New insights into the relationship between conversion and enantioselectivity for the asymmetric hydrogenation of alkyl pyruvate

Article abstract of DOI:10.1016/j.jcat.2003.10.001

The initial transient period in the enantioselective hydrogenation of alkyl pyruvate esters is probed using the sequential reactions of ethyl and methyl pyruvate. The reaction of methyl pyruvate, subsequent to the hydrogenation of ethyl pyruvate, led to a higher e.e. when compared to the coreaction of these reactants, or prehydrogenation with methyl pyruvate followed by reaction of ethyl pyruvate. The initial transient effect, in which e.e. increases with conversion, is observed in both periods of the sequential reaction and the origin of this effect is discussed.

Full text of DOI:10.1016/j.jcat.2003.10.001

Journal of Catalysis 221 (2004) 653–656
www.elsevier.com/locate/jcat
Priority Communication
New insights into the relationship between conversion and
enantioselectivity for the asymmetric hydrogenation of alkyl pyruvate
∗
Xiabao Li, Richard P.K. Wells, Peter B. Wells, and Graham J. Hutchings
Department of Chemistry, Cardiff University, PO Box 912, Cardiff CF10 3TB, UK
Received 26 September 2003; accepted 3 October 2003
Abstract
The initial transient period in the enantioselective hydrogenation of alkyl pyruvate esters is probed using the sequential reactions of ethyl
and methyl pyruvate. The reaction of methyl pyruvate, subsequent to the hydrogenation of ethyl pyruvate, led to a higher e.e. when compared
to the coreaction of these reactants, or prehydrogenation with methyl pyruvate followed by reaction of ethyl pyruvate. The initial transient
effect, in which e.e. increases with conversion, is observed in both periods of the sequential reaction and the origin of this effect is discussed.

2003 Elsevier Inc. All rights reserved.
Keywords: Enantioselective hydrogenation of alkyl pyruvates; Cinchonidine-modified Pt/Al O catalysts
2
3
1
. Introduction
At present, two schools of thought exist concerning the ori-
gin of this effect. Baiker and co-workers [11,13] consider
that the enhancement in e.e. is caused by removal of im-
purities present on the catalyst surface at the start of the
reaction, for example, oxygen or the degradation products
of dissociative chemisorption of the substrate. During the
initial reaction in the presence of hydrogen, these impurities
are removed. In contrast, Blackmond and co-workers [9,12]
consider the effect to be due to a “reaction-driven equilibra-
tion” of the catalyst surface and have shown that the effect of
rising rate and enantioselection in the initial reaction period
was linked solely to substrate conversion and was indepen-
dent of reaction conditions. To aid our understanding of this
intriguing process, we have designed a sequential experi-
ment that adds new insight into this fascinating problem.
The study of enantioselective catalysts continues to re-
ceive significant attention due to the importance of pure
enantiomers in the manufacture of pharmaceuticals and
agrochemicals. Much progress has been achieved with ho-
mogeneous catalysts [1–3], but significant attention remains
focused on heterogeneous asymmetric catalysts where a
number of catalysts have been identified that give very high
enantioselection, e.g., the hydrogenation of ethyl pyruvate
using cinchona-modified supported Pt catalysts [4,5] and
the aziridination of alkenes using (bis)oxazoline-modified
Cu -exchanged zeolite Y [6] and the subject of immo-
bilised homogeneous catalysts has recently been reviewed
by Sing and Lee [7]. One intriguing observation with some
heterogeneous asymmetric catalysts is that the enantiose-
lection is enhanced during the initial period of conversion;
whereas, typically, the reverse is observed in many systems.
The effect of this initial transient period has been well stud-
ied for the cinchona-modified Pt catalysts [8–11]. However,
the origin of the enhanced enantioselection during the initial
reaction remains a matter of considerable debate [12,13].
Herein, we deal with the origin of the increase in
enantioselection with conversion for cinchona-modified
Pt/Al2O3 catalysts for the hydrogenation of alkyl pyruvates.
2
+
2
. Experimental
A 5% Pt/Al2O3 (Johnson Matthey, type SR94B) was pre-
◦
treated with hydrogen at 400 C prior to use. Cinchonidine
(
>
Fluka, 98%) was used as received. Ethyl pyruvate (Fluka,
97%) and methyl pyruvate (Fluka, > 97%) were purified
with great care prior to use. The pyruvate (50 ml) was mixed
with CH2Cl2 (50 ml) and treated with KHCO3 (0.1 mol/l,
5
0 ml). The organic layer was collected and distilled un-
*
der reduced pressure over MgSO to remove CH Cl . The
Corresponding author.
4
2
2
E-mail address: hutch@cardiff.ac.uk (G.J. Hutchings).
residue was distilled over anhydrous CaCl2 under vacuum
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2003.10.001

654
X. Li et al. / Journal of Catalysis 221 (2004) 653–656
and the purified pyruvate was used immediately. Each pyru-
vate ester contained no detectable impurities; in particular,
they contained no lactate, for example, racemic lactates.
CH2Cl2 was purified by distillation over anhydrous CaCl2
and stirred over a 4A molecular sieve. Toluene was purified
by distillation over a 4A molecular sieve and stored over a
4
A molecular sieve.
The enantioselective hydrogenation reactions were car-
ried out using a Parr 50-ml stirred autoclave reactor. Two
experimental procedures were used and all experiments were
◦
conducted at 20 C in duplicate. Procedure I: pyruvate sub-
strate (66 mmol), CH2Cl2 solvent (12.5 ml), cinchonidine
(
2.5 mg), and 5% Pt/Al2O3 catalyst (0.25 g) were sealed into
the autoclave. The autoclave was purged three times with
hydrogen or argon to a pressure of 30 bar to remove resid-
ual air. Hydrogen was then admitted to the required reaction
pressure (30 or 50 bar). At the same time, stirring was com-
menced (1200 rpm) to start the reaction and the reaction was
stopped by turning off the stirrer, removing the hydrogen at-
mosphere, and filtering off the catalyst, and the product was
isolated and analysed by GC and chiral GC. Procedure II:
this was adopted for the sequential experiments. The initial
reaction of ethyl pyruvate was carried out as described in
Procedure I, except that a decreased amount of ethyl pyru-
vate (33 mmol) was used and a small pressure vessel (10-ml
volume) was connected to the autoclave. The small pres-
sure vessel was initially filled with hydrogen (50 bar) using
three purge cycles. Methyl pyruvate (33 mmol) in CH2Cl2
Fig. 1. Dependence of enantiomeric excess and conversion of ethyl pyruvate
hydrogenation. Key: •, ethyl pyruvate conversion; ꢀ, (R)-ethyl lactate; ꢁ,
ꢀe.e. Conditions: 5% Pt/Al2O3 (0.25 g), CH2Cl2 (12.5 ml), cinchonidine
◦
(
2.5 mg), ethyl pyruvate (66 mmol), PH = 50 bar, 20 C, 1200 rpm stirring
2
speed.
ments, we added the solvent, catalyst, cinchona alkaloid, and
ethyl pyruvate to the reactor in that order and purged the sys-
tem with hydrogen several times to remove residual oxygen.
Following this, the hydrogen was introduced and the stirrer
started. Experiments showed that, under our conditions, the
e.e. was independent of stirring speed above 800 rpm and
it took 10 s to achieve the optimal stirring rate of 1200 rpm.
During this time, only 1% substrate conversion was observed
(
Fig. 1) and, since the significant enhancement in e.e. is ob-
(
5 ml) was added to the hydrogen-filled small pressure ves-
served up to ca. 20% conversion, we do not consider that
the e.e. enhancement is due to initial diffusion effects. In
addition, experiments using different amounts of cinchona
modifier and substrate concentrations did not affect the na-
ture of the initial transient period. However, in carrying out
these experiments, it is clear that the initial transient period
could be influenced by factors that are difficult to control,
since the reactor stirring and addition of hydrogen are req-
uisites to start the reaction, but they take a finite time to
reach optimal levels. To overcome these problems, we de-
vised the following set of experiments using the sequential
hydrogenation of ethyl pyruvate and of methyl pyruvate.
First, the enantioselective hydrogenation of the two alkyl
pyruvates was investigated separately; both gave the ex-
pected initial transient effect and the final e.e.s observed
at 100% substrate conversion are given in Table 1. Signif-
icantly, when reacted separately, the pyruvate esters give
almost identical e.e.s for the lactates, and the increase in e.e.
with conversion was also indistinguishable. Second, the hy-
drogenation of the ethyl pyruvate and methyl pyruvate was
investigated simultaneously using half the normal concen-
trations of each substrate, both being added at the beginning
sel carefully via a syringe and this was purged by three
successive additions of hydrogen (50 bar) using switch-
ing valves to add the hydrogen and to depressure the ves-
sel subsequently. The small pressure vessel containing the
methyl pyruvate/CH2Cl2 was then filled with H2 (50 bar).
Following reaction of ethyl pyruvate at 30 bar, the methyl
pyruvate/CH2Cl2 solution was added to the autoclave reac-
tor under the increased H2 pressure without admission of air.
3
. Results and discussion
The observation of the initial transient period is shown
in Fig. 1 for the hydrogenation of ethyl pyruvate in an auto-
clave reactor using cinchona-modified Pt/γ -Al2O3. For each
data point in Fig. 1, a separate experiment was carried out in
duplicate. At the designated time, the reaction was stopped
by release of the hydrogen reactant pressure and the prod-
uct isolated. The observed enhancement in enantioselection
is apparent in both the overall enantiomeric excess (e.e.) de-
termined at the specified reaction times, and the differential
e.e. (ꢀe.e.) determined from the increase in e.e. over that
of the previous data point, which shows a more significant
1
increase in e.e. at the low reaction times. In our experi-
have the lowest e.e., there is a cumulative effect for the e.e. at any time,
which is the sum of all the earlier less enantioselective reactions. ꢀe.e. gives
the e.e. for the conversion between the reaction time, at which it is recorded,
and the preceding reaction time. This provides a better representation of the
enantioselection achieved at the catalyst surface.
1
ꢀ
e.e. is the mean value of the e.e. provided by the reaction between
data points. As the e.e. is increasing with conversion and the initial turnovers

X. Li et al. / Journal of Catalysis 221 (2004) 653–656
655
Table 1
Comparison of the enantiomeric excess for the hydrogenation of alkyl pyru-
vates using Procedures I and II
b
Experimental
Substrate
e.e. (%)
a
procedure
R-Ethyl lactate R-Methyl lactate
c
I
I
I
Ethyl pyruvate
80.5 ± 0.5
–
c
Methyl pyruvate
–
79.0 ± 0.5
Ethyl and methyl
pyruvate simultaneous
Sequential: ethyl
pyruvate followed
by methyl pyruvate
74.0 ± 0.5
75.0 ± 0.5
d
II
81.0 ± 0.5
91.0 ± 0.5
d
a
See experimental, CH Cl (12.5 ml), cinchonidine (2.5 mg), 5%
2
2
◦
Pt/Al O catalyst (0.25 g), 20 C, 1200 rpm stirring speed.
2
3
(a)
b
c
d
At end of the reaction.
Pyruvate (66 mmol).
Ethyl pyruvate (33 mmol) and methyl pyruvate (33 mmol).
of the reaction. Interestingly, the final e.e.s for this simul-
taneous experiment are significantly lower for both alkyl
lactate products when compared to the e.e. achieved for the
single substrate (Table 1) and the e.e.s for the two lactate
products were identical with respect to their increase with
conversion. Third, a sequential experiment was carried out
in which the reaction of methyl pyruvate was carried out
following the initial hydrogenation of ethyl pyruvate using
chinona-modified Pt/Al2O3. In these experiments, the initial
hydrogenation of ethyl pyruvate was carried out with 30 bar
hydrogen pressure. Following this reaction, methyl pyruvate
was rapidly added under hydrogen pressure while the stir-
ring was continued in such a way that no oxygen could be
admitted to the reaction. The second hydrogenationwas con-
sequently conducted at a higher reaction pressure (50 bar),
but, as noted previously, the observation of the initial tran-
sient effect is independent of the reaction conditions [9,12].
The results of the sequential experiment are shown in Fig. 2a
and are most revealing. Most importantly, two effects are
apparent; first, the initial transient period is observed for
the hydrogenation of methyl pyruvate and, second, the final
e.e. observed in R-methyl lactate formation (91%) is signif-
icantly higher than that observed in the single (79% e.e.) or
simultaneous (75% e.e.) experiments. It should be noted that
each data point in Fig. 2 is for a separate experiment carried
out in duplicate and, therefore, is a composite of 22 separate
experiments for Fig. 2a and 16 experiments for Fig. 2b. In
addition, the e.e. of the R-ethyl lactate remained unchanged
during the subsequent formation of R-methyl lactate. The
observation of enhanced e.e. for the sequential hydrogena-
tion of methyl pyruvate is totally unexpected and represents
one of the most highly enantioselective hydrogenations of
this substrate recorded.
(b)
(c)
Fig. 2. Sequential hydrogenation experiment using reaction Procedure II.
a) Ethyl pyruvate reacted to 100% conversion prior to addition of methyl
(
pyruvate; (b) ethyl pyruvate reacted to 50% conversion prior to addition of
methyl pyruvate. Key: ꢁ, ethyl pyruvate conversion; •, e.e. in (R)-ethyl
lactate, ꢂ, methyl pyruvate conversion; ꢀ, e.e. in (R)-methyl lactate. Con-
ditions: 5% Pt/Al O (0.25 g), CH Cl (12.5 ml), cinchonidine (2.5 mg),
2
3
2
2
◦
ethyl pyruvate (33 mmol), methyl pyruvate (33 mmol), 20 C, 1200 rpm
stirring speed, reaction pressure 30 bar H for ethyl pyruvate hydrogena-
2
To ensure that the effect was not due to the complete con-
version of ethyl pyruvate occurring prior to the addition of
the methyl pyruvate, a further experiment was conducted
where the ethyl pyruvate reaction was taken to 50% con-
version prior to the addition of methyl pyruvate (Fig. 2b).
Again, the same initial transient for methyl pyruvate was ob-
tion, 50 bar H for methyl pyruvate introduction; (c) methyl pyruvate re-
2
acted to 100% conversion prior to addition of ethyl pyruvate.
served, even though the catalyst was now converting both
ethyl and methyl pyruvate. Significantly, the e.e. observed
for methyl pyruvate at the end of the reaction (91%) was still

656
X. Li et al. / Journal of Catalysis 221 (2004) 653–656
significantly higher than that observed if the methyl pyru-
vate had been added at the start of the reaction (Table 1,
erogeneous asymmetric catalysts. The effect may be caused
by selective poisoning of particular surface sites that lead
to racemic reaction, and this effect warrants further study.
However, our data also show that pretreatment can also give
deleterious effects since, when methyl pyruvate is initially
reacted prior to ethyl pyruvate, the resulting (R)-ethyl lac-
tate has a significantly lower e.e. than when ethyl pyruvate
is reacted with a nonpretreated catalyst. We consider that the
steric bulk of the alkyl substitute could have an important
role in defining the interaction of the reactant with the cin-
chona modifier on the surface of the platinum nanocrystals.
It is possible that the modifier conformation on the surface,
rather than in solution, is affected by the structure of the
reactant and this is an effect that warrants further detailed
study.
7
5%).
The experiment was repeated with the order of reagent
addition reversed, i.e., methyl pyruvate conversion first, fol-
lowed by ethyl pyruvate. Again, in this experiment, the ini-
tial hydrogenation of methyl pyruvate was carried out with
3
0 bar pressure and the initial transient effect was observed
(
Fig. 2c). Following complete conversion of the methyl
pyruvate, ethyl pyruvate was added as described previously.
An increase in e.e. is observed with increasing conversion
but the e.e. is not enhanced. This is an interesting result
since it shows that the order of reactant addition is of cru-
cial importance and, although ethyl pyruvate reaction prior
to methyl pyruvate leads to a significant enhancement in
e.e. of the (R)-methyl lactate, the same is not true for the
reverse experiment. Further experiments were carried out us-
ing toluene as solvent and similar results were obtained.
As noted previously, Baiker and co-workers [11,13] at-
tribute the observation of the initial transient effect to im-
purities such as oxygen, or other residues, formed on the
surface due to the pyruvate adsorption at the start of the
reaction. The reaction has the effect of cleaning the sur-
face of these impurities and it, therefore, takes a significant
number of reaction turnovers to achieve the optimal chirally
modified surface operating under steady-state conditions. In
our experiments, the sequential hydrogenation of ethyl and
methyl pyruvates was carried out in such a way that did not
expose the catalyst to air or interrupt the reaction progress
In conclusion, we have described a sequential hydrogena-
tion experiment for the reaction of alkyl pyruvates using
cinchona-modified Pt/Al O that shows the enhancement in
2
3
e.e. observed with increasing conversion is related to the in-
teraction of the modifier and the substrate. Furthermore, this
effect can be used to provide markedly enhanced enantiose-
lection in specific cases and this could be of importance in
the design of enantioselective heterogeneous catalysts.
Acknowledgments
This work was supported by the EPSRC. Xiabao Li
thanks the Committee of the Vice-Chancellors and Princi-
pals of the Universities of the United Kingdom for an ORS
award.
(
Fig. 2). The chiral-active sites at the cinchonidine/Pt surface
had been established and were functioning at high enan-
tioselection. The observation that the initial transient effect
is still observed with the sequential hydrogenation of ethyl
and methyl pyruvates indicates that the increase in e.e. with
conversion for this reaction is not attributable to impurity ef-
fects, since the optimal chiral-active sites had been achieved
from the reaction of ethyl pyruvate. The results support the
ideas proposed by Blackmond and co-workers [9,12] con-
cerning the “reaction-drivenequilibrium of the chiral surface
environment.” However, our experiments suggest (Table 1,
Fig. 2) that the cinchona modifier adsorbed on the metal sur-
face may interact differently with the two reactants and the
active chiral site is dependent upon the nature of the sub-
strate. The different nature of this interaction for the two
pyruvate esters is also indicated by the simultaneous reaction
data which show identical behaviour, but the e.e. is signifi-
cantly different from that obtained for the reaction of each
alone. However, our observation of markedly enhanced e.e.
for the sequential hydrogenation of methyl pyruvate follow-
ing ethyl pyruvate indicates that there is potential to premod-
ify chiral surfaces to give enhanced enantioselection. This
observation could, in effect, be the most significant, since it
suggests that surface pretreatment/conditioning could be a
fruitful area of investigation in the design of improved het-
References
[
[
1] Angew. Chem., Int. Ed. Engl. 41 (2002) 998.
2] D.A. Evans, K.A. Woerpel, M.H. Hinman, M.M. Paul, J. Am. Chem.
Soc. 113 (1991) 726.
[3] R. Noyori, Asymmetric Catalysis in Organic Synthesis, Wiley, New
York, 1994.
[
[
4] H.U. Blaser, Chem. Rev. 92 (1992) 935; Chem. Commun. (2003) 293.
5] B. Minder, T. Mallat, A. Baiker, G. Wang, T. Heinz, A. Pfaltz, J. Ca-
tal. 154 (1995) 371.
[6] S. Taylor, J. Gullick, P. McMorn, D. Bethell, P.C. Bulman Page, F.E.
Hancock, F. King, G.J. Hutchings, J. Chem. Soc., Perkin Trans. 2
(2001) 1714.
[
[
7] C.E. Sing, S. Lee, Chem. Rev. 102 (2002) 3495.
8] U.K. Singh, R.N. Landau, Y. Sun, C. LeBlond, D.G. Blackmond, S.K.
Tanselyan, R.L. Augustine, J. Catal. 154 (1995) 91.
9] J. Wang, Y. Sun, C. LeBlond, R.N. Landau, D.G. Blackmond, J. Ca-
tal. 161 (1996) 752.
[
[
[
[
10] Y. Sun, G. Wang, C. LeBlond, R.N. Landau, D.G. Blackmond, J. Ca-
tal. 161 (1996) 759.
11] T. Mallat, Z. Bodnar, B. Minder, K. Borszeky, A. Baiker, J. Catal. 168
(1997) 183.
12] D.G. Blackmond, J. Catal. 176 (1998) 267.
[13] T. Mallat, A. Baiker, J. Catal. 176 (1998) 271.