Hydrogenation of α-Keto Ethers: Dynamic Kinetic Resolution with a Heterogeneous Modified Catalyst and a Heterogeneous Base

Article abstract of DOI:10.1002/1615-4169(200207)344:5<511::AID-ADSC511>3.0.CO;2-D

The first successful example of the asymmetric hydrogenation of substituted α-keto ethers with Cinchona-modified Pt/Al₂O₃ is reported. In the absence of an additional base, kinetic resolution of the racemic starting material was observed with high diastereoselectivity and ee's up to 98% at conversions of <50%. Addition of KOH gave a strong reaction acceleration but racemic product. Immobilization of OH^- on solid ion exchangers resulted in the desired dynamic kinetic resolution, and ee's of >80% were obtained at >95% conversion. These effects are rationalized on the basis of a simple kinetic and structural model.

Full text of DOI:10.1002/1615-4169(200207)344:5<511::AID-ADSC511>3.0.CO;2-D

COMMUNICATIONS
Hydrogenation of a-Keto Ethers: Dynamic Kinetic Resolution
with a Heterogeneous Modified Catalyst and a Heterogeneous
Base
Martin Studer,* Hans-Ulrich Blaser, Stephan Burkhardt
Solvias AG, R-1055.6, P.O. Box, 4002 Basel, Switzerland
Fax: ( ‡ 41) 61-686-6311, e-mail: Martin.studer@solvias.com
Received: April 7, 2002; Accepted: April 27, 2002
the base are not easily separated from the product after
the reaction. An elegant solution for this problem would
be the use of a heterogeneous catalytic system.
Abstract: The first successful example of the asym-
metric hydrogenation of substituted a-keto ethers
with Cinchona-modified Pt/Al2O3 is reported. In the
absence of an additional base, kinetic resolution of
the racemic starting material was observed with high
diastereoselectivity and ee×s up to 98% at conver-
sions of <50%. Addition of KOH gave a strong
reaction acceleration but racemic product. Immobi-
lization of OH on solid ion exchangers resulted in
the desired dynamic kinetic resolution, and ee×s of
>80% were obtained at >95% conversion. These
effects are rationalized on the basis of a simple
kinetic and structural model.
Cinchona-modified heterogeneous Pt catalysts are
technically useful for the hydrogenation of a number of
ketones. Despite considerable search efforts for suitable
substrates, up to now only strongly activated ketones
such as a-keto acid derivatives[3] and cyclic analogues
thereof,[3b] certain trifluoromethyl ketones,[3c] a-keto
acetals,[4] and a-diketones[5] can be hydrogenated with
satisfactory selectivities. Intrigued by the results of the
homogeneous catalysts, we investigated Pt-Cinchona
catalysts for the enantioselective hydrogenation of a-
keto ethers (Scheme 1). Here, we report the successful
combination ofa chiral heterogeneous Pt catalyst and an
insoluble, strongly basic ion exchanger for the enantio-
and diastereoselective hydrogenation of racemic a-
alkoxy ketones with good yields and moderate to good
stereoselectivities.
Keywords: asymmetric hydrogenation; dynamic ki-
netic resolution; heterogeneous modified catalyst; a-
keto ethers; ligand accelerated catalysis
For a preliminary investigation, we chose a proven Pt/
Al2O3 catalyst modified with HCd derivatives under
Dynamic kinetic resolution is a powerful tool to obtain reaction conditions optimized for a-keto esters. Table 1
products with high enantioselectivity from racemic shows the results using racemic substrates 1a 1e at
starting materials.[1] If the asymmetric reaction creates conversions between 7 and 53%. As expected, kinetic
a second stereogenic center, the diastereoselectivity of resolution took place. AcOH or toluene gave the highest
the product can also be controlled. A synthetically ee×s, and the diastereoselectivities were always very
interesting reaction of this type is the hydrogenation of high, with erythro-(R,S)-2[6] as the major product. With
a-keto ethers giving very high de and ee using homoge- the exception of 1c, ee×s of 90 98% were obtained.
neous Ru catalysts in the presence of KOH.[2] Despite Compared to the unmodified systems, ligand-acceler-
the good selectivity, this methodology suffers from a ated catalysis was observed in all substrates carrying an
drawback because both the homogeneous catalyst and aliphatic R group (1a, 1b, and 1c, for details see below).
Scheme 1. Structures and naming of the starting materials, products and Cinchona modifiers.
Adv. Synth. Catal. 2002, 344, No. 5
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002 1615-4150/02/34405-511 515 $ 17.50+.50/0
511

Hydrogenation of a-Keto Ethers: Dynamic Kinetic Resolution
COMMUNICATIONS
Table 1. Hydrogenation of various a-keto ethers.[a]
Substrate
Solvent
Modifier
Time [min]
Conversion [%]
2 [%]
ee 2 [%]
3 [%]
Initial Rate
(mmol H2/min  g)
1a
1b
1b
1b
1b
1c
1d
1d
1e
AcOH
toluene
toluene
i-PrOH
AcOH
toluene
toluene
toluene
toluene
4b
12
145
24
638
1654
54
23
6
13
22
33
44
38
52
7
20
25
42
79
88
98
92
<1
7
< 1
< 1
2.1
0.5
2.4
n.a.
4a
4a
4b
4a
2
13.7
5
29
75
2
5
91< 1
< 1
0.2
1.5
0.7
1.0
34
42
10
4a
4a
3
42
90
10
[a]
Reaction conditions: 1 2 g 1, 20 25 mL solvent, 50 100 mg 5% Pt/Al2O3, 5 10 mg modifier, 25 8C, 60 bar.
With all substrates, the addition of the chiral modifier of ca. 5up toa conversion of 50% and then asignificantly
led to significantly higher diastereoselectivity. These slower reaction a clear sign for a kinetic resolution. The
finding are in contrast to results obtained with methoxy- cis/trans ratio 2b/3b increased from 4 to >30 in the
acetone where only 12% ee and a very low activity was presence of the modifier (3b not shown in Figure 1b).
observed.[7]
Similar effects were observed for all aliphatic ketones
The effect of the modifier on the course of the reaction whereas a slight deceleration of the overall reaction was
was investigated in detail for the cyclic 1b. As a noticed for the aromatic ketones in the presence of the
comparison of Figures 1a and 1b shows, the modifier modifier (see Table 1). The course of the reactions (lines
causes an acceleration of the overall reaction by a factor in Figures 1 3) was modeled by applying a simple
Figure 2. Hydrogenation of 1b in the presence of ion ex-
changer. a) upper panel, no modifier, 1.0 g 1b, 100 mg 5% Pt/
Al2O3, 25 mL toluene, 1.6g Amberlite IRA-900 OH acti-
vated, 60 bar, 25 8C. b) lower panel, with 10 mg HCd.
Reaction conditions as in a).
Figure 1. Hydrogenation of 1b in toluene. Reaction condi-
tions see Table 1. a) upper panel, no modifier, b) lower panel,
HCd.
512
Adv. Synth. Catal. 2002, 344, 511 515

Martin Studer et al.
COMMUNICATIONS
kinetic model as shown in Scheme 2 and discussed
below.
While the very high initial ee×s were impressive, it was
also clear that this method with yields of <50% and
gradually decreasing ee×s is of little preparative value.
The obvious solution would be dynamic kinetic reso-
lution as reported for the homogeneous system.[2] In
fact, with > 5 mM KOH in i-PrOH a very high rate was
obtained for 1b (100% conversion in 6min!) but the ee
was 0%. Obviously, something dramatic had happened
to the modified catalyst. One possible explanation is
that the adsorbed chiral modifier is displaced by OH ,
another one that the strong base interfered with the
interaction of the modifier with the adsorbed ketone
which is thought to occur via an H-bridge (see below).
We speculated that a solid base could still racemize the
keto ether but would not disturb the chirally modified
surface on the Pt due to physical separation by the
support. To test this idea, we screened a number of
strongly basic anion exchangers and solid bases. With
OH activated Amberlite IRA-402 (gel type) and
Amberlite IRA-900 (macroreticular), dynamic kinetic
resolution was indeed observed in i-PrOH, the optimal
solvent for the homogeneous system.[2] (R,S)-2b was
obtained with ee×s of up to 56% at >95% conversion.
MgO on silica, sepiolite, or hydrotalcite had no effect,
either because these materials were not basic enough or
because the concentration of basic sites was too low. In
toluene, even higher ee×s of >80% at >95% conversion
Figure 3. Hydrogenation of 1d in the presence of ion ex-
changer. a) upper panel, no modifier. 2.0 g 1d, 200 mg 5% Pt/
Al2O3, 25 mL toluene, 1.6g Amberlite IRA-900 OH acti-
vated, 60 bar, 25 8C. b) lower panel with 20 mg HCd.
with less than 1% of 3b were obtained but, in this Reaction conditions as in a).
solvent, only macroreticular ion exchangers showed the
desired effect. Similar experiments were conducted with
1d and an ee of 90% was achieved at 88% conversion.
simplification, the very small kSS and kRR were assumed
Typical concentration profiles observed for 1b and 1d to be equal (kRR/SS). This model[6] has only three (no ion-
in the presence of OH activated Amberlite IRA-900 exchanger) or four adjustable parameters (with ion-
are depicted in Figures 2 and 3. The kinetic data were exchanger). The values for the various constants for 1b
analyzed for 1b and 1d using the reaction scheme shown and 1d are summarized in Table 2. The surprisingly good
in Scheme 2 (for the non-dynamic case krac ˆ 0 was fit (data for 1b shown in Figures 1 3) makes us
assumed). It is clear that for a rigorous analysis, a confident that our major conclusions are real and not
Langmuir-Hinshelwood approach would be necessary artifacts of the simplifications.
as, e.g., carried out for the hydrogenation of ethyl
Table 2 shows the effect of the modifier and/or the ion
pyruvate with the same catalytic system.[8] However, for exchanger on the various rate constants and hence on
a simplified analysis, we decided to lump all kinetic and the selectivity for the hydrogenation of 1b and 1d. With
adsorption terms into pseudo-first-order rate constants both substrates, addition of the modifier (with and
kXY, as already done for a-diketones.[5] As an additional without OH ) led to a moderate to very strong increase
Scheme 2. Possible reaction pathways for the hydrogenation of 1b.
Adv. Synth. Catal. 2002, 344, 511 515
513

Hydrogenation of a-Keto Ethers: Dynamic Kinetic Resolution
COMMUNICATIONS
Table 2. Pseudo first order rate constants[a] for 1b and 1d.
Substrate Modifier Base[b] kRS kSR kRR/SS Sk krac
group but interacts with the modifier via a hydrogen
bridge as well. Our model has obvious limits because it
certainly does not explain why methoxyacetone gives
such a low ee.
1b
1b
1b
1b
1d
1d
1d
1d
No
Yes
No
Yes
No
Yes
No
Yes
No
No
Yes
Yes
No
No
Yes
Yes
2.0 2.0 0.9
5.8
95
13
19
23
14
21
22
94
0.5 0.1
In conclusion, we have shown that substituted ali-
phatic and aromatic a-keto ethers are suitable sub-
strates for the enantioselective hydrogenation catalyzed
by Cinchona-modified Pt catalysts and that both kinetic
resolution and dynamic kinetic resolution is possible. At
conversions of <50%, ee×s of up to 98% were obtained
when starting with a racemic substrate (kinetic resolu-
tion). With the addition of KOH, a strong acceleration
of the reaction but 0% ee were observed, most likely
because the chirally modified surface was disturbed/
displaced. In order to get dynamic kinetic resolution the
OH ions had to be immobilized on a solid ion
[c]
6.1 6.1 0.9
18
0.9 0.1
6.8
9.4 9.4 1.7
13
0.60.2
[c]
6.5 6.5 3.9
20
0.9 0.5
11
[a]
[b]
[c]
1
h /g catalyst, see also Experimental Section.
OH bound on Amberlite IRA-900.
Cannot be obtained from our experimental data.
of kRS and a notable decrease of all other rates. Whereas exchanger, and ee×s of >80% were obtained at >95%
the overall reaction was significantly faster for aliphatic conversion.
keto ethers (ligand accelerated catalysis[3a]), the aro-
matic ketones 1d and 1e showed a lower rate in presence
of HCd. Because the krac values are smaller than kRS, the
ee decreases slowly during the reaction (see Figures 2b
and 3b).
Experimental Section
Materials
While there is still considerable debate on the detailed
mode of action of Cinchona-modified catalysts, there is
little doubt that the enantioselective reaction occurs on
an active site formed by a number of surface Pt atoms
and one adsorbed modifier.[3] Obviously, free OH ions
can either displace the chiral modifier or deprotonate
the complex shown in Scheme 3, giving strongly acti-
vated sites with no induction. Some mobile OH ions
might also be responsible for the slightly lower initial
ee×s observed for the OH loaded ion exchanger.
All materials used were commercially available or synthesized
according to literature methods. NMR analysis was done in
CDCl3 on a Bruker 300 MHz instrument equipped with an
autosampler. The ee×s of the products were determined on a
chiral GLC column (hydrogen carrier gas, Beta-dex 110
column supplied from Supelco) or on HPLC (see below).
Typical Hydrogenation Experiment of 1d
The observed stereoselectivities can be rationalized
by applying a variant of the model proposed by Pfaltz
10 mg dihydrocinchonidine (HCd, 4a) were placed in a 50-mL
pressure autoclave equipped with a magnetic stirring bar and
and Baiker[3] (Scheme 3). The most important feature is baffles. 100 mg catalysts (JMC 94, Batch 14017/01, Supplier:
Johnson Matthey, pretreated 2 h at 400 8C under a flow of
hydrogen) were suspended in 2 mL toluene and transferred to
the autoclave. 1 g 1d was dissolved in 18 mL toluene and also
transferred to the autoclave. After sealing, the autoclave was
purged three times with argon and three times with hydrogen
and then pressurized with hydrogen to 60 bar. The reaction was
started by turning the magnetic stirrer on, and the temperature
was kept constant at 25 8C with the help of a cryostat. The
pressure in the autoclave was kept constant at 60 bar during the
reaction by the use of a pressure transducer. The hydrogen
a hydrogen bond between the quinuclidine of the
Cinchona molecule and one or two O atoms of the
concomitantly adsorbed ketone. For the hydrogenation
of a-keto esters only one N-H-O-C hydrogen bond was
proposed. However, a second interaction would explain
the positive effect of an a-oxygen in a cis position to the
C ˆ O group. It also rationalizes the preferred formation
of the (R,S)-2 isomers due to the steric effect of the
methyl group in the adsorbed state. This means that the
electronegative a-substituent not only activates the keto consumption was measured by the pressure drop in a reservoir
Scheme 3. Artists view of the adsorbed reaction intermediates.
514
Adv. Synth. Catal. 2002, 344, 511 515

Martin Studer et al.
COMMUNICATIONS
with a known volume. After 63 minutes, a sample of approx-
imately 0.5 mL was withdrawn from the solution. After
128 min, the reaction was stopped, the pressure was released,
and the autoclave was purged with argon for three times. The
catalyst was filtered off and the reaction mixture was evapo-
rated to dryness; yield: 0.85 g (85%). HPLC analysis was
Kinetic Model Used
For the determination of the rate constants we assumed that
the reactions of 1 were first order in substrate and in catalyst,
leading at constant hydrogen pressure to the following
equations:
¾
carried out on an HP 1100 with a Chiracel OD (Daicel)
d[(R,S)-2]/dt ˆ [cat](kRS[(S)-1])
d[(S,S)-2]/dt ˆ [cat](kSS[(S)-1])
d[(S,R)-2]/dt ˆ [cat](kSR[(R)-1])
d[(R,R)-2]/dt ˆ [cat](kRR[(R)-1])
column of 0.46 Â 25 cm and hexane/isopropanol (98:2) as
eluent and detection at 210 nm. Retention time: (S)-1d
12.2 min, (R)-1d 20.3 min, and for the erythro product
24.3 min for (S,R)-2dand 30.6min for ( R,S)-2dweremeasured.
For the threo diastereomers (S,S)-3d and (R,R)-3d, 15.8 and
18.7 min were measured, but no assignment of the absolute
configuration was possible for the threo diastereomers (below
2% for all systems containing modifier).
d[(R)-1]/dt ˆ [cat][(R)-1](kSR ‡ kRR
)
d[(S)-1]/dt ˆ [cat][(S)-1](kRS ‡ kSS)
In case of dynamic resolution (ion exchanger present), the
equations for (R)-1 and (S)-1 were modified to
d[(R)-1]/dt ˆ [cat]([(R)-1](kSR ‡ kRR ‡ krac
)
)
[(S)-1] krac
[(R)-1] krac
)
)
Dynamic Kinetic Resolution
d[(S)-1]/dt ˆ [cat]([(S)-1](kSS ‡ kRS ‡ krac
1.6g Amberlite IRA-900 (strongly basic anion exchanger,
converted to the OH form by washing with 0.1 M NaOH until
chloride free) were added after the substrate. The rest of the
experiment was carried out as described above.
The time-dependent concentration of all species was
calculated by numerically integrating these equations. The
values of the rate constants were determined by a least square
fit of the calculated and measured values (Microsoft EXCEL).
Determination of the Absolute Configuration of 2d
and Related erythro Isomers
References and Notes
For analysis, an isolated mixture was chromatographed on
silica gel (Merck 60 F 254) with ethyl acetate/hexane (15/85) as
eluent. By comparison to the NMR spectra of the starting
material, the earlier eluting material was identified as starting
material. The later eluting fraction was identified as 2d [erythro
mixture of (R,S)- and (S,R)-enantiomers, 81% ee according to
HPLC) by comparison to the 1H NMR reported in the
literature.[9] For the determination of the absolute configura-
tion of the major product in 2d, [a]D of the purified sample was
measured to be ˆ‡ 18.858 (DIP 181 polarimeter, 5 cm cuvette,
c 4.2, in CHCl3). The positive sign of the optical rotation
confirmed the expected preferential formation of (R,S)-2d and
the ee calculated by comparison with the literature value[9] was
81%, the same as determined by HPLC. For the hydrogenation
of 1a 1c and 1e, the main product was identified to be the
[1] a) J. M. Keith, J. F. Larrow, E. N. Jacobsen, Adv. Synth.
Catal. 2001, 343, 5; b) V. Ratovelomanana-Vidal, J. P.
Genet, Can. J. Chem. 2000, 78, 846; c) M. T. El Gihani,
J. M. Williams, J. Curr. Opin. Chem. Biol. 1999, 3, 11.
[2] a) T. Matsumoto, T. Murayama, S. Mitsuhashi, T. Miura,
Tetrahedron Lett. 1999, 40, 5043; b) K. Murata, K. Okano,
M. Miyagi, H. Iwane, R. Noyori, T. Ikariya, Org. Lett.
1999, 1, 1119.
[3] a) H. U. Blaser, H. P. Jalett, M. M¸ller, M. Studer,
Catalysis Today 1997, 37, 441; b) A. Pfaltz, T. Heinz,
Topics in Catalysis 1997, 4, 229; c) A. Baiker, J. Mol. Catal.
A: Chemical 1997, 115, 473.
[4] a) M. Studer, S. Burkhardt, H. U. Blaser, Chem. Commun.
1999, 1727; b) B. Tˆrˆk, K. Felfˆdi, K. Balazsik, M.
Bartok, Chem. Commun. 1999, 1725.
1
erythro product by comparison of the H NMR with known
[5] M. Studer, V. Okafor, H. U. Blaser, Chem. Commun.
1998, 1053.
[6] For the assignment of the stereoisomers and the detailed
kinetic model see experimental part.
[7] H. U. Blaser, H. P. Jalett, D. M. Monti, J. F. Reber, J. T.
Wehrli, Stud. Surf. Sci. Catal. 1988, 41, 153.
spectra. In analogy to 1d, the absolute configuration of the
main product was assumed to be always (R,S). This is also in
accord with the fact that in the hydrogenation of all known
activated ketones, cinchonidine derivatives such as 4a an 4b
induce the (R)-alcohols.
[8] H. U. Blaser, H. P. Jalett, M. Garland, M. Studer, H. Thies,
A. Wirth-Tijani, J. Catal. 1998, 173, 282.
[9] F. A. Davis, M. S. Haque, R. M. Przeslawski, J. Org.
Chem. 1989, 54, 2021.
Adv. Synth. Catal. 2002, 344, 511 515
515