Structural Changes of the Gold−Support Interface during CO Oxidation

interface changes as CO reacts with the gold to give gold carbonyls, which react with ... loaded into a cell (In-situ Research Institute, Inc., South...
0 downloads 0 Views 191KB Size
Langmuir 2005, 21, 5693-5695

5693

Structural Changes of the Gold-Support Interface during CO Oxidation Catalyzed by Mononuclear Gold Complexes Bonded to Zeolite NaY: Evidence from Time-Resolved X-ray Absorption Spectroscopy Juan C. Fierro-Gonzalez and Bruce C. Gates* Department of Chemical Engineering and Materials Science, University of California, Davis, California 95616 Received March 10, 2005. In Final Form: April 11, 2005 Mononuclear gold complexes synthesized from AuIII(CH3)2(acac) in zeolite NaY were characterized by time-resolved X-ray absorption spectroscopy and infrared spectroscopy as they catalyzed CO oxidation at 298 K and 760 Torr in flow systems. Initial contact with a CO + O2 mixture led to the rapid formation of cationic gold complexes in which Au was bonded to approximately two zeolite O atoms, on average. Further contact with CO + O2 led to breaking of an Au-surface oxygen bond, giving a gold carbonyl anchored to approximately one O atom. The process was reversed in the absence of CO and O2.

Introduction

Experimental Section

The discovery of unanticipated properties of gold as a catalyst has motivated a surge in research on mononuclear (cationic) gold complexes and supported gold nanoparticles as catalysts for homogeneous1,2 and heterogeneous3,4 reactions. Both gold nanoclusters5,6 and gold cations7,8 have been implicated in CO oxidation catalyzed by highly dispersed gold on supports. The identities of the catalytically active species are complicated by the possibility that they may contain both zerovalent and cationic gold, the latter possibly present at the gold-support interface. To investigate this interface and how it changes during CO oxidation catalysis, we prepared samples from an organogold complex, Au(CH3)2(C5H7O2) (dimethyl acetylacetonate Au(III)), on a well-defined support, zeolite NaY. The supported mononuclear complex catalyzes CO oxidation at room temperature in the absence of detectable zerovalent gold.9 We now report time-resolved X-ray absorption spectra and infrared (IR) spectra of the functioning catalysts, showing how the gold-support interface changes as CO reacts with the gold to give gold carbonyls, which react with O2 to give CO2. The results demonstrate the breaking and forming of chemical bonds between Au and O atoms of the support caused by changes in the reactive atmosphere. They show that the structure of the catalytically active gold can be tuned by changes in the reactant composition.

Sample syntheses and transfers were carried out with exclusion of air and moisture. The sample, containing 1 wt % Au, was prepared by bringing Au(CH3)2(C5H7O2) (Strem, 98%) in contact with zeolite NaY (W. R. Grace and Co.), which had been calcined in flowing O2 and then dried under vacuum at 573 K, as described elsewhere.9 Time-resolved extended X-ray absorption fine structure (EXAFS) spectra were recorded in transmission mode at beamline MR-CAT of the Advanced Photon Source at the Argonne National Laboratory as the initially prepared sample was treated in a mixture of CO and O2 in a flow reactor/cell10 at 298 K and 760 Torr. The total feed flow rate to the reactor was 150 mL (NTP) min-1 with CO and O2 partial pressures of 7.0 and 14.0 Torr, respectively, and the remainder He. The sample mass was 0.3 g. An X-ray absorption spectrum was recorded every 2 min, as before.9 A cryogenic double-crystal Si(111) monochromator was used. The spectra were collected in transmission mode, with the X-ray beam passing through N2-filled ionization chambers before and after passing the reactor/cell. In a separate experiment, the initially prepared sample was treated in a mixture of flowing CO and O2 in a standard once-through, nearly isothermal tubular packed-bed flow reactor under the same conditions, as described above. The sample (0.30 g, mixed in a mass ratio of 1:50 with particles of inert, nonporous R-Al2O3) was loaded into the reactor and transferred to a flow system without contacting air. The conversions of CO to CO2 as a function of time on stream (TOS) were determined by gas chromatographic analysis of the product stream; conversions were determined on the basis of both CO and O2 consumption and CO2 formation, with an accuracy of about (5%. IR spectra were recorded as the sample was treated in a mixture of flowing CO and O2 under the same conditions. The sample was pressed into a self-supporting wafer and loaded into a cell (In-situ Research Institute, Inc., South Bend, IN) in an Ar-filled glovebox. The cell was connected to a vacuum/adsorption system, which allowed recording of spectra while He, CO, and O2 flowed through and around the wafer. Spectra were recorded with a Bruker IFS 66v

* To whom correspondence should be addressed. E-mail: bcgates@ ucdavis.edu. (1) Chen, H.; Oldmstead, M. M.; Smith, D. P.; Maestre, M. F.; Fish, R. H. Angew. Chem., Int. Ed. Engl. 1995, 34, 1514. (2) Lai, S.-W.; Chan, M. C. W.; Peng, S.-M.; Che, C.-M. Angew. Chem., Int. Ed. 1999, 38, 669. (3) Sanchez, R. M. T.; Ueda, A.; Tanaka, K.; Haruta, M. J. Catal. 1997, 168, 125. (4) Stangland, E. E.; Stavens, K, B.; Andres, R. P.; Delgass, W. N. J. Catal. 2000, 191, 332. (5) Cunningham, D. A. H.; Vogel, W.; Kageyama, H.; Tsubota, S.; Haruta, M. J. Catal. 1998, 177, 1. (6) Valden, M.; Lai, X.; Goodman, D. W. Science 1998, 281, 1647. (7) Guzman, J.; Gates, B. C. J. Am. Chem. Soc. 2004, 126, 2672. (8) Nkosi, B.; Adams, M. D.; Coville, N. J.; Hutchings, G. J. J. Catal. 1991, 128, 378. (9) Fierro-Gonzalez, J. C.; Gates, B. C. J. Phys. Chem. B 2004, 108, 16999.

(10) Ozdak, J. F.; Argo, A. M.; Lai, F. S.; Gates, B. C.; Pandya, K.; Feraria, L. Rev. Sci. Instrum. 2001, 72, 3943.

10.1021/la0506574 CCC: $30.25 © 2005 American Chemical Society Published on Web 05/25/2005

5694

Langmuir, Vol. 21, No. 13, 2005

Letters

Table 1. EXAFS Results Characterizing the Sample Synthesized by Bringing AuIII(CH3)2(C5H7O2) in Contact with the Calcined Zeolite NaYa sample treatment/time on stream (min) flowing He/20

shell

Au-Au Au-O Au-Al flowing CO + O2/2 Au-Au Au-O Au-Al flowing CO + O2/20 Au-Au Au-O Au-Al flowing He (after catalysis)/30 Au-Au Au-O Au-Al

N

R 103∆σ2 (Å) (Å2)

b 2.0 0.9 b 2.1 1.0 b 0.9 0.8 b 1.6 0.9

b 2.08 3.20 b 2.16 3.21 b 2.24 3.20 b 2.17 3.21

b 3.80 7.62 b 3.81 11.00 b 5.12 8.21 b 5.20 6.06

∆E0 (eV) b 9.30 -1.01 b 6.22 -6.14 b 11.35 -5.15 b 10.06 -5.45

a The sample was characterized in flowing He and during CO oxidation catalysis in a flow system at 298 K and 760 Torr. Notation: N, coordination number; R, distance between absorber and backscatterer atoms; ∆σ2, Debye-Waller factor; ∆E0, inner potential correction. b Undetectable; the Au-Au contribution was included in models for which fitting was carried out, and the results show Au-Au coordination numbers indistinguishable from zero. Expected errors: N, (10%; R, (0.02 Å; ∆σ2, (20%; ∆E0, (20%.

spectrometer with a spectral resolution of 4 cm-1. Each reported spectrum is the average of 128 scans. Results and Discussion EXAFS results characterizing the initially prepared zeolite-supported sample as it was treated in flowing He show no evidence of Au-Au contributions, consistent with the presence of mononuclear, site-isolated gold complexes (Table 1). The spectra also show that each Au atom on average was bonded to approximately two O atoms at a distance of 2.08 Å, as in Au(CH3)2(C5H7O2), consistent with physisorption of this precursor in the zeolite.9 When the sample was treated with a flowing mixture of CO and O2, catalytic formation of CO2 occurred immediately, with initial CO conversions of approximately 60%, with the total flow rate and gas composition as stated above. The CO conversion declined to approximately 40% within 20 min and then stabilized (Supporting Information). When IR spectra were recorded in a separate experiment as the initially prepared sample was treated under the same conditions, a band appeared at 2169 cm-1 within 2 min (prior to recording of the first spectrum). This band has been assigned to CO bonded to cationic gold,11 and its appearance indicates the formation of gold carbonyls. The intensity of the 2169-cm-1 band remained nearly constant after 20 min TOS (Supporting Information), suggesting that the gold was not reduced during catalysis. Bands at 1690 and 1735 cm-1, characteristic of formate ligands, were also observed in the spectra. It has been proposed12 that formate ligands on gold might be reaction intermediates in CO oxidation catalyzed by supported gold. Consistent with the IR spectra showing the lack of reduction of the gold, the time-resolved X-ray absorption near edge structure (XANES) spectra recorded with the sample during CO oxidation catalysis show no features characteristic of zerovalent gold (Supplementary Information). These results are bolstered by the time-resolved EXAFS spectra, showing no detectable Au-Au contributions at any time (Table 1), consistent with the lack of (11) Because no νCO bands were observed in the IR spectra of the bare zeolite treated in flowing CO under the same conditions as those used for the zeolite-supported gold samples, the observed νCO band was attributed to CO bonded to the gold.9 (12) Mohamed, M. M.; Ichikawa, M. J. Colloid Interface Sci. 2000, 232, 381.

Figure 1. Radial distribution functions determined from timeresolved EXAFS spectra recorded at 298 K and 760 Torr as the sample made from AuIII(CH3)2(C5H7O2) and zeolite NaY was treated in flowing He (no shading), followed by treatment in flowing CO + O2 (light gray), and then treated in flowing He after cessation of the CO and O2 flows (gray). Feed CO and O2 partial pressures during catalysis were 7.0 and 14.0 Torr, respectively, with the remainder being He. The total feed flow rate was 150 mL (NTP) min-1, and the catalyst mass was 0.30 g. The radial positions were not corrected for phase shifts.

Figure 2. Changes in the Au-O coordination numbers (circles) and distances (triangles) estimated from time-resolved EXAFS spectra recorded as the sample made from AuIII(CH3)2(C5H7O2) and zeolite NaY was treated in flowing He, followed by CO oxidation catalysis, followed by treatment in flowing He after the CO flow had been stopped. Conditions are given in Figure 1. The vertical dotted lines denote the time at which the step changes in gas composition occurred.

aggregation of gold. The data thus demonstrate that the gold remained as mononuclear (cationic) complexes during catalysis, consistent with our previous results for the same sample treated under similar conditions.9 Although the gold did not aggregate, the EXAFS data show that the gold-support interface changed. Within 2 min TOS (prior to recording of the first time-resolved EXAFS spectrum), the Au-O distance changed from its initial value of 2.08 to 2.16 Å (Figures 1 and 2), the former representing the Au-O distance in the physisorbed precursor and the latter being a typical M-O bonding distance in zeolite- and oxide-supported mononuclear group 8 metal complexes.13 During the change, the Au-O

Letters

Figure 3. Radial distribution functions of the time-resolved EXAFS spectra recorded at 298 K and 760 Torr as the sample made from AuIII(CH3)2(C5H7O2) and zeolite NaY was treated successively in flowing He (no shading, gray, darker gray) and with a catalytically reacting mixture of CO and O2 (light gray, dark gray).

coordination number remained essentially unchanged at approximately 2 (Figure 2). These results indicate the rapid formation of supported gold complexes bonded on average to approximately two O atoms of the zeolite. Within the next 10 min of continuing flow of CO + O2, the Au-O distance increased from approximately 2.16 to 2.24 Å (Figure 2), with a concomitant decrease in the Au-O coordination number from approximately 2 to 1. The latter value remained essentially unchanged as the flow of CO + O2 (and CO oxidation catalysis) continued. After the preceding experiments, the flow was switched to He and the appearance of CO2 in the product stream decreased; after 20 min, no more CO2 was observed, evidently having then been purged from the flow system. EXAFS spectra give no evidence of Au-Au contributions in the sample treated under these conditions in the EXAFS cell, consistent with the presence of mononuclear gold complexes (Table 1). However, the Au-O distance decreased within 10 min, from 2.24 to 2.17 Å (Table 1 and Figure 2), with a concomitant increase in the Au-O coordination number from approximately 1 to approximately 1.6 (Table 1 and Figure 2). These results indicate that as CO2 desorbed and was purged from the system, a new Au-surface oxygen bond formed, again giving mononuclear gold complexes bonded (on average) to approximately two O atoms of the support.14 (13) Guzman, J.; Gates, B. C. Dalton Trans. 2003, 17, 3303.

Langmuir, Vol. 21, No. 13, 2005 5695

The process was repeatable (Figure 3), the reversible interconversion of the two types of mononuclear gold complexes being dictated by the presence or absence of CO and O2 in the gas phase. In summary, the data show a reversible change in the coordination of gold to the support, as the Au-O coordination number cycled between approximately 1 and 2. In view of the errors in the EXAFS data (Table 1), we do not regard these numbers as exact and do not rule out the possible presence of mixtures of surface species with various Au-O coordination numbers. The change in coordination of the gold is also indicated by IR spectra recorded under the same conditions, showing the appearance of gold carbonyls and formates in the presence of flowing CO and O2, and their disappearance as the flow of these gases stopped (Supporting Information). We infer that the formation of bonds between the gold and CO to give gold carbonyls (as well as ligands derived from the carbonyls, such as formates) is part of the process by which an Au-surface oxygen bond is broken, to give a gold complex bonded to only a single O atom of the support.15 These results are the first evidence of changes in the gold-support interface during CO oxidation catalyzed by supported gold. They show that the coordination of gold to a support can be tuned by changes in the composition of the reacting mixture. Because at least some catalysts incorporating supported zerovalent gold clusters also contain cationic gold (perhaps at the gold-support interface16,17), this result may pertain to supported gold catalysts with some generality. The results also emphasize the importance of characterizing the metal-support interface in such catalysts in the functioning state. Acknowledgment. This research was supported by the National Science Foundation, Grant Number CTS0121619. We thank the staff at beamline MR-CAT at the Advanced Photon Source at Argonne National Laboratory; the work performed at MR-CAT was supported in part by funding from the Department of Energy, Grant Number DEFG0200ER45811. Supporting Information Available: IR and XANES spectra of the sample synthesized by bringing Au(CH3)2(C5H7O2) in contact with zeolite NaY as it was treated in flowing CO and O2. This material is available free of charge via the Internet at http://pubs.acs.org. LA0506574 (14) When only the flow of CO was stopped, with O2 remaining in the feed, EXAFS spectra again showed the formation mononuclear gold complexes bonded to approximately two O atoms of the support. (15) Although the presence of gold carbonyls and formates is indicated by the IR spectra, the EXAFS data were not sufficient to demonstrate their presence (it was not possible to fit any Au-C contributions in the EXAFS spectra). (16) Fu, L.; Wu, N. Q.; Yang, J. H.; Qu, F.; Johnson, D. L.; Kung, M. C.; Kung, H. H. J. Phys. Chem. B 2005, 109, 3704. (17) Haruta, M.; Date´, M. Appl. Catal., A 2001, 222, 427.