25586
2006, 110, 25586-25589 Published on Web 11/29/2006
Oxidic or Metallic Palladium: Which Is the Active Phase in Pd-Catalyzed Aerobic Alcohol Oxidation? Jan-Dierk Grunwaldt,* Matteo Caravati, and Alfons Baiker Institute of Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zurich, Ho¨nggerberg-HCI, CH-8093 Zurich, Switzerland ReceiVed: October 23, 2006; In Final Form: NoVember 17, 2006
In situ X-ray absorption spectroscopy combined with on-line catalytic measurements using FT-IR spectroscopy unequivocally identified that metallic palladium is the more active phase in the aerobic oxidation of benzyl alcohol than palladium oxide. The aerobic oxidation of benzyl alcohol in cyclohexane at 50 °C was low over oxidized 0.5%Pd/Al2O3 and 5%Pd/Al2O3 catalysts. XANES and EXAFS showed that the catalysts in the as-received state were almost fully oxidized and no reduction of the palladium constituent was observed during time-on-stream. After in situ reduction by hydrogen-saturated cyclohexane, the catalysts were much more active (over 50 times) than before reduction. Both XANES and EXAFS uncovered that the palladium constituent was mainly in a reduced state under these conditions of high catalytic activity. This demonstrates that metallic palladium is the active phase for alcohol dehydrogenation.
Introduction Aerobic alcohol oxidation over noble-metal-based catalysts is an important process in fine chemistry and has attracted considerable attention during the past decades.1 It is a green chemistry approach because molecular oxygen is used as the sole oxidant and water is the only byproduct. However, the mechanism is still a matter of discussion and unambiguous proof of the nature of the active noble metal sites appears of fundamental importance.2-4 In several studies it was proposed that metallic palladium or platinum are the active species and thus the dehydrogenation step is prevalent.2,3 These studies have been mainly based on electrochemical measurements investigating alcohol oxidation in water and on X-ray absorption spectroscopy studies in both aqueous media and organic solvents. The role of oxygen is in those cases mainly traced back to the removal of surface hydrogen, which could also be shown indirectly by using hydrogen acceptors.5 On the other hand, it has been speculated that the oxidized noble metal particles (e.g., palladium oxide) are responsible for the catalytic activity.4 The catalytic activity was reported to be higher on oxidized noble metal particles than after the reduction by the reactant alcohol.4b However, the spectroscopic studies were not performed “truly” in situ, as required for the determination of structure-performance relationships.6 As discussed in detail in a previous study,6b for such investigations the catalyst should be ideally present as powder or shell-impregnated particles and the catalytic activity should be simultaneously measured on-line. This prompted us to design a setup that allows following structural changes in liquid organic media by X-ray absorption spectroscopy and the catalytic activity by on-line infrared * Corresponding author. E-mail:
[email protected]. Phone: +4144-632 30 93. Fax: +41-44-632 11 63.
10.1021/jp066949a CCC: $33.50
SCHEME 1: Oxidation of Benzyl Alcohol over Pd/Al2O3 Catalysts
spectroscopy. In the present study we compare a palladium catalyst in its oxidized and reduced state. The catalytic activity and the catalyst structure were investigated during the aerobic oxidation of benzyl alcohol (cf. Scheme 1) before and after reduction of the catalyst in hydrogen, applying in both cases the same reaction conditions. The reaction temperature was set at 50 °C, where the alcohol alone was not expected to reduce palladium oxide but the catalytic activity was high enough to be monitored by on-line FT-IR. Experimental Section A complete description of the experimental setup is given in the Supporting Information. In brief, it consists of a spectroscopic reaction cell in which the catalyst is placed (as powder and/or shell-impregnated catalyst). The cell mimics the conditions of a continuous-flow fixed-bed reactor and is equipped with X-ray transparent windows for the spectroscopic studies. A liquid pump allowed feeding the liquid reactants into the in situ reaction cell, which can be heated by a specially designed oven. Monitoring of reaction progress was achieved by a transmission flow-through IR cell, which was connected to the exit of the in situ reaction cell. The experiments were performed at beamline X1 at HASYLAB (DESY, Hamburg). Both a commercial 5%Pd/Al2O3 catalyst in the form of a powder (ca. 50 µm mean particle size) and a commercial 0.5%Pd/Al2O3 catalyst in the form of pellets (shell-impregnated, sieved fraction of the crushed catalyst 200 µm < d < 300 µm, active catalytic layer of ca. 60 µm, as determined by SEM) were used for the © 2006 American Chemical Society
Letters
J. Phys. Chem. B, Vol. 110, No. 51, 2006 25587
Figure 2. In situ extended X-ray absorption fine structure (EXAFS) spectra (a) and the corresponding Fourier transformed data (b) of 5%Pd/ Al2O3 during the different steps of the selective benzyl alcohol oxidation depicted in Figure 1.
Figure 1. In situ X-ray absorption spectra of 5%Pd/Al2O3 during the selective oxidation of benzyl alcohol (top) and the relative catalytic activity measured on-line by determining the absorption of the CdO stretching band (bottom) at different steps of the reaction: (a) as-received catalyst in cyclohexane at 50 °C; (b) catalyst in benzyl alcohol/O2/cyclohexane at 50 °C; (c) during reduction in H2/cyclohexane at 50 °C; (d) catalyst in benzyl alcohol/O2/cyclohexane at 50 °C. The cyclohexane solution was saturated with oxygen and hydrogen, respectively, and the concentration of benzyl alcohol was 200 µL/100 mL of cyclohexane.
studies. The spectra were analyzed using the Athena 0.8.049 software7 and the EXAFS spectra were fitted using the FEFF 6.0 code8 (details in ESI). Results and Discussion Figure 1 shows the results from the combined study using in situ X-ray absorption spectroscopy and on-line product analysis by FTIR for the 5%Pd/Al2O3 catalyst. While the as-prepared catalyst was heated to the reaction temperature of 50 °C in
cyclohexane and exposed to the reaction mixture of benzyl alcohol/oxygen/cyclohexane (200 µL of benzyl alcohol in 100 mL of cyclohexane, saturated with oxygen, steps a and b), the structure of the catalyst remained in the same partially oxidized state. Hardly any catalytic activity was measured under these conditions (bottom of Figure 1) as also supported by corresponding laboratory experiments (not shown). No change in the region 1700-1720 cm-1 (carbonyl stretching region) was observed and the C-O stretching vibration at 1018 cm-1 was constant and clearly visible. During exposure to hydrogensaturated cyclohexane, the catalyst was completely reduced after about 30 min, as derived from both the XANES (Figure 1, step c) and EXAFS data (Figure 2). From the EXAFS analysis it further emerged that palladium hydride is formed (backscattering from nearest neighbor Pd atoms decreases and the Pd-Pd distance increases). This leads both to a more elongated EXAFS function (Figure 2a) and to a shift in the Fourier-transformed EXAFS spectra to higher R values (Figure 2b). Upon exposure of the catalyst to the reaction mixture benzyl alcohol/oxygen/ cyclohexane at 50 °C, a much higher catalytic activity of the noble metal was achieved, as indicated by the on-line FT-IR measurements (Figure 1, bottom, step d). In the XANES spectra hardly any changes were observed, indicating that palladium remained metallic during the aerobic oxidation of the alcohol
25588 J. Phys. Chem. B, Vol. 110, No. 51, 2006 (Figure 1, step d). The EXAFS data showed that the hydride was immediately decomposed again, but no oxidation of the Pd particles was observed. The on-line FT-IR measurements showed a certain deactivation of the catalyst with time-onstream, which may be due to accumulation of some byproducts on the catalyst surface.9a However, also after longer time-onstream the structure of the catalyst remained in the same state (cf. Figure 2b) and the catalytic activity was strikingly higher (>50 times) than that observed when the oxidized palladium catalyst was used. No effect of the X-ray beam was detected; the catalytic activity was the same if the beam was switched on or off. Because Lee et al.4b reported an important role of the metal dispersion and of surface palladium oxide, we additionally studied a low-concentration commercial and highly dispersed shell-impregnated catalyst. Due to the higher dispersion of the palladium particles, in this case the Pd particles in the asreceived catalyst were completely reoxidized after the prereduction performed by the manufacturer. Nevertheless, the same results were obtained (cf. Supporting Information). No change of the oxidation state and hardly any catalytic activity were observed before in situ reduction. The catalyst was rapidly reduced in hydrogen-saturated cyclohexane (within less than 12 min), due to the shell-type macro structure and the lower Pd content. Also in this case, after reduction, the catalyst was much more active in the oxidation of benzyl alcohol, as the on-line FT-IR measurements showed. The catalytic activity was by a factor of 4-5 lower than that of the 5%Pd/Al2O3. Nevertheless, the Pd constituent remained in a metallic state. Due to the similarity of the results to those obtained with the 5%Pd/Al2O3 catalyst, we can exclude that noble metal loading or the macro structure of the catalyst (shell-impregnated or homogeneous powder catalyst) affects the interpretation of our results. Note that in oxygen/cyclohexane a rapid reoxidation of the palladium constituent occurred (see additional details in the Supporting Information). In further experiments the Pd-based catalysts were reoxidized in a mixture of oxygen/cyclohexane at 30 °C, and after this treatment the catalytic activity was lower but increased with time-on-stream. Raising the temperature using the as-prepared catalysts (without any pretreatment) typically showed a higher activity at 80 °C.3 This can be explained by the fact that the alcohol is able to reduce the Pd constituent efficiently at 80 °C.3b Therefore a high catalytic activity on the as-prepared catalysts may be found at higher temperature as, e.g., reported in ref 4b, but also in this case the origin of the high catalytic activity is probably connected to the formation of metallic sites upon reduction by the alcohol. In summary, our results demonstrate that palladium oxide exhibits hardly any catalytic activity for alcohol oxidation at 50 °C, whereas metallic Pd particles are much more active. This is in agreement with the dehydrogenation mechanism, depicted in Figure 3. The first step is the oxidation of the alcohol by dehydrogenation, which results in the formation of the aldehyde and adsorbed hydrogen on the Pd surface. In the second step hydrogen is removed from the Pd surface by oxidation (with oxygen). This mechanism is supported by several studies in literature that describe the use of hydrogen acceptors other than oxygen, cyclovoltammetric experiments in liquid phase, and also the investigation of promoter atoms during the oxidation of 1-phenylethanol.2,3,5 In these studies it was also found that “overoxidation” of the Pd surface leads to lower activity of the catalyst, as also supported by the present study (cf. deactivation by oxidation of Pd particles in Figure 3). In fact, Weaver and
Letters
Figure 3. Mechanism and deactivation pathways during Pd-catalyzed alcohol oxidation. Metallic noble metal sites are required for the dehydrogenation step; oxygen (e.g., adsorbed on the Pd surface) acts as hydrogen acceptor. This mechanism does not require the Pd surface to be completely reduced (details see text).
co-workers showed a quite similar effect for methanol oxidation on Pd-based catalysts with the use of in situ SERS.9 In addition, deactivation can occur by formation and adsorption of degradation products, which is probably at the origin of the initially higher catalytic activity after the hydrogen treatment.10 Hence, for alcohol oxidation one should choose conditions that keep the Pd constituent in a metallic state but allow an optimal availability of oxygen to effectively remove the formed hydrogen and degradation products from the catalyst surface. The most beneficial situation is that metallic Pd sites are available connected with rapid access to adsorbed oxygen. For strongly reducing alcohols such as benzyl alcohol, this situation can be advantageously achieved in supercritical fluids.3b The advantage of the present approach was that by choosing a highly active alcohol and a moderate temperature of 50 °C, the alcohol was not able to reduce the palladium oxide itself, whereas once reduced, the activity was high enough to prevent “overoxidation” of the Pd catalysts. In addition, we were able to observe the catalytic activity on-line by FT-IR, while simultaneously probing the structure of the catalyst by X-ray absorption spectroscopy. The presented approach can be regarded as an important step forward to generally derive structure-activity relationships under dynamic reaction conditions also in liquid-phase reactions, which up to now have been much less studied than gas-phase reactions, where EXAFS is typically combined with on-line mass spectrometry.11 Acknowledgment. We thank HASYLAB at DESY (Hamburg, Germany) for providing beamtime for the in situ XANES/ EXAFS studies and the beam line staff for help and support. The work at the synchrotron radiation source was supported by the European Community-Research Infrastructure Action under the FP6 “Structuring the European Research Area” program (through the Integrated Infrastructure Initiative “Integrating Activity on Synchrotron and Free Electron Laser Science”, Contract RII3-CT-2004-506008). Special thanks are due to Dr. D. Ferri for fruitful discussions, and P. Tru¨ssel (mechanical workshop, ETH Zurich) is acknowledged for his continuous technical assistance. Supporting Information Available: Experimental details, EXAFS and catalytic data during benzyl alcohol oxidation over Pd/Al2O3 catalysts as well as IR-spectra during on-line catalytic measurements are available free of charge via the Internet at http://pubs.acs.org.
Letters References and Notes (1) Besson, M.; Gallezot, P. Catal. Today 2000, 57, 127. Kluytmans, J. H. J.; Markusse, A. P.; Kuster, B. F. M.; Marin, G. B.; Schouten, J. C. Catal. Today 2000, 57, 143. Mallat, T.; Baiker, A. Chem. ReV. 2004, 104, 3037. (2) Vinke, P.; de Wit, D.; Goede, A. T. J. W.; van Bekkum, H. Catal. 1992, 72, 1. Mallat, T.; Bodnar, Z.; Hug, P.; Baiker, A. J. Catal. 1995, 153, 131. (3) (a) Grunwaldt, J.-D.; Keresszegi, C.; Mallat, T.; Baiker, A. J. Catal. 2004, 225, 138. Keresszegi, C.; Grunwaldt, J.-D.; Mallat, T.; Baiker, A. J. Catal. 2004, 222, 268. (b) Caravati, M.; Grunwaldt, J.-D.; Baiker, A. Catal. Today 2004, 91-92, 1. Grunwaldt, J.-D.; Caravati, M.; Baiker, A. J. Phys. Chem. B 2006, 110, 9916. (4) (a) Smits, P. C. C.; Kuster, B. F. M.; van der Wiele, K.; van der Baan, H. S. Appl. Catal. 1986, 33, 83. (b) Lee, A. F.; Wilson, K. Green Chem. 2004, 6, 37. Lee, A. F.; Hackett, S. F. J.; Hargreaves, J. S. J.; Wilson, K. Green Chem. 2006, 8, 549. (5) Keresszegi, C.; Mallat, T.; Baiker, A. New J. Chem. 2001, 25, 1163.
J. Phys. Chem. B, Vol. 110, No. 51, 2006 25589 (6) (a) Somorjai, G. A.; Thomas, J. M., Eds. Top. Catal. 1999, 8. Topsøe, H. J. Catal. 2003, 216, 155. Weckhuysen, B. M. Chem. Commun. 2002, 97. Banares, M. A. Catal. Today 2005, 100, 71. (b) Grunwaldt, J.D.; Caravati, M.; Hannemann, S.; Baiker, A. Phys. Chem. Chem. Phys. 2004, 6, 3037. (7) Ravel, B.; Newville, M. J. Synchrotron Rad. 2005, 12, 537. (8) Zabinsky, S. I.; Rehr, J. J.; Ankudinov, A.; Albers, R. C.; Eller, M. J. Phys. ReV. B 1995, 52, 2995. (9) Chan, H. Y. H.; Williams, C. T.; Weaver, M. J.; Takoudis, C. G. J. Catal. 1998, 174, 191. (10) (a) Keresszegi, C.; Ferri, D.; Mallat, T.; Baiker, A. J. Phys. Chem. 2005, 109, 958. (b) Keresszegi, C.; Bu¨rgi, T.; Mallat, T.; Baiker, A. J. Catal. 2002, 211, 244. (11) Clausen, B. S. Catal. Today 1998, 39, 293. Revel, R.; Bazin, D.; Seigneurin, A.; Barthe, P.; Dubuisson, J. M.; Decamps, T.; Sonneville, H.; Poher, J. J.; Maire, F.; Lefrancois, P. Nucl. Instrum. Methods Phys. Res. B 1999, 155, 183. Grunwaldt, J.-D.; Clausen, B. S. Topics Catal. 2002, 18, 37. Newton, M. A.; Jyoti, B.; Dent, A. J.; Fiddy, S. G.; Evans, J. Chem. Commun. 2004, 2382.
