A remarkable solvent effect on reductive amination of ketones

Article abstract of DOI:10.1016/j.mcat.2018.05.017

We report the first systematic study of solvent effect on reductive amination of ketones with ammonia and dihydrogen (H₂) over Ru/C, Rh/C, Pd/C and Pt/C catalysts. Protic (water, methanol, ethanol and isopropanol), aprotic polar (dioxane and tetrahydrofuran) and aprotic apolar (cyclohexane and toluene) solvents were investigated. Reaction kinetic model was built to reveal solvent-dependent reaction pathway and solvent-related rate constant for individual steps. Primary amine is produced via two distinct routes, i.e., hydrogenation of imine and hydrogenolysis of Schiff base adduct. These two routes co-exist in organic solvents, while the preference of which route to take heavily depends on the nature of the solvent. In contrast, the formation of imine and Schiff base are not favored in water, resulting in high selectivity towards alcohol. Methanol is identified as the best solvent for reductive amination of ketones, attributed to the highest rates for imine and Schiff base formation compared to other solvents, as well as high hydrogenation activity.

Full text of DOI:10.1016/j.mcat.2018.05.017

MolecularCatalysis454(2018)87–93
Contents lists available at ScienceDirect
Molecular Catalysis
journal homepage: www.elsevier.com/locate/mcat
A remarkable solvent effect on reductive amination of ketones
Song Song, Yunzhu Wang, Ning Yan⁎
Department of Chemical and Biomolecular Engineering, National University of Singapore, 4 Engineering Drive 4, 117585 Singapore, Singapore
A R T I C L E I N F O
A B S T R A C T
Keywords:
We report the first systematic study of solvent effect on reductive amination of ketones with ammonia and
dihydrogen (H2) over Ru/C, Rh/C, Pd/C and Pt/C catalysts. Protic (water, methanol, ethanol and isopropanol),
aprotic polar (dioxane and tetrahydrofuran) and aprotic apolar (cyclohexane and toluene) solvents were in-
vestigated. Reaction kinetic model was built to reveal solvent-dependent reaction pathway and solvent-related
rate constant for individual steps. Primary amine is produced via two distinct routes, i.e., hydrogenation of imine
and hydrogenolysis of Schiff base adduct. These two routes co-exist in organic solvents, while the preference of
which route to take heavily depends on the nature of the solvent. In contrast, the formation of imine and Schiff
base are not favored in water, resulting in high selectivity towards alcohol. Methanol is identified as the best
solvent for reductive amination of ketones, attributed to the highest rates for imine and Schiff base formation
compared to other solvents, as well as high hydrogenation activity.
Reductive amination
Solvent effect
Ketone
Amines
Hydrogenation
1. Introduction
transformation of heptaldehyde to 1-heptylamine [25].
In addition to catalysts, the choice of solvent often has a great effect
Primary amines are important chemical intermediates with wide
applications in the synthesis of polymers, drugs, dyes and pharmaceu-
ticals [1–3]. Reductive amination of carbonyl compounds with am-
monia (NH3) and dihydrogen (H2) to primary amines represents a more
atom efficient synthetic method compared to other processes that use
organic amines [4–7] and stoichiometric reductants (such as borohy-
dride [8–10] and formate [11–13]). It is also a key reaction for re-
newable amine production from biomass [14,15]. The easy formation
of side products, such as secondary amines, tertiary amines or corre-
sponding alcohols, is the main problem encountered in reductive ami-
nation of carbonyl compounds [16–19]. Significant achievements have
been made to overcome this obstacle over heterogeneous [17,20–30],
homogeneous [31–33] and enzymatic catalysis [34,35]. Most efforts in
the development of heterogeneous catalysts have so far been focused on
Group VIII metals [22–26,28,29], to reveal the effect of the metal center
and the support. For example, Lercher et al. reported that the catalyst
selectivity in reductive amination of butyraldehyde was determined by
the type of metal and the substrate: Ru- and Rh- based catalysts favored
the formation of primary amine, while Pd- and Pt- based catalysts were
more effective in producing secondary amine [23]. Nakamura et al.
reported the amination of ketones with ammonia in o-xylene over Pt
based catalysts, in which the authors claimed that the Lewis acid sites
on the support played an important role in the synthesis of primary
amine [24]. Dong et al. revealed that amphoteric supports were su-
perior to purely basic and relative acidic supports for the
on reaction activity [36–38]. To our knowledge, however, systematic
study concerning solvent effect on reductive amination was not avail-
able [25]. Recently, Chatterjee et al. reported transformation of furfural
to furfuryl amine in aqueous ammonia [26]. When commercial Ru/C
was used as the catalyst, secondary amine (97.9% selectivity) was de-
tected as the main product in water. However, Komanoya et al. found
that in methanol a moderated yield of furfuryl amine (31%) was
achieved over the same catalyst [28]. In another report, co-solvent
(water-ethanol) was used to increase the yields of primary amines for
some specific substrates using Ru/ZrO2 catalyst [29]. It appears that the
selectivity towards primary amine was not only based on the choice of
catalysts, but also affected strongly by the selection of solvent. The
reaction pathway in reductive amination is also under debate
[19,26,28]. Whether the formation of primary amine results from the
hydrogenation of imine or from hydrogenolysis of the adduct Schiff
base is not clear. The apparent contradictory reports on reaction
pathway may due to the different solvent system used in each study.
Since inspections into the literature suggest solvent may play a pi-
votal role, we conducted a detailed study on solvent effects in reductive
amination of ketones with NH3 and H2 over Group VIII metal-based
catalysts. Different solvents including protic (water, methanol, ethanol
and isopropanol), aprotic polar (dioxane and tetrahydrofuran) and
aprotic apolar (toluene and cyclohexane) solvents were investigated
and compared. Using cyclohexnone as a model substrate, the reaction
pathway and the reaction rate constants for individual steps in various
⁎
Corresponding author.
E-mail address: ning.yan@nus.edu.sg (N. Yan).
https://doi.org/10.1016/j.mcat.2018.05.017
Received 30 March 2018; Received in revised form 9 May 2018; Accepted 15 May 2018
2468-8231/©2018ElsevierB.V.Allrightsreserved.

S. Song et al.
MolecularCatalysis454(2018)87–93
Scheme 1. Reaction pathway for reductive amination of cyclohexanone over the Ru/C catalyst.
solvents have been estimated based on experimental data and kinetic
modelling.
pore size of 0.45 μm. The liquid products were analyzed with GC
(Agilent GC 7890A with a flammable ionization detector). Qualitative
analysis of reaction products was conducted with GC–MS (GC–MS, an
Agilent 7890A GC system and 5975C inert MSD with triple-axis de-
tector) with column HP-5.
2. Experimental
2.1. Chemicals
2.4. Kinetic modeling of the reaction
All chemicals were purchased from commercial suppliers: cyclo-
hexanone (Sigma-Aldrich, > 99.0% purity), cyclohexanol (Sigma-
Aldrich, 99% purity), cyclohexylamine (Sigma-Aldrich, > 99.9%
purity), methanol (Sigma-Aldrich, 99.9% purity), ethanol (Sigma-
Aldrich, 99.5% purity), toluene (Fisher Chemical, 99.9% purity), cy-
clohexane (Sigma-Aldrich, anhydrous, 99.5% purity), tetrahydrofuran
(Sigma-Aldrich, anhydrous, > 99.9% purity), isopropanol (Sigma-
Aldrich, anhydrous, 99.5% purity), dioxane (Sigma-Aldrich, anhydrous,
99.8% purity,) and ultra-pure water (Siemens Ultra-Clear TWF Water
Purification Systems).
The concentration of dissolved gas in solvent was affected by tem-
perature and gas partial pressure. The solubility of H2 in selected sol-
vents was limited based on the mole fraction of H2 (XH2) in Table S1,
the calculated equations are listed as follow [40,41]:
nH
2
XH
=
2
nH + nliquid
2
nH
2
C (H2) =
V
where nH2 is the moles of hydrogen dissolved in solvent, nliquid is the
moles of solvent, C[H2] is the mole concentration of H2 in solvent and V
is the volume of solvent at 298 K. The mole conversion of hydrogen was
below 20% in all solvents. Therefore, the concentration of H2 in all
solvents was assumed to be a constant in the reaction kinetic model.
The solubility of ammonia in solvent is much higher than the so-
lubility of hydrogen based on the mole fraction of NH3 (XNH3) shown in
Table S1. The ammonia pressurized procedure is the same in all sol-
vents, thus, C[NH3] = 2 M was used in all solvents. The estimation is as
follows: during 10 times purging, ca. 5 mmol NH3 is dissolved in the
solvent (0.5 mmol each time, based on ideal gas law). During the final
charging of 4 bar NH3, an additional 1.5 mmol NH3 is dissolved. As
such, the total amount of dissolved NH3 is estimated to be 6.5 mmol,
providing a NH3 concentration of ca. 2 M considering the total volume
is 3 mL. The conversion of NH3 was also below 20%. Therefore, the
concentration of NH3 was assumed to be a constant in the reaction
kinetic model.
2.2. Catalysts
Commercial 5% Ru/C, 5% Rh/C and 5% Pt/C catalysts were pur-
chased from Shaanxi Kaida Chemical Engineering Co., Ltd. 5% Pd/C
catalyst was synthesized in our lab following a literature procedure
[39]. Before reaction, the catalysts were treated with 5% H2/N2 at
673 K for 1 h with a ramping rate of 10 K/min, then cooled down to
room temperature.
2.3. Catalytic reactions
The reductive amination of cylcohexanone was conducted in a
14 mL autoclave reactor with a magnetic stirrer (1000 rpm). Typically,
1 mmol cyclohexanone was added into 3 mL solvent, then 0.02 g cata-
lyst was added and the autoclave was sealed and purged with 2 bar NH3
ten times to remove the remaining air in the reactor. Each time, NH3
was purged back to 2 bar when the pressure dropped to ca. 1 bar. After
that, the NH3 pressure was adjusted to 4 bar, and gas valve was closed.
When the pressure decreased to atmosphere (since ammonia quickly
dissolves in the solvent), the autoclave is further pressurized with H2
and the final pressure was adjusted to 10 bar. Then, the reaction was
conducted at room temperature (298 K) under stirring for certain times.
After reaction, the catalyst was filtered by a PTEE membrane with a
The reaction pathway for reductive amination of cyclohexanone to
cyclohexylamine was depicted in Scheme 1. C[cyclohexanone],
C[imine], C[cyclohexylamine], C[Schiff base] and C[cyclohexanol] are
used to represent the concentration of substrates, intermediates and
products, respectively. NH3 and H2 are both in large excess so they are
treated as pseudo-first-order. Moreover, the hydrogen pressure has a
first order dependence on both C]O and C]N bond hydrogenation
88

S. Song et al.
MolecularCatalysis454(2018)87–93
(Table S2). As such, first-order reaction was assumed for all steps in-
volved in Scheme 1. The number of catalyst active sites was assumed to
keep constant in the entire reaction process [42,43]. Based on the
Scheme 1 and these assumptions, kinetic equations are listed as follows:
Table 1
Reductive amination of cyclohexanone in different solvents over Ru/C catalysts.
Entry
Solvent
Con. (%)
Product distributions (%)
dC [cyclohexanone]
= −k1*C [cyclohexanone]*C [NH3] − k4
dt
*C [cyclohexanone]*C [H2]
1
2
3
4
5
6
7
8
Methanol
Ethanol
Isopropanol
Water
100
87.8
36.1
8.9
25.7
0
0
61.3
27.4
7.8
5.9
2.1
74.2
0
0
12.2
15.6
0
4.4
91.8
82.1
89.5
36.1
30.9
67.9
69.8
18.7
40.6
0
97.0
100
4.9
3.8
39.3
48.4
0.1
3.0
0
− k5*C [cyclohexanone]*C [cyclohexylamine] + k2*C [imine]*C [H2O]
dC [imine]
= k1*C [cyclohexanone]*C [NH3] − k3*C [imine]*C [H2] − k2
THF
dt
Dioxane
Toluene
Cyclohexane
*C [imine]*C [H2O]
21.6
53.2
dC [cyclohexylamine]
= k3*C [imine]*C [H2] − k5*C [cyclohexanone]
Reaction
conditions:
Cyclohexanone
1 mmol,
catalyst
0.02 g
dt
(Ru:cyclohexanone = 1:100), solvent 3 mL, H2 = 9 bar, T = 298 K, agitator
speed = 1000 rpm, time = 2 h.
*C [cyclohexylamine]
+ 2*k6*C [Schiff base]*C [H2]*C [NH3]
established a solvent dependent reaction pathway and reaction kinetic
profile, to provide rationales for the observed solvent effect.
dC [Schiff base]
= k5*C [cyclohexanone]*C [cyclohexylamine]
dt
− k6*C [Schiff base]*C [H2]*C [NH3]
3.2. Modeling of reaction kinetics
dC [cyclohexanol]
Based on GC–MS analysis of product mixtures and a previous report
[19], a reaction pathway for the reductive amination of cyclohexanone
over Ru/C catalyst was depicted in Scheme 1. Cyclohexanone first re-
acts with ammonia to form imine, and further hydrogenation of imine
affords the desired product cyclohexylamine. However, cyclohex-
ylamine is not inert in the system; it reacts with cyclohexanone to form
a Schiff base, as often observed as an intermediate in reductive ami-
nation [21]. The Schiff base further reacts with ammonia to form an
adduct, and then converts to two molecules of cyclohexylamine via
hydrogenolysis [23]. This means, cyclohexylamine is first produced
from imine hydrogenation, and additionally from a self-propagation
process where one amine molecule becomes two via Schiff base as an
intermediate. In the literature, the Schiff base may be directly reduced
to a secondary amine by hydrogenation [18]. Nevertheless, only trace
amount of secondary amine was detected in all reactions, in accordance
with a previous finding that the hydrogenation of the Schiff base was
not favored over Ru-based catalysts [23]. As such, formation of sec-
ondary amine is not further analyzed in the kinetic modeling, and we
only consider direct hydrogenation of cyclohexanone to cyclohexanol
as the main side reaction.
Reaction kinetic model has been established based on the reaction
pathway (Scheme 1), where first-order reaction was assumed for all
steps. The adduct (addition of ammonia to Schiff base) was un-
detectable, indicating that reaction rate constant k7 was much larger
than k6. Therefore, the rate constant of Schiff base to adduct (k6) was
used to represent the rate constant of Schiff base to cyclohexylamine.
The reversed reaction rate constant of Schiff base formation, i.e., Schiff
base reacts with water, and decomposes into cyclohexanone and cy-
clohexylamine (correspond to rate constant k8) was much lower than k5
according to modeling results. Thus, k8 was ignored in the reaction
kinetic model. All kinetic equations for reactant, intermediates and
products are listed in Experimental Section 2.4. On the basis of ex-
perimental data in selected solvents, reaction rate constants for each
step calculated from the kinetic model in Scheme 1 were summarized in
Table 2. The fitting curves were shown in Figs. 1–5, which well cap-
tured the key features of the concentration of various species as a
function of time for selected solvents (methanol, ethanol, water, to-
luene and dioxane).
= k4*C [cyclohexanone]*C [H2]
dt
Water is produced from the dehydration steps, and the concentra-
tion of water is equal to the amounts of nitrogen-containing compounds
based on Scheme 1 (except for water as solvent), as shown below:
C [H2O] = C [imine] + C [cyclohexylamine] + C [Schiff base]
The yield of cyclohexylamine from imine hydrogenation was cal-
culated based on the equation shown below. The equation of C[imine]
was generated by the curve fitting toolbox of MATLAB based on si-
mulated concentration data.
t
Yield of cyclohexylamine from imine hydrogenation =
k3*C [H2]
∫
0
*C [imine]dt
The time-on-stream concentration of substrate, intermediates and
products in different solvents were used to simulate the reaction kinetic
constants. Reaction kinetic parameters were simulated based on the
least squares fitting algorithm of MATLAB lsqcurvefit [43–48]. Several
constraints have been used to make sure that the fitting curve is rea-
sonable.
3. Results and discussion
3.1. Solvent effects on reductive amination of cyclohexanone with ammonia
and H2
Reductive amination of cyclohexanone with NH3 and H2 over Ru/C
catalyst was chosen as a model reaction to investigate the solvent ef-
fects [16,17]. Eight solvents including protic solvents (water, methanol,
ethanol and isopropanol), aprotic polar solvents (dioxane and tetra-
hydrofuran) and aprotic apolar solvents (cyclohexane and toluene)
were investigated at 298 K for 2 h. The results were summarized in
Table 1. The highest conversion was obtained with methanol as solvent
(entry 1, 100%), followed by water and other alcohols (entries 2–4,
∼80–90%), then aprotic apolar solvents (entries 7–8, ∼60–70%), and
finally aprotic polar solvents (entries 5–6, ∼30–40%). Cyclohex-
ylamine or Schiff base were the main products in alcohols and aprotic
apolar solvents, while imine was the main product in aprotic polar
solvents. In water, on the other hand, the main product was cyclohex-
anol. These results clearly highlighted a critical influence of solvent in
both substrate conversion and product distribution. Prompt by this, we
The fitting curves for methanol were shown in Fig. 1. The rate
constant of cyclohexanone to imine (k1) was three times higher than the
rate constant of cyclohexanone to cyclohexanol (k4), which was con-
sistent with the fact that imine was observed as the major product in the
89

S. Song et al.
MolecularCatalysis454(2018)87–93
Table 2
Reaction rate constants (min−1) calculated from the kinetic model in Scheme 1.
Solvent
k1
k2
k3
k4
k5
k6
Methanol
Ethanol
Water
Toluene
Dioxane
0.229
0.023
0.064
0.004
0.737
0.104
3.944
0.793
3.880
0.364
0.045
0.726
0.264
0.137
0.052
0.081
0.049
1.584
0.043
0.003
0.140
0.029
0.201
0.009
0.002
5.000
1.299
–
0.122
0.015
3.194
0.810
0.255
0.149
–
0.522
0.201
0.048
0.030
k = k1*k3/k2 = 15.193
0.005
0.123
0.001
0.073
0.533
1.963
1.132
0.023
0.287
0.009
0.026
0.004
0.107
0.082
Fig. 2. Experimental (cycles) and modeled (solid lines) concentration of cy-
clohexanone and different products in the reductive amination of cyclohex-
anone with ethanol as solvent over Ru/C catalyst.Figure 3. Experimental (cy-
cles) and modeled (solid lines) concentration of cyclohexanone and different
products in the reductive amination of cyclohexanone with water as solvent
over Ru/C catalyst.
Fig. 1. Experimental (cycles) and modeled (solid lines) concentration of cy-
clohexanone and products in the reductive amination of cyclohexanone with
methanol as solvent over Ru/C catalysts.
first twenty minutes (Fig. 1b). The rate constant of imine to cyclohex-
anone (k2) was comparable with that of imine to cyclohexylamine (k3),
but k2 is less than 20% of the rate constant of cyclohexanone with cy-
clohexylamine to Schiff base (k5). As such, a large amount of cyclo-
hexanone has been consumed in the initial five minutes, and further
suppressed the formation of cyclohexanol.
The fitted curve for the yield of cyclohexylamine against reaction
time exhibited an inflection point at 40–50 min. The slope of the curve
gradually increased to a peak value until the concentration of the Schiff
base reached its maximum. Afterwards, the slope of the curve decreased
as the concentration of the Schiff base decreased. Accordingly to the
reaction network, cyclohexylamine is produced via two routes, in-
cluding the hydrogenation of imine and the hydrogenolysis of Schiff
base adduct. MATLAB was employed to identify the percentage of cy-
clohexylamine from each route (detailed in experimental section). In
methanol, approximately 50% of cyclohexylamine originated from the
hydrogenation of imine, and the rest was produced from the hydro-
genolysis of Schiff base adduct (Table S4).
The yield trends of products (Fig. 2) were similar in ethanol to that
in methanol (Fig. 1), but the reaction rate constants were drastically
different. The biggest difference lay in k1, which was almost ten times
lower in ethanol, well justifies that the inferior activity was due to the
much slower formation rate of imine from cyclohexanone. Other steps
were inhibited as well. From Table 2, the reaction rate constants of k2 to
k6 in ethanol were 2–6 times smaller than that in methanol. This in-
duced a significantly enhanced percentage of intermediates in the
product mixture after 3 h reaction. Nonetheless, the ratio of cyclohex-
ylamine from the hydrogenolysis of Schiff base adduct was close to 50%
with ethanol as solvent, which was similar to that in methanol (Table
S4).
Fig. 3. Experimental (cycles) and modeled (solid lines) concentration of cy-
clohexanone and different products in the reductive amination of cyclohex-
anone with water as solvent over Ru/C catalyst.
Therefore, k = k1*k3/k2 was used to represent the formation rate from
cyclohexanone to cyclohexylamine (detailed equations provided in the
Supporting Information). Only a trace amount of the Schiff base
(< 0.1% yield) was detected during the entire reaction process sug-
gesting the formation of the Schiff base is not favored. We further
conducted a control experiment where cyclohexanone and cyclohex-
ylamine were added into aqueous ammonia (Table S3). The yield of the
Schiff base was only 1%, indisputably highlighted the strong inhibition
Two major products were identified using water as the solvent,
namely cyclohexylamine and cyclohexanol (Fig. 3). Imine was not de-
tected, not unexpected though, due to the instability of imine in water.
90

S. Song et al.
MolecularCatalysis454(2018)87–93
Fig. 4. Experimental (cycles) and modeled (solid lines) concentration of cy-
clohexanone and different products in the reductive amination of cyclohex-
anone with toluene as solvent over Ru/C catalyst.
Fig. 5. Experimental (cycles) and modeled (solid lines) concentration of cy-
clohexanone and different products in the reductive amination of cyclohex-
anone with dioxane as solvent over Ru/C catalyst.
effect of water for the formation of Schiff base. Thus, the rate of cy-
clohexanone with cyclohexylamine to Schiff base (k5) was ignored in
the kinetic model. After these simplification, we obtained the rate
constant of cyclohexanone to cyclohexanol (k4), as well as that for cy-
clohexanone to cyclohexylamine (k). Strikingly, k4 in water is 20 times
and 30 times more than that in methanol and ethanol, respectively,
suggesting hydrogenation of C]O double bond to alcohol is greatly
enhanced. This is an important discovery, implying that the low se-
lectivity towards amine products in water is not only due to the in-
stability of imine, but also due to the remarkably enhanced ability of
the catalyst in C]O double bond hydrogenation.
Toluene was chosen as an example of apolar aprotic solvents, and
the fitting result was shown in Fig. 4. The conversion of cyclohexanone
was slower than earlier cases. In the first 45 min, imine was observed as
the main product, but the concentration was significantly lower than
that in alcohols. Indeed, the k1 value in toluene is at least one order of
magnitude lower than that in other solvents (except water). At the same
time, the reaction rate constant of Schiff base (k5) in toluene was a few
to ten percent compared to the values in alcohols, in accordance with
control experiments on Schiff base formation in different solvents
where toluene was the second worst solvent after water (Table S3).
Despite these disadvantages, the rate constant of imine hydrogenation
(k3) was the largest, while the rate constant of cyclohexanone hydro-
genation was the second smallest in toluene. For these reasons, the yield
of cyclohexylamine was 85 mol% at full conversion of cyclohexanone,
comparable to that in methanol (88 mol%) and in ethanol (86 mol%).
In toluene, close to 70% of cyclohexylamine originated from Schiff base
(Table S4), which was different with that in protic solvents.
The simulated result for an aprotic polar solvent, dioxane, was
shown in Fig. 5. A strong inhibition was observed for both cyclohex-
anone and imine hydrogenation (Table 1). Therefore, the percentage of
amine formation from imine hydrogenation was very small (16% based
on fitting results, see Table S4). A dominant portion of the product
came from the hydrogenolysis of Schiff base adduct, more significant
than that in toluene, and entirely different from reactions in alcohols
and water.
Based on reaction kinetic research, we conclude that there are dif-
ferent reaction pathways in various solvents for reductive amination.
Cyclohexylamine is the major final product in organic solvents. In al-
cohols, cyclohexylamine is produced equally from the hydrogenation of
imine and hydrogenolysis of Schiff base adduct. In aprotic solvents,
cyclohexylamine is mainly produced from the hydrogenolysis of Schiff
base adduct with a much slower overall rate. In water, the formation of
cyclohexanol is dominate.
3.3. Nature of solvent effects on reductive amination of cyclohexanone
Hildebrand-Hansen and Kamlet-Taft parameters are two common
sets of parameters used to explain the solvent effects [49–51]. We first
attempted to associate catalytic performance with solvent properties,
but no clear correlations between solvent parameters and cyclohex-
ylamine yield can be observed (as shown in Fig. S1 and Fig. S2).
Therefore, the remarkable solvent effect on reductive amination cannot
be simply attributed to the intrinsic properties of the solvent as de-
scribed by Hilderbrand-Hansen or Kamlet-Taft parameters. Instead,
solvent has a profound influence on the catalyst and substrates resulting
in a modification of specific reactivity.
In aprotic polar solvents, such as dioxane and tetrahydrofuran, the
main product was imine (Table 1, entry 5 and 6). The higher selectivity
towards imine implied that hydrogenation of imine was inhibited,
which was further testified by the small hydrogenation rate constants
(k3 and k4 in dioxane, see Table 2). Such an inhibition effect in hy-
drogenation, plausibly, is due to the strong adsorption of the solvent
molecules on the catalysts blocking the accessibility of the active sites
[52]. For protic and aprotic apolar solvents, the solvent-catalyst inter-
action is weaker so that the hydrogenation reaction is not inhibited.
Indeed, comparable hydrogenation rate constants (k3 and k4) were
found in methanol, ethanol and toluene (Table 2). Therefore, the
drastically different product distributions in these solvents was a result
of solvent-dependant imine and Schiff base formation rate. Protic sol-
vents are more active promoting the reaction between ammonia and
ketone, as well as reaction between imine and ketone than aprotic
apolar solvents. The lowest selectivity of cyclohexylamine observed in
water is due to two reasons: 1) the instability of imine and Schiff base in
water, which is well known, and 2) the significantly enhanced C]O
double bond hydrogenation activity. One explanation for the second
factor is that water directly participated in the reaction by acting as a
hydrogen donor, which has been suggested by DFT calculations for C]
O hydrogenation under basic conditions [53].
To identify whether the solvent effects observed on Ru catalysts are
extendable to other metal catalysts, reductive amination of cyclohex-
anone over Pt/C, Rh/C and Pd/C catalysts in methanol and water were
investigated (Fig. 6, Ru/C data were included for easier comparison).
The exact trend observed over Ru/C catalysts, i.e., the selectivity to-
ward cyclohexylamine in methanol was substantially higher than that
in water, was also observed over Pt/C (10% vs 0.8%), Rh/C (89% vs
42%) and Pd/C (72% vs 13%) catalysts, suggesting the drastic solvent
effect in amination is not specific to Ru, but is a general phenomenon.
91

S. Song et al.
MolecularCatalysis454(2018)87–93
Acknowledgments
The authors thank the Young Investigator Award from National
University of Singapore, and Tier-1 & Tier-2 projects from Singapore
Ministry of Education for financial supports (R-279-000-462-112, R-
279-000-464-133 and R-279-000-479-112). We thanks Mr. Fatih Yakar
for helping with recycling experiments.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the
online version, at http://dx.doi.org/10.1016/j.mcat.2018.05.017.
References
[1] K.S. Hayes, Appl. Catal. A-Gen. 221 (2001) 187–195.
[2] C. Gunanathan, D. Milstein, Angew. Chem. Int. Ed. 47 (2008) 8661–8664.
[3] T.C. Nugent, Chiral Amine Synthesis: Methods, Developments and Applications,
John Wiley & Sons, 2010, 2018.
[4] J.L. Klinkenberg, J.F. Hartwig, Angew. Chem. Int. Ed. 50 (2011) 86–95.
[5] P.V. Ramachandran, P.D. Gagare, K. Sakavuyi, P. Clark, Tetrahedron Lett. 51
(2010) 3167–3169.
[6] R. Apodaca, W. Xiao, Org. Lett. 3 (2001) 1745–1748.
[7] W. Yang, L. Wei, F. Yi, M. Cai, Catal. Sci. Technol. 6 (2016) 4554–4564.
[8] A.F. Abdel-Magid, S.J. Mehrman, Org. Process. Res. Dev. 10 (2006) 971–1031.
[9] A.F. Abdel-Magid, C.A. Maryanoff, K.G. Carson, Tetrahedron Lett. 31 (1990)
5595–5598.
[10] B.C. Ranu, A. Majee, A. Sarkar, J. Org. Chem. 63 (1998) 370–373.
[11] S. Ogo, K. Uehara, T. Abura, S. Fukuzumi, J. Am. Chem. Soc. 126 (2004)
3020–3021.
[12] B. Basu, S. Jha, M.M.H. Bhuiyan, P. Das, Synlett 2003 (2003) 0555–0557.
[13] R. Kadyrov, T.H. Riermeier, Angew. Chem. Int. Ed. 42 (2003) 5472–5474.
[14] M.J. Hülsey, H. Yang, N. Yan, ACS Sustain. Chem. Eng. 6 (2018) 5694–5707.
[15] W. Deng, Y. Wang, S. Zhang, K.M. Gupta, M.J. Hülsey, H. Asakura, L. Liu, Y. Han,
E.M. Karp, G.T. Beckham, P.J. Dyson, J. Jiang, T. Tanaka, Y. Wang, N. Yan, Proc.
Natl. Acad. Sci. U. S. A. 115 (2018) 5093–5098, http://dx.doi.org/10.1073/pnas.
1800272115.
[16] K.V. Chary, K.K. Seela, D. Naresh, P. Ramakanth, Catal. Commun. 9 (2008) 75–81.
[17] S.R. Kirumakki, M. Papadaki, K.V. Chary, N. Nagaraju, J. Mol. Catal. A: Chem. 321
(2010) 15–21.
[18] F. Qi, L. Hu, S. Lu, X. Cao, H. Gu, Chem. Commun. 48 (2012) 9631–9633.
[19] A.W. Heinen, J.A. Peters, H.V. Bekkum, Eur. J. Org. Chem. 2000 (2000)
2501–2506.
Fig. 6. Product distributions of reductive amination of cyclohexanone over
Group VIII metal-based catalysts in methanol and water. Reaction conditions:
Cyclohexanone 1 mmol, catalyst 0.02 g, solvent 3 mL, H2 = 9 bar, T = 298 K,
agitator speed = 1000 rpm, time = 2 h.
Recycling experiments of cyclohexanone reductive amination over
Ru/C catalyst were conducted. The conversion of cyclohexanone de-
creased from 94.8% to 87.7% after five cycles. The yield of imine in-
creased, while the yields of Schiff base, cyclohexylamine and cyclo-
hexanol decreased (as shown in Fig. S3). These suggest that the Ru/C
catalyst is reusable and reasonably stable in methanol.
[20] C.F. Winans, J. Am. Chem. Soc. 61 (1939) 3566–3567.
[21] S. Gomez, J.A. Peters, T. Maschmeyer, Adv. Synth. Catal. 344 (2002) 1037–1057.
[22] S. Gomez, J.A. Peters, J.C. van der Waal, W. Zhou, T. Maschmeyer, Catal. Lett. 84
(2002) 1–5.
4. Conclusion
[23] J. Bodis, L. Lefferts, T.E. Muller, R. Pestman, J.A. Lercher, Catal. Lett. 104 (2005)
23–28.
In organic chemistry, the solvent effect is well established and
widely practiced. This was not the case for either gas-liquid-solid three
phase catalytic reactions in general, or metal catalyzed reductive ami-
nation reaction in particular. In this study, we report a dramatic solvent
effect on reductive amination of ketones over metal catalysts in protic,
aprotic polar and aprotic apolar solvents. A complete, solvent-depen-
dent reaction network was established and quantitatively analyzed.
Imine and Schiff base are two key intermediates for amine formation,
while direct hydrogenation of ketone to alcohol is the major side re-
action. In alcohols, cyclohexylamine is produced equally from the hy-
drogenation of imine and hydrogenolysis of Schiff base adduct while in
toluene and dioxane, cyclohexylamine is mainly originated from hy-
drogenolysis of Schiff base adduct.
[24] Y. Nakamura, K. Kon, A.S. Touchy, K. Shimizu, W. Ueda, ChemCatChem 7 (2015)
921–924.
[25] B. Dong, X.C. Guo, B. Zhang, X.F. Chen, J. Guan, Y.F. Qi, S. Han, X.D. Mu, Catalysts
5 (2015) 2258–2270.
[26] M. Chatterjee, T. Ishizaka, H. Kawanami, Green Chem. 18 (2016) 487–496.
[27] C. Schäfer, B. Nişanci, M.P. Bere, A. Daştan, B. Török, Synthesis 48 (2016)
3127–3133.
[28] T. Komanoya, T. Kinemura, Y. Kita, K. Kamata, M. Hara, J. Am. Chem. Soc. 139
(2017) 11493–11499.
[29] G. Liang, A. Wang, L. Li, G. Xu, N. Yan, T. Zhang, Angew. Chem. Int. Ed. 129 (2017)
3096–3100.
[30] J.J. Martínez, E. Nope, H. Rojas, M.H. Brijaldo, F. Passos, G. Romanelli, J. Mol.
Catal. A: Chem. 392 (2014) 235–240.
[31] J. Gallardo-Donaire, M. Hermsen, J. Wysocki, M. Ernst, F. Rominger, O. Trapp,
A.S.K. Hashmi, A. Schäfer, P. Comba, T. Schaub, J. Am. Chem. Soc. 140 (2018)
355–361.
[32] J. Gallardo‐Donaire, M. Ernst, O. Trapp, T. Schaub, Adv. Synth. Catal. 358 (2016)
Water is not a preferable solvent, since it prevents imine/Schiff base
formation while promotes undesired C]O hydrogenation. Aprotic
polar solvents have strong solvent-catalyst interactions that inhibit the
hydrogenation activity of the catalyst, resulting in a slow overall re-
action rate towards amine. The major issue with aprotic apolar solvents
is that the formation of two key intermediates, i.e., imine and Schiff
base, is sluggish. Methanol is the most suitable solvent due the highest
rates for imine and Schiff base formation compared to other solvents,
and a high rate for hydrogenation. This study highlights the critical role
of solvents in determining the activity and selectivity in reductive
amination, which should not be neglected in designing new catalytic
systems and rationalizing catalyst performance.
358–363.
[33] S. Enthaler, ChemSusChem 3 (2010) 1024–1029.
[34] A.K. Holzer, K. Hiebler, F.G. Mutti, R.C. Simon, L. Lauterbach, O. Lenz, W. Kroutil,
Org. Lett. 17 (2015) 2431–2433.
[35] J.H. Schrittwieser, S. Velikogne, W. Kroutil, Adv. Synth. Catal. 357 (2015)
1655–1685.
[36] H. Yoshida, Y. Onodera, S. Fujita, H. Kawamori, M. Arai, Green Chem. 17 (2015)
1877–1883.
[37] S. Song, G. Wu, W. Dai, N. Guan, L. Li, J. Mol. Catal. A: Chem. 420 (2016) 134–141.
[38] T.W. Walker, A.K. Chew, H. Li, B. Demir, Z.C. Zhang, G. Huber, R.C. Van Lehn,
J. Dumesic, Energy Environ. Sci. 11 (2018) 617–628.
[39] J. Moreira, P. Del Angel, A. Ocampo, P. Sebastian, J. Montoya, R. Castellanos, Int. J.
Hydrogen Energy. 29 (2004) 915–920.
[40] Q. Liu, F. Takemura, A. Yabe, J. Chem. Eng. Data 41 (1996) 1141–1143.
[41] E. Brunner, J. Chem. Eng. Data 30 (1985) 269–273.
92

S. Song et al.
MolecularCatalysis454(2018)87–93
[42] Y. Wan, M. Zhuang, S. Chen, W. Hu, J. Sun, J. Lin, S. Wan, Y. Wang, ACS Catal. 7
(2017) 6038–6047.
[48] M. Osman, M.M. Hossain, S. Al-Khattaf, Ind. Eng. Chem. Res. 52 (2013)
13613–13621.
[43] S. Siankevich, G. Savoglidis, Z. Fei, G. Laurenczy, D.T. Alexander, N. Yan,
P.J. Dyson, J. Catal. 315 (2014) 67–74.
[44] S.-H. Kwack, M.-J. Park, J.W. Bae, K.-S. Ha, K.-W. Jun, React. Kinet. Mech. Catal.
104 (2011) 483–502.
[49] J. Zhang, N. Yan, ChemCatChem 9 (2017) 2790–2796.
[50] X. Wang, R. Rinaldi, ChemSusChem 5 (2012) 1455–1466.
[51] M. Khodadadi-Moghaddam, A. Habibi-Yangjeh, M.R. Gholami, Appl. Catal. A-Gen.
341 (2008) 58–64.
[45] H. Ait Rass, N. Essayem, M. Besson, ChemSusChem 8 (2015) 1206–1217.
[46] P. Castaño, J.M. Arandes, B. Pawelec, J.L.G. Fierro, A. Gutiérrez, J. Bilbao, Ind. Eng.
Chem. Res. 46 (2007) 7417–7425.
[52] H.J. Wan, A. Vitter, R.V. Chaudhari, B. Subramaniam, J. Catal. 309 (2014)
174–184.
[53] A. Rossin, G. Kovács, G. Ujaque, A. Lledós, F. Joó, Organometallics 25 (2006)
5010–5023.
[47] K. Lamminpää, J. Ahola, J. Tanskanen, Ind. Eng. Chem. Res. 51 (2012) 6297–6303.
93