Why are α-hydroxycarboxylic acids poor chiral modifiers for Pt in the hydrogenation of ketones?

Article abstract of DOI:10.1016/j.jcat.2005.11.009

Mandelic acid (1) and its derivatives and 1-naphthylglycolic acid (8) were used as modifiers for the asymmetric hydrogenation of ketopantolactone to pantolactone on supported Pt, Ru, and Rh catalysts. A systematic variation of the modifier structure showed that 8 was the best modifier, affording up to 28% ee on Pt; the other metals were barely efficient. The catalytic studies were completed with FTIR spectroscopic analysis of the modifier-modifier and modifier-substrate interactions in solution. Attenuated total reflection infrared spectroscopy was applied to study the adsorption of the α-hydroxy-carboxylic acid type modifiers on a Pt/Al₂O ₃ model catalyst in the presence of hydrogen. The study indicates that the modifiers are present on the metal surface as monomers with an internal hydrogen bond between the hydroxyl group and the carboxyl carbonyl group, and the phenyl or naphthalene rings are in a tilted position relative to the surface. The moderate enantioselection is attributed to a hydrogen bond interaction between the carboxylic OH group of the modifier and the carbonyl O atom of the substrate. The poor enantioselectivities attained in ketopantolactone hydrogenation may be explained by the overly weak adsorption of the modifiers on Pt and by the weakness of the OHO-type H bond between the substrate and the modifiers. It seems that the carboxyl group has a double role, allowing interaction with the substrate and the Pt surface.

Full text of DOI:10.1016/j.jcat.2005.11.009

Journal of Catalysis 237 (2006) 230–236
www.elsevier.com/locate/jcat
Why are α-hydroxycarboxylic acids poor chiral modifiers for Pt
in the hydrogenation of ketones?
Mihaela Maris, Davide Ferri, Lucia Königsmann, Tamas Mallat, Alfons Baiker ∗
Department of Chemistry and Applied Biosciences, Swiss Federal Institute of Technology, ETH Zürich, CH-8093 Zurich, Switzerland
Received 17 October 2005; revised 8 November 2005; accepted 11 November 2005
Available online 15 December 2005
Abstract
Mandelic acid (1) and its derivatives and 1-naphthylglycolic acid (8) were used as modifiers for the asymmetric hydrogenation of ketopantolac-
tone to pantolactone on supported Pt, Ru, and Rh catalysts. A systematic variation of the modifier structure showed that 8 was the best modifier,
affording up to 28% ee on Pt; the other metals were barely efficient. The catalytic studies were completed with FTIR spectroscopic analysis of
the modifier–modifier and modifier–substrate interactions in solution. Attenuated total reflection infrared spectroscopy was applied to study the
adsorption of the α-hydroxy-carboxylic acid type modifiers on a Pt/Al O model catalyst in the presence of hydrogen. The study indicates that
2
3
the modifiers are present on the metal surface as monomers with an internal hydrogen bond between the hydroxyl group and the carboxyl carbonyl
group, and the phenyl or naphthalene rings are in a tilted position relative to the surface. The moderate enantioselection is attributed to a hydrogen
bond interaction between the carboxylic OH group of the modifier and the carbonyl O atom of the substrate. The poor enantioselectivities attained
in ketopantolactone hydrogenation may be explained by the overly weak adsorption of the modifiers on Pt and by the weakness of the O–H–O-type
H bond between the substrate and the modifiers. It seems that the carboxyl group has a double role, allowing interaction with the substrate and the
Pt surface.
2005 Elsevier Inc. All rights reserved.
Keywords: Ketopantolactone; Asymmetric; Enantioselective; Hydrogenation; (R)-Mandelic acid; (S)-Naphthylglycolic acid; FTIR; ATR-IR; Hydrogen bonding;
Pt; Rh; Ru
1. Introduction
anchor the modifier onto the Ni surface, whereas the OH func-
tions can interact via hydrogen bonds with the carbonyl groups
of the reactant.
The two chirally modified metal catalysts most effective in
the hydrogenation of ketones are Ni modified with tartaric acid
[1–4] and Pt modified with cinchona alkaloids [5–12]. Not
only the modifiers, but also the application ranges are very
specific for the metals; Ni is used for the hydrogenation of β-
functionalized and unfunctionalized ketones, whereas Pt is out-
standing in the hydrogenation of activated (α-functionalized)
ketones to the corresponding chiral alcohols. In case of the
Ni/tartaric acid system, substitution of more than one of the hy-
droxyl groups or one of the carboxyl groups of the modifier
leads to an almost complete loss of enantioselection [2,3,13]. It
was suggested that in the hydrogenation of a β-ketoester (e.g.,
methylacetoacetate), the two carboxyl groups are necessary to
Mechanistic studies of cinchona-modified Pt, including the
systematic variation of the structure of cinchonidine [14] and
some other chiral amine modifiers and their derivatives [15–22],
have revealed that the crucial structural parts of the modifiers,
besides the stereogenic center(s), are the flat aromatic ring for
anchoring onto the Pt surface and the basic aliphatic nitrogen
function for interacting with the substrate. Beside amines and
amino alcohols, several other types of chiral compounds have
been tested as modifiers in the Pt-catalyzed hydrogenation of
ketones, but none of them could compete with cinchonidine or
its O-methyl derivative [23–26].
One reaction of special interest for industry is the hydro-
genation of ketopantolactone (KPL) to (R)-pantolactone ((R)-
PL), which is an intermediate in the production of vitamin B5
and a constituent of co-enzyme A [27]. Cinchonidine-modified
*
Corresponding author.
E-mail address: alfons.baiker@chem.ethz.ch (A. Baiker).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.11.009
2005 Elsevier Inc. All rights reserved.

M. Maris et al. / Journal of Catalysis 237 (2006) 230–236
231
(R)-α-hydroxybenzeneacetamide (4) was synthesized from
methyl (R)-mandelate and ammonia, as reported elsewhere
[31]; its structure was confirmed by NMR. (S)-1-Naphthylgly-
colic acid (8) was synthesized from bromoform, 1-naphthylal-
dehyde, and potassium hydroxide with lithium chloride as a cat-
alyst [32,33]. The (S)-1-naphthyl glycolic acid enantiomer was
separated by crystallization, after refluxing a saturated solution
of the racemic mixture with cinchonine in ethanol. Its structure
was identified by NMR, and its enantiomeric purity was veri-
fied by HPLC, using a CHIRALCEL OD column. The NMR
analyses were performed using a Bruker Avance 500 spectrom-
eter.
Scheme 1. Enantioselective hydrogenation of ketopantolactone (KPL) to pan-
tolactone (PL), on supported Pt, Rh and Ru catalysts, modified by α-hydroxy-
carboxylic acids and derivatives.
Dichloromethane (Baker, 99.5%) was dried and stored
over activated molecular sieves. All other solvents were used
as received. The additive 1,8-diazabicyclo[5.4.0]undec-7-ene
(DBU, Aldrich) was used as received.
2.2. Catalytic hydrogenation
The 5% Pt/Al2O3 catalyst (Engelhard 4759) was prereduced
at 400 ◦C for 1 h as described in detail previously [34]. The
same procedure was applied to the 5% Pt/SiO2 (Engelhard, Es-
catTM 2351), Pt/C (Fluka), Rh/Al2O3 (Engelhard 8001 ESCAT
34), and Ru/Al2O3 (Degussa AG H213 B/D 5%) catalysts.
Hydrogenations under standard conditions were carried out
in a parallel pressure reactor system (EndeavorTM) with eight
mechanically stirred 15-mL stainless steel reactors [22]. Hydro-
genations at 60 bar were performed in a 100-mL autoclave with
a 50-mL glass liner and PTFE cover and magnetic stirrer [29].
Total pressure and H2 uptake were controlled by a comput-
erized constant-volume, constant-pressure device (Büchi BPC
9901). For hydrogenation at 15 ◦C, the system was cooled with
a Julabo F25 cryostat.
Under standard conditions, 10 mg of prereduced catalyst,
0.437 mmol of KPL, 19.7 µmol of modifier, and 5 mL of sol-
vent were stirred (500 rpm) at 10 bar and room temperature
(23–25 ◦C), if not otherwise stated. In case of difficulties in dis-
solving the modifier, the slurry was sonicated before the catalyst
was added.
Fig. 1. Structure of the chiral modifiers and the best ee’s achieved at full con-
version, on Pt/Al O .
2
3
Pt afforded 91.6% ee [28]; other amine-type modifiers were less
effective [17,29,30]. A new chiral non–amine-type modifier,
1-naphthyl-1,2-ethandiol gave only 30% ee [26]. The probable
mechanism involves O–H–O-type hydrogen bonds between the
two OH groups and the keto-carbonyl group of KPL. Because
the OH groups are weak hydrogen bond donors, the low enan-
tioselectivity might be due to the weak interaction between the
modifier and the substrate. We assumed that the more protic α-
hydroxy carboxylic acids that function excellently with Ni may
offer better enantioselectivities in the hydrogenation of KPL as
well.
Here we report the study of this reaction (Scheme 1) on sup-
ported Pt, Rh, and Ru catalysts modified by mandelic acid and
its derivatives (Fig. 1). The catalytic experiments have been
completed with IR spectroscopic measurements to aid in un-
derstanding the function of these modifiers.
Conversion and enantioselectivity were determined by an
HP 6890 gas chromatograph, using a Chirasil-DEX CB
(Chrompack 7502; 25 m × 0.25 mm × 250 nm) capillary col-
umn. Conditions were split injection (250 ◦C; 20:1), He carrier
gas (42 cm/s), FID detector (250 ◦C), and 80–180 ◦C column
temperature. The reproducibility of ee was within 0.5%.
2. Experimental
2.1. Materials
2.3. Infrared measurements
Ketopantolactone (KPL; 4,4-dimethyldihydrofuran-2,3-dio-
ne; F. Hoffmann–La Roche AG), (R)-(–)-α-hydroxyphenylace-
tic acid (1; (R)-mandelic acid; Aldrich, 99%), (R)-(–)-α-
methoxyphenylacetic acid (2; Fluka, 99.5%), (R)-(–)-α-hy-
droxyphenylacetic acid methyl ester (3; Aldrich, 99%), (S)-
(+)-2-hydroxypropanoic acid (5; (S)-lactic acid; Lancaster,
99%), (S)-(–)-hydroxysuccinic acid (6; Sigma, 98–100%), and
L-(+)-dihydroxysuccinic acid (7; L-tartaric acid; Across, 99%)
were used as received.
All infrared measurements were performed on a Bruker Op-
tics IFS66 spectrometer after co-addition of 200 scans at a res-
olution of 4 cm−1. FTIR transmission spectra were recorded
using a liquid cell of 1 mm path length equipped with KBr
windows. (R)-(–)-α-hydroxyphenylacetic acid (1) and (R)-(–)-
α-methoxyphenylacetic acid (2) were analyzed at the same
concentrations as those in the reaction mixture (19.7 µmol
modifier in 5 mL CH2Cl2). Additional spectra of (S)-(+)-2-
hydroxypropanoic acid (5) and acetic acid were recorded for

232
M. Maris et al. / Journal of Catalysis 237 (2006) 230–236
comparison. Two series of KPL/1 mixtures at molar ratios rang-
ing from 0.25 to 2 were analyzed in CH2Cl2 at constant sub-
strate concentration and constant acid concentration. The neat
solvent served as the reference for the spectra of the solutes.
Adsorption of 1, 5, and 8 on Pt/Al2O3 was monitored in
situ using attenuated total reflection infrared (ATR-IR) spec-
troscopy. The Pt/Al2O3 model films (1 nm Pt, 100 nm Al2O3)
were prepared and characterized as reported elsewhere [35].
The experimental procedure for adsorption studies is analogous
to that used for other chiral modifiers [36]. Briefly, the metal
surface was cleaned using H2-saturated solvent before solute
adsorption. A spectrum was recorded after 10 min on stream,
which served as background for the next spectra. Then a solu-
tion of the acid was admitted to the thermostatted stainless steel
cell, and ATR-IR spectra were recorded as a function of time.
The solute was replaced by H2-saturated solvent after about
50 min on stream to remove weakly adsorbed and dissolved
species. All measurements were carried out at a temperature of
20 ◦C, a flow rate of 1 mL/min, and an acid concentration of
1 mM (if not otherwise specified).
Table 1
Hydrogenation of KPL to (R)-pantolactone in different solvents, using (R)-
mandelic acid (1) as modifier, under standard conditions
Entry Catalyst
Solvent
Time
(h)
Conver-
sion (%)
ee%
(R)-PL
1
2
3
4
5
6
7
Pt/Al O
Dichloromethane
Dichloromethane
Chloroform
1,2-Dichloroethane
1,2-Dichlorobenzene 0.5
0.6
1.5
1
100
100
78
47
51
20
10
5
5
6
2
3
a
3
b
Trifluorotoluene
Toluene
2
1
86
100
11
13
8
9
10
11
12
13
Pt/SiO
Dichloromethane
Dichloromethane
Chloroform
1,2-Dichloroethane
1,2-Dichlorobenzene
Toluene
3
3
3
3
3
1
100
100
71
14
78
19
20
2
5
6
2
a
100
8
14
15
16
17
18
Pt/C
Dichloromethane
Dichloromethane
Chloroform
1,2-Dichloroethane
1,2-Dichlorobenzene
2
2
2
2
2
100
100
99
87
95
19
6
1
5
2
a
3. Results
19
20
21
22
23
Rh/Al O
Dichloromethane
Dichloromethane
Chloroform
1,2-Dichloroethane
Toluene
2
2
2
4
1
55
54
4
13
100
6
4
2
8
2
3
3
a
3.1. Enantioselective hydrogenation on mandelic
acid-modified Pt-group metals
Racemic
The hydrogenation of KPL to (R)-PL was investigated on
supported Pt, Rh, and Ru catalysts modified with 1; the impor-
tant results of these experiments are summarized in Table 1. Pt
afforded the highest ee (about 20%) in dichloromethane (Ta-
ble 1, entries 1, 8, and 14); the type of support had only a minor
effect on ee. More significant differences were obtained when
using the catalysts without reductive treatment at elevated tem-
peratures (Table 1, entries 2, 9, and 15). Note that the positive
effect of reductive catalyst treatment frequently observed since
the first report by Orito et al. [8] is probably due to cleaning or
restructuring of the metal surface [34,37–39].
Addition of the chiral modifier 1 diminished the reaction
rate. For example, the conversion was 39% after 5 min in the
reaction quoted in Table 1, entry 1, and it increased to 91.5%
in the absence of 1. The lower rate in the presence of 1 is anal-
ogous to the slower hydrogenation of activated ketones on Pt
in the presence of strong acid additives, such as trifluoroacetic
acid [21,40,41].
The enantioselectivity dropped below 10% when replacing
Pt/Al2O3 by Rh/Al2O3 and Ru/Al2O3 which performed even
worse. Only racemic PL was formed in very slow reactions (Ta-
ble 1).
Next, improvement of the enantioselectivity of Pt/Al2O3 was
attempted by varying the conditions. A “chiral” restructuring
of the catalyst by prestirring the slurry containing the prere-
duced Pt/Al2O3 and 1 in dichloromethane for 1 h under N2 or
H2 did not result in the expected improvement [37], but the ee
decreased to 18 and 13%, respectively.
24
25
Ru/Al O
Dichloromethane
Tetrahydrofuran
8
8
<1
7
–
2
Racemic
a
Untreated catalyst.
Partially dissolved modifer.
b
important that dichloromethane is carefully dried before use;
otherwise, even traces of water diminish the enantioselectivity.
For example, addition of 10 µL water to the reaction mixture in
dichloromethane resulted in a decrease of ee from 20% (Ta-
ble 1, entry 1) to 5% under otherwise standard conditions.
Attempts to hydrogenate KPL on Pt/Al2O3 in polar solvents,
including ethanol, tetrahydrofuran, dioxane, water, and acetic
acid, resulted in the complete loss of enantioselectivity. Only
racemic product was formed also on Rh/Al2O3 when the reac-
tion was carried out in nonchlorinated solvents.
The pressure had only a marginal effect on catalyst perfor-
mance. For example, the ee decreased from 20 to 18% (at full
conversion) when the reaction in Table 1, entry 1 was carried
out at 5 bar instead of 10 bar (standard conditions) and no
change was detectable between 10 and 60 bar. Temperatures
above room temperature (23–25 ◦C) had a negative effect on the
enantioselectivity. In the same reaction, at 35 ◦C the ee dropped
from 20 to 9% at full conversion, and no improvement could be
achieved at lower temperatures (e.g., 18.5% ee at 15 ◦C). These
experiments show that 1 is a poor modifier in the hydrogena-
tion of KPL and its application is limited to Pt in weakly polar
solvents.
3.2. Structural effects
Hydrogenation of KPL in the presence of 1 was strongly
solvent-dependent. Enantioselectivity was induced only in
halogenated solvents and toluene, as shown in Table 1. It is
Additional experiments were performed to explore the role
of the three functions of modifier 1—OH, COOH, and the

M. Maris et al. / Journal of Catalysis 237 (2006) 230–236
233
phenyl ring—in the interaction with the substrate and the metal
Table 2
Hydrogenation of KPL to (S)-PL in different solvents, using (S)-1-naphthyl-
glycolic acid (8) as modifier, under standard conditions (20 mg catalyst)
surface. In this study several α-hydroxycarboxylic acids and
derivatives were used as modifiers of Pt in the hydrogenation
of KPL; their structures are presented in Fig. 1. The experiment
shown in Table 1, entry 1 was used as reference reaction. Note
also that (R)-1 and its derivatives afforded always (R)-PL as
the major enantiomer, whereas (S)-PL was produced in excess
when Pt was modified by (S)-1 and its derivatives.
Solvent
Catalyst
Time
(h)
Conver-
sion (%)
ee%
(S)-PL
Dichloromethane
Dichloromethane
1,2-Dichlorobenzene
1,2-Dichloroethane
Isopropanol
Tetrahydrofuran
Toluene
1,2-Dichlorobenzene
Dichloromethane
1,2-Dichloromethane
1,2-Dichlorobenzene
1,2-Tetrahydrofuran
1,2-Dichloromethane
1,2-Dichlorobenzene
Pt/Al O
0.5
1
0.5
2
1.25
1.25
1
0.8
0.6
2.75
1
25
100
56
7
14
24
9
5.5
8
14
5.5
Racemic
Racemic
Racemic
Racemic
Racemic
Racemic
2
3
3
3
3
3
3
3
a
Pt/Al O
2
Pt/Al O
2
Pt/Al O
37
2
Protection of the –OH group of 1 by methylation in mod-
ifier 2 resulted in a drop in enantioselectivity from 20 to 5%
ee. The effect of protecting the carboxyl group by methylation
was even more detrimental; a racemic mixture was obtained
when Pt was modified by (R)-(–)-α-hydroxyphenylacetic acid
methyl ester (3). Obviously, the free carboxyl group is crucial
for enantioselection. This conclusion is supported by the com-
plete loss of enantioselectivity when the reference reaction was
carried out in the presence of 1,8-diazabicyclo-[5.4.0]undec-7-
ene (DBU, 5 µL). The strong bulky base deprotonates 1 and
presumably prevents the interaction of the carboxylate with
the substrate. In contrast, replacement of the carboxyl group
by an amido group in 4 decreased the ee only to 15%. The
likely explanation for this finding is that the amido group of
4 can still interact with the substrate via H-bonding. Similar
to 1, the α-hydroxyamide type modifier 4 was effective only in
dichloromethane; a racemic mixture was formed in any other
solvent.
Pt/Al O
2
100
100
100
100
100
100
11
22
<1
<1
Pt/Al O
2
a
Pt/Al O
2
a
Pt/SiO
Pt/SiO
2
2
a
a
Rh/Al O
2
3
3
3
3
3
a
Rh/Al O
2
Ru/Al O
8
8
8
2
Ru/Al O
2
Ru/Al O
2
a
20 bar, 20 mg catalyst.
higher modifier coverage and fewer free surface sites available
for the hydrogenation reaction.
Further evidence for the stronger adsorption of 8 on Pt was
obtained by studying the nonlinear phenomenon [42–45]. When
the hydrogenation of KPL was carried out with mixtures of the
two modifiers (R)-1 and (S)-8, in all cases 8 controlled the
enantioselectivity. For example, under standard conditions in
dichloromethane, a mixture of 1 and 8 containing only 1 mol%
8 afforded 2% (S)-PL in excess even though the calculated
(theoretical) value was >19% ee to (R)-PL. This difference in-
dicates that the naphthyl ring increased the adsorption strength
of the modifier.
Next we investigated the importance of the phenyl group
of 1 as the potential “anchoring moiety” of the modifier. Af-
ter replacement of the phenyl ring by a methyl group ((S)-lactic
acid, 5), the ee dropped to less than half. Modifier 5 may adsorb
via the carboxyl group, and this adsorption mode is sufficient to
provide a small ee.
3.3. Infrared spectroscopy
Replacement of the phenyl group with a carboxymethyl
group in (S)-(–)-hydroxysuccinic acid (6) resulted in the loss
of enantioselectivity. The hydrogenation was repeated under
standard conditions in other solvents, such as 2-propanol,
tetrahydrofuran, dioxane, and water, that dissolve 6 better than
dichloromethane, but no enantioselection was obtained.
In the presence of tartaric acid (7), which contains two addi-
tional functional groups (OH and COOH) relative to mandelic
acid (1), a very small but significant ee of 2% was obtained in
tetrahydrofuran. Other solvents, including 2-propanol, dioxane,
and water, did not improve enantioselectivity.
Improved enantioselectivity was achieved only when the
phenyl ring of 1 was replaced by a naphthyl group in (S)-1-
naphthylglycolic acid (8). Under optimized conditions (0.86
mmol KPL, 19.8 µmol 8, 20 mL dichlorobenzene, 20 mg re-
duced Pt/Al2O3, 20 bar, room temperature), (S)-PL was pro-
duced with 28% ee. More details on the role of solvent and
catalyst composition are given in Table 2. Clearly, only Pt is
useful in this reaction. The stronger adsorption of 8 on Pt com-
pared to that of 1 is indicated by the enantioselection obtained
in polar solvents, such as tetrahydrofuran and isopropanol, in
which 1 was completely ineffective. The twofold slower conver-
sion of KPL observed by replacing 1 with 8 is probably another
indication of the stronger adsorption of the latter that leads to
Fig. 2 shows the IR spectra of 1, 5, and acetic acid in di-
chloromethane in the ν(O–H) region. The weak band extending
from 3300 to below 2700 cm−1 in the spectra of acetic acid
reveals the formation of intermolecular hydrogen bonds [46].
This band is very weak in 5 and is missing in 1. Therefore,
these spectra indicate that association via both intermolecular
and intramolecular hydrogen bonds is possible, although not
favored, in 5 [47], whereas 1 exists exclusively as monomer in
solution with an internal hydrogen bond between the hydroxyl
group and the carbonyl group of the acid function, in agreement
with previously published data [48] (this part of the spectrum
not shown). This structure could be retained on adsorption on
Pt, as discussed below.
Titration of KPL with 1 did not reveal any significant change
in the carbonyl and in the hydroxyl region, indicating that only
a very weak substrate–modifier interaction exists in solution.
Fig. 3 shows the ATR-IR spectra of 5, 1, and 8 in solu-
tions in contact with the Pt/Al2O3 model film. Similar features
can be found in the transmission spectra of the acid modifiers,
which are shown for 1 and 8 in Fig. 3e and f. The signal at
ca. 2000 cm−1 indicates the formation of CO adsorbed on top,
probably originating from decomposition of the acid or of im-
purities of solvent and solute. An intense signal at 1450 cm−1

234
M. Maris et al. / Journal of Catalysis 237 (2006) 230–236
Fig. 2. Transmission IR spectra of (R)-mandelic acid (1) (—), (S)-lactic acid
(5) (- - -) and acetic acid (—) in CH Cl in the OH region. The concentration
2
2
(3.94 mM) is equivalent to that used in the hydrogenation reactions. The nega-
tive bands are due to incomplete solvent compensation.
that is not seen in the spectrum of the neat acids (top panel
of Fig. 2) is assigned to the symmetric stretch mode of a car-
boxylate group [νS(OCO)], suggesting that the acids adsorb
partly as anion on Pt. The broad feature extending between
1800 and 1600 cm−1 is probably due to an overlap of a weak
signal of CO adsorbed in the bridged geometry, which is more
evident toward higher frequencies in 5 and signals of unpro-
tonated acid molecules, in comparison with the transmission
spectra. The presence of these entities is corroborated by the
signal at ca. 1145 cm−1 [δ(O–H)acid], which should be silent
if only deprotonated species were present on Pt. Other signals
in the 1600–1400 cm−1 spectral region indicate that the phenyl
and naphthyl rings are tilted with respect to the metal surface,
either adsorbed or hanging toward the solution. The signal at
ca. 1340 cm−1 belongs to the deformation mode of the alco-
holic –OH, which is coupled with C–H in plane deformation
modes of the ring in the aromatic acids. The signal indicates
that the intramolecular hydrogen bond characteristic of 1 is re-
tained on adsorption.
All of the aforementioned signals display little loss of in-
tensity on replacement of the solute with neat solvent. Thus
the spectra shown in Fig. 3 are characteristic of adsorbed
species. Inspecting these spectra reveals that at least two ad-
sorption modes are feasible through interaction of the carboxyl
group with the surface, thus affording carboxyl and carboxylate
species, as mentioned earlier. This group could be the domi-
nant anchoring point to the surface in the carboxylate species.
The spectra suggest that orientation of the aromatic moiety is
Fig. 3. In situ ATR-IR spectra of (b) (S)-lactic acid (5), (c) (R)-mandelic acid
(1) and (d) (S)-naphthylglycolic acid (8) adsorbed on Pt/Al O in the presence
2
3
of H and CH Cl solvent. Spectrum (a) represents the state of the Pt surface
2
2
2
before adsorption. Spectra (e) an (f) are the transmission IR spectra of a solution
◦
of 1 and 8 in CH Cl , respectively. Conditions: C
acid
= 1 mM, 20 C.
2
2
not parallel to Pt, although signals of an absolutely parallel ring
would not be allowed by the surface selection rule [49]. This
suggests the presence of species with the aromatic moiety ad-
sorbed tilted or pendant. The variety of adsorption modes is
likely given by the high surface coverage [50] obtained with
the relatively high concentration used in the ATR-IR measure-
ments. However, experiments at tenfold-lower concentrations
released equivalent spectra with less contribution from those
below 1300 cm−1 due to solution-like species (carboxylic acid).
Additional information can be extracted from the spectra
shown in Fig. 3. In contrast to the ATR-IR spectra of strongly
adsorbing chiral modifiers, such as the cinchona alkaloids
[36,51], the present spectra do not display any significant neg-
ative signal at ca. 1400 cm−1. This signal is likely indicative of
the adsorption strength of the adsorbate because it is assigned
to the removal of other hydrocarbon surface species resulting
from decomposition of dichloromethane (solvent) on Pt during
the 10-min cleaning of the Pt/Al2O3 film in hydrogen [36]. In
the case of cinchona alkaloids, adsorption is dictated primar-
ily by the nearly “flat” adsorbing quinoline moiety and secon-
darily by the quinuclidine moiety [52], with a strong negative

M. Maris et al. / Journal of Catalysis 237 (2006) 230–236
235
Fig. 5. Probable interaction of ketopantolactone (a) with 1 and 8, and (b) with 4.
and naphthylglycolic acid 8. Under optimized conditions, KPL
was hydrogenated on Pt/Al2O3 with only 20 and 28% ee, re-
spectively. Structural variations in the modifiers revealed that
the carboxyl group is fundamental for enantioselection and ad-
sorption. We assume that this function is responsible for the
substrate–modifier interaction, presumably via a weak O–H–O-
type H-bond toward the keto-carbonyl O atom of KPL.
Catalytic and spectroscopic measurements showed that ad-
sorption of 1 on Pt was weak, and that replacing the phenyl ring
with a naphthyl ring in 8 led to stronger adsorption. Hence, the
better ee achieved with 8 may be attributed to its stronger ad-
sorption, although the increasing steric hindrance induced by
the extended aromatic system may also contribute to the dif-
ference. ATR-IR analysis showed that the aromatic rings of 1
and 8 adopted a tilted position relative to the Pt surface and that
even 8 adsorbed weakly on Pt compared with the most effec-
tive modifier cinchonidine. On the basis of the ATR-IR spectra
and a comparison with effective chiral amine-type modifiers,
adsorption of 1 and 8 via the carboxyl (or carboxylate) function
likely is more important than that via the aromatic rings.
We assume that another major reason for the poor perfor-
mance of α-hydroxy-carboxylic acids as modifiers of Pt is their
weak interaction with the ketone substrate. This weak interac-
tion is the likely reason for the complete loss of enantioselec-
tivity when Pt was modified by 1 in polar solvents that allow
H-bond interactions and can compete with the substrate. (An-
other feasible explanation for this is related to their stronger
adsorption on Pt.) Mandelic acid, which is present exclusively
as intramolecular hydrogen bonded species in solution (with
this structure probably retained on adsorption on Pt), interacts
very weakly with the substrate. A feasible modifier–substrate
interaction is depicted schematically in Fig. 5. The H-bond type
interaction is supported by the observation that deprotonation of
the carboxyl group of 1 by DBU or transformation to an ester
group (3) hindered the enantioselection, but that replacement
of the carboxyl group by an amido group (still amenable to H
bonding) had only a small negative effect on ee.
Fig. 4. In situ ATR-IR spectra of (a) (R)-mandelic acid (1) adsorbed on Pt
and (b) of a solution containing excess DBU in contact with the same pre-
equilibrated Pt surface. Spectrum (c) corresponds to the state of the surface after
◦
removing the solution containing DBU. Conditions: C
= 1 mM, 20 C.
acid
signal observed at 1400 cm−1. Similarly, adsorption of (R)-
2-(1-pyrrolidinyl)-1-(1-naphthyl)ethanol (PNE) predominantly
via the naphthyl ring at a concentration 100 times lower than
that in the present study afforded the same band without further
signal attributable to ring modes [53], thus indicating that the
naphthyl ring of PNE adsorbs strongly on Pt parallel to the sur-
face. In contrast, the spectra shown in Fig. 3 suggest that despite
the aromatic ring, the chiral acids do not adsorb strongly on Pt.
The stability of adsorbed 1 is greatly affected by adding a
strong base, such as DBU, as indicated by the catalytic data.
Fig. 4b shows that admission of an H2-saturated solution of
DBU in large excess with respect to the chiral acid to an acid
pre-equilibrated metal surface induces desorption of a major
fraction of 1. This is best seen in the spectrum obtained af-
ter changing again to H2-saturated solvent to remove dissolved
and weakly adsorbed species [Fig. 4c]. The strong signal at
1641 cm−1 indicates protonation of DBU [54], occurring only
due to adsorbed acid. DBU not only displaces the chiral acid
from the metal surface, but also adsorbs strongly and thus re-
moves the hydrocarbon fragments present on Pt, as indicated
5. Conclusion
by the negative signal at 1400 cm−1
.
Various α-hydroxy-carboxylic acids and their derivatives
were tested as chiral modifiers of Pt in the hydrogenation of ke-
topantolactone to pantolactone. The <30% ee’s achieved after
a limited optimization are far below those values attainable by
chiral amine-type modifiers. The probable reason for the low
4. Discussion
The two best modifiers among the various α-hydroxy-
carboxylic acids and derivatives tested here are mandelic acid 1

236
M. Maris et al. / Journal of Catalysis 237 (2006) 230–236
efficiency is that in α-hydroxy-carboxylic acids the “anchor-
ing” and “interacting” functions are not separated; that is, the
carboxyl group plays a dual role in adsorption on Pt and in-
teraction with the substrate. This is a fundamental difference
between these modifiers and the known good amine and amino
alcohol-type chiral modifiers of ketone hydrogenation, such as
cinchona alkaloids and 1-(1-naphthyl)ethylamine derivatives,
in which the aromatic ring is responsible for adsorption and the
amine function is responsible for interacting with the ketone.
[20] C. Exner, A. Pfaltz, M. Studer, H.U. Blaser, Adv. Synth. Catal. 345 (2003)
1253.
[21] E. Toukoniitty, I. Busygin, R. Leino, D.Y. Murzin, J. Catal. 227 (2004)
210.
[22] R. Hess, A. Vargas, T. Mallat, T. Bürgi, A. Baiker, J. Catal. 222 (2004)
117.
[23] A. Tungler, T. Tarnai, T. Mathe, J. Petro, J. Mol. Catal. 70 (1991) L5.
[24] E. Sipos, A. Tungler, I. Bitter, J. Mol. Catal. 198 (2003) 167.
[25] G. Szöllösi, C. Somlai, P.T. Szabó, M. Bartók, J. Mol. Catal. A: Chem. 170
(2001) 165.
[26] A. Marinas, T. Mallat, A. Baiker, J. Catal. 221 (2004) 666.
[27] R. Schmid, Chimia 50 (1996) 110.
Acknowledgments
[28] M. Schürch, N. Künzle, T. Mallat, A. Baiker, J. Catal. 176 (1998) 569.
[29] S. Diezi, A. Szabó, T. Mallat, A. Baiker, Tetrahedron: Asymmetry 14
(2003) 2573.
[30] S. Diezi, T. Mallat, A. Szabó, A. Baiker, J. Catal. 228 (2004) 162.
[31] E.J. Ebbers, G.J.A. Ariaans, A. Bruggink, B. Zwanenburg, Tetrahedron:
Asymmetry 10 (1999) 3701.
Financial support from the Swiss National Foundation and
the Foundation Claude and Giuliana is gratefully acknowl-
edged.
[32] E.L. Compere, J. Org. Chem. 33 (1968) 2565.
[33] N. Momiyama, H. Yamamoto, J. Am. Chem. Soc. 127 (2005) 1080.
[34] M. Schürch, O. Schwalm, T. Mallat, J. Weber, A. Baiker, J. Catal. 169
(1997) 275.
[35] D. Ferri, T. Bürgi, A. Baiker, J. Phys. Chem. B 105 (2001) 3187.
[36] D. Ferri, T. Bürgi, J. Am. Chem. Soc. 123 (2001) 12074.
[37] R. Hess, F. Krumeich, T. Mallat, A. Baiker, Catal. Lett. 92 (2004) 141.
[38] H.U. Blaser, H.P. Jalett, D.M. Monti, J.T. Wehrli, Appl. Catal. 52 (1989)
19.
References
[1] T. Sugimura, S. Nakagawa, A. Tai, Bull. Chem. Soc. Jpn. 75 (2002) 355.
[2] T. Osawa, T. Harada, O. Takayasu, Top. Catal. 13 (2000) 155.
[3] A. Tai, T. Sugimura, in: D.E. De Vos, I.F.J. Vankelekom, P.A. Jacobs
(Eds.), Chiral Catalyst Immobilization and Recycling, Wiley–VCH, Wein-
heim, 2000, p. 173.
[4] Y. Izumi, Adv. Catal. 32 (1983) 215.
[39] H. Lieske, G. Lietz, H. Spindler, J. Volter, J. Catal. 81 (1983) 8.
[40] M. von Arx, T. Bürgi, T. Mallat, A. Baiker, Chem.-Eur. J. 8 (2002) 1430.
[41] B. Török, K. Balázsik, K. Felföldi, M. Bartók, Stud. Surf. Sci. Catal. 130
(2000) 3381.
[42] W.R. Huck, T. Bürgi, T. Mallat, A. Baiker, J. Catal. 216 (2003) 276.
[43] L. Balazs, T. Mallat, A. Baiker, J. Catal. 233 (2005) 327.
[44] K.E. Simons, P.A. Meheux, A. Ibbotson, P.B. Wells, Stud. Surf. Sci.
Catal. 75 (1993) 2317.
[5] A. Baiker, J. Mol. Catal. A: Chem. 115 (1997) 473.
[6] H.U. Blaser, H.P. Jalett, M. Muller, M. Studer, Catal. Today 37 (1997)
441.
[7] P.B. Wells, A.G. Wilkinson, Top. Catal. 5 (1998) 39.
[8] Y. Orito, S. Imai, S. Niwa, J. Chem. Soc. Jpn. (1979) 1118.
[9] E. Toukoniitty, P. Maki-Arvela, M. Kuzma, A. Villela, A.K. Neyestanaki,
T. Salmi, R. Sjoholm, R. Leino, E. Laine, D.Y. Murzin, J. Catal. 204
(2001) 281.
[45] A. Tungler, K. Fodor, T. Máthé, R.A. Sheldon, Stud. Surf. Sci. Catal. 108
(1997) 157.
[46] D. Hadzi, S. Detoni, The Chemistry of Acid Derivatives, Wiley, New York,
1979, p. 212.
[47] G. Cassanas, M. Morssli, E. Fabregue, L. Bardet, J. Raman Spectrosc. 22
(1991) 409.
[10] J. Zhang, X.P. Yan, H.F. Liu, J. Mol. Catal. A 175 (2001) 125.
[11] M. Bodmer, T. Mallat, A. Baiker, in: F.E. Herkes (Ed.), Catalysis of Or-
ganic Reactions, Dekker, New York, 1998, p. 75.
[12] B. Török, K. Felföldi, K. Balàzsik, M. Bartók, Chem. Commun. (1999)
1725.
[13] R. Raval, CATTECH 5 (2001) 12.
[48] M. Nobuo, A. Yuzuru, I. Toshiko, T. Yojiro, Bull. Chem. Soc. Jpn. 42
(1969) 482.
[14] H.U. Blaser, H.P. Jalett, W. Lottenbach, M. Studer, J. Am. Chem. Soc. 122
(2000) 12675.
[49] R.G. Greenler, J. Chem. Phys. 44 (1966) 310.
[50] C.L. Brosseau, M.S. Maurice, S.L. Bearne, S.G. Roscoe, Electrochim.
Acta 50 (2005) 1289.
[51] N. Bonalumi, A. Vargas, D. Ferri, T. Bürgi, T. Mallat, A. Baiker, J. Am.
Chem. Soc. 127 (2005) 8467.
[52] A. Vargas, D. Ferri, A. Baiker, J. Catal. 236 (2005) 1.
[53] D. Ferri, T. Bürgi, A. Baiker, unpublished data.
[54] D. Ferri, T. Bürgi, A. Baiker, J. Chem. Soc., Perkin Trans. 2 (2002) 437.
[15] K.E. Simons, G.Z. Wang, T. Heinz, T. Giger, T. Mallat, A. Pfaltz, A. Bai-
ker, Tetrahedron: Asymmetry 6 (1995) 505.
[16] M. Schürch, T. Heinz, R. Aeschimann, T. Mallat, A. Pfaltz, A. Baiker,
J. Catal. 173 (1998) 187.
[17] E. Orglmeister, T. Mallat, A. Baiker, Adv. Synth. Catal. 347 (2005) 78.
[18] B. Minder, M. Schurch, T. Mallat, A. Baiker, T. Heinz, A. Pfaltz, J. Ca-
tal. 160 (1996) 261.
[19] M. Bartók, M. Sutyinszki, K. Felföldi, J. Catal. 220 (2003) 207.