Enantioselective hydrogenation of ketopantolactone

Article abstract of DOI:10.1006/jcat.1997.1674

The enantioselective hydrogenation of ketopantolactone to R-(-)-pantolactone was investigated on 5 wt% Pt/Al₂O₃ chirally modified with cinchonidine. The influence of catalyst pretreatment conditions, hydrogen pressure, temperature, solvent polarity, and catalyst, reactant, and modifier concentrations was studied in a slurry reactor. An enantiomeric excess (ee) of 79% at full conversion was achieved in toluene after optimization of pressure, temperature, and amount of modifier. Good ee could be obtained only after rigorous removal of traces of oxygen and water during catalyst pretreatment and from the hydrogenation reaction mixture. Molecular modeling studies (performed using molecular mechanics, semiempirical, and ab initio methods) provided a feasible structure for the diastereomeric transition complex formed between cinchonidine and ketopantolactone and an explanation for the observed enantiodifferentiation in apolar medium. The calculations indicate that formation of the complex affording R-(-)-pantolactone is energetically favored with cinchonidine, whereas the near enantiomer cinchonine favors S-pantolactone, in agreement with experimental observations. Interestingly, in apolar solvents, where the alkaloid modifier is not protonated, the modeling suggests similar structures for the diastereomeric transition complexes for the hydrogenation of ketopantolactone and methyl pyruvate.

Full text of DOI:10.1006/jcat.1997.1674

JOURNAL OF CATALYSIS 169, 275–286 (1997)
ARTICLE NO. CA971674
Enantioselective Hydrogenation of Ketopantolactone
�
�
,
�
1,
�
M. Sch u¨ rch, O. Schwalm, † T. Mallat, J. Weber,† and A. Baiker
�
Department of Chemical Engineering and Industrial Chemistry, ETH Zentrum, CH-8092 Z u¨ rich, Switzerland; and †Departement of Physical
Chemistry, University of Geneva, 30 quai Ernest-Ansermet, CH-1211 Geneva 4, Switzerland
Received September 4, 1996; revised February 10, 1997; accepted February 24, 1997
The enantioselective hydrogenation of ketopantolactone to
R-(� )-pantolactone was investigated on 5 wt% Pt/Al2O3 chirally
modified with cinchonidine. The influence of catalyst pretreatment
conditions, hydrogen pressure, temperature, solvent polarity, and
catalyst, reactant, and modifier concentrations was studied in a
slurry reactor. An enantiomeric excess (ee) of 79% at full conver-
sion was achieved in toluene after optimization of pressure, tem-
perature, and amount of modifier. Good ee could be obtained only
after rigorous removal of traces of oxygen and water during cata-
lyst pretreatment and from the hydrogenation reaction mixture.
Molecular modelingstudies(performed usingmolecular mechanics,
semiempirical, and ab initio methods) provided a feasible structure
for the diastereomeric transition complex formed between cinchoni-
dine and ketopantolactone and an explanation for the observed
enantiodifferentiation in apolar medium. The calculations indicate
that formation of the complex affording R-(� )-pantolactone is en-
ergetically favored with cinchonidine, whereas the near enantiomer
cinchonine favors S-pantolactone, in agreement with experimental
observations. Interestingly, in apolar solvents, where the alkaloid
modifier is not protonated, the modeling suggests similar structures
for the diastereomeric transition complexes for the hydrogenation
SCHEME 1
in �-position, a structure similar to that of the most stud-
ied �-ketoesters. A heterogeneous catalytic method for
the production of PL was patented almost a decade ago
(
15), but the best enantiomeric excess (ee) of 52% with a
cinchonidine-modified Pt/C system is far below the acces-
sible values of up to 95% ee for ethyl pyruvate hydrogena-
tion (13).
The homogeneous catalytic route has attracted more in-
terest in the past. Various rhodium and ruthenium com-
plexes (16–22) have been successfully tested as catalysts in
this hydrogenation reaction with ee’s up to 99% under very
mild conditions (20).
of ketopantolactone and methyl pyruvate.
�c 1997 Academic Press
INTRODUCTION
To obtain more information about this important and in-
teresting reaction and to broaden our general knowledge
on heterogeneously catalyzed enantioselective reactions,
we reinvestigated this reaction using a cinchona-modified
Pt/alumina catalyst.
In the past few years the heterogeneous enantioselective
hydrogenation of �-ketoesters has become a field of consid-
erable interest for several research groups. The most stud-
ied reaction is the hydrogenation of ethyl and methyl pyru-
vate over supported Pt catalysts modified with cinchona
alkaloids (1–6), some other alkaloids (7), or simple chiral
N compounds (8–11). Only a few investigations have been
made with other �-ketoesters such as methyl benzoylfor-
mate (12) and ethyl 4-phenyl-2-oxobutyrate (13).
EXPERIMENTAL
1
0,11-Dihydrocinchonidine (HCD) (Scheme 1) was ob-
tained byhydrogenation under atmosphericpressure ofcin-
chonidine (CD, 98% Fluka) in dilute aqueous HCl over
a 5 wt% Pd/alumina catalyst. Under standard conditions,
ketopantolactone (KPL, 97% Aldrich) was dissolved in
toluene (Fluka) and dried by azeotropic distillation before
each reaction. A 5 wt% Pt/alumina catalyst (Engelhard
The enantioselective hydrogenation of ketopantolactone
(
KPL, dihydro-4,4-dimethyl-2,3-furandione) to R-(� )-pan-
tolactone (PL, 2-hydroxy-3,3-dimethyl-� -butyrolactone) is
an industrially important process (Scheme 1). PL is a key
intermediate in the synthesis of pantothenic acid (14). In
KPL the keto carbonyl group is activated by an ester group
4
759) was used for all experiments. The metal dispersion
after heat treatment was 0.27 as calculated from the TEM
1
To whom correspondence should be addressed. Fax: (+41-1) 632 11 63. images (23).
75
2
0021-9517/97 $25.00
Copyright �c 1997 by Academic Press
All rights of reproduction in any form reserved.

¨
SCHURCH ET AL.
2
76
Catalyst pretreatment was performed in a fixed-bed
�
1
reactor by flushing the catalyst with 12.5 ml min ni-
trogen (99.995% N2, O2 < 10 ppm, H2O < 10 ppm) at
6
3
3
73 K for 30 min followed by a reductive treatment in
0 ml min hydrogen (99.999% H2, O2 < 2 ppm, N2 <
ppm, H2O < 5 ppm, CnHm < 0.5 ppm) for another 90 min.
�
1
The catalyst was then cooled to room temperature either in
hydrogen or in nitrogen and transferred into the autoclave
under exclusion of oxygen. The catalyst was first contacted
with the solvent containing the proper amount of modifier.
The hydrogenation of KPL was carried out in a 100-ml
stainless-steel autoclave (Baskerville) with a 50-ml glass
liner and PTFE cover, to provide clean conditions. Un-
der standard conditions, 150 mg pretreated catalyst (cooling
in hydrogen after pretreatment), 18 mg HCD, 0.5 g KPL,
and 20 ml toluene were used at 70 bar and 285 K. The
atmospheric pressure hydrogenations were performed in
a 100-ml magnetically mixed glass reactor connected to a
glass burette. The reaction was followed by taking samples
and analyzing with GC.
SCHEME 2
ization, the sample was reduced in hydrogen at 673 K for
0 min and evacuated at the same temperature at 10 Pa
�
3
9
for 30 min.
Molecular modeling for the conformational analysis of
CD (and also the complexes formed between cinchoni-
dine and the reactant) was performed using the MM2 force
field as implemented in the HyperChem package (24).
The charges of CD atoms were calculated using the AM1
semiempirical molecular orbital method (25). In this anal-
ysis all the geometrical parameters of the modifier were
optimized without any constraint except the two dihedral
angles, T1 and T2 (Scheme 2), which were used as param-
Reaction temperatures in figures and tables represent
the controlled temperature of the bath into which the au-
toclave was immersed. The pressure was held at a constant
value by computerized constant volume–constant pressure
equipment (B u¨ chi BPC 9901). In this system a standard
volume of hydrogen is injected when the pressure inside
the reactor drops below a preset value. By monitoring the
pressure inside the vessel and the injected pulses, the hydro-
gen uptake can be followed. The reactor was magnetically
�
�
�
eters and varied from � 180.0 to +180.0 by steps of 10.0 .
All the calculations were performed on a PC, applying the
default values of the software. The CD conformers were
then fully optimized by the ab initio method using the 3-21G
basis set at the SCF level (26). The Hondo 8.5 program (27)
running on an IBM SP2 parallel machine was used for this
purpose. As such calculations of the relative energies of
conformers of organic systems do not generally need in-
troduction of electron correlation nor the use of extensive
basis sets, one can estimate the error bars on such energies
as about � 1 kcal/mol. The same error probably applies to
MM2 energies, as the parameters of this force field have
been consistently derived so as to perform accurate evalu-
ations of the relative energies of conformers.
�
1
mixed (n = 1000 min ). It was proved that under standard
conditions the rate of hydrogen uptake was independent of
�
1
the mixing frequency between 500 and 1000 min
.
ee and conversion were determined with an HP 5890A
gas chromatograph with a chiral WCOT cyclodextrin-�-
2
,3,6-M-19 (Chrompack) capillary column. The enantiose-
lectivity is expressed as ee (% ) = 100 � (|[R] � [S])|/([R] +
S]). ee values were determined at full conversion, if not
[
otherwise stated.
Catalyst samples for transmission electron microscopy
(
TEM) were suspended into hexane and loaded onto per-
forated, thin carbon films supported on copper grids. High-
resolution TEM was carried out on a Philips CM30ST at
3
00 kV, with a point resolution of 0.19 nm. Particles were
measured from photographic enlargements of the micro-
graphs. The most important parameter obtained from the
size distribution is the mean diameter from the surface area-
weighted particle size data, hdia defined by
RESULTS
Influence of Catalyst Pretreatment
s
It hasbeen proposed (15) that before the enantioselective
hydrogenation of KPL takes place the Pt catalyst should be
modified with CD in an ethanolic solution at reflux temper-
ature. This modified 5 wt% Pt/C catalyst provided 36% ee in
P
ni di2
hdia =
P
.
[1]
di
CO chemisorption on Pt/alumina was carried out in a Mi- benzene at 60bar (15). The application ofthisprocedure un-
cromeritics ASAP 2000 chemisorption apparatus, at 308 K der partly different conditions (5 wt% Pt/alumina instead of
and a pressure ranging from 0 to 45 kPa. Prior to character- 5 wt% Pt/C, toluene instead of benzene, and 20 bar instead

ENANTIOSELECTIVE HYDROGENATION OF KETOPANTOLACTONE
277
TABLE 1
Role of Catalyst Pretreatment and Added Water in the Enantioselective Hydrogenation of Ketopantolactonea
Catalyst
pretreatment
H2 pressure
(bar)
[reactant]/
[modifier]
Time
(h)
Conversion
(% )
ee
(% )
Entry
1
2
3
4
5
6
7
8
Reflux in EtOH/CDb
Reflux in EtOH/CD
H2 at 673 K, cooling in N2
No pretreatment
H2 at 673 K, cooling in N2
H2 at 637 K, cooling in H2
H2 at 673 K, cooling in H2c
60
20
20
90
90
90
90
90
—
—
2
3
1
1
1
1
1
—
80
36
21
56
25
52
68
76
40
240
240
480
480
480
480
480
100
100
100
100
100
100
c,d
H2 at 673 K, cooling in H2
a
Reactions were performed in toluene at 285 K using 50 mg of Pt/Al2O3.
Solvent: benzene instead of toluene, data taken from Ref. (15).
b
c
d
1
56 mg catalyst.
Addition of 2.8 mmol water.
of 60 bar) resulted in only 21% ee (Table 1.) A change vorable for the enantioselectivity, ashasbeen demonstrated
of the modifier from CD to HCD had no significant effect previously (3).
on the ee (deviation less than 1% ). Besides, there was no
detectable variation of ee above 30% conversion.
An important part of the pretreatment procedure is the
cooling of the catalyst after reduction in hydrogen. Cooling
When searching for reasons of poor enantioselectivity,
it must be considered that ethanol is not an inert solvent
in carbonyl hydrogenations over Pt (28, 29). Destructive
adsorption of ethanol (and other primary alcohols) pro-
duces CO- and CxHy-type fragments on the Pt surface (29,
3
0). In addition, ethanol remaining on the catalyst after
the pretreatment procedure forms hemiketal from the acti-
vated keto group of the reactant. The hemiketal formation
is catalyzed by the basic quinuclidine N atom of CD (1, 28).
Accordingly, the catalyst pretreatment in refluxing
ethanol was replaced by a prereduction at elevated tem-
perature in flowing hydrogen, a procedure that has been
found most suitable for �-ketoester hydrogenation (28).
CD was simply added to the catalyst in the solvent used
for the hydrogenation reaction. This change in pretreat-
ment conditions resulted in an increase of 35% in ee, un-
der otherwise identical conditions (Table 1, entries 2 and
3
). Temperature-programmed oxidation and FTIR analy-
sis of the processes occurring during heat treatment of the
catalyst showed that some water, CO2 (adsorbed on the alu-
mina support), and organic impurities were removed from
the surface (31). Removal of surface impurities is one fea-
sible explanation for the enhancement of ee after thermal
pretreatment (Table 1).
Another important influence of thermal treatment is in-
dicated by TEM investigation of Pt/alumina before and
after heat treatment. As illustrated in Fig. 1, the average
particle size increases during thermal treatment at 673 K
by ca. 40% , and the calculated dispersion drops from 0.38
to 0.27. For comparison, a platinum dispersion of 0.24 was
found with CO chemisorption for the pretreated catalyst.
The formation of relatively large, flat crystalline Pt parti-
FIG. 1. Frequenciesofthe equivalent circle diameterscalculated from
cles, characteristic of the hydrogen-treated catalyst, is fa- the TEM pictures (a) before and (b) after heat treatment.

¨
SCHURCH ET AL.
2
78
at ambient temperature (after heat treatment and cooling in
hydrogen).
For comparison, in the general pretreatment procedure
applied before the hydrogenation of ethyl pyruvate (8, 28),
the catalyst was always cooled after pretreatment in a ni-
trogen flow. The oxygen contamination of nitrogen had no
detectable negative effect on the ee of the reaction. In con-
trast, stirring the reaction mixture (ethanolic solution) in
the presence of air after the heat treatment and before the
hydrogenation reaction even increased the ee and the reac-
tion rate (28, 32). The latter effect has been explained partly
by the formation of acetic acid from ethanol and subsequent
protonation of the quinuclidine nitrogen of CD (28).
Influence of Pressure
FIG. 2. Influence of pressure on the enantiomeric excess. The catalyst
is cooled in hydrogen (� ) or nitrogen (᭺) after heat treatment (150 mg
catalyst, 0.067 mmol HCD, 3.9 mmol KPL, 285 K, 70 bar H2, toluene).
There is a considerable enhancement of enantioselectiv-
ity with increasing hydrogen pressure, as shown in Fig. 2.
Above 40 bar, ee reaches a plateau. This behavior is simi-
lar to that observed in the hydrogenation of ethyl pyruvate
the catalyst in nitrogen lowers the ee (Table 1, entries 5, 6),
compared with the ee obtained after cooling in hydrogen.
The difference is a function of reaction conditions (hydro-
gen pressure, amount of catalyst), as can be seen in Figs. 2
and 3. Although this effect is not yet fully understood, a
possible explanation of this observation is the chemisorp-
tion of oxygen, contained as impurity in N2 (<10 ppm),
and subsequent formation of PtOx on the Pt surface. Af-
ter the hydrogenation reaction is started in the autoclave,
(
2, 32). The relatively high surface hydrogen concentration
attainable at high hydrogen pressure seems to be favorable
for the enantiodifferentiation. The rate of hydrogen uptake
was not influenced by mass transport (hydrogen dissolution
and diffusion in the liquid phase) when using 150 mg cata-
lyst at 70 bar pressure. Note that working under standard
conditions the reaction rate could not be determined reli-
ably because of the small amount of KPL reactant and, con-
sequently, low hydrogen consumption. A complete kinetic
analysis including optimization of the reaction conditions
is presently underway.
0
PtOx is reduced to Pt and H2O. Contamination of sur-
face Pt atoms with a submonolayer of water can lower
the ee considerably. In a control experiment 50 �l wa-
ter (0.25 vol% related to the solvent) was added to the
reaction mixture and the ee dropped from 76 to 40%
As discussed above, the ee is considerably lower when
the catalyst is cooled in N2 after heat treatment in H2
(
Fig. 2). In addition, the reproducibility of ee under these
(
Table 1, entries 7 and 8). Note that the effect was consider-
conditions is very poor (unusually high standard deviation).
ably smaller, only 5% , when the catalyst was exposed to air
Influence of Catalyst Amount
The role of catalyst amount is shown in Fig. 3. Pt together
with HCD modifier is considered the active and selective
catalyst system; accordingly, the catalyst/modifier ratio was
always kept constant. There is hardly any change in ee
with the amount of catalyst when cooling the catalyst in
hydrogen after pretreatment. A small drop in selectivity
at very low catalyst loading is likely due to experimental
difficulties in transferring the prereduced catalyst to the
autoclave under exclusion of oxygen, as discussed above.
In this region of catalyst loading, ee is extremely low when
the catalyst is coold in nitrogen after prereduction at 673 K.
Influence of Solvent
In the hydrogenation of ethyl and methyl pyruvate with
cinchonidine-modified Pt, ee decreases with increasing sol-
vent polarity (13, 28, 23). Quite similar behavior is observed
FIG. 3. Influence of catalyst amount on the enantiomeric excess after
cooling the catalyst in hydrogen (� ) or in nitrogen (᭺) (catalyst/HCD
weight ratio: 7.5, 3.9 mmol ketopantolactone, 285 K, 70 bar H2, toluene). in the hydrogenation of KPL, as illustrated in Fig. 4. Good

ENANTIOSELECTIVE HYDROGENATION OF KETOPANTOLACTONE
279
Other possible reasons for the very low ee achieved in
polar solvents could be the interaction of solvent with the
modifier, resulting in a less favorable conformation (36–38),
and the strong adsorption of the solvent molecules on the
active sites.
Influence of Modifier
A broad maximum in ee has been observed as a func-
tion of modifier concentration (Fig. 5). Interestingly, this
maximum corresponds to a rather low HCD/Pts (surface
Pt atoms) molar ratio of about 0.065. The corresponding
cinchona alkaloid/Pts molar ratio was found to be 0.5 in
the enantioselective hydrogenation of ethyl pyruvate un-
der similar conditions [toluene, 5 wt% Pt/alumina (39)].
When cinchonine, the near enantiomer of CD, was used,
FIG. 4. Influence of the solvent on the enantiomeric excess listed in
the order of the empirical solvent parameter, ET (34). The experiments
(
S)-pantolactone was formed in excess (ee = 73% ).
N
were carried out with 150 mg catalyst, 0.067 mmol modifier, 3.9 mmol
ketopantolactone, at 285K and 70bar H2. The solventsare (1) cyclohexane,
Influence of Temperature
(
2) hexane, (3) toluene, (4) diethyl ether, (5) THF, (6) acetic acid, (7)
The enantioselectivity reaches its maximum at around
88 K in toluene at 70 bar, as illustrated in Fig. 6. There
ethanol, (8) formamide, (9) water.
2
is no unambiguous explanation yet for the volcano-type
curve. For comparison, in the enantioselective hydrogena-
tion of �-ketoesters, ee increases with decreasing reaction
temperature, reaching a plateau below room temperature
ee can be obtained only in an apolar medium, such as
toluene. In polar solvents, ee drops to zero or close to
it. The trend of decreasing enantioselectivity with increas-
ing solvent polarity is even more pronounced than in the
case of ethyl pyruvate hydrogenation on Pt/alumina (28).
The negative correlation between solvent polarity and ee
is not rigorous. For example, in toluene, ee is higher by
about 25% than the ee measured in diethyl ether, though
the empirical solvent parameters (34) are almost identical
(
2, 32). On the other hand, a maximum in ee as a func-
tion of temperature was observed in the enantioselective
hydrogenation of pyruvic acid oxime (40).
Influence of Reactant Concentration
Reactant concentration in the range 0.05–2 mmol l 1 had
�
for these two solvents. A likely reason for the unexpect- only a minor influence on ee, as illustrated in Fig. 7. At
edly low ee in the most apolar solvents hexane and cyclo- low KPL/catalyst ratios, ee decreased by 3% , likely due
hexane is the limited solubility of KPL and CD. In these to the relatively large contribution of the initial transient
instances, the reactions were performed in the presence of period (increase in surface hydrogen concentration to the
undissolved KPL. Note that when applying the dielectric
constants as values characteristic of solvent polarity, there
are also strong deviations from the linear correlation.
No by-products (from side reactions such as acid- or base-
catalyzed ring opening and dimerization) could be detected
by GC and NMR analysis, when using acetic acid or ethanol
as solvents. Nevertheless, by-product formation in small
amounts and strong adsorption on the metal surface cannot
be excluded. (The detection limit was about 0.5 mol% .)
The influence of solvent polarity may be explained partly
by the change in hydrogen solubility with the empirical sol-
vent parameter. There is a general tendency for the solubil-
ity of hydrogen to decrease with increasing solvent polarity,
independent of pressure (35); however, because of the com-
plex phenomena occurring on the catalyst surface before
steady state is achieved (removal of impurities, competi-
tive adsorption of reaction components), a direct relation
FIG. 5. Influence of modifier (HCD) concentration on the enan-
between hydrogen solubility and enantioselectivity is not
apparent (30).
tiomeric excess (150 mg catalyst, 3.9 mmol ketopantolactone, 285 K, 70 bar
H2, toluene).

¨
SCHURCH ET AL.
2
80
the initial transient period during enantioselective hydro-
genations over Pt and Pd is due mainly to (i) competitive
adsorption of reactant modifier, hydrogen, and solvent and
(
ii) removal of surface impurities.
Optimization
A limited optimization including the most important re-
action parameters (pressure, temperature and amount of
modifier) using the gradient method resulted in an ee of
7
9% . The corresponding conditions are 70 bar and 285 K,
using 140 mg 5 wt% Pt/alumina, 3.8 mmol KPL, 0.18 mg
HCD, and 20 ml toluene. Remarkable is the very small
amount of modifier necessary to obtain high enantioselec-
tivity. It is likely that involvement of other parameters (e.g.,
catalyst nature, solvent) in the optimization process will
provide further improvement in ee.
FIG. 6. Influence of temperature on the enantiomeric excess (150 mg
catalyst, 0.067 mmol HCD, 3.9 mmol ketopantolactone, 70 bar H2,
toluene).
Rate Acceleration by Cinchona Alkaloids
steady-state value and possible removal of surface impuri-
ties by the competitive adsorption of hydrogen; see below).
There is a considerable rate acceleration in the hydro-
genation of KPL in the presence of cinchona alkaloids. An
example illustrating this “ligand acceleration” effect (39)
is shown in Table 2. A relatively high KPL concentration
Influence of Conversion at Atmospheric Pressure
An interesting feature of Pt- and Pd-catalyzed enantio- was applied to obtain reliable hydrogen consumption rate
selective hydrogenation of activated carbonyl compounds data. The initial rate of hydrogen consumption increased
and olefins, respectively, is that both the ee and the reac- by a factor of almost 6 in the presence of HCD. A compar-
tion rate increase in the first part of the reaction (6, 30). In ison of modified and racemic reactions is also made based
contrast, there was no significant change in ee as a function on the overall rate or reaction time. The reaction time nec-
of KPL conversion when the reactor was operated in the essary for 75% conversion of KPL was found to be seven
kinetic regime. A considerable increase in ee (from 20 to times lower in the presence of HCD, as compared with the
3
5% between 5 and 90% conversion) was observed at at- racemic reaction (Table 2). In the enantioselective hydro-
mospheric pressure in toluene. In addition, a maximum in genation of ethyl pyruvate over the same catalyst system,
reaction rate was obtained at 40–50% conversion; however, the rate acceleration was between 4 and 25 depending on
these changes were found to be due to the inefficient hydro- the reaction conditions (2).
gen transport in the system. A practically constant ee was
Molecular Modeling
measured even at 1 bar, when the influence of mass trans-
port was eliminated. It has recently been shown (30) that
The possible mechanism of enantiodifferentiation in the
hydrogenation of KPL was studied by theoretical calcu-
lations. We focused first on the interaction between the
cinchona alkaloid modifier and the activated carbonyl com-
pound reactant, and then on the adsorption of the activated
complex. Both molecular mechanics (MM) and ab initio
calculations were performed to determine the minimum
TABLE 2
Rate Acceleration Observed in the Presence of HCD,
at Close to the Optimum Conditions (150 mg Catalyst,
�
1
9.5 mmol KPL, 25 ml Toluene, 70 bar, 12 C)
HCD
ee
(% )
r0
t75% a
(min)
(mmol h 1(gcat)� 1
�
)
(
mg)
0
0
—
70
420
2470
25
3.5
2
FIG. 7. Influence of reactant concentration on the enantiomeric ex-
a
cess (150 mg catalyst, 0.067 mmol HCD, 285 K, 70 bar H2, toluene).
Reaction time required for 75% KPL conversion.

ENANTIOSELECTIVE HYDROGENATION OF KETOPANTOLACTONE
281
contour plots calculated (MM) as a function of these di-
hedral angles are represented in Fig. 8. Six different en-
ergy minima (dark area) can be identified as a function of
T1 and T2. The numbering of minima 1–4 corresponds to
that used by Dijkstra et al. (37). In the so-called “closed”
conformation the quinuclidine nitrogen is oriented toward
the aromatic rings, whereas in the “open” conformation it
pointsawayfrom the aromaticrings. For illustration see also
Fig. 9.
It is interesting to compare the results of our ab initio cal-
culations with those obtained by Dijkstra et al. using one-
and two-dimensional NMR techniques, combined with MM
and semiempirical molecular orbital calculations (38). Ac-
cording to these methods, the first four minima only were
identified; the open(5) and open(6) conformers were not
found. It is notheworthy that additional calculations per-
formed by us indicated also six energy minima for cincho-
nine (CN), the near enantiomer of CD.
The key question is now which of these six minimum
energy conformations should be chosen for further calcula-
tions. To find the most stable conformer, conformers (1–6)
were optimized without constraints; the corresponding rel-
ative energies (1E) and the dihedral angles T1 and T2 are
collected in Table 3. Both ab initio and MM calculations
indicate that the open(3) conformation is the global min-
imum. Moreover, this conformer corresponds to that ob-
served by Oleksyn using X-ray crystallography (41). The
�
�
calculated T1 and T2 dihedral angles are 100.5 and 156.2
(
� �
ab initio) for open(3) and 101.4 and 158.1 in the X-ray
structure.
Figure 9 displays the three-dimensional space-filling
models of open(3) and closed(1) conformers. These confor-
mations suggested by ab initio calculations were observed
in solution by NMR (38). Closed(1) exhibits the lowest en-
FIG. 8. Isoenergy contour plots calculated (MM) for cinchonidine as
a function of the dihedral angles T1 and T2 (show in Scheme 2). The energy
spacing between the contours is 0.5 kcal/mol.
energy conformation of CD. It is known from the liter- ergy among the closed conformations, but it is still higher in
0
ature (36–38) that rotations around C4 –C9 and C9–C8 energy by 3.3 kcal/mol than open(3). It is clearly seen that
bonds are the most important. The corresponding dihe- in the closed(1) conformation the lone electron pair of the
dral angles T1 and T2 are shown in Scheme 2. Isoenergy quinuclidine N atom points toward the quinoline rings and
FIG. 9. Space-filling models of the two minimum energy conformations closed(1) and open(3). Colors: H, white; C, gray; O, N, black; white “N”
indicates the quinuclidine N atom.

¨
SCHURCH ET AL.
2
82
TABLE 3
It has recently been proposed (42, 43) that one of
the closed conformations of CD would interact with the
�-ketoester reactant, and the � orbitals of the quinoline
ring of CD would stabilize the reactant via �–� overlap-
ping with the conjugated double bond; however, no details
of the MM calculations or any data supporting this “specific
shielding effect” have been provided by the authors. More-
over, the experimental evidence on which the model was
based, namely that ee is close to zero at very low pyruvate
conversion, has recently been proved to be erroneous (30).
The structures and energies of complexes formed on in-
teraction of KPL and CD have been calculated by MM
T1 and T2 Dihedral Angles and the Relative Energies
E of the Six Conformers of Cinchonidine Fully Op-
timized Using the Force Field MM2 and the ab Initio
Method (3–21G Basis Set at the SCF Level)
a
1
Conformer
Closed(1)
MM2
3–21G
T1 (deg)
T2 (deg)
� 106.5
48.8
� 107.1
54.6
1
E (kcal/mol)
1.0
3.3
Closed(2)
T1 (deg)
T2 (deg)
63.1
45.0
1.5
70.1
45.7
6.4
(
MM2 force field). In these calculations all the geometric
parameters have been optimized without any constraint.
The atomic charges have been calculated at the semiempir-
ical level using the AM1 method. Figure 10 shows the cal-
culated minimum energy conformations of the CD–KPL
adduct leading to the formation of (R)- or (S)-PL on hy-
drogenation. In these models, the optimum conformations
of the complexes have been accommodated on an ideal
Pt(111) “surface,” assuming flat adsorption of the aromatic
rigs parallel to the Pt surface over two adjacent Pt atoms.
This plausible orientation of the modifier has been evi-
denced by recent adsorption studies using scanning tun-
neling microscopy for the naphthalene/Pt system (44) and
H/D exchange experiments for HCD/Pt system (45).
1
E (kcal/mol)
Open(3)
T1 (deg)
T2 (deg)
E (kcal/mol)
Open(4)
T1 (deg)
T2 (deg)
E (kcal/mol)
Open(5)
T1 (deg)
T2 (deg)
E (kca/mol)
Open(6)
T1 (deg)
T2 (deg)
93.2
156.4
0.0
100.5
156.2
0.0
1
� 96.6
129.2
4.4
� 94.1
137.5
3.7
1
92.7
� 82.7
3.7
89.5
� 57.1
10.1
1
� 95.4
� 59.4
2.0
� 93.8
� 61.0
7.9
The top views of the space-filling models of complexes
and the corresponding formulas are shown in the lower part
of Fig. 10. In the CD–KPL complex the carbonyl O atom
of KPL is bound to the quinuclidine N atom of alkaloid
via H bonding. The activated complex resembles the half-
hydorgenated state of the carbonyl group (late transition
state). Note that the stabilization of the half-hydrogenated
state of ethyl pyruvate by CD has already been proposed by
Wells and co-workers (46). The complex leading to the for-
mation of the (S)-product on hydrogenation was found to
1
E (kcal/mol)
a
1
E = energy of the conformer related to that of open(3).
it is not accessible to interaction with the carbonyl group of
the reactant, adsorbed nearby on the Pt surface. A further
support to the choice of open(3) is that this is the preferred
conformation in many apolar and protic polar solvents as
found by Dijkstra et al. (38). These considerations led us to
use the open(3) minimum in further calculations.
�
1
be energetically less favorable by 2.2 kcal mol (Table 4).
For comparison, the same calculations have been per-
formed to determine the nature of the interaction between
Application of various derivatives of CD as the source of
chiral information in the enantioselective hydrogenation of
�
-ketoesters indicated that the basic quinuclidine N atom
(
or its protonated form) is responsible for the interaction
TABLE 4
with the reactant (13). The successful application of some
simple chiralaminesasmodifiers(8–11) showsthat the pres-
ence of the 1,2-aminoalcohol structural part in the modifier,
characteristic of cinchona alkaloids, is not a prerequisite for
achieving high ee. It is very likely that the (C9)–O atom of
Interaction Energies Eint Calculated Using the MM2 Force Field
for the Cinchonidine (CD)–Ketopantolactone (KPL), CD–Methyl
a
Pyruvate (MP), and Cinchonine (CN)–MP Interactions
Eint (kcal/mol)
CD is not involved in the reactant–modifier interaction, but
it seems to be important in stabilizing the proper conformer
of CD, adsorbed on Pt. Our previous MM calculations (not
reported here) questioned the feasibility of the formation MP(H) (trans)–CD
of a six-membered ring intermediate involving the quinu-
clidine N and (C9)–O atoms of CD as electron donors to the
Complex
Pro (R)
Pro (S)
1Eb
KPL(H)–CD
� 20.2
� 20.9
� 13.4
� 18.0
� 14.3
� 20.3
2.2
6.6
6.9
MP(H) (trans)–CN
a
The transition complexes resemble the half-hydrogenated state of
reactants, which is indicated by “H” in brackets.
carbonyl C atoms of the reactant, as proposed by Augustine
et al. (5).
b
|Eint(R) � Eint(S)|.

ENANTIOSELECTIVE HYDROGENATION OF KETOPANTOLACTONE
283
FIG. 10. Space-filling models of the transition complexes formed between CD and ketopantolactone which yield (R)- and (S)-PL (left and right,
respectively). Top: side view over an ideal Pt(111) “surface.” Bottom: top view of the space-filling model, together with the formula. Colors as in Fig. 9.
CD and the simplest �-ketoester, methyl pyruvate (MP). the aromatic rings of CD and the two carbonyl groups of
The minimum energy conformation of MP was determined MP are adsorbed parallel to the flat Pt surface. The cal-
by ab initio calculations. In the lowest energy state the two culated interaction energies are collected in Table 4. The
carbonyl groups are in trans configuration. The calculated formation of (R)-lactate is energetically favored when us-
minimum energy conformations leading to the formation of ing CD as modifier, but the (S)-enantiomer is the preferred
(
R)- and (S)-methyl lactates on hydrogenation are depicted product when CN is used in the calculations. The results of
in Fig. 11. The space-filling models are accommodated in these calculations are in agreement with the experimental
side views over a schematic Pt surface, assuming that both observations (1, 32).

¨
SCHURCH ET AL.
2
84
FIG. 11. Space-filling models of the transitions complexes formed between CD and MP which yield (R)- and (S)-lactate (left and right, respectively)
on hydrogenation. Top: side view over an ideal Pt(111) “surface.” Bottom: top view of the space-filling model, together with the formula. Colors as in
Fig. 9.
DISCUSSION
low solvent polarity, and relatively high surface hydrogen
concentration on the enantiodifferentiation. An advanta-
The enantioselective hydrogenation of the cyclic lactone geous feature of the reaction is the rather small amount of
KPL over cinchona-modified Pt/alumina exhibitscharacter- chiral modifier necessary to induce the maximum ee. The
istics similar to those of the hydrogenation of �-ketoesters. calculated ratio of 1 molecule CD to 15 surface Pt atoms
Typical examples are the positive influences of catalyst pre- includes the “inefficient” fraction of alkaloid, i.e., the modi-
treatment at elevated temperature in flowing hydrogen, the fier that is adsorbed on the alumina surface or present in the

ENANTIOSELECTIVE HYDROGENATION OF KETOPANTOLACTONE
285
toluenic solution in steady state. It is very likely that only lactone and the �-ketoester in an apolar medium, where
a minority of surface Pt atoms are covered by CD, even if the alkaloid is not protonated (39).
we consider the relatively large size of active site ensembles
covered by one single modifier molecule.
The hydrogenation of pyruvate esters provides the high-
est ee when the CD modifier is protonated (13, 28). For this
The preliminary optimization of the reaction conditions, case we have already proposed a model for the enantio-
involving only a few important parameters, afforded 79% differentiation based on molecular modeling calculations
ee in toluene, at below room temperature and high hydro- (48, 49). When the protonated CD interacts with the acti-
gen pressure. This value corresponds to an increase of 27% vated carbonyl compound the crucial interaction can again
compared with the only previous report on the enantiose- be described as a hydrogen bonding between the quinucli-
lective reduction of KPL catalyzed by the Pt–CD system. dine N atom and the �-carbonyl O atom.
This improvement is attributed to three main reasons:
CONCLUSIONS
1. The supported Pt is properly preconditioned before
the hydrogenation reaction. Catalyst prereduction at ele-
vated temperature in flowing hydrogen is a simple and ef-
ficient way to provide clean, relatively large, and “flat” Pt
particles preferable for enantiodifferentiation (3, 12).
The present study demonstrates that the hydrogenation
of ketopantolactone to (R)- or (S)-pantolactone can be
performed with good enantioselectivity over cinchonidine-
or cinchonine-modified Pt/alumina, respectively, provided
the catalyst is properly preconditioned and traces of oxy-
gen and water are rigorously removed. It is likely that in-
volvement of a broader range of reaction parameters in
the optimization will result in even higher ee’s, similar
to those characteristic of �-ketoester hydrogenation over
cinchona-modified Pt catalysts. Molecular modeling of the
diastereomeric transition complexes formed between ke-
topantolactone and the modifiers cinchonidine and cincho-
nine indicated that the complex affording (R)-pantolactone
is energetically favored with cinchonidine, whereas the near
enantiomer cinchonine favors (S)-pantolactone formation.
The structures of these transition complexes are strikingly
similar to those calculated for the interaction of these modi-
fiers with methyl pyruvate, the crucial interaction being the
hydrogen bonding between the quinuclidine nitrogen of the
modifier and the oxygen at the �-carbonyl group of the
reactant. Thus, the theoretical studies extend the previous
models suggested for the cinchona alkaloid–pyruvate inter-
action (48), where the modifier exists in protonated form.
2. The use of reactive solvents, such as primary alco-
hols, diols, and some carbonyl compounds (47), has been
avoided. In general, polar solvents including traces of wa-
ter are detrimental to the enantioselectivity.
3. Rigorously oxygen-free conditions have been applied
during catalyst pretreatment and the hydrogenation of
KPL.
The second and third points represent interesting devia-
tions from the characteristics of the well-known enantiose-
lective hydrogenation of methyl and ethyl pyruvates.
Molecular modeling calculations provided a feasible ex-
planation for the observed enantiodifferentiation. These
calculations suggest that in the transition complex the ba-
sic quinuclidine N atom of CD is bound to the carbonyl
group of KPL via a hydrogen bond interaction (N–H–O
connection), stabilizing the half-hydrogenated state of the
reactant (46). Because of this interaction, the formation of
(
R)-pantolactone is energetically favored compared with
that of (S)-pantolactone. When applied to CN, the calcula-
tions indicate the preferential formation of the (S)-product,
in agreement with the experimental observations.
ACKNOWLEDGMENTS
Note that these calculations take into account only the
reactant–chiral modifier interaction. Future studies should
include the important role of the adsorption bonds of the
Financial support of this work by the Swiss National Science Founda-
tion (Program CHiral 2) is gratefully acknowledged. Thanks are also due
R. Wessicken (Department of Solid State Physics, ETH, Z u¨ rich) for the
diastereomeric transition complex. In the present calcula- electron microscopic measurements.
tion we made the simplifiying assumption that the transi-
tion complex adsorbs on Pt in a position that the quinoline
rings of CD and the carbonyl groups of KPL are adsorbed
parallel to a flat (ideal) Pt(111) surface. The interaction be-
tween the solvent and the transition complex, which has not
been considered, is likely to be less important in an apolar
medium.
REFERENCES
1
2
. Orito, Y., Imai, S., and Niwa, S., J. Chem. Soc. Japan 8, 1118 (1979).
. Blaser, H. U., Jalett, H. P., Monti, D. M., Reber, J. F., and Wehrli, J. T.,
Stud. Surf. Sci. Catal. 41, 153 (1988).
. Wehrli, J. T., Baiker, A., Monti, D. M., and Blaser, H. U., J. Mol. Catal.
49, 195 (1989).
3
4
5
6
. Sutherland, I. M., Ibbotson, A., Moyes, R. B., and Wells, P. B., J. Catal.
The calculations have been extended to the more studied
1
25, 77 (1990).
. Augustine, R. L., Tanielyan, S. K., and Doyle, L. K., Tetrahedron Asym.
, 1803 (1993).
. Singh, U. K., Landau, R. N., Yongkui, S., LeBlond, C., Blackmond,
D. G., Tanielyan, S. K., and Augustine, R. L., J. Catal. 154, 91 (1995).
�
-ketoester hydrogenation reaction (pyuvate → lactate).
The ab initio and MM calculations demonstrated the
similarity of the diastereomeric transition complex when
comparing the enantioselective hydrogenation of the cyclic
4

¨
2
86
SCHURCH ET AL.
7
. Tungler, A., Mathe, T., Fodor, K., Sheldon, R. A., and Gallezot, P., 28. Minder, B., Mallat, T., Skrabal, P., and Baiker, A., Catal. Lett. 29, 115
J. Mol. Catal. A 108, 145 (1996).
(1994).
8
. Heinz, T., Wang, G., Pfalz, A., Minder, B., Sch u¨ rch, M., Mallat, T., and
Baiker, A., J. Chem. Soc. Chem. Common., 1421 (1995).
. Wang, G., Heinz, T., Pfaltz, A., Minder, B., Mallat, T., and Baiker, A.,
J. Chem. Soc. Chem. Commun., 2047 (1994).
29. Iwasita, T., and Nart, F. C., in “Advances in Electrochemical Science
and Engineering” (H. Gerischer and C. W. Tobias, Eds.), Vol. 4, p. 123.
VCH, Weinheim/New York, 1995.
30. Mallat, T., Bodnar, Z., Minder, B., Borszeky, K., and Baiker, A., J.
Catal. 168, 183 (1997).
9
1
0. Minder, B., Sch u¨ rch, M., Mallat, T., Baiker, A., Heinz, T., and Pfaltz,
A., J. Catal. 160, 261 (1996).
1. Baiker, A., J. Mol. Catal. A: Chemical 115, 473 (1997).
31. Blaser, H. U., Jalett, H. P., Monti, D. M., and Wehrli, J. T., Appl. Catal.
52, 19 (1989).
1
1
2. Orito, Y., Imai, S., Niwa, S., and Nguyen-Gia-Hung, J. Synth. Org. 32. Meheux, P. A., Ibbotson, A., and Wells, P. B., J. Catal. 128, 387
Chem. Japan 37, 173 (1979).
(1991).
1
1
1
3. Blaser, H. U., Jalett, H. P., and Wiehl, J., J. Mol. Catal. 68, 215 (1991).
4. Schmid, R., Chimia 50, 110 (1996).
5. Niwa, S., Imamura, Y., and Otsuka, Jpn. Kokai Tokkyo Koho 62,
33. Wehrli, J. T., Baiker, A., Monti, D. M., Blaser, H. U., and Jalett, H. P.,
J. Mol. Catal. 57, 245 (1989).
34. Reichardt, C., in “Solvents and Solvent Effects in Organic Chemistry”
(H. F. Ebel, Ed.), p. 359. VCH, Weinheim/New York, 1988.
1
58268 (1987); Chem. Abstr. 109, 128815f (1988).
1
1
6. Carpentier, J.-F., Agbossou, F., and Mortreux, A., Tetrahedron Asym. 35. Young, C. L., “Solubility Data Series.” Pergamon Press, Oxford, 1981.
6
, 39 (1995).
36. Dijkstra, G. D. H., Kellogg, R. M., Wynberg, H., Svendsen, J. S., Marko,
I., and Sharpless, K. B., J. Am. Chem. Soc. 111, 8069 (1989).
37. Dijkstra, G. D. H., Kellogg, R. M., and Wynberg, H., Recl. Trav. Chim.
Pays-Bas 108, 195 (1989).
7. Gen eˆ t, J. P., Pinel, C., Ratovelomanana-Vidal, Mallart, S., Pfister, X.,
Bischoff, L., Ca n˜ o De Andrade, M. C., Darses, S., Galopin, C., and
Laffitte, J. A., Tetrahedron Asym. 5, 675 (1994).
1
8. Hapiot, F., Agbossou, F., and Mortreux, A., Tetrahedron Asym. 5, 515
38. Dijkstra, G. D. J., Kellogg, R. M., and Wynberg, H., J. Org. Chem. 55,
6121 (1990).
(
1994).
1
2
9. Ojima, I., and Kogure, T., J. Organomet. Chem. 195, 239 (1980).
0. Roucoux, A., Devocelle, M., Carpentier, J.-F., Agbossou, F., and
Mortreux, A., Synlett 4, 358 (1995).
39. Garland, M., and Blaser, H. U., J. Am. Chem. Soc. 112, 7048 (1990).
40. Borszeky, K., Mallat, T., Aeschimann, R., Schweizer, W. B., and Baiker,
A., J. Catal. 161, 451 (1996).
2
1. Takahashi, H., Hattori, M., Chiba, M., Morimoto, T., and Achiwa, K., 41. Oleksyn, B. J., Acta Crystallogr. Sect. B 38, 1832 (1982).
Tetrahedron Lett. 27, 4477 (1986). 42. Margitfalvi, J. L., and Heged u¨ s, M., J. Mol. Catal. 107, 281 (1996).
2. Roucoux, A., Suisse, I., Devocelle, M., Carpentier, J.-F., Agbossou, F., 43. Margitfalvi, J. L., Heged u¨ s, M., and First, E. T., Tetrahedron Asym. 7,
2
and Mortreux, A., Tetrahedron Asym. 7, 379 (1996).
571 (1996).
2
2
3. Mallat, T., and Petr o´ , J., React. Kinet. Catal. Lett. 11, 307 (1979).
4. HyperChem 4.0, Hypercube Inc., 419 Philips Street, Waterloo, 45. Bond, G., and Wells, P. B., J. Catal. 150, 329 (1994).
44. Hallmark, V. M., and Chiang, S., Surf. Sci. 286, 190 (1993).
Ontario, Canada N2L 3X2 (1994).
5. Dewar, M. J. S., Zoebisch, E. G., Healy, E. F., and Stewart, J. J. P., J.
Am. Chem. Soc. 107, 3902 (1985).
46. Bond, G., Meheux, P. A., Ibbotson, A., and Wells, P. B., Catal. Today
10, 371 (1991).
47. Leung, L.-W. H., and Weaver, M. J., Langmuir 6, 323 (1990).
48. Schwalm, O., Minder, B., Weber, J., and Baiker, A., Catal. Lett. 23, 271
(1994).
2
2
2
6. Binkley, J. S., Pople, J. A., and Hehre, W., J. Am. Chem. Soc. 102, 939
(
1980).
7. Dupuis, M., Johnston, F., and Marquez, A., IBM Co., Neighborhood
Road, Kingston, NY, 1994.
49. Schwalm, O., Weber, J., Minder, B., and Baiker, A., Int. J. Quantum
Chem. 52, 191 (1994).