Enantioselective hydrogenation of ketones over a tartaric acid-modified raney nickel catalyst: Substrate-modifier interaction strength and enantioselectivity

Article abstract of DOI:10.1246/bcsj.20180046

Chiral (R,R)-tartaric acid and NaBr-doubly modified Raney nickel (TA-MRNi) is a promising heterogeneous catalyst for enantioselective hydrogenation of prochiral β-keto esters. To obtain deeper insights into the factors ruling the enantioselectivity, enantiodifferentiating hydrogenation of substituted ketones was studied over TA-MRNi and NaBr-modified RNi by use of combined individual-competitive hydrogenation techniques. Relative equilibrium adsorption constants of the substrates were estimated to evaluate their relative interaction strength with adsorbed tartaric acid moiety. DFT calculations were also performed to estimate the interaction energy through hydrogen bonding, providing clear support to the kinetic analysis and surface model. It is concluded with the enantioselective hydrogenation of ketones over TA-MRNi that the enantioselectivity increases as the substrate-modifier interaction strength increases: Methyl acetoacetate (MAA) > acetylacetone (AA) ～ 4-hydroxy-2-butanone (HB) > 2-octanone (2O).

Full text of DOI:10.1246/bcsj.20180046

Enantioselective Hydrogenation of Ketones over a Tartaric
Acid-Modified Raney Nickel Catalyst: Substrate-Modifier
Interaction Strength and Enantioselectivity
Azka Azkiya Choliq, Eitaro Murakami, Shota Yamamoto, Tomonori Misaki, Morifumi Fujita,
Yasuaki Okamoto, and Takashi Sugimura*
Graduate School of Material Science, University of Hyogo, 3-2-1 Kohto, Kamigori, Ako-gun, Hyogo 678-1205, Japan
E-mail: sugimura@sci.u-hyogo.ac.jp
Received: February 15, 2018; Accepted: May 17, 2018; Web Released: May 30, 2018
Takashi Sugimura
Dr. Takashi Sugimura, Professor at University of Hyogo, graduated from Osaka University in 1979 and
received his PhD in 1984. After working as a postdoctoral fellow at Ohio State University, USA for two years,
he got a job at Himeji Institute of Technology in 1986. He has been a full professor at University of Hyogo,
Graduate School of Material Science, since 2006. He has been studying in the fields of both organic
synthesis and asymmetric hydrogenation catalysis. He received an award in 1995 for Excellent Young
Scientists from The Society of Organic Synthetic Chemistry, Japan.
Abstract
industrial applications because of its easy handling, separation
and reuse.
1
,2
Chiral (R,R)-tartaric acid and NaBr-doubly modified Raney
nickel (TA-MRNi) is a promising heterogeneous catalyst for
enantioselective hydrogenation of prochiral β-keto esters. To
obtain deeper insights into the factors ruling the enantio-
selectivity, enantiodifferentiating hydrogenation of substituted
ketones was studied over TA-MRNi and NaBr-modified RNi
by use of combined individual-competitive hydrogenation
techniques. Relative equilibrium adsorption constants of the
substrates were estimated to evaluate their relative interaction
strength with adsorbed tartaric acid moiety. DFT calculations
were also performed to estimate the interaction energy through
hydrogen bonding, providing clear support to the kinetic
analysis and surface model. It is concluded with the enantio-
selective hydrogenation of ketones over TA-MRNi that the
enantioselectivity increases as the substrate-modifier interaction
strength increases: methyl acetoacetate (MAA) > acetylacetone
It has been well established that chiral tartaric acid and
NaBr-doubly modified Raney nickel (TA-MRNi) is an excellent
chiral-modified heterogeneous catalyst for enantiodifferentiat-
3
ing hydrogenation of prochiral ketones. Intensive research
has been performed from various points of view by several
groups to disclose the origins of enantiodifferentiation as well
3­6
as to expand its applications. Tartaric acid (TA), as a chiral
modifier in TA-MRNi, has two important roles: (1) as inter-
action sites for a prochiral substrate to fix a certain enantioface
of the substrate delivering the reaction towards enantioselec-
tive hydrogenation and (2) as a poisoning agent, together with
NaBr, by covering the nickel surface preventing racemic
hydrogenation reaction from occurring. It has been reported
that the optimum coverage of tartaric acid on nickel surface
7
,8
is around 20%. After modification, there are two types of
active sites available on nickel surface: (1) enantiodifferentiat-
ing sites where tartaric acid is adsorbed on the surface induc-
ing enantioselective hydrogenation reaction by the interac-
tions with the substrate and (2) non-enantiodifferentiating sites
where tartaric acid is absent on the surface, generating racemic
products. Modification of non-enantiodifferentiating sites with
NaBr has been reported to significantly improve the enantio-
(AA) µ 4-hydroxy-2-butanone (HB) > 2-octanone (2O).
Keywords: Enantioselective hydrogenation
j
Raney nickel catalyst Tartaric acid
j
1
. Introduction
3
Selective synthesis of optically pure compounds is among
the most attractive and rewarding research especially in the
field of pharmaceuticals and agrochemicals to provide medi-
cines, pesticides, disinfectants and so forth without biological
selectivity. However, the increase in enantioselectivity is also
accompanied by significant decrease in hydrogenation rate.
It has been proposed that two hydroxyl groups of surface
tartaric acid moiety (tartrate or monosodium tartrate species)
1
3
side effects. Among several established synthesis methods,
are essential to achieve high enantioselectivity. Since the
heterogeneous catalysis induced by surface modification with
chiral molecules is one of the most promising ways for
modifier occupies the nickel surface sites for the adsorption of
the substrate and hydrogen, it acts as a poison suppressing the
Bull. Chem. Soc. Jpn. 2018, 91, 1325–1332 | doi:10.1246/bcsj.20180046
© 2018 The Chemical Society of Japan | 1325

hydrogenation reaction. However, two-point hydrogen bond
interactions between two hydroxyl groups of the tartaric acid
moiety and the carbonyl groups of β-keto esters accelerate
enantiodifferentiating hydrogenation (ligand acceleration) by
increasing the surface concentration of the TA-substrate inter-
action intermediates.8 Recently, Osawa et al. suggested the
formation of surface bisodium tartrate species, the sodium ion
Individual Hydrogenation
O
OH
H
2
, TA-MRNi or MRNi
THF, CH COOH
∗
3
R
R
Competitive Hydrogenation
,9
O
+
OH
∗
and surface hydroxyl group of which act as anchoring sites for
R
H
2
, TA-MRNi or MRNi
R
enantiodifferentiation.10
+
THF, CH COOH
3
O
O
OH O
∗
In general, enantio-differentiating hydrogenation of β-keto
ester affords β-hydroxy ester with high enantioselectivity over
O
O
3
,4
TA-MRNi. The asymmetric hydrogenation of methyl aceto-
acetate (MAA), the simplest analogue of β-keto ester, over TA-
Scheme 1. Enantioselective hydrogenation of various
ketones over TA-MRNi or MRNi (R = ­CH2CO2CH3
(methyl acetoacetate, MAA), ­CH2COCH3 (acetylacetone,
AA), ­CH2CH2OH (4-hydroxy-2-butanone, HB), ­n-C6H13
2-octanone, 2O). (In competitive hydrogenation, methyl
acetoacetate was used as a standard substrate).
MRNi gave methyl 3-hydroxybutyrate with chemical purity of
3
8
6%ee (enantiomeric excess). The highest enantioselectivity,
>
98%ee, was reported for the hydrogenation of methyl 3-
(
11
cyclopropyl-3-oxopropanoate over TA-MRNi. On the other
hand, enantiodifferentiating hydrogenation of other ketones,
such as diketones and β-hydroxy ketones, yielded relatively
lower enantioselectivity compared to β-keto ester. Hydrogena-
tion of acetylacetone (AA) gave a product with enantioselec-
tivity of 74%ee. On the other hand, hydrogenation of 4-
hydroxy-2-butanone (HB) and 2-octanone (2O) provided the
two-point hydrogen bond interactions between surface tartrate
and the substrate, generating a lower enantioselective prod-
1
4
uct (52%ee at 333 K and 10 MPa). On the other hand, the
methoxy group at the p-position of phenyl disturbs phenyl-Ni
surface interactions, resulting in higher enantioselectivity with
methyl 3-p-methoxyphenyl-3-oxopropanoate (72%ee at 333 K
corresponding products with enantioselectivity of 69%ee and
f,12
3
3%ee, respectively.3 Origins of the difference in the enan-
1
4
tioselectivity among these ketones are still under debate.
There are several factors proposed to govern enantioselec-
tivity in the asymmetric hydrogenation of ketones over TA-
MRNi: the interaction strength and conformation of surface
modifier-substrate interaction species and the ratio of hydro-
genation reaction rate on enantioselective sites to the total reac-
tion rate including the reaction on non-selective sites. However,
it is not easy to obtain direct evidence by, for example, spec-
troscopic observations with Raman spectroscopy to evaluate
their respective contributions because of the complicated nature
of the reaction systems.1 In the present study, we focused on
the substrate-modifier interaction strength from the viewpoint
of reaction kinetics by using combined individual-competitive
hydrogenation techniques to provide further understandings
of the dominant factors ruling the enantioselectivity of the
enantiodifferentiating hydrogenation reaction of ketones over a
chirally modified Raney nickel catalyst, TA-MRNi.
and 10 MPa).
In the current study, we tried to obtain further insights into
the controlling factors in the enantioselective hydrogenation
of a series of ketones: acetylacetone (AA), 4-hydroxy-2-
butanone (HB), 2-octanone (2O) and MAA, over TA-MRNi
from kinetic points of view (Scheme 1). Both individual and
competitive hydrogenation modes were used in this study to
extract the interaction strength between the substrate-surface
tartrate species. MAA was employed as a standard substrate in
the competitive hydrogenation. Both TA-MRNi and MRNi,
which was modified with NaBr alone, were used to obtain
better insights into the generation of new functions by TA
modification. In addition, we also performed DFT calculations
to theoretically substantiate the experimental approach. It is
evidently demonstrated in the present study that the enantio-
selectivity increases as the substrate-modifier interaction
strength increases: methyl acetoacetate > acetylacetone µ 4-
hydroxy-2-butanone > 2-octanone.
­3
The reaction kinetics parameters can be obtained by using
competitive hydrogenation techniques, in which two or more
substrates are competitively hydrogenated under the same
reaction conditions, with one of the substrates being used as a
standard substrate. This method allows the determination of the
reaction kinetics parameters relative to those of the standard
substrate: relative adsorption strength and reactivity of the
2
. Experimental
2.1 Materials. Methyl acetoacetate (MAA), acetylacetone
(AA) and acetic acid were purchased from Kishida Chemical
Co., Ltd. and used as received. 4-hydroxy-2-butanone (HB) and
2-octanone (2O) were obtained from Tokyo Chemical Industry
Co., Ltd. and used as received. Ion exchanged water for the
catalyst preparation was supplied from Kishida Chemical Co.,
Ltd. (R,R)-tartaric acid sodium salt was supplied from Nacalai
Tesque, Inc and used as received.
8
,9,13
substrate.
In a previous study, we conducted competitive
hydrogenation to examine the effect of substituent (methyl,
phenyl and p-methoxy phenyl) at the β-position of MAA on
9
enantioselectivity from the viewpoints of kinetics. The hydro-
genation rates of MAA and methyl 3-p-methoxyphenyl-3-
oxopropanoate were significantly increased by the modification
of Ni metal surface with TA compared with malic acid (MA)-
modified RNi (ligand acceleration), while no clear acceleration
effect was observed with methyl 3-phenyl-3-oxopropanoate. It
was proposed that strong phenyl-Ni interaction caused distorted
2.2 Catalyst Preparation.
prepared from a Ni/Al alloy (42/58, Kawaken Fine Chemicals,
Ltd. Japan) by the W-2 type development method, followed by
washing with water under ultrasonic irradiation. An aliquot of
the alloy (1.0 g) was treated in an alkaline solution of NaOH
(4.5 g) in deionized water (20 mL) at 373 K for 1 h. The modi-
Raney nickel (RNi) was
9
1326 | Bull. Chem. Soc. Jpn. 2018, 91, 1325–1332 | doi:10.1246/bcsj.20180046
© 2018 The Chemical Society of Japan

fication was performed to prepare TA-MRNi (0.4 g) by heat-
ing the RNi at 373 K for 1 h in a 50 mL-aqueous solution of
tartaric acid sodium salt as a chiral modifier (6.6 mmol) and
NaBr (5 g) as an auxiliary one, after adjustment of the pH at 3.2
with NaOH. After the modification, the solution was removed
by decantation, followed by thorough washing with water, then
with methyl alcohol, and finally with tetrahydrofuran (THF) to
prepare TA-MRNi. The preparation method of MRNi was
similar to that of TA-MRNi, except for the modification step
where only NaBr (5 g) was added in the modification solution.
species.3d The energy of hydrogen bond interactions, E , was
calculated using eq 2,
H
EH ¼ ES-GA ꢂ ES ꢂ EGA
ð2Þ
where ES-GA, ES and EGA denote the total energy of the
substrate-GA interaction species, separate GA molecule and
separate substrate molecule, respectively. In the case of HB,
a dimer model of HB was also employed for the DFT calcu-
lations, taking into consideration possible interactions of two
HB molecules in a liquid phase through hydrogen bonding
between the hydroxyl and carboxyl groups. Accordingly, the
energy of hydrogen bond interactions for HB-GA interaction
species was also calculated using eq 3,
2
.3 Hydrogenation.
To a 100 mL autoclave, a single
substrate (MAA, AA, HB or 2O, 101 mmol) for individual
hydrogenation or a mixture of the substrate (AA, HB or 2O,
5
0.5 mmol) and MAA (50.5 mmol) for competitive hydro-
EH ¼ ES-GA ꢂ 0:5 ꢀ ES-dimer ꢂ EGA
ð3Þ
genation was placed with acetic acid (0.2 mL) and TA-MRNi or
MRNi (400 mg). Hydrogen was injected (10 MPa) and the auto-
clave was heated to 333 K. Reciprocating shaking was applied
to initiate the catalytic reaction. The reaction time was limited
to 1 h with MRNi to obtain a low conversion of the substrate,
while to 3 h with TA-MRNi. The accuracy of the conversion
was estimated from reproducibility to be within 5% of the
observed one. The reaction mixture was separated from the
catalyst by decantation. NMR was used to determine the con-
version of the substrate.
where ES-dimer denotes the energy of the HB-dimer model.
3
. Results and Discussion
Both individual and competitive enantioselective hydrogen-
ation reactions were performed using a reciprocating shaking
reactor (the strongest agitation technique available in our
group) at 333 K under H -pressure of 10 MPa to eliminate as
2
much as possible the effects of diffusion on the reactions. This
reaction setup and procedures made the reaction rate analysis
possible in a Langmuir-Hinshelwood formula, as shown pre-
Enantioselective hydrogenations of AA and HB were carried
out over TA-MRNi to confirm the reported enantioselectivity
under the previous reaction conditions, where the substrate
8
,9,15
viously.
With the competitive enantioselective hydrogen-
ation reaction, an equimolar mixture of AA, HB or 2O and
MAA (the standard substrate) was used for simplicity. The
conversions of the substrates were kept lower than 20% except
for that of MAA (43% for MRNi). The (initial) reaction rate
(mmol/h g-cat) was calculated from the conversion of the
substrate thus obtained. The enantioselectivities for AA and
HB were obtained for the respective individual hydrogenation
reactions at almost full conversions. It was found that the %ee
values were 79 and 78%ee for AA and HB, respectively. These
values are in between those of MAA and 2O, as reported
(1.0 g), acetic acid (0.2 mL) and THF (10 mL) were used. Prior
to determining enantioselectivity, the reaction products were
acetylated with acetic anhydride and pyridine. Enantiomeric
excess of the products was determined with gas chromatog-
raphy (Shimadzu GC-17A equipped with a CP-Chirasil DEX-
CB capillary column). Enantiomeric excess (%ee) is defined as,
%ee ¼ 100 ꢀ ð½Rꢁ ꢂ ½SꢁÞ=ð½Rꢁ þ ½SꢁÞ
ð1Þ
where [R] and [S] denote the amounts of the (R)- and (S)-
products.
3
f,12
previously.
2
.4 DFT Calculations. DFT calculations were performed
Table 1 summarizes the reaction rates of the hydrogenation
of the substrates in both individual and competitive reactions
over MRNi (1 h) or TA-MRNi (3 h). As for MRNi without
the chiral modifier, the reaction rate in the individual hydro-
genation depends on the substrate and decreases in the order,
MAA º AA µ HB > 2O. The hydrogenation rates of the
with Gaussian 09 software using B3LYP hybrid functional
and a basis set of 6-311+G (2d.p). Several calculations with
different starting geometries were conducted to energetically
optimize the configuration. Glyceric acid (GA) was used as a
model of adsorbed tartrate or monosodium tartrate surface
Table 1. Reaction rates of the substrates in individual and competitive hydrogenation reactions over MRNi
and TA-MRNi catalysts.
Reaction rate (mmol/h g-cat)d
Entry
Catalyst
Hydrogenation
Individuala
MAA
109
AA
HB
2O
1
2
3
4
MRNi
MRNi
TA-MRNi
TA-MRNi
42.9
16.5/14.1
11.3
40.9
20.3/16.5
4.2
17.7
5.8/27.0
6.6
b
Competitive
a
Individual
13.6
Competitiveb
5.5/8.4
2.5/6.5
0.67/9.1
a
Substrate 101 mmol, acetic acid 0.2 ml, catalyst 400 mg, H2-pressure of 10 MPa, temperature of 333 K.
b
An equimolar mixture of the substrate and MAA (2 © 50.5 mmol), acetic acid 0.2 ml, catalyst 400
c
mg, H2-pressure of 10 MPa, temperature of 333 K. ee% values of the substrates, MAA (methyl
acetoacetate); 86%, AA (acetylacetone); 79%, HB (4-hydroxy-2-butanone); 78% and 2O (2-octanone);
d
3
3%. Reaction rate of the substrate for the individual hydrogenation. Reaction rates of the substrate/
MAA for the competitive hydrogenation.
Bull. Chem. Soc. Jpn. 2018, 91, 1325–1332 | doi:10.1246/bcsj.20180046
© 2018 The Chemical Society of Japan | 1327

substrate and MAA in the competitive hydrogenation are
strongly affected by each other by the simultaneous presence of
the other substrate. The reaction behavior in Table 1 is rea-
sonably understood on the basis of a Langmuir-Hinshelwood
estimated rates over SA-MRNi. Accordingly, it is evident that
the hydrogenation rate of all the substrates in Table 1 is
increased by the modification with TA, possibly, through
hydrogen bonding between the substrate and surface TA. With
the competitive hydrogenation over TA-MRNi, the rate ratios
of the hydrogenation of the substrates in AA/MAA and HB/
MAA are rather close to those observed for the individual
hydrogenations. It is suggested from eq 6 that the interaction
strengths between the surface TA moiety and MAA, AA, or HB
are not very different from each other in contrast to the adsorp-
tion strength on MRNi. On the other hand, the hydrogenation
rate of 2O is strongly suppressed by the simultaneous presence
of MAA, indicating much stronger adsorption of MAA com-
pared to 2O over TA-MRNi.
The relative adsorption strength of the substrate is qualita-
tively noted above from Table 1 for the (enantioselective)
hydrogenations on MRNi and TA-MRNi. The objective of the
present study is to (semi)-quantitatively discuss the relation-
ship between the interaction strength of the substrate with the
surface modifier and the enantioselectivity. For the present
purpose, we modify the L-H equations for the individual and
competitive hydrogenations, taking into account the present
reaction conditions. The reactions were conducted at high con-
centrations of the substrates (neat conditions without solvent)
and the initial concentrations of the substrate and MAA were
(L-H) formalism for competitive hydrogenations, assuming that
the substrates and hydrogen are competitively adsorbed on
Ni metal surface and that the first addition of an dissociatively
adsorbed hydrogen atom to the adsorbed substrate molecule
is the rate-determining step of the hydrogenation.8 The reac-
,9
tion rates of the substrate S, r , and of MAA, rMAA, can be
S
expressed as
rS ¼ kSKSCSKH1=2PH1=2
ð1 þ KSCS þ KMAACMAA þ KH1=2PH
1=2 2
=
Þ
ð4Þ
1
=2
1=2
rMAA ¼ kMAAKMAACMAAKH PH
1=2 2
=
ð1 þ KSCS þ KMAACMAA þ KH1=2PH
Þ
ð5Þ
ð6Þ
rS=rMAA ¼ ðkS=kMAAÞðKS=KMAAÞðCS=CMAAÞ
where k (k or k
) is a rate constant, K (or KMAA) and KH
S
MAA
S
the equilibrium adsorption constants of the substrate S (or
MAA) and H , C (or CMAA) the concentration of the substrate
2
S
S (or MAA), and PH the pressure of H . In the case of HB/
2
MAA, the reaction rate of HB exceeds that of MAA, although
the rate of MAA hydrogenation is much larger than that of HB
hydrogenation in the individual reaction. Taking into consid-
eration the L-H formalism in eq 6, this finding strongly sug-
gests that the equilibrium adsorption constant of HB is consid-
erably larger than that of MAA, namely, HB is more strongly
adsorbed than MAA on MRNi modified only with NaBr, sup-
pressing the hydrogenation of MAA. With AA/MAA, it is
suggested that the substrate AA is more strongly adsorbed than
MAA on MRNi, but to a lesser extent compared to HB/MAA.
On the other hand, the hydrogenation rate ratio of 2O to MAA
in the competitive reaction (0.21) is similar to that in the indi-
vidual reaction (0.16), suggesting that the adsorption strength
of 2O on MRNi is slightly larger than or similar to that of
MAA.
chosen to be almost identical (C µ CMAA) in the competitive
S
hydrogenations. The H -pressure was virtually kept constant
2
throughout the reaction. Thus, the relative reaction rate, eq 6,
of the substrate in the competitive hydrogenation on MRNi or
TA-MRNi is reduced to eq 7.
C
ðrS=rMAAÞ ¼ ðkS=kMAAÞðKS=KMAAÞ
ð7Þ
In the case of the individual hydrogenation, the reaction rate
rS is expressed by eq 8.
rS ¼ kSKSCSKH1=2PH1=2=ð1 þ KSCS þ KH1=2PH
1=2 2
Þ
ð8Þ
With the individual hydrogenation on TA-MRNi, the rate
depends on the substrate but in a different manner from that on
MRNi, MAA µ AA º 2O µ HB. On the basis of a systematic
study on the Raney nickel catalysts modified with succinic acid
Equation 8 is reduced to eq 9, assuming that C is large enough
S
1/2
1/2
or K C º 1 + K
H
PH . It was actually shown in our pre-
S
S
vious study that the hydrogenation of MAA was expressed by
a rate equation with a negative order with respect to C
without a solvent, as described by eq 9. Thus, the hydrogena-
MAA
8
(SA with no OH group), malic acid (MA with one OH group),
or TA (with two OH groups) for the enantioselective hydro-
genation of MAA, we have shown previously that the modifi-
cation of Ni metal surface with TA induces a ligand accel-
eration effect on the MAA hydrogenation rate as well as the
enhanced enantioselectivity.8 It was shown that the surface
tion rate of the substrate relative to that of MAA is approx-
imately expressed by eq 10 for the individual hydrogenation as
long as the concentration of the substrate is almost the same as
that of MAA.
,9
rS ꢃ kSKH1=2PH1=2=KSCS
ðrS=rMAAÞ ¼ ðkS=kMAAÞðKMAA=KSÞ
ð9Þ
ð10Þ
coverages of SA, MA, and TA were almost the same (ca.
I
8
2
0%). The reaction rate of MAA hydrogenation was increased
5 fold on TA-MRNi compared with that on SA-MRNi, where
1
Combining eq 7 and eq 10, eq 11 is obtained to estimate the
surface Ni metal was simply poisoned or blocked by adsorbed
SA without the substrate-modifier interactions as observed for
the hydrogenation on MRNi. It is thus expected that the
hydrogenation rate of MAA over SA-MRNi is about 1 mmol/h
g-cat under the present reaction conditions. Similarly, the
hydrogenation rates of AA, HB and 2O are roughly estimated
to be 0.4, 0.4 and 0.2 mmol/h g-cat, respectively, from the
corresponding individual hydrogenation rates over MRNi in
Table 1. The individual reaction rates of the substrates over
TA-MRNi (Table 1) are apparently much greater than these
adsorption constant of the substrate relative to that of MAA.
C
I
2
ðrS=rMAAÞ =ðrS=rMAAÞ ¼ ðKS=KMAAÞ
ð11Þ
Table 2 presents the relative equilibrium adsorption con-
stants (K /KMAA) of the substrates estimated from the reac-
S
tion rates in Table 1 using eq 11. With MRNi, (KS/KMAA) de-
creases in the order, HB µ AA º 2O µ MAA. It is evident that
HB and AA are more strongly adsorbed on Ni metal surface
than MAA. The relatively lower adsorption strength of MAA
may be due, possibly, to steric effects of the ester group to
1328 | Bull. Chem. Soc. Jpn. 2018, 91, 1325–1332 | doi:10.1246/bcsj.20180046
© 2018 The Chemical Society of Japan

Table 2. Relative equilibrium adsorption constant of the
substrate on unmodified and TA-modified catalysts and
normalized KS/KMAA.
KS/KMAA
Entry
Catalyst
MAA
AA
HB
2O
1
2
3
MRNi
TA-MRNi
Normalized
1.00
1.00
1.00
1.72
0.88
0.51
1.81
1.11
0.61
1.15
0.38
0.33
a
a[(KS/KMAA) for TA-MRNi] divided by [(KS/KMAA) for
MRNi].
Figure 2. Relationship between normalized (KS/KMAA) and
enantioselectivity (%ee). The present ( ) and previous ( )
%ee values are shown.
where E and E denote adsorption energies of the substrate
S
MAA
and MAA, respectively. The validity of this normalization is
presented in the supplementary material. The normalized rela-
norm
tive adsorption constants, (K /KMAA) , thus obtained are
S
summarized in Table 2 for AA, HB, and 2O.
The normalized (K /KMAA) of the substrate increases in the
Figure 1. Proposed two-point hydrogen bond interaction
model between methyl acetoacetate (MAA) and surface
tartrate.
S
order of 2O < AA µ HB < MAA, indicating that the inter-
action strength of the substrate-TA also increases in the order.
This is consistent with the order of the enantioselectivity for
the substrates, 2O (33%ee) < AA (79%ee) µ HB (78%ee) <
hinder the adjacent carbonyl group from interacting with Ni
metal surface. On the other hand, the order of the relative
adsorption strength is significantly changed by the modification
of Ni metal surface with TA and decreases in the order, HB µ
MAA µ AA º 2O. The relative adsorption constants of HB,
MAA, and AA are close to each other, but much larger than
that of 2O. The significant change in the relative adsorption
strength by the surface modification evidently indicates the
change in the interaction strength between the substrate and the
surface TA moiety in the enantioselective hydrogenation.
As illustrated in Figure 1, it has been proposed that the sub-
strate interacts with surface TA (tartrate species and/or mono-
MAA (86%ee). Figure 2 shows a parallel correlation between
norm
the adsorption strength as represented by ln (K /KMAA)
and
S
the enantioselectivity in the asymmetric hydrogenation of the
ketones over TA-MRNi. It has been tentatively proposed that
high enantioselectivity of MAA is accompanied by ligand
8
,9
acceleration due to its interaction with surface TA and that
the surface TA and MAA geometrically well fit each other,
resulting in the formation of stable surface MAA-TA inter-
mediates with two-point interactions (Figure 1). The interac-
tions are strong enough to fix the enantioface of the substrate
prior to hydrogenation. On the other hand, it has been hypoth-
esized that 2O, having only one carbonyl group, forms only
one-point hydrogen bond interaction with surface TA resulting
in less stable surface intermediate 2O-TA, consequently leading
to a low enantioselectivity. In the case of AA and HB, both of
them are structurally capable of forming two-point interac-
tion with surface TA similar to MAA. However, the present
kinetic analysis suggests that the interaction strength is not
large enough to achieve a very high enantioselectivity, being
consistent with the observed %ee values for AA and HB. It is
concluded from Figure 2 that the enantioselectivity of the
asymmetric hydrogenation of ketones over TA-MRNi increases
as the interaction strength between the substrate and surface TA
moiety increases. It should be noted that the effects of enhanced
interactions between the substrate and surface modifier are
two fold, that is, the stronger interaction is more favorite for
enantioface differentiation as well as for the increase in the
surface concentration of the substrate-modifier interaction inter-
mediates, thus resulting in a higher enantioselectivity accom-
panied by stronger ligand acceleration. The strong ligand accel-
eration apparently contributes to enhanced enantioselectivity by
3
sodium tartrate species), inducing enantiodifferentiation. It is
evident that the adsorption strength or the magnitude of the
equilibrium adsorption constant reflects the interaction strength
between the substrate and the surface tartrate as well as the
interaction between the substrate and Ni metal surface. It is
considered that the former interactions mainly determine the
enantioselectivity by controlling the enantioface of the sub-
strate. To extract the interaction strength between the substrate
and surface tartaric acid, the relative adsorption constant on
TA-MRNi is normalized to the relative adsorption constant on
MRNi by dividing [(K /K
) for TA-MRNi] by [(K /KMAA)
S
S
MAA
for MRNi], assuming that the substrate-Ni metal surface inter-
actions are not considerably modified by the substrate-surface
TA interactions. The normalized (K /K
), denoted by (K /
S
S
MAA
KMAA)
norm, is correlated to the interaction energy of the sub-
strate with surface TA, relative to that of MAA, as follows.
norm
ꢂ RT lnðKS=KMAAÞ
¼
½ðES on TA-MRNiÞ ꢂ ðES on MRNiÞꢁ
ꢂ ½ðEMAA on TA-MRNiÞ ꢂ ðEMAA on MRNiÞꢁ ð12Þ
Bull. Chem. Soc. Jpn. 2018, 91, 1325–1332 | doi:10.1246/bcsj.20180046
© 2018 The Chemical Society of Japan | 1329

Figure 3. Substrate-glyceric acid (GA) and 4-hydroxy-2-butanone (HB) dimer interaction models optimized geometrically and
energetically by DFT calculations. MAA, methyl acetoacetate; HB, 4-hydroxy-2-butanone; AA, acetoacetate; 2B, 2-butanone; and
GA, glyceric acid.
a
increasing the hydrogenation rate on modified sites relative to
that on possible unmodified sites.
Table 3. Hydrogen-bond energy obtained from DFT calcu-
lations for the substrate-glyceric acid interaction model.
In order to theoretically substantiate the interaction strength
between the substrate and surface modifier estimated above
from the kinetic data in Table 2, we conducted density func-
tional theory (DFT) calculations in the present study. It has
been previously proposed that TA is adsorbed on nickel metal
surface as a tartrate or monosodium tartrate, where one of the
carboxylate ions interacts with surface nickel metal, while the
other carboxylic group remains intact or is deprotonated to
form monosodium tartrate, thus the adsorbed tartrate species
slightly facing upwards (Figure 1). Owing to a big challenge in
performing DFT calculations for the substrate-surface modifier
interaction model as illustrated in Figure 1, we simplified the
interaction model in the present calculations by adopting an
isolated substrate-modifier interaction model in a gaseous
phase without Ni metal surface for the first approximation.
Thus, in our calculations, we replaced a surface tartrate species
with glyceric acid (GA, lacking one carboxyl group compared
to TA). We also substituted 2O with 2-butanone (2B) with
assumption that 2B is enough to estimate the hydrogen bond
energy due to one-point interactions. The DFT calculations
were optimized in the total energy and configuration. Table 3
summarizes the interaction energies, which were calculated by
use of eq 2, for the energetically optimized substrate-GA inter-
action species. Figure 3 presents the optimized configurations
for the simplified models. It is obvious that the interactions
between the substrate and GA are induced by hydrogen bond-
ing and that the MAA-GA interaction species is described by
a two-point interaction model in agreement with our previous
proposition for the enantioselective hydrogenation over TA-
Substrate
MAA AA
HB
2O
Hydrogen bond energy (kJ/mol) 45.0 38.3 44.1b 29.9d
c
3
4.8
a
Hydrogen bond energies are presented as ¹EH in eq 2 or eq 3.
b
c
Based on an isolated HB molecule model. Based on the HB
d
dimer model. Calculated using 2-butanone (2B) as the sub-
strate instead of 2O.
MRNi.3,8,9 A similar two-point hydrogen-bond interaction
model is also applied to AA-GA. With HB-GA, on the other
hand, Figure 3 suggests that a more complicated hydrogen-
bond interaction mode (a kind of three-point interaction mode)
is formed between the hydroxyl groups and carbonyl group in
addition to the common hydrogen bond interaction mode. GA-
2B configuration is described as a one-point interaction model,
3
as expected previously.
It is evident from Table 3 that MAA-GA has the strongest
hydrogen bond energy (45.0 kJ/mol) among the substrates
examined here, followed by HB-GA (44.1 kJ/mol). It is con-
sidered that the weaker interactions of HB-GA than MAA-GA
result from the formation of distorted three-point hydrogen
bond interactions between HB and GA because of geometrical
restrictions. On the other hand, AA-GA shows a lower inter-
action energy (38.3 kJ/mol) than MAA-GA, although two-
point hydrogen bond interaction similar to MAA-GA is formed
as shown in Figure 3. This is probably due to the presence of
the ester group in MAA, which induces a higher electron
density on the carboxyl oxygen atoms of MAA, resulting in
1330 | Bull. Chem. Soc. Jpn. 2018, 91, 1325–1332 | doi:10.1246/bcsj.20180046
© 2018 The Chemical Society of Japan

Figure 4. Relationship between the enantioselectivity (%ee)
and hydrogen bond energy obtained from DFT calculations
for the hydrogenation of the ketones over TA-MRNi. The
present ( ) and previous ( ) %ee values are shown.
Figure 5. Correlation between normalized (K /KMAA) and
hydrogen bond energy obtained from DFT calculations
S
(
= MAA, = AA, = HB and = 2O).
stronger hydrogen bonds of MAA-GA compared to those of
AA-GA. 2B-GA shows the lowest interaction energy in
Table 3 (29.9 kJ/mol), as expected from one-point hydrogen
bond interaction. Figure 4 shows the enantioselectivity (%ee)
as a function of the substrate-GA interaction energy. It is evi-
dently concluded from Figure 4 that the enantioselectivity of
the products in the enantioselective hydrogenation of prochiral
ketones over TA-MRNi increases as the interaction energy
between the substrate and surface tartrate increases.
Finally, we will discuss the correlation between the experi-
mental interaction strength of the substrate and the interaction
energy obtained by the DFT calculations. As discussed above,
the interaction strength is estimated here from the equilibrium
adsorption constant, which reflects the energy difference
between the substrate in a liquid phase and the one interacting
with the surface tartrate. With the substrates in the present
study, no specific strong interactions are expected except for
HB having a hydroxyl group and a carbonyl group in the
molecule. It is considered that these functional groups induce
hydrogen bonding between the molecules in a liquid phase. We
examined a dimer model for HB in a liquid phase for the DFT
calculations, as illustrated in Figure 3. In the case of HB-GA,
the interaction energy was calculated from eq 3 (HB dimer
model) to estimate relative adsorption constant.
genation of prochiral β-keto esters. In the present study, we
successfully performed the experimental and theoretical studies
on the enantioselective hydrogenation of various ketones over
TA-MRNi to obtain further insights into the factors control-
ling the enantioselectivity. It is demonstrated that the rela-
tive equilibrium adsorption constants of the substrates are
(semi)-quantitatively evaluated by combining individual and
competitive hydrogenation techniques to estimate the relative
interaction strength of the substrate. It is evidently substan-
tiated that the substrate-modifier interaction plays important
roles in deciding the enantioselectivity of the products and
that the enantioselectivity increases as the interaction strength
between the substrate and surface tartaric acid moiety in-
creases; methyl acetoacetate > acetylacetone µ 4-hydroxy-2-
butanone > 2-octanone.
The present approach can be extended to understand the
enantioselectivity of the other substrates on a molecular level
as well as the effects of auxiliary additives such as NaBr on the
performances of TA-MRNi. Furthermore, the present conclu-
sion provides obvious guidelines for the molecular design of
modifier-substrate combinations to yield almost perfect enan-
tioselectivity in the enantiodifferentiating hydrogenation over
heterogeneous catalysts.
Supporting Information
Figure 5 presents the substrate-TA interaction strength rela-
norm
tive to that of MAA-TA, as represented by ln (K /KMAA)
,
The derivation and verification of equation (12) are pre-
sented for clarity in the material. This material is available on
http://dx.doi.org/10.1246/bcsj.20180046.
S
as a function of the interaction energy obtained from the
DFT calculations. A satisfactory linear correlation in Figure 5
norm
demonstrates that ln (K /KMAA)
represents, at least approx-
S
imately, the substrate-TA interaction energy relative to that of
MAA-TA, substantiating the validity of our approach to the
present very complicated reaction system through kinetics com-
bined with DFT calculations. Accordingly, it is also concluded
that the competitive hydrogenation techniques are very useful
to estimate the relative substrate-modifier interaction energy
from the kinetic analysis.
References
1
a) M. Bartók, Chem. Rev. 2010, 110, 1663. b) H.-U. Blaser,
C. Malan, B. Pugin, F. Spindler, H. Steiner, M. Studer, Adv. Synth.
Catal. 2003, 345, 103. c) M. Kitamura, T. Ohkuma, S. Inoue, N.
Sayo, H. Kumobayashi, S. Akutagawa, T. Ohta, H. Takaya, R.
Noyori, J. Am. Chem. Soc. 1988, 110, 629.
2
a) T. Mallat, E. Orglmeister, A. Baiker, Chem. Rev. 2007,
07, 4863. b) Asymmetric Solutions, ed. by H. U. Blaser, E.
Schmidt, Wiley-VCH, Verlag, Weinheim, 2004.
a) Y. Izumi, M. Imaida, H. Fukawa, S. Akabori, Bull.
1
4
. Conclusion
Chiral tartaric acid-modified Raney nickel (TA-MRNi) is a
promising heterogeneous catalyst for enantioselective hydro-
3
Chem. Soc. Jpn. 1963, 36, 21. b) Y. Izumi, Angew. Chem., Int. Ed.
Bull. Chem. Soc. Jpn. 2018, 91, 1325–1332 | doi:10.1246/bcsj.20180046
© 2018 The Chemical Society of Japan | 1331

Engl. 1971, 10, 871. c) Y. Izumi, Adv. Catal. 1983, 32, 215. d) T.
Sugimura, Catal. Surv. Jpn. 1999, 3, 37. e) A. Tai, T. Harada, in
Tailored Metal Catalysis, ed. by Y. Iwasawa, D. Reidel, Dordrecht,
f) M. A. Keane, Can. J. Chem. 1994, 72, 372.
T. Osawa, M. Kitano, T. Harada, O. Takayasu, Catal. Lett.
2009, 128, 413.
7
1
986, pp. 265­324. doi:10.1007/978-94-009-5261-4_5. f) A. Tai,
8
T. Sugimura, S. Nakagawa, N. Kamata, T. Tei, T. Tajiri, R.
T. Sugimura, in Chiral Catalyst Immobilization and Recycling, ed.
by D. E. De Vos, I. F. J. Vankelecom, P. A. Jacobs, Wiley-VCH,
Weinheim, 2000, pp. 173­209. doi:10.1002/9783527613144.
ch08. g) T. Osawa, T. Harada, O. Takayasu, Curr. Org. Chem.
Tsukiyama, T. Okuyama, Y. Okamoto, Bull. Chem. Soc. Jpn. 2015,
88, 271.
9
A. A. Choliq, J. Watanabe, T. Misaki, Y. Okamoto, T.
Sugimura, Tetrahedron: Asymmetry 2016, 27, 657.
10 a) T. Osawa, Y. Tanabe, M. Fujiwara, Catal. Lett. 2017,
147, 686. b) T. Osawa, T. Harada, O. Takayasu, Top. Catal. 2000,
13, 155.
2
006, 10, 1513. h) A. Tai, T. Harada, M. Yamamoto, H. Ozaki, Y.
Izumi, Bull. Chem. Soc. Jpn. 1983, 56, 1414.
a) T. Osawa, K. Yoshino, K. Takimoto, O. Takayasu, T.
4
Harada, Catal. Lett. 2006, 112, 167. b) T. Osawa, Y. Nakagawa, M.
Ando, T. Harada, O. Takayasu, J. Mol. Catal. A: Chem. 2009, 302,
11 a) S. Nakagawa, T. Sugimura, A. Tai, Chem. Lett. 1997, 26,
859. b) T. Sugimura, S. Nakagawa, A. Tai, Bull. Chem. Soc. Jpn.
2002, 75, 355.
4
3. c) T. Osawa, M. Ando, S. Sakai, T. Harada, O. Takayasu,
Catal. Lett. 2005, 105, 41. d) T. Osawa, T. Harada, A. Tai, Catal.
Today 1997, 37, 465.
12 a) A. Tai, K. Ito, T. Harada, Bull. Chem. Soc. Jpn. 1981, 54,
223. b) S. Murakami, T. Harada, A. Tai, Bull. Chem. Soc. Jpn.
1980, 53, 1356.
13 K. Balázsik, K. Szöri, G. Szöllösi, M. Bartók, Chem.
Commun. 2011, 47, 1551.
14 S. Nakagawa, A. Tai, T. Okuyama, T. Sugimura, Top.
Catal. 2000, 13, 187.
15 Y. Nitta, T. Imanaka, S. Teranishi, J. Catal. 1983, 80, 31.
5
a) F. Zaera, J. Phys. Chem. C 2008, 112, 16196. b) F. Zaera,
Acc. Chem. Res. 2009, 42, 1152.
a) G. Kyriakou, S. K. Beaumont, R. M. Lambert, Langmuir
011, 27, 9687. b) V. Humblot, S. Haq, C. Muryn, W. A. Hofer, R.
6
2
Raval, J. Am. Chem. Soc. 2002, 124, 503. c) T. E. Jones, C. J.
Baddeley, Surf. Sci. 2002, 513, 453. d) M. A. Keane, G. Webb,
J. Catal. 1992, 136, 1. e) M. A. Keane, Catal. Lett. 1993, 19, 197.
1332 | Bull. Chem. Soc. Jpn. 2018, 91, 1325–1332 | doi:10.1246/bcsj.20180046
© 2018 The Chemical Society of Japan