Alumina during Chiral

Jun 21, 2007 - Therefore, the adsorption strength of the different species is as follows: flat ≫ tilted > CDH6. Evidence for the formation of the fl...
2 downloads 0 Views 130KB Size
Download PDF
Langmuir 2007, 23, 8087-8093

8087

Surface Processes Occurring on Rh/Alumina during Chiral Modification by Cinchonidine: An ATR-IR Spectroscopy Study Erik Schmidt, Davide Ferri, and Alfons Baiker* Institute for Chemical and Bioengineering, ETH Zurich, Ho¨nggerberg, HCI, CH - 8093 Zurich, Switzerland ReceiVed March 22, 2007. In Final Form: May 18, 2007 Cinchona alkaloids are frequently used for chiral modification of supported noble metal catalysts employed in heterogeneous enantioselective hydrogenation. In order to gain molecular insight into the surface processes occurring at the metal/liquid interface, cinchonidine (CD) adsorption on vapor-deposited Rh/Al2O3 films has been studied in the presence of solvent and hydrogen by means of attenuated total reflection infrared (ATR-IR) spectroscopy. The spectrum of CD adsorbed on Rh exhibited two dominant signals at 1593 and 1511 cm-1, which are characteristic of a surface species having a quinoline ring tilted with respect to the metal. Interestingly, no adsorbed modifier in the flat geometry (quinoline parallel to the metal plane) was observed. During desorption, these signals vanished, and a new prominent signal appeared at 1601 cm-1, which belongs to a species with the quinoline ring hydrogenated on the heteroaromatic side. Concentration-dependent experiments and the reversibility of the observed phenomenon indicate that CD was readily hydrogenated to 1′,2′,3′,4′,10,11-hexahydrocinchonidine (CDH6) on Rh. The ATR-IR spectra also reveal that the flat species was indeed immediately hydrogenated when CD was provided from solution, and the only visible adsorbed species was the tilted species, which displaced the hydrogenation product from the metal surface. In the absence of dissolved CD, during desorption, the tilted species was converted to the flat species and rapidly hydrogenated. The hydrogenation product was stable on the metal surface only in the absence of CD. Therefore, the adsorption strength of the different species is as follows: flat . tilted > CDH6. Evidence for the formation of the flat species and its role as an intermediate to the hydrogenation product is given by an experiment in which CD was adsorbed in the absence of dissolved hydrogen after surface cleaning. The adsorption and hydrogenation of CD on Rh deviate significantly from that observed earlier on Pt and Pd under similar conditions, where the flat species could be observed even in the presence of hydrogen. This difference is attributed to the weaker interaction and lower hydrogenation rate occurring on Pt and Pd.

Introduction Asymmetric catalysis is a key technology to produce enantiomerically pure compounds that meet the stringent requirements of the pharmaceutical industry. Despite its intrinsic technical advantages in separation and handling,1 enantioselective heterogeneous catalysis with chirally modified supported metal nanoparticles has not achieved a breakthrough yet, the main reason being the limited number of potential systems2-6 affording high and better enantiomeric excess (ee) compared to traditional asymmetric homogeneous catalysts. The enantioselective hydrogenation of activated ketones over supported Pt catalysts modified by cinchona alkaloids7 (the so-called Orito reaction) is still one of the most efficient asymmetric heterogeneous catalytic systems. Besides platinum, rhodium is the only noble metal able to hydrogenate activated ketones with satisfactory ee. After initial attempts to hydrogenate methyl pyruvate with 5 wt % Rh/Al2O3,8 novel rhodium-based catalysts have recently been prepared9-11 and research has been focused on the broadening * Corresponding author. E-mail: [email protected]; fax: +41 44 632 11 63. (1) Baiker, A.; Blaser, H.-U. In Handbook of Heterogeneous Catalysis; (Ertl, G., Kno¨zinger, H., Weitkamp, J., Eds.; VCH: Weinheim, Germany, 1997; Vol. 5, pp. 2422 (2) Studer, M.; Blaser, H.-U.; Exner, C. AdV. Synth. Catal. 2003, 345, 45. (3) Bu¨rgi, T.; Baiker, A. Acc. Chem. Res. 2004, 37, 909. (4) Murzin, D. Y.; Ma¨ki-Arvela, P.; Toukoniitty, E.; Salmi, T. Catal. ReV. Sci. Eng. 2005, 47, 175. (5) Osawa, T.; Harada, T.; Takaysu, O. Curr. Org. Chem. 2006, 10, 1513. (6) Barto´k, M. Curr. Org. Chem. 2006, 10, 1533. (7) Orito, Y.; Imai, S.; Niwa, S. J. Chem. Soc. Jpn. 1979, 1118. (8) Blaser, H. U.; Jalett, H. P.; Monti, D. M.; Reber, J. F.; Wehrli, J. T. Stud. Surf. Sci. Catal. 1988, 41, 153.

of the scope of suitable substrates. Commercially available 5 wt % Rh/Al2O3 has been used with reasonable success to hydrogenate ketopantolactone,12 1-phenylpropane-1,2-dione,13 2-hydroxy-1(4-methoxy-phenyl)-ethanone,14 and 3,5-di-(trifluoromethyl)acetophenone.15 The partial success of rhodium catalysts is also due to the limited scope of chiral modifiers used so far compared to platinum catalysts. Recently, β-isocinchonine was found to produce a chirally modified Rh catalyst superior to that prepared using cinchonidine (CD) in the enantioselective hydrogenation of ketopantolactone, which did not require high modifier/substrate ratios.16 Understanding of the surface processes involved in the adsorption of a chiral modifier is a necessary prerequisite for the rational design of enantioselective metal catalysts. It is known that the cinchona alkaloid bestows local chirality to metal surfaces such as Pt, Pd, Rh and controls the orientation of the substrate, thereby generating enantioselective catalysts. Because of their complex molecular structure, the cinchona alkaloids exhibit a rich surface chemistry on metal surfaces. The typical molecular structure of cinchona alkaloids is reflected by that of CD (Scheme (9) Huang, Y. L.; Chen, J. R.; Chen, H.; Li, R. X.; Li, Y. Z.; Min, L. E.; Li, X. J. J. Mol. Catal. A: Chem. 2001, 170, 143. (10) Huang, Y. L.; Li, Y. Z.; Hu, J. Y.; Cheng, P. M.; Chen, H.; Li, R. X.; Li, X. J.; Yip, C. W.; Chan, A. S. C. J. Mol. Catal. A: Chem. 2002, 189, 219. (11) Ma, H. X.; Chen, H.; Zhang, Q.; Li, X. J. J. Mol. Catal. A: Chem. 2003, 196, 131. (12) Maris, M.; Mallat, T.; Baiker, A. J. Mol. Catal. A: Chem. 2005, 242, 151. (13) Toukoniitty, E.; Franceschini, S.; Vaccari, A.; Murzin, D. Y. Appl. Catal. A 2006, 300, 147. (14) Sonderegger, O.; Ho, G. M. W.; Bu¨rgi, T.; Baiker, A. J. Catal. 2005, 230, 499. (15) Hess, R.; Krumeich, F.; Mallat, T.; Baiker, A. J. Mol. Catal. A: Chem. 2004, 212, 205. (16) Hoxha, F.; Mallat, T.; Baiker, A. J. Catal. 2007, 248, 11.

10.1021/la700835j CCC: $37.00 © 2007 American Chemical Society Published on Web 06/21/2007

8088 Langmuir, Vol. 23, No. 15, 2007 Scheme 1. Molecular Structure of CD and Its Hydrogenation Reaction on a Rh-Based Catalyst

1). It includes a quinoline ring, a quinuclidine moiety bearing a vinyl group, and an OH functionality at the stereogenic C(9) atom, which, together with C(8), generates the chiral environment dictating the sense of enantiodifferentiation. This structure is highly complex also because of the conformational flexibility allowing for numerous conformers in solution and especially on a metal surface. Adsorption on noble metal surfaces is typically governed by the extended aromatic ring of quinoline. The interaction with the metal generates species different in the orientation of the ring with respect to the surface plane. Recent results indicate that the quinuclidine moiety is also involved in adsorption on Pt.17,18 At low coverage, the energetically favored geometry exhibits the aromatic ring parallel to the surface and the quinuclidine moiety oriented toward the metal surface in a geometry that has been named surface quinuclidine bound (SQB).19 The aromatic ring tilts away from the metal with increasing surface crowding, thus generating a so-called tilted species.19 Adsorbed species of cinchona alkaloids exhibit different infrared spectra from each other and from the free modifier, which makes vibrational spectroscopy a suitable method to investigate the adsorption of cinchona alkaloids on metal surfaces. Attenuated total reflection infrared (ATR-IR) spectroscopy,20-22 reflection absorption infrared spectroscopy,23 and Raman spectroscopy24 studies of CD adsorbed on polycrystalline Pt and Pd have been reported up to now, whereas adsorption on Rh surfaces has not yet been studied. These techniques have the advantage of enabling the investigation of the adsorption process at the solid-liquid interface and thus under conditions close to those encountered during catalytic reaction. Depending on the nature of the metal surface, the adsorbed molecules are able to resist, to some extent, high hydrogen pressure, an important feature for their use as chiral modifiers in enantioselective hydrogenation. Pt appears to be the metal on which the highest stability is achieved, which typically decreases with increasing hydrogen pressure. Pd and Rh show additional degradation processes.25 The strong interaction between the anchoring group of the chiral modifier adsorbing in a flat geometry and the metal catalyst allows for the generation of hydrogenation products where the ring is partially and totally hydrogenated. This process decreases the strength of this interaction and therefore the stability of the chiral agent, thus resulting in a diminished (17) Calvo, S. R.; LeBlanc, R. J.; Williams, C. T.; Balbuena, P. B. Surf. Sci. 2004, 563, 57. (18) Vargas, A.; Ferri, D.; Baiker, A. J. Catal. 2005, 236, 1. (19) Vargas, A.; Baiker, A. J. Catal. 2006, 239, 220. (20) Ferri, D.; Bu¨rgi, T. J. Am. Chem. Soc. 2001, 123, 12074. (21) Ferri, D.; Bu¨rgi, T.; Baiker, A. J. Catal. 2002, 210, 160. (22) Meier, D. M.; Mallat, T.; Ferri, D.; Baiker, A. J. Catal. 2006, 244, 260. (23) Kubota, J.; Zaera, F. J. Am. Chem. Soc. 2001, 123, 11115. (24) Chu, W.; LeBlanc, R. J.; Williams, C. T. Catal. Commun. 2002, 3, 547. (25) Bond, G.; Wells, P. B. J. Catal. 1994, 150, 329.

Schmidt et al.

enantioselective potential of the modified surface.26 The different chemistry toward ring hydrogenation observed for CD and its pseudoenantiomer cinchonine on Pt/Al2O327 suggests that different cinchona alkaloids might also possess different degrees of interaction with the surface and that the stability of the anchoring group can, in principle, be modulated to stabilize the chiral modifier. Moreover, hydrogenolysis of the quinuclidine moiety was also reported on Rh, and hydrogen/deuterium exchange experiments predicted that the chemistry of cinchona alkaloids should be even more complex on this metal than on Pt.25 All side reactions involving the chiral modifier obviously decrease the amount of modifier effectively taking part in the enantioselective hydrogenation. Therefore, on metals such as Rh, a high modifier/substrate ratio is often required to achieve high ee.12,14 On the other hand, the presence of a number of side reactions on Rh catalysts also indicates that the interaction of cinchona alkaloids with this metal is rather strong, and this has been confirmed by density functional theory (DFT) calculations, at least for the adsorption of quinoline.28 With the above information in mind, we have investigated in the present work the surface processes occurring upon the adsorption of CD on vapor-deposited Rh/Al2O3 films under conditions closely resembling those encountered during enantioselective hydrogenation using ATR-IR spectroscopy. The studies reveal significant differences in the adsorption and hydrogenation behavior of CD on Rh compared to that known for Pt and Pd. Experimental Materials. CD (Alfa Aesar, 99%) and 1,2,3,4-tetrahydroquinoline (Fluka, 95%) were used as received. The solvent dichloromethane (Baker) was stored over activated 5-Å molecular sieves. All gases were supplied by PANGAS (N2 99.995 vol %, H2 99.999 vol %, and 10.15 vol % CO/Ar). Rh wires (99.9%) and Al2O3 pellets (99.98%) used as targets for electron beam vapor deposition were supplied by Umicore. Film Preparation. The Rh/Al2O3 model films were produced on a trapezoidal Ge internal reflection element (IRE, 50 × 20 × 2 mm3, Komlas) by physical vapor deposition.29 The Ge IRE was cleaned before each film deposition by polishing with 0.25 µm diamond paste and rinsing with ethanol. A BAE-370 (Balzers) vacuum system was used for evaporation of the target materials from a graphite crucible (8 kV and up to 0.13 mA) at a maximum base pressure of 1 × 10-5 mbar. First, a 100 nm thick film of Al2O3 was deposited at a rate of 1.0 Å/s followed by 1 nm Rh without breaking the vacuum (0.5 Å/s). The film thickness was measured using a quartz crystal microbalance. Infrared Spectroscopy. ATR-IR spectra of the solid-liquid interface were recorded on a Bruker IFS-66 spectrometer equipped with a commercial ATR accessory (Optispec) and a liquid nitrogencooled MCT detector. The coated Ge IRE was mounted in a homemade stainless steel flow-through cell and maintained at 15 °C during the measurements. After cell mounting and optics alignment, the probe chamber was purged with nitrogen overnight in order to remove gas-phase water and carbon dioxide. Adsorption studies were carried out after flowing N2-saturated dichloromethane at 0.8 mL/min over the sample for about 90 min until stabilization of the signals was reached. Prior to adsorption, the Rh surface was cleaned by flowing H2-saturated dichloromethane for 10 min, which has been previously shown to be efficient for Pt20,29 and Pd21 thin films. After surface cleaning, a H2-saturated solution (10-4 M) of CD in dichloromethane was admitted to the (26) Morawsky, V.; Pru¨sse, U.; Witte, L.; Vorlop, K.-D. Catal. Commun. 2000, 1, 15. (27) Szo¨llo¨si, G.; Forgo, P.; Bartok, M. Chirality 2003, 15, S82. (28) Santarossa, G.; Vargas, A.; Baiker, A. Institute for Chemical and Bioengineering, ETH Zurich, Switzerland. To be submitted for publication. (29) Ferri, D.; Bu¨rgi, T.; Baiker, A. J. Phys. Chem. B 2001, 105, 3187.

Surface Processes of Chirally Modified Rh/Al2O3

Langmuir, Vol. 23, No. 15, 2007 8089

cell. The concentration dependence of the signals of adsorbed species was studied by admitting solutions of CD at an increasing concentration (10-6 to 5 × 10-4 M). CO adsorption was examined by circulating CO-saturated dichloromethane over the sample. For the competition experiments between CO and CD, a solution of the modifier (10-4 M) was saturated with CO and admitted to the cell.

Results and Discussion Film Characterization. The physical vapor-deposited Rh films were already characterized in some detail by X-ray photoelectron spectroscopy and CO adsorption from gas phase (5 bar) in a previous work.30 Both techniques indicated that Rh is in the metallic state. CO adsorption occurred with the evolution of two prominent signals at 2018 (on-top, COL) and 1880 cm-1 (bridge, COB). In the present contribution, CO adsorption was monitored from the liquid phase (CH2Cl2) to characterize the state of the surface prior to adsorption of the chiral modifier CD. Microscopic images of similarly prepared Pt,29 Pd,21 Au, and bimetallic31 films on Al2O3 demonstrated that the vapor-deposited films are composed of particles, and it is therefore expected that the present Rh films show comparable morphology. It is known from catalytic and spectroscopic measurements that the metal surface needs to be reduced or “cleaned” in order to observe the adsorption of cinchona modifiers. Cleaning is achieved in the spectroscopic cell upon admission of an H2saturated solvent at the working temperature (typically 10-20 °C) for ca. 10 min, and the effect of cleaning has been already described thoroughly.20,29 Owing to the use of dichloromethane, solvent decomposition products readily form on the noble metals as soon as these come in contact with hydrogen. The most prominent product on Rh was ethylidyne, which exhibited sharp signals at 1338 and 1120 cm-1 (≡C-CH3, not shown), in agreement with the formation of such species on Rh(111)32 and on Al2O3-supported Rh.33 The extent of solvent decomposition appeared to be less compared to that of Pt and Pd, but “rearrangement” of the water layer and removal of carbonates upon hydrogenation and CO formation were observed as well.29 The state of the surface was checked upon the adsorption of CO after different extents of H2 cleaning, that is, on an untreated Rh surface and on Rh treated for 2 and 10 min with hydrogen. The corresponding ATR-IR spectra are shown in Figure 1. The spectrum acquired after CO adsorption at saturation level on untreated Rh (Figure 1a) shows two signals of comparable intensity at 2037 and 1852 cm-1. As mentioned before, the two signals have been assigned to COL and COB adsorbed on metallic rhodium.34 No signals of CO bonded to oxidized rhodium species (characteristic doublet at ca. 2080 and 2020 cm-1) were found, indicating that, under these conditions, (i) CO adsorption occurs only on metallic rhodium particles and (ii) cationic Rh species are not stable (or do not exist). The surface obtained by the adsorption of CO on the untreated Rh was then treated with H2-saturated CH2Cl2 for ca. 10 min. Figure 1b demonstrates that the intensity of both CO signals was enhanced. The signal at 2037 cm-1 shifted to 2009 cm-1 and exhibited a 7-fold higher intensity than before hydrogen treatment, whereas the signal at 1852 cm-1 shifted to 1826 cm-1 with a 10-fold increase in intensity. Since care was taken that no additional CO could reach the metal surface in this time, this enhanced intensity is not the result of further CO adsorption from solution and does not (30) Burgener, M.; Ferri, D.; Grunwaldt, J. D.; Mallat, T.; Baiker, A. J. Phys. Chem. B 2005, 109, 16794. (31) Ferri, D.; Behzadi, B.; Kappenberger, P.; Hauert, R.; Ernst, K. H.; Baiker, A. Langmuir 2007, 23, 1203. (32) Koel, B. E.; Bent, B. E.; Somorjai, G. A. Surf. Sci. 1984, 146, 211. (33) Beebe, T. P.; Yates, J. T. J. Phys. Chem. 1987, 91, 254. (34) Yang, A. C.; Garland, C. W. J. Phys. Chem. 1957, 61, 1504.

Figure 1. ATR-IR spectra of CO adsorption from CH2Cl2 on Rh/ Al2O3 thin films after different pretreatments: (a) CO on the asprepared rhodium surface (scaled 10:1); (b) after purging the surface obtained in spectrum a with H2-saturated solvent for 10 min; (c) CO on a rhodium surface pretreated with H2-saturated solvent for 2 min; (d) CO on a rhodium surface pretreated with H2-saturated solvent for 10 min.

originate from decomposition products formed upon the admission of hydrogen. As suggested for other similar metal films,21,29 a surface enhancement effect is more appropriate to describe the phenomenon.35 The observed red-shift of both signals is primarily an electronic effect due to the enhanced back-donation of rhodium into the antibonding π*-orbital of CO; it also reflects the different environment within which the CO domains were confined before and after H2 treatment. Adsorption of CO on a rhodium surface pretreated for 2 min with hydrogen resulted in the spectrum shown in Figure 1c. The CO signals appeared at higher frequencies (2040 and 1886 cm-1) and displayed higher intensities compared to spectrum b because of the higher CO coverage achieved after hydrogen treatment. The reversed ratio COL/COB in spectra b and c likely reveals the efficiency of the hydrogen treatment at this temperature, with 2 min being enough to create large Rh domains that resemble extended Rh surfaces.34 A longer treatment with hydrogen (10 min, Figure 1d) obviously improved the extent of surface cleaning and therefore increased the CO coverage. The apparent red-shift of the COL signal (∆ν ) 7 cm-1) can be attributed to the appearance of the negative pole on the high-energy side of the signal and to the corresponding change of the optical properties of the Rh surface after being covered by a dense CO layer.36 Adsorption of CD. The hydrogen pretreatment is a prerequisite to follow the adsorption of CD on a noble metal surface. Under the present experimental conditions (i.e., CH2Cl2 solvent and dissolved hydrogen), the adsorption of this chiral modifier is typically competitive with the fragments generated by decomposition of the CH2Cl2 solvent upon contact with the reduced surface. Besides the signals of adsorbed CD, negative signals at ca. 1400 and 1338 cm-1 (for the spectral range of interest) are (35) Osawa, M. Top. Appl. Phys. 2001, 81, 163. (36) Bu¨rgi, T. Phys. Chem. Chem. Phys. 2001, 3, 2124.

8090 Langmuir, Vol. 23, No. 15, 2007

Schmidt et al. Table 1. Assignment of Selected Vibrational Modes of CD Adsorbed on Pt, Pd, and Rh Surfaces and of Adsorbed Speciesa frequency (cm-1)

assignment mode δ(C-H), quinuclidine δ(C-H), quinuclidine ring def., quinoline (phenyl) ring def., quinoline ring def., quinoline ring def., quinoline (pyridine) ring def., quinoline ν(CdC), vinyle

speciesb

sol.20

Pt20

Pd21

Rhc

all all tilted

1454 sh. 1509

1458 sh. 1511

1456 sh. 1509

1460 1480 1511

R-quinolyl flatd tilted

n.o. 1570 1593

1530 1570 1590

n.o. 1567 1586

n.o. 1572 1593

tilted

1615 1635

1610 n.o.

1612 n.o.

1615 n.o.

a Def.) deformation; n.o.) not observed; sh.) shoulder. b The most representative species for each signal. c This work. d This signal is, in principle, common to all species, but it is the major signal on Pt because of the presence of the flat species. e The vinyl group is immediately hydrogenated on these surfaces.

Figure 2. ATR-IR spectra of (a) CD on Rh/Al2O3 and (b,c) after rinsing with H2-saturated solvent. Spectrum d is the transmission spectrum of 1,2,3,4-tetrahydroquinoline (scaled 1:500).

typically observed in the ATR-IR spectra, which belong to the same species formed during the cleaning treatment, that is, CHx (possibly containing Cl atoms) and ethylidyne species. The adsorption of CD has been extensively studied on various metal surfaces,20,21,23,24,37,38 and the observed spectral features are attributed to the population of specific adsorption geometries on the basis of the assignment of selected vibrational modes. Additionally, DFT calculations provide a picture of these complex structures on the metal surface that matches the species observed spectroscopically.19 The adsorption of CD from a 0.1 mM solution in CH2Cl2 on a Rh/Al2O3 thin film resulted in the spectra shown in Figure 2. Two prominent signals quickly developed with time at 1511 and 1593 cm-1, and reached saturation within ca. 10 min. These signals are associated with aromatic ring vibrations of the quinoline ring of CD, and quantum chemical calculations indicate that the signal at high frequency has to be associated with a ring stretch of the heteroatom-containing ring, whereas that at low frequency is predominantly a stretch of the benzene ring of quinoline.20,37 Detection of these two signals implies observation of a local adsorption geometry in which the quinoline ring is tilted with respect to the Rh surface and therefore interacts weakly with it. This can be deduced by comparison with the spectrum of CD in solution which much resembles spectrum a. The weaker signals at 1615 and 1572 cm-1 (aromatic ring vibrations) and at 1460 cm-1 (quinuclidine CH2-deformation) are also associated with vibrations of all known adsorption geometries, which are summarized in Table 1 and compared with the adsorption of CD on similar noble metal films. The additional signal at approximately 1480 cm-1 belongs to a deformation mode of the methylene groups of quinuclidine, which, for example, on Pt (37) Chu, W.; LeBlanc, R. J.; Williams, C. T.; Kubota, J.; Zaera, F. J. Phys. Chem. B 2003, 107, 14365. (38) Behzadi, B.; Vargas, A.; Ferri, D.; Ernst, K. H.; Baiker, A. J. Phys. Chem. B 2006, 110, 17082.

appears only as a weak shoulder on the high-energy side of the signal at ca. 1460 cm-1.20 The better separation between the two signals is interpreted as the result of the weaker amplitude of the latter signal on Rh compared to that on Pt and Pd, as it will become clear in the following. In contrast to adsorption on platinum, no R-H-abstraction and therefore formation of the R-quinolyl-species (1530 cm-1) was observed (Figure 2 and Table 1), although this species has been observed on Rh under different conditions before.39 This indicates that this species is not stable under the present experimental conditions and already gives an indication of the reactivity of the chiral modifier on the surface. More importantly, only little if any adsorption of the quinoline moiety in the flat geometry was detected under the conditions employed, as it can be concluded from the very low intensity of the signal at 1572 cm-1. This signal is typically the most prominent for identical adsorption conditions on Pt films20 and on other polycrystalline Pt surfaces,23,24 and the assignment agrees with the particular orientation of the quinoline ring in the flat geometry. The considerably reduced intensity of the signal associated with the flat adsorbed species of CD indicates that the adsorption and/or hydrogenation of CD on Rh differs significantly from that on Pt. This is revealed by the spectra recorded during rinsing the Rh surface covered by CD with H2-saturated CH2Cl2. Figure 2b shows that the intensity of both signals at 1593 and 1511 cm-1 quickly decreased concomitant with the disappearance of the weaker signals at 1615 and 1572 cm-1 and the attenuation of those at 1480 and 1460 cm-1. The signal at 1593 cm-1 initially broadened and, after 9 min of rinsing, disappeared (as the signal at 1511 cm-1), resulting in a new signal at 1601 cm-1. Two additional signals also appeared at 1515 and 1498 cm-1. The fast change in the spectra and the appearance of a new surface species upon the admission of the H2-saturated solvent strongly suggest that hydrogenation of the quinoline ring of CD occurred under these conditions. Comparison with the transmission spectrum of 1,2,3,4-tetrahydroquinoline (Figure 2d) confirmed this interpretation and therefore the hydrogenation of the heteroaromatic part of the quinoline ring with the formation of 1′,2′,3′,4′tetrahydrocinchonidine. The signal at 1601 cm-1 corresponds to the quadrant stretch (ν(C-C) and δ(C-H)) of the benzene ring of the hydrogenation product. The signals at around 1500 cm-1 include the semicircle stretch and the deformation of the CH2 groups of the aliphatic ring. The weaker signal at 1460 cm-1 (39) Mate, C. M.; Somorjai, G. A.; Tom, H. W. K.; Zhu, X. D.; Shen, Y. R. J. Chem. Phys. 1988, 88, 441.

Surface Processes of Chirally Modified Rh/Al2O3

likely belongs to the quinuclidine moiety of the new species. Similarly to Pt and Pd vapor-deposited films, the vinyl group of CD is also readily hydrogenated on Rh, as the absence of the signal at 1635 cm-1 suggests. Therefore, 1′,2′,3′,4′,10,11hexahydrocinchonidine (CDH6) is produced (Scheme 1). This species did not disappear over the next 30 min while flowing H2-saturated solvent and was therefore stable on Rh. Since the quinoline ring is the anchoring group of CD in all adsorbed species, its orientation (flat, tilted) is a crucial factor in determining the adsorption strength and the stability of the species on the surface. The hydrogenation of the ring containing the heteroatom becomes feasible when the ring is (nearly) parallel to the metal surface. Hence, CDH6 should originate from the flat adsorbed species of CD, which is, however, not so clearly observed on Rh as it is on Pt. Hydrogenation of the tilted species occurs in a second phase because of the inappropriate orientation of the quinoline ring. This indicates that the tilted species may act as a reservoir of flat species in the absence of free CD. The absence of the signal at ca. 1570 cm-1 in the spectra suggests that the heteroaromatic ring of the flat adsorbed CD is continuously hydrogenated, affording CDH6. However, Figure 2 also suggests that, during contact of the solution of CD with the surface, the hydrogenation product was immediately displaced from the surface, and only the signals of tilted adsorbed CD are observed. The hydrogenation product was enriched on the surface during rinsing; both tilted CD and CDH6 coexisted on the surface shortly after changing to H2-saturated solvent, resulting in one broad signal at 1598 cm-1 (Figure 2b). Longer desorption time resulted in the observation of only CDH6, which was stable in the absence of (adsorbed and dissolved) CD. In essence, these observations reveal that the tilted species of CD is more strongly adsorbed than CDH6 and that it is also hydrogenated to the same product during rinsing with solvent rather than simply desorbed. The competition between CD and CDH6 for adsorption sites was studied by continuously increasing the concentration of CD in the solution flowing over the Rh/Al2O3 film. The ATR-IR spectra recorded under steady-state conditions for each selected concentration are shown in Figure 3. At low concentrations (10-6 to 2 × 10-6 M), the most prominent signals are located at 1601 cm-1 and approximately 1504 and 1460 cm-1. Therefore, CDH6 dominated in the concentration range in which flat adsorbed CD is the dominant species on Pt.20 Upon increase of the concentration to 4 × 10-6 M, the signals of adsorbed CD clearly appeared at the expense of those of the hydrogenation product. Figure 3d shows that, at a concentration of 7 × 10-6 M, the signal at 1572 cm-1 became as intense as that at 1593 cm-1, whereas the latter and the signal at 1511 cm-1 increased to a larger extent with increasing concentration. Therefore, there is a concentration regime (ca. 2 × 10-6 to 10-5 M) where flat CD may also be observed. For concentrations larger than 10-5 M, tilted CD dominated the infrared spectra on Rh. These data reveal that two superimposed equilibria between flat CD and CDH6 (fast hydrogenation of the quinoline ring) and between tilted CD and CDH6 exist on Rh. The former originates from the fast hydrogenation of flat CD, which is likely very strongly interacting with Rh, but is not visible in the ATR-IR spectra because of the fast transformation. The second equilibrium intervenes when enough CD is available from solution. Then CD can display both flat and tilted geometries: the flat species is readily hydrogenated, whereas tilted CD displaces the hydrogenation product and adsorbs on Rh. A qualitative estimation of the Gibbs free enthalpy of adsorption can be derived from the concentration-dependent experiment. Assuming that adsorption follows a Langmuir isotherm and using

Langmuir, Vol. 23, No. 15, 2007 8091

Figure 3. ATR-IR spectra of CD on Rh/Al2O3 at increasing solution concentration: (a) 10-6 M; (b) 2 × 10-6 M; (c) 4 × 10-6 M; (d) 7 × 10-6 M; (e) 10-5 M; (f) 2 × 10-5 M; (g) 5 × 10-5 M; (h) 10-4 M; (i) 5 × 10-4 M. The top spectrum (j) is a transmission spectrum of CD in dichloromethane solution.

Figure 4. Time-dependent behavior of the integrated absorbance of the signals at 1511 cm-1 (0, CD) and 1601 cm-1 (9, CDH6) from ATR-IR spectra collected when alternating the feed to the cell between a solution of CD (10-4 M) in CH2Cl2/H2 for 10 min and CH2Cl2/H2 for 10 min for a total of three cycles.

the signal at 1460 cm-1 for the estimation because it represents all adsorbed species,20,21 an adsorption constant (K) of ≈ 2.49 × 105 and a Gibbs free enthalpy of 29.8 kJ/mol have been calculated, which is slightly lower than the value found for Pt.20 This value is significantly underestimated since it characterizes only tilted adsorbed CD and does not correlate with the high adsorption energy calculated for quinoline adsorbed on a Rh cluster in the vacuum.28 Moreover, unlike that on Pt and Pd, the value is substantially affected by the hydrogenation of the chiral modifier. Figure 4 displays the time-dependent behavior of the integrated absorbance of the signals at 1511 and 1601 cm-1 in an experiment where the CD solution and H2-saturated solvent were alternately

8092 Langmuir, Vol. 23, No. 15, 2007

admitted to the cell. The experiment is basically a replica of that shown in Figure 2, simply repeated three times on the same Rh surface. The signal related to CD (1511 cm-1) reached saturation approximately after ca. 5 min and vanished almost completely within 10 min after switching to H2-saturated solvent, whereas the signal at 1601 cm-1 exhibited the opposite behavior. Interestingly, the latter signal vanished much faster (within 2 min) when the solution of CD had been admitted again to the cell, compared to the adsorption-desorption experiment shown in Figure 2. This behavior is best seen in Figure 4 during the last cycle where H2-saturated solvent was fed to the cell for ca. 30 min after CD adsorption and clearly indicates that CDH6 remained adsorbed on Rh only if no CD was provided from solution. Also, the signals at 1480 and 1460 cm-1 (not shown) were attenuated during the solvent flow but did not vanish completely. Because of their nature (i.e., deformation modes of the methylene groups of the quinuclidine moiety), both signals are common to all adsorbed species of CD and to the hydrogenation product as well. The preferred hydrogenation of the heteroaromatic part of the quinoline ring is in good agreement with catalytic investigations of the hydrogenation of quinoline over Rh-based catalysts. Hydrogenation of the heteroaromatic ring is reported to be favored in cinchona alkaloids, and saturation of the quinoline ring is typically more difficult.40 CDH6 obtained from the hydrogenation of CD on Pt/Al2O3 in H2SO4 solution at high H2 pressure was characterized recently using NMR spectroscopy.27,41 Hydrogenation of the quinoline ring as a competitive reaction to enantioselective hydrogenations over cinchona-modified noble metals has been followed using UV spectroscopy for supported Rh,14 Pd,42 and Pt43 catalysts, with the order of extent of quinoline degradation being Rh > Pd > Pt. Therefore, higher chiral modifier/substrate ratios are typically required for hydrogenations on Rh and Pd to achieve high enantioselectivity. Comparing the data collected on vapor-deposited Rh to those of investigations on similar Pd films,21 it emerges that the extent of hydrogenation on Rh is larger under the applied conditions, indicating that the hydrogenation rate on Pd (and Pt) is lower than that on Rh. The observation that the tilted species of CD dominates in the infrared spectra on the Rh surface and that the flat species is quickly hydrogenated at the anchoring ring well agrees with various catalytic studies of enantioselective hydrogenations on supported Rh particles. Considering that the surface species responsible for inducing enantiodifferentiation is that exhibiting a flat geometry,3 the hydrogenation of this species on the vapordeposited polycrystalline Rh is a crucial factor explaining why supported Rh catalysts show inferior performance in enantioselective hydrogenations compared to supported Pt catalysts. Interestingly, as mentioned above, theoretical calculations indicate a stronger interaction of the quinoline ring of CD with Rh than with Pt. Thus the flat geometry is in principle favored on Rh, but, in the presence of hydrogen, it is rapidly hydrogenated, which explains why it could not be detected under these conditions. The existence of the flat species of CD on Rh was verified in an experiment in which CD adsorption on Rh was followed in the absence of H2 under otherwise identical conditions to those used in Figure 2. This experiment is shown in Figure 5b, (40) Campanati, M.; Vaccari, A.; Piccolo, O. J. Mol. Catal. A: Chem. 2002, 179, 287. (41) Szo¨llo¨si, G.; Chatterjee, A.; Forgo´, P.; Barto´k, M.; Mizukami, F. J. Phys. Chem. A 2005, 109, 860. (42) Maris, M.; Huck, W.-R.; Mallat, T.; Baiker, A. J. Catal. 2003, 219, 52. (43) Hess, R.; Vargas, A.; Mallat, T.; Bu¨rgi, T.; Baiker, A. J. Catal. 2004, 222, 117.

Schmidt et al.

Figure 5. ATR-IR spectra of CD on Rh/Al2O3 adsorbed from (a) H2- and (b) N2-saturated solution. In both experiments, the surface was contacted with H2-saturated solvent (10 min) prior to adsorption of the chiral modifier.

which is the spectrum collected after admission of the solution of CD saturated with nitrogen following the surface cleaning by hydrogen. As discussed above, in the presence of hydrogen, the signals of tilted CD (1593 and 1511 cm-1) dominate the spectrum (Figure 5a). In the absence of hydrogen in the solution of CD (Figure 5b), these signals were significantly weaker, and the signal at 1572 cm-1 was strongly enhanced. The signal reached saturation after about 40 min on stream, which is more slowly than that on Pt likely because of the consumption of adsorbed hydrogen in the early stages of CD adsorption. Note that the spectrum in Figure 5b is remarkably similar to that obtained upon adsorption of CD on Pt, on which the signal of the flat adsorbed species dominates at this concentration. Additionally, the intensity of the signal at ca. 1460 cm-1 also became more important than that in the presence of hydrogen, probably as a result of the (better) interaction of the quinuclidine N atom with the metal surface in the SQB geometry.18 These data also demonstrate further detail of the complex infrared spectra of CD adsorbed on metals. The signal associated with the deformation of the methylene groups of the quinuclidine moiety (typically at ca. 1460 cm-1) is composed of the contributions of this moiety in the flat and tilted species because of its interaction with the metal surface.19 This confirms that the Open(3) conformer of CD, the most stable conformer in vacuum and in apolar solvents,44 is not the most appropriate model to describe the adsorption of this chiral modifier on a metal surface. Hence, the flat adsorbed species is favored on rhodium, as predicted by the theoretical calculations.28 However, it is more quickly hydrogenated than the tilted species. These results reveal a more complex interplay of surface processes (adsorption, reaction) than that described for Pt/Al2O3.20 This interplay is schematically represented in Figure 6. The adsorption of CD on Rh generates both flat and tilted species. Hydrogenation of the quinoline ring occurs quickly and predominantly on the flat species (44) Bu¨rgi, T.; Baiker, A. J. Am. Chem. Soc. 1998, 120, 12920.

Surface Processes of Chirally Modified Rh/Al2O3

Langmuir, Vol. 23, No. 15, 2007 8093

modifiers for which hydrogenation of the anchoring group is hindered to some extent should be used in order to achieve high enantioselectivity with Rh-based catalysts.

Conclusions

Figure 6. Schematic presentation of surface processes occurring upon CD adsorption from H2-saturated solution on Rh/alumina films. The hydrogenation of the flat species affording adsorbed CDH6 is very fast, affecting the equilibrium between tilted and flat species. The hydrogenated product desorbs (is replaced) upon adsorption of CD from solution. Note that, in the absence of free CD in solution, the desorption of CDH6 (dashed arrow) is suppressed. Flat adsorbed CD is only detectable by IR in the absence of hydrogen.

and generates CDH6. Tilted species compete for adsorption with this product and displace it from the metal surface. In the absence of dissolved CD, tilted species convert into flat species and are in turn hydrogenated. The product of hydrogenation of the quinoline ring preferably adsorbs with the phenyl ring tilted with respect to the surface plane, which could be the reason for the much slower hydrogenation of this part of the quinoline ring and for the stability of the species on Rh against desorption. It should be remembered that the quinuclidine moiety can stabilize the molecule by interaction with the surface through the N atom. The experiments shown above allow the assessment of the relative adsorption strength as CDflat . CDtilted > CDH6. The surface processes of CD observed on the Rh surface obviously have striking implications for the enantioselective hydrogenations on Rh/Al2O3 catalysts. The fast desorption of the hydrogenation product of CD implies that, during enantioselective hydrogenation, the surface concentration of CD is restored upon adsorption from solution and that a fraction of newly adsorbed modifier is again consumed by side reaction(s). Therefore, higher concentrations of modifier are required that exceed the few micromoles of CD needed to induce maximum enantioselectivity on Pt/Al2O3. It becomes obvious that chiral

The surface processes occurring upon the adsorption of CD on Rh/Al2O3 have been investigated using ATR-IR spectroscopy. The adsorption of CD appears more complex than that observed on Pt and Pd, and new aspects on the interaction between metal and chiral modifier have been uncovered. On Rh, hydrogenation of the heteroaromatic part of the quinoline ring readily occurs, whereas a similar behavior was not observed on Pt and Pd using vibrational spectroscopy. Adsorbed CD in the flat geometry is the intermediate of this reaction, whereas tilted CD reacts only in the absence of dissolved modifier and needs to be converted into the flat species. The flat species was observed on Rh when adsorption was performed in the absence of dissolved hydrogen, and corroborates the strong interaction of CD with Rh. This interaction is responsible for the quick hydrogenation of the quinoline ring and does not allow detection of the flat species in the presence of dissolved hydrogen. This contrasts the behavior observed on Pt and Pd reported in the literature where flat species were observed even in the presence of hydrogen. This phenomenon is attributed to the much lower rate of quinoline ring hydrogenation on these two metals. CD is more strongly adsorbed than its hydrogenation product: the latter is only stable on Rh in the absence of the intact chiral modifier. These data demonstrate that hydrogenation of the quinoline ring destabilizes CD and reduces with time the fraction of active species required to induce enantioselectivity. This species can be provided from solution, but higher concentrations of modifier are required compared to Pt and Pd where the ring hydrogenation is slower. Acknowledgment. Financial support by the Swiss National Science Foundation and the Foundation Claude and Giuliana is kindly acknowledged. LA700835J