Chiral Modification of Platinum by CoAdsorbed ...

5 downloads 311 Views 2MB Size Report
Dec 30, 2013 - phase-domain (f=0!3608), by using PSD provides further sensitivity ...... served reduced enantiocontrol with increasing conversion.[23c].
DOI: 10.1002/chem.201303261

Full Paper

& Asymmetric Catalysis

Chiral Modification of Platinum by Co-Adsorbed Cinchonidine and Trifluoroacetic Acid: Origin of Enhanced Stereocontrol in the Hydrogenation of Trifluoroacetophenone Fabian Meemken,[a] Alfons Baiker,*[a, b] Sebastian Schenker,[a] and Konrad Hungerbhler[a]

Abstract: Cinchonidine (CD) adsorbed onto a platinum metal catalyst leads to rate acceleration and induces strong stereocontrol in the asymmetric hydrogenation of trifluoroacetophenone. Addition of catalytic amounts of trifluoroacetic acid (TFA) significantly enhances the enantiomeric excess from 50 to 92 %. The origin of the enantioselectivity bestowed by co-adsorbed CD and TFA is investigated by using in situ attenuated total reflection infrared spectroscopy and modulation excitation spectroscopy. Molecular interactions between the chiral modifier (CD), acid additive (TFA) and the trifluoro-activated substrate at the solid–liquid interface are elucidated under conditions relevant to catalytic hydrogenations, that is, on a technical Pt/Al2O3 catalyst in the presence of H2 and solvent. Monitoring of the unmodified and modified surface during the hydrogenation provides an insight into the phenomenon of rate enhancement and the crucial

Introduction Optically active, fluorine-containing alcohols are widely applied as chiral building blocks in material and biological sciences and their enantiopure production is of major importance for the fine chemical and pharmaceutical industries.[1] An elegant and sustainable strategy to produce enantiomers in high purity is asymmetric catalysis[2] because an ideal stereochemically defined catalyst, inducing chiral amplification, can lead to the formation of only the desired optical isomer. Although homogeneous asymmetric catalysis with transition-metal complexes is industrially well-established due to its higher versatility and efficiency,[3] heterogeneous catalytic systems would offer important technical benefits connected with catalyst separation, regeneration and recycling, and also facilitate transfer

[a] F. Meemken, Prof. Dr. A. Baiker, Dr. S. Schenker, Prof. Dr. K. Hungerbhler Department of Chemistry and Applied Biosciences Institute for Chemical and Bioengineering ETH Zrich, Hçnggerberg, HCI, 8093 Zrich (Switzerland) Fax: (+ 41) 44-632-11-63 E-mail: [email protected] [b] Prof. Dr. A. Baiker Chemistry Department, Faculty of Science King Abdulaziz University, P.O. Box 80203, Jeddah 21589 (Saudi Arabia) Supporting information for this article is available on the WWW under http://dx.doi.org/10.1002/chem.201303261. Chem. Eur. J. 2014, 20, 1298 – 1309

interactions of CD with the ketone, corresponding product alcohol, and TFA. Comparison of the diastereomeric interactions occurring on the modified surface and in the liquid solution shows a striking difference for the chiral preferences of CD. The spectroscopic data, in combination with calculations of molecular structures and energies, sheds light on the reaction mechanism of the heterogeneous asymmetric hydrogenation of trifluoromethyl ketones and the involvement of TFA in the diastereomeric intermediate surface complex: the quinuclidine N atom of the adsorbed CD forms an NHO-type hydrogen-bonding interaction not only with the trifluoro-activated ketone but also with the corresponding alcohol and the acid additive. Strong evidence is provided that it is a monodentate acid/base adduct in which the carboxylate of TFA resides at the quinuclidine N-atom of CD, which imparts a better stereochemical control.

to continuous process operations.[4] However, the inherent heterogeneity and complexity of chiral surfaces make research in this field a challenging task. The limited understanding of the underlying reaction mechanisms has considerably hampered the development of an efficient heterogeneous asymmetric catalysis.[5] Although fundamental studies on chiral single-crystal surfaces at vacuum conditions[6] have provided valuable information about the properties of chiral surfaces on an atomic and molecular scale, gaining comparable information on technical catalysts under reaction conditions is still very challenging.[7] However, a proper understanding of catalytic processes at the chiral solid–liquid interphase is crucial for a rational design of heterogeneous asymmetric catalysts. Here, we made an attempt to gain fundamental insight into the mechanism of an asymmetric hydrogenation reaction catalyzed by a chirally modified technical Pt catalyst under the reaction conditions. Among the few strategies for bestowing chirality to an intrinsically achiral catalytic surface,[8] chiral modification of Pt group metal catalysts by cinchona alkaloids is still considered to be one of the most promising.[9] In the so-called Orito reaction, simple addition of small amounts of cinchonidine (CD) to the reaction mixture results in strong stereocontrol with an enantiomeric excess (ee) of up to 98 % in the hydrogenation of a-keto esters on supported Pt.[10] However, the efficiency of the asymmetric hydrogenation depends on various factors,

1298

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper such as structure of the reactant ketone and the chiral modifier, solvent, temperature, hydrogen partial pressure, and catalyst properties.[8a, e, 11] Whereas considerable knowledge has been gathered about the hydrogenation of keto esters, relatively little is known about the reaction of other activated ketones, such as trifluoro-activated ketones.[12] General mechanistic models reported in the literature[11, 13–16] widely agree upon the decisive role of a 1:1 modifier–reactant ketone complex in the enantiodifferentiating step occurring on the noble metal surface. Furthermore, three structural requirements of the chiral modifier have been identified that lead to an efficient chiral modification of metal catalysts: an aromatic ring system responsible for anchoring the modifier to the metal surface,[11, 13] the absolute configuration of the chiral linker at C8’ C9’ defining the asymmetric site,[17] and the tertiary amine functionality as the stereodirecting driving force.[18, 19] Mechanistic similarities for the enantioinduction on the metal surface in the hydrogenation of a-keto esters and trifluoromethyl ketones can be expected. Involvement of the quinuclidine nitrogen in the stereoselective surface complex was derived from the experimental observation that blocking of the basic amine function of the modifier leads to a complete loss in ee for ketones, either activated by a-ester-group or atrifluoromethyl-group.[20] However, there are also indications of different reaction mechanisms, depending on the specific substrate classes.[21] The experimental findings from batch hydrogenations shown in Scheme 1 points out some unique characteristics of the asymmetric hydrogenation of trifluoro-activated

Scheme 1. Influence of modifier, hydrogen pressure, and addition of trifluoroacetic acid (TFA) on ee and turnover frequency (TOF) in the hydrogenation of TFAP. [a] From Ref. [12b]: 42 mg cat., CD (0.68 mmol), TFAP (1.85 mmol), toluene (5 mL), 10 bar H2, RT. [b] From Ref. [22]: TFAP (1.78 mmol), 273 K, otherwise identical conditions. [c] At 1 bar H2. Chem. Eur. J. 2014, 20, 1298 – 1309

www.chemeurj.org

ketones, for example, a,a,a-trifluoroacetophenone (TFAP) to (R)-a,a,a-(trifluoromethyl)benzyl alcohol ((R)-TFBA) on chirally modified Pt/Al2O3. In contrast to the minor role assigned to the OH function of the modifier in the hydrogenation of a-keto esters, the Omethyl ether of CD (MeOCD) was shown to be an ineffective modifier for the asymmetric hydrogenation of several different aryl-substituted a,a,a-trifluoroacetophenones (0–4 % ee (S), the opposite enantiomer).[22] Regarding the hydrogen surface coverage, unlike for most of the other activated ketones increased H2-pressure leads to a decline of enantioselection.[12b] In addition to the general sensitivity of the catalytic system regarding reaction conditions, not only substrate concentration strongly influences the enantiodiscrimination[23] but also the acidic product alcohol is detrimental for the stereoselective reaction route.[23c] Interestingly, addition of a small amount of trifluoroacetic acid (TFA) to the reaction mixture was found to greatly enhance the stereocontrol on the metal surface in the hydrogenation of trifluoro-activated ketones.[12a, 24] Whereas acidity and solvent composition are also known to influence the conformation of the modifier and its interaction with the substrate in liquid phase,[25] dependence of stereocontrol on the solvent[8c, 26] deviates strongly for the various substrate classes. Acetic acid or toluene/acetic acid mixtures generally yield optimal ee values for keto esters and the positive effect of carboxylic acid is mostly explained by protonation of the quinuclidine N-atom resulting in an improved NHO-type hydrogen-bonding interaction between the co-adsorbed modifier and ketone.[24, 27] On the contrary, in acetic acid the ee value drops dramatically for trifluoromethyl ketones and the best results are achieved in dichlorobenzene or toluene.[28] By using these solvents, optimal enhancement is found with 0.1 vol % TFA (1–3 mol TFA/mol CD)[12b, 22] and enantioselectivity is almost completely lost with 2 vol % TFA.[29] Supported by IR evidence, a three-step reaction mechanism was postulated in which an intermediate complex of modifier–TFA–ketone is formed.[29, 30] Because only catalytic amounts of the strong acid TFA yield the positive effect, such a remarkable enhancement cannot be attributed alone to protonation of the amine function for this substrate class and blockage of racemic sites by acid molecules. The latter has been postulated by McBreen and co-workers[31] based on scanning tunneling microscopy (STM) investigations of the co-adsorption of TFAP and TFA under ultrahigh vacuum conditions, which showed that TFA disrupts aryl CH···O bonding through insertion into TFAP dimers to form isolated trimolecular and bimolecular assemblies. Specifically, they proposed that the acid additive operates in the enantioselective hydrogenation of TFAP through the inhibition of the racemic reaction at sites remote from the chiral modifier. Enantiodiscriminating complexes including TFA have been proposed based on experimental results of catalytic hydrogenations of ethyl-4,4,4-trifluoroaceto acetate and ex situ-IR investigations in dichloromethane (CH2Cl2).[29] However, the role of the catalytic surface was neglected in the liquid phase studies and CH2Cl2 is not a suitable solvent for hydrogenations on

1299

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper Pt due to possible HCl formation by hydrogenolysis and concomitant catalyst poisoning.[32] Missing in situ spectroscopic studies of the asymmetric hydrogenation of trifluoro-activated ketones and the lack of information about the role of TFA in the enantiodiscriminating step prompted us to investigate this catalytic process by using in situ attenuated total reflection infrared (ATR-IR) spectroscopy in combination with modulation excitation spectroscopy (MES) and phase-sensitive detection (PSD). We investigated the dynamic interactions occurring on the surface of the cinchonamodified Pt/Al2O3 catalyst under reaction conditions, that is, in the presence of H2-saturated toluene and the reactant trifluoroacetophenone. Special attention was given to the elucidation of the beneficial action of TFA in the enantiodifferentiating process. A surprisingly different behavior is discovered for the interactions of the diastereomer CD in solution and in its adsorbed state on the metal surface. Analysis of the surface spectra supported by liquid phase measurements and quantum chemical calculations shed light on the modifier–trifluoro compound interactions and active enantiodifferentiating species on the platinum surface.

Results and Discussion Interaction between CD and trifluoro compounds at the solid–liquid interface The strong effects of solvent[8c, 26] and hydrogen pressure[12b] on the stereocontrol in the asymmetric hydrogenation on cinchona-modified Pt, stress the necessity of in situ surface studies for gaining a deeper understanding of these catalytic systems. Particularly for the hydrogenation of TFAP, a change of solvent can even lead to the formation of the opposite enantiomer,[38] for example, a reaction using CD and toluene yields the (R)enantiomer with a very good ee value, whereas in alcoholic solvents the (S)-enantiomer is favored.[23c, 39] Therefore, we chose H2-saturated toluene as reaction medium to avoid additional H-bond-donor or -acceptor abilities from the weakly polar, aprotic solvent. A suitable method to study the adsorption/desorption process of substrate and product molecules on the catalyst under conditions relevant to heterogeneous asymmetric hydrogenations is ATR-IR spectroscopy combined with MES.[40] The latter spectroscopic technique analyses the response of a system on an externally triggered perturbation caused by periodic modulation of an external parameter (concentration, absolute configuration of chiral molecules, etc.). The combination of the two spectroscopic methods can be applied to monitor the processes occurring at the catalytic solid– liquid interface. This combination provides significant enhancement of the signal-to-noise (S/N) ratio and allows the discrimination between active species and spectators.[34a] Mathematical transformation of the transient MES data from time- into phase-domain (f = 0!360 8), by using PSD provides further sensitivity enhancement and kinetic information can be extracted with improved accuracy according to Equation (1): Chem. Eur. J. 2014, 20, 1298 – 1309

www.chemeurj.org

2

Þ¼T Ak cosðfk  fdelay k

Z

T

AðtÞ sinðkwt þ fk Þdt

ð1Þ

0

in which T is the length of a cycle, f is the demodulation frequency, fk is the demodulation phase angle, fdelay is the phase k delay, k is the demodulation index (k = 1 in this study and for simplicity omitted hereafter), and A(t) and Ak are the active species response in time- and phase-domain, respectively. In our transient in situ-measurements, spectra in phasedomain (PD) contain the spectral changes induced by the periodic switching between two solutions in a quasi-steady state. But, in contrast to the data acquired in time-domain, after the transformation to PD vibrational features are a function of the phase angle f, which is altered from 0 to 360 8 with 10 8 increments resulting in 36 phase-resolved spectra. In accordance to the sinusoidal transformation, for every vibrational mode its amplitude, which is the equivalent of absorbance in PD, goes through a maximum at a certain phase angle. The switching between dissolved ketone and pure toluene solutions corresponds to phase angles 0 and 180 8, so that the amplitudes of vibrational modes of dissolved substrate in the first solution are maximum and minimum at exactly these angles, respectively. Reflecting the transport to the catalytic site and the sorption process signals from adsorbed species are normally delayed with respect to the phase of stimulation (switching of liquid concentrations). Therefore, after excitation of the surface process with dissolved ketone, the strongest response from an active species adsorbing on the catalytic surface will be monitored at a phase angle greater than 0 8, which is then called the in-phase angle of these species. Furthermore, signals belonging to the same active species will be in-phase at the same f, whereas a response from a subsequently forming species, for example, intermediates or products, will show its strongest signal output with more phase delay (greater f). In Figure 1, phase-resolved surface spectra of concentration modulation experiments (120 s adsorption–120 s desorption) display the dynamic adsorption period (f = 90!90 8) caused by concentration excitation with trifluoromethyl compounds on unmodified (a) and CD-modified (b–d) Pt/Al2O3 at 283 K. Note that in-phase spectra are highlighted in blue or green. Admission of dissolved TFAP to the unmodified catalyst (Figure 1 a) yields features at 1722 (C=O stretch) and at 1205– 1150 cm1 (asymmetric and symmetric CF3 stretching modes). With a phase delay of 60 8 a broad band centered at around 1680 cm1 is detected, which is assigned to C=O stretch vibration of the carbonyl group interacting with the Pt surface (see the Supporting Information for additional adsorption experiments with He-saturated solution and on an Al2O3-layer in Figures S1 and S2, respectively). Although the orientation of the detected C=O vibration has to be non-parallel to the metal surface, the dynamic nature of such an interaction leads to the band broadening and the blueshift indicates a reduced bond strength. On the CD-modified surface in Figure 1 b, no accumulation of PtC=O at 1680 cm1 but additional vibrational modes at 1269 (CCF3 stretch) and 1067 cm1 (CO stretch) are observed, which are assigned to the product alcohol (compare with Figure S8 in the Supporting Information). Obviously,

1300

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper cycling of low concentration of the strong trifluoro-activated acid TFA on CD–Pt/Al2O3 results in the appearance of a broad absorption band centered at around 2600 cm1. Formation of an active species involving the protonation of the quinuclidine N-atom is likely to originate from an acid/base interaction. Observed bands for asymmetric carboxylate stretches at around 1674 cm1 are in agreement to an ion-pair formation with the base and obey the same slow kinetics (f = 80 8). Table 1 summarizes vibrational frequencies assigned to TFAP, TFBA, and TFA detected in the MES experiments described in Figure 1.

Table 1. Assignment of selected absorption bands for TFAP, TFBA, TFA, and their interaction with co-adsorbed CD on Pt. In case of possible distinction from dissolved species, absorption bands originating exclusively from absorbed species are indicated. Absorption band [cm1]

the presence of CD leads to an accelerated hydrogenation and the modulation of formed product alcohol is included in the adsorption/desorption cycles with a phase delay of 30 8. In the literature the phenomenon of an enhanced reaction rate in presence of strongly adsorbing CD-molecules is generally referred to as “rate enhancement” or “rate acceleration” but the origin of the accelerating effect is still debated.[41] In addition to growing product peaks, a broad band following the admission of TFAP is observed in the surface spectra at around 2650 cm1, which is assigned to NHO-type hydrogen-bonding interaction between the quinuclidine N-atom of the co-adsorbed modifier and the carbonyl group of substrate, with is in agreement with an earlier discovery indicating the interaction between keto esters and CD.[18, 42] Although broadness and shape indicate a dynamic state of the observed vibrational mode, due to overlap of the strongest signal of the substrate (f = 10 8) and the formed product (f = 30 8), identification of the H-bond acceptor species is not unambiguous. The additional experiment with (R)-TFBA in Figure 1 c shows that the pure product alcohol also gives rise to the NHO absorption band with a slightly altered band shape. The comparison of Figure 1 b and 1 c indicates that the transformation of the substrate is accompanied by a competition of the ketone and product alcohol to dock with the modifier. This observation on the chirally modified catalyst is an explanation for the declining ee value with increasing substrate conversion.[23c] The crucial role that the tertiary amine function of the quinuclidine moiety plays as an active docking part of the adsorbed chiral modifier is further supported in the modulation experiment with TFA. Similar to ketone and alcohol adsorption, www.chemeurj.org

TFAP n˜ (C=O) n˜ (CF3) n˜ (CF3) n˜ (CF3) n˜ (C=O) on Pt n˜ (NH) of CD on CD-Pt TFBA n˜ (OH) n˜ (CCF3) n˜ (CF3) n˜ (CF3) n˜ (CF3) n˜ (CO) n˜ (NH) of CD on CD-Pt

1722 1205 1180 1149 1680 2650

Figure 1. Phase-domain spectra during adsorption cycles of TFAP on unmodified (a) Pt/Al2O3 and of (b) TFAP, (c) (R)-TFBA, and (d) TFA on CD-modified Pt/Al2O3, in H2-saturated toluene at 283 K. In-phase angles are indicated in green and blue. Absorbance is stated in mA.

Chem. Eur. J. 2014, 20, 1298 – 1309

Assignment

3560 1269 1205 1168 1129 1067 2650

TFA n˜ (C=O) monomer n˜ (C=O) dimer n˜ (CF3) n˜ (CF3) n˜ (CF3) n˜ (NH) of CD on CD-Pt

1802 1789 1215 1169 1128 2600

Enantiodiscrimination on CD–Pt/Al2O3 To better understand the nature of the chiral pocket induced by the adsorption of CD, we also performed absolute configuration modulation. Whereas conventional IR-spectroscopy is mute to the stereochemical orientation of distinct groups and their interaction with other chiral molecules, absolute configuration modulation responds exclusively to the switch in chirality of two enantiomers revealing a particular insight into the chiral environment on the CD-modified solid–liquid interface. In Figure 2, phase domains of concentration modulation between 0 and 1 mm (R)-TFBA (a) and absolute configuration modulation between 1 mm (R)- and 1 mm (S)-TFBA (b) on the CD-modified catalyst are displayed. Compared with the adsorption/desorption dynamics induced by the concentration change (f = 30 8), the switch in chirality shows a slower uptake of product at the chiral sites (f = 608). Clearly, periodic switch-

1301

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper Interactions between TFA and the modifier in a liquid solution

Figure 2. Phase domains during concentration modulation between 0 and 1 mm (R)-TFBA (a) and during absolute configuration modulation between 1 mm (R)-TFBA and 1 mm (S)-TFBA (b) on CD-modified Pt/Al2O3 in H2-saturated toluene at 298 K. Absorbance is stated in mA.

ing from (S)- to (R)-alcohol reveals positive peaks for CCF3and CF3-stretching modes, which proves transfer of the righthanded chiral information to the Pt catalyst. Whereas (R)-TFBA is able to better access CD-modified Pt, rejection of (S)-TFBA involves positive signal outputs from adsorbed modifier observed at its ring-breathing modes of the quinoline ring at 1590, 1578, and 1510 cm1, and CH deformation modes of quinuclidine at 1463 and 1321 cm1.[43] However, in this experiment the catalytic surface was modified by the CD solution prior to admission of alcohols and this finding does not object the possibility of a preceding TFBA–CD interaction in solution leading to an unfavored adsorption conformation of the modifier.[23c] A second striking observation is the absence of the C O stretch vibration at 1067 cm1 in Figure 2 b. Possessing the required stereochemical arrangement to occupy the chiral re face adsorption site, (R)-TFBA is able to dock with the OH function to the quinuclidine N-atom of adsorbed CD while the aromatic ring and CO bond are nearly in the same plane and in close contact to the Pt surface. The orientation of the CO bond of the product molecule within the chiral pocket is apparently close to parallel to the Pt surface so that the induced dipole moment change is diminished by the paramagnetic metal surface (Greenler’s surface selection rule),[44] and only the non-parallel CCF3 and CF3 vibrational modes are detected. In this adsorption state the trifluoro group has to point in the direction of the OH function of CD, which is in very good agreement to DFT calculations for the docking of TFAP to CD on Pt.[17] Probably, an additional but weak attractive interaction between the functional groups assists in pinning down the molecule to the re face. In the following section, the involvement of TFA in the enantiodiscriminating interaction and accompanied alterations of the asymmetric sites are addressed. Chem. Eur. J. 2014, 20, 1298 – 1309

www.chemeurj.org

As pointed out in the introduction, a peculiarity of the asymmetric hydrogenations of trifluoro-activated ketones is that addition of small amounts of TFA to the reaction mixture significantly improves enantioselection but lowers the catalytic activity.[12b, 22] As shown in Scheme 1, the enhancement of stereocontrol depends on the ratio of TFA/CD with optimal ratios found between 1–3 mol TFA/mol CD.[12b, 20] To address this unique characteristic of the catalytic system and to gain molecular insight into the role of TFA in the enantiodiscriminating surface complex, we investigated at first the interaction between the carboxylic acid TFA and the basic chiral modifier CD as well as that of related amine bases by using ATR-IR spectroscopy. These investigations were performed in toluene solution at different molar ratios by using concentration ranges relevant to catalytic reactions. Very low concentrations and weak ATR-IR sensitivity in the higher wavenumber region[45] make a detailed analysis of OH stretching vibrations cumbersome but the formation of acid/base adducts can easily be followed by their strongly IR-absorbing carboxylate stretching modes. Comparison to the ATR-IR measurements of related quinuclidine derivatives as well as with theoretical spectra calculated by the B3LYP/DZP method support the assignment of carboxylate and carbonyl bands to several coexisting ion-pair complexes and the identification of functional groups involved in the formation of acid/base adducts between TFA and CD in toluene. In Figure 3, the carbonyl-stretching region (1800–1700 cm1, n(C=O)), the asymmetric- (1700–1600 cm1, nas(OCO)) and the

Figure 3. ATR-IR spectra of toluene solutions containing (a) 2.5 mm CD, (b) 10 mm TFA (b), and mixtures of TFA and the amine bases (c) quinuclidine, (d) 3-quinoclidinol, (e) MeOCD, and (f) CD with the following concentrations (TFA/base) in mm: 5:5 (c); 5:5 (d); 5:5, 10:5, 15:5 (e); 1.25:2.5, 5:5, 10:5, 15:5 (f). Absorbance is stated in mA.

symmetric (1430–1390 cm1, ns(OCO)) carboxylate-stretching region as well as the spectral region for CF3-stretching modes (1250–1100 cm1, n(CF3)) for toluene solutions containing CD (a), TFA (b), and TFA-CD mixtures at varying molar ratios (f) are presented. Spectra of acid/base adducts with quinuclidine (c), 3-quinuclidinol (d), and the methyl ether of CD, MeOCD, with varying stoichiometry (e) are also included. As shown in Fig-

1302

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper ure 3 f, upon addition of 1.25 mm TFA to 2.5 mm CD solution, a nas(OCO) absorption band[29, 46, 47] is observed at 1663 cm1, whereas the n(C=O) peaks of free acid molecules are completely absent, indicative for a strong interaction between TFA and CD. At an acid/modifier ratio of 1:1 (5:5 mm) the carboxylate vibration nas(OCO) grows stronger and a shoulder at 1699 cm1 is observed. In the symmetric carboxylate-stretching region a band grows at 1430 cm1 with a minor feature at 1420 cm1 and in the CF3-stretching region two major peaks grow at 1199 and 1148 cm1, which are shifted with respect to pure TFA bands in toluene solution (compare with Figure 3 b)). Assuming a dynamic equilibrium between co-existing acid/ base complexes in solution, low acid concentration rather leads to the formation of 1:1 acid/base adducts, whereas excess acid brings along higher-ordered complexes, for example, 2:1, in which protonated TFA binds to existing 1:1 complexes. When adding excess TFA to the 5 mm CD solution, observed bands grow in intensity but also new features appear. In particular, the broad, overlapping signals in the asymmetric carboxylate stretching region indicate a high flexibility of the system, dynamic processes, and various possible configurations near the energetic minima.[48] At a concentration ratio of 2:1, a distinct band is seen at 1643 cm1, which is assigned to nas(OCO) of a structurally different ion-pair complex of higher order.[46, 47] Particularly, the shoulder at 1699 cm1 grows stronger, whereas the presence of additional acid molecules causes the other asymmetric carboxylate vibration to slightly redshift its center to peak at 1670 cm1. In addition to the symmetric carboxylate peak at 1430 cm1, a higher acid concentration leads to the appearance of two additional bands centered at around 1410 and 1390 cm1. Concerning the CF3-stretching modes, contributions from TFA species are observed at 1219 and 1173 cm1, which match the positions of TFA features in pure acid solution. This is an indication that these observed bands belong to protonated, weakly bound acid molecules involved in hydrogen-bonding, as can be observed for acid dimers.[47] In the spectrum of 3:1 acid excess, the features assigned to protonated TFA species grow stronger and are accompanied by carbonyl vibrations in the acid dimer region at around 1780 cm1 corroborating the assignment to H-bonded acid molecules. Obviously, for the strong acid and the modifier several acid/base complexes with different stoichiometry coexist in toluene in a tautomeric equilibrium: TFA + CDÐTFA + CD + , which lies almost completely on the right side at equimolar ratios and the excess acid molecules bind to the adducts by additional hydrogen-bonding, as had also been observed by Szafran and co-workers for TFA and 2,4,6-trimethylpyridine in benzene.[46] Regarding the effect of the ion-pair formation on the major absorption frequencies of the modifier, the position of the ring-deformation mode of the quinoline ring shift from 1590 to 1595 cm1 and the CO stretch of the alcohol group at 1098 cm1 is attenuated upon complexation with TFA. Broad bands centered around 3260 cm1 (OH stretching modes) and 2585 cm1 (NH + stretching modes) also prove the involvement of these functional groups,[29] which show dynamic shifts depending on molar ratios (see Figure S5, the Supporting Information). Chem. Eur. J. 2014, 20, 1298 – 1309

www.chemeurj.org

Next, we investigated the involvement of the functional groups of CD and their coordination with TFA by comparing the spectral similarities and dissimilarities of acid/base adducts with quinuclidine derivates (Figure 3 c–e). The TFA–quinuclidine adduct in Figure 3 c gives rise to one main nas(OCO) band identified at 1676 cm1 and one ns(OCO) at 1410 cm1, accompanied by similar shifts in the CF3-stretching region. However, only the presence of the OH group in 3-quinuclidinol (compare with Figure 3 d) leads to a carbonyl shoulder at 1694 cm1, tail broadening in the nas(OCO) band region, and additional bands at 1420 as well as at 1173 cm1. This is an indication that the presence of the alcohol function located within the distance of two C atoms allows, even at an equimolar acid/base ratio, the formation of a stable 2:1 complexes of the form A···HB + ···HA in which the involvement of the second, protonated acid molecule (HA) is evident from the carbonyl shoulder at around 1694 cm1. Due to the absence of a carbonyl shoulder in case of quinuclidine, the observed 2:1 complex in Figure 3 d is likely to be cyclic, including an attractive interaction between the protonated acid and the alcohol function of 3-quinuclidinol. If the OH function of CD is methylated (MeOCD), only excess acid leads to significant broadening in the carbonyl region and appearance of the second carboxylate peak nas(OCO) at 1634 cm1. Obviously, blocking of the alcohol function of the modifier significantly retards the formation of 2:1 complexes at low acid concentrations and mainly one carboxylate band originating from the adduct with the quinuclidine function is identified at 1676 cm1. Compared with the carboxylate peak of the TFA–CD adduct centered at 1663 cm1, the lower frequency position and lower IR absorption intensity indicate a weaker ionization of the acid group by MeOCD and hence a more protonated state of TFA if only interacting with the quinuclidine N-atom. With the aid of DFT calculations, the coordination of several co-existing TFA/CD complexes and their assignment to distinct IR absorption frequencies can be derived from the spectral analysis in toluene solution and their proposed molecular structures are summarized in Scheme 2. Solvation effects were not considered for calculated gasphase frequencies leading to discrepancies with regard to the band positions measured in the liquid phase but the general shift in frequency for different carboxylates and carbonyl vibrations are in good agreement. For the present investigation, identification of the asymmetric carboxylate bands at 1676, 1663, and 1643 cm1 as well as carbonyl bands and their assignment to TFA–CD interactions is crucial. Molar acid/base ratios and energy-optimized structures obtained by using calculations (see Tables S1 and S2 in the Supporting Information) emphasize the existence of mainly 1:1 and 2:1 TFA/CD complexes for which the two most relevant structures are presented. According to the computational results, in the most stable 1:1 structure the deprotonated trifluoroacetic acetate is hydrogen-bonded to the protonated quinuclidine N atom but also to the OH function of CD as shown in Scheme 2 B. A similar cyclic structure was found by ab initio calculations including acetic acid and CD.[25b] The monodentate complex involving only the NH + ···O hydrogen-bonding interaction is signifi-

1303

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper Table 2. Selected vibrational frequencies of TFA, TFA/CD 1:1 acid/base complexes (monodentate and bidentate) and TFA/CD 2:1 acid/base complexes (acetate and hydrogen-bonded acid). n˜ (C=O) n˜ as(OCO) n˜ s(OCO) n˜ (CF3) n˜ (CF3) n˜ (CF3) TFA 1803 TFA–CD (CD + A) monodentate TFA–CD (CD + A) bidentate TFA–CD (CD + A···HA) acetate TFA (CD + A···HA) 1699 H bonded

1673

1410

1215 1199

1173 1148

1663

1430

1199

1148

1643

1390

1199

1148

1219

1173

1130

1130

Scheme 2. Proposed molecular structures of TFA/CD acid/base complexes. Experimental (exptl) and theoretical (calcd) vibrational frequencies of the identified complexes for the asymmetric carboxylate- (nas) and the carbonyl(nC=O) stretching modes are indicated.

cantly less stable leading to a redshift of the carboxylate band. In fact, only for MeOCD such a stable monodenate complex was found in gas-phase calculations in which TFA is still protonated giving rise to a carbonyl instead of a carboxylate band. However, it is expected that solvation in the liquid phase can stabilize weaker complexes and the experimental findings of stable acid/base adducts involving quinuclidine or MeOCD prove the existence of carboxylate in toluene coordinated only to the nitrogen atom giving rise to nas(OCO) at 1676 cm1 and ns(OCO) at 1410 cm1. Regarding higher-ordered complexes, cyclic and half-cyclic stable complexes with the molecular coordination of 2:1 were identified in accordance to previously reported results involving acetic acid.[25b] Based on the calculations, the position of the carboxylate band in the 2:1 cyclic structure (Scheme 2 c) is higher in frequency than the half-cyclic complex in which an additional acid molecule binds to the bidentate complex as shown in Scheme 2 d (an open 2:1 complex, a possible structure for MeOCD, was not within the scope of the theoretical study and the resulting frequency shifts were neglected). For both complexes though, the additional H-bonded TFA molecule gives rise to a carbonyl band at around 1710 cm1, which is in very good agreement to the observed carbonyl shoulder at 1699 cm1. Assignment of the absorption bands originating from TFA and its adducts with CD are summarized in Table 2. Further insight into the molecular interactions relevant to the studied multiphase reaction system is gained by exposing the diastereomeric equimolar CD–TFA complex to increasing concentrations of the two product enantiomers in the liquid phase, which mimics the composition of the reaction solution Chem. Eur. J. 2014, 20, 1298 – 1309

www.chemeurj.org

Figure 4. ATR-IR spectra of solutions of 2.5 mm TFA and 2.5 mm CD in toluene with increasing molar equivalents of 0, 2.4, 5, and 14.4 of (a) (S)-TFBA and (b) 0, 2.6, 5.4, and 14.1 of (R)-TFBA. Arrows indicate the change in intensity of bands following the addition of TFBA. Absorbance is stated in mA.

throughout the catalytic hydrogenation. The deviating behavior experienced during the addition of distinct optical isomers is striking. The spectra in Figure 4 a show that a slight molar excess of (S)-alcohol with respect to CD/TFA is sufficient to decrease bands of the diastereomer-acid complex, particularly observable by decreasing bands at 1699, 1663, 1430, 1196, and 1146 cm1. In the higher wavenumber region an increasing excess of (S)-TFBA observable at its OH stretch at 3560 cm1 (as well as at 1267 and 1067 cm1) leads to the consecutive decrease of broad bands in the region around 3260 (also redshifts) and 2585 cm1, which originate from hydrogen-bonded OH and NH + ···O of cyclic CD–TFA-complexes, respectively.[29] Attenuation of bands assigned to the TFA–CD complex must arise from a competing interaction of (S)-TFBA hindering the access to the modifier. In case of the right-handed enantiomer, only a significant excess of 14:1:1 = (R)-TFBA/TFA/CD led to a considerable interference with the acid/base complex. This is a strong indication of at least two attractive or repulsive interactions between modifier and chiral alcohol molecule making it stereochemically more demanding for (R)-TFBA to interact properly with CD in solution. In an earlier NMR study it was also found that (S)-TFBA is more prone to bind with its acidic

1304

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper OH group to the diastereomer CD in CH2Cl2 solution[23c] (note that the low concentration mixtures of CD and TFBA in toluene gave no significant rise to any detectable signal assignable to their interaction). In contrast, on Pt/Al2O3 preferred adsorption of (R)-TFBA at the CD-modified site was shown (compare Figure 2 b, in the absence of TFA). Obviously, the metal surface must impose geometrical constrains and CD adsorbed on Pt creates the chiral pocket for preferred docking of (R)-trifluoromethyl alcohol. For addition of both enantiomers though at higher alcohol concentration the decrease of the nas(OCO) band is accompanied by a redshift to 1673 cm1, which was assigned to the carboxylate species of TFA only bound to the quinuclidine nitrogen atom of CD (complex a in Scheme 2). It appears that the bidentate TFA–CD complex can be cleaved at the second interaction by the presence of the chiral product alcohol leading to the weakening of the sensitive asymmetric nas(OCO) vibration. In the opened, monodentate CD–TFA complex, both the OH function of the modifier and the carboxylate residing at the quinuclidine N-atom are available for interactions with other molecules, for example, substrate or product. In other words, the trifluoromethyl alcohol is able to interact with the TFA–CD complex leading to the transformation of bidentate to monodentate complex. Therefore, it is reasonable to assume that the product molecule breaks the OH···O interaction and inserts with its CF3 or its alcohol function remaining at least either at the carboxylate of TFA or the OH function of CD. Enantiodiscrimination on TFA–CD–Pt/Al2O3 The different behavior of product enantiomers interacting with CD-modified Pt and with CD in solution highlights the importance of the additional steric hindrance created by adsorption onto the active metal surface. To further the identification of active surface species involving TFA we studied the interaction of (R)- and (S)-TFBA on the chirally modified Pt/Al2O3 in the presence of small amounts of the acid. In the phase-resolved spectra in Figure 5 a, the response on the catalytic surface triggered by addition of 1 mm (R)-TFBA to the solution containing 1 mm CD and 0.25 mm TFA is shown. Upon admission of the (R)-alcohol, its adsorption is evident from characteristic absorption peaks at 1267, 1190, 1164, 1128, and 1068 cm1. Compared to the concentration modulation experiment in absence of TFA in Figure 2 a the absorption amplitude is lowered by a factor of almost 4. This is an indication that TFA and product alcohol compete for the same adsorption site and strongly adsorbed TFA species interfere with the accumulation of TFBA on the surface. Exposing the TFA–CD-modified surface to product molecules also shows overlapping features in the carboxylate-stretching region. Carbonyl bands at 1700 cm1 and CF3-stretching modes at 1215 cm1 belonging to hydrogen-bonded TFA molecules are removed by the alcohol. Furthermore, the bidentate modifier-carboxylate species identified at 1663 cm1 is reduced, whereas the asymmetric stretch at 1678 cm1 of the carboxylate coordinated only to Chem. Eur. J. 2014, 20, 1298 – 1309

www.chemeurj.org

Figure 5. Phase domains during concentration modulation between 0 and 1 mm (R)-TFBA (a) and during absolute configuration modulation between 1 mm (R)-TFBA and 1 mm (S)-TFBA (b) on CD-modified Pt/Al2O3 in H2-saturated toluene with 0.25 mm TFA at 298 K. Absorbance is stated in mA.

the quinuclidine N-atom shows the highest active signal output in the same phase as the product alcohol (f = 308). The behavior of the TFA–CD complex detected on the surface seems to resemble the dynamics of the carboxylate species detected in the liquid solution: The product alcohol interferes with TFA for adsorbed CD leading to the preferred formation of the monodentate complex evident also from the active signal output of the symmetric carboxylate stretch ns(OCO) at 1408 cm1. The interaction between the adsorbed CD–TFA ionpair complex and product alcohol leads to the redshift of the nas(OCO) mode and appearance of the band at 1678 cm1. A positive response of the adsorbed quinoline ring of the diastereomer-acid complex is also detected at 1597 and 1578 cm1, which corroborates the active role of the detected species. Upon switching the liquid solution back to 0 mm (R)-TFBA (f =  180 8), accumulation of the protonated TFA (f = 110 8) is followed by a slow recovery of the bidentate TFA–CD complex at the expense of the monodentate species observed with an in-phase angle of 70 8. Next we explored the participation of TFA in the enhanced enantiodifferentiating process on the catalytic surface by absolute configuration modulation between solutions containing 1 mm (R)- and 1 mm (S)-TFBA in the presence of 0.25 mm TFA on CD-modified Pt/Al2O3. Note that in the absolute modulation excitation experiment all monitored signals originate from molecules in close proximity to the adsorbed diastereomeric modifier and therefore have to be close to the paramagnetic surface. Hence, the detected dipole moment changes have to be tilted with respect to the surface while parallel vibrational

1305

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper modes are diminished. In Figure 5 b, the switch in chirality of the product alcohol imposes deviations in the ion-pair formation between adsorbed CD and TFA as evident from the spectral changes induced by the modulation. The surface spectra reveal that interactions of acid molecules around the adsorbed diastereomeric modifier change depending on the presence of the two enantiomers, firmly indicating the involvement of TFA in the enantiodifferentiating action. Monodentate carboxylate peaks at 1678 and 1408 cm1 show the strongest signal output after switching to the solution containing the right-handed product, which is also the chiral preference of CD in the absence of TFA (see Figure 2 b). Although the determination of the kinetic sequence is ambiguous in the asymmetric stretching region due to band overlaps, the quinoline ring of CD and the symmetric monodentate carboxylate stretch clearly show the same phase behavior giving the strongest active signal response with a phase delay of f = 60 8 with respect to the Figure 6. Phase domain during concentration modulation between 0 and switch to (R)-TFBA. It can be inferred that it is this monoden1 mm TFAP on CD-modified Pt/Al2O3 in H2-saturated toluene with 0.25 mm TFA at 298 K. Absorbance is stated in mA. tate TFA–CD complex with the carboxylate coordinated to the quinuclidine N-atom of CD that preferably interacts with the right-handed enantiomer and hence plays an active role translating the desired stereochemical information for the asymmetble to assume the formation of a similar co-adsorbed CD–TFA– ric reaction. In the modulation experiments without TFA in TFAP complex responsible for imparting enhanced enantioinFigure 2 the same kinetic behavior was discovered: Concentraduction toward the right-handed enantiomer in which the very tion modulation on CD-modified Pt/Al2O3 showed a phase acidic carboxylic acid is residing at the quinuclidine nitrogen. delay of f = 30 8, whereas the stereochemical excitation in the The proposed enantiodifferentiating intermediate surface comabsolute configuration modulation obeyed the same slower kiplexes between the a-trifluoromethyl-activated ketone and CD netics with f = 60 8. in the absence and presence of TFA are shown in Scheme 3. Upon switching to the (S)-product solution, the adsorbed Conformation of adsorbed CD and NHO interaction beTFA–CD complex blocks its docking sites by transforming to tween the quinuclidine N-atom and the carbonyl group of the the bidentate complex, which is also accompanied by the acketone adsorbed with its re face allow for a second attractive cumulation of hydrogen-bonded TFA (f = 110 8). An unexinteraction between OH and CF3 functions. Blocking of the pected spectral feature is the negative response of the CCF3 asymmetric site by acidic chiral product alcohol is also depicted. Considering the interaction of the TFA–CD-modified site stretch of product alcohol at a phase angle of f = 60 8, which with reactant ketone two feasible prochiral diastereomeric suris the only isolated signal assignable to TFBA. Two explanations face complexes are proposed, which could be responsible for seem feasible: Either the relative population of (R)-product is enhanced stereocontrol. In these two complexes the role of decreased on the TFA–CD-modified surface, or the change in the acid is of a different kind. In the complex “CD–TFA– orientation of the (R)-product adsorbed at the better defined ketone 1” TFA would serve as a linker between the basic amine asymmetric environment is close to parallel to the metal surfunction of CD and the oxygen of the keto-carbonyl group, face diminishing this molecular vibration. which provides a better fixation of the substrate close to the Finally, the formation of a CD–TFA complex involving the surface and also simplifies a second interaction between the prochiral ketone is addressed in the modulation experiment with 1 mm TFAP on TFA–CDmodified Pt/Al2O3. In Figure 6, admission of substrate leads to the detection of a stable surface species with a phase delay of f = 40 8. Bands associated with the previously identified monodentate CD–TFA complex feature at around 1678, 1598, 1582, and 1410 cm1. The active surface response triggered by prochiral substrate is in very good agreement with the signals assigned Scheme 3. Enantiodifferentiating intermediate surface complexes between co-adsorbed ketone TFAP and chiral to the liquid-phase complex a in modifier CD in absence and presence of TFA. The NHO-type hydrogen-bonding interaction between CD and (R)Scheme 2. Hence, it is reasona- TFBA is also depicted. For the adsorbed modifier, the conformation open(4)[49] is assumed. Chem. Eur. J. 2014, 20, 1298 – 1309

www.chemeurj.org

1306

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper OH and the CF3 functions. In the complex “CD-TFA-ketone 2”, the role of TFA on the surface would mainly be a stabilizing function of the open conformation of CD allowing for an improved access to the alcohol function. Although the experimental observations highlight the importance of the OH function of CD, the question whether it plays an active role in the enantiodifferentiation, or if the steric hinderance of alkyl ether derivatives of the cinchona alkaloids retards the stereoselective docking, remains unclear and has to be addressed in future research regarding the asymmetric hydrogenation of trifluoro-activated ketones on chirally modified metal surfaces.

Conclusion Excellent stereocontrol in the asymmetric hydrogenation of the a,a,a-trifluoromethyl ketone TFAP can be achieved on CDmodified Pt/Al2O3 at ambient hydrogen pressure, in particular, by addition of small amounts of TFA. The present in situ investigation, applying ATR-IR spectroscopy in combination with MES and PSD, confirms the earlier postulated crucial interaction between the reactant ketone and the chiral modifier on the surface, which are believed to be at the origin of enantioselection. Formation of an NHO-type hydrogen-bonding between CD and TFAP was observed, which is in close agreement with related studies involving ester-activated ketones. In contrast to the transformation of keto esters, it is shown that the fluoro-activated, acidic product alcohol is also prone to form such a hydrogen bond on the surface leading to the earlier observed reduced enantiocontrol with increasing conversion.[23c] Furthermore, the occurrence of the “rate enhancement” could be followed on the catalytic solid–liquid interface. In the presence of adsorbed CD the accumulation of the less reactive carbonylPt species was not detected, which could be linked to rate enhancement originating from an assisted guidance of the carbonyl group to a favorable and more reactive adsorption configuration. Correlation between IR-spectra obtained in liquid solutions and monitoring of adsorbed species on the catalytic surface revealed a striking difference between the dissolved and adsorbed diastereomeric modifier interacting with chiral compounds. Whereas (S)-TFBA interacts more strongly with CD in solution, on the active metal surface the chiral preference is inverted. Adsorption of chiral product alcohol at the asymmetric adsorption site is mainly directed by NHO interaction leading to a parallel orientation of the aromatic ring and CO bond with respect to the surface. Due to the significant enhancement of the ee value achieved by addition of catalytic amounts of trifluoro-activated acid TFA to the hydrogenation reaction mixture of this substrate class, investigations on the role of the additive were included in this study. ATR-IR measurements combined with DFT calculations show that TFA forms several co-existing acid/base adducts involving not only the quinuclidine N-atom but also the OH function of CD in toluene. In situ surface spectra prove the involvement of TFA in the chiral footprint on the catalytic surface. Modulation excitation experiments including chiral product and prochiral subChem. Eur. J. 2014, 20, 1298 – 1309

www.chemeurj.org

strate provide strong evidence that the enantiodiscrimination of CD for trifluoro-activated ketones is enhanced by a TFA molecule residing at the quinuclidine group. Two possible molecular structures of the enantiodifferentiating surface complex are proposed, which rationalize the beneficial effect of addition of TFA by improved access to both functional group of the chiral modifier or by stabilization of the open conformer on the surface leading to preferential adsorption of the ketone with the re face. However, cleaving of the bidentate complex is required to access the modified site and it could be this additional energy barrier together with some blocking of active sites[30] that lead to the considerable decrease in catalytic activity in the presence of TFA. Refinement of existing mechanistic models for the Pt-catalyzed asymmetric hydrogenations of activated ketones and improved understanding of enantioselective hydrogenations of fluoro-activated ketones should stimulate the design of synthetic modifiers specifically designed for this class of substrates.

Experimental Section Materials Cinchonidine, CD (Sigma–Aldrich, 96 %), methyl-O-CD, MeOCD (Ubichem Research Limited), quinuclidine, Qu (Fluka, 97 %), 3-quinuclidinol, QuOL (Aldrich, 99 %), trifluoroacetophenone, TFAP (Sigma–Aldrich, 99 %), (R)-(trifluoromethyl)benzyl alcohol, (R)-TFBA (Fluka, 99 %, R:S 98:2), (S)-(trifluoromethyl)benzyl alcohol, (S)-TFBA (Fluka, 99 %, S:R 98:2), trifluoroacetic acid, TFA (Fluka, 98 %), and toluene (Sigma–Aldrich,  99.9 %) were used as received. The commercial 5 wt. % Pt/g-Al2O3 catalyst[13] (BASF-Engelhard 4759; platinum dispersion, 0.22, BET surface area, 168 m2g1) was reduced in a fixed-bed tubular reactor under flowing hydrogen for 1 h at 673 K and subsequently a thin layer of the material was deposited onto a ZnSe crystal. After mounting the coated crystal in a flowthrough cell the catalyst layer was reduced again by subjecting it to flowing toluene saturated with hydrogen at 323 K for 2 h prior to use.

Preparation of the catalyst layer A slurry was prepared with the catalyst (90 mg) and H2O (12 mL) and stirred for several hours to obtain a well-mixed suspension. Subsequently, 1.0 mL of the slurry was placed on a ZnSe internal reflection element (IRE, bevel of 45 8, 52  20  2 mm, Crystran Ltd.) and H2O was evaporated at 80 8C for 2 h. The as-prepared catalyst layer adhered to the IRE such that no loss of catalyst was observed over the course of several hours under flow-through conditions.

IR spectroscopy of the solid–liquid interface ATR-IR spectra were recorded on a Bruker IFS66/S spectrometer equipped with a liquid nitrogen-cooled MCT detector at 4 cm1 resolution. A home-build stainless steel flow-through cell mounted on an ATR-IR attachment was used for all the adsorption/desorption measurements. The cell has been characterized in detail elsewhere.[32] The measurement temperatures (283–298 K) were regulated by a thermostat (Julabo, F25). For adsorption/desorption experiments, a standard solution of CD (and TFA) in toluene was prepared, which was then split into two

1307

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper separate solutions. In accordance to the required modulation excitation (concentration or absolute configuration) the second solution was prepared by addition of trifluoro compounds. Both solutions were filled in two separated glass-made bubble tanks in which they were saturated with pure hydrogen (PanGas, 99.999 %). The tanks were connected to the flow-through cell by Teflon tubing. Solutions were continuously admitted at a flow rate of 0.5 mL min1 by using a peristaltic pump (ISMATEC, Reglo 100) installed at the outlet of the cell. Two interconnected pneumatically activated Teflon valves (PARKER, PV-1–2324) were synchronized with the acquisition of IR spectra by the spectrometer software (Bruker Optics, OPUS). Application of MES and PSD are described elsewhere.[34]

[7] [8]

[9] [10]

IR Spectroscopy of solutions ATR-IR spectra were recorded on a Bruker Vertex 70 spectrometer equipped with a liquid nitrogen-cooled MCT detector at 4 cm1 resolution. The commercially available Bruker single reflection Platinum ATR-IR accessory equipped with a diamond crystal was used. Before measurements all toluene solutions were well mixed in a sonication bath for 5 min and immediately analyzed thereafter. Liquid samples were measured by filling the sample holder surrounding the 2  2 mm2 crystal and covering the solution with a glass plate. The spectra were recorded by co-adding 128 scans and the background was recorded with pure solvent.

[11] [12]

Computations

[18] [19]

All molecules were geometry optimized by using the gas-phase B3LYP[35]/DZP[36] method as implemented in Turbomole 6.3.[37] Optimized structures were confirmed as minima by a full gas phase frequency calculation. IR spectra were calculated within the harmonic approximation using gas-phase frequency calculations. IR shifts were scaled by a correction factor of 0.94, which was obtained by matching the CF3 region of the calculated spectra of TFA and the TFA dimer to the measured spectrum of TFA.

[13] [14] [15]

[16] [17]

[20]

[21] [22] [23]

Acknowledgements Financial support by the Foundation Claude & Giuliana is kindly acknowledged.

[24]

[25]

Keywords: asymmetric catalysis spectroscopy · ketones · platinum

IR

[26] [27]

[1] a) G. Resnati, Tetrahedron 1993, 49, 9385; b) P. Bravo, G. Resnati, Tetrahedron: Asymmetry 1990, 1, 661; c) Chirality in Industry: The Commercial Manufacture and Applications of Optically Active Compounds (Eds.: A. N. Collins, G. N. Sheldrake, J. Crosby), John Wiley & Sons, Chichester, 1995. [2] R. A. Sheldon, J. Chem. Technol. Biotechnol. 1996, 67, 1. [3] R. Noyori, T. Ohkuma, Angew. Chem. 2001, 113, 40; Angew. Chem. Int. Ed. 2001, 40, 40. [4] N. Knzle, R. Hess, T. Mallat, A. Baiker, J. Catal. 1999, 186, 239. [5] H. U. Blaser, B. Pugin, F. Spindler, J. Mol. Catal. A 2005, 231, 1. [6] a) R. Raval, J. Phys. Condens. Matter 2002, 14, 4119; b) V. Demers-Carpentier, G. Goubert, F. Masini, R. Lafleur-Lambert, Y. Dong, S. Lavoie, G. Mahieu, J. Boukouvalas, H. L. Gao, A. M. H. Rasmussen, L. Ferrighi, Y. X. Pan, B. Hammer, P. H. McBreen, Science 2011, 334, 776; c) V. Demers-Carpentier, G. Goubert, F. Masini, Y. Dong, A. M. H. Rasmussen, B. Hammer, P. H. McBreen, J. Phys. Chem. Lett. 2012, 3, 92; d) J. D. Horvath, A. J. Gellman, Top. Catal. 2003, 25, 9; e) R. M. Hazen, D. S. Sholl, Nat. Mater. 2003, 2, 367; f) G. A. Attard, J. Phys. Chem. B 2001, 105, 3158; g) T. E. Jones, C. Baddeley, Surf. Sci. 2002, 519, 237; h) M. O. Lorenzo, C. J. Bad-

[28] [29] [30] [31]

Chem. Eur. J. 2014, 20, 1298 – 1309

·

hydrogenation

www.chemeurj.org

·

[32] [33] [34] [35]

[36] [37]

1308

deley, C. Muryn, R. Raval, Nature 2000, 404, 376; i) D. Stacchiola, L. Burkholder, W. T. Tysoe, J. Am. Chem. Soc. 2002, 124, 8984; j) M. Schunack, E. Laegsgaard, I. Stensgaard, I. Johannsen, F. Besenbacher, Angew. Chem. 2001, 113, 2693; Angew. Chem. Int. Ed. 2001, 40, 2623. G. Goubert, P. H. McBreen, ChemCatChem 2013, 5, 683 – 685. a) T. Mallat, E. Orglmeister, A. Baiker, Chem. Rev. 2007, 107, 4863; b) M. Heitbaum, F. Glorius, I. Escher, Angew. Chem. 2006, 118, 4850; Angew. Chem. Int. Ed. 2006, 45, 4732; c) M. Bartk, Current Organic Chemistry 2006, 10, 1533; d) F. Zaera, Acc. Chem. Res. 2009, 42, 1152; e) D. Y. Murzin, P. Maki-Arvela, E. Toukoniitty, T. Salmi, Catal. Rev. Sci. Eng. 2005, 47, 175; f) H. U. Blaser, B. Pugin, F. Spindler, M. Thommen, Acc. Chem. Res. 2007, 40, 1240. A. Baiker, Catal. Today 2005, 100, 159. a) Y. Orito, S. Imai, S. Niwa, G.-H. Nguyen, J. Synth. Org. Chem. Jpn. 1979, 37, 173; b) Y. Orito, S. Imai, S. Niwa, Nippon Kagaku Kaishi 1979, 1118; c) M. Sutyinszki, K. Szçri, K. Felfçldi, M. Bartk, Catal. Commun. 2002, 3, 125. H. U. Blaser, H. P. Jalett, M. Mller, M. Studer, Catal. Today 1997, 37, 441. a) T. Mallat, M. Bodmer, A. Baiker, Catal. Lett. 1997, 44, 95; b) T. Varga, K. Felfçldi, P. Forgo, M. Bartk, J. Mol. Catal. A 2004, 216, 181; c) V. Demers-Carpentier, P. H McBreen, J. Phys. Chem. C 2011, 115, 6513. A. Baiker, J. Mol. Catal. A 1997, 115, 473. S. Lavoie, M. A. Laliberte, I. Temprano, P. H. McBreen, J. Am. Chem. Soc. 2006, 128, 7588. J. L. Margitfalvi, M. Hegedus, E. T. First in 11th Proc. Intern. Congress on Catalysis, 40th Anniversary, Pts. A and B, Vol. 101 (Eds.: J.W Hightower, W. N. Delgass, E. Iglesia, A. T. Bell), Elsevier, Amsterdam, 1996, p. 241. F. Zaera, J. Phys. Chem. C 2008, 112, 16196. E. Schmidt, C. Bucher, G. Santarossa, T. Mallat, R. Gilmour, A. Baiker, J. Catal. 2012, 289, 238. N. Bonalumi, T. Brgi, A. Baiker, J. Am. Chem. Soc. 2003, 125, 13342. S. Lavoie, G. Mahieu, P. H. McBreen, Angew. Chem. 2006, 118, 7564; Angew. Chem. Int. Ed. 2006, 45, 7404. M. von Arx, T. Mallat, A. Baiker in Supported Catalysts and Their Applications (Eds.: D. C. Sherrington, A. P. Kybett), RSC, Cambridge, 2001, p. 247. F. Meemken, N. Maeda, K. Hungerbhler, A. Baiker, Angew. Chem. 2012, 124, 8336; Angew. Chem. Int. Ed. 2012, 51, 8212. M. von Arx, T. Mallat, A. Baiker, Tetrahedron: Asymmetry 2001, 12, 3089. a) M. Bodmer, T. Mallat, A. Baiker in Catalysis of Organic Reactions, Vol. 75 (Ed.: F. K. Herkes), Marcel Dekker, New York, 1998; b) K. Szçri, K. Balazsik, S. Cserenyi, G. Szçllçsi, M. Bartk, Appl. Catal. A 2009, 362, 178; c) Z. Cakl, S. Reimann, E. Schmidt, A. Moreno, T. Mallat, A. Baiker, J. Catal. 2011, 280, 104. B. Tçrçk, K. Balzsik, K. Felfçldi, M. Bartk, Studies in Surface Science and Catalysis, Vol. 130 (Eds.: A. Corma, F. V. Melo, S. Mendioroz., J. L. G. Fierro) Elsevier, Granada, 2000, p. 3381. a) T. Brgi, A. Baiker, J. Am. Chem. Soc. 1998, 120, 12920; b) D. Ferri, T. Brgi, A. Baiker, J. Chem. Soc. Perkin Trans. 2 1999, 1305. Z. Ma, F. Zaera, J. Phys. Chem. B 2005, 109, 406. O. Schwalm, J. Weber, J. Margitfalvi, A. Baiker, J. Mol. Struct. 1993, 297, 285. M. von Arx, T. Mallat, A. Baiker, J. Catal. 2000, 193, 161. M. von Arx, T. Brgi, T. Mallat, A. Baiker, Chem. Eur. J. 2002, 8, 1430. W. R. Huck, T. Brgi, T. Mallat, A. Baiker, J. Catal. 2001, 200, 171. J. Brunelle, V. Demers-Carpentier, R. Lafleur-Lambert, G. Mahieu, G. Goubert, S. Lavoie, P. H. McBreen, Top. Catal. 2011, 54, 1334. I. Maupin, L. Pinard, J. Mijoin, P. Magnoux, J. Catal. 2012, 291, 104. A. Urakawa, R. Wirz, T. Brgi, A. Baiker, J. Phys. Chem. B 2003, 107, 13061. a) D. Baurecht, U. P. Fringeli, Rev. Sci. Instrum. 2001, 72, 3782; b) A. Urakawa, T. Brgi, A. Baiker, Chem. Eng. Sci. 2008, 63, 4902. a) P. A. M. Dirac, Proc. R. Soc. London Ser. A 1929, 123, 714; b) J. C. Slater, Phys. Rev. 1951, 81, 385; c) S. H. Vosko, L. Wilk, M. Nusair, Can. J. Phys. 1980, 58, 1200; d) A. D. Becke, Phys. Rev. A 1988, 38, 3098; e) A. D. Becke, J. Chem. Phys. 1993, 98, 5648; f) C. T. Lee, W. T. Yang, R. G. Parr, Phys. Rev. B 1988, 37, 785. A. Schfer, H. Horn, R. Ahlrichs, J. Chem. Phys. 1992, 97, 2571. Universitt Karlsruhe and Forschungszentrum Karlsruhe GmbH, TURBOMOLE V6.3 2011, 1989 – 2007.  2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper [38] R. Pereniguez, G. Santarossa, T. Mallat, A. Baiker, J. Mol. Catal. A 2012, 365, 39. [39] R. Hess, A. Vargas, T. Mallat, T. Brgi, A. Baiker, J. Catal. 2004, 222, 117. [40] J. M. Andanson, A. Baiker, Chem. Soc. Rev. 2010, 39, 4571. [41] a) G. Szçllçsi, S. Csernyi, I. Bucsi, T. Bartk, F. Flçp, M. Bartk, Appl. Catal. A 2010, 382, 263; b) M. A. Laliberte, S. Lavoie, B. Hammer, G. Mahieu, P. H. McBreen, J. Am. Chem. Soc. 2008, 130, 5386; c) E. Talas, J. L. Margitfalvi, O. Egyed, J. Catal. 2009, 266, 191; d) E. Toukonitty, D. Y. Murzin, J. Catal. 2006, 241, 96; e) M. Garland, H. U. Blaser, J. Am. Chem. Soc. 1990, 112, 7048. [42] a) N. Maeda, K. Hungerbhler, A. Baiker, J. Am. Chem. Soc. 2011, 133, 19567; b) Y. Matsuda, T. Ebata, N. Mikami, J. Chem. Phys. 1999, 110, 8397 – 8407. [43] a) D. Ferri, T. Brgi, J. Am. Chem. Soc. 2001, 123, 12074; b) W. Chu, R. J. LeBlanc, C. T. Williams, J. Kubota, F. Zaera, J. Phys. Chem. B 2003, 107, 14365; c) Z. Ma, I. Lee, J. Kubota, F. Zaera, J. Mol. Catal. A 2004, 216, 199.

Chem. Eur. J. 2014, 20, 1298 – 1309

www.chemeurj.org

[44] a) R. G. Greenler, J. Chem. Phys. 1966, 44, 310; b) R. G. Greenler, D. R. Snider, D. Witt, R. S. Sorbello, Surf. Sci. 1982, 118, 415. [45] E. Roedel, A. Urakawa, S. Kureti, A. Baiker, Phys. Chem. Chem. Phys. 2008, 10, 6190. [46] P. Barczynski, Z. Degaszafran, M. Szafran, J. Chem. Soc. Perkin Trans. 2 1987, 901. [47] F. Ito, Chem. Phys. 2011, 382, 52. [48] P. Larkin, Infrared and Raman Spectroscopy: Principles and Spectral Interpretation, 1st. ed., Elsevier, Waltham, 2011. [49] R. A. Olsen, D. Borchardt, L. Mink, A. Agarwal, L. J. Mueller, F. Zaera, J. Am. Chem. Soc. 2006, 128, 15594.

Received: August 19, 2013 Published online on December 30, 2013

1309

 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim