Design of new modifiers for the enantioselective hydrogenation of ethyl pyruvate

Article abstract of DOI:10.1006/jcat.1997.1919

All efficient chiral modifiers for Pt in the enantioselective hydrogenation of α-ketoesters possess a basic, secondary or tertiary N atom for interacting with the carbonyl group of the reactant and an aromatic ring system for adsorptive anchoring of the activated complex on Pt. Analysis of the available data suggested that an enlargement of the naphthalene or quinoline anchoring moiety should improve the enantioselection. Accordingly, 1-(9-anthracenyl)-2-(1-pyrrolidinyl)ethanol (10) has been synthesized and tested in the hydrogenation of ethyl pyruvate. The best enantiomeric excess achieved with the new modifier was 87%, which is 12% higher than the optimized value obtained with the corresponding naphthalene derivative (4). A further advantage of the new modifier is the higher stability against self-hydrogenation. Hydrogenation of ethyl pyruvate in the presence of modifier mixtures indicated the following order of adsorption strength on Pt: cinchonidine > 10 > 4. This ranking correlates with the best enantiomeric excesses obtained with these modifiers. Compared with the other modifiers, the number of possible conformations for 10 is reduced because of the symmetry of the anthracenyl ring system. Molecular mechanics calculations suggest that the energy and geometry of the transition complexes between ethyl pyruvate and 10 or 4 are similar. Accordingly, the better efficiency of 10 should be due to its stronger adsorption on Pt and higher acceleration of the modified reaction compared with the competing nonenantioselective (unmodified) reaction. Substituting the 9-anthracenyl group of 10 with a 9-triptycenyl moiety led to a complete loss of enantiodifferentiation, demonstrating that the extended flat aromatic ring system is a crucial structural element of efficient modifiers for α-ketoester hydrogenation.

Full text of DOI:10.1006/jcat.1997.1919

JOURNAL OF CATALYSIS 173, 187–195 (1998)
ARTICLE NO. CA971919
Design of New Modifiers for the Enantioselective Hydrogenation
of Ethyl Pyruvate
ꢀ
ꢀ
ꢀ
Markus Schu¨rch, Thomas Heinz,† Roland Aeschimann, Tamas Mallat,
,1
ꢀ
Andreas Pfaltz,† and Alfons Baiker
ꢀ
Department of Chemical Engineering and Industrial Chemistry, Swiss Federal Institute of Technology, ETH-Zentrum, CH-8092 Zu¨rich, Switzerland;
and †Max-Planck-Institut fu¨r Kohlenforschung, D-5470 Mu¨lheim an der Ruhr, Germany
Received June 23, 1997; revised September 10, 1997; accepted September 10, 1997
to the metal surface. It is sufficient to add traces of the chiral
modifier [a few hundred ppm relative to the reactant (5, 6)]
to induce impressive changes in the product distribution.
The most studied reaction, the transformation of methyl
pyruvate or ethyl pyruvate (EP) to the corresponding lac-
tate, is shown in Scheme 1. Strong efforts have been made
to understand the role of the chiral modifier and to find
new efficient substitutes for the cinchona alkaloids. Three
distinctly different research strategies have been applied.
At first, several chiral compounds with natural origin were
tested in a purely empirical way and with limited success.
Codeine, strychnine, and brucine (7) provided only 2–12%
enantiomeric excess (ee). Another alkaloid, dihydrovin-
pocetin (8), afforded 30% ee, which is still well below
the 90–92% achievable with cinchonidine (CD) or 10,11-
dihydrocinchonidine (9).
The other two directions, the structural modification of
CD (10, 11) and the synthesis of some simple chiral amino
alcohols and amino esters (12–15), provided important in-
formation on the role ofvariousstructuralpartsofthe modi-
fier, which will be summarized briefly, because it constitutes
the basis for the rational design pursued in this work.
It was early established (11) that, besides the crucial
stereogenic center at C-8, CD possesses two important
structural parts (Scheme 1): the quinuclidine N atom for
interaction with the pyruvate molecule (“docking” moiety)
All efficient chiral modifiers for Pt in the enantioselective hydro-
genation of ꢀ-ketoesters possess a basic, secondary or tertiary N
atom for interacting with the carbonyl group of the reactant and
an aromatic ring system for adsorptive anchoring of the activated
complex on Pt. Analysis of the available data suggested that an en-
largement of the naphthalene or quinoline anchoring moiety should
improve the enantioselection. Accordingly, 1-(9-anthracenyl)-2-
(1-pyrrolidinyl)ethanol (10) has been synthesized and tested in
the hydrogenation of ethyl pyruvate. The best enantiomeric excess
achieved with the new modifier was 87%, which is 12% higher than
the optimized value obtained with the corresponding naphthalene
derivative (4). A further advantage of the new modifier is the higher
stability against self-hydrogenation. Hydrogenation of ethyl pyru-
vate in the presence of modifier mixtures indicated the following or-
der of adsorption strength on Pt: cinchonidine > 10 > 4. This rank-
ing correlates with the best enantiomeric excesses obtained with
these modifiers. Compared with the other modifiers, the number of
possible conformations for 10 is reduced because of the symmetry
of the anthracenyl ring system. Molecular mechanics calculations
suggest that the energy and geometry of the transition complexes
between ethyl pyruvate and 10 or 4 are similar. Accordingly, the
better efficiency of 10 should be due to its stronger adsorption on Pt
and higher acceleration of the modified reaction compared with the
competing nonenantioselective (unmodified) reaction. Substituting
the 9-anthracenyl group of 10 with a 9-triptycenyl moiety led to a
complete loss of enantiodifferentiation, demonstrating that the ex-
tended flat aromatic ring system is a crucial structural element of ef-
ficient modifiers for ꢀ-ketoester hydrogenation.
ꢁc 1998 Academic Press
INTRODUCTION
The enantioselective hydrogenation of ꢀ-ketoesters over
cinchona-modified Pt, discovered by Orito et al. (1), has
gained great attention (2–4). An attractive feature of the
reaction is that there is no need for a troublesome (and
expensive) pretreatment to provide the chiral information
1 To whom correspondence should be addressed. Fax: +411 632 1163.
E-mail: a.baiker@tech.chem.ethz.ch.
SCHEME 1
187
0021-9517/98 $25.00
c
Copyright ꢁ 1998 by Academic Press
All rights of reproduction in any form reserved.

¨
SCHURCH ET AL.
188
A crucial question is the adsorption of the modifier on
Pt. H/D exchange experiments indicated that the quinoline
rings of CD adsorb parallel to the Pt surface, while the
quinuclidine part of the modifier is not in direct contact with
Pt (23). On the contrary, a perpendicular adsorption of the
quinoline rings via the N atom was proposed on the basis of
“single turnover” studies (3). However, this proposal is not
in line with catalytic results obtained with the modifiers 3
and 4 shown in Scheme 4, which differ only in the anchoring
moiety. At low pressure (1–10 bar) they provide similar ee’s
for (R)-lactate under otherwise identical conditions (12).
This is an indication that the extended aromatic ꢂ-electron
system is important in the adsorption process, whereas the
quinoline N atom seems to have little influence. Note that
the adsorption of naphthalene parallel to the Pt surface was
evidenced by STM and other physicochemical methods (24,
25).
A third model has been proposed recently (26, 27), ac-
cording to which CD would adsorb on Pt in a tilted posi-
tion, forming a “roof” and providing a specific “shielding”
effect for the pyruvate molecule. However, no evidence of
this special adsorption mode has been provided yet, and it
stands in contrast to the results of isotope exchange experi-
ments mentioned above (23).
SCHEME 2
and the aromatic ring system for adsorption on Pt (“anchor-
ing” moiety).
With respect to the chemical nature of the docking moi-
ety, all the efficient modifiers contain a secondary or tertiary
N atom in ꢀ or ꢁ position to the stereogenic center. Primary
amines are not stable under the reaction conditions and are
rapidly alkylated by EP (condensation of the amino group
with the activated carbonyl group of EP followed by re-
duction of the imine). An example is the transformation
of naphthylethylamine 1 to 2 (Scheme 2), which acts as an
efficient modifier affording 82% ee under optimized condi-
tions (15–17).
It has been proposed (18) that the quinuclidine N stabi-
lizes the half-hydrogenated state of EP by a N H O-type
– –
Comparison of the modifiers 3 and 4 in Scheme 4 also
demonstrated that the stability against hydrogenation of
the naphthalene ring system at pressures above 10 bar is
considerably lower than that of the quinoline moiety (12).
interaction. This proposal is in line with recent molecu-
lar modeling studies (19, 20). Similarly, theoretical calcu-
lations (21, 22) suggested that in protic polar solvents the
protonated quinuclidine N atom of CD interacts with the
carbonyl O atom of EP by H bonding, providing again a
– –
N H O-type interaction.
– – –
The O C C N structural part is characteristic of cin-
chona alkaloids (Scheme 1) and other 1,2-amino alcohol-
type chiral auxiliaries. An important role in enantiodiffer-
entiation was attributed earlier also to the O atom (3, 11).
However, there are several efficient modifiers of EP hydro-
genation that possess no O atom at all. Two of these com-
pounds, prepared by reductive alkylation of 1, are shown in
Scheme 3, together with the ee values achieved (15). These
examples indicate that it is the basic N that is responsible
for the reactant–modifier interaction, though the adjacent
OH group present in other modifiers can be advantageous.
SCHEME 3
SCHEME 4

ENANTIOSELECTIVE HYDROGENATION OF ETHYL PYRUVATE
189
SCHEME 5
The steric hindrance exerted by the aromatic ring system fier (10 in Scheme 6) have been synthesized and tested in
of the modifier has been shown to be important. (R)-2- the enantioselective hydrogenation of EP. In addition, 11
(1-Pyrrolidinyl)-1-(1-naphthyl)ethanol (4 in Scheme 4) af- (Scheme 6) was also prepared which enabled us to demon-
forded 75% ee under optimized conditions (14, 17). When strate the role of the flat aromatic ring system in the ad-
changing the point of attachment of the pyrrolidinyl- sorption of the modifier on Pt.
ethanol moiety to the naphthalene ring by one position
(5), the excess of (R)-lactate dropped by 26% under other-
EXPERIMENTAL
wise identical conditions. This effect was attributed to the
In the following, the generally used abbreviations (CN,
smaller steric hindrance for the formation of (S)-lactate.
CD) are applied for the cinchona alkaloids, but for the
The most important requirement is that the modifier
sake of simplicity the synthetic modifiers are represented
should be anchored onto Pt by at least two adsorption sites
by boldface numbers as indicated in Schemes 2–6.
[Scheme 5 (13, 16)]. This is the case when the modifier pos-
sesses a naphthalene (4) or quinoline (3) ring system or a
phenyl group substituted with a strongly adsorbing func-
tional group (6). [Note that 6 is only the precursor of the
real modifier (15). At the beginning of the reaction the nitro
group israpidlyreduced to an amino group and the aliphatic
amino group is alkylated by EP, as discussed above.] When
one of the adsorption sites is elimated, i.e., the modifier
contains a benzene (7, 9) or a pyridine (8) ring as anchor-
Synthesis of New Modifiers
The synthesis of modifiers 10 and 11 is illustrated in
Scheme 6. A short summary of steps i–ix can be found be-
low. More details of the synthesis and identification of the
intermediates and products are available from the authors
on request.
Step (i) (28). Phenyllithium 58.7 mmol in 37 ml cyclo-
ing moiety, the ee drops to zero under otherwise identical hexane/ether was added dropwise to a stirred suspension of
conditions. A feasible explanation is that in this case the 62.1 mmol methyltriphenylphosphonium bromide in 165 ml
modifier does not adsorb in a way that would allow the re- THF, at 0ꢂC. The mixture was warmed to room temperature
quired interaction. Note that the inefficiency of modifiers and stirred for 30 min. Then 11.8 g (57 mmol) anthracene-
7–9 possessing only a single aromatic ring is further ex- 9-carbaldehyde in 42 ml THF was added dropwise at 0ꢂC.
perimental evidence against the validity of the “shielding” The mixture was stirred at room temperature for 15 h, then
model. According to this model the decreasing size of the 1 ml MeOH was added. The slurry was diluted with 100 ml
aromatic ring system should result in somewhat lower, but petroleum ether and the supernatant solution was decanted
still good enantioselectivity (27).
and filtered through celite. The solids remaining in the mask
On the basis of the comparative experiments illustrated were washed with three 100-ml portions of petroleum ether,
in Scheme 5 we assumed that a modifier, similar to com- and the supernatant solutions were also filtered through
pound 4, but possessing an anthracenyl instead of the naph- celite. The filtrate was concentrated in vacuo. Purification
thylgroup asanchoringmoiety, could be even more efficient by flash chromatography with hexane afforded 10.3 g (81% )
in enantiodifferentiation. Both enantiomers of this modi- of 9-vinyl-anthracene as a yellow solid.

¨
190
SCHURCH ET AL.
SCHEME 6
Step ii (29, 30). K3[Fe(CN)6] (49.6 g, 150 mmol), K2CO3 (1 : 1) afforded 967 mg (89% ) (S)-4-(9-anthracenyl)-2,2-di-
(20.8 g, 150 mmol) (DHQ)2-PHAL (395 mg, 0.5 mmol) methyl-1,3-dioxolane as a yellow oil. [ꢀ]D = ꢃ5 (c = 1.39,
and K2OsO2(OH)4 (37 mg, 0.1 mmol) in 250 ml t-butyl CHCl3).
alcohol and 250 ml water were cooled to 0ꢂC. 9-Vinyl-
Step vi (32). A solution of 490 mg (3.58 mmol) an-
anthracene (10.3 g, 50 mmol) was added and the slurry
thranilic acid in 3 ml THF was added over 3 h to a reflux-
stirred at 0ꢂC for 87 h. The mixture was quenched with 76 g
ing mixture of 940 mg (3.4 mmol) anthracenyl-dimethyl-
Na2SO3, warmed to room temperature, and stirred for 1 h.
dioxolane and 0.5 ml (3.8 mmol) amyl nitrite in 12 ml
Extraction with EtOAc followed by flash chromatography
CH2Cl2. The solution was concentrated in vacuo. Six
with hexane/EtOAc (1 : 1) afforded 10.1 g (84% ) of (S)-1-
milliliters of o-xylene and 250 mg maleic anhydride were
(9-anthracenyl)-1,2-ethanediol. Greater than 99.5% ee was
added, and the solution was refluxed for 15 min. After cool-
obtained via recrystallization from CH2Cl2/hexane.
ing to room temperature, 15 ml water and 20 ml CH2Cl2
Step iii. Anthracenyl-ethanediol (6.5 g, 27 mmol) and
NEt3 (5.7 ml, 41 mmol) were dissolved in 500 ml CH2Cl2.
Mesitylene sulfonyl chloride (5.9 g, 27 mmol) was added
and the reaction mixture stirred at room temperature for
2 weeks. Removal of the solvent in vacuo followed by flash
chromatography with hexane/EtOAc (2 : 1) afforded 10.0 g
(88% ) of (S)-2-hydroxy-2-(9-anthracenyl)ethyl mesitylene
sulfonates as a yellow paste. [ꢀ]D = +40 (c = 1.37, CHCl3).
Step iv. The product of step iii was stirred in 200 ml
pyrrolidine at 40ꢂC for 10 days. The excess amine was re-
moved in vacuo. The residue was stirred in 350 ml ether and
100 ml saturated NaHCO3 solution for 2 h. After separa-
tion the aqueous phase was extracted with ether, dried over
K2CO3, and concentrated in vacuo. Purification by flash
chromatography with EtOAc/NEt3 (100 : 1) afforded (S)-1-
(9-anthracenyl)-2-(1-pyrrolidinyl)ethanol (10) as crystals.
Crystallization from hot i-PrOH yielded 3.6 g (52% ) yellow
solid. [ꢀ]D = +8 (c = 1.05, CHCl3), >99.5% ee by HPLC.
The (R)-enantiomer of 10 was prepared in an analogous
manner.
were added. The layers were separated, and the organic
phase was extracted three times with 10 ml of 12% aqueous
KOH, dried over MgSO4 and concentrated in vacuo. Purifi-
cation by flash chromatography with hexane/EtOAc (5 : 1)
afforded 952 mg (80% ) (S)-4-(9-triptycenyl)-2, 2-dimethyl-
1,3-dioxolane. Crystallization from CH2Cl2/hexane yielded
yellow crystals (30% ). [ꢀ]D = +45 (c = 0.98, CHCl3).
Step vii. A solution of 0.54 g (1.5 mmol) triptycenyl-di-
methyl-dioxolane and 1.45 g (7.6 mmol) p-toluene-sulfonic
acid in 40 ml methanol was refluxed for 6 days. Removal
of the solvent in vacuo followed by flash chromatography
with hexane/EtOAc (1 : 1) afforded 333 mg (70% ) (S)-1-
(9-triptycenyl)-1,2-ethanediol as a white solid. [ꢀ]D = ꢃ5
(c = 1.09, EtOH).
Step viii. The procedure wasanalogousto step iii. Purifi-
cation by flash chromatography with hexane/EtOAc (3 : 1)
afforded (S)-2-hydroxy-2-(9-triptycenyl)ethyl mesitylene
sulfonates as a yellow paste (81% ).
Step ix. The procedure was analogous to step iv. Purifi-
Step v(31). Anthracenyl-ethanediol(930mg, 3.9mmol) cation by flash chromatography with EtOH/NEt3 (200 : 1)
and p-toluenesulfonicacid (80mg, 0.42mmol) in 100mlace- followed by crystallization from hot methanol yielded (S)-
tone were stirred for 3 h. Removal of the solvent in vacuo 1-(9-triptycenyl)-2-(1-pyrrolidinyl)ethanol (11) as pale yel-
followed by flash chromatography with hexane/EtOAc low crystals (46% ). [ꢀ]365 = +7 (c = 1.06, CHCl3).

ENANTIOSELECTIVE HYDROGENATION OF ETHYL PYRUVATE
RESULTS
191
Catalytic Hydrogenation
Ethyl pyruvate (EP, Aldrich) was freshly distilled in
vacuo before each reaction. AcOH (Riedel de Hae¨n), cin-
chonine (CN, Fluka, 99% ), and cinchonidine (CD, Fluka,
>98% ) were used without further purification.
Five weight percent Pt/alumina (Engelhard 4759) was
pretreated in flowing nitrogen at 400ꢂC for 30 min, fol-
lowed by a reductive treatment in 30 ml minꢃ1 hydrogen
for another 90 min. The catalyst was then cooled to room
temperature in flowing hydrogen and transferred to the au-
toclave under exclusion of oxygen. The catalyst was first
contacted with the solvent containing the proper amount
of modifier. The metal dispersion after heat treatment was
27% as calculated from the TEM images.
Comparisonof1-(9-Anthracenyl)-2-(1-pyrrolidinyl)ethanol
(10) and 1-(1-Naphthyl)-2-(1-pyrrolidinyl)ethanol (4)
Both enantiomers of the new modifier 10 have been syn-
thesized, as described under Experimental, and tested in
the hydrogenation of EP. The preliminary study on the
performance of 10 indicated that the influence of reaction
parameters was in many respects similar to that observed
earlier when using the naphthalene derivative 4 (14). For
example, reductive pretreatment of Pt/alumina at 400ꢂC
improved the ee considerably. Acetic acid was the best sol-
vent, suggesting that protonation of the pyrrolidine N atom
improves the enantiodifferentiation (21, 22).
The new modifier 10 was more efficient with respect to
ee and rate acceleration [“ligand acceleration” effect (5)].
This is illustrated by the kinetic results in Figs. 1 and 2. With
both modifiers, higher modifier concentration increased ee
up to a concentration limit of 0.15 mmol literꢃ1 (Fig. 1). This
concentration corresponds to a modifier : reactant molar ra-
tio of 1 : 30,000. Above this concentration the changes were
minor and modifier 10 provided 12–13% higher ee than
modifier 4 [data for 4 are taken from Ref. (13)]. In contrast
to modifier 4, no maximum of the reaction rate could be
observed with modifier 10 up to a concentration of 1 mmol
literꢃ1 (Fig. 2). When applying the modifiers in sufficiently
high concentration, a rate acceleration (rate of reaction, re-
lated to that of the unmodified reaction) up to a factor of
20–21 was observed with 10, as compared with a factor of
7–8 induced by 4 (14).
The hydrogenation of EP was carried out in a 100-ml
stainless-steel autoclave (Baskerville) with a 50-ml glass
liner and PTFE cover. The reaction temperature was
controlled by the bath in which the reactor was immersed.
The mixture was stirred magnetically at 1000 rpm. The
pressure was kept constant with computerized constant
volume–constant pressure equipment (Bu¨chi BPC 9901)
which allowed calculation of the rate of hydrogen consump-
tion. The (initial) reaction rate was determined between
0 and 20% conversion by linear regression. The influence
of interparticle mass transport was found to be negligible,
based on experiments with varying amounts of catalyst.
The enantiomeric excess was determined at full con-
version with a HP 5890A gas chromatograph using
a chiral WCOT Cyclodextrin-ꢁ-2,3,6-M-19 (Chrompack)
capillary column. The enantioselectivity is expressed as
ee (% ) = |[R] ꢃ [S]|/[R] + [S].
A further advantage of 10, as compared with the former
successful synthetic modifiers 2 and 4, is the significantly
Theoretical Calculations
Molecular mechanics calculations were performed to
gain some insight into the structure of the modifier–reactant
complex, which is assumed to be adsorbed parallel to the
platinum surface via the aromatic ring system and the car-
bonyl group of EP. This assumption is justified based on
previous H/D isotope exchange experiments (18) and the
fact that the adsorption enthalpy (33) is much higher than
the hydrogen bond interaction energy which is calculated
by molecular mechanics.
Calculations were carried out using the Discover Molec-
ular Simulation Program (Version 4.0.0) from MSI with
the CFF91 force field (Version 2.0) (34). The default val-
ues were used during geometry optimization except that
the aromatic ring system of the modifiers and the car-
bonyl groups of EP were kept in one plane (restraint
type: out of plain, force constant: 10,000). The calcula-
tions were repeated with the HyperChem program from
Hypercube (Version 4.5) using the MM+ force field with-
out any constrains (smallest significant difference: 0.2 kcal/
mol).
FIG. 1. Influence of modifier concentration on the enantiomeric ex-
cess of (S)- or (R)-lactate in the hydrogenation of EP over Pt/alumina
modified with 10 or 4 [(S) or (R) enantiomer, respectively]. Reaction con-
ditions for modifier 10: 10ꢂC, 40 mg catalyst, 10 ml AcOH, 45 mmol EP,
70 bar H2; for modifier 4: 25ꢂC, 100 mg catalyst, 20 ml AcOH, 90 mmol
EP, 10 bar H2.

¨
SCHURCH ET AL.
192
FIG. 4. Influence of hydrogen pressure on the initial rate of EP hydro-
FIG. 2. Influence of modifier concentration on the initial rate of EP
genation over Pt/alumina modified with 10 or 4 [both (R)-enantiomers].
hydrogenation over Pt/alumina modified with 10 or 4 [(S) or (R) enan-
For conditions see Fig. 3.
tiomer, respectively]. For conditions see Fig. 1.
higher stability at high hydrogen pressures. This behavior
is illustrated in Figs. 3 and 4. When applying 10, the reaction
rate increased with increasing pressure, expectedly, but the
ee barely changed between 10 and 100 bar (not all data are
shown in Figs. 3 and 4). The poor performance of modi-
fier 4 at pressures higher than 10 bar was proved to be due
to the partial hydrogenation of the naphthalene ring sys-
tem, resulting in weaker adsorption on Pt (14). Note that
10 provided significantly higher ee even at 10 bar (Fig. 3),
when the saturation of the naphthalene rings was found to
be negligible.
ee by about 2% . This difference may be due to the presence
of traces of impurities and to analytical error.
Another new modifier, (S)-1-(9-triptycenyl)-2-(1-pyrro-
lidinyl)ethanol (11, Scheme 6), which differed from the
(S)-conformer of 10 only in the anchoring moiety, was also
synthesized. Interestingly, no enantiodifferentiation (<5%
ee) could be detected in EP hydrogenation when using 11 as
chiral modifier for Pt/alumina. The influence of substituting
the flat aromaticanthracenylringsystem in 10bythe “three-
dimensional” triptycenyl moiety in 11 is discussed later.
The highest ee achieved with the (R)-enantiomer of 10 Comparison of Modifiers 10 and 4
was 87% (conditions: 38 mg Pt/alumina, 3.4 mg modifier,
and 5 ml EP in 10 ml AcOH were hydrogenated at 10ꢂC
and 70 bar). The (S)-enantiomer provided a slightly lower
with Cinchona Alkaloids
The hydrogenation of EP was performed in the presence
of two different modifiers to obtain some information on
their relative strength of adsorption on Pt under the reac-
tion conditions. CN was always used as one of the compo-
nents, which alone provides the (S)-lactate in excess. The
other modifier was (R)-10, (R)-4, or CD (for comparison);
all three compounds provide (R)-lactate in excess, when
used alone as the source of chiral information. The results
of these experiments are shown in Figs. 5 and 6. When ap-
plying CD or CN alone, the ee values are similar. A de-
viation from the linear correlation between ee and molar
fraction of CD in favor of (R)-lactate formation was found
when using a mixture of the two alkaloids. This is an indica-
tion that CD adsorbs on Pt more strongly than CN and has
greater influence on the direction of enantiodifferentiation
over the entire composition range. Note that similar results
were obtained earlier, though a strikingly different cata-
lyst pretreatment procedure and reaction conditions were
applied (35).
FIG. 3. Influence of hydrogen pressure on the enantiomeric excess of
(R)-lactate in the hydrogenation of EP over Pt/alumina modified with 10
or 4 [both (R)-enantiomers]. Reaction conditions for modifier 10: 10ꢂC,
40 mg catalyst, 10 ml AcOH, 45 mmol EP, 3.4 ꢃmol modifier; for modifier
4: 25ꢂC, 100 mg catalyst, 20 ml AcOH, 90 ꢃmol EP, 6.2 ꢃmol modifier.
Deviation in favor of (S)-lactate formation was observed
when mixing CN with either synthetic modifier 10 or 4.
Interestingly, even a CN : 4 = 1 : 9 mixture provided 44%

ENANTIOSELECTIVE HYDROGENATION OF ETHYL PYRUVATE
193
aromatic rings of the modifier and the carbonyl groups of
EP are adsorbed parallel to a flat (ideal) Pt surface (23,
25, 36–38). The structure of the transition complex is as-
sumed to resemble the transition state and enantiodiffer-
entiation occurs due to energetical favoring of one of the
diasteromeric complexes. In contrast to the previous cal-
culations (21, 22), the simulation of the flat adsorption is
achieved by keeping the aromatic rings of 4 and 10 and
the carbonyl groups of EP in a plane during energy mini-
mization of the modifier–EP complex. For both modifiers
the protonated forms were used for calculations. The calcu-
lated geometries, illustrated in Scheme 7, are very similar
for modifiers 4 and 10. In both cases the formation of (R)-
lactate is favored, when using the (R)-conformers of the
modifiers, though the energy differences between the com-
plexes leading to (R)- or (S)-lactate formation are rather
small (around 0.35 kcal molꢃ1). For comparison, similar cal-
culations have been performed with the Hyperchem pro-
gram (MM+ force field), which provided almost the same
optimized geometries, but considerably higher energy dif-
ferences in favor of (R)-lactate formation (2–4 kcal molꢃ1).
This comparative study shows that although the absolute
values of the calculated energies may differ depending on
the software used, the direction of enantiodifferentiation is
clearly indicated.
FIG. 5. Enantiomeric excess of (R)- or (S)-lactate in EP hydrogena-
tion, when using a mixture of two modifiers [CN and one of CD, 10 (R)
and 4 (R)]. Reaction conditions: 70 bar H2, 10ꢂC, 100 mg Pt/alumina, 20 ml
AcOH, 91 mmol EP, total amount of 8 ꢃmol modifier.
excess of (S)-lactate, suggesting that 4 is less strongly ad-
sorbed than CN on the Pt surface. This behavior is due at
least partly to the partial hydrogenation of the naphthalene
ring system during reaction at 70 bar, as discussed above.
This side effect could be eliminated when performing the
experiments at 10 bar or below, but these conditions are
far from the optimum for the cinchona alkaloids and the
comparison would be less interesting.
Independent of the size of the energy differences
between the modifier–EP transition complexes (Scheme 7),
the almost identical values suggest that the higher ee
achieved with 10 (as compared with 4) cannot be due to
an energetically more favored interaction with EP or some
extra steric hindrance between 10 and EP.
There are big differences in the rate of EP hydrogenation
in the presence of various modifier mixtures (Fig. 6). Both
4 and 10 decrease the reaction rate as compared with the
–
rate achieved with CN alone. Adversely, CD and CN CD
mixtures provide higher rates than CN alone.
Theoretical Calculations
As in previous computational studies of the structure of
the diastereomeric transition states, we assumed that the
FIG. 6. Reaction rate in EP hydrogenation, when using a mixture of
two modifiers [CN and one of CD, 10 (R) and 4 (R)]. For conditions see
Fig. 5.
SCHEME 7

¨
SCHURCH ET AL.
194
mer results (14). When applying N-methyl-pyrrolidine or
pyrrolidinyl-ethanol, the rate of EP hydrogenation related
to the rate of unmodified reaction under similar conditions
in AcOH (rrel) is less than unit. Both compounds can inter-
act via the basic N atom with the carbonyl O atom of EP;
however, this interaction does not result in rate accelera-
tion. Only those compounds (10 and 4) that can properly
anchor the N base–EP adduct onto the Pt surface, are able
to increase the reaction rate. This indicates that the origin
of rate acceleration is the same as that of enantiodifferen-
SCHEME 8
– –
tiation: an N H O-type interaction between reactant and
Molecular modeling provided a likely explanation of the
inefficiency of modifier 11. The minimum energy conforma-
tion of 11 could not be positioned in any feasible way onto
a flat Pt surface. This is an indication that the absence of
enantiodifferentiation with 11 is due to the “propeller-like”
structure of the aromatic ring system, which hinders the ꢂ
bonding interaction with the surface Pt atoms (Scheme 8).
modifier, both adsorbed appropriately on the Pt surface.
The positive correlation between rate acceleration and ee
is also illustrated by the comparative studies in Figs. 1–6.
The catalytic test of modifier 10 demonstrated that substi-
tution ofthe naphthalene ringsystem ofthe alreadyefficient
modifier 4 with an anthracenyl anchoring moiety improved
the ee significantly. The better performance of 10 compared
with 4 could be explained by a change in the modifier–
EP interaction and/or in the adsorption of the modifier–
EP adduct. The former possibility seems unlikely based on
molecular modeling. The calculations strongly suggested
that there is no significant deviation in the interaction en-
ergies or any extra steric hindrance against the formation
of (R)- or (S)-lactate, when applying modifier 10 instead
of 4.
With respect to the interaction of Pt with the modifier–
EP complex, the hydrogenation of EP in the presence of
modifier mixtures indicated that 10 adsorbs more strongly
than 4 on Pt. The difference at high hydrogen pressure is
due partly to the higher stability of the anthracenyl moiety
against hydrogenation (16).
The comparative catalytic experiments showed that the
strength of adsorption and the observed rate acceleration
are strongly correlated features of the modifiers. Accord-
ing to a former simple kinetic model (39), lactate is formed
from pyruvate via a modified enantioselective route and
an unmodified route leading to racemic product. The mea-
sured overall rate is the sum of the rates of modified and
unmodified reactions. The ee can be improved by increasing
the rate of the modified reaction (resulting in higher overall
rate, i.e., rate acceleration) or by suppressing the unmodi-
fied reaction. The latter is also likely in the case of 10 due
to the larger site requirement of the anthracenyl moiety,
compared with the naphthyl group of 4.
DISCUSSION
Let us consider first the rate acceleration, i.e., the higher
rate observed in the presence of chiral modifier as com-
pared with the rate of the unmodified (racemic) reaction. It
was proposed earlier (18) that the rate acceleration is due
to the stabilization of the half-hydrogenated state of EP by
CD. The importance of this interaction in apolar medium
has also been emphasized recently by theoretical calcula-
tions (20). It has also been shown by molecular modeling
(21) that in acetic acid (i) there is an interaction already
between the protonated form of CD and EP (i.e., before the
–
half-hydrogenated form is produced), and (ii) the CD EP
complex resembles the half-hydrogenated state of the pyru-
vate molecule.
– –
The N H O-type interaction between the basic N atom
of the modifier and the carbonyl O atom of pyruvate is an
important but not sufficient requirement for achieving rate
acceleration (and enantiodifferentiation). This is illustrated
by the data in Scheme 9, based on the present work and for-
CONCLUSIONS
In recent years we have focused on the development
of new, structurally simple chiral amines as modifiers for
Pt in the enantioselective hydrogenation of ꢀ-ketoesters.
An efficient member of this family is 1-(1-naphthyl)-2-
(1-pyrrolidinyl)ethanol (4), which afforded 75% ee at
low pressures (1–10 bar) (14). Based on the influence of
SCHEME 9

ENANTIOSELECTIVE HYDROGENATION OF ETHYL PYRUVATE
195
structural modifications on the efficiency of these new syn- 13. Minder, B., Mallat, T., Baiker, A., Wang, G., Heinz, T., and Pfaltz, A.,
J. Catal. 154, 371 (1995).
14. Minder, B., Schu¨rch, M., Mallat, T., Baiker, A., Heinz, T., and Pfaltz,
A., J. Catal. 160, 261 (1996).
15. Blaser, H. U., Jalett, H. P., Monti, D. M., Baiker, A., and Wehrli, J. T.,
thetic modifiers, we predicted that the enlargement of the
anchoring moiety should improve ee, on the condition
that the aromatic ring system still possesses a flat, two-
dimensional structure. Synthesis and catalytic testing of
Stud. Surf. Sci. Catal. 67, 147 (1991).
1-(9-anthracenyl)-2-(1-pyrrolidinyl)ethanol (10) and 1-(9- 16. Minder, B., Schu¨rch, M., Mallat, T., and Baiker, A., Catal. Lett. 31, 143
(1995).
triptycenyl)-2-(1-pyrrolidinyl)ethanol (11) confirmed our
assumption. When comparing the best ee values achieved,
modifier 10 was found to be almost as efficient as the cin-
chona alkaloids. Moreover, the studies of several struc-
17. Heinz, T., Wang, G., Pfalz, A., Minder, B., Schu¨rch, M., Mallat, T., and
Baiker, A., J. Chem. Soc. Chem. Commun., 1421 (1995).
18. Bond, G., Meheux, P. A., Ibbotson, A., and Wells, P. B., Catal. Today
10, 371 (1991).
turally simple chiral amino alcohols and amino esters 19. Simons, K. E., Meheux, P. A., Griffiths, S. P., Sutherland, I. M.,
Johnston, P., Wells, P. B., Carley, A. F., Rajumon, M. K., Roberts, M.
W., and Ibbotson, A., Recl. Trav. Chim. Pays-Bas 113, 465 (1994).
20. Schu¨rch, M., Schwalm, O., Mallat, T., Weber, J., and Baiker, A.,
J. Catal. 169, 275 (1997).
21. Schwalm, O., Minder, B., Weber, J., and Baiker, A., Catal. Lett. 23, 271
(1994).
(12, 15) provided new insights concerning the reactant–
modifier–metal surface interaction. This knowledge may
serve as a guide in the search for modifiers for other enan-
tioselective hydrogenation reactions over Pt.
22. Schwalm, O., Weber, J., Minder, B., and Baiker, A., Int. J. Quantum
Chem. 52, 191 (1994).
ACKNOWLEDGMENT
23. Bond, G., and Wells, P. B., J. Catal. 150, 329 (1994).
24. Hallmark, V. M., and Chiang, S., Phys. Rev. Lett. 70, 3740 (1993).
25. Hallmark, V. M., Chiang, S., Brown, J. K., and Wo¨ll, C., Phys. Rev.
Lett. 66, 48 (1991).
Financialsupport ofthiswork bythe SwissNationalScience Foundation
(Program Chiral 2) is gratefully acknowledged.
REFERENCES
26. Margitfalvi, J. L., Hegedu¨s, M., and Tfirst, E., Tetrahedron: Asym. 7,
571 (1996).
1. Orito, Y., Imai, S., and Niwa, S., J. Chem. Soc. Jpn. 8, 1118 (1979).
2. Baiker, A., J. Mol. Catal. A 115, 473 (1997).
27. Margitfalvi, J. L., Hegedu¨s, M., and Tfirst, E., Stud. Surf. Sci. Catal.
101, 241 (1996).
3. Augustine, R. L., Tanielyan, S. K., and Doyle, L. K., Tetrahedron: 28. Leopold, E. J., Org. Synth. 64, 164 (1986).
Asym. 4, 1803 (1993).
29. Sharpless, K. B., Amberg, W., Bennani, Y. L., Gerard, A. C., Hartung,
4. Blaser, H.-U., Jalett, H.-P., Mu¨ller, M., and Studer, M., Catal. Today
37, 441 (1997).
J., Kuy-Sung, J., Kwong, H.-L., Morikawa, K., Wang, Z.-M., Xu, D.,
and Zhang, X.-L., J. Org. Chem. 57, 2768 (1992).
5. Garland, M., and Blaser, H. U., J. Am. Chem. Soc. 112, 7048 (1990).
30. Corey, E. J., and Noe, M. C., J. Am. Chem. Soc. 118, 11038 (1996).
6. Bond, G., Simons, K. E., Ibbotson, A., Wells, P. B., and Whan, D. A., 31. Hayashi, H., Nakanishi, K., Brandon, C., and Marmur, J., J. Am. Chem.
Catal. Today 12, 421 (1992).
Soc. 95, 5749 (1973).
7. Griffiths, S. P., Johnston, P., Vermeer, W. A. H., and Wells, P. B.,
J. Chem. Soc. Chem. Commun., 2431 (1994).
8. Tungler, A., Ma´the´, T., Fodor, K., Sheldon, R. A., and Gallezot, P.,
J. Mol. Catal. A 108, 145 (1996).
9. Blaser, H. U., Jalett, H. P., Monti, D. M., Reber, J. F., and Wehrli, J. T.,
Stud. Surf. Sci. Catal. 41, 153 (1988).
32. Friedmann, L., and Logullo, F. M., J. Am. Chem. Soc. 85, 1549 (1963).
33. Gavezzotti, A., Ortoleva, E., and Simonetta, M., Chem. Phys. Lett. 98,
536 (1983).
34. Hwang, M. J., Stockfisch, T. P., and T., H. A., J. Comput. Chem. 15, 162
(1994).
35. Simons, K. E., Meheux, P. A., Ibbotson, A., and Wells, P. B., Stud. Surf.
Sci. Catal. 75, 2317 (1993).
36. Ponec, V., Appl. Catal. A 149, 27 (1997).
10. Blaser, H. U., Jalett, H. P., and Wiehl, J., J. Mol. Catal. 68, 215
(1991).
11. Wang, G., Heinz, T., Pfaltz, A., Minder, B., Mallat, T., and Baiker, A., 37. Delbecq, F., and Sautet, P., J. Catal. 152, 217 (1995).
J. Chem. Soc. Chem. Commun., 2047 (1994).
12. Simons, K. E., Wang, G., Heinz, T., Giger, T., Mallat, T., Pfaltz, A., and
Baiker, A., Tetrahedron: Asym. 6, 505 (1995).
38. Carley, A. F., Rajumon, M. K., Roberts, M. W., and Wells, P. B.,
J. Chem. Soc. Faraday Trans. 91, 2167 (1995).
39. Wehrli, J., Ph.D. thesis, ETH-Zu¨rich, No. 8833 (1989).