Hydrogenation without a transition-metal catalyst: On the mechanism of the base-catalyzed hydrogenation of ketones

Article abstract of DOI:10.1021/ja016152r

The hydrogenation of unsaturated organic substrates such as olefins and ketones is usually effected by homogeneous or heterogeneous transition-metal catalysts. On the other hand, a single case of a transition-metal-free and purely base-catalyzed hydrogena

Full text of DOI:10.1021/ja016152r

Published on Web 06/26/2002
Hydrogenation without a Transition-Metal Catalyst: On the
Mechanism of the Base-Catalyzed Hydrogenation of Ketones
Albrecht Berkessel,* Thomas J. S. Schubert, and Thomas N. M u¨ ller
Contribution from the Institut f u¨ r Organische Chemie der UniVersit a¨ t zu K o¨ ln, Greinstrasse 4,
D-50939 K o¨ ln, Germany
Received May 7, 2001
Abstract: The hydrogenation of unsaturated organic substrates such as olefins and ketones is usually
effected by homogeneous or heterogeneous transition-metal catalysts. On the other hand, a single case
of a transition-metal-free and purely base-catalyzed hydrogenation of ketones was reported by Walling
2
and Bollyky some 40 years ago. Unfortunately, the harsh reaction conditions (ca. 200 °C, >100 bar H ,
potassium tert-butoxide as base) limit the substrate spectrum of this reaction to robust, nonenolizable ketones
such as benzophenone. We herein present a mechanistic study of this process as a basis for future rational
improvement. The base-catalyzed hydrogenation of ketones was found to be irreversible, and it shows
first-order kinetics with respect to the substrate ketone, hydrogen, and catalytic base. The rate of the reaction
depends on the type of alkali ion present (Cs > Rb ≈ K . Na . Li). Using D
2 2
instead of H revealed a
rapid base-catalyzed isotope exchange/equilibration between the gas phase and the solvent as a
concomitant reaction. The degree of deuteration of the product alcohols did not indicate a significant kinetic
isotope effect. It is proposed that both ketone reduction and isotope exchange proceed via similar six-
2 2
membered cyclic transition states involving the H (D )-molecule, the alkoxide base, and the ketone (solvent
alcohol in the case of isotope exchange). Mechanistic analogies are pointed out which apparently exist
between the base-catalyzed hydrogenation of ketones studied here and the Ru-catalyzed asymmetric ketone
hydrogenation developed by Noyori. In both cases, heterolysis of the hydrogen molecule appears to be
assisted by a Brønsted-base (i.e., alkoxide), the latter being bound to the substrate ketone or the catalyst
ligand, respectively, by a bridging Lewis-acidic alkali ion.
Introduction
Scheme 1
Hydrogenation is one of the most important chemical
processes, both in industry and in the synthesis-oriented research
laboratory. Numerous heterogeneous and homogeneous transi-
tion-metal catalysts have been developed to remarkable perfec-
tion, with applications ranging from the mass production of
hydrogenated bulk materials to the synthesis of enantiomerically
pure chemicals by asymmetric hydrogenation.1-5 On the other
hand, a single transition-metal-free hydrogenation of an organic
substrate - in this case of benzophenone (1) - was observed
already some 40 years ago by Walling and Bollyky. At high
temperatures and H2-pressure, and in the presence of potassium
tert-butanolate, the formation of benzhydrol (2) was reported
6
(Scheme 1). Clearly, the harsh reaction conditions (ca. 200 °C,
strong base) limit the substrate spectrum to stable, nonenolizable
ketones such as benzophenone. Nevertheless, a mechanism-
based improvement of this hydrogenation procedure might
eventually lead to milder reaction conditions and thus to broader
applicability. With this in mind, we found it worthwhile to study
the base-catalyzed hydrogenation in detail. In summary, a
mechanistic proposal for this intriguing transformation could
be derived on the basis of kinetic data, as well as on structural
variations of substrate and base. As it turned out, some
mechanistic similarities to the Ru-catalyzed asymmetric hydro-
genation of ketones developed by Noyori et al. apparently
*
To whom correspondence should be addressed. Phone: int+49-221-
4
70-3283. Fax: int+49-221-470-5102. E-mail: berkessel@uni-koeln.de.
1) Brown, J. M. Hydrogenation of Functionalized Carbon-Carbon Double
Bonds. In ComprehensiVe Asymmetric Catalysis; Jacobsen, E. N., Pfaltz,
A., Yamamoto, H., Eds.; Springer: Berlin, Heidelberg, 1999; pp 121-
(
1
82.
(
2) Halterman, R. L. Hydrogenation of Non-Functionalized Carbon-Carbon
Double Bonds. In ComprehensiVe Asymmetric Catalysis; Jacobsen, E. N.,
Pfaltz, A., Yamamoto, H., Eds.; Springer: Berlin, Heidelberg, 1999; pp
5
,7,8
1
83-195.
exist.
(
(
(
3) Ohkuma, T.; Noyori, R. Hydrogenation of Carbonyl Groups. In Compre-
hensiVe Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H.,
Eds.; Springer: Berlin, Heidelberg, 1999; pp 199-246.
4) Blaser, H.-U.; Spindler, F. Hydrogenation of Imino Groups. In Compre-
hensiVe Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H.,
Eds.; Springer: Berlin, Heidelberg, 1999; pp 247-265.
(6) Walling, C.; Bollyky, L. J. Am. Chem. Soc. 1964, 86, 3750-3752; 1961,
83, 2968-2969.
(7) Noyori, R.; Ohkuma, T. Angew. Chem. 2001, 113, 41-75; Angew. Chem.,
Int. Ed. 2001, 40, 40-73.
5) Noyori, R.; Yamakawa, M.; Hashiguchi, S. J. Org. Chem. 2001, 66, 7931-
(8) Ohkuma, T.; Koizumi, M.; Ikehira, H.; Yokozawa, T.; Noyori, R. Org.
Lett. 2000, 2, 659-662.
7
944.
10.1021/ja016152r CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 8693-8698
9
8693

A R T I C L E S
Berkessel et al.
Figure 1. Time course of the base-catalyzed hydrogenation of various
ketones. Reaction conditions: 210 °C, 135 bar H2, t-BuOH, 20 mol %
t-BuOK.
Results
Figure 2. Base-catalyzed hydrogenation of benzophenone (1) at various
pressures.
We first concentrated on the reproduction of the observation
made by Walling and Bollyky (WB).6 Thus, the catalytic
reduction of benzophenone (1) to benzhydrol (2) in tert-butanol
as solvent was followed as a function of time. In the presence
of 20 mol % of potassium tert-butanolate and 135 bar H2-
pressure at 210 °C, a final yield of ca. 95% benzhydrol (2) was
reached after ca. 25 h (Figure 1). This result is in good
6
agreement with the data given by WB (98% after 25 h). In
addition, our capillary GLC analysis revealed the formation of
minor quantities of diphenylmethane (2-4%). No reduction
occurred in the absence of base or when hydrogen was
exchanged for argon (at identical pressure). The latter controls
(not reported by WB) unambiguously established that H2 is
indeed the reductant. We were delighted to see that other
nonenolizable ketones such as pivalophenone (3), di-tert-butyl
ketone (4), or 2,2,5,5-tetramethylcyclopentanone (5) could
cleanly be hydrogenated as well, albeit at lower rates in the
cases of the aliphatic substrates 4 and 5. As will be discussed
later, the higher reactivity of the phenone substrates 1 and 2
may be interpreted by cation-π interaction of the alkali ion
and the phenyl residue(s) involved.
Figure 3. Effect of temperature and type of base (20 mol % in all cases)
on the base-catalyzed hydrogenation of benzophenone (1). Traces 1,3,
KOCHPh2; traces 2,4,5, KOt-Bu.
from 135 bar (Figure 2, trace 1) to 45 bar (Figure 2, trace 2)
11
resulted in a reduction of the reaction rate. When the
hydrogenation was started at 45 bar and the H2-pressure was
increased to 135 bar after ca. 12 h, the product formation was
accelerated, and the reaction profile switched almost perfectly
from the original 45 bar trace to the 135 bar curve (Figure 2,
trace 3). However, subsequent release of pressure (after ca. 24
h) did not reverse the reaction (Figure 2, trace 3). This indication
of irreVersibility was further supported by the finding that
treatment of the product benzhydrol (2) with potassium tert-
butanolate under “45 bar conditions” did not give rise to
significant quantities of benzophenone (e5%).
When the solvent tert-butanol was exchanged for diglyme,
our experiments yielded results in contrast to those reported by
WB. When rigorously purified diglyme was employed, no
transformation of the substrate benzophenone (1) occurred at
9
all. In unpurified “off the shelf” diglyme, the substrate
benzophenone (1) was largely converted to benzoic acid (ca.
5
0%, in addition to ca. 8% triphenylmethanol), independent of
the presence or absence of hydrogen. The conversion of
benzophenone (1) to these products could be induced in absolute
diglyme by addition of catalytic amounts of the radical initiator
AIBN. Apparently, a hitherto unknown radical chain conversion
of the ketone occurs that requires both a base and a radical
Performing the hydrogenation of benzophenone in the pres-
ence of potassium tert-butanolate at different temperatures
revealed that this process actually follows biphasic kinetics. At
1
0
initiator. Further experiments thus concentrated exclusively
on tert-butanol as solvent.
100 °C, the product formation leveled off at ca. 15% conversion
(Figure 3, trace 2). At 150 °C, a comparatively fast initial
Kinetic Studies, Variation of Hydrogen Pressure and
Temperature. As expected, lowering of the hydrogen pressure
reduction was followed by a slow but steady increase of the
benzhydrol concentration (Figure 3, trace 4). Finally, at 210
°
C, the “normal” smooth and almost quantitative reduction of
(
9) For example, WB reported a 52% yield of benzhydrol (2) after 18 h at 170
°
C, 100 bar H pressure, 20 mol % of t-BuOK (ref 6).
2
benzophenone (1) occurred (Figure 3, trace 5). The conversion
achieved at 100 °C corresponds to somewhat less than one
turnover of substrate per tert-butanolate (20 mol %). We assume
that at this temperature, potassium tert-butanolate is the only
(
10) For example, the radical initiator could oxidize the hemiacetal anion which
may be formed in small quantities from benzophenone (1) and tert-
butanolate to an alkoxyl radical. Intramolecular H-abstraction from a tert-
butyl methyl group (similar to a Norrish type II reaction), â-fission with
loss of isobutene, and extrusion of a phenyl radical furnish benzoic acid
and propagate the radical chain. This mechanistic assumption also explains
the formation of triphenylmethanol, that is, the addition of a phenyl moiety
to benzophenone.
(11) Final yields of benzhydrol (2): 135 bar, 95%; 90 bar (not shown), 90%;
45 bar, 79%.
8694 J. AM. CHEM. SOC.
9
VOL. 124, NO. 29, 2002

Hydrogenation without a Transition-Metal Catalyst
A R T I C L E S
Figure 4. Reaction rates as a function of benzophenone and base
concentrations (210 °C, 135 bar H2).
Figure 5. Isotope effect in the base-catalyzed hydrogenation of benzophen-
one (1).
catalytically active base. At higher temperature, the product
benzhydrolate becomes effective in heterolyzing H2 as well, and
catalytic turnover results. This assumption is in agreement with
the lower basicity of benzhydrolate as compared to that of tert-
butanolate. Obviously, the biphasic behavior poses serious
problems for a kinetic analysis of the overall process. We
therefore resorted to using benzhydrolate as the base right from
the start of the reactions. As expected, only very slow conversion
occurs under these conditions at 100 °C (Figure 3, trace 1). At
resulted in a proportional decrease of reduction rate (graph not
shown). Consequently, the reaction appears to be first order also
in hydrogen.
Effect of the Alkali Ion on the Rate of Hydrogenation.
For comparison, the hydrogenation of benzophenone (1) was
carried out in the presence of lithium, sodium, potassium,
rubidium, and cesium benzhydrolate, respectively (20 mol %
in all cases). At 210 °C and 135 bar H2, K and Rb proved
equally active (relative initial rate ) 100), whereas the Cs base
was even slightly faster (ca. 105). Sodium benzhydrolate proved
much less active (relative initial rate ca. 50), followed by lithium
1
2
1
50 °C, clean monophasic reduction of benzophenone sets in,
at a rate identical to that of the slow region of the biphasic curve
14
observed in the presence of KOt-Bu (trace 3).
(relative initial rate ca. 7). In the case of cesium, the reduction
Kinetic Orders in Benzophenone, Hydrogen, and Base.
With the monophasic reaction conditions in hand, we proceeded
to the determination of the kinetic orders of substrates and
catalyst. In Figure 4, reaction rates are plotted against the
corresponding concentrations of substrate, that is, benzophenone
of benzophenone (1) was complete after 7 h. When the latter
reaction mixtures were analyzed after 24 h at 210 °C and 135
bar H2, a minor amount of diphenylmethane was found as well
(ca. 1%). Some reduction of benzophenone (1) to benzhydrol
(2) also occurred in the presence of 20 mol % KOH (22% after
24 h), whereas KOAc was completely inactive.
1
5
(
1). For trace 1, 20 mol % of base was employed, whereas
traces 2 and 3 summarize the analogous experiments done with
0 and 5% of potassium benzhydrolate, respectively. Figure 4
Exchange of Hydrogen for Deuterium. When hydrogen was
exchanged for deuterium, the kinetic course of the hydrogenation
of benzophenone (1) was indistinguishable from that in the
presence of H2 (Figure 5, traces 1, 2). In the isolated product
1
nicely illustrates the linear relationship between substrate
concentration and reaction rate; that is, the reduction is first
order in benzophenone (1).
1
13
benzhydrol (2), H- and C NMR indicated D-incorporation at
CH(D)-OH of ca. 60%, and not quantitative, as originally
expected. This result pointed to a concomitant base-catalyzed
exchange between the gas phase and the protic solvent. Such
exchange phenomena are well documented in the literature.17
To test the validity of our assumption, the deuterium content
Furthermore, the rate constants found at base loadings of 20,
10, and 5 mol % (Figure 4, slopes of traces 1, 2, and 3) are
proportional to the base concentrations. In other words, the
reduction is also first order in base.16 Finally, the dependence
of the reaction rate on the hydrogen pressure was studied at
1
of the solution was monitored by H NMR, in the presence
1
35, 90, and 45 bar (210 °C, 20 mol % of potassium
(Figure 6, trace 1) and absence (Figure 6, trace 2) of the substrate
benzhydrolate). Quite similar to the effect of lowering the base
concentration (Figure 4), lowering of the hydrogen pressure
benzophenone (1). As shown in Figure 6, rapid isotope exchange
between the gas phase and the solution occurs, independent of
the presence/absence of the substrate ketone. After ca. 2 h at
210 °C and 135 bar [ca. 1/10 of the time needed for the full
conversion (i.e., hydrogenation) of benzophenone (1)], equilib-
rium is reached at a t-BuOD/t-BuOH ratio of ca. 6:4. We
furthermore analyzed the isotopic composition of the gas phase
by means of low-temperature GC at equilibrium conditions.18
The molar ratio of H2:HD:D2 was found to be 1.0:2.1:3.04
((0.05). Finally, the reduction of benzophenone (1) with D2
(
a
12) The pK of benzhydrol (2) is ca. 2-3 units lower than that of tert-butanol:
ref 13.
(
13) (a) Arnett, E. M.; Small, L. E. J. Am. Chem. Soc. 1977, 99, 808-816. (b)
Reeve, W.; Erikson, C. M.; Aluotto, P. F. Can. J. Chem. 1979, 57, 2747-
2
754.
(
14) In practice, potassium benzhydrolate was prepared in situ by dissolving
equivalent amounts of benzhydrol (2) and potassium tert-butanolate in tert-
butanol.
(
15) At a given conversion, the slope of a time/conversion curve is equal to the
reaction rate at the concentration of substrate still present at this point.
Experiments were run at least in duplicate, and the error bars represent
deviations from the mean value.
16) Similarly, doubling the amount of base to 40 mol % resulted in a further
increase of the reaction rate. However, the high reaction rate, as compared
to the time needed for the thermal equilibration of the autoclave, prohibited
a kinetic analysis under these conditions.
(
(17) Klingler, R. J.; Rathke, J. W. Prog. Inorg. Chem. 1991, 39, 113-180.
(18) GC on Fe-doped alumina at liquid nitrogen temperature, see Experimental
Section for details.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 29, 2002 8695

A R T I C L E S
Berkessel et al.
Scheme 2
Figure 6. Base-catalyzed isotope exchange between the gas phase and the
solvent. Trace 1, in the presence of benzophenone (1); trace 2, no
benzophenone (1).
was repeated in t-BuOD as solvent. Under these conditions, the
isotope exchange between the gas phase and the solution is
degenerate, and only D2 can react. As shown in Figure 5 (trace
3
), the reduction becomes extremely slow under these condi-
merically enriched phenyl-tert-butylcarbinol (58% ee) was
subjected to the reaction conditions, the recovered material did
not show a significant loss of enantiomeric purity (55% ee).
Consequently, the low enantioselectivity observed in our experi-
ment must be attributed to insufficient discrimination of the
diastereomeric transition states that lead to the two product
enantiomers. The high reaction temperature probably accounts
for the low effect of the subtle differences in transition-state
energy.
tions. This result indicates that H2/HD effect the formation of
benzhydrol in the D2/t-BuOH system. Control experiments, that
is, subjecting Ph2CD-OH to the strongly basic reaction condi-
tions, revealed that the deuterium label at the benzhydrylic
carbon atom is not stable and undergoes exchange with the
solvent. This exchange phenomenon frustrated a simple distinc-
tion - based on the deuterium incorporation in the product
benzhydrol (2) - whether or not there is a kinetic isotope effect
kH /kHD. However, in the purely aliphatic and less CH-acidic
2
2,2,4,4-tetramethyl-3-propanol [the reduction product of di-tert-
Discussion
butyl ketone (4, Figure 1)], no isotope exchange at the 3-position
occurs under the reaction conditions. When the ketone 4 was
treated with D2 in t-BuOH under identical conditions, NMR
analysis of the product carbinol revealed D-incorporation of ca.
For the sake of clarity, the most important features of the
base-catalyzed hydrogenation of ketones shall be summarized
again:
(i) Reversibility: The reaction is irreversible.
5
0%, that is, a value compatible only with the absence of a
(ii) Kinetic orders: The reaction is first order in ketone,
kinetic isotope effect kH /kHD.
2
hydrogen, and base.
iii) Effect of alkali ion: In the series of alkali benzhydrolates
Finally, the question of asymmetric induction in the base-
catalyzed ketone hydrogenation was addressed. If a chiral base
is employed as catalyst, face selection in the formal hydride
transfer to a prochiral ketone may be expected. The irrevers-
ibility of the process is clearly an advantage, but the high
reaction temperatures required so far are not. In preliminary
experiments, we found that the hydrogenation of prochiral
pivalophenone (3) in the presence of 20 mol % of potassium
R-1-phenylethanolate at 140 °C and 135 bar H2 for 24 h gave
ca. 50% of the product alcohol and a reproducible ee of 12%
(
as catalytic bases, the rate of the hydrogenation decreases in
the order Cs > Rb ≈ K . Na . Li.
(iv) Deuterium versus hydrogen: The product alcohols are
partially deuterated. The source of deuterium is HD, resulting
from a rapid, base-catalyzed isotope exchange between the gas
phase and the solution.
All of these experimental observations can be accommodated
by the mechanism shown in Scheme 2. The key features for
both ketone hydrogenation and isotope exchange are (i) the
binding of substrate and base to the alkali cation, and (ii) the
joint action of a Brønsted-base and a Lewis-acid (the ketone
carbonyl-C or the alcohol proton, respectively) on the H2
molecule. For both reactions, a cyclic, six-membered transition
state can be formulated. This representation might suggest that
H-H bond cleavage is the rate-determining step, but our
experimental evidence speaks against that. In the presence of
D2 instead of H2, the incorporation of D at the benzhydrylic
position clearly proceeds via intermediate HD, and both the
observed rates and the extent of deuterium labeling indicate that
there is no significant discrimination of H2 over HD. In other
words, the reaction does not show a significant kinetic H/D
effect which in turn speaks against H-H fission as the rate-
determining step. A comparison with the low relative rate
observed in the D2/t-BuOD experiment appears problematic.
19
(
major enantiomer: R-2,2-dimethyl-1-phenyl-1-propanol). Con-
trol experiments in the absence of hydrogen established that
there was no competing Meerwein-Ponndorf-Verley reduction
of the substrate ketone by the chiral secondary alkoholate used
as base. To the best of our knowledge, this is the first example
of asymmetric induction in a purely base-catalyzed hydrogena-
tion. The relatively low ee observed prompted us to investigate
whether concomitant racemization of the chiral base or the chiral
product alcohol may obscure an intrinsically higher enantiose-
lectivity. However, the 1-phenylethanol liberated from the base
after the reaction (by acidification) had very close to the same
ee as the 1-phenylethanolate employed as base (base, 85% ee;
recovered 1-phenylethanol, 83% ee). Similarly, when enantio-
(
19) Brown, H. C.; Park, W. S.; Cho, B. T. J. Org. Chem. 1986, 51, 1934-
1
936.
8696 J. AM. CHEM. SOC.
9
VOL. 124, NO. 29, 2002

Hydrogenation without a Transition-Metal Catalyst
A R T I C L E S
Scheme 3
Scheme 4
ca. 10 equiv (relative to Ru-catalyst) of potassium tert-
butanolate. A recent study by Chen and Hartmann revealed that
both the alkoxide base and the proper choice of the alkali ion
are crucial for efficient catalysis. It was proposed that the
hydrogen molecule is coordinated to the Ru-center and that the
Deuteride transfer from D-D is unique as it contains a “double-
primary isotope effect”, not to mention differences in pKa’s and
O-H versus O-D bond strengths in deuterated versus protic
2
7
2
0
tert-butanol. It may also be argued that the low rate observed
under these conditions might indicate a rate-limiting transfer
heterolysis of the H molecule occurs in the ternary complex 7,
again via a six-membered transition state (Scheme 4, right). As
2
+
+
of H versus D to the product alcoholate. Instead, we assume
that the assembling of highly ordered transition states such as
the ones shown in Scheme 2 limits the rate of reaction. It is
interesting to note that a bent arrangement of H2 and an oxygen
base was calculated by Radom et al. as the transition state of
the hydrogenation of an imidazolidinium cation, mediated by a
carboxylate base.21 This calculation was intended to model a
naturally occurring case of transition-metal-free hydrogenation,
that is, the catalysis effected by the so-called metal-free
the final result, H is split into a protic NH and a hydridic Ru-H
2
(8). The transfer of the H-atoms to the substrate ketone finally
occurs as depicted in Scheme 3 (right). Chen thus pointed out
the crucial role of the potassium ion as the Lewis acid that
properly positions the alkoxide base for the heterolysis of the
H -molecule. Furthermore, an explanation was offered why, in
2
the case of the Noyori catalyst 6, potassium is the most active
27
alkali ion (K > Na ≈ Rb >Li). This order of reactivity follows
neither the covalency of the bases nor the Lewis acidities of
the alkali ions. Instead, a molecular modeling study revealed
that the deprotonated amide nitrogen atom, together with two
aryl substituents, forms a preorganized binding pocket for the
22
hydrogenase found in methanogenic archaea.
Please note that for the mechanism postulated in Scheme 2
left), an analogy exists in the highly efficient and extremely
(
enantioselective ruthenium-catalyzed transfer hydrogenation of
ketones developed by Noyori et al.5,7,23 Theoretical studies
strongly support a six-membered transition state for this process,
as shown in Scheme 3 (right). In the direction of ketone
reduction, the NH-proton of the ligand bound to Ru interacts
with the oxygen atom of the substrate, and a hydride is
transferred by the Ru-H moiety to the carbonyl C-atom. Norton
recently reported the first example of an ionic and enantiose-
27
alkali cation. The complexation of alkali ions to aromatic
2
8-31
π-systems is well documented in the literature.
In this
particular case, the size of the binding site is optimal for
potassium. How about the transition-metal-free hydrogenation
discussed here? Clearly, also benzhydrolate can “chelate” alkali
ions by interaction with the negatively charged oxygen atom
and additional π-complexation. Several studies have shown that
in coordinating solvents (as opposed to, e.g., the gas phase),
24
lective Ru-catalyzed hydrogenation of iminium cations. In the
latter process, a Ru-hydride species results from the fission of
H2 by the cationic CpRu(diphosphine)-catalyst and subsequent
deprotonation. Again, hydride is transferred to the substrate from
the Ru-H-intermediate, followed by protonation of the resulting
+
π-complexation of benzene residues is stronger with K than
+
28-31
it is with Na .
As shown in Figure 1, the aromatic ketones
1 and 3 are hydrogenated more rapidly than purely aliphatic
ones (4, 5). In the latter cases, the product alcohols and thus
the catalytically active alkoxide bases do not carry aromatic
residues. Their lower rates of hydrogenation may be considered
as additional support for π-complexing of the potassium ions.
However, lower steric accessibility of the carbonyl C-atoms in
the cases of 4 and 5 or the electron-withdrawing effect of the
phenyl residues in ketones 1 and 3 may account for their lower
reactivity as well.
-
+
amine (a stepwise additon of H and H to the CdN-double
2
4
bond). The related Mo- or W-catalyzed hydrogenation of
2
5
ketones reported by Bullock and Voges follows a similar
mechanism, but with the opposite order of the proton/hydride
transfer.
Finally, the dependence of the hydrogenation rate on the
nature of the alkali ion shall be discussed. The covalency of
the metal oxygen bonds increases in the order Cs-Rb-K-
2
6
Na-Li. However, we believe that this is not the exclusive
cause for the observed reactivity order. Again, a mechanistic
analogy to the Ru-catalyzed asymmetric hydrogenation of
ketones seems to exist. Formula 6 (Scheme 4) represents a
ruthenium complex, developed by Noyori et al., that allows for
the asymmetric hydrogenation of a variety of ketones with
Conclusions
We have shown that a series of nonenolizable ketones can
be hydrogenated in the presence of substoichiometric amounts
of base, and in the absence of a transition-metal catalyst. Low
but significant asymmetric induction was observed when a
prochiral ketone was hydrogenated in the presence of a chiral
base. Kinetic studies revealed that the base-catalyzed hydroge-
5
,7,8
excellent yields and enantioselectivities (up to 99%).
Inter-
estingly, the hydrogenation protocol involves the addition of
(
(
(
(
20) Scheiner, S.; C˜ uma, M. J. Am. Chem. Soc. 1996, 118, 1511-1521.
21) Scott, A. P.; Golding, B. T.; Radom, L. New J. Chem. 1998, 1171-1173.
22) Berkessel, A. Curr. Opin. Chem. Biol. 2001, 5, 486-490.
(27) Hartmann, R.; Chen, P. Angew. Chem. 2001, 113, 3693-3697; Angew.
Chem., Int. Ed. 2001, 40, 3581-3585.
(28) Gokel, G. W.; De Wall, S. L.; Meadows, E. S. Eur. J. Org. Chem. 2000,
2967-2978.
23) Yamakawa, M.; Ito, H.; Noyori, R. J. Am. Chem. Soc. 2000, 122, 1466-
1
478.
(29) De Wall, S. L.; Meadows, E. S.; Barbour, L. J.; Gokel, G. W. J. Am. Chem.
Soc. 1999, 121, 5613-5614.
(30) Ma, J. C.; Dougherty, D. A. Chem. ReV. 1997, 97, 1303-1324.
(31) Dougherty, D. A. Science 1996, 271, 163-168.
(24) Magee, M. P.; Norton, J. R. J. Am. Chem. Soc. 2001, 123, 1778-1779.
(25) Bullock, R. M.; Voges, M. H. J. Am. Chem. Soc. 2000, 122, 12594-12595.
(26) Bains, M. S. Can. J. Chem. 1964, 42, 945-946.
J. AM. CHEM. SOC.
9
VOL. 124, NO. 29, 2002 8697

A R T I C L E S
Berkessel et al.
nation is first order in ketone, hydrogen, and base, in agreement
with the assumption of a six-membered cyclic transition state
involving the H2-molecule and an aggregate of the catalytic base,
alkali cation, and ketone. There seem to be analogies to the
ruthenium catalysts developed by Noyori which transfer hy-
drogen to ketones by a two-stage mechanism. First, the
heterolysis of the H2-molecule is effected by binding to the Ru-
center and subsequent attack of a Brønsted base. The resulting
ensemble of hydridic Ru-H and protic N-H reduces the
substrate ketone. In the absence of a hydrogen-binding and
hydride-accepting transition-metal center, the main problem of
a purely base-catalyzed hydrogenation of ketones seems to lie
in the proper preorientation of the substrate ketone and the
catalytically active base. Apparently, the fission of the H2-
molecule is not rate-limiting. Instead, poor population of the
As a control experiment, deuterated benzhydrol (Ph
subjected to the above reaction conditions. After 48 h at 210 °C and
35 bar H
, 91% of benzhydrol was reisolated. 1H NMR analysis
2
CD-OH) was
1
2
indicated that the deuterium label was almost completely replaced by
hydrogen. When 2,2,4,4-tetramethyl-3-propanol-3-d was treated analog-
1
ously for 72 h, reisolation (87%) and H NMR analysis revealed that
g98% of the deuterium label was still present.
Attempted Base-Catalyzed Hydrogenation of Benzophenone (1),
Diglyme as Solvent. Benzophenone (1, 364 mg, 2.00 mmol) and
potassium tert-butanolate (45-673 mg, 20-300 mol %) were dissolved
in 5 mL of abs. diglyme. The solution was transferred into the autoclave,
pressurized at ca. 20 °C with 54 bar H
to 130 °C. After 5 h, the mixture was poured into water, acidified with
HCl, and extracted with ether. After being dried over anhydrous Na
SO , the solvent was removed, and the residue was analyzed by GLC.
2
or Ar, respectively, and heated
2
-
4
Monitoring of the H/D-Exchange between Gas Phase and
Solution. The autoclave was charged with a solution of 0.90 g (8.00
mmol) of potassium tert-butanolate in 100 mL of abs. tert-butanol.
After repeated evacuation and flushing with deuterium, the autoclave
was heated to 210 °C, and the deuterium pressure was adjusted to 135
bar. Samples were withdrawn by means of the sampling tube and
transferred directly into an NMR tube, equipped with an inner capillary
“
reactive” conformation of the substrate-base assembly most
likely accounts for the relatively low efficiency of the process
at its current stage. Further work in our laboratory aims at
improving the preorganization of substrate and base.
Experimental Section
3
containing CDCl as reference and lock solvent. The deuterium content
1
of the tert-butanol solvent was determined by H NMR integration of
the tert-butyl resonance and the OH-signal. The composition of the
Solvents and Chemicals. tert-Butanol, tert-butanol-d, and diglyme
were purchased from Aldrich, purified by repeated distillation from
sodium, and stored under argon. Commercially available benzophenone
2 2
gas phase (H , HD, D ) was analyzed as follows: When the NMR
analysis of the solution indicated that equilibrium was reached (ca. 2
h), a sample of the gas phase was withdrawn and analyzed by GC,
using a 1.1 m × 2 mm column packed with Fe(III)-doped alumina at
liquid nitrogen temperature and heat capacity detection. Calibration with
(1) was purified by repeated recrystallization from ethanol and drying
in vacuo. Pivalophenone (3), di-tert-butyl ketone (4), and 2,2,5,5-
tetramethylcyclopentanone (5) were prepared according to published
3
2
procedures. The corresponding alcohols were obtained from the
ketones by LAH- and LAH-d -reduction, respectively. Lithium and
H
2
, HD, D
2
-mixtures of known composition allowed for the quantitative
:HD:D ) 1.0:2.1:3.04 ((0.05).
4
analysis: H
2
2
sodium tert-butanolate were purchased from Fluka, potassium tert-
butanolate from Riedel-de Haen. These alkali tert-butanolates were
purified by sublimation at 180 °C, ca. 1 Torr, prior to use. Rubidium
and cesium tert-butanolate were prepared by dissolving the alkali metals
Hydrogenation of Pivalophenone (3) in the Presence of a Chiral
Base. Potassium R-1-phenylethanolate was prepared from commercially
available R-1-phenylethanol and potassium hydride. This material (64
mg, 0.40 mmol) was suspended in 324 mg (2.00 mmol) of pivalophen-
one (3). The mixture was transferred into the autoclave. After two cycles
of evacuation-flushing with argon, the autoclave was pressurized to
(Aldrich) in abs. tert-butanol. Pressurized deuterium (2.7) was supplied
by Messer-Griesheim.
Base-Catalyzed Hydrogenations of Ketones in tert-Butanol,
Kinetic Measurements. All reactions were run in a 300 mL Parr
Instruments Ltd. Series 4560 autoclave, equipped with a temperature
control unit, a glass liner, and a sampling tube. Typically, the autoclave
was charged under argon (glovebox) with 40 mmol of the ketone and
2
83 bar H at 25 °C and heated to 140 °C for 12 h. The reaction mixture
was worked up as described above and analyzed by GLC. The
enantiomeric excess of the 2,2-dimethyl-1-phenyl-propan-1-ol formed
was analyzed on a WCOT-FS CP-Chiralsil-Dex column (25 m). The
configuration of the major enantiomer was found to be R by co-injection
with a sample of enantiomerically enriched R-2,2-dimethyl-1-phenyl-
propan-1-ol, prepared according to ref 19.
8
.0 mmol (20 mol %) of the alkali tert-butoxide in 100 mL of tert-
butanol. For reactions catalyzed by alkali benzhydrolates, 1 equiv
relative to the base) of benzhydrol (2) was added, too. The autoclave
(
When the reaction was run with potassium R-1-phenylethanolate of
was evacuated/flushed with hydrogen (or deuterium) twice and pres-
surized to an extent that the desired final pressure was reached upon
heating. For the kinetic analyses, samples of ca. 200 µL were withdrawn
by means of the sampling tube. These samples were quenched
immediately by addition to a mixture of 1 mL of methanol, 100 µL of
concentrated hydrochloric acid, and 2 mL of dichloromethane. After
filtration through a plug of neutral alumina, the dichloromethane phases
were analyzed by GLC on a 25 m HP-5 column. Calibration curves
were recorded for all components for quantitative analysis. For each
substrate, the hydrogenation experiments were worked up preparatively,
and the products (and starting ketone in case of nonquantitative
conversion) were isolated by column chromatography (silica gel, eluting
with dichloromethane). In all cases, the mass balances were g95%.
8
5% ee, the 1-phenylethanol recovered after acidification of the reaction
mixture had an ee of 83%. When enantiomerically enriched 2,2-
dimethyl-1-phenyl-propan-1-ol (58% ee) was subjected to the reaction
conditions, the recovered alcohol had an ee of 55%.
Acknowledgment. This work is dedicated to A.B.’s scientific
mentor, Professor Ronald Breslow, on the occasion of his 71st
birthday. Support by the Fonds der Chemischen Industrie and
by the BASF AG, Ludwigshafen, is gratefully acknowledged.
The authors wish to thank Dr. G. Glugla and S. Gr u¨ nhagen,
Karlsruhe Research Center - Technology and Environment, for
performing the gas phase H2, HD, D2 analyses.
(32) Millard, A. A.; Rathke, M. W. J. Org. Chem. 1978, 43, 1834-1835.
JA016152R
8698 J. AM. CHEM. SOC.
9
VOL. 124, NO. 29, 2002