calcination at 923 K in air for 3 h to remove the residual organic template materials, yielding the final mesoporous HMS material. 2.2. Characterization of Samples. Specific surface areas of the samples are measured by nitrogen adsorption at 77 K (Micromeritics Tristar ASAP 3000), using the Brunauer-EmmettTeller (BET) method. The copper loadings are determined by the inductively coupled plasma method (ICP, thermo E.IRIS). The small-angle XRD patterns below 2θ ) 6° were carried out on a Rigaku Multiflex instrument operated at 1.5 kW, using Cu KR radiation (1.5406 Å) at 40 kV and 40 mA. The wideangle XRD patterns were collected on a Bruker D8 Advance X-ray diffractometer, using nickel-filtered Cu KR radiation (λ ) 0.15406 nm) with a scanning angle (2θ) of 20-80°, a

Yin et al. scanning speed of 2 deg · min-1, and a voltage and current of 40 kV and 40 mA, respectively. The full width at half-maximum (FWHM) of CuO (111) reflection was measured for calculating crystallite sizes using the Scherrer equation. TEM micrographs are obtained on a JEOL JEM 2010 transmission electron microscope. FT-IR characterization of the catalysts was performed with a Bruker Vector 22 spectrometer equipped with a DTGS detector. The samples were finely grounded, dispersed in KBr, and pelletized. TPR profiles were obtained on a Tianjin XQ TP5080 autoadsorption apparatus. A 20 mg sample of the calcinated catalyst was outgassed at 473 K under Ar flow for 2 h. After cooling to room temperature under Ar flow, the inline gas was switched to 5% H2/Ar, and the sample was heated to 773 K at a ramping rate of 10 deg · min-1. The H2 consumption was monitored by a TCD detector. The metallic Cu surface area was measured by decomposition of N2O at 363 K, using a pulsed method with N2 as the carrier gas.31 The consumption of N2O was detected also by a TCD detector. The specific area of metallic copper was calculated from the total amount of N2O consumption with 1.46 × 1019 copper atoms per m2. XPS spectra are recorded under ultrahigh vacuum (99%) in methanol and H2 were fed into the reactor at a H2/ DMO molar ratio of 100 and a system pressure of 2.5 MPa. The reaction temperature was first set at 473 K and the room temperature LHSV of DMO was set at the range from 0.15 to 0.65 h-1. The products were condensed and then analyzed on a gas chromatograph (Finnigan Trace GC Ultra) fitted with an HP-5 capillary column and a flame ionization detector (FID). 3. Results and Discussion 3.1. Structural Evolution and Textural Properties of the Catalysts. The physicochemical properties of the calcined catalysts and the bare support (HMS) are summarized in Table 1. As can be seen, after incorporating copper species, a moderately reduced BET surface area was identified for all Cucontaining samples. This trend continued with the further increase in copper loading on the surface. A clear effect in both pore volume and pore diameter can be observed upon incorporating the copper species onto the silica substrate. The phenomenon that declines in BET surface area (from 1154 to 290 m2 g-1) and decreases in mesopore volume (from 0.86 to 0.38 cm3 g-1) may be attributed to a certain aggregation of copper species and collapse or reorganization of the silica structure caused by higher copper loading during the preparation process. However, the average pore diameter increases with the increasing of copper loading, which indicated the expansion of some mesopores or some mesopores collapsed and reorganized to accumulated pores. The changes in the structural properties of the Cu-HMS material for the increase of the copper content suggest the incorporation of Cu ions into the walls of mesoporous silicate framework. Interestingly, with an increase in copper content from 0 to 30 wt % in the current synthesis, the

The Synergetic Effect between Cu0 and Cu+

J. Phys. Chem. C, Vol. 113, No. 25, 2009 11005

TABLE 1: Physicochemical Properties and Catalytic Properties of Synthesized Cu-HMS Catalysts Prepared by the OPS Method Compared with the Blank HMS and 20Cu-HMS-im Prepared by the Impregnation Method sample

Cu loadinga (wt %)

SBET (m2 g-1)

SCub (m2 g-1)

Vp (cm3 g-1)

dp (nm)

HMS 5Cu-HMS 10Cu-HMS 15Cu-HMS 20Cu-HMS 25Cu-HMS 30Cu-HMS 20Cu-HMS-im

0 4.5 8.6 14.4 17.5 18.2 19.5 19.4

1154 933 820 799 675 305 290 556

1.6 4.3 9.2 11.6 10.2 5.1 2.9

0.86 0.80 0.88 0.77 0.62 0.43 0.38 0.36

2.5 2.7 3.7 3.0 4.8 5.9 8.1 2.2

dCuOc (nm)

TOFDMOd (h-1)

7.7 7.5 10.4 11.3 22.3

7.3 5.3 4.5 3.8 2.6 1.9 1.4

a Determined by ICP-AES analysis. b Cu metal surface area determined by the N2O titration method. c Calculated from the XRD data based on the Scherrer equation. d The TOF value was calculated as moles of ester reacted in the initial 1 h per mole of surface copper calculated from the copper dispersion.

Figure 1. N2 adsorption-desorption isotherms of (A) the calcined catalysts of different copper loadings and (B) BJH pore size distribution of the calcined catalysts of different copper loadings.

surface areas of the Cu-HMS materials decreased from 1154 to 290 m2 g-1, basically where a linear relationship with the nominal copper content can be addressed. In other words, by adopting our synthesis method, one can adjust porous properties of these materials by regulating the copper content. However, the real Cu loading of these catalysts as determined by ICPAES technique showed that the maximum copper loading prepared by this OPS method could be up to 18.5 wt %. The texture structure would be changed or even destroyed if the support incorporated higher content of copper. As a result, the regular structure of the catalyst could only be obtained under lower copper loading. It is found that, at higher nominal Cu loading, the resulting gel was deep blue, indicating the incomplete precipitation of the Cu species. Although the samples from 20Cu-HMS and 20Cu-HMS-im did not show much difference in the same BET surface area, the copper metal surface areas determined by N2O titration method are totally different, indicating that the preparation method has a great influence on the dispersion of the copper species. The diversity in the copper loading level may be related to the different pore structure properties of the support originating from the various adsorption-desorption properties and different pore diameter distribution properties (as shown below). Figure 1 shows the N2 adsorption-desorption isotherms and the pore size distribution curves of the catalysts and the bare support. All the samples exhibit Langmuir type IV isotherms with a H4-type hysteresis loop,32 corresponding to a typical mesoporous material with size-homogeneous 1D slit channels. Capillary condensation of nitrogen with uniform mesopores occurred, causing an abrupt steep increase in nitrogen uptake in the characteristic relative pressure (P/P0) range of 0.3-0.5

for the samples with the copper loading lower than 20 wt %, suggesting typical mesoporous structure with uniform pore diameters. The shift to higher relative pressures indicates an increase in framework pore size with increasing copper loading. Additionally, each sample displayed a significant N2 adsorptiondesorption hysteresis at high relative pressure of P/P0 > 0.9 and a sign of a high degree of textural porosity.7 This kind of loop is associated with porous materials that consist of agglomerate or compact packing of approximately regular and uniform spheres, which therefore have relatively narrow pore size distributions. However, with continuous increasing of copper loading (larger than 20 wt %), the sudden steep increase in nitrogen uptake in the range of 0.3-0.5 disappeared, and the hysteresis loop became flat, which could be attributed to the collapse of some pore structure. Because HMS exhibited a uniform hexagonal array of mesopores connected by smaller micropores, the broad hysteresis loop in the isotherms for the copper-doped samples is an indication of long mesopores, limiting the emptying and filling of the accessible volume.33 The pore size distribution curves derived from the desorption branch (Figure 1b) show that, at higher copper loading, the contribution of pores at ca. 2.8 nm to the total pore volume increased considerably at the expense of pores at ca. 3.9 nm, which is consistent with the decrease of the average pore volume from 0.86 to ca. 0.38 cm3 g-1. However, the pore size of the catalyst prepared by the impregnation method was almost the same as that of HMS, which may originate from the much severe aggregation and the nonuniform dispersion of the copper species. The powder XRD patterns obtained for the Cu/HMS catalyst precursors and copper-impregnated HMS samples after calcina-

11006

J. Phys. Chem. C, Vol. 113, No. 25, 2009

Yin et al.

Figure 2. XRD patterns of catalysts of different copper loadings: (A) after calcinations at low angle range; (B) after calcinations at high angle range; and (C) after reaction at high angle range.

tion at 723 K are presented in Figure 2. A single low-angle intense diffraction (100) peak at about 2θ ) 2° characteristic of a wormhole mesoporous framework is obviously observed for samples with copper loading less than 15 wt %, indicating the presence of the hexagonal regularity of porous structure of the sample and the long-range order of the HMS framework (Figure 2a). However, with further increase in the copper loading amount, the (100) peak disappeared, indicating the deorganization at long range of the mesoporous structure. Although the collapse of mesoporous channels could be one reason, we think that the incorporation of copper species into the mesopores would also cause such irregularity at long range. This finding could be further evidenced by the facts that the diffraction peak shifted to lower angle, the intensity of (100) peak decreased, and the width of (100) peak became broader with an increase of copper content in these samples. The XRD patterns of high diffraction angles of all the samples after calcination are shown in Figure 2b. No diffraction peaks of any crystalline phase of copper species were observed when the Cu loading was lower than 10%; however, with the continuous increasing in Cu loading, the intensity of the CuO diffraction peak at 2θ of 35.5° and 38.9° (JCPDS 05-0661) pertained to a typical structure of tenorite, a monoclinic structure of CuO with the crystal planes of (111) and (200) increased, demonstrating that small crystallites of CuO aggregated on the HMS surface. The diffraction peaks of CuO for OPS-derived catalyst are much broader and less intense than those of the IM-derived one at the same Cu loading, suggesting that CuO/SiO2 prepared by the OPS method shows much smaller crystalline size (7.5 vs. 22.3 nm) and higher

copper dispersion than that of the IM-derived samples. The diffraction pattern of SiO2 is only seen at around 2θ ) 21.7° with a broad and diffuse diffraction peak, which is attributed to amorphous silica for both catalysts. According to the Scherrer equation, the CuO crystallite sizes of the 20Cu-HMS and 20CuHMS-im are 7.5 and 22.3 nm, which illustrated that smaller particles could be obtained by using the OPS method. No diffraction peaks could be observed at ca.31.2 and 35.8°, implying the absence of copper phyllosilicate formed by IE and DP methods, and further evidence will be given below. After reduction (Figure 2c), the peaks from copper oxides disappeared along with the appearance of the peaks from metallic copper. In addition, a diffraction peak at 2θ ) 36.8° attributed to Cu2O is also observed in the reduced catalyst. These findings infer that the metallic copper and Cu2O coexisted in the working catalyst. Because of the weak interaction between the Cu0 crystallites and the silica support, and the mobility of metallic copper during the reduction process, copper species in the IM catalyst aggregated to larger particles (27.0 nm).34 In contrast, via the OPS method, the particle size of copper species did not grow obviously after reduction. The sintering of copper for IM-derived catalyst during the reduction process implied its bad thermal stability if compared with those Cu particles in the OPS-derived one. Figure 3a-f compares TEM images of the representative CuHMS samples after reaction and after reduction in H2/Ar flow at 573 K. The corresponding dispersion of copper species with the increase in copper loading can be directly observed. The TEM image of the 15Cu-HMS sample after reduction (Figure

The Synergetic Effect between Cu0 and Cu+

J. Phys. Chem. C, Vol. 113, No. 25, 2009 11007

Figure 3. TEM images for different copper loading catalysts after reduction and reaction: (a) 15Cu-HMS-re; (b) 20Cu-HMS-re; (c) 25Cu-HMS25-re; (d) 30Cu-HMS-re; (e) 20Cu-HMS-po; and (f) 30Cu-HMS-30-po.

3a) shows well-dispersed copper particles with uniform morphology, typically a wormhole mesoporous structure, due to the self-assembly of the TEOS in the synthesizing process. Also the image for the 15Cu-HMS sample did not show any evidence of bulk copper species condensed on the HMS surface, which further confirmed that the copper species were thoroughly dispersed in the HMS mesoporous walls after the synthesis process and the following high-temperature thermal treatment. In the TEM image of the 20Cu-HMS sample after reduction (Figure 3b), light gray silica particles are identified along with the dark ones assigned to lower valence copper species. The latter reduced in amount with 20 wt % copper loading, suggesting the improved dispersion of copper species or the formation of phases other than CuO. With the continuous increase of the copper loading, obvious aggregation of bulk copper crystallites in the framework of HMS and the deterioration of the mesoporous structure could be observed, in good accordance with the corresponding XRD patterns. After reaction, as can be seen from Figure 3e,f, the dispersion of the 20CuHMS sample is better than that of 30Cu-HMS. Unlike the 20CuHMS sample, the image of the impregnated one with the same copper loading (not shown here) showed bulk copper crystallites on the surface of silica, resulting in the aggregation of copper particles and the lower BET surface area. The big difference of the active copper surface area between 20Cu-HMS and 20CuHMS-im may originate from the differences in copper particle size and dispersion. The FT-IR spectra of siliceous HMS and CuO-HMS samples with different copper loadings are shown in Figure 4a. The absorption bands at approximately 1090, 800, and 470 cm-1 are assigned to the different vibration modes of the Si-O bonds in the amorphous SiO2.7 The broad absorption band in the range of 3600-3200 cm-1 is due to the overlapping of the OH stretching of adsorbed water and silanols.35 The widening of the band is related to the degree of hydrogen bonding with

neighboring OH groups. Similarly, the band near 1640 cm-1 corresponds to the bending mode of OH groups of adsorbed water.36 Because of their structural OH groups, IR spectroscopy is commonly adopted to discriminate the species from copper hydroxide, copper nitrate hydroxide, and copper hydrosilicate.37 Therefore, for silica-supported copper materials, the discrimination between copper hydroxide and copper hydrosilicate can be made from the frequencies of the δOH bands (for copper hydroxide at around 938 and 694 cm-1, for copper hydrosilicate at ca. 670 cm-1).37 As can be seen from Figure 4, no bending absorption of the Cu-O-H bond of the sample can be observed at 690 cm-1, suggesting the absence of Cu(OH)2 after calcination. Because of the presence of a broad band at 470 cm-1 from the support, the vibrations of the CuO bond that appear at 575, 500, and 460 cm-1 cannot be observed.38 The shoulder at around 600 cm-1 is observed, which could be supposed as an indication of the Cu(II)-O species.39 In addition, new band absorption at around 960 cm-1 appeared, suggesting the bond formation of Cu-O-Si33 during the catalyst preparation. The ratio of band intensities at 970 and 807 cm-1 could be a criterion of this incorporation for the reason that the intensity of the band at about 970 cm-1 increased whereas that of the band at 807 cm-1 did not change. It is worthwhile to note that the I970/I807 ratio only gives a qualitative estimation of the amount of incorporated copper, because the extinction coefficients of the corresponding IR bands are not known. Figure 4b clearly shows that the relative amount of incorporated copper in the form of Cu-O-Si in calcined Cu-HMS samples presents a volcano shape with the increase of copper loading and maximizes at 20 wt %. The calculated ratio of the two band intensities for Cu-HMS showed that this value is higher than that for pure HMS. This observation indicates that copper is probably included in the structure of Cu-HMS. All the data of XRD and IR changed in a similar way with the changes in Cu/Si ratio, suggesting that Vegard’s rule was obeyed.40 Such a result further indicated strong

11008

J. Phys. Chem. C, Vol. 113, No. 25, 2009

Yin et al.

Figure 4. (A) FT-IR spectra of catalysts with different copper loadings. (B) The I957/I806 intensity ratio representing the relative amount of incorporated copper species.

Figure 5. TPR profiles of catalysts with different copper loadings.

interaction between heteroatoms and silicon could be present in the framework of HMS. Thus, the spectra shown in Figure

4 clearly show that the copper species in the calcined samples are mainly present in the form of CuO and Cu-O-Si, but not copper phyllosilicate, which is known to form in the CuO/SiO2 catalysts prepared by IE41 and DP37 methods. 3.2. TPR of the Catalysts. TPR measurements were carried out to investigate the reducibility of the copper species in various CuO/HMS catalyst precursors. Figure 5 presents the reduction profiles of these samples. For the 30Cu-HMS sample, besides the main reduction peak at 543 K, there was a shoulder peak at ca. 560 K, which vanished at lower Cu loading. Compared with the IM-derived sample and combined with the XRD and TEM results, this shoulder peak can be assigned to the reduction of large CuO particles to metallic Cu. On the other hand, the main reduction peak at lower temperature is assigned to the reduction of highly dispersed Cu species with small particle size. It is worthwhile to note that the 20Cu-HMS catalyst displays the lowest reduction peak at 532 K, 10 K lower than that for the 30Cu-HMS catalyst, suggesting the presence of a copper-support interaction in the present samples, which facilitates the reduction of copper oxides. However, this does not mean the 20Cu-HMS

The Synergetic Effect between Cu0 and Cu+

J. Phys. Chem. C, Vol. 113, No. 25, 2009 11009

Figure 6. Cu2p photoelectron spectra of catalysts with different copper loadings: (A) the catalysts after calcinations and (B) the catalysts after reduction.

has the lowest reduction temperature as displayed on the TPR profile due to the not obvious shift of the reduction peak. Both bulk and dispersed CuO are reduced by a process involving autocatalytic reduction, and it is well documented that supported CuO can be reduced more readily than bulk CuO. In this situation, it is interesting to note that among the Cu-HMS series catalyst (Figure 5), the 20Cu-HMS sample shows the lowest reduction temperature, indicating that the copper loading is an important parameter in optimizing the redox properties of the catalysts. On the other hand, this finding suggests the presence of small CuO or Cu particles and the high dispersion of active species. These results correlate well with the metallic copper surface area measurements, which revealed the highest dispersion in the 20Cu-HMS sample. Moreover, good correlation with the XRD results is also observed, showing the presence of small sized CuO and Cu particles in this material. In addition, the reduction profile is narrow and almost symmetrical, indicating a narrow particle size distribution and homogeneous material. In this case, it should be pointed out that the TPR findings cannot be explained solely in terms of copper dispersion.42 Indeed, closer comparison of the TPR results with the XRD and N2O chemisorption measurements reveals that, in addition to dispersion, other factors, including CuO crystallinity and the interaction between the metallic copper and the oxide support, may affect the reducibility of the catalytic materials. The observed shift in reduction temperature may be ascribed to the different copper particle sizes, different interactions between copper oxide and silica, and different copper oxide dispersions. Chen et al.33 reported that the lack of reduction peaks at temperatures lower than 473 K indicated the absence of oxocations (Cu-O-Cu)2+, while the lack of peaks at temperatures higher than 533 K indicated the absence of a copper crystal phase. According to what has already been observed on Cu/SiO2 samples,43 the unique peak observed in Figure 5 can be attributed to the reduction of well-dispersed Cu2+ species. The formation of large CuO crystallite with weak or no metal-support interaction in the calcined IM-derived catalyst as illustrated above is most probably due to a poor interaction between the Cu2+ and the surface of the silica support under higher acidity. However, under high pH conditions (>7) during the OPS procedure, both the silica surface and microparticles of the copper oxide precursor are of high reactivity, so they may further react to form a small amount of copper silicate.

Additionally, copper silicate can also be formed during the calcination procedure. It should be noted that there is no detection of such a copper silicate phase in the XRD patterns due to its amorphous phase or very small particle size. 3.3. Surface Chemical States of the Catalysts. The XPS spectra of the calcined and H2-reduced OPS samples as well as the X-ray induced Auger spectra (XAES) of the reduced Cu/ HMS catalysts are illustrated in Figure 6. The intense and broad photoelectron peak at above 933.4 eV (Cu2p3/2) along with the presence of the characteristic shakeup satellite peaks suggests that the copper oxidation state is +2 in all the calcined samples. However, the XPS peak shapes of these samples are different. Considering the asymmetry of the Cu2p3/2 envelope, the peak of the calcined OPS samples can be deconvoluted into two contributions centered at around 936.0 and 934.1 eV, implying the existence of two Cu2+ species with different chemical circumstances.44 CuO species was determined at BE ) 934.1 eV, thus the relatively big positive BE shift of the Cu2p core level for the calcined catalysts is indicative of a charge transfer from the metal ions toward the support matrix, that is, a strong interaction between the metal ions and the matrix.45 Therefore, the XPS results inferred the presence of well-dispersed Cu2+ ions interacting with the silica support (936.0 eV) in the calcined sample, which is in good agreement with the TPR results. The low-temperature reduction peak could be attributed to the highly dispersed copper species and the high-temperature reduction peak could be assigned to the segregated copper particles. The two characterizations are in good agreement with each other. In the case of the reduced samples, the BE of Cu2p3/2 core levels reduced to 932.3-932.6 eV and the satellite lines disappeared, indicating that copper became reduced (Cu+ or Cu0). Because the BE values of Cu+ and Cu0 are almost identical, the distinction between these two species present on the catalyst surface is feasible only through the examination of XAES spectra. Two overlapping Cu LMM Auger kinetic energy peaks centered at about 919.6 and 916.1 eV are observed in the reduced OPS catalyst (Figure 7), whereas only one peak at 918.3 eV is observed in the reduced IM-derived catalyst. These results indicate that both Cu0 and Cu+ coexist on the surface of the reduced OPS catalyst, while Cu0 is mainly present on the surface of the reduced IM catalyst. Nevertheless, we believe that inadequately reduced Cu2+ species exist on the surface of the reduced OPS catalysts, because this kind of dispersed Cu2+

11010

J. Phys. Chem. C, Vol. 113, No. 25, 2009

Yin et al.

Figure 7. Cu LMM XAES spectra of the reduced catalysts with different copper loadings.

TABLE 2: Surface Cu Component of the Reduced Catalysts Based on Cu LMM Deconvolution a

KE (eV) +

0

b

AP (eV) +

0

catalyst

Cu

Cu

Cu

Cu

5Cu-HMS 10Cu-HMS 15Cu-HMS 20Cu-HMS 25Cu-HMS 30Cu-HMS 20Cu-HMS-im

916.2 916.2 916.5 916.1 916.2 916.3

919.3 918.9 919.7 919.1 919.6 919.4 918.3

1848.9 1848.9 1848.9 1848.5 1849.2 1849.1

1852.0 1851.6 1852.1 1851.5 1852.5 1852.2 1851.0

Cu 2p3/2 XCu0/Cu+c BE (eV) (mol/mol) RCu/Sid RCu/Sie 932.7 932.7 932.4 932.4 932.9 932.8 932.5

0.87 0.88 0.98 1.26 1.06 1.03 0

0.01 0.02 0.02 0.06 0.16 0.19 0.01

0.04 0.04 0.09 0.15 0.21 0.23 0.04

a Kinetic energy. b Auger parameter. c XCu0/Cu+ ) Cu0/ (Cu+) × 100%. d Mole ratio of Cu to Si of the catalysts before reduction determined by XPS. e Mole ratio of Cu to Si of the catalysts after reduction determined by XPS.

ions has strong metal-support interaction. The Cu2p BE of the 20Cu-HMS sample is much higher than that of the 20Cu-HMSim sample, in which the dominant Cu species can be considered as CuO. The much higher BE of Cu2p in the 20Cu-HMS sample reveals that the Cu ions are surrounded by silicon, most likely in the structure of Cu-O-Si; the higher electron affinity of silicon causes movement of electrons from copper to silicon and then the increase in BE of Cu2p. The modified Auger parameter R, which represents the summation of the kinetic energy (KE) of the Cu LMM Auger electron and the BE of the Cu 2p3/2 photoelectron, was employed to distinguish the Cu0 and Cu+ species. Deconvolution of the original CuLMM peaks was thus carried out and the peak positions as well as their contributions extracted from the deconvolution are listed in Table 2. As shown in Table 2, the Auger parameter of copper varied from 1848.9 to 1849.8 eV, suggesting the presence of Cu+ species along with the metallic copper.44 The R value at ca. 1851.0 eV is ascribed to Cu0 and ca. 1847.0 eV to Cu+. The smaller R value for Cu+ than the bulk one is attributed to the strong interaction between Cu+ and HMS. It is reported that when copper is in a highly dispersed state and in intimate contact with the supports, R can be 2-3 eV lower than the bulk values.46 As listed in Table 2, the Cu0/Cu+ intensity ratio derived by fitting the Cu LMM peak increased with the increase of the Cu loading, and maximized at 20 wt % with the Cu0/Cu+ ratio of 1.26 by the OPS method. The existence of Cu+ for the reduced catalysts signifies the stronger interaction between copper and silica. XPS intensity ratios of the metal cations in the supported metal oxide to those in the oxide support can provide important

Figure 8. Effect of copper loading on DMO conversion and EG, MG, ethanol, and 1, 2-BDO selectivities in DMO hydrogenation. Reaction conditions: P ) 2.5 MPa, T ) 473 K, H2/DMO ) 50 (mol/mol), and DMO LHSV ) 0.45 h-1.

information regarding the dispersion and crystallite size of supported particles.47 The Cu/Si ratio on the calcined OPS sample obtained by XPS analysis is 0.06, which is 6 times higher than that of the calcined IM sample (0.01) with essentially the same bulk composition, suggesting that the OPS method can greatly improve the dispersion of Cu2+ species on SiO2 surface and thus generate more active sites on the support surface. After reduction, a certain amount of copper species transferred to the surface and resulted in the increased ratio of Cu/Si. For the samples from 5 to 30 wt %, the surface contents of copper are much lower than the bulk one. This finding may imply that certain amounts of Cu species are buried in the framework of the channel walls. Further XPS analysis performed on 20CuHMS with the Ar+ etching technique at different time shows that the Cu/Si ratio changed from 0.15 (0 min) to 0.16 (15 min) and 0.18 (30 min). This result supports the speculation that the copper species was buried by silica in the pore walls. In conclusion, it may be proposed that the mesostructured CuSiO2 composite precipitated in a sandwich-like form, that is, the copper species is in fact surrounded by silicon species. 3.4. Catalytic Activities. DMO hydrogenation reaction was carried out to study the catalytic properties of both OPS- and IM-derived catalysts, and their activity and selectivity as a function of Cu loading were shown in Figure 8. Because of a much higher dispersion of copper species with significantly small particle size mentioned above, OPS catalysts presented a much higher DMO hydrogenation activity and EG selectivity than that of the IM-derived one. The 20Cu-HMS catalyst showed a high activity (100% conversion) for DMO hydrogenation with 96% of EG selectivity. Nevertheless, only 41% of DMO conversion with lower selectivity (68%) toward EG was obtained over IM catalyst at the same Cu loading. The OPS catalyst also showed much better catalytic performance than the previously reported Cu/SiO2.12,29 These results clearly demonstrated the high efficiency of the OPS catalyst in the selective hydrogenation of DMO to EG. The reason for the enhanced catalytic performance as compared to that for the IM catalyst might originate from the dramatic decrease of the particle size and proper ratio of Cu0/Cu+ for the OPS catalyst. The number of active sites with higher activity such as the corners of crystallites increases as the particle size decreases,48 thus giving a good explanation for the enhancement in catalytic activity for the OPS catalyst. The special pore structure of HMS provides good dispersion of copper species on the surface of the catalyst. Figure 9 shows

The Synergetic Effect between Cu0 and Cu+

Figure 9. Copper metal surface area measured by N2O titration and copper dispersion in copper-containing catalysts as a function of copper loading.

the copper mass concentration as a function of copper surface area and the copper dispersion. As can be seen, when the copper mass concentration reaches 20 wt %, the highest copper surface area (11.6 m2 g-1) and the largest dispersion of copper (61%) can be obtained, which may contribute to its excellent catalytic performance. 4. Discussions 4.1. Copper Species on Calcined Cu/HMS Samples. The copper species on the Cu-SiO2 catalysts will change with different preparation methods. Two types of supported Cu2+ species prepared by the selective adsorption method as grafted Cu2+ ions and copper phyllosilicate were reported by Toupance et al.49 The copper species prepared by the IE method are presented in two other forms: one is the immobilized Cu ions on the SiO2 surface by exchange with two silanol groups, the other is a well-dispersed CuO layer over the ion-changed Cu-O-Si layer, which originates from the calcinations of Cu(OH)2. This crystal phase was suggested to appear by the hydrolyzation of Cu(NH3)42+ trapped in the filter cake during the washing process.33 These two methods resemble to some extent the OPS method utilized in the present work. For these two methods, SiO2 is immersed with Cu2+ at room temperature until adsorption and ion-exchange reach a dynamic equilibrium. The difference is that SiO2 was in situ synthesized by hydroxylation of TEOS. Thus, three kinds of copper species are proposed to form via the OPS method based on the detailed characterizations mentioned above. One is the Cu ion incorporated into the HMS framework identified by FT-IR, the second is the grafted Cu ions resulting from electrostatic adsorption of Cu2+ in solution onto the silica, and the third is a copper silicate other than copper phyllosilicate formed by the self-assembly of copper and silica species during the sol-gel procedure, which could be confirmed by the FT-IR result. Thus we suggest that on calcined Cu-HMS samples, besides the presence of a small amount of large CuO particles identified by XRD and H2-TPR, there are ion-exchanged Cu-O-Si layer and a well-dispersed CuO layer. Toupance et al. revealed a partial decomposition of copper phyllosilicate after being calcined at 723 K.49 Therefore, there should be another kind of well-dispersed CuO due to partial decomposition of copper phyllosilicate during calcinations at 723 K for 4 h. The formation of large CuO particles is attributed to the aggregation of some loosely bonded and welldispersed CuO during calcinations. Due to the strong interaction between the copper species and silicate under high pH value (>7), the Cu-O-Si bond could

J. Phys. Chem. C, Vol. 113, No. 25, 2009 11011

Figure 10. Yield of EG plotted against copper metal surface area for catalysts with different copper loadings.

be formed in the framework of the catalyst, which could be observed to form the band adsorption around 970 cm-1 in the FT-IR spectrum. The framework copper silicate species could not only improve the dispersion of copper species, but also immobilize the active copper species during the hydrogenation of DMO. Thus, the metallic copper reduced from the highly dispersed copper oxide species and the Cu+ reduced from the copper silicate coparticipates in the hydrogenation of DMO. 4.2. Active Sites on Cu-HMS Catalysts for DMO Hydrogenation. Our previous study speculated that a synergism between Cu0 and Cu+ oxidation states might exist during hydrogenation of DMO to EG,29 which is consistent with the behavior reported for crotonaldehyde hydrogenation on copperbased catalyst50 and IPA dehydrogenation on Cu powder.51 Dandekar et al. have shown that the turnover frequency in the IPA dehydrogenation reaction, based on the surface Cu atoms, is dependent upon the ratio of Cu0/Cu+ in surface sites, with a value near unity found to be optimal.51 Fridman and Davydov have reported that the oxidation state of Cu can affect not only the dehydrogenation activity but also the selectivity, and in their study of cyclohexanol dehydrogenation over Cu-Mg and Cu-Zn-Al catalysts, they found that Cu+ was significantly more active than Cu0 for cyclohexanone formation, while Cu0 was active for the aromatization reaction of cyclohexanol to phenol.52 With regard to the essential nature of active copper species for the hydrogenation of the DMO reaction, the metallic copper surface area was assumed to be the main factor in the structure-activity correlation for Cu-based catalysts. Although both the larger BET surface area and special pore structure are helpful to improve the dispersion of copper species and the interaction between the copper species and the support, which has been observed in the investigation of TPR, TEM, and FTIR, the intrinsic structure-activity relationship still remains unclear. To gain further insight into the nature of the copper species in the present as-synthesized Cu-HMS catalysts in relation to the hydrogenation activity, the effect of copper loading on the EG yield as well as the specific area as determined by surface titration with N2O is illustrated in Figures 9 and 10. The variation in copper surface area shows a perfect correlation with the corresponding EG yield and a positive correlation can also be illustrated between the EG yield and the copper dispersion. This finding indicates that the Cu surface area contributes mainly to the catalytic performance of Cu-HMS in the hydrogenation of DMO to EG, displaying a rational explanation for the copper loading effect on active catalyst. The lack of straightforward

11012

J. Phys. Chem. C, Vol. 113, No. 25, 2009

Yin et al. performance of Cu-HMS in the hydrogenation of DMO to EG. The proper ratio of Cu0/Cu+ can greatly improve the catalytic performance of the catalyst. On the 20Cu-HMS catalyst, an EG yield of 98% was obtained under optimized hydrogenation conditions. The TOF value of the 20Cu-HMS catalyst is 2.7 times larger than that of the 20Cu-HMS-im catalyst. Acknowledgment. This work is financially supported by the Major State Basic Resource Development Program (Grant No. 2003CB615807), NSFC (Project 20573024), and the Natural Science Foundation of Shanghai Science & Technology Committee (06JC14004). References and Notes

+

Figure 11. Yield of EG plotted and Cu /(Cu + Cu ) against copper loadings. 0

0

correlation between the copper particle size and the activity in the present study may be understood by keeping in mind that besides the Cu surface area, other factors, such as the chemical circumstance of copper species, metal-support interaction, and special pore structural properties, all affect the catalytic hydrogenation performance over the OPS-derived Cu-HMS catalysts. It has been suggested that only Cu0 acts as the active sites in ester hydrogenation.53 But in methyl acetate hydrogenation, Poels and Brands54 reported while Cu0 dissociatively adsorbs H2, Cu+ stabilizes the methoxy and the acyl species, which are also important intermediates in DMO hydrogenation. The broad and asymmetric peak shape of the Cu LMM XAES spectra indicated that Cu existed in more than one chemical state in reduced Cu/HMS catalysts. Deconvolution treatment of Figure 7 revealed that the surface copper species contained a certain proportion of Cu+ after reduction. Therefore, Cu+ may play an indispensably important role in the DMO hydrogenation reaction. Moreover, Cu+ may function as electrophilic or Lewis acid sites to polarize the CdO bond via the electron lone pair on oxygen, thus improving the reactivity of the ester group in DMO. It is notable that the mole ratio of Cu0/Cu+ is found to increase steadily at lower copper loading, maximized on the 20Cu-HMS catalyst, and dropped with continuous increase of the copper loading (shown in Figure 11). So we tentatively propose that the optimal catalytic activity on the 20Cu-HMS catalyst lies in the synergetic effect of Cu0 and Cu+. Although by using other methods, such as the selective adsorption method and the ammonia evaporation method, one can also obtain Cu0 and Cu+ species, the loading of Cu is much lower due to the intrinsic limitation of the method itself. The OPS method supplies an effective way to prepare Cu/SiO2 catalyst with higher copper loading as well as special pore structure, which may be important for catalytic reactions, such as selective hydrogenation of DMO exemplified here. 5. Conclusions In summary, copper-containing HMS catalysts with higher copper loading (17.5 wt %) and high dispersion of active sites are successfully synthesized by the OPS method. Three kinds of copper species were proposed as follows: aggregated CuO clusters, an ion-exchanged Cu-O-Si layer, and well-dispersed CuO species. The active copper species have been assigned to the one with strong synergetic effect between Cu0 and Cu+. The Cu surface area contributes mainly to the catalytic

(1) Bordoloi, A.; Halligudi, S. B. J. Catal. 2008, 257, 283. (2) Cole-Hamilton, D. J. Science 2003, 299, 1702. (3) Kresge, C. T.; Leonowicz, M. E.; Roth, W. J.; Vartuli, J. C.; Beck, J. S. Nature 1992, 359, 710. (4) Huo, Q.; Margolese, D. I.; Ciesla, U.; Feng, P.; Gier, T. E.; Sieger, P.; Leon, R.; Petroff, P. M.; Schuth, F.; Stucky, G. D. Nature 1994, 368, 317. (5) Corma, A. Chem. ReV. 1997, 97, 2373. (6) Jackie, Y.; Ying, C. P. M. M. S. W. Angew. Chem., Int. Ed. 1999, 38, 56. (7) Pauly, T. R.; Liu, Y.; Pinnavaia, T. J.; Billinge, S. J. L.; Rieker, T. P. J. Am. Chem. Soc. 1999, 121, 8835. (8) Sudhakar Reddy, J.; Liu, P.; Sayari, A. Appl. Catal. A 1996, 148, 7. (9) Clark, J. H.; Macquarrie, D. J. Chem. Commun 1998, 853. (10) Wang, B. W.; Zhang, X.; Xu, Q.; Xu, G. H. Chin. J. Catal. 2008, 29, 275. (11) Liu, H.; Lu, G.; Guo, Y.; Guo, Y.; Wang, J. Microporous Mesoporous Mater. 2008, 108, 56. (12) Matteoli, U.; Bianchi, M.; Menchi, G.; Frediani, P.; Piacenti, F. J. Mol. Catal. 1985, 29, 269. (13) Matteoli, U.; Menchi, G.; Bianchi, M.; Piacenti, F. J. Mol. Catal. 1988, 44, 347. (14) Matteoli, U.; Menchi, G.; Bianchi, M.; Piacenti, F. J. Mol. Catal. 1991, 64, 257. (15) Teunissen, H. T.; Elsevier, C. J. Chem. Commun. 1997, 667. (16) van Engelen, M. C.; Teunissen, H. T.; de Vries, J. G.; Elsevier, C. J. J. Mol. Catal. A: Chem. 2003, 206, 185. (17) Yin, D.; Li, W.; Yang, W.; Xiang, H.; Sun, Y.; Zhong, B.; Peng, S. Microporous Mesoporous Mater. 2001, 47, 15. (18) Zhang, P.; Zhang, Z.; Wang, S.; Ma, X. Catal. Commun. 2007, 8, 21. (19) Xu, J.; Yu, J. S.; Lee, S. J.; Kim, B. Y.; Kevan, L. J. Phys. Chem. B 2000, 104, 1307. (20) Yeh, M. S.; Yang, Y. S.; Lee, Y. P.; Lee, H. F.; Yeh, Y. H.; Yeh, C. S. J. Phys. Chem. B 1999, 103, 6851. (21) Henglein, A. J. Phys. Chem. B 2000, 104, 1206. (22) Chusuei, C. C.; Brookshier, M. A.; Goodman, D. W. Langmuir 1999, 15, 2806. (23) Ohde, H.; Hunt, F.; Wai, C. M. Chem. Mater. 2001, 13, 4130. (24) Dong, X.; Potter, D.; Erkey, C. Ind. Eng. Chem. Res. 2002, 41, 4489. (25) Michael, A.; Karakassides, K. G. F. A. T. D. P. AdV. Mater. 1998, 10, 483. (26) Karakassides, M. A.; Bourlinos, A.; Petridis, D.; Coche-Guerente, L.; Labbe, P. J. Mater. Chem. 2000, 10, 403. (27) Derrien, G.; Charnay, C.; Zajac, J.; Jones, D. J.; Roziere, J. Chem. Commun. (Cambridge, U.K.) 2008, 3118. (28) Yin, A.; Guo, X.; Dai, W.-L.; Li, H.; Fan, K. Appl. Catal. A 2008, 349, 91. (29) Chen, L.-F.; Guo, P.-J.; Qiao, M.-H.; Yan, S.-R.; Li, H.-X.; Shen, W.; Xu, H.-L.; Fan, K.-N. J. Catal. 2008, 257, 172. (30) Tanev, P. T.; Pinnavaia, T. J. Science 1995, 267, 865. (31) Evans, J. W.; Wainwright, M. S.; Bridgewater, A. J.; Young, D. J. Appl. Catal. 1983, 7, 75. (32) Schmidt, R.; Hansen, E. W.; Stoecker, M.; Akporiaye, D.; Ellestad, O. H. J. Am. Chem. Soc. 1995, 117, 4049. (33) Chen, L.; Horiuchi, T.; Osaki, T.; Mori, T. Appl. Catal. B 1999, 23, 259. (34) Marchi, A. J.; Fierro, J. L. G.; Santamara, J.; Monzon, A. Appl. Catal. A 1996, 142, 375. (35) Chaminand, J.; Djakovitch, L.; Gallezot, P.; Marion, P.; Pinel, C.; Rosier, C. Green Chem. 2004, 6, 359. (36) Henrist, C.; Traina, K.; Hubert, C.; Toussaint, G.; Rulmont, A.; Cloots, R. J. Cryst. Growth. 2003, 254, 176.

The Synergetic Effect between Cu0 and Cu+ (37) Toupance, T.; Kermarec, M.; Lambert, J.-F.; Louis, C. Conditions of Formation of Copper Phyllosilicates in Silica-Supported Copper Catalysts Prepared by Selective Adsorption. J. Phys. Chem. B 2002, 106, 2277. (38) Diaz, G.; Perez-Hernandez, R.; Gomez-Cortes, A.; Benaissa, M.; Mariscal, R.; Fierro, J. L. G. J. Catal. 1999, 187, 1. (39) Cordoba, G.; Arroyo, R.; Fierro, J. L. G.; Viniegra, M. J. Solid State Chem. 1996, 123, 93. (40) Tichit, D.; Bennani, M. N.; Figueras, F.; Ruiz, J. R. Langmuir 1998, 14, 2086. (41) Vandergrift, C. J. G.; Elberse, P. A.; Mulder, A.; Geus, J. W. Appl. Catal. 1990, 59, 275. (42) Sun, Q.; Zhang, Y.-L.; Chen, H.-Y.; Deng, J.-F.; Wu, D.; Chen, S.-Y. J. Catal. 1997, 167, 92. (43) Carniti, P.; Gervasini, A.; Modica, V. H.; Ravasio, N. Appl. Catal. B 2000, 28, 175. (44) Dai, W.-L.; Sun, Q.; Deng, J.-F.; Wu, D.; Sun, Y.-H. Appl. Surf. Sci. 2001, 177, 172. (45) Jernigan, G. G.; Somorjai, G. A. J. Catal. 1994, 147, 567.

J. Phys. Chem. C, Vol. 113, No. 25, 2009 11013 (46) Kohler, M. A.; Curry-Hyde, H. E.; Hughes, A. E.; Sexton, B. A.; Cant, N. W. J. Catal. 1987, 108, 323. (47) Braun, S.; Appel, L. G.; Camorim, V. L.; Schmal, M. J. Phys. Chem. B 2000, 104, 6584. (48) Ryu, B. H.; Lee, S. Y.; Lee, D. H.; Han, G. Y.; Lee, T.-J.; Yoon, K. J. Catal. Today 2007, 123, 303. (49) Toupance, T.; Kermarec, M.; Louis, C. J. Phys. Chem. B 2000, 104, 965. (50) Rioux, R. M.; Vannice, M. A. J. Catal. 2003, 216, 362. (51) Dandekar, A.; Baker, R. T. K.; Vannice, M. A. J. Catal. 1999, 184, 421. (52) Fridman, V. Z.; Davydov, A. A. J. Catal. 2000, 195, 20. (53) Mokhtar, M.; Ohlinger, C.; Schlander, J. H.; Turek, T. Chem. Eng. Technol. 2001, 24, 423. (54) Poels, E. K.; Brands, D. S. Appl. Catal. A 2000, 191, 83.

JP902688B