22982
2006, 110, 22982-22986 Published on Web 10/31/2006
Discrimination of Active Palladium Sites in Catalytic Liquid-Phase Oxidation of Benzyl Alcohol Davide Ferri,† Cecilia Mondelli,‡ Frank Krumeich,† and Alfons Baiker*,† Department of Chemistry and Applied Biosciences, ETH Zurich, HCI, CH-8093 Zurich, Switzerland, and Department of Inorganic, Metallorganic and Analytical Chemistry, UniVersity of Milan, Via Venezian 21, I-20133 Milan, Italy ReceiVed: September 5, 2006; In Final Form: October 13, 2006
Knowledge of the structure of active sites is a prerequisite for the rational design of solid catalysts. Using site-selective blocking by CO and isotope labeling combined with in situ attenuated total reflection infrared (ATR-IR) spectroscopy, we were able to discriminate the different sites involved in the liquid-phase oxidation of benzyl alcohol on Pd/Al2O3. The main reaction, that is, the oxidative dehydrogenation of the alcohol to the corresponding aldehyde, showed only little dependence on structure and occurred on all exposed Pd faces, whereas the undesired product decarbonylation occurred preferentially on hollow sites on (111) Pd faces. This explains why specific blocking of the latter sites, as realized in the industrially used Pd-Bi/Al2O3 catalysts, leads to improved catalytic performance.
Introduction The discrimination of catalytically active sites remains a great challenge when the solid catalyst operates in a liquid environment because application of surface analytical methods is restricted to only few techniques capable of providing adequate information under reaction conditions.1-4 Although UHV studies on single crystals5,6 and model catalysts7 represent an extremely valuable approach for gaining fundamental insight into the functioning of such catalytic systems, the interpretation of the data requires extrapolation to normal pressure and in particular to the liquid phase, which is often connected with great uncertainty. Well-defined supported metal nanoparticles have been applied successfully to bridge the “material gap” between single crystals and polycrystalline materials to understand the structure of active sites.8 They allow mimicking catalytic particles9 and metal-support interfaces10 as found in real catalysts. The liquid-phase oxidation of alcohols is an example of a relevant industrial catalytic process where the use of single crystals combined with UHV techniques contributed greatly to unravel fundamental aspects of the reaction mechanism.11 This operates via an oxidative dehydrogenation taking place on metallic Pd.12 Recent in situ spectroscopic studies on commercial Pd/Al2O3 catalysts including attenuated total reflection infrared (ATR-IR) spectroscopy13,14 and X-ray absorption spectroscopy (XAS)15 provided unique information on the state of the metal surface during reaction together with the observation of the complex reaction mechanism in which metal particles are involved. The molecular structure of the alcohol allows for a number of side reactions, which run parallel to the main alcohol * Corresponding author. E-mail:
[email protected]. † Department of Chemistry and Applied Biosciences, ETH Zurich. ‡ Department of Inorganic, Metallorganic and Analytical Chemistry, University of Milan.
10.1021/jp065779z CCC: $33.50
dehydrogenation reaction, thus revealing that the catalyst surface is composed of different active sites enabling one (or more) specific reaction(s). Here, we report the discrimination of active sites exposed on a technical Pd/Al2O3 catalyst during the liquid-phase oxidative dehydrogenation of benzyl alcohol. The sites active for the desired dehydrogenation reaction as well as those catalyzing the undesired decarbonylation of the product are identified using a combined approach including selective site blocking by CO adsorption, isotope labeling, and in situ attenuated total reflection infrared (ATR-IR) spectroscopy to probe simultaneously the catalyst surface during reaction and the reaction progress. Experimental Section Attenuated total reflection infrared (ATR-IR) measurements of benzyl alcohol oxidation were performed as described elsewhere.14 Briefly, the reaction was monitored at 50 °C first during contact of the catalyst deposited on the ZnSe ATR-IR crystal with an Ar-saturated solution of the alcohol in cyclohexane (20 mM). Then Ar was replaced by air in the same solution without interrupting the liquid flow. Before reaction, the catalyst coating was contacted with H2-saturated solvent for about 20 min. This treatment affords reduced Pd.15 In case of pre-equilibration of the Pd/Al2O3 with CO, after catalyst reduction (H2-saturated solvent, 50 °C, 20 min) COsaturated cyclohexane was admitted to the ATR-IR cell for about 1 h. Then the flow was switched to the Ar-saturated solution of benzyl alcohol (or 13C-benzyl alcohol). Transmission electron microscopy (TEM) investigations were performed with a CM30ST microscope (Philips; LaB6 cathode, operated at 300 kV, point resolution ∼2 Å), while scanning transmission electron microscopy (STEM) investigations were © 2006 American Chemical Society
Letters
J. Phys. Chem. B, Vol. 110, No. 46, 2006 22983
Figure 2. High-resolution transmission electron microscopy (TEM, top left) and scanning transmission electron microscopy (STEM, bottom left) of the pristine Pd/Al2O3 catalyst. Top right is a representation of the idealized cuboctahedron shape of a Pd crystallite exposing (111) and (100) planes adapted from ref 18. Bottom right are spectra of CO adsorbed on Pd/Al2O3 (a) reduced in situ at 50 °C in flowing H2-saturated cyclohexane (ATR-IR) and (b) reduced in situ at 300 °C in flowing hydrogen (DRIFTS).
Figure 1. ATR-IR spectra of a solution of benzyl alcohol on reduced Pd/Al2O3 under inert (Ar) and oxidative (air) conditions. Conditions: Calcohol ) 20 mM; 4 mg catalyst; cyclohexane solvent; 50 °C; liquid flow rate ) 0.6 mL/min.
carried out using a Tecnai F30 microscope (FEI (Eindhoven); field emission cathode, operated at 300 kV). Further information on materials and methods including details of ATR-IR cell is given in the Supporting Information. Results and Discussion Figure 1 shows the ATR-IR spectra observed when flowing a solution of benzyl alcohol over reduced Pd/Al2O3 under inert (Ar) and oxidative (air) conditions. Besides the formation of benzaldehyde (ca. 4% yield, signal at 1713 cm-1), decomposition of the product to CO (1853 cm-1)16,17 and hydration of benzaldehyde to adsorbed benzoate species (signals below 1600 cm-1)14 occurred as indicated by the infrared signals observed when the catalyst surface was exposed to dehydrogenation conditions (alcohol and inert gas). The catalytic activity increased by a factor 3-4 (up to ca. 15% yield) with respect to the dehydrogenation when oxygen was added to the feed, also yielding a higher amount of product decarbonylation on Pd. The enhanced formation of benzoate species and of surface water corroborates the deactivation of the catalyst under oxidizing conditions. After a relatively long period of coexistence, CO is displaced from Pd by oxygen. Comparison of the signals of adsorbed CO obtained under dehydrogenation conditions (Figure 1) and after adsorption of CO (Figure 2, spectrum a) on the reduced catalyst indicates that only specific sites are occupied by CO during benzyl alcohol oxidation on Pd. Nevertheless, the dehydrogenation reaction occurs even in the case of site blocking by CO. The ATR-IR spectra of CO adsorbed on Pd/Al2O3 (Figure 2) provide valuable information on the morphology of metal
particles owing to the sensitivity of the C-O bond to the location of the CO molecule on the particle. Comparison with adsorption of CO on alumina-supported Pd catalysts,18 where the assignment is made on the base of studies on well-defined Pd nanoparticles, provides excellent reference for CO adsorbed on crystalline facets and defect sites.8 The signal at 1968 cm-1 corresponds to bridge CO on (100) planes and defect sites (edges, corners). The shoulders at lower energy indicate the population of hollow sites on (111) planes and bridge sites on (100) planes. The signal at 2074 cm-1 is associated with ontop CO on defect sites and is flanked by a signal at ca. 2050 cm-1, indicating occupation of edge sites. The assignment is supported strongly by high-resolution transmission electron microscopic (HR-TEM) images of the commercial Pd/Al2O3 catalyst (Figure 2). The image corresponding to the pristine material shows nearly round-shaped particles of ca. 3-4 nm with defined (111) and (100) crystalline planes, that can be approximated by the cuboctahedron shown in Figure 2. The particle boundaries are faded due to the strong Bragg contrast of the crystalline alumina support. STEM confirms that the Pd crystallites are coalesced to form worm-like aggregates. Reduction under bubbling hydrogen in cyclohexane did not change the appearance of the catalyst particles. The diffuse reflectance infrared (DRIFT) spectrum of CO adsorbed on the same Pd/Al2O3 reduced at 300 °C (Figure 2, spectrum b) appears different from the one obtained in the liquid phase, but correlates perfectly with the literature.18 Similar sites are occupied but the relative population of these sites is substantially different, occupation of (111) planes being favored in the gas-phase experiment. This difference can originate from different reasons. The interference of solvent molecules residing in proximity of the metal surface with CO dipoles can change the appearance of the spectrum of adsorbed CO compared to the gas-phase experiment, as demonstrated in the co-adsorption of CO and water.19 Additionally, the largely different reduction conditions (liquid vs gas phase and temperature) may also affect
22984 J. Phys. Chem. B, Vol. 110, No. 46, 2006 the environment to which adsorbing CO is exposed, thus providing differences in the shape of the CO signals. Interestingly, the spectrum of CO adsorbed from the gas phase (reduction at 300 °C) is more similar to that measured during reaction in the liquid phase than that of CO adsorbed from the liquid phase. This is tentatively attributed to the strong reducing potential of alcohols at low temperature,15 which might affect the actual state of the Pd surface and of the adsorbate layer. The assignment of the signals corresponding to adsorbed CO is important for the following discussion. Figure 1 shows that during the liquid-phase oxidation of benzyl alcohol on Pd/Al2O3, CO from decarbonylation of benzaldehyde peaks at 1853 cm-1 with a number of shoulders. Similar signals were found for CO in an equivalent experiment with benzaldehyde under identical conditions.17 According to the assignment made for adsorbed CO, the low frequency of the signal indicates that during dehydrogenation CO occupies predominantly hollow sites on (111) planes. This assignment agrees quite well with the preferential adsorption of CO on the (111) faces of well-defined Pd nanoparticles pre-equilibrated with methanol.20 The pretreatment with methanol ensures the formation of carbon deposits, which selectively occupy defect sites as inferred from the vanishing of the signal at 1960 cm-1 compared with CO adsorption on a pristine Pd surface. Similarly, it has been observed that Xe selectively perturbs the vibration of CO adsorbed on step sites but not that on terrace sites.21 Figure 3a shows the ATR-IR spectra of a solution of benzyl alcohol on the Pd/Al2O3 catalyst pre-equilibrated with CO at 50 °C for 1 h. Upon admission of the alcohol solution saturated with Ar, the signal at 2074 cm-1 is quickly attenuated and that at 1968 cm-1 rapidly shifts to ca. 1950 cm-1 and is enhanced. This is likely due to disruption of CO dipole-dipole interactions induced by benzyl alcohol. After few minutes on stream the intensity of the latter signal starts to decrease slowly and the signal shifts by ca. 30 cm-1, whereas its low-energy flank gains intensity. During this period no catalytic activity was recorded in the ATR-IR and in the online FT-IR spectra. The signal at 1713 cm-1 of dissolved benzaldehyde appears in the spectra only when the signal originally at 1968 cm-1 is suddenly consumed. At this point a broad CO signal extends between 1950 and 1700 cm-1, which has no equivalent signal in Figure 2a but resembles the CO feature of Figure 1. The timedependence of the intensity of the signal originally at 1968 cm-1 is reported in Figure 4a together with that of the signal at 1713 cm-1. Benzaldehyde formation under dehydrogenation conditions (absence of oxygen) clearly occurs when CO is completely removed from bridge sites (defects and (100) planes). CO adsorbed in the on-top geometry on corner sites (2074 cm-1) does not appear strictly related to the catalytic activity because these sites are liberated almost immediately. On the contrary, the on-top geometry on edge sites (ca. 2050 cm-1) follows a similar trend to the signal at 1968 cm-1. Figure 3 also shows that what remains on Pd after reaction ignition is CO adsorbed predominantly on (111) planes. This CO is strongly bound. However, it does not originate from benzaldehyde decarbonylation as Figure 1 would suggest. It is residual CO from the pre-equilibration of Pd/Al2O3 with CO. This is demonstrated in an equivalent experiment to that shown in Figure 3a but with benzyl alcohol labeled with 13C at the methylene group. This approach is based on the fact that dehydrogenation of this alcohol should afford benzaldehyde labeled at the carbonyl carbon atom and the decarbonylation of this product should provide 13CO. Figure 3c and d shows the
Letters
Figure 3. ATR-IR spectra of a solution of (a) benzyl alcohol and (b) 13 C-benzyl alcohol under dehydrogenation conditions (Ar) on reduced Pd/Al2O3 pre-equilibrated with CO (1 h, red spectra). Blue spectra represent the end of the experiments (ca. 2 h after admission of the alcohol solution). Dotted spectra are intermediate spectra taken at approximately similar times on stream. (c) Spectra of 13C-benzyl alcohol reaction in the presence of (d) Ar and (c) air, similar to the spectra shown in Figure 1. Red and blue spectra represent the beginning and the end of the Ar and air phases, respectively. Conditions: Calcohol ) 20 mM; 4 mg catalyst; cyclohexane solvent; 50 °C; liquid flow rate ) 0.6 mL/min.
ATR-IR spectra of 13C-benzyl alcohol oxidative dehydrogenation on Pd/Al2O3 under identical conditions to those shown in Figure 1. Multibonded 13CO appears at 1776 cm-1, whereas a very weak signal at 1947 cm-1 corresponds to on-top 13CO. 13Cbenzaldehyde is observed at 1675 cm-1 together with signals of 13C-labeled benzoate species at 1592 and 1515 cm-1. The red shift of the signals of evolving carbon-containing species is in agreement with the isotope effect on the vibrational energy. Figure 3b shows the equivalent experiment to that reported in Figure 3a but with 13C-benzyl alcohol. After pre-equilibration with CO for about 1 h and admission of the solution of the labeled alcohol, the CO signal behaves identically to the case of unlabeled benzyl alcohol. Importantly, no 13CO is detected when 13C-labeled benzaldehyde appears and the shape of the remaining CO signal is identical to that obtained with benzyl alcohol (Figure 3a) and to that shown in Figure 1. The selective site blocking achieved using CO (Figure 3) provides evidence that benzyl alcohol displaces CO from bridge sites (defects and (100) planes) and reacts to benzaldehyde only
Letters
J. Phys. Chem. B, Vol. 110, No. 46, 2006 22985
Figure 4. (a) Time dependence of the signal of bridge CO (initially at 1968 cm-1, plain circles) and of benzaldehyde (1713 cm-1, diamonds, 10 times magnified) when admitting a solution of benzyl alcohol under Ar on Pd/Al2O3 pre-equilibrated with CO as in Figure 3. Insets depict snapshots of the CO signals and show the sudden appearance of the signal at 1713 cm-1 simultaneously with the depletion of the signal at 1968 cm-1. (b) Artist view on the structural dependence of the oxidative dehydrogenation of benzyl alcohol on an ideal cuboctahedron Pd crystallite. Alcohol dehydrogenation occurs on all Pd sites (red arrows), whereas product decomposition occurs predominantly on (111) planes (blue arrow). Only dehydrogenation on edges and (111) planes is represented for clarity. The black dot represents CO on hollow sites (top view).
when these sites are liberated. Additionally, it provides important information on the product decarbonylation, which contributes to catalyst deactivation in the absence of oxygen. Benzaldehyde desorbs intact from the sites from which CO is displaced by benzyl alcohol and does not decompose on those sites where CO is more strongly bound (envelope between 1950 and 1700 cm-1). Because the signal at 1853 cm-1 observed during benzyl alcohol oxidation on the untreated Pd catalyst (Figure 1) resembles the feature observed during the same reaction on the catalyst pre-equilibrated with CO (Figure 3a), product decarbonylation occurs preferentially on such sites. These sites are predominantly hollow sites on (111) faces. The spectroscopic data obtained in the liquid phase indicate that benzyl alcohol dehydrogenation occurs on all faces exposed by the Pd particles. However, on (111) faces the product can decompose to CO and carbon-containing fragments, like benzene. The improved adsorption of the product is likely the origin of the activity of (111) faces toward benzaldehyde decarbonylation. The phenyl ring of benzaldehyde can stabilize the carbonyl group on the metal surface, adsorbed in the η2 mode,22 thus increasing the residence time of the product on the regular metal surface and therefore rendering C-C scission more probable.23 On the contrary, benzaldehyde adsorbs only “temporarily” on more open surfaces and defects and desorbs intact. This agrees with the preferential adsorption of acetophenone, the product of 1-phenylethanol oxidation that does not decarbonylate,17,24 on on-top and bridge sites.24 The structure sensitivity of aldehyde decarbonylation implied by the present results confirms previous UHV studies, which attributed the apparent structure sensitivity of alcohol reactions on Pd to the different stability of the product on different surfaces.25 It is not possible to assess which sites are also responsible for alcohol dehydration affording toluene, the typical byproduct accompanying dehydrogenation. UHV studies indicate that dehydration occurs on (111) planes but not on more open surfaces.26
All of these findings provide a reasonable explanation why supported Pd catalysts promoted by bismuth or lead are the most efficient industrial catalyst for alcohol oxidation. Pd-Bi catalysts are characterized by Bi deposited selectively on the Pd particles.27 The ATR-IR spectra of a 5 wt % Pd-1 wt % Bi/ Al2O3 showed no CO formation under identical conditions to those presented in the spectra of Figure 1.28 The data presented here clearly indicate that Bi occupies predominantly (111) planes and therefore prevents benzaldehyde decarbonylation. In conclusion, in situ ATR-IR spectroscopy combined with selective site blocking by coadsorbed CO and isotope labeling allowed us to discriminate the different sites active in the liquidphase oxidative dehydrogenation of benzyl alcohol on Pd/Al2O3 (Figure 4b). Alcohol dehydrogenation to the carbonyl product is found to be rather structure-insensitive, occurring on virtually all exposed Pd sites. However, some of these sites (bridge sites) release the intact product and are liberated for adsorption of the reactant, whereas other sites (threefold or hollow) promote product degradation (decarbonylation). The sites responsible for product degradation are likely distributed on (111) terraces. In a more general sense, the combined method described in this study could give access to the discrimination of active sites in various other solid catalyzed liquid-phase reactions and this under working conditions. Acknowledgment. We acknowledge financial support from the Foundation Claude and Giuliana. C.M. is thankful for a scholarship from the University of Milan. Supporting Information Available: Complete materials and methods including experimental details of ATR-IR measurements and electron microscopy. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Williams, C. T.; Yang, Y.; Bain, C. D. Langmuir 2000, 16, 2343. (2) Bu¨rgi, T.; Baiker, A. AdV. Catal. 2006, 50, 227.
22986 J. Phys. Chem. B, Vol. 110, No. 46, 2006 (3) Grunwaldt, J. D.; Baiker, A. Phys. Chem. Chem. Phys. 2005, 7, 3526. (4) Roeffaers, M. B. J.; Sels, B. F.; Uji-i, H.; DeSchryver, F. C.; Jacobs, P. A.; DeVos, D. E.; Hofkens, J. Nature 2006, 439, 572. (5) Somorjai, G. A. Chem. ReV. 1996, 96, 1223. (6) Over, H.; Kim, Y. D.; Seitsonen, A. P.; Wendt, S.; Lundgren, E.; Schmid, M.; Varga, P.; Morgante, A.; Ertl, G. Science 2000, 287, 1474. (7) Gunter, P. L. J.; Niemantsverdriet, J. W.; Ribeiro, F. H.; Somorjai, G. A. Catal. ReV. Sci. Eng. 1997, 39, 77. (8) Ba¨umer, M.; Freund, H. J. Prog. Surf. Sci. 1999, 61, 127. (9) Silvestre-Albero, J.; Rupprechter, G.; Freund, H. J. J. Catal. 2006, 240, 58. (10) Schalow, T.; Brandt, B.; Laurin, M.; Schauermann, S.; Guimond, S.; Kuhlenbeck, H.; Libuda, J.; Freund, H. J. Surf. Sci. 2006, 600, 2528. (11) Mavrikakis, M.; Barteau, M. A. J. Mol. Catal. A: Chem. 1998, 131, 135. (12) Mallat, T.; Baiker, A. Chem. ReV. 2004, 104, 3037. (13) Bu¨rgi, T.; Bieri, M. J. Phys. Chem. B 2004, 108, 13364. (14) Keresszegi, C.; Ferri, D.; Mallat, T.; Baiker, A. J. Phys. Chem. B 2005, 109, 958. (15) Keresszegi, C.; Grunwaldt, J.-D.; Mallat, T.; Baiker, A. J. Catal. 2004, 222, 268.
Letters (16) Davis, J. L.; Barteau, M. A. Surf. Sci. 1992, 268, 11. (17) Keresszegi, C.; Ferri, D.; Mallat, T.; Baiker, A. J. Catal. 2005, 234, 64. (18) Lear, T.; Marshall, R.; Lopez-Sanchez, J. A.; Jackson, S. D.; Klapo¨tke, T. M.; Ba¨umer, M.; Rupprechter, G.; Freund, H. J.; Lennon, D. J. Chem. Phys. 2005, 123, 174706. (19) Yee, N. C.; Chottiner, G. S.; Scherson, D. A. J. Phys. Chem. B 2005, 109, 7610. (20) Schauermann, S.; Hoffmann, J.; Johanek, V.; Hartmann, J.; Libuda, J.; Freund, H. J. Angew. Chem., Int. Ed. 2002, 41, 2532. (21) Xu, J.; Henriksen, P. N.; Yates, J. T. Langmuir 1994, 10, 3663. (22) Delbecq, F.; Sautet, P. Surf. Sci. 1993, 295, 353. (23) Grimaud, C. M.; Radosavkic, D.; Ustaze, S.; Palmer, R. E. Appl. Surf. Sci. 2001, 178, 1. (24) Chen, C. S.; Lin, J. H.; Chen, H. W. Appl. Catal., A 2006, 298, 161. (25) Shekhar, R.; Barteau, M. A. Catal. Lett. 1995, 31, 221. (26) Shekhar, R.; Barteau, M. A. Surf. Sci. 1994, 319, 298. (27) Keresszegi, C.; Mallat, T.; Grunwaldt, J. D.; Baiker, A. J. Catal. 2004, 225, 138. (28) Mondelli, C.; Ferri, D.; Grunwaldt, J. D.; Krumeich, F.; Baiker, A., to be submitted for publication.
