Continuous Flow Asymmetric Hydrogenation with Supported Ionic Liquid Phase Catalysts Using Modified CO2 as the Mobile Phase: From Model Substrate to an Active Pharmaceutical Ingredient

Article abstract of DOI:10.1021/acscatal.8b00216

The continuous flow asymmetric hydrogenation of (hetero)aromatic enamides has been realized using a Rh-Quinaphos catalyst immobilized in a supported ionic liquid phase (SILP) and employing supercritical CO₂ modified with toluene (modCO₂) as the mobile phase. This approach allows expansion of the scope of the original SILP/scCO₂ system to nonvolatile substrates with poor solubility in pure CO₂. The potential of a SILP catalyst in combination with modCO₂ was demonstrated for an industrial case study using the continuous flow hydrogenation for the synthesis of a key intermediate of an active pharmaceutical ingredient (API) from AstraZeneca's portfolio. Toluene was selected as the most promising modifier, and the influence of the ratio of modifier to CO₂ was evaluated in detail. The catalyst support was found to play a major role for maintaining constant performance and the use of hydrophobic fluorous reverse-phase silica (FRP-SiO₂) instead of dehydroxylated silica strongly enhanced the long-term stability under continuous flow operation. Virtually a single enantiopure product was obtained over a prolonged time-on-stream of 90 h (quantitative single-pass conversion, ee > 99%) reaching a total turnover number of 10 300 at a space-time yield (STY) of 24 g L^-1 h^-1. No metal contamination was detected in the product solutions, indicating effective catalyst retention.

Full text of DOI:10.1021/acscatal.8b00216

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
Research Article
pubs.acs.org/acscatalysis
Cite This: ACS Catal. 2018, 8, 3297−3303
Continuous Flow Asymmetric Hydrogenation with Supported Ionic
Liquid Phase Catalysts Using Modified CO2 as the Mobile Phase: from
Model Substrate to an Active Pharmaceutical Ingredient
Daniel Geier,† Pascal Schmitz,† Jędrzej Walkowiak,† Walter Leitner,*,†,‡ and Giancarlo Francio*,†
̀
†Institut fur Technische und Makromolekulare Chemie, RWTH Aachen University, Worringerweg 2, 52074 Aachen, Germany
̈
‡Max-Planck-Institut fur Chemische Energiekonversion, Stiftstraße 34-36, 45470 Mulheim an der Ruhr, Germany
̈
̈
S
* Supporting Information
ABSTRACT: The continuous flow asymmetric hydrogenation of
(hetero)aromatic enamides has been realized using a Rh-Quinaphos
catalyst immobilized in a supported ionic liquid phase (SILP) and
employing supercritical CO2 modified with toluene (modCO2) as the
mobile phase. This approach allows expansion of the scope of the original
SILP/scCO2 system to nonvolatile substrates with poor solubility in pure
CO2. The potential of a SILP catalyst in combination with modCO2 was
demonstrated for an industrial case study using the continuous flow
hydrogenation for the synthesis of a key intermediate of an active
pharmaceutical ingredient (API) from AstraZeneca’s portfolio. Toluene
was selected as the most promising modifier, and the influence of the ratio
of modifier to CO2 was evaluated in detail. The catalyst support was found
to play a major role for maintaining constant performance and the use of
hydrophobic fluorous reverse-phase silica (FRP-SiO2) instead of dehydroxylated silica strongly enhanced the long-term stability
under continuous flow operation. Virtually a single enantiopure product was obtained over a prolonged time-on-stream of 90 h
(quantitative single-pass conversion, ee > 99%) reaching a total turnover number of 10 300 at a space−time yield (STY) of 24 g
L−1 h−1. No metal contamination was detected in the product solutions, indicating effective catalyst retention.
KEYWORDS: continuous flow synthesis, asymmetric hydrogenation, SILP catalysts, carbon dioxide, catalyst immobilization, rhodium,
quinaphos
examples show that this requirement is met if both substrates
and products are in the gas phase, as ionic liquids have
extremely low volatility (Figure 1).8−10 With the aid of a
controlled evaporation mixing device using helium as carrier gas
INTRODUCTION
■
Continuous flow techniques have been attracting increasing
attention as a promising technology for sustainable production
of fine chemicals and pharmaceuticals.1,2 For transition metal-
catalyzed reactions, the catalyst immobilization method plays a
crucial role for the realization of an integrated continuous flow
process, ideally achieving high selectivity and productivity at
minimized metal contamination in the product.3 Preferred
approaches are noncovalent immobilization strategies4 which
allow the use of off-the-shelf catalysts and can lead to a catalyst
retention comparable to that achieved with the more tedious
and time-demanding covalent anchoring procedures.5 More-
over, covalent immobilization strategies lead in most cases to a
depletion of activity and selectivity performances of the
immobilized catalyst as compared to the original molecular
catalyst. This drawback is much less pronounced by non-
covalent immobilization approaches where the catalyst
structure is maintained.6 Among noncovalent immobilization
strategies, the supported ionic liquid phase (SILP) method is a
straightforward approach where the medium, in which the
catalyst is dissolved, is immobilized.7 A prerequisite for efficient
continuous flow reactions with SILP catalysts is the
preservation of the solvent film on the support. Several
Figure 1. Selected examples of continuous flow reactions using SILP
catalysts.
Received: January 17, 2018
Revised: March 6, 2018
© XXXX American Chemical Society
3297
DOI: 10.1021/acscatal.8b00216
ACS Catal. 2018, 8, 3297−3303

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
ACS Catalysis
Research Article
and methanol, also relatively high boiling compounds like
methyl pyruvate (bp 135 °C) and methyl acetoacetate (bp 169
°C) could be used as substrates.11,12 The delivery of nonvolatile
substrates containing functional groups in the gas phase,
however, is not a suitable option in most cases. In particular,
when the targeted reaction is an asymmetric transformation, the
high temperatures required to generate a gas phase can be
beyond the thermal stability of the substrates or products, or
lead to reduced chemo- or enantioselectivity. Highly
enantioselective continuous flow asymmetric catalysis with
SILP catalysts involving medium volatile substrates have been
achieved using supercritical CO2 (scCO2) as the mobile phase,
as this medium effectively combines gas-phase transport
properties with liquid-like solvation properties.13−18 Sufficient
solubility of substrates and products in the scCO2 phase is an
indispensable requirement for the realization of such processes.
Functionalized substrates of medium polarity and low volatility,
which are common in the fine-chemical and pharma industry,
do often not possess useful solubilities, however, hampering a
broad applicability of this methodology.
In this work, we present a significant extension of the original
scope of SILP catalysis to substrates with no appreciable
solubility in scCO2. This was achieved using a strategy widely
applied in supercritical fluid chromatography (SFC), i.e., mixing
scCO2 with a cosolvent, a so-called entrainer or modifier.23 The
use of modified scCO2 (modCO2) results in a mobile phase
with higher solvent strength for the solutes, yet maintaining the
compatibility with SILP catalysts as demonstrated by stable
catalyst performance over 90 h on stream in the continuous
flow Rh-catalyzed asymmetric hydrogenation of enamides with
excellent enantioselectivities of >99% ee. The mobile phase is
characterized by medium polarity and low viscosity and hence
does neither dissolve nor strip the IL film from the support,
allowing for effective IL/catalyst retention. Factors controlling
the long-term catalyst stability have been investigated, and
finally, this strategy has been successfully applied in the
asymmetric hydrogenation of a real industrial case study, an
active pharmaceutical ingredient (API) of AstraZeneca’s
portfolio.
Scheme 1. Asymmetric Hydrogenation of 1
therefore seemed an ideal case study to explore the possibility
to adapt the SILP/scCO2 system for these kinds of substrates.
In preliminary experiments, we used N-(1-phenylvinyl)-
acetamide (1a) as a model substrate possessing a structure
similar to the actual target molecule of this investigation, N-(1-
(5-fluoropyrimidin-2-yl)vinyl)acetamide (1b). Enamides 1a
and 1b are polar solids with a melting point of 89 and 123
°C, respectively. Both substrates have a negligible solubility of
<0.1 mg L−1 in scCO2 at 40 °C and pressures up to 200 bar (see
SI for determination procedure) and, hence, are not suitable for
direct use with scCO2 as a mobile phase for continuous
hydrogenation. As a possible and practical way to circumvent
this problem, we reasoned to dissolve the substrate in a suitable
solvent and then to mix this solution into the flow of scCO2 and
H2 before entering the reactor. Thus, the solvent would play
the role of a modifier for scCO2, resulting in a mobile phase
exhibiting sufficient solvent strength for substrates and products
without stripping or dissolving the ionic liquid film from the
support. In this way, reaction conditions compatible with a
highly selective transformation should be applicable. Moreover,
the delivery of the substrates as a liquid solution would be
beneficial for keeping the continuous flow setup as simple as
possible.
A sufficient solubility of the substrates 1a and 1b31 in the
modifier solvent is the initial selection criterion, as the final
polarity of the mobile phase, which will critically affect catalyst
and IL retention, can be modulated to some extent by adjusting
the ratio between substrate solution and scCO2. As possible
modifier solvents, apolar linear and cyclic alkanes were
discarded because they are not able to dissolve the polar
substrates 1 in appreciable amounts. The aromatic solvent
toluene, the medium polar dichloromethane (DCM), and
methanol as a highly polar and protic solvent were selected for
testing. As expected, the solubility of 1a increases significantly
with the polarity of the solvent (Figure 2). Next, we studied the
phase behavior of solutions of 1a in the modifier at 40 °C at
increasing CO2 pressures in order to identify a suitable process
window. For optimum mass transfer in the reaction chamber, it
is required that all components merge into a single
homogeneous phase and no substrate precipitation occurs, as
this event can cause detrimental clogging of the continuous
flow setup. In a typical experiment, the desired amount of 1a
and modifier was loaded into a 10 mL window autoclave
equipped with a magnetic stirring bar. The solution was heated
to 40 °C, and while stirring, CO2 was added stepwise via a
needle valve. After waiting for the system to equilibrate, the
phase behavior was visually monitored.
RESULTS AND DISCUSSION
■
The asymmetric hydrogenation of enamides is one of the most
explored fields of enantioselective catalysis.24 This convenient
method provides access to biologically active chiral amides and
amines and has found a number of applications at the industrial
level.25 Quite rare, however, are examples of asymmetric
hydrogenation of enamides under continuous flow condi-
tions.26,27 A prototypical example is provided by the chiral
amine (S)-5-fluoropyrimidin-2-yl-ethanamine, a key building
block for the preparation of a JAK2 kinase 119 inhibitor
previously tested at the clinical level at AstraZeneca.28 Two
different medicinal chemistry methods were devised for
producing this intermediate, the asymmetric hydrogenation
with Rh/Et-DuPhos of the enamide 1b (Scheme 1) and the
biocatalyzed transamination of the corresponding aldehyde.29,30
Whereas the latter methodology was scaled up to produce the
material on a kilogram scale, the batch-wise asymmetric
hydrogenation with Rh/Et-DuPhos was not pursued further
because of the high costs of the catalyst used at 1 mol % loading
and the tedious workup procedure for removing metal
contamination. The potential to overcome these limitations
by continuous flow operation was demonstrated recently.27 It
Pressurizing with CO2 a solution of 1a (1.0 M) in methanol
(1 mL) in a window autoclave (10 mL) resulted in a biphasic
system to up to 75 bar consisting of a gaseous phase and a
liquid phase that expanded with increasing pressure.32 At
pressures of 80 bar or higher, the system turned monophasic,
3298
DOI: 10.1021/acscatal.8b00216
ACS Catal. 2018, 8, 3297−3303

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
ACS Catalysis
Research Article
Scheme 2. Asymmetric Hydrogenation of 1 with a SILP
Catalyst Based on Rh-Quinaphos
Figure 2. Maximum solubility of 1a in pure organic solvent at rt (solid
bars) and maximum concentration in pure scCO2 and modCO2 (1.0
mL of modifier) allowing for monophasic conditions upon
pressurization with CO2 (T = 40 °C, p = 200 bar, V(tot) = 10 mL;
striped bars).
solvent adjusting the pore filling degree of the support material
to the optimal value of α = 0.36.17 Preliminary test reactions in
batch were carried out in window autoclaves equipped with a
porous glass inlet placed in the upper part of the reactor
chamber. The solid catalyst was charged in the porous glass
inlet, and a solution of the substrate 1a (50 mg) dissolved in
the modifier was added at the bottom of the reactor. The
autoclave was heated to 50 °C and pressurized with H2 (30 bar)
and then with CO2 to attain a total pressure of 200 bar and a
homogeneous phase in the entire autoclave. This procedure
allows avoiding the contact of the catalyst with the substrate
solution prior to pressurization with CO2. After 18 h, the
autoclave was cooled down to 0 °C and slowly depressurized.
Conversion (1H NMR), IL contamination (19F-NMR), and ee
(GC analysis) were determined (see Table 1).
With methanol as modifier, a low enantioselectivity of up to
62% ee was achieved, and a significant leaching of the IL was
observed correlating with the amount of the solvent used
(Table 1, entries 1 and 2). Thus, the use of methanol as a
modifier is not compatible with the catalyst system, and no
further experiments have been carried out with this solvent.
The reaction performed with DCM as a modifier resulted in a
more rapid hydrogenation (TOF 32 h−1) with high
enantioselectivity (ee 98%). However, also in this case, a
substantial amount of ionic liquid leached into the product
phase, therefore excluding DCM for this application (Table 1,
entry 3). In contrast, toluene showed good results, leading to an
ee of 91% and 85% conversion at a low catalyst loading of ca.
0.1 mol % (Table 1, entry 4). Notably, no IL leaching was
detected by NMR, and the Rh content in the product solution
was below the detection limit of ICP-OES (1 ppm), indicating
an effective catalyst retention. Most gratifyingly, the target
substrate 1b could be hydrogenated using toluene as a modifier
and Rh-Quinaphos@SiO2-[EMIM][NTf2] as a catalyst achiev-
ing a high conversion of 90% within 18 h and an excellent ee >
99% (Table 1, entry 5). Again, no IL-leaching could be detected
under these conditions.
thus fulfilling the basic requirement. The same behavior was
observed at lower substrate concentrations, whereas a saturated
methanol solution (4.7 M) remained biphasic even at a
pressure of up to 360 bar. Using DCM (1 mL) as a modifier,
the system became monophasic at a pressure above 100 bar if
the concentration was ≤0.25 M. At higher concentrations of 1a,
DCM formed a single phase with compressed CO2 at pressures
in the range of 200 bar, but the enamide precipitated. The best
compromise between substrate concentration and processing
window was obtained when toluene was used as a modifier. At
an initial volume of 0.5 mL of a saturated solution of 1a in
toluene (0.25 M), the system became monophasic at a pressure
of 85 bar, but a precipitation of 1a was observed at a pressure of
110 bar. When using 1.0 mL of a 0.25 M solution, this
phenomenon was not observed and the system stayed
monophasic up to a pressure of 200 bar, defining a broad
operating window. The maximum solubility of the industrial
case study substrate 1b in toluene is 0.2 mol L−1 and thus in the
same range as that of 1a. The phase behavior was tested for 1.0
and 2.0 mL of a 0.18 M solution of 1b in toluene. Both systems
became monophasic at around 100 bar, and neither
precipitation nor phase separation occurred up to a pressure
of 200 bar.
In the next set of experiments, the solvents toluene, DCM,
and methanol were tested as modifier in the batch-wise
asymmetric hydrogenation of 1 (Scheme 2) under conditions
ensuring homogeneity of the modCO2 phase according to
Figure 2. As molecular catalyst in the SILP system, we chose
Rh-Quinaphos,33,34 a very effective catalyst for highly
enantioselective transformations, including the asymmetric
hydrogenation of enamides.34 This catalyst has already been
successfully immobilized using the SILP strategy for highly
enantioselective asymmetric hydrogenation under a continuous
flow with pure scCO2 as a mobile phase.17,18
Building up on these promising results, a series of continuous
flow experiments was carried out using a computer-controlled
in-house built setup.35 A stainless steel tube with a volume of 8
mL (diameter = 8.5 mm; length = 14 cm) containing the SILP-
catalyst material (2.07 g, 3.3 μmol) was used as the reactor, and
The SILP catalyst Rh-Quinaphos@SiO2-[EMIM][NTf2] was
prepared according to a literature procedure.17 Dehydroxylated
SiO2 was used as support material, dried and degassed 1-ethyl-
3-methylimidazolium bis(trifluoromethylsulfonyl)imide
([EMIM][NTf2]) as IL, and [Rh(cod){(Sa,Rc)-1-Naph-
Quinaphos}][NTf2] or [Rh(cod){(Ra,Sc)-1-Naph-
Quinaphos}][NTf2]34 as catalyst precursor (Scheme 1). The
immobilization was done via wet impregnation with DCM as a
̇
the flow rates were adjusted as follows: V(tol) = 0.037 mL
−1
−1
−1
̇
̇
min , V(H2) = 18 mLN min , and V(CO2) = 60 mLN min .
The temperature was set to 50 °C and the (total) pressure to
150 bar. Before starting the experiment, the system was flushed
3299
DOI: 10.1021/acscatal.8b00216
ACS Catal. 2018, 8, 3297−3303

Evaluation Only. Created with Aspose.PDF. Copyright 2002-2021 Aspose Pty Ltd.
ACS Catalysis
Research Article
a
Table 1. Batch Hydrogenation of 1 with Rh-Quinaphos@SiO2-[EMIM][NTf2] in modCO2
b
entry
modifier (mL)
ptotal bar
sub
sub/cat mol/mol
conv. %
TOF h−1
ee %
IL leaching %
1
2
3
4
5
methanol (0.4)
methanol (1.0)
DCM (0.8)
200
200
150
150
150
1a
1a
1a
1a
1b
350
350
925
925
35
81
62
85
90
7
31
62
16
39
34
<1
<1
16
32
44
46
98
toluene (1.8)
toluene (1.8)
91
925
>99
a
b
1
Reaction conditions: 1 = 0.31 mmol, T = 50 °C, pH2 = 30 bar, t = 18 h. Determined via H and 19F NMR.
with CO2 for removing oxygen and water from the setup. The
initial conversion of 74% rapidly decreased to 12% after 24 h
(see Table S1 and Figure S11). Notably, the enantioselectivity
remained constant at 98% ee during the experiment. However,
significant IL and Rh leaching was observed under these
conditions, indicating that the composition of the mobile phase
was unsuitable for sufficient IL and catalyst retention.
second continuous flow experiment under the same conditions
as before. Indeed, this simple procedure allowed to significantly
enhance the catalyst lifetime. The initial almost quantitative
conversion declined slowly, reaching 84% after 90 h and 68%
after 120 h on stream, while the ee stayed stable above 99% for
the entire experiment (see Figure 4 and Table S3). A tTON of
Thus, in a following experiment, the properties of the mobile
phase were adjusted by increasing the CO2 flow to 80 mLN
min−1 under otherwise identical conditions corresponding to an
enhancement of the mCO /mtol ratio from 3.64 to 4.86. Very
2
good results were obtained in the first 20 h on stream with a
single pass conversion above 90% and enantioselectivities of
98% ee. However, the performance of the catalyst could not be
stabilized, and a significant reduction of activity was observed
with conversions dropping to 48% after 90 h on stream. Again,
the enantioselectivity remained constant at 98% ee during the
entire experiment, and a total turnover number (tTON) of
8000 with an average STY of 16 g L−1 h−1 was obtained (see
Figure 3 and Table S2). Notably, the Rh content in the
Figure 4. Continuous hydrogenation of 1b using Rh-Quinaphos@
SiO2-[EMIM][NTf2] as catalyst. The setup was evacuated for 3 days
prior to the experiment. Conditions: T = 50 °C, ptotal = 150 bar,
−1
̇
̇
V(substrate) = 0.037 mL min (c(1b) = 0.18 M in toluene), V(H2) =
18 mLN min , V(CO2) = 80 mLN min−1, n(catalyst) = 3.3 μmol.
−1
̇
13 000 was achieved, and the average STY could be improved
to 21 g L−1h−1. The Rh content in the product samples was
again below the ICP-OES detection limit of 1 ppm, indicating a
negligible leaching of the catalyst.
To improve the catalyst’s stability further, two additional
measures were taken. First, a predrying column containing 3 Å
molecular sieves was installed upstream in front of the
reactor35,17 to remove water traces from the feed and thus
preventing water accumulation in the system. Second, instead
of dehydroxylated silica, highly hydrophobic fluorous reverse-
phase silica (FRP-SiO2) was used as support for the SILP
catalyst.17 FRP-SiO2 is obtained by modifying the surface of
silica with triethoxyperfluorooctylsilane.36,37 The combination
of these two measures with the evacuation of the setup prior to
the experiment led to excellent results in the hydrogenation of
1b (Figure 5, Table S4).
Single pass conversions above 99% and 99% ee were achieved
during the entire 90-h-long experiment. After 28 h, the feed-
flow rates were doubled for the next 15 h while still obtaining
full conversion, thus indicating that the upper limit for the
catalyst activity was not reached yet. Returning to the original
flow rates restored the original performance completely, which
was retained fully over the entire period of investigation. Thus,
a tTON of 10 300 was obtained within 90 h with a maximum
STY of 24 g L−1 h−1.38
Figure 3. Continuous hydrogenation of 1b using Rh-Quinaphos@
SiO2-[EMIM][NTf2] as catalyst. Conditions: T = 50 °C, ptotal = 150
−1
̇
bar, V(substrate) = 0.037 mL min (c(1b) = 0.18 M in toluene),
V(H2) = 18 mLN min , V(CO2) = 80 mLN min−1, n(catalyst) = 3.3
−1
̇
̇
μmol.
collected product samples was below the detection limit of
ICP-OES of 1 ppm, indicating that a suitable mobile phase
composition was used. Thus, the drop in conversion cannot be
ascribed to catalyst or IL leaching but rather suggests a
chemical deactivation of the catalyst decomposing to a
nonproductive species.
Previous investigations have shown that traces of water in the
continuous flow setup may eventually lead to deactivation of
this type for the Rh-Quinaphos catalyst in SILP-type
systems.17,35 Hence, to remove any residual water from the
system, the setup was evacuated for 3 days before running a
3300
DOI: 10.1021/acscatal.8b00216
ACS Catal. 2018, 8, 3297−3303