Kinetic Studies of the Enantioselective Hydrogenation of Ethyl Pyruvate Catalyzed by a Cinchona Modified Pt/Al2O3 Catalyst

Article abstract of DOI:10.1006/jcat.1997.1925

The kinetics of the hydrogenation of ethyl pyruvate to ethyl lactate on a 5% Pt/Al₂O₃ catalyst in toluene was investigated both in absence and in presence of the chiral modifier 10,11-dihydrocinchonidine. It was shown that all important prerequisites for obtaining reliable kinetic data for the reaction were fulfilled. The effects on rate and enantiomeric excess of catalyst loading, modifier and substrate concentrations, hydrogen pressure, and temperature were determined for the unmodified and the modified system. The modified reaction was approximately 20-30 times faster than the unmodified reaction. A significant increase of the enantiomeric excess from 1 to 40 bar was noticed. Apparent activation energies were estimated to be 4-6 kcal/mol. Rate equations were developed for various kinetic schemes on the basis of the Langmuir-Hinshelwood-Hougen-Watson formalism and fitted to the kinetic data. Several such schemes described the measured data reasonably well and in most cases explanations other than the one considered to be the most plausible were also in agreement with our data. For the unmodified catalyst, we propose a competitive adsorption of the α-ketoester and hydrogen and the addition of the first hydrogen atom to be rate determining. On a chiral site, the rate determining step (RDS) to the major enantiomer is proposed to be the addition of the second hydrogen, whereas the RDS for the minor enantiomer remains the first H addition. On the basis of this interpretation, different proposals advanced in the literature for the mode of action of the cinchona modified Pt catalyst were compared.

Full text of DOI:10.1006/jcat.1997.1925

JOURNAL OF CATALYSIS 173, 282–294 (1998)
ARTICLE NO. CA971925
Kinetic Studies of the Enantioselective Hydrogenation of Ethyl Pyruvate
Catalyzed by a Cinchona Modified Pt/Al2O3 Catalyst
Hans-Ulrich Blaser,1 Hans-Peter Jalett, Marc Garland,2 Martin Studer, Hans Thies, and Amina Wirth-Tijani
Catalysis and Synthesis Services, Novartis Services AG, R-1055.6, Basel, CH-4002 Switzerland
Received January 2, 1997; revised May 16, 1997; accepted September 10, 1997
ral heterogeneous catalysts have received less attention,
even though they have potential technological advantages
and offer some interesting conceptual opportunities (3–5).
Kinetic studies for the enantioselective hydrogenation of
-ketoesters have been reported for tartrate-modified Ni
catalysts both in the liquid and the gas phase (4, 6, 7). Based
on these results and on other observations, several mecha-
nistic schemes were developed for this reaction (4, 6–8).
Preliminary kinetic results were also reported for Pt/Al2O3
and Pt/SiO2 catalysts modified with cinchona alkaloids for
the hydrogenation of -ketoesters in the liquid phase (3,
9–15).
The Pt cinchona system was first described by Orito and
co-workers in 1979 (16) (Fig. 1). The most fascinating aspect
of this catalyst is the fact that the presence of the cinchona
modifiers leads not only to enantioselection but also to a
rate acceleration by a factor of 10–100 (12, 17). This effect
was studied in more detail for the hydrogenation of ethyl
pyruvate (etpy) catalyzed by Pt/Al2O3 modified with dihy-
drocinchonidine (HCd) (10). An equilibrium was proposed
between unmodified, unselective Pt sites with low activity
and modified, enantioselective sites with high activity. How-
ever, these studies could not answer the question about the
mode of action of the chiral modifier.
The kinetics of the hydrogenation of ethyl pyruvate to ethyl
lactate on a 5% Pt/Al2O3 catalyst in toluene was investigated
both in absence and in presence of the chiral modifier 10,11-
dihydrocinchonidine. It was shown that all important prerequisites
for obtaining reliable kinetic data for the reaction were fulfilled.
The effects on rate and enantiomeric excess of catalyst loading,
modifier and substrate concentrations, hydrogen pressure, and
temperature were determined for the unmodified and the modi-
fied system. The modified reaction was approximately 20–30 times
faster than the unmodified reaction. A significant increase of the
enantiomeric excess from 1 to 40 bar was noticed. Apparent ac-
tivation energies were estimated to be 4–6 kcal/mol. Rate equa-
tions were developed for various kinetic schemes on the basis of the
Langmuir–Hinshelwood–Hougen–Watson formalism and fitted to
the kinetic data. Several such schemes described the measured data
reasonably well and in most cases explanations other than the one
considered to be the most plausible were also in agreement with
our data. For the unmodified catalyst, we propose a competitive
adsorption of the -ketoester and hydrogen and the addition of the
first hydrogen atom to be rate determining. On a chiral site, the
rate determining step (RDS) to the major enantiomer is proposed
to be the addition of the second hydrogen, whereas the RDS for
the minor enantiomer remains the first H addition. On the basis of
this interpretation, different proposals advanced in the literature
for the mode of action of the cinchona modified Pt catalyst were
compared.
c
1998 Academic Press
We reasoned that a comparison of the kinetics of the
modified and the unmodified catalyst might provide more
insight into the role of the cinchona molecule during the
hydrogenation. For example, a very important question is
that of whether the modifier simply causes a change in
the rate determining step or leads to a completely differ-
ent mechanism for the hydrogenation. In this contribution,
we present a kinetic study investigating the influence of
hydrogen pressure, etpy concentration, modifier concen-
tration, catalyst loading, and temperature on rate and the
enantiomeric excess (e.e.). The results are interpreted in
the framework of the Langmuir–Hinshelwood–Hougen–
Watson (LHHW) formalism and consistent mechanistic
schemes are proposed. This paper discusses in depth data
that were published partly in (19) and, more importantly,
INTRODUCTION
Enantioselective hydrogenation is a topic of current in-
terest not only from a synthetic but also from a mecha-
nistic point of view (1). Until recently, most efforts con-
cerning the elucidation of the reaction mechanism have
been dedicated to homogeneous catalytic systems (2). Chi-
1 To whom correspondence should be addressed. Fax: ++41-61-697-
8453. E-mail: hans-ulrich.blaser@sn.novartis.com.
2 Present address: Department of Chemical Engineering, National
University of Singapore, 10 Kent Ridge Crescent, Singapore 119260.
282
0021-9517/98 $25.00
c
Copyright
1998 by Academic Press
All rights of reproduction in any form reserved.

KINETIC STUDIES OF ETHYL PYRUVATE HYDROGENATION
283
temperature and 1 bar H2. After neutralization with 50%
NaOH under stirring, the precipitated HCd was filtered off
and washed with 1.5 liter deionized water until chloride-
free (AgNO3 test). The white solid was dried in vacuum
at 100 C. Yield 87.4 g (86.8% ). mp 233–235 C. Elemen-
tal analysis (C19H24N2O 0.04 H2O, 297.13). Calc.: C 76.80,
H 8.17, N 9.43, H2O 0.24. Found: C 76.49, H 8.08, N 9.26,
H2O 0.24. 5% Pt/Al2O3 (Engelhard 4759) with a Pt dis-
persion of 0.24 (determined by CO adsorption (18)) was
prereduced for 2 h at 400 C under flowing H2. This catalyst
could be stored for two weeks under air with no significant
loss in selectivity or activity. Toluene (Fluka, puriss. p.a.)
was used as received.
FIG. 1. Hydrogenation of etpy. Reaction scheme, catalyst, and mod-
ifier structure.
Kinetic Measurements
gives the derivation of all kinetic models that were consid-
ered.
All reactions were carried out in a thermostated 50 ml
autoclave equipped with baffles, two thermocouples, and a
magnetic stirrer (л = 5 mm, l = 20 mm). A reservoir and a
pressure regulator allowed experiments to be carried out
under isobaric conditions. Pressure and temperature were
measured in the autoclave and in the reservoir every 10 s
and recorded with a Micromec data recorder (Suprag AG,
Zu¨rich, Switzerland).
For a description of the measured datasets see Table 1.
In the unmodified system (Series U1–U9) 5–450 mg pre-
reduced 5% Pt/Al2O3, 1–30 ml etpy, and 0–29 ml toluene
were introduced into the autoclave in this sequence. The
METHODS
Reagents
Etpy was purchased from Fluka (purum) and Boehringer
(purum). Before use, it was distilled and kept at 5 C for a
maximum of 3 weeks. 10,11-Dihydrocinchonidine (HCd)
was prepared from commercially available cinchonidine
(100 g, Fluka, purum) by hydrogenation with 5% Pd/C
(10 g, Engelhard 4522) in 600 ml 1 M aqueous HCl at room
TABLE 1
Datasets U1–U9 and M1–M10
Activity
coefficient
pH2
(bar)
[Mod]
(mM)
[E]
(M)
Cat
(mg)
T
( C)
Number of
datapoints
Series
Supplier
U1
U2
U3
U4
U5
U6
U7
U8
U9
Boeringer
Fluka
Fluka
Fluka
Boeringer
Fluka
Fluka
Boeringer
Fluka
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1.00
21
101
0–34
0–161
0–101
21
101
21
101
0
0
0
0
0
0
0
0
0
3.0
3.6
1.4
1.4
3.0
0–9.0
0–9.0
3.0
50–350
50–450
100
400
250
250
400
250
400
20
22
20
22
20
20
22
0–70
0–70
4
9
20
11
6
13
11
8
1.4
8
M1
M2
M3
M3a
M4
M4a
M4b
M5
M6
M7
Boeringer
Boeringer
Boeringer
Fluka I
Fluka II
Fluka I
1.00
1.00
1.00
0.86
2.21
0.86
0.86
1.00
1.00
2.21
—
21
0–101
0–101
1–101
0–141
0–41
0–101
21
21
101
21
101
1.1E-03
1.1E-03
1.1E-03
1.1E-03
6.7E-04
6.7E-04
6.7E-04
1.1E-03
1.1E-03
6.7E-04
1.1E-03
6.7E-04
0-1.1E.3
3.0
3.0
3.0
3.0
5.0
1.8
5.0
0–3.9
0–9.0
0–9.0
3.0
5–150
50
50
50
25
25
50
50
50
25
50
25
50
20
20
20
20
20
20
20
20
20
7
6
7
8
13
6
11
12
15
15
8
Fluka I
Boeringer
Boeringer
Fluka II
Boeringer
Fluka I
22
M8
M9
M10
0–70
0–70
22
—
0.86
1.8
3.0
8
22
Fluka I
101

284
BLASER ET AL.
autoclave was purged five times with argon (21 bar) while and “linear.” In some cases, e.g., model MV2 and MV6 for
stirring and three times with hydrogen (21 bar) without stir- the unmodified system (see below), local minima were ob-
ring. Then both the reservoir and the autoclave were pres- served. To circumvent this problem, different sets of initial
surized with hydrogen. Negligible hydrogen absorption oc- estimates were tested. With our data, however, only mod-
curred prior to stirring, and the experiments were started els with a maximum of three (unmodified system) or five
by turning the magnetic stirrer on (900–1300 rpm). Under parameters (modified system) always gave unambiguous
the reaction conditions, the liquid phase became saturated results.
with hydrogen within 1–3 min. After 10–60 min, the auto-
clave was purged with argon, and the solution was filtered.
After evaporating the toluene, the products were analyzed.
In all experiments, the conversion was higher than 15% .
For the modified system (Series M1–M10) experiments
were similar to those described above. However, only
3–75 mg prereduced 5% Pt/Al2O3 was used and 5–20 mg
HCd was added after loading the catalyst.
Conversion to ethyl lactate (etlac) was calculated by the
pressure drop in the reservoir and confirmed by glc (2 m
OV 101; 45 C). The e.e.’s were determined by glc after
derivatization ofetlacwith isopropylisocyanate (25m capil-
lary Chirasil-L-Val; 140 C); in some instances both con-
version and e.e. were determined by analyzing the crude
reaction mixture (30 m capillary -Dex 100; 75 C). The
R-enantiomer of etlac was the major enantiomer in all ex-
periments with HCd modifier.
The initial rates of hydrogenation were calculated from
the pressure drop in the reservoir (r1 = dH2/dt in mmol
H2/min, takingthe compressibilityofhydrogen into account
(20)). Usually, the rate of hydrogenation was evaluated at
reaction times of 3–8 min. Both a graphical and a linear
least square method were used for determination of the
initial rates. The results were similar for both methods with
a reproducibility of 10% .
RESULTS
In this section, two issues are addressed. First, the pre-
requisites for obtaining reliable kinetic data are discussed.
Next, the dependencies of the observed rates and enan-
tioselectivities on catalyst loading, etpy concentration, hy-
drogen pressure, and temperature will be described and
discussed. A preliminary description of selected results has
appeared (19).
Prerequisites for Obtaining Reliable Kinetic Data
Stoichiometry and Stability of the System Components
Under the conditions used in this experimental study, the
selectivity to etlac was >99% as determined by glc, and the
amount of hydrogen consumed was equal to the amount of
etlac formed. Problems due to deterioration of both etpy
and cinchona alkaloids are known to occur. However, such
difficulties usually occur under reaction conditions consid-
erably different from those in this study and at longer re-
actions times (23–25). To minimize these problems, only
initial rate data were used in this study.
Reproducibility of Rate and e.e. Measurement
For all calculations below, the actual hydrogen concen-
tration in the reaction mixture was used. The solubility of
hydrogen in toluene and mixtures of toluene and etpy was
determined by measuring the amount of gas dissolved at
20 and 100 bar and at five different etpy concentrations. A
simple linear approach with two adjustable parameters a
and b described the concentration dependence of hydro-
gen on pressure and varying etpy/toluene ratios well. For
the temperature dependence, the approach of (21) was used
(exponential part of the equation). The constant c was de-
termined from a third data set with varying temperatures
((22), data not shown). The following expression allows one
to calculate the hydrogen concentration:
The evaluation of the measured rates by the graphical
or the least square procedure allowed a reproducibility of
approximately 10% . However, in the course of this inves-
tigation, which lasted several years, a number of different
batches of etpy, catalyst, and solvents were used. In addi-
tion, different autoclaves were used and different investiga-
tors carried out experiments. Not surprisingly, variations of
the optical yields and especially of the rates were observed
for series having the same initial conditions but performed
at different dates during the investigation. The etpy quality
was the major factor affecting the observed rates (26, 27). In
order to be able to fit the kinetic models to the data points
of all series, it was necessary to introduce an “activity fac-
tor” for the modified system. For relative rates see Table 1.
At the moment we have no reasonable explanation for this
phenomenon. For the unmodified system, the differences
were much smaller and no correction was made.
1
293
1
)
[H2] = pH2(a + b[etpy])ec(
273+T
mol
3
a = 2.633 10
liter bar
1
5
Induction Periods
b = 7.5 10
c = 764 K.
bar
An increase in e.e. with conversion was described for the
All curve fitting was done using the Excel-Solver subrou- hydrogenation of -keto esters using tartrate-modified Ni
tine (MS Excel 7.0) with the settings “Newton,” “Forward,” catalysts in the liquid (7) and the gas phase (28). Careful

KINETIC STUDIES OF ETHYL PYRUVATE HYDROGENATION
285
analysis of our reaction mixtures has confirmed the find-
ings recently published by the group of Blackmond (29, 30)
that the initial e.e.’s determined at very low conversions are
consistently lower than subsequent e.e.’s. This can lead to
a significant difference of 3–30% between measured e.e. at
a specific reaction time and the actual incremental e.e. be-
ing produced (19). Indeed, it is important to recognize that
measured e.e.’s are the average value obtained over the
preceding reaction period. In the present investigation, the
e.e. was consistently measured at conversions greater than
15% . Consequently, calculations indicated that the differ-
ence between the measured and the incremental e.e.’s is
smaller than 5% . As reported by other investigators (13,
17, 31), S-shaped curves can arise in plots of pressure ver-
sus time and similar behavior was occasionally obtained in
this study. In these cases, the initial rates were lower than
the maximum rates but the difference was less than 10–
20% . Such induction phenomena indicate that these mod-
ified heterogeneous catalysts can require appreciable time
before they reach maximal activity. However, the exact rea-
son for this behavior is still unknown.
FIG. 2. Dependence of e.e. on the polarity of the reaction mixture.
(᭜) [etpy] = 0.4–9.0 M in toluene, 22 C, data from M7; (ᮀ) [etpy] = 3.4 M
in solvents of various d.e., 20 C, 100 mg 5% Pt/Al2O3 (Engelhard 4759),
(33). d.e. for reaction mixtures from a linear interpolation between the
values of the solvent and etpy.
of the reaction mixture. The d.e. dependence found here
is very similar to the one observed with different solvents
(34). From this we conclude that the observed e.e. changes
are due mainly to changes in polarity. On the other hand,
the effect of etpy concentration on the rate is due to the
changing substrate concentration (see below).
Mass Transfer
In a previous study, mass transfer effects on the rate and
e.e. during the hydrogenation of etpy with cinchona modi-
fied Pt/Al2O3 was studied in detail (11). It was shown that
the overall mass transfer rate for hydrogen into the liquid
phase significantly affected both rate and e.e. (10, 11). Up-
per limits for catalyst loading and minimum stirring speeds
were identified for the autoclaves used in this study, and
consequently, it was possible to avoid regions of hydrogen
transfer limitation. Except for reactions with very high cata-
lyst loading, we can exclude that hydrogen diffusion was
rate limiting. One to three minutes was required to satu-
rate the liquid phase with hydrogen for the experiments
presented here. These initial mass transfer difficulties are
one factor that can contribute to the observed induction
period and the lower initial e.e.’s.
Conclusions with Respect to the Present Investigation
We must admit that some of the above prerequisites are
not met as well as we thought at the beginning of our investi-
gation. In some respects, the hydrogenation of -ketoesters
on Pt catalysts modified with cinchona alkaloids is more
complex than we (and others as well) expected. Neverthe-
less, we are convinced that the data we have gathered allow
not only a qualitative but also a quantitative interpretation.
On the other hand, it is also clear that the quality of our
results does not allow the discrimination of closely related
kinetic models.
Kinetic Measurements
Effects of Changing Polarity of the Reaction Mixture
Table 1 describes the datasets used for this investigation
Changes in the dielectric constant of the reaction mix- (series U1–U9 for the unmodified system, series M1–M10
ture might also affect the e.e. and rate. Two effects have to for the modified system; tables with the detailed results are
be considered: the conversion of etpy to etlac during the available from the authors). Calculated curves in Figs. 3–7
reaction and the changing of the etpy/toluene ratio. The di- represent the best fits obtained with the kinetic model vari-
electric constant (d.e.) of etlac is 15.4 at 30 C (32), while that ants MV1 (unmodified) and MV8 (modified, see below). In
of etpy is not known. Since molecular modeling gives very the preliminary publication (19), the fitting was carried out
similar dipole moments for both etpy and etlac (33), we es- with slightly different assumptions and kinetic models.
timate a d.e. for etlac of approximately 15. This means that
The effect of catalyst concentration for the unmodified
conversion of etpy to etlac does not affect the polarity of system was investigated for the interval of 50–350 mg cata-
the reaction mixture very much. This is different for chang- lyst at 21 bar and for the interval 50–450 mg catalyst at
ing etpy concentrations in toluene, where the d.e. changes 101 bar (U1 and U2). The effect of catalyst loading for the
from approximately 3 at very low etpy concentration to 15 modified system was examined with 3 M etpy and 1.1 mM
when the reaction is run in neat etpy. Figure 2 illustrates HCd in the reaction mixture at 21 bar (M1) for the interval
the apparent dependence of e.e. on the dielectric constant 5–150 mg catalyst. In both cases, the initial rates showed a

286
BLASER ET AL.
FIG. 5. Effect of hydrogen pressure on rate and e.e. of the modified
FIG. 3. Effect of hydrogen pressure on rate for the unmodified sys-
system: Series M4a and M4b (3.0 M etpy). (᭺) Measured rate for series
M4a; ( ) measured rate for series M4b, ( ) and (—) calculated rates; (ᮀ)
measured e.e. for series M4b; (- - -) calculated e.e.
tem. ( ) Measured rate for series U3; (᭺) measured rate for series U4;
(—) calculated rate.
first-order dependence on the catalyst concentration except
for the modified catalyst above 50 mg catalyst, when the
reaction order in hydrogen began to decrease due to non-
negligible mass transport resistance for hydrogen. The op-
tical yields obtained were constant (e.e. = 76 1% ) for the
interval 5–75 mg catalyst, but began to decrease at higher
catalyst loading, i.e., when the system got hydrogen starved.
These results are in agreement with those in several stud-
ies for the hydrogenation of etpy with modified Pt catalysts
(11, 35–37).
The effect of the hydrogen concentration for the unmod-
ified system in the liquid phase was studied for hydrogen
pressures of 3–160 bar at etpy concentrations of 1.4 and
3.0 M (see Fig. 3). The specific measured rates (rate per
gram catalyst) were similar for 1.4 and 3 M etpy concentra-
tion. The rates increased with increasing hydrogen partial
pressure, but there is a clear trend to lower reaction order
above 60 bar. This type of saturation curve for hydrogen
is rather uncommon, and we are unaware of similar results
for ketone hydrogenations in the literature.
For the modified system, the effect of the hydrogen con-
centration was studied at hydrogen pressures of 2–140 bar
at etpy concentrations of 1.8 and 3.0 M (see Figs. 4 and
5). The rate of hydrogenation was essentially proportional
to the hydrogen partial pressure over the entire interval.
The observed e.e.’s increased noticeably in the interval of
0–40 bar, and then remained constant.
The effect of hydrogen pressure has been investigated
by several groups and most reported a first-order rate de-
pendence. Increases in enantioselectivity were reported by
Augustine et al. (35), the group of Blackmond (29, 38), and
by Bo¨hmer et al. (39). Unsystematic variations (9, 13) and
a drop (36) in e.e. with increasing pressure were described
as well, plausibly explained for the latter case by a fast hy-
drogenation of the modifier at higher pressures.
The effect of the ethyl pyruvate concentration in absence
of HCd was studied in a concentration range of 0.2–9 M
(pure etpy), at both 21 and 101 bar hydrogen pressure (see
Fig. 6). In both cases, maxima are observed. They occur at
0.6 M etpy for the 21 bar experiments and at 2.0 M for the
FIG. 4. Effect of hydrogen pressure on rate and e.e. of the modified
FIG. 6. Effect of etpy concentration for the unmodified system.
system: Series M3a (1.8 M etpy). ( ) Measured rate; (—) calculated rate; (᭹) Measured rate for series U6; ( ) measured rate for series U7; (
(ᮀ) measured e.e.; ( ) calculated e.e. and (—) calculated rates.
)

KINETIC STUDIES OF ETHYL PYRUVATE HYDROGENATION
287
1/T showed in both cases a linear behavior over the
whole temperature range. The resulting values for the
apparent activation energy were low, namely Ea = 6.1
0.4 kcal/mol (26 1.6 kJ/mol, 21 bar) and 4.5 0.3 kcal/mol
(19 + 1.4 kJ/mol, 101 bar). In presence of modifier, the re-
action was studied between 0 and 70 C at both 21 bar (3 M
etpy, M8) and 101 bar (1.8 M etpy, M9). The plot of ln
robs versus 1/T showed in both cases a linear behavior be-
tween 0 and 50 C and a slight deviation for 70 C. The re-
sulting values for the apparent activation energy between
0 and 50 C were also low, namely Ea = 4.1 0.3 kcal/mol
(17.3 1.3 kJ/mol, 21 bar) and 5.06 0.21 kcal/mol (21.2
0.9 kJ/mol, 101 bar). The e.e. was in both cases constant
from 0 to 50 C, but decreased noticeably from 50 to 70 C,
especially at 101 bar. An activation energy of 38 kJ/mol
was reported by Sutherland et al. (12) but both rate and e.e.
strongly decreased above 50 C. These deviations are prob-
ably due to desorption of the modifier from the Pt surface
at higher temperatures.
FIG. 7. Effect of etpy concentration for the modified system, series
M6 (21 bar). ( ) Measured rate; (—) calculated rate; (ᮀ) measured e.e.;
(- - -) calculated e.e.
101 bar experiments. At higher concentrations of etpy the
rates decrease. In neat etpy (9 M) the rates are approxi-
mately 60% of their maximal value.
DISCUSSION
In the modified system, the effect of the etpy concentra-
tion wasstudied in the range of0.4to 9M etpyat both 21and
101 bar hydrogen pressure (M6, M7; Fig. 7 and (19)). As for
the unmodified system, the rates showed a maximum but
this was less pronounced. For the reaction at 21 bar there
was some problem with the reproducibility but a maximum
was observed between 1 and 3 M etpy. At 101 bar, the maxi-
mum rate was obtained at approximately 5 M etpy and it
was 32 times higher than in the unmodified system. At both
low and high pressure, a weak e.e. maximum was observed
at approximately 2 M etpy. As discussed above, this effect
is reproducible and is attributed to changes in the polarity
of the reaction solution. In comparison, literature reports
described dependencies similar to those described above
(36, 39), almost constant rates (23, 29) but an increase of
e.e. from 20–30% to >60% (23) as well as strong substrate
inhibition (35).
Anyadequate mechanisticmodelmust be able to account
for the following significant kinetic effects:
the first-order dependence of the observed rate on the
catalyst loadingfor both the unmodified and the unmodified
system;
the local rate maximum as a function of the etpy con-
centration in both the unmodified and the unmodified sys-
tem;
the different dependencies of rate on hydrogen partial
pressure observed with the unmodified and the unmodified
system;
the hydrogen pressure dependence of the e.e.;
the dependence of the rate on modifier concentration
(“first order” with saturation).
Development of Kinetic Models and of Rate
and e.e. Equations
The effect of the modifier concentration for a fully modi-
fied catalyst was studied, and a HCd concentration between
Several descriptions and discussions of the kinetics of
0.03 and 0.6 mM is necessary in the enantioselective hydro- the hydrogenation of ketones have appeared (41, 42). Usu-
genation of etpy in toluene (10, 40). This condition was ful- ally, the kinetic results are discussed in the framework of
filled for all modified data sets except for M10, where the the Langmuir–Hinshelwood (LH) formalism; i.e., it is as-
modifier concentration was varied between 0 and 1.1 mM. sumed that the reaction occur in the adsorbed state. For
M10 shows the typical saturation curve with maximum e.e. the practical application of this formalism, two approaches
and rate already at very low modified concentration as de- have been used. In most cases the LHHW approximation
scribed by Garland et al. (10). M10 was incorporated in our (43) is applied, assuming that adsorbed species are in fast
kinetic analysis and it was shown that these data can be fit- equilibrium before the rate limiting step, characterized by
ted well even with a simplified model (adsorption of only adsorption constants Ki and reacting with rate constants
one HCd) with only small deviation at higher HCd concen- ki. The observed rate is then proportional to the surface
trations.
coverage of species that react in the rate determining step
The effect of the temperature on the initial rate was stud- and the rate constant k, leading to the typical LHHW rate
ied between 0 and 70 C at both 21 bar (3 M etpy, U8) equations (kinetic terms in the numerator and the adsorp-
and 101 bar (1.4 M etpy, U9). The plot of ln robs versus tion term in the denominator). The second approach takes

288
BLASER ET AL.
the forward and back reaction of every elementary step the back reaction was not considered. Depending on which
before the RDS into account and applies a steady state step in this sequence is assumed to be rate determining,
assumption, i.e., assumes that the concentration of these different rate expressions are finally obtained.
intermediates are constant. This approach was for exam-
ple used by Halpern for the Rh-catalyzed homogeneous
Adsorption and
Reaction step
rate constants
KH , (kH and k
hydrogenation of enamides (44) and recently also for the
hydrogenation of etpy with the Pt cinchona system (15).
Even though we fully realize that the LHHW formalism
is an oversimplification, we use it here exclusively for the
analysis of both the modified and the unmodified system.
The main reason for this choice is that the steady-state ap-
proach (15) would involve more adjustable constants than
we can handle with our experimental basis. Asa further sim-
plification, we assume only one kind of active center that
can competitively adsorb hydrogen, etpy, or HCd and we
H2 + 2
2 -H
)
[1]
[2]
H
E +
-H + -E
-H + -EH
-E
KE , (kE and k
)
E
-EH +
-EH2 +
K
K
EH , (kEH and k
)
[3]
EH
, (k
and k
EH2
) [4]
[5]
EH2
EH2
-EH2 → + EH2
kD , fast
Using classical LHHW calculations, the following rate equ-
ations were determined:
√
kEH KE [E] KH [H2]
√
Model variant MV1: RDS first H-addition
Model variant MV2: RDS second H-addition
rcalc
rcalc
=
=
[6]
[7]
(1 + KE [E] + KH [H2])2
kEH KE KH KEH [E][H2]
2
√
√
(1 + KE [E] + KH [H2] + KE KEH [E] KH [H2])2
√
kEH0 [E] KH [H2]
√
Model variant MV3: RDS reaction of etpy
from solution with -H (Eley-Rideal)
rcalc
=
=
=
=
[8]
[9]
1 + KH [H2]
kH [H2]
(1 + KE [E])2
Model variant MV4: RDS dissociative adsorption rcalc
of hydrogen (etpy adsorbed reversibly)
kE [E]
√
Model variant MV5: RDS ketone adsorption
(hydrogen adsorbed reversibly)
rcalc
rcalc
[10]
[11]
1 + KH [H2]
kD KE KEH KEH KH [E][H2]
2
√
√
Model variant MV6: RDS desorption of etlac
1 + KE [E] + KH [H2] + KE KEH [E] KH [H2] + KE KEH KEH [E]KH [H2]
2
neglect the adsorption of etlac (because of low conversion) Many other schemes could be considered, e.g., adsorption
as well as of the solvent. Generally, different combinations of the -ketoester on two sites as proposed by Simons et al.
of elementary steps can give similar or equal sets of rate ex- (45), or on a dual site, the simultaneous addition of two hy-
pressions in the LHHW formalism (43). It should be noted drogen atoms in one step, etc., as discussed in great detail
that the number of parameters increases rapidly with the for the acetone hydrogenation by Simonikova et al. (42).
complexity of a kinetic model often affording a fit as good as Several of their more complex model variants were indeed
or better than a simpler expression, due only to the increase fitted with about the same results as MV1 and MV2. Be-
in the number of parameters and not necessarily because of cause these models have more adjustable variables but do
an intrinsically better model. Our goal was to find the sim- not give more insight into the reaction mechanism, only the
plest plausible model with the lowest number of adjustable most simple ones were further pursued.
parameters that can explain all significant effects and that
gives a reasonable fit to our measured data.
Discussion of the Unmodified Catalyst
Kinetic Models
Discrimination of Kinetic Models
The data sets U1–U9 were used to test and discriminate
the kinetic models MV1–MV6. The quality of fit was judged
by the unweighted least square method as well as by vi-
sual inspection. A simple visual inspection of the observed
rate data rules out the models MV3–MV5. The model MV6
gives a reasonable fit. However, five variables are required
to fit the data, and there are no indications that the des-
orption of etlac is the slow step. Therefore, MV6 was also
excluded from further consideration. The principal differ-
ence between the rate equations of models MV1 and MV2
is a half-order dependence of the rate on [H2] in MV1 and
For the hydrogenation of etpy, the following equilibria
and elementary steps are considered: adsorption of hydro-
gen and etpy (Eqs. [1] and [2]) on the active sites, fast sur-
face diffusion, stepwise addition of the dissociated H2 to the
==
C
O bond (Eqs. [3] and [4]), and desorption of the product
alcohol (Eq. [5]). Based on our previous calculations and
experimental studies, significant bulk diffusion resistance
is neglected (11). Since all rates obtained are initial rates,

KINETIC STUDIES OF ETHYL PYRUVATE HYDROGENATION
289
sites of the Pt surface; the symbol ✪ is used to describe a
modified site. It is plausible that not all racemic active sites
can be modified. A similar approach was used by Tai and
Harada for the surface of Raney nickel (7). However, ki-
netic models cannot discriminate between a fully or only
partially modifiable surface and indeed both scenarios can
describe existing kinetic results (10, 40). The kinetic treat-
ment can formally be carried out as for the unmodified site,
except for an important difference: Since ✪ is chiral, the
interactions with the prochiral etpy lead to intermediates
that are diastereomeric, i.e., they are different in geometry
as well as in energy and reactivity. This gives more compli-
cated kinetic schemes because the catalytic cycles leading
to (R)-etlac and to (S)-etlac have to be considered sepa-
rately. The same approach indeed was used by Landis and
Halpern (44) as well as by Tai and Harada (7). For the sake
of simplicity we assumed that the activation of H2 occurs
only on unmodified sites and that adsorbed hydrogen and
etpy undergo fast surface migration to other active sites.
It must be stressed that these assumptions are not neces-
sary requirements to arrive at the results and conclusions
described below. For several kinetic models we have ob-
tained the same quality of fit to our data assuming partially
modifiable surfaces, or by H2 activation on ✪.
TABLE 2
Kinetic Constants and Least Square Sums for Models
MV1, MV2, and MV6
MV1
MV2
MV6
Least square:
42.2
110
40.92
133
56.77
23.0
2
I
k (1000
mol ):
g
min
KH (M 1):
0.475
0.497
—
16,500
17.4
0.0044
—
0.00162
66700
2820
KE (M 1):
KEH
KEH
:
:
—
0.00330
2
a first-order dependence in MV2. In spite of this significant
difference, it was possible to fit the data U1–U9 equally well
with both MV1and MV2(Table 2) because ofthe saturation
behavior. This means that we cannot distinguish between
the two alternatives. However, since the adsorption con-
stants for MV1 are more reasonable than those for MV2,
and MV2 needs one variable more to obtain a similar fit,
MV1 is slightly favored. Accordingly, the calculated curves
in Figs. 5a, 6, and 8 were obtained using this model variant.
Both models are in agreement with the reaction intermedi-
ates most often depicted for the hydrogenation of ketones
on noble metal surfaces (see Fig. 8).
The Kinetic Models
Discussion of the Modified Catalyst
For the modified system the following additional assump-
tions compared to the unmodified case were made:
The Modified Site (✪)
The unmodified sites react as described by the model
MV1 or MV2 (same adsorption and rate constants).
A modified site, ✪ consists of one cinchona molecule
adsorbed reversibly on a platinum site, :
As discussed by Blaser et al. (10, 40), the modification of
the Pt catalyst can be described quite well by assuming a
reversible adsorption of the cinchona molecules on active
+ HCd
✪
Km0 .
[12]
The addition of the H-atoms to the adsorbed etpy oc-
curs from the side of the metal. Therefore, etpy adsorbed
via the re-face of the keto group leads to (S)-etlac whereas
that adsorbed via the si-face leads to (R)-etlac.
Detailed derivation of MV7. In this variant it is assumed
that all active sites are modifiable and that etpy is reversibly
adsorbed on a modified site and reacts with hydrogen that
is activated at an unmodified site ( -H). In order to have
as few adjustable parameters as possible, KsAiE and KrAeE are
taken as equal even though the two species are known to
be diastereomeric. As a consequence, the fitting will lead
to apparent kinetic constants KrAeEH , KsAiEH , and KsAiEH
:
2
E + ✪ ✪-si/re-E KEA.
[13]
FIG. 8. Artist’s view of the structure of adsorbed reaction interme-
diates on the unmodified Pt surface. (a) Adsorbed etpy, si-face; (b) half-
hydrogenated ketone (first H addition to carbonyl oxygen, si-face); (c) half
hydrogenated ketone (first addition to C, R-form); (d) weakly adsorbed
(R)-lactate.
For the formation of (S)-etlac it is assumed that the first
addition of H to the adsorbed etpy is rate determining

290
BLASER ET AL.
(analogous to MV1):
For the formation of (S)-etlac it is again assumed that the
first addition of ✪-H to -etpy give ✪-re-EH (Eq. [30]) is
rate determining, followed by a fast addition of the second
H and desorption of the product (Eqs. [15] and [16]).
✪-re-E + H- → ✪-re-EH +
krAeEH (RDS to (S)-etlac)
[14]
✪-re-EH + H- → ✪-(S)-EH2 +
fast
[15]
✪-H + -E → ✪-re-EH +
krBeEH (RDS to (S)-etlac)
[30]
✪-(S)-EH2 → ✪ + (S)-etlac fast.
[16]
For (R)-etlac the second addition of H is assumed to be rate
determining (as for MV2):
For (R)-etlac it is also assumed that the second addition
of H is assumed to be rate determining, followed by fast
desorption (Eqs. [31], [18], and [19]):
[17]
✪-si-E + H-
✪-si-EH +
KsAiEH
✪-si-EH + H- → ✪-(R)-EH2 + ksAiEH (RDS to (R)-etlac)
✪-H + -E
✪-si-EH +
KsBiEH
.
[31]
2
[18]
The balance for the modified active sites (10% of the total
[19]
Pt atoms) is
✪-(R)-EH2 → ✪ + (R)-etlac fast.
The mass balance for the active sites is
[ ]+[ -H]+[ -E]+[✪]+[✪-H]+[✪-si-EH] = 0.1. [32]
[ ] + [ -H] + [ -E] + [✪] + [✪-si-E] + [✪-re-E]
+ [✪-si-EH] = 1.
Solving for gives the adsorption term
[20]
0.1
B1 = [ ] =
.
p
p
1 +
KH [H2] + KE [E] + Km0 [HCd]{1 + KMB
KH [H2](1 + KE [E])}
Using the equilibria defined above, the following adsorp-
tion term A is obtained:
[33]
1
A = [ ] =
.
p
p
The balance for the unmodified site (90% of the total) is
1 +
KH [H2] + KE [E] + Km0 [HCd]{1 + 2KEA[E](1 + KsAiEH
KH [H2])}
[ ] + [ -H] + [ -E] = 0.9.
[34]
[21]
The individual rates for formation of the (S)- and the (R)-
enantiomers of etlac on the modified (rmod) and unmodified
(rrac) sites can then be calculated as
Solving for gives the adsorption term
0.9
p
B2 =
[35]
1 + KH [H2] + KE [E]
rmodS = krAeEH [✪-E][ -H]
p
= krAeEH KEA[E] KH [H2]Km0 [HCd]A2
[22]
leading to the rate equations for (S)- and (R)-etlac, pro-
duced on the modified sites
rmodR = ksAiEH [✪-si-EH][ -H]
2
rmodS = krBeEH [ -E][✪-H]
= ksAiEH KsAiEH KEA[E]KH [H2]Km0 [HCd]A2 [23]
p
2
= krBeEH KE [E]Km0 [HCd]KMB KH [H2]B1 B2
[36]
[37]
p
rrac = kEH KE [E] KH [H2]A2 (see MV1). [24]
rmodR = ksBiEH [✪-si-EH][ -H]
2
The observed rates and e.e.’s can be calculated as
= ksBiEH KE KsBiEH Km0 [HCd]KMB KH [H2]B1 B2
2
p
rcalc = rmodR + rmodS + rrac
rR = rmodR + 0.5rrac
rS = rmodS + 0.5rrac
[25]
[26]
[27]
[28]
rrac = kEH KH [H2]KE [E](B1 + B2)2 (see MV1). [38]
The observed rates and e.e.’s can be calculated as shown
above (Eqs. [25]–[27]).
e.e.calc = 100(rR rS)/(rR + rS).
Other model variants. The same procedure can be car-
ried out assuming different RDS for both the two model
variants described above. Several such model variants were
actually derived and tested. Table 3 lists the rate determin-
ing steps for the most plausible models MV7–MV12.
Derivation ofMV8. Hydrogen isactivated on (Eq. [1])
and is transferred to ✪ giving a protonated adsorbed cin-
chona modifier. Etpy is also adsorbed on a nonmodified site
(Eq. [2]) and can migrate to the protonated modified site.
In addition, it is arbitrarily assumed that only 10% of the
active sites are modifiable:
Discrimination of Kinetic Models
Data sets M1–M10 were used to test the models for the
-H + ✪ ✪-H +
KmB.
[29] modified system. The quality of fit to both the rate and

KINETIC STUDIES OF ETHYL PYRUVATE HYDROGENATION
291
and hydrogen. when comparing the rate constants for the
modified and the unmodified system, the formation of the
half-hydrogenated intermediate is accelerated much more
for the (R)-enantiomer than for the (S)-enantiomer, lead-
ing to both enantiodiscrimination and ligand acceleration.
For the calculated curves of rate and e.e. as depicted in
Figs. 4 and 7 we used model variant MV8. Indeed with the
exception of the rates in Fig. 7, the fit is satisfactory.
We also tried to fit our data to the steady-state model dis-
cussed by Wang et al. (15). Their treatment is based on the
assumption that the reaction occursthrough irreversible ad-
dition of adsorbed hydrogen to adsorbed etpy as depicted
in Eqs. [14] and [17], contributions from unmodified sites
are neglected. Steady-state treatment leads to rate equa-
tions for rR and rS with seven adjustable parameters that
are able to model the dependence of e.e. on pressure very
TABLE 3
Characterization of Different Model Variants
for the Modified System
Formation of
(R)-etlac:
RDS addn of RDS addn of
Formation of
(S)-etlac:
Percentage of
modifiable
sites
Model
variant
Unmodified
sites
MV7
MV8
MV9
MV10
MV11
MV12
MV1
MV1
MV1
MV1
MV1
MV2
Second H
Second H
First H
Second H
First H
First H
First H
First H
Second H
Second H
First H
100
10
100
100
100
100
Second H
e.e. data was again judged by the unweighted least square
method as well as by visual inspection. For the unmodi-
fied sites the above-determined parameters for MV1 and
MV2 were used. Simple visual inspection of the measured
data leaves MV7, MV8, and MV12 as the only valid candi-
dates. MV9 and MV11 (first H-addition is rate determining
for the formation of the R-enantiomer) are not consistent
with the observed first-order dependence of rate on hydro-
gen pressure. MV10 (second H addition is rate determining
for the formation of both R- and S-etlac) gives a good fit
to the rate of reaction for the data sets M1–M10 but be-
cause the formation of the R- and the S-enantiomer have
the same order in hydrogen, the e.e. dependence on hy-
drogen concentration is not adequately modeled (the same
holds for MV9).
well. However, the hydrogen dependence of the rate is cal-
p
culated to be [H2] or lower. This means that the data of
our modified system, which shows first-order rate depen-
dence on the hydrogen pressure over a wide range, cannot
be explained by the approach. For this reason, the steady-
state approach was no longer pursued.
Proposed Mechanistic Schemes
The detailed reaction mechanism of the heterogeneous
hydrogenation of ketones is much less understood than the
==
analogoushydrogenation ofC C bonds(46). Nevertheless,
there is general agreement that the addition of hydrogen to
==
the C O group occurs in the adsorbed state in a stepwise
manner. Accordingly, the rate determining step is assumed
to be either the first or the second hydrogen addition. There
is less agreement as to whether the first addition occurs at
the oxygen (probably as a proton) or at the carbon atom of
the carbonyl group (probably as a hydride). Both reaction
pathways have been proposed to explain results observed
for different catalysts and different ketones (47). Our inter-
pretation of the kinetic results obtained for the unmodified
system fits quite well into this framework. But kinetic data
provide of course no information on the structure of the
decisive species, e.g., whether the addition occurs at the
carbon or oxygen end of the carbonyl group or how these
intermediates are bound to the Pt surface.
The relatively simple models MV7, MV8, and MV12 de-
scribe with only five adjustable parameters the rate and e.e.
dependence on hydrogen, etpy and modifier concentration
for a wide range of pressure and substrate concentration.
Table 4 presents the best constants obtained for MV7, MV8,
and MV12. Not unexpectedly, allthree modelsgive a similar
fit and it is not possible to discriminate between them. The
adsorption constant KEA is in the same order of magnitude
as KE, and KsAiEH has a value similar to KEH in MV2. The ad-
sorption of the modifier is much stronger than for both etpy
TABLE 4
For the modified catalyst, a review of the different reac-
tion mechanisms proposed up to now was presented in (5).
Most proposals address only the mode of enantiodiscrim-
ination, i.e., the structure of the product-controlling reac-
tion intermediate, even though most investigators probably
agree that rate enhancement and enantiocontrol are closely
connected effectsand should be discussed together. There is
general agreement that very specific interactions between
the modifier and the substrate are necessary in order to
give the observed rate enhancement and such high enantio-
selectivities. However, there is less agreement concerning
the structure of the reaction intermediates and about the
Kinetic Constants and Least Square Sums for Models
MV7, MV8, and MV12
MV7
MV12
15,500
MV8
Least square:
16,900
Least square:
13,400
ksAiEH
ksBiEH
2
2
2
2
I
I
(
mol ): 441,000 345,700
(
mol ): 4,200
1000 min
g
1000 min
g
krAeEH (min 1):
KEA (M 1):
Km0 (M 1):
751
130
krBeEH (min 1):
KMB (M 1):
Km0 (M 1):
70
0.098
5,280
0.0832
0.134
0.282
122,000
1.51
200,700
0.0000981 KsBiEH
:
KsAiEH
:

292
BLASER ET AL.
role of the platinum surface in these transformations. Two
distinct types of mechanisms have been proposed thus far
and will be discussed below. The first type of mechanism
assumes the adsorption of the modifier on platinum to give
modified Pt sites, product formation being due to the reac-
tion between adsorbed species (5, 10, 35, 45). The second
type of mechanism proposed by Margitfalvi et al. (23, 26)
assumes the existence of a substrate-modifier complex in
the liquid phase as the controlling species for the modified
hydrogenation.
Proposals Involving a “Modified Catalyst”
From kinetic evidence, Blaser et al. (10) suggested that a
stereoselective site consists of one cinchona molecule ad-
sorbed on an ensemble of approximately 15 platinum sur-
face atoms. From the effect of catalyst dispersion it was
speculated that a flat arrangement of the Pt atoms should
be favorable. Adsorption of the modifier was thought to be
with the quinoline part, probably parallel to the surface (via
the system). The question of the structure of the reaction
intermediates was left open.
Sutherland et al. (12) considered at first that an array
of adsorbed cinchona molecules might be responsible for
enantiocontrol. They explained the observed rate enhance-
ment by three effects: (1) the activation of the substrate by
the quinuclidine nitrogen (higher rate when quinuclidine
is added); (2) a higher hydrogen coverage of the modified
catalyst (enhanced H2/D2 exchange on the modified cata-
lyst); and (3) the electronic effect of the adsorbed quinoline
part (higher rate when quinoline is added). This interest-
ing idea of an ordered adsorption had to be abandoned for
several reasons. The most important one was that the ob-
served dependence of e.e. and rate on the modifier concen-
tration was not in agreement with the formation of an array
of modifier molecules (45). Instead, the one-to-one model
described above was used with the suggestion that the ad-
sorbed cinchonidine would stabilize the half-hydrogenated
etpy via the quinuclidine nitrogen, thereby leading to the
observed rate enhancement.
From molecular modeling studies, Baiker et al. and
Schwalm et al. (5, 48, 49) deduced that it is the protonated
cinchona modifier that interacts with the oxygen of the car-
bonyl bond, thereby activating etpy for hydrogenation and
at the same time also exerting the enantioface discrimina-
tion of the ketone adsorption. This proposal is very similar
to the mode of action proposed for the hydrogenation of
methyl ketones catalyzed by tartrate modified nickel cata-
lysts. There, a hydrogen bond between one OH group of
the adsorbed tartaric acid and the oxygen of the carbonyl
group is proposed to control the adsorption mode of the
ketone (7).
FIG. 9. Artist’s view of the structure of adsorbed reaction interme-
diates: (a) Interaction of protonated HCd with si-form; (b) interaction of
HCd with half-hydrogenated ketone (si-form); (c) interaction of HCd with
etpy adsorbed at Pt adatom (Augustine et al. (51)); (d) solution complex
between Cd and etpy (Margitfalvi et al. (20)).
hydrogen atoms and etpy are adsorbed. In this proposal
the nitrogen atom of the quinuclidine interacts with the
keto group whereas the ester carbonyl of etpy interacts with
the lone pair of the hydroxy group, giving a six-membered
arrangement that is thought to be important for controlling
the stereochemistry. After the addition of the first hydrogen
atom, the intermediate c in Fig. 9 is assumed.
All these proposals are basically in agreement with the
kinetic results discussed above except for the array model.
The two suggestions by Wells and by Baiker concerning
the hydrogen-bridged species are almost identical. As the
source of the proton, Baiker has proposed acetic acid or
a protic solvent. In our model we propose that the pro-
ton comes from dissociatively adsorbed hydrogen (it is not
excluded that the transfer can occur via the solvent, espe-
cially in presence of acetic acid). In this scenario, the second
H-atom would be added as hydride and the Pt bulk would
serve as a reservoir for the electrons involved. At the mo-
ment this reaction sequence seems the most plausible to
us. It should be noted that we get about the same fit in ki-
netic modeling assuming a reaction between H- and an
adsorbed etpy on a modified site to give the stabilized half-
hydrogenated ketone as proposed by Augustine et al. (50)
except that in our model hydrogen and etpy are adsorbed
on different Pt atoms (MV7). Augustine’s proposal of an
adsorption on the same Pt adatom is not in agreement with
the rate decrease observed at high etpy which is interpreted
as competition between the adsorption of etpy and H2.
Proposals Involving a “Modified Substrate”
(Shielding Effect)
In contrast, Augustine et al. (34, 50) postulated that the
cinchona molecule isadsorbed either via the -system or via
Margitfalvi et al. (23, 26) have proposed that in solu-
the N of the quinoline close to a Pt ad-atom on which both tion etpy forms a complex with the aromatic part of the

KINETIC STUDIES OF ETHYL PYRUVATE HYDROGENATION
CONCLUSIONS
293
effective modifiers (d in Fig. 9) and that this activated com-
plex is hydrogenated. The authors have described two dif-
ferent scenarios of how this might happen. In the first (26),
the role of the metal is described as being “to provide ad-
sorbed hydrogen for the hydrogenation,” implying some
sort of Eley–Rideal mechanism. In the second (23), the
substrate-modifier complex was proposed to adsorb on the
metal surface where it is then hydrogenated.
The first proposal concerning an Eley–Rideal mechanism
can be excluded with considerable certainty due to the sat-
uration effects observed at extremely low HCd concentra-
tion as well as at increasing etpy concentration. Adsorption
of a preformed modifier-substrate complex, however, leads
to the same rate and e.e. dependencies if the RDS for the
formation of both the (R)- and the (S)-enantiomer is a re-
action between adsorbed species. Indeed, fitting our results
to the corresponding kinetic model gave a fit similar to that
of MV8. On the other hand, these assumptions reduce the
importance of the modifier-substrate complex in solution
to a mere pre-equilibrium because all important reactions
again would take place between adsorbed species.
It was shown that all important prerequisites can be met
for obtaining reliable data for the kinetics of the hydrogena-
tion of etpy to etlac on a 5% Pt/Al2O3 catalyst in toluene
both in absence and in presence ofthe chiralmodifier, 10,11-
dihydrocinchonidine (HCd). The kinetic investigation of
the hydrogenation of etpy on Pt catalysts is complicated
by several problems that demand a careful evaluation of
the experimental conditions. Nevertheless, the experimen-
tal data presented here allowed the identification of a few
possible schemes as well as the exclusion of many other
reaction schemes.
From the observed dependencies of the reaction rates
on the catalyst, etpy, hydrogen, and HCd concentrations
and from the effect of these parameters on the optical yield
in presence of HCd, it was possible to propose coherent
reaction schemes both for the reaction with and without
modifier. The data allowed the exclusion of the following
steps as rate determining: dissociative adsorption of hydro-
gen, adsorption of etpy, and desorption of etlac. Also ex-
cluded were all forms of Eley–Rideal mechanisms where
adsorbed hydrogen reacts with etpy directly from solution.
The adsorption of hydrogen and etpy on the catalyst was
shown to be competitive. However, it was not possible to
ascertain that all Pt surface atoms can adsorb a modifier
molecule. In absence of the modifier, it was difficult to dis-
tinguish between the first and the second H-addition being
rate limiting. Mainly on the basis of the adsorption constant
for hydrogen, the addition of the first hydrogen was con-
sidered as the more likely RDS. The results in presence of
HCd were more complex to interpret. The rate for major
enantiomer is clearly controlled by the second H-addition,
as indicated by a first order dependence on H2 concentra-
tion. It was possible to fit the data, assuming that the RDS
of the minor etlac enantiomer remains the first hydrogen
addition. A proposal was made concerning the origin of
the rate acceleration as well as the enantiodiscrimination
on the basis of the models proposed by Baiker and Wells.
Mechanism: Conclusions
With the information summarized above, and at thisstage
of understanding, the most convincing explanation for both
the enantiodiscrimination and the rate acceleration is the
preferentialstabilization ofone ofthe two possible diastere-
omeric intermediates (or transition states) toward the half-
hydrogenated etpy. As to the precise structure of these in-
termediates, no direct information is currently available. At
this time, we favor the following reaction sequence for the
major pathway to (R)-etlac: Adsorption of etpy on a non-
modified site; migration to a modified site due to the attrac-
tive interactions between the ketone oxygen and the proto-
nated cinchona modifier (Baiker’s complex, a in Fig. 9); fast
transfer of the proton to the oxygen (Well’s complex, b in
Fig. 9) followed by the slow addition of the second hydro-
gen and desorption. For the unmodified catalyst as well as
for the formation of (S)-etlac we propose that the addition
of the first hydrogen is rate limiting. The difference in RDS
between the two enantiomeric pathways would be due to
a more effective stabilization of the si-complex. Compared
to the reaction without modifier or the reaction with the re-
complex, the first addition would be accelerated, thereby
leading to the second addition being rate limiting. We re-
alize of course that this picture is probably too simplistic.
With differences of 2–3 kcal/mol for the activation energies
of the rate determining steps we are on the limits of the
LHHW formalism which does not consider gradual changes
in the RDS. A second concern is that such a good enantios-
electivity should have it’s cause in a rather loose modifier–
catalyst–substrate complex and, further, that this particular
hydrogen bridge should have such an extraordinary char-
acter.
ACKNOWLEDGMENTS
We thank J. Mastio, D. Imhof, and J. Salathin for carrying out some
of the experimental work; D. G. Blackmond for helpful discussion; and
R. Bader for the continuous support of this work.
REFERENCES
1. (a) Koenig, E. K., in “Asymmetric Synthesis” (J. D. Morrison, Ed.),
Vol. 5, Chapter 3. Academic Press, New York, 1985; (b) Takaya,
H., Ohta, T., and Noyori, R., in “Catalytic Asymmetric Synthesis”
(I. Ojima, Ed.), p. 1. VCH, Weinheim, 1993.
2. See for example Halpern, J., in “Asymmetric Synthesis” (J. D.
Morrison, Ed.), Vol. 5, Chapter 2. Academic Press, New York, 1985.
3. Blaser, H. U., and Mu¨ller, M., Stud. Surf. Sci. Catal. 59, 73 (1991).
4. Izumi, Y., Adv. Catal. 32, 215 (1983).

294
BLASER ET AL.
5. Baiker, A., Mallat, T., Minder, B., Schwalm, O., Simons, K. E., and
29. Wang, J., Sun, Y., LeBlond, C., Landau, R. N., and Blackmond, D. G.,
J. Catal. 161, 752 (1996).
Weber, J., in “Chiral Reactions in Heterogeneous Catalysis” (G.
Jannes and V. Dubois, Eds.), p. 95. Plenum Press, New York, 1995.
6. Sachtler, W. M. H., Chem. Ind. 189, 22 (1985).
7. Tai, A., and Harada, T., in “Tailored Metal Catalysts” (Y. Iwasawa,
Ed.), p. 265. Reidel, Dordrecht, 1986.
30. Singh, U. K., Landau, R. N., Sun, Y., LeBlond, C., Blackmond, D. G.,
Tanielyan, S. K., and Augustine, R. L., J. Catal. 154, 91 (1995).
31. Margitfalvi, J. L., Minder, B., Talas, E., Botz, L., and Baiker, A., in “Pro-
ceedings, 10th International Congress on Catalysis, Budapest, 1992”
(L. Guczi, F. Solymosi, and P. Tetenyl, Eds.), p. 2471. Akade´miaiKiado´,
Budapest, 1993.
8. Klabunovskii, E. I., Izv. Akad. Nauk. SSSR, Ser. Khim., 505 (1984).
9. Webb, G., and Wells, P. B., Catal. Today 12, 319 (1992).
10. Blaser, H. U., Garland, M., and Jalett, H. P., J. Catal. 144, 569 32. “Landolt-Bo¨rnstein, Numerical Data and Functional Relationships
(1993).
in Science and Technology,” New Series, Group 4, Vol. 6, p. 80, #517.
Springer-Verlag, Berlin/New York.
11. Garland, M., Jalett, H. P., and Blaser, H. U., Stud. Surf. Sci. Catal. 59,
177 (1991).
33. Wolf, R., Ciba-Geigy AG, personal communication.
12. Sutherland, I. M., Ibbotson, A., Moyes, R. B., and Wells, P. B., 34. Wehrli, J. T., Baiker, A., Monti, D. M., Blaser, H. U., and Jalett, H. P.,
J. Catal. 125, 77 (1990). J. Mol. Catal. 57, 245 (1989).
13. Meheux, P. A., Ibbotson, A., and Wells, P. B., J. Catal. 128, 387 35. Augustine, R. L., Tanielyan, S. K., and Doyle, L. K., Tetrahedron:
(1991). Asymmetry 4, 1803 (1993).
14. Bond, G., Meheux, P. A., Ibbotson, A., and Wells, P. B., Catal. Today 36. Minder, B., Mallat, T., Baiker, A., Wang, G., Heinz, T., and Pfaltz, A.,
10, 371 (1991).
J. Catal. 154, 371 (1995).
15. Wang, J., LeBlond, C., Orella, C. J., Sun, Y., Blackmond, D. G., and
Bradley, J. S., Stud. Surf. Sci. Catal. 108, 183 (1997).
37. Reschetilowski, W., Bo¨hmer, U., and Wiehl, J., Stud. Surf. Sci. Catal.
84C, 2021 (1994).
16. Orito, Y., Imai, S., Niwa, S., and Nguyen, G. H., J. Synth. Org. Chem. 38. Sun, Y., Landau, R. N., Wang, J., LeBlond, C., and Blackmond, D. G.,
Jpn. 37, 173 (1979); Orito, Y., Imai, S., and Niwa, S., J. Chem. Soc. Jpn.,
1118 (1979), 670 (1980), and 137 (1982).
17. Blaser, H. U., Jalett, H. P., Monti, D. M., Reber, J. F., and Wehrli, J. T.,
Stud. Surf. Sci. Catal. 41, 153 (1988).
18. Blaser, H. U., Jalett, H. P., Monti, D. M., Baiker, A., and Wehrli, J. T.,
Stud. Surf. Sci. Catal. 67, 147 (1991).
19. Blaser, H. U., Jalett, H. P. Mu¨ller, M., and Studer, M., Catal. Today 37,
441 (1997).
J. Amer. Chem. Soc. 118, 1348 (1996).
39. Bo¨hmer, U., Morgenschweis, K., and Reschetilowski, W., Catal. Today
24, 195 (1995).
40. Blaser, H. U., Imhof, D., and Studer, M., Stud. Surf. Sci. Catal. 108, 175
(1997).
41. see for example Geneste, P., Bonnet, M., and Rodriguez, M., J. Catal.
57, 147 (1979); Godolets, G. I., Kinetika i Kataliz 28, 35 (1987).
42. Simonikova, J., Hillaire, L., Panek, J., and Kochloefl, K., Z. Phys. Chem.
83, 287 (1973).
20. Redlich, O., and Kwong, J. N. S., Chem. Rev. 44, 233 (1949); Soave, G.,
Chem. Eng. Sci. 27, 1197 (1972); Peng, D. Y., and Robinson, D. B., Ind. 43. Hougen, O. A., and Watson, K. M., in “Chemical Process Principles,”
Eng. Chem. Fundm. 15, 59 (1976).
Kinetics and Catalysis, Vol. 3. Wiley, New York, 1947.
44. Landis, C. R., and Halpern, J., J. Amer. Chem. Soc. 109, 1746 (1987).
45. 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).
46. Cremer, P. S., and Somorjai, G. A., J. Chem. Soc. Faraday Trans. 91,
3671 (1995).
21. Koelliker, R., and Thies, H., J. Chem. Eng. Data 38, 437 (1993).
22. Mastio, G., and Thies, H., Novartis Services AG, unpublished
results.
23. Margitfalvi, J. L., Hegedu¨s, M., and Tfirst, E., Tetrahedron: Asymmetry
7, 571 (1996).
24. Margitfalvi, J. L., in “Catalysis of Organic Reactions” (M. G. Scaros
and M. L. Prunier, Eds.), Chem. Ind. 62, p. 189. Dekker, New York, 47. Tanaka, K., in “Catalytic Hydrogenation” (L. Cerveny, Ed.), Chapter
1995.
3. Elsevier, Amsterdam, 1986.
25. Margitfalvi, J. L., and Hegedu¨s, M., J. Mol. Catal. 107, 281 (1996).
26. Margitfalvi, J. L., J. Catal. 156, 175 (1995).
48. Schwalm, O., Minder, B., Weber, J., and Baiker, A., Catal. Lett. 23, 271
(1994).
27. Blaser, H. U., Jalett, H. P., and Spindler, F., J. Mol. Catal. A: Chemical 49. Schwalm, O., Weber, J., Minder, B., and Baiker, A., J. Mol. Struct.
107, 85 (1996). (Theochem) 330, 353 (1995).
28. Yasumori, I., Yokozeki, M., and Inoue, Y., Faraday Discuss. 72, 385 50. Augustine, R. L., and Tanielyan, S. K., J. Mol. Catal. A: Chemical 112,
(1981). 93 (1996).