Kinetic study of the asymmetric hydrogenation of methyl acetoacetate in the presence of a ruthenium binaphthophosphepine complex

Article abstract of DOI:10.1002/adsc.200800531

The asymmetric hydrogenation of methyl acetoacetate (MAA) in methanol using dibromobis{( S)-4-phenyl-4,5-dihydro-3H-dinaphtho[2,1-c: 1′,2′-e] phosphepine}-ruthenium was studied in detail. For the determination of the reaction network, data from kinetic experiments were compared to different possible reaction networks using the kinetic software Presto Kinetics. The simulation was optimised to describe the reaction accurately with a minimal set of process parameters and reaction equations. For the best model the reaction orders, collision factors and activation energy of all reaction steps were determined. Additionally, the influence of reaction temperature and hydrogen pressure on the enantiomeric excess (ee) of the reaction was studied. It was found that high reaction temperatures and high hydrogen pressures result in increasing enantioselectivities.

Full text of DOI:10.1002/adsc.200800531

FULL PAPERS
DOI: 10.1002/adsc.200800531
Kinetic Study of the Asymmetric Hydrogenation of Methyl
Acetoacetate in the Presence of a Ruthenium
Binaphthophosphepine Complex
a
a
b
b,
Eva ꢀchsner, Bastian Etzold, Kathrin Junge, Matthias Beller, *
a,
and Peter Wasserscheid *
Lehrstuhl fꢁr Chemische Reaktionstechnik, Universitꢂt Erlangen-Nꢁrnberg, Egerlandstr. 3, 91058 Erlangen, Germany
a
Fax : (+49)-9131-852-7421; e-mail: Peter.Wasserscheid@crt.cbi.uni-erlangen.de
Leibniz-Institut fꢁr Katalyse e.V. an der Universitꢂt Rostock, Albert-Einstein-Str.29a, 18059 Rostock, Germany
b
Fax : (+49) 381-1281-5000; e-mail: Matthias.Beller@catalysis.de
Received: August 25, 2008; Revised: October 31, 2008; Published online: December 19, 2008
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adsc.200800531.
Abstract: The asymmetric hydrogenation of methyl activation energy of all reaction steps were deter-
acetoacetate (MAA) in methanol using dibromo- mined. Additionally, the influence of reaction tem-
bis{(S)-4-phenyl-4,5-dihydro-3H-dinaphtho[2,1-c:
perature and hydrogen pressure on the enantiomeric
1’,2’-e]phosphepine}-ruthenium was studied in detail. excess (ee) of the reaction was studied. It was found
For the determination of the reaction network, data that high reaction temperatures and high hydrogen
from kinetic experiments were compared to different pressures result in increasing enantioselectivities.
possible reaction networks using the kinetic software
Presto Kinetics. The simulation was optimised to de-
scribe the reaction accurately with a minimal set of Keywords: asymmetric catalysis; asymmetric hydro-
process parameters and reaction equations. For the genation; hydrogenation; keto esters; ketones; kinet-
best model the reaction orders, collision factors and ic study
Introduction
allylic alcohols as well as ketones could be hydrogen-
ated with high selectivity and yield. In 2000, several
Today, asymmetric hydrogenation plays an important monodentate phosphorus ligands with a 2,2’-binaph-
[
9,10,11]
role in the production of chiral compounds and inter- thol core were introduced.
The easier prepara-
mediates. In particular, asymmetric hydrogenations tion of these monodentate ligands proved to be an es-
are applied for the commercial manufacture of phar- sential advantage compared to most known bidentate
[
12]
maceuticals, for example, naproxen, vitamin E, b- ligands.
[
1,2]
lactam antibiotics and adrenaline.
Moreover, asym-
Surprisingly, for asymmetric hydrogenation reac-
metric hydrogenation processes are also used in the tions detailed kinetic and thermodynamic studies are
synthesis of perfumery ingredients, intermediates of rarely found in literature. However, for a deeper un-
fine chemicals and for the development of new mate- derstanding of the respective reaction it is essential to
rials like ferro-electric liquid crystals and biodegrad- describe its properties with kinetic and thermodynam-
[1,3]
able polymers.
ic parameters, in particular if more than one reaction
In the past the development of asymmetric hydro- mechanism is principally possible. In such case a de-
genation was significantly based on the synthesis of tailed kinetic analysis can help to identify the most
[
4,5,6,7]
novel catalysts and suitable chiral ligands.
A par- appropriate mechanistic model as the latter describes
ticularly successful type of chiral ligand is the axially the experimental results with the smallest set of inde-
dissymmetric bisphosphine ligand BINAP which has pendent parameters and in the most accurate way.
[
8]
been developed in the early 1980s. Using this new
In this study kinetics models are used to deduce
ruthenium/BINAP complexes a wide range of func- possible reaction pathways for the asymmetric hydro-
tionalised olefins ranging from a-arylacrylic acids genation of methyl acetoacetate (MAA). Therefore, a
through a,b- and b,g-unsaturated carboxylic acid to large number of hydrogenation experiments have
Adv. Synth. Catal. 2009, 351, 235 – 245
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
235

Eva ꢀchsner et al.
FULL PAPERS
been carried out and kinetic models have been de- asymmetric hydrogenation of MAA with phosphorus
rived from the respective concentration/time curves at ligands has been reported to show the highest activity
different temperatures and hydrogen pressures. We and enantioselectivity in alcoholic solvents, in particu-
[
13,14]
want to introduce this approach as a valuable and fast lar in methanol.
Therefore, we used methanol as
method for the discrimination of different reaction the solvent for our kinetic studies.
pathways. It is noteworthy, however, that our study
If the hydrogenation reaction of MAA is per-
builds uniquely on the analysis of reaction products formed in methanol, ketal and hemiketal intermedi-
detectable by GC from the reaction mixture. There- ates are formed. Obviously, these parallel reactions
fore we cannot exclude that short-living intermediates are reversible equilibrium reactions. Figure 2 shows
play a role in the molecular mechanism of the reac- the formal reaction scheme as it was known from the
[
15]
tion.
literature when we started the detailed kinetic stud-
In this contribution, the kinetics of the hydrogena- ies described in this publication.
tion of methyl acetoacetate (MAA) to R/S-methyl hy-
Note that the reported formal reaction scheme
droxybutyrate (MHB) are described in detail. In our shows the formation of ketal and hemiketal com-
study dibromobis[(S)-4-phenyl-4,5-dihydro-3H-di- pounds, but no hydrogenation of these compounds.
naphtho[2,1-c:1’,2’-e]phosphepine]ruthenium is used Hence, a major aim of this study was to check the
as the homogeneous catalyst (Figure 1). In 2005 this proposed reaction network of the hydrogenation reac-
[
12a]
Ru complex was first published by some of us
and tion in methanol by comparison of the experimental
the ligand has been used in various asymmetric hy- kinetic results with several possible kinetic models.
[12b–e]
drogenations.
For the most suitable model, reaction orders, collision
In most catalytic asymmetric hydrogenations, alco- factors and the activation energy have been deter-
hols and halogen-containing hydrocarbons like mined for all reaction steps.
CH Cl are applied as solvents. As an example, the
2
2
Figure 1. Hydrogenation of MAA using dibromobis[(S)-4-phenyl-4,5-dihydro-3H-dinaphtho[2,1-c:1’,2’-e]phosphepine]ruthe-
nium as catalyst.
[15]
Figure 2. Formal reaction scheme of MAA hydrogenation similar to ref.
236
asc.wiley-vch.de
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 235 – 245

Kinetic Study of the Asymmetric Hydrogenation of Methyl Acetoacetate
FULL PAPERS
Results and Discussion
modelling, equations for formation and consumption
of all reactants [Eq. (1)] and power law rate expres-
sions [Eq. (2)] for all reaction steps were assumed
Determination of the Reaction Network
(
for details see section “Estimation of Kinetic Param-
Figure 3 shows the concentration profiles of all liquid eters” below).
reactants during a typical hydrogenation experiment
(
more experimental data including reproduction runs
and hydrogenation runs at different temperatures and
pressures can be found in the Supporting Informa-
tion). The results of the kinetic modelling (discussed
in detail in the section “Estimation of Kinetic Param-
eters” below) are indicated as dotted lines.
In the first 20 min of the experiment a fast decrease
of MAA is observed which is obviously not due to
MAA hydrogenation but is caused by the rapid for-
mation of the ketal and hemiketal intermediates.
For solving the differential equations, the following
During the course of the reaction, the concentrations simplifications and assumption were made. All experi-
of ketal and hemiketal intermediates run through a ments were carried out with the same catalyst concen-
maxima (tmax,ketal =20 min, tmax,hemiketal =100 min) while tration. Therefore the catalyst concentration was not
the formation of the MHB product increases in an accounted for in the modelling. The concentration of
almost linear fashion over reaction time.
the solvent methanol was in excess to the substrate
Apparently, these results agree in principle with the MAA and remained quasi constant during reaction.
formal reaction scheme in Figure 2. However, we The concentration of water during the reaction is un-
were interested to check whether this reported reac- known. Only water-free solvent and reactants were
tion scheme would indeed give the best description of utilised, but water produced during the reaction could
the observed concentration vs. time curves or whether not be analysed. Thus, no water-dependent reaction
alternative models of the reaction network would rates were derived. The hydrogen solubility was as-
allow for a better agreement with the experimental sumed to change only marginally during the experi-
kinetic data. Therefore alternative reaction networks ments. Consequently, the hydrogen partial pressure
have been tested in the kinetic modelling with the was utilised directly in the reaction rate equations in-
aim to identify the model that describes with the stead of the hydrogen concentration.
lowest number of parameters the experimental results
in the most accurate manner.
Ketal and hemiketal formation are strongly influ-
enced by thermodynamics. Therefore, the equilibrium
All modelling work has been carried out using the constants need to be known for an accurate descrip-
software Presto Kinetics (CiT GmbH, Germany, ver- tion of the kinetics. With these constants [determined
sion 7.2.1). To obtain the differential equations for according to Eq. (3)] the reaction rate constant of the
Figure 3. Concentration profiles of all liquid reactants during the MAA hydrogenation in methanol. Reaction conditions: T=
8
08C, p(H )=10 bar, stirrer speed=1200 rpm, S/C=259.
2
Adv. Synth. Catal. 2009, 351, 235 – 245
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
237

Eva ꢀchsner et al.
FULL PAPERS
minimisation of the least squares objective function of
absolute values by the damped Gauss–Newton algo-
rithm) is determined for all tested models. The model
with the lowest residual fits best if the number of in-
dependent parameters in all different models is com-
reversible reactions was calculated [see Eqs. (10)–(15) parable.
in section “Estimation of Kinetic Parameters” below]. As a first benchmark, the proposed reaction
In the studied temperature region (60–1208C), the scheme was adapted for its utilisation as a kinetic
[
15]
0
standard
reaction
enthalpy
(D H
=
reaction network. In this adaption, all reaction steps
R
i MAA–K
À1
À1
0
À22.96 kJmol ,
D H
=42.75 kJmol , including experimentally non-detectable intermedi-
R
i K–HK
0
À1
D H
=À40.01 kJmol )
and
entropy ates (namely Int 1 and Int 2 from Figure 2) were
0
R
i MAA–HK
0
À1 À1
(
D S
=À57.05 Jmol K ,
D S
R
=
omitted. Consequently, only the hydrogenation of
R
i MAA–K
i K–HK
À1 À1
0
À1 À1
113 Jmol K , D S
=114.4 Jmol K ) are MAA and the parallel equilibrium reactions of MAA
R
i MAA–HK
assumed to be constant. The values have been derived to K, K to HK and MAA to HK are considered in
by fitting Eq. (3) to the measured equilibrium con- model 1 (Figure 4 left).
stants. The measured equilibrium constants were de-
The residual (representing the degree of fit be-
termined from concentration vs. reaction time plots. It tween the simulated and the measured data) was
is assumed that equilibrium is reached when the found to be 0.296 for model 1. As a first variation of
ratios over the reaction time converge to a final this model, we tested the elimination of the reversible
value. This procedure is possible as the reversible re- reaction between MAA and HK in model 2. Interest-
actions are faster than the hydrogenation reactions.
ingly, this variation did not result in a significant
Table 1 shows the calculated equilibrium constants change of the model fit, the same residual of 0.296
for the reaction of MAA to hemiketal, from MAA to was obtained. Consequently, to fulfil the aim of a ki-
ketal and from ketal to hemiketal intermediate.
netic description with a minimum of parameters in a
At this point, all basic equations and settings for more formalistic manner, this reaction step should be
the modelling are defined and the modelling results omitted from the kinetic model. However, a reasona-
for the different reaction network can be compared to ble molecular mechanistic view on the reaction would
the experimental data. For this comparison, the resid- always require to consider the MAA–HK equilibrium
ual (given from the parameter estimation done by the reaction step as the direct formation of ketal is diffi-
cult to imagine without a hemiketal intermediate. For
this reason and in the light of the identical fit of the
Table 1. Calculated equilibrium constants for the ketal/hemi- two models, we decided to include the MAA–HK
[a]
ketal formation of MAA in methanol.
equilibrium in all further models.
A more detailed comparison of the experimental
and simulated results for models 1 and 2 revealed
that the equilibrium reactions are described accurate-
ly, but the concentration of the product MHB is simu-
lated systematically wrong for high pressures and
temperatures. This led us to the assumption that the
hydrogenation reaction to MHB differs from models
T [8C]
KMAA–K
KK–HK
KMAA–HK
6
8
9
1
1
0
0
0
00
20
4.2
2.6
2.1
1.7
1.2
0.2
0.4
0.6
0.8
1.7
2.0
0.9
0.6
0.4
0.2
[
a]
1
and 2. Hence, also models that do not include a
Reaction conditions: T=60–1208C, p(H )=5–20 bar, stir-
2
rer speed=1200 rpm, S/C=259.
direct hydrogenation of MAA have been tested
Figure 4. Reaction network models for the hydrogenation of MAA: model 1 (left) and model 2 (right).
238
asc.wiley-vch.de
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 235 – 245

Kinetic Study of the Asymmetric Hydrogenation of Methyl Acetoacetate
FULL PAPERS
Figure 5. Reaction networks for the hydrogenation of MAA: model 3 (left) and model 4 (right).
(
(
Figure 5). Hydrogenation of the ketal intermediate data. The residual of model 5 was determined as
model 3) and hydrogenation of the hemiketal inter- 0.257, the calculations based on model 6 resulted in a
mediate (model 4) were assumed to be the only residual of 0.215. To illustrate the difference in the fit
MHB-forming steps, respectively.
quality between the formal reaction network similar
[
15]
Both models show high deviation from experimen- to the literature
(model 1) and the expanded
tal data as the residual of model 3 was determined to models tested in this work, Figure 7 shows the parity
.332 and the residual of model 4 to 0.484. These re- plots of the MHB formation for models 1 and 6 in
0
sults correspond to literature reports, in which the comparison. All parity plots based on model 6 show
asymmetric hydrogenation of MAA with similar li- the best agreement of experimental and modelled
gands has also been described in a non-alcoholic sol- data for the reactants. It is noteworthy that the addi-
[
13]
vent, such as dichloromethane.
The simulations using models 3 and 4 revealed that intermediate in model 6 offered no improvement.
the experimental concentration profiles can in princi- As a result of our kinetic studies and the compari-
tional consideration of a hydrogenation of the ketal
ple also be fitted by hydrogenation reactions from the son of the experimental data with different kinetic
ketal or hemiketal intermediates to MHB, but in models we conclude that the literature proposed reac-
these cases the results of the temperature and hydro- tion network for the hydrogenation of MAA in meth-
[
15]
gen pressure variation is not represented well. This in- anol with Ru-BINAP complexes
differs from our
dicates that these reactions should not be ignored Ru-phenylphosphepine-catalysed reaction network.
completely, and should be taken into consideration The here proposed reaction network of model 6
besides the direct hydrogenation of MAA.
shows a much better fit to the experimental data and
Consequently, this led to models 5 and 6 (Figure 6) explains easily the experimental fact that the asym-
both containing the direct hydrogenation from MAA metric hydrogenation of MAA is significantly acceler-
to MHB. Model 5 includes additionally the hydroge- ated in alcoholic solvents. The additional hydrogena-
nation reaction from the ketal intermediate to MHB tion step of hemiketal intermediate – which is only
while model 6 accounts for the additional hydrogena- formed in alcoholic solvents – opens an additional
tion of the hemiketal intermediate to MHB.
pathway to the MHB formation and is necessary for
Both networks showed – compared to models 1 to the accurate kinetic description.
– a clearly improved fit to the experimental kinetic
4
Figure 6. Reaction networks for the hydrogenation of MAA: model 5 (left) and model 6 (right).
Adv. Synth. Catal. 2009, 351, 235 – 245
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
239

Eva ꢀchsner et al.
FULL PAPERS
Figure 7. Parity plots for MHB (exemplary) based on model 1 and 6. Reaction conditions: T=60–1208C, p(H )=5–20 bar,
2
stirrer speed=1200 rpm, S/C=259.
Estimation of Kinetic Parameters
For all six models kinetic parameters were fitted to
calculate the residual of all models. In this section
only the parameters for the most appropriate reaction
model 6 are discussed. Figure 8 shows the simplified
reaction network that has been generated from model
6.
For the kinetic modelling, the following equations
for the formation and consumption of all liquid reac-
tants were applied.
The individual reaction rates for the hydrogenation
steps are given by Eqs. (8) and (9). The reaction rates
for the equilibrium reaction are defined according to
Eqs. (10)-(15).
To determine the reaction order of MAA, ketal in-
termediate K and hemiketal intermediate HK, experi-
ments with constant hydrogen pressure have been car-
ried out. This guaranteed the reaction rate to be unaf-
fected by the availability of hydrogen in this set of ex-
periments. The orders of all organic reactants were
determined by the integral method, which involves fit-
ting of the resulting concentration vs. time plots. As a
result, the reaction orders of all organic liquid reac-
tants (MAA, K and HK) were found to be 1. Similar
data have been obtained in the literature for the
Figure 8. Reaction network developed from model 6 for the
MAA hydrogenation in methanol.
240
asc.wiley-vch.de
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 235 – 245

Kinetic Study of the Asymmetric Hydrogenation of Methyl Acetoacetate
FULL PAPERS
asymmetric hydrogenation of unsaturated carboxylic erolytic split of hydrogen by the ruthenium-olefin
[
16]
acids with Ru-BINAP systems (reaction order=1)
complex is the rate-determining step (rds). At higher
and for the asymmetric hydrogenation of tiglic acid hydrogen pressure, however, the reaction of the
with Ru-BINAP systems (reaction order=0.881– ruthenium complex with olefin becomes rate-deter-
[
17]
[17]
0.929).
mining and the reaction is zero order in hydrogen.
The activation energies and collision factors have
In another set of experiments the reaction order
with respect to hydrogen was derived experimentally been determined from the experiments at different
by varying the hydrogen pressure from 5 to 20 bar. temperatures in the range between 608C and 1208C.
The variation results in a linear relation between the In Arrhenius plots, which are shown in Figure 10 and
rate coefficients and hydrogen pressure (Figure 9). Figure 11, reaction coefficients were plotted against
Thus a first order with respect to hydrogen pressure different reaction temperatures for all reaction path-
was determined for all hydrogenation reactions. These ways. The activation energy of the reaction has been
results are in agreement with the work of Ashby calculated from the slope of the Arrhenius plot and
[16]
[17]
et al.
and Dong et al.
They described Ru-cata- the collision factor has been determined from the in-
lysed hydrogenation reactions to be of first order in tersection with the y-axis. The resulting values are
hydrogen at low hydrogen pressures because the het- summarised with all kinetic parameters in Table 2.
Figure 9. Reaction rates of hydrogenation steps (MAA–
MHB and HK–MHB) as a function of different hydrogen
Figure 10. Arrhenius plots for different hydrogenation steps
in the hydrogenation of MAA. Reaction conditions: T=60–
pressures. Reaction conditions: T=808C, p(H )=5–20 bar,
2
stirrer speed=1200 rpm, S/C=259.
1208C, p(H )=5–20 bar, stirrer speed=1200 rpm, S/C=259.
2
Adv. Synth. Catal. 2009, 351, 235 – 245
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
241

Eva ꢀchsner et al.
FULL PAPERS
À1
(
À721.3 kJ·mol ) were determined out of the com-
bustion enthalpies. The formation enthalpy of hydro-
À1
[19]
gen (À0.379 kJ·mol ) was found in literature. The
reaction enthalpy was calculated from the enthalpies
of formation of every single reactant.
The reaction enthalpy was determined to
À1
À63.2 kJ·mol indicating a moderately exothermic re-
action.
Influence of Reaction Temperature and Pressure on
Figure 11. Arrhenius plot of the ketal and hemiketal forma- the Enantiomeric Excess
tion reaction in the hydrogenation of MAA in methanol.
Reaction conditions: T=60–1208C, p(H )=5–20 bar, stirrer
speed=1200 rpm, S/C=259.
2
The effect of the reaction temperature on the enantio-
selectivity was investigated in the range of 608C–
208C. Figure 12 shows the ee profile and the average
1
The activation energies for the hydrogenation reac- ee (ee ) over the studied temperature range.
av
À1
tions are in a range of 74.8–85.6 kJ·mol . This range
Remarkably, the ee is strongly dependent on the re-
corresponds quite well to published literature data for action temperature with higher temperatures leading
other hydrogenation reactions with Ru complexes. to significantly higher enantiomeric excess (average
For example, the group of Dong studied the asymmet- ee of 81.3% at 608C vs. 95.0% at 1208C). Within the
ric hydrogenation of tiglic acid catalysed by Ru- reaction time the ee increases at the beginning of the
BINAP and Ru-(p-OCF )-BINAP systems and deter- reaction in the temperature range between 608C and
3
mined
activation
energies
between
80.1– 1008C and remains constant after some kind of “acti-
À1 [17]
82.7 kJ·mol .
Mahfud et al. calculated the activa- vation period”. This is different at 1208C where the
tion energy of the hydrogenation of 1-hydroxy-2- ee shows an almost constant value over the whole re-
propanone with RuCl (PPh ) to be
81Æ action time. A possible explanation for the increasing
The activation energies of the ketal ee in the early stages of the reaction at lower reaction
and hemiketal formation reactions were found to temperature is the slow formation of the most selec-
AHCTUNGTRENNUNG
2
3 3
À1 [18]
1
7 kJ·mol .
À1
[12]
show lower values (41.4–68.8 kJ·mol ) compared to tive catalyst species. In the literature the unsteady
the hydrogenation reactions. ee of Ru complexes of binaphthophosphepine ligands
We also determined the reaction enthalpy for the is explained by the fact that the most selective cata-
total reaction from MAA to MHB. For this purpose lyst species only forms after addition of the substrate.
the combustion enthalpies of MAA and MHB were From our results we conclude that this formation pro-
measured in a calorimeter. The formation enthalpies cess occurs much faster at 1208C and therefore an
À1
of
MAA
(À657.7 kJ·mol )
and
MHB
Table 2. Collision factors and activation energies for the hydrogenation of MAA in methanol assuming a power law model.
Variable
Value
ACHTUNGTRENNUNG[ Unit]
0
1
2
5
1
À1 À1
k0, MAA–MHB
k0, HK–MHB
k0, MAA–K
k0, K–HK
6.01ꢄ10
5.15ꢄ10
3.28ꢄ10
2.61ꢄ10
2.13ꢄ10
74.8
85.6
41.4
68.8
48.3
Pa ·s
À1 À1
Pa ·s
À1
s
s
s
À1
À1
k0
, HKÀMAA
À1
À1
À1
À1
À1
EA, MAA–MHB
EA, HK–MHB
EA, MAA–K
EA, K–HK
kJ·mol
kJ·mol
kJ·mol
kJ·mol
kJ·mol
EA, HK–MAA
242
asc.wiley-vch.de
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 235 – 245

Kinetic Study of the Asymmetric Hydrogenation of Methyl Acetoacetate
FULL PAPERS
Figure 12. Effect of reaction temperature on enantioselectiv-
ity for the MAA hydrogenation in methanol. Reaction con-
Figure 13. Effect of hydrogen pressure on the enantioselec-
tivity of the hydrogenation of MAA in methanol. Reaction
ditions: T=60–1208C, p(H )=10 bar, stirrer speed=
conditions: T=808C, p(H
1200 rpm, S/C=259.
2
)=5–20 bar, stirrer speed=
2
1200 rpm, S/C=259.
almost constant value of 95% ee is observed over the thoxy-2-naphthyl)propenoic acid in methanol decreas-
whole reaction time. ing enantioselectivities with lower pressure in the
An additional explanation for the improved ee at pressure range below 30 bar.
high temperatures may arise from the mechanistic
considerations discussed earlier in this paper. As it
can be seen from Table 1, a high reaction temperature Conclusions
shifts significantly the equilibrium from MAA to
ketal and hemiketal intermediates. This results in a In summary, detailed kinetic studies of the asymmet-
higher concentration of ketal and hemiketal inter- ric hydrogenation of methyl acetoacetate (MAA)
mediates and favours the hydrogenation of the hemi- using a dibromobis[(S)-4-phenyl-4,5-dihydro-3H-di-
ketal intermediate. Also the higher activation energy naphtho[2,1-c:1’,2’-e]phosphepine]ruthenium catalyst
À1
of the hemiketal hydrogenation (85.6 kJ·mol ) com- in methanol have been reported. The reaction pro-
À1
pared to the hydrogenation of MAA (74.8 kJ·mol ) ceeds via a reaction network consisting of hydrogena-
prefers the hemiketal hydrogenation at high reaction tion steps and the reversible formation of hemiketal
temperatures. As higher temperature favours the for- and ketal intermediates. Six possible reaction network
mation of hemiketal and leads to significantly higher models have been designed and tested for their suita-
ee we conclude that the hydrogenation of the hemike- bility to describe this asymmetric hydrogenation in
tal leads to a higher enantioselectivity compared to the most accurate way using the lowest number of in-
the direct hydrogenation of MAA.
dependent parameters. By comparing the experimen-
Finally, the influence of hydrogen pressure on the tal kinetic data with fitted model data – determined
enantioselectivity was examined in the pressure range by the kinetic software Presto – the best modelling re-
from 5 to 20 bar (Figure 13). Interestingly, an increas- sults were obtained for model 6 (residual=0.215)
ing hydrogen pressure resulted in clearly higher enan- which includes hydrogenation steps from methyl ace-
tioselectivities. At 808C, higher hydrogen pressures toacetate (MAA) to methyl hydroxybutyrate (MHB)
also led to a faster increase of ee in the course of the and from the hemiketal intermediate to MHB. For
reaction. While at 20 bar hydrogen pressure an almost this network all kinetic parameters were determined.
constant ee of 92% was reached after 2400 s reaction Reaction orders of all organic reactant and hydrogen
time, the ee was increasing steadily over 8000 s reac- were found to be 1. The activation energy of the hy-
tion time for the experiments at 5 bar and 10 bar drogenation reaction from MAA to MHB is in the
À1
reaching a maximum ee of 88.1% (5 bar H ) and range of 74.8–85.6 kJ·mol .
2
9
0.9% (10 bar H ), respectively.
Additionally, the influence of temperature and hy-
Again, these results are in good agreement with lit- drogen pressure on the enantioselectivity was investi-
erature reports of other Ru-catalysed asymmetric hy- gated. It was shown that reaction conditions favouring
2
[
20]
drogenation reactions. Combes et al.
reported for the hydrogenation of the hemiketal intermediate (i.e.,
the Ru/BINAP-catalysed hydrogenation of 2-(6-me- high temperature influencing both the hemiketal
Adv. Synth. Catal. 2009, 351, 235 – 245
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
243

Eva ꢀchsner et al.
FULL PAPERS
equilibrium and the hemiketal hydrogenation kinet-
ics) led to significantly higher enantioselectivities.
Moreover, a higher reaction temperature resulted in
faster formation of the most selective catalyst species
with almost constant ee of 95% over the whole reac-
tion time for a reaction temperature of 1208C. A ben-
eficial effect of higher hydrogen partial pressures was
revealed in the pressure range of 5 to 20 bar.
Experimental Section
Chemicals
All chemicals applied in this study were commercially avail-
able and were used without further purification steps. MAA
(
(
99%) was purchased from Aldrich, hydrobromic acid
48%) from Fluka. Commercial dry methanol and acetone
from Acros (both 99.9%, water content less than 50 ppm),
bis(2-methoxyethyl) ether from Acros (99%, used as inter-
nal GC-standard) and [bis(2-methylallyl)(1,5-cyclooctadie-
ne)ruthenium(II)] from Acros were used. (S)-4-Phenyl-4,5-
dihydro-3H-dinaphtho[2,1-c:1’,2’-e]phosphepine was synthes-
[12]
Figure 14. Experimetal set-up as used in the catalytic hydro-
genation experiments: (1) gas entrainment stirrer, (2)
sample line with frit, (3) cooling coil with cooling water, (4)
thermocouple.
ised according to the literature.
Catalyst Preparation
The catalyst was prepared in close analogy to former re-
[
12]
ports.
A mixture of one equivalent of [bis(2-methylall-
ples were filtered with syringe filters and were analysed by
gas chromatography (Varian 3900 equipped with a Lipodex
E column).
yl)(1,5-cyclooctadiene)ruthenium(II)] and two equivalents
of (S)-4-phenyl-4,5-dihydro-3H-dinaphtho[2,1-c:1’,2’-e]phos-
phepine was placed in a Schlenk flask under argon atmos-
phere. Dry acetone (5 mL) and a solution (0.29 molar) of
hydrobromic acid (0.33 mL) in methanol were added and
the solution was stirred 30 min at room temperature. This
solution was heated to 308C in an oil bath and the solvent
was removed under high vacuum for 1 h. After this proce-
dure, the catalyst was obtained as a brown solid powder.
For the kinetic experiments, reaction temperatures (60–
1
208C) and hydrogen pressures (5–20 bar) were varied at a
constant catalyst concentration (S/C=259).
To determine the heat of formation (DH ) of MAA and
R
the hydrogenation product methyl hydroxybutyrate (MHB)
the IKA-calorimeter “C400 adiabatic” was applied.
Catalytic Hydrogenation
All hydrogenation experiments were carried out in a 300-
mL Parr stainless steel autoclave (type 316, see Figure 14) References
with gas entrainment stirrer and glass liner. The reactor was
equipped with an internal cooling coil (cooling agent:
water), a thermocouple (Ni-Cr-Ni), a pressure transducer, a
purge gas valve, a sampling valve, a relief pressure valve, a
gas inlet and a gas outlet.
[1] S. Akutagawa, Appl. Catal. A: General 1995, 128,171–
207.
[2] F. D. Klingler, Acc. Chem. Res. 2007, 40, 1367–1376.
[3] L. A. Saudan, Acc. Chem. Res. 2007, 40, 1309–1319.
[4] W. S. Knowles, M. J. Sabacky, J. Chem. Soc. (D)
1968,1445–1446.
Prior to every hydrogenation experiment, the catalyst was
dissolved in 160 mL methanol and the resulting solution was
stirred for one hour. Afterwards, the catalyst solution was
transferred into the reactor under an argon atmosphere and
MAA and the internal GC-standard bis(2-methoxyethyl)
ether were added. Then the reactor was closed and purged
with argon to remove the air. The autoclave was heated to
the desired reaction temperature and charged with hydro-
gen. The reaction was initiated by starting the gas entrain-
ment stirrer (1200 rpm). Samples were taken periodically
through a sampling tube. The sampling line was completely
flushed before every sampling to avoid falsification of re-
sults. All experiments were semi-batch as the hydrogen pres-
sure was kept constant during reaction. The collected sam-
[5] W. S. Knowles, M. J. Sabacky, B. D. Vineyard, J. Chem.
Soc., Chem. Commun. 1972,10.
[6] W. S. Knowles, Angew. Chem. 2002, 114, 2096–2107;
Angew. Chem. Int. Ed. 2002, 41, 1998–2007.
[7] W. S. Knowles, R. Noyori, Acc. Chem. Res. 2007, 40,
1238–1239.
[8] R. Noyori, Angew. Chem. 2002, 114, 2108–2123;
Angew. Chem. Int. Ed. 2002, 41, 2008–2022.
[9] M. T. Reetz, T. Sell, Tetrahedron Lett. 2000, 41, 6333–
6336.
[10] L. A. Arnold, R. Imobos, A. Manoli, A. H. M. de Vries,
R. Naasz, B. Feringa, Tetrahedron 2000, 56, 2865–2878.
244
asc.wiley-vch.de
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2009, 351, 235 – 245

Kinetic Study of the Asymmetric Hydrogenation of Methyl Acetoacetate
FULL PAPERS
[
11] C. Claver, E. Fernandez, A. Gillon, K. Heslop, D. J.
Hyett, A. Martorell, A. G. Orpen, P. G. Pringle, Chem.
Commun. 2000, 961–962.
dron: Asymmetry 2007, 18, 1288–1298; g) S. Enthaler,
G. Erre, K. Junge, J. Holz, A. Bçrner, E. Alberico, I.
Nieddu, S. Gladiali, M. Beller, Org. Process Res. Dev.
2007, 11, 568–577.
[12] a) B. Hagemann, K. Junge, S. Enthaler, M. Michalik, T.
Riermeier, A. Monsees, M. Beller, Adv.Synth.Catal.
[13] A. Wolfson, I. F. J. Vankelecom, S. Geresh, P. A.
2
005, 347, 1978–1986; b) K. Junge, G. Oehme, A. Mon-
Jacobs, J. Mol. Catal. A 2003,198, 39–45.
[14] A. Wolfson, I. F. J. Vankelecom, S. Geresh, P. A.
Jacobs, J. Mol. Catal. A 2004, 217, 21–26.
sees, T. Riermeier, U. Dingerdissen, M. Beller, Tetrahe-
dron Lett. 2002, 43, 4977–4980; c) K. Junge, G. Oehme,
A. Monsees, T. Riermeier, U. Dingerdissen, M. Beller,
J. Organomet. Chem. 2003, 675, 91–96; d) K. Junge, B.
Hagemann, S. Enthaler, G. Oehme, M. Michalik, A.
Monsees, T. Riermeier, U. Dingerdissen, M. Beller,
Angew. Chem. 2004, 116, 5176–5179; Angew. Chem.
Int. Ed. 2004, 43, 5066–5069; e) B. Hagemann, K.
Junge, S. Enthaler, M. Michalik, T. Riermeier, A. Mon-
sees, M. Beller, Adv. Synth. Catal. 2005, 347, 1978–
[15] L. Bartek, M. Drobek, M. Kuzam, P. Kluson, L. Cerve-
ny, Catal. Commun. 2005, 6, 61–65.
[16] M. T. Ashby, J. Halpern, J. Am. Chem. Soc. 1991, 113,
589–594.
[17] X. Dong, C. Erkey, J. Mol. Catal. A 2004, 211, 73–81.
[18] F. H. Mahfud, F. Ghijsen, H. J. Heeres, J. Mol. Catal. A
2007, 264, 227–236.
[19] http://cea.grc.nasa.gov/index.html, accessed 4.5.08.
[20] G. Combes, E. Coen, F. Dehghani, N. Foster, Ind. Eng.
Chem. Res. 2006, 45, 1281–1290.
1
986; f) S. Enthaler, G. Erre, K. Junge, D. Michalik, A.
Spannenberg, F. Marras, S. Gladiali, M. Beller, Tetrahe-
Adv. Synth. Catal. 2009, 351, 235 – 245
ꢃ 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
245