Al2O3 Catalysts Prepared

10612

Langmuir 2004, 20, 10612-10616

FTIR Studies of CO Adsorption on Rh-Ge/Al2O3 Catalysts Prepared by Surface Redox Reactions Gwendoline Lafaye,†,‡ Corina Mihut,‡ Catherine Especel,† Patrice Mare´cot,† and Michael D. Amiridis*,‡ Laboratoire de Catalyse en Chimie Organique, UMR CNRS 6503 Universite´ de Poitiers, 40 Avenue du Recteur Pineau, F-86022 Poitiers Cedex, France, and Department of Chemical Engineering, University of South Carolina, Columbia, South Carolina 29208 Received February 4, 2004. In Final Form: September 7, 2004 A series of bimetallic Al2O3-supported Rh-Ge catalysts was prepared by surface redox reactions under controlled hydrogen atmosphere. The surface properties of these catalysts were probed via in-situ FTIR spectroscopic studies of adsorbed CO and were compared to those of monometallic Rh catalysts that had undergone similar treatments. The results indicate that Ge addition results in the formation and stabilization of smaller rhodium ensembles at the expense of larger Rh0 surfaces. A charge-transfer mechanism from Ge to Rh is also inferred by the IR results for the high Ge loading samples. Air exposure of the catalysts leads to an irreversible segregation of the two metals and formation of large Rh crystallites.

Introduction The catalytic behavior of rhodium is of great interest due to its wide array of heterogeneous catalytic applications in automotive exhaust, hydrogenation/ dehydrogenation, and hydroformylation applications. Attempts have been made in the past to improve the catalytic behavior of Rh by the introduction of a second metal.1-6 However, significant variations in catalyst morphology, and subsequently reactivity, can be observed in bimetallic systems as a result of preparation procedures. Very often, for example, the use of simple co-impregnation procedures from individual metal salt precursors leads to the formation of segregated monometallic particles due to the thermodynamic instability of the alloy of the two metals involved. An alternative approach is the utilization of a redox reaction, which takes place in the liquid phase between a prereduced parent monometallic catalyst or a reducer adsorbed on this parent metal and a solution containing an oxidized form of the second metal.3,6-10 This approach has been successful in the past in yielding stable supported bimetallic particles for different metal combinations. The characterization of the surface morphology and properties of bimetallic particles is not trivial. Frequently, * Author to whom correspondence should be addressed: Fax: (803)777-8265. E-mail: [email protected]. † Universite ´ de Poitiers. ‡ University of South Carolina. (1) Didillon, B.; Candy, J. P.; Le Peletier, F.; Ferretti, O. A.; Basset, J. M. Stud. Surf. Sci. Catal. 1993, 78, 163. (2) Coupe, J. N.; Jorda˜o, E.; Fraga, M. A.; Mendes, M. J. Appl. Catal. A 2000, 199, 45. (3) Lafaye, G.; Micheaud-Especel, C.; Montassier, C.; Marecot, P. Appl. Catal. A 2002, 230, 19. (4) Sordelli, L.; Psaro, R.; Vlaic, G.; Cepparo, A.; Recchia, S.; Dossi, C.; Fusi, A.; Zanoni, R. J. Catal. 1999, 182, 186. (5) Quirmbach, M.; Kless, A.; Holz, J.; Tararov, V.; Bo¨rner, A. Tetrahedron: Asymmetry 1999, 10, 1803. (6) Szabo, S. Int. Rev. Phys. Chem. 1991, 10, 207. (7) Barbier, J. Advances in Catalyst Preparation, Study Number 4191 CP; Catalytica Studies Division: Mountain View, CA, 1992. (8) Pieck, C. L.; Marecot, P.; Barbier, J. Appl. Catal. A. 1996, 134, 319. (9) Bachir, R.; Marecot, P.; Didillon, B.; Barbier, J. Appl. Catal. A. 1997, 164, 313. (10) Dumas, J. M.; Geron, C.; Hadrane, H.; Marecot, P.; Barbier, J. J. Mol. Catal. 1992, 77, 87.

probe molecules or probe reactions are involved in these efforts. In the case of Rh-containing catalysts, the study of CO adsorption can be revealing since previous studies11-13 have documented the presence of several different types of adsorbed CO species on Rh and have correlated these species with different Rh structures. In particular, a bridge-bonded CO species (with a characteristic band at 1850 cm-1) together with a linearly-bonded CO species (with a characteristic band at approximately 2040-2070 cm-1) are formed on large Rh surfaces, while a geminal dicarbonyl species (with characteristic symmetric and antisymmetric bands at 2030 and 2090-2100 cm-1, respectively) is formed on small clusters or atomically dispersed Rh1+ sites. In this paper, we report the results of our characterization studies of bimetallic alumina-supported Rh-Ge catalysts. These catalysts were synthesized via the surface redox reaction approach. Previous studies of these catalysts for the dehydrogenation of cyclohexane3 and the selective hydrogenation of citral3,14 have shown that this method of preparation leads to the formation of bimetallic Rh-Ge particles. FTIR spectroscopic studies of adsorbed CO are utilized in this report in an effort to understand better the effect of Ge addition on the surface properties of Rh. Experimental Section Catalyst Preparation: The monometallic 1.0 wt% Rh catalyst was prepared via cationic exchange of the hydroxyl groups of the γ-alumina support (240 m2/g; particle size of 0.04-0.1 mm) with chloropentaamine rhodium [RhCl(NH3)5]2+ in an aqueous solution in the presence of NH4+ ions (pH ) 11). Following this exchange process, the catalyst was calcined in flowing air for 4 h at 300 °C, and reduced in pure hydrogen for 4 h at 500 °C. This sample will be further referred to as “monometallic” 1Rh catalyst. Using the same procedure, a monometallic 3.6 wt% Ge sample (further referred to as monometallic 4Ge) was also prepared from a GeCl4 precursor under inert (N2) atmosphere. All bimetallic Rh-Ge catalysts were prepared from the same batch of 1Rh catalyst (“parent catalyst”) via the surface redox (11) Yang, A. C.; Garland, G. W. J. Phys. Chem. 1957, 61, 1504. (12) Solymosi, F.; Pasztor, M. J. Phys. Chem. 1986, 90, 5312. (13) Dictor, R.; Roberts, S. J. Phys. Chem. 1989, 93, 2526. (14) Lafaye, G.; Micheaud-Especel, C.; Marecot, P. Appl. Catal. A 2004, 257, 107.

10.1021/la049692l CCC: $27.50 © 2004 American Chemical Society Published on Web 10/22/2004

FTIR Studies of CO Adsorption on Rh-Ge/Al2O3 Catalysts

Langmuir, Vol. 20, No. 24, 2004 10613

Table 1. Elemental Analysis and Rh Dispersion Data for the Rh and Rh-Ge Catalysts Used in This Study content (%) catalyst

Rh

“blank” 1Rh Ge 1Rh1Ge 1Rh3Ge 1Rh5Ge

1.0 1.0 1.0 1.0

Ge expa 3.6 1.0 3.0 5.0

Ge actb 3.6 1.0 2.5 3.5

Cl

Rh dispersion (%)

1.2

36.0 n/a 11.8 8.5 5.3

1.0 1.2 1.4

a Expected final Ge loading on the catalyst if the full amount of Ge in solution would deposit on the catalyst. b Actual Ge loading on the catalyst

reaction between hydrogen activated on rhodium particles and GeCl4 dissolved in water (“catalytic reduction” method). A known amount of the prereduced parent catalyst was introduced to the reactor (described in detail elsewhere10), and activated for 1 h under H2 flow at 300 °C. Appropriate solutions of the germanium precursor, calculated to yield final Ge contents of 1, 3, and 5 wt% and previously degassed under H2 flow, were then added to the reactor at room temperature. After 1 h of reaction time under H2 flow, the solution was filtered and the catalyst was dried overnight at 120 °C under flowing H2. Finally, each bimetallic catalyst sample (further referred to as 1Rh1Ge, 1Rh3Ge, and 1Rh5Ge) was reduced under flowing H2 for 1 h at 300 °C (2 °C/ min heating rate). Using the same approach a “blank” rhodium catalyst was also prepared by replacing the germanium salt with a hydrochloric acid solution of the same pH, so that the level of chlorine on this catalyst would match that of the bimetallic samples. The blank rhodium sample was further treated identically to the bimetallic ones (i.e., drying overnight at 120 °C and reduction for 1 h under flowing H2 at 300 °C). 1Rh3Ge and 1Rh5Ge samples were also exposed to air and then re-reduced at 300 °C. This treatment was performed because previous kinetic results suggest substantial changes in the catalytic activity of these materials following exposure to air.14 Elemental analysis of the Rh, Ge, and Cl contents of the different catalyst samples, as well as the GeCl4 solutions used for their preparation, was performed via atomic absorption measurements. The results of these measurements are summarized in Table 1. Rh dispersion data obtained via hydrogen chemisorption measurements at room temperature are also included in Table 1. Prior to these measurements, the samples were reduced insitu in flowing H2 at 300 °C for 1 h, followed by evacuation with Ar at the same temperature for 1 h. The samples were then cooled to room temperature and exposed to H2 for the chemisorption measurements. Rh dispersions were obtained utilizing the H2 uptake (since Ge does not chemisorb H2), the Rh content (obtained from the elemental analysis) and assuming a H/Rh ratio of 1:1. Fourier Transform Infrared (FTIR) Spectroscopy Studies of Adsorbed CO. Transmission FTIR spectra were collected in the single-beam mode, with a resolution of 2 cm-1, using a Nicolet Nexus 470 FT-IR spectrometer equipped with an MCT-B detector. A 10-cm long stainless steel IR cell, with NaCl windows cooled by flowing water, was used to collect in situ spectra. A heating element wrapped around the cell allowed spectra collection at elevated temperatures. The cell temperature was monitored by a thermocouple placed in close proximity with the catalyst sample. Reference spectra of the clean surfaces in He were collected at room temperature. Difference spectra between the samples and the corresponding reference are shown in this paper. Catalyst samples were prepared as self-supported wafers with a diameter of 12 mm and a “thickness” of approximately 40 mg/cm2. The wafers were pressed and loaded in the IR cell under inert atmosphere in order to avoid contact with air. Prior to CO adsorption experiments, all catalysts were reduced in-situ for 2 h in a flowing 5% H2/He mixture at 300 °C. The flow rate during this, as well as all other subsequent treatments, was maintained at 70 mL/min. Following this reduction step, the samples were flushed with He at 300 °C and cooled in He to room temperature. At room temperature, a flowing 5% CO/He mixture

Figure 1. FTIR spectrum of CO adsorbed at room temperature on the blank 1Rh sample. (purified to remove trace amounts of O2 and H2O) was introduced to the cell and spectra were collected until the steady state was reached. Helium was then purged through the cell to remove any weakly bonded CO. When necessary for peak assignments, identical studies were also conducted with a flowing 5% 13CO/He mixture. Spectral deconvolution was performed using the Galactic PeakSolve peak-fitting program. All deconvoluted spectra shown in this paper are converged solutions, with correlation factors above 0.999 and standard errors below 0.004.

Results and Discussion Monometallic Rh catalysts: The deconvoluted roomtemperature FTIR spectrum of CO adsorbed on the blank 1Rh sample reduced at 300 °C is shown in Figure 1. Five different features are present in this spectrum: two strong peaks at 2034 and 2101 cm-1, two shoulders at 2065 and 2020 cm-1, and a broad peak centered at approximately 1865 cm-1. Similar results have been previously reported for the adsorption of CO on Rh/Al2O3, and the different spectral features observed have been assigned to the symmetric and antisymmetric stretching vibrations of a geminal dicarbonyl species and the stretching vibrations of linear- and bridge-bonded CO on different types of Rh.12,15-17 These assignments are summarized in Table 2. In particular, the bridge- and linear-bonded bands (1865 and 2065 cm-1, respectively) can be assigned to CO adsorbed on Rh0 surfaces. In contrast, the geminal dicarbonyl species (symmetric and antisymmetric stretches at 2034 and 2101 cm-1, respectively) is believed to be present on dispersed two-dimensional rafts18 or even atomically dispersed Rh1+.19,20 The formation of highly dispersed Rh1+ is believed to be the result of an oxidative disruption of the Rh-Rh bonds in the Rh crystallites at (15) Rice, C. A.; Worley, S. D.; Curtis, C. W.; Guin, J. A.; Tarrer, A. R. J. Chem. Phys. 1981, 74, 6487. (16) Panayotov, D.; Basu, P.; Yates, J. T., Jr. J. Phys. Chem. 1988, 92, 6066. (17) Trautmann, S.; Baerns, M. J. Catal. 1994, 150, 335. (18) Yates, D. J. C.; Murrell, L. L.; Prestridge, E. B. J. Catal. 1979, 57, 41. (19) Yates, J. T.; Duncan, T. M.; Worley, S. T.; Vaughan, R. W. J. Chem. Phys. 1979, 70, 1219. (20) Yates, J. T.; Duncan, T. M.; Worley, S. T.; Vaughan, R. W. J. Chem. Phys. 1979, 71, 3908.

10614

Langmuir, Vol. 20, No. 24, 2004

Lafaye et al.

Table 2. Peak Assignments and Relative Surface Areas under Each Peak Observed during the Adsorption of CO on Al2O3-supported Monometallic Rh, and Bimetallic Rh-Ge Samples catalyst sample “blank” 1Rh 1Rh1Ge 1Rh3Ge 1Rh5Ge

bridge on Rh0x

bridge on Rh+1

gem asym (Rh+) Rh(CO)2

linear Rh0x RhCO

gem sym (Rh+) Rh(CO)2

linear (Rh>1+) ORhCO

1865 4.6% 1854 2.5%

2025 20.6% 2017 23.0% 2018 13.4% 2016 8.1%

2034 30.1% 2028 20.2% 2034 15.5% 2038 12.9%

2065 21.8% 2074 19.2% 2067 18.3% 2085 11.3%

2101 22.9% 2098 35.1% 2101 50.6% 2102 51.6%

2129 2.2% 2132 16.1%

room temperature and under CO atmosphere, due to the higher energy of the Rh-CO bond as compared to the Rh-Rh one.21 Consequently, the presence of the geminal dicarbonyl bands in the spectra should not be interpreted to suggest the presence of isolated Rh1+ ions on the support, but rather isolated Rh1+ islands on the larger Rh0 crystallites induced by the presence of CO. The geminal dicarbonyl bands appear immediately following exposure of the sample to CO, indicating that the disruptive effect of CO is very fast. Using 12CO/13CO isotopic exchange, Yates et al.16,22,23 showed that the oxidative disruption takes place via a nondissociative mechanism involving isolated -OH groups from the support. Consequently, supports with higher Lewis acidity (such as Al2O3) are more likely to enhance the formation of Rh1+ species. The presence of the spectral feature at approximately 2020 cm-1 is observed in the deconvoluted spectra of most catalysts; introduction of such a peak was required in order to obtain converged solutions during the deconvolution process. Spectra of adsorbed 13CO show this peak more clearly as a shoulder. The presence of such a spectral feature has been also postulated by Rice et al.15 to account for the higher intensity of the geminal antisymmetric band in their 10% Rh/Al2O3 sample as compared to that observed for the 0.5 and 2.2% Rh samples. Furthermore, a similar peak at 2017 cm-1 has been observed by Trautmann et al.17 on a Rh/TiO2 sample and was assigned to bridgebonded species on Rh1+ sites. In accordance with these previous reports, we assign the 2020 cm-1 peak in our spectra to a bridge-bonded species on rhodium sites of higher oxidation state (+1 or +2). Bimetallic Rh-Ge catalysts: Room-temperature spectra of CO adsorbed on different bimetallic Rh-Ge catalysts reduced at 300 °C are shown in Figure 3. The corresponding peak assignments and the relative areas under each peak are summarized in Table 2. Similar spectral features were observed with the bimetallic samples, as with the monometallic Rh one shown in Figure 1. This is an anticipated result since no CO adsorption was observed on the monometallic 4Ge sample. Nevertheless, as can be seen in Table 2, the relative intensities of the different peaks change as a function of Ge loading. Chemisorption results summarized in Table 1 indicate that the total amount of adsorbed H2 strongly decreases with the germanium content of these catalysts. However, previous HRTEM characterization3 showed a mean comparable particle size between the bimetallic catalysts and the blank 1Rh one, suggesting that Ge covers the Rh surface. The results of Table 2 indicate that the different surface CO species are not affected proportionally by the presence of Ge. In fact, the effect of Ge is substantially (21) Van’t Bilk, H. F. J.; Van Zon, J. B. A. D.; Huizinga, T.; Vis, J. C.; Koningsberger, D. C.; Prins, R. J. Phys. Chem. 1983, 87, 2264. (22) Basu, P.; Panayotov, D.; Yates, J. T., Jr. J. Phys. Chem. 1987, 91, 3133. (23) Basu, P.; Panayotov, D.; Yates, J. T., Jr. J. Am. Chem. Soc. 1988, 110, 2074.

Figure 2. FTIR spectra of CO adsorbed at room temperature on (a) the blank 1Rh, (b) 1Rh1Ge, (c) 1Rh3Ge, and (d) 1Rh5Ge samples.

more pronounced on the decrease of the bridged species, as well as on species assigned to Rh0x. The peak corresponding to bridge-bonded CO species on Rh0 disappears from the spectra of the samples with the highest germanium contents. In contrast, the gem-symmetric species on Rh1+ becomes dominant on these catalysts. These observations can be interpreted to suggest that the presence of Ge leads to the creation of smaller rhodium ensembles at the expense of the larger Rh0 surfaces and support the hypothesis that Ge gradually covers the Rh surface. Solely on the basis of the FTIR results, we cannot determine whether the necessary charge for the creation of Rh1+ sites responsible for the gem-dicarbonyl species is transferred to Rh from Ge or is generated as a result of the CO oxidative disruption processsalso observed with the monometallic Rhswhich may proceed at a greater extend in the presence of Ge. Moreover, the deconvolution of the spectra of the 1Rh3Ge and 1Rh5Ge samples reveal the presence of a new high-frequency peak at approximately 2130 cm-1 (noticed as a shoulder in the original spectra). A similar peak has been observed previously for CO adsorbed on oxidized Rh/Al2O3 catalysts and has been assigned to CO linearly bonded on higher-oxidation-state rhodium ions, such as Rh2+ and/or Rh3+.17,24,25 Since no oxygen was present during our experiments, the formation of Rh2+ or Rh3+ could be due to either a CO dissociation process or a charge-transfer effect between Rh and Ge. However, the 2130 cm-1 peak was not detected in the spectra of the monometallic Rh or the 1Rh1Ge samples, as would be expected if CO dissociation was responsible for the oxidation of Rh.17 Correlated with the observation that (24) Kondarides, D. I.; Zhang, Z.; Verykios, X. E. J. Catal. 1998, 176, 536. (25) Wey, J. P.; Neely, W. C.; Worley, S. D. J. Catal. 1992, 134, 378.

FTIR Studies of CO Adsorption on Rh-Ge/Al2O3 Catalysts

Langmuir, Vol. 20, No. 24, 2004 10615

Figure 3. FTIR spectra of CO adsorbed at room temperature on (a) the blank 1Rh and (b) 1Rh5Ge samples. Top spectra: CO adsorbed on fresh samples. Bottom spectra: CO adsorbed on samples that have already undergone a CO adsorption/desorption.

increased Ge loadings result in an increase in the intensity of the 2130 cm-1 peak (Figure 3c and d), these findings suggest that a charge transfer is indeed taking place between Rh and Ge, leading to the appearance of rhodium ions in higher oxidation states, more noticeable at higher germanium loadings. Such a conclusion is also consistent with TPR results, which suggest that hydrogen treatment at 300°C is not sufficient for the full reduction of Ge.14 A similar charge-transfer effect has also been reported for Pt-Ge catalysts, with Ge once again being the electron acceptor.26, 27 Furthermore, a comparison of the intensities of the two geminal dicarbonyl bands indicates that increased Ge contents result in a significant increase in the intensity of the symmetric mode at the expense of the antisymmetric one (Figure 3, Table 2). The relative intensities of the two geminal vibration modes can be directly correlated to the angle between the two vibrating CO molecules on the same Rh1+ site according to the following relationship:28

Isym/Iasym ) ctg R where Isym is the intensity of the symmetric geminal dicarbonyl mode, Iasym is the intensity of the antisymmetric geminal dicarbonyl mode, and ctg R is the cotangent of the angle, R, between the two vibrating CO dipoles The significant increase in the intensity of the symmetric mode indicates that the angle, R, between the geminally adsorbed CO molecules decreases upon increasing Ge content. Results of previous studies regarding the adsorption of CO on Rh indicate that a CO adsorption/desorption/ readsorption process at either room17 or higher temperatures13,29 leads to an irreversible reductive agglomeration of Rh1+ clusters to Rh0, as indicated by an increase in the intensity of the linear- and bridge-bonded CO peaks along with a corresponding decrease in the intensity of the geminal dicarbonyl ones. To investigate what effect the Ge presence has on this process, we performed a series of CO adsorption/desorption experiments with the blank 1Rh, as well as several bimetallic samples. The results, shown in Figure 3, are in agreement with the previous (26) Bowman, R.; Biloen, P. J. Catal. 1977, 48, 209. (27) Borgna, A.; Garetto, T. F.; Apesteguia, C. R. Appl. Catal. A 1999, 182, 189. (28) Cotton, F. A.; Wilkinson, G. Advanced Inorganic Chemistry, 3rd ed; Interscience Publishers: New York, 1972. (29) Solymosi, F.; Knozinger, H. J. Chem. Soc., Faraday Trans. 1990, 86, 389.

Figure 4. FTIR spectra of CO adsorbed at room temperature on (a) the fresh 1Rh5Ge sample and on the same sample after air exposure and re-reduction at either (b) 300 °C or (c) 500 °C.

literature reports and indicate that Rh0 crystallite growth takes place at the expense of the Rh1+ clusters in the case of the blank 1Rh sample (Figure 3a), as indicated by the decrease in the intensity of the peaks assigned to the geminal dicarbonyl species associated with Rh1+ sites. A similar resultsalthough less pronouncedswas also obtained with the 1Rh1Ge sample. In contrast, no such effect was observed in the case of the 1Rh5Ge sample (Figures 4b), suggesting that the presence of increased levels of germanium stabilizes rhodium in the form of small Rh1+ clusters. All experiments on bimetallic Rh-Ge catalysts were performed under inert He atmosphere to avoid contact with atmospheric oxygen and the subsequent oxidation of Ge. To examine the effect of air exposure on the surface properties of these catalysts, bimetallic samples were exposed to air and then re-reduced at 300 °C, as described in the Experimental Section. A significant increase was observed in the intensity of the linearly-bonded CO peaks relative to the intensities of the geminal and the bridgebonded ones, for all catalyst samples exposed to air, as shown in the examples of Figure 4. This effect was even more pronounced when a higher reduction temperature (500 °C) was used. In addition, the high-frequency peak at 2130 cm-1 disappeared completely from the spectra of the samples exposed to air, supporting the idea that this feature is indeed related to a charge transfer between Rh and Ge, and thus, is no longer present when Ge is oxidized to an inactive form. These results suggest that air exposure

10616

Langmuir, Vol. 20, No. 24, 2004

and re-reduction of the bimetallic samples leads to segregation of the two metals and formation of large Rh0 crystallites. Conclusions We have studied the effect of Ge addition via surface redox techniques on the surface properties of Rh supported on Al2O3. The analysis of FTIR spectra of adsorbed CO on monometallic Rh and bimetallic RhGe catalysts indicates that the presence of Ge leads to the creation of smaller rhodium clusters at the expense of larger Rh0 surfaces. Successive CO adsorption/desorption experiments suggest

Lafaye et al.

that the Ge presence has a stabilizing effect on these dispersed Rh1+ ensembles. Furthermore, a charge transfer is taking place between Rh and Ge, leading to the formation of rhodium ions in higher oxidation states, more noticeable at higher Ge loadings. Finally, exposure of the bimetallic Rh-Ge catalysts to air leads to an irreversible segregation of the two metals. Acknowledgment. Authors in the US express their gratitude to the U.S. Department of Energy (DE-FG0296ER14666) for financial support of this work. LA049692L