Mechanistic aspects of hydrogen addition during enantioselective rhodium-catalysed reduction of C=C double bonds with formic acid/triethylamine or molecular hydrogen

Article abstract of DOI:10.1039/b108774f

Deuterium labelling experiments reveal a remarkably different hydrogen addition mode during homogeneously catalysed C=C bond reduction of itaconic acid derivatives 1a-d using molecular hydrogen (hydrogenation) or formic acid/triethylamine (transfer hydrogenation). The expected vicinal addition of two hydrogen atoms across the double bond prevails for all substrates in conventional hydrogenation, whereas the deuterium pattern depends largely on the nature of the carboxyl group in the β (or allylic) position during transfer hydrogenation. Vicinal addition is observed only in case of itaconic acid 1a and α-methylitaconate 1c, while 1,3-addition is preferred with dimethylitaconate 1b and β-methylitaconate 1d. Significant amounts of polydeuterated products are formed also during hydrogenation and transfer hydrogenation. Monitoring the deuterium pattern as a function of time reveals that deuterium scrambling is responsible for polydeuteration, but not for the change of the addition mode. The use of monodeuterated formic acid isotopomers shows that the incorporation from the hydridic formyl position occurs preferentially at the terminal end of the double bond (C-3) whereas the protic hydrogen is directed either in the higher substituted olefinic (C-2) or the methylene (C-1) position. Control experiments using mesaconic (2) and citraconic (3) acids demonstrate that double bond migration in 1a-d is negligible under the reaction conditions. These results are best rationalised on the basis of a common mechanism for hydrogenation and transfer hydrogenation that involves (i) the generation of Rh-H intermediates, (ii) reversible hydride transfer to coordinated substrate to form two isomeric σ-alkyl intermediates, and (iii) irreversible product liberation through protolytic Rh-C cleavage. The key intermediates are similar if not identical for hydrogenation and transfer hydrogenation. The change of the hydrogen transfer pattern can be explained on the basis of the relative rates of the individual steps within the catalytic cycle as compared to the rate of isomerisation of the σ-alkyl intermediates. The Royal Society of Chemistry 2002.

Full text of DOI:10.1039/b108774f

View Article Online / Journal Homepage / Table of Contents for this issue
Mechanistic aspects of hydrogen addition during enantioselective
rhodium-catalysed reduction of C᎐C double bonds with formic
᎐
acid/triethylamine or molecular hydrogen†
Susanne Lange‡ and Walter Leitner*
Max-Planck-Institut für Kohlenforschung, Kaiser-Wilhelm-Platz 1, 45470 Mülheim, Germany.
E-mail: leitner@mpi-muelheim.mpg.de
Received 26th September 2001, Accepted 19th October 2001
First published as an Advance Article on the web 21st January 2002
Deuterium labelling experiments reveal a remarkably different hydrogen addition mode during homogeneously
catalysed C᎐C bond reduction of itaconic acid derivatives 1a–d using molecular hydrogen (hydrogenation) or formic
᎐
acid/triethylamine (transfer hydrogenation). The expected vicinal addition of two hydrogen atoms across the double
bond prevails for all substrates in conventional hydrogenation, whereas the deuterium pattern depends largely on the
nature of the carboxyl group in the β (or allylic) position during transfer hydrogenation. Vicinal addition is observed
only in case of itaconic acid 1a and α-methylitaconate 1c, while 1,3-addition is preferred with dimethylitaconate 1b
and β-methylitaconate 1d. Significant amounts of polydeuterated products are formed also during hydrogenation and
transfer hydrogenation. Monitoring the deuterium pattern as a function of time reveals that deuterium scrambling is
responsible for polydeuteration, but not for the change of the addition mode. The use of monodeuterated formic acid
isotopomers shows that the incorporation from the hydridic formyl position occurs preferentially at the terminal end
of the double bond (C-3) whereas the protic hydrogen is directed either in the higher substituted olefinic (C-2) or the
methylene (C-1) position. Control experiments using mesaconic (2) and citraconic (3) acids demonstrate that double
bond migration in 1a–d is negligible under the reaction conditions. These results are best rationalised on the basis
of a common mechanism for hydrogenation and transfer hydrogenation that involves (i) the generation of Rh–H
intermediates, (ii) reversible hydride transfer to coordinated substrate to form two isomeric σ-alkyl intermediates,
and (iii) irreversible product liberation through protolytic Rh–C cleavage. The key intermediates are similar if not
identical for hydrogenation and transfer hydrogenation. The change of the hydrogen transfer pattern can be
explained on the basis of the relative rates of the individual steps within the catalytic cycle as compared to
the rate of isomerisation of the σ-alkyl intermediates.
formic acid can result from a tandem process comprising
independent hydrogen donor decomposition and subsequent
Introduction
The reduction of C᎐C double bonds is one of the most fund-
᎐
conventional hydrogenation using two different catalytic
systems for the two half reactions.9 The majority of efficient
and especially enantioselective transfer hydrogenations seems to
follow more complex pathways, however, where the same active
centre is responsible for the cleavage of the hydrogen donor and
the transfer of the hydrogen equivalents to the substrate.5–8 The
present study sheds some light on the intimate mechanistic
details of the hydrogen addition step for rhodium-catalysed
transfer hydrogenation using the formic acid/triethylamine
azeotrope as the hydrogen source (HCO2H/NEt3 5 : 2) (Scheme
1). The isomeric unsaturated dicarboxylic acids 1a, 2, and 3
amental synthetic transformations and plays a key role in the
manufacturing of a wide variety of bulk and fine chemicals.
Hydrogenation of olefinic substrates can be achieved readily
with molecular hydrogen in many cases, but transfer hydrogen-
ation methods using suitable donor molecules such as formic
acid or alcohols are receiving increasing attention as possible
synthetic alternatives.1 Transfer hydrogenation is attractive
especially for the small scale production of diverse products as
it requires no special equipment and avoids the handling of
potentially hazardous gaseous hydrogen. Highly stereoselective
methods have emerged from transfer hydrogenation techniques
based on the use of suitable chiral transition metal complexes in
homogeneous solution.2 Especially, chiral rhodium phosphine
complexes have proven extremely effective for both hydrogen-
ation and transfer hydrogenation of C᎐C double bonds.2b,3
᎐
Although the mechanism of rhodium-catalysed homogeneous
hydrogenation of C᎐C bonds with molecular hydrogen has
᎐
been studied in great detail and all key intermediates have been
identified by spectroscopic techniques,4 much less is known
about the corresponding transfer hydrogenation sequence.5–8
In the simplest case, transfer hydrogenation reactions with
† Based on the presentation given at Dalton Discussion No. 4, 10–13th
January 2002, Kloster Banz, Germany.
‡ Current address: Department of Chemistry, University of Leicester,
UK.
Scheme 1 Enantioselective rhodium-catalysed transfer hydrogenation
of isomeric unsaturated acids and esters 1–3. Initially formed esters
are converted to the free acid 5 during hydrolytic work-up (see
Experimental section).
752
J. Chem. Soc., Dalton Trans., 2002, 752–758
This journal is © The Royal Society of Chemistry 2002
DOI: 10.1039/b108774f

View Article Online
Table 1 Conversion and enantioselectivity as a function of the reac-
tion time for transfer hydrogenation under the conditions of Scheme 1
Conversion (%)
ee (S) (%)
Substrate
10 min
1 h
8 h
10 min
1 h
8 h
1a
1b
1c
1d
2
8
30
100
2
28
70
—
7
100
100
—
80
100
8
80
26
63
36
2
78
25
—
28
4
73
27
—
16
3
Fig. 1 Numbering scheme and nomenclature of isotopomers exem-
plified for 1a and 5. Stereochemically distinct positions at C-3 (1a) and
C-1 (5) are indicated as 3/3Ј and 1/1Ј, respectively.
8
25
3
n.d.a
n.d.a
n.d.a
n.d.a
8
a n.d. = not detectable.
ments allows the unambiguous assignment of all possible
isotopomers of 5 and 1a.11 The additive shifts of the carbon
resonances by substitution of deuterium for hydrogen in the
simultaneously proton and deuterium decoupled {1H,2D}-
13C-NMR spectra is particularly useful for qualitative and
quantitative analysis of such isotopomeric mixtures.5,12 As a
representative example, Fig. 2 shows the relevant parts of the
and the esters 1b–d were chosen as substrates to study the
effects of systematic structural variation. Methylsuccinic acid 5
is obtained as the only product from all these substrates after
hydrolytic work-up. The chiral complex [(bppm)Rh(hfacac)] 4
(bppm = (2S,4S)-1-tert-butoxycarbonyl-4-(diphenylphosphino-
methyl)pyrrolidine; hfacac = 1,1,1,5,5,5-hexafluoroacetylacet-
onate) was used as catalyst precursor for transfer hydrogenation
as well as for control experiments using molecular hydrogen or
deuterium. (S)-5 was formed as the preferred enantiomer from
all substrates regardless of the reaction conditions.
Results and discussion
The reaction rates and enantioselectivities observed for the
transfer hydrogenation of substrates 1a–d vary considerably
under identical reaction conditions, but there is no direct
correlation between rate and selectivity (Table 1). With α-
methylitaconate 1c quantitative conversion is achieved within
less than 10 min, while the reduction of β-methylitaconate 1d
requires more than 24 h to reach completion. The highest
enantioselectivity of 73% at complete conversion is obtained
with itaconic acid 1a and the enantioselectivity of 1c is only
slightly lower with 63%.10 The acid group in the β-position (or
allylic position) seems to be necessary for high selectivity, as the
enantioselectivities of 1b and 1d are much lower with 27% and
16%, respectively. The two acids 2 and 3 lacking an allylic carb-
oxyl group give an almost racemic product. (Z)-Configured 3
is reduced at an extremely low rate and only 75% conver-
sion could be reached even after further treatment with excess
formic acid and an extended reaction time of five days. Only a
small decrease of the enantioselectivity during the reaction is
observed with 1a, but it is more significant in the case of 1d.
Labelling experiments were carried out as described in detail
in the Experimental section. The acidic positions of the carb-
oxylic acids were exchanged for deuterium when D2, DCOOD
or HCOOD were used as reducing agents. No exchange with
the H/D-groups of the solvent or triethylamine was observed by
direct NMR analysis of reaction mixtures in selected experi-
ments. Similarly, no H/D exchange was observed for isolated
5 under either transfer hydrogenation or hydrogenation con-
ditions. Therefore, all incorporation and scrambling processes
discussed below must occur before the irreversible liberation of
the product from the catalyst.
Fig. 2 Methyl (top trace), methylene (middle trace) and methine
region (lower trace) of the 13C{1H,2D}-NMR-spectrum of an
isotopomeric mixture of 5 obtained from deuteration of 1a with D2
after 92% conversion (E = starting material).
Mechanistic conclusions from deuterium labelling studies
require a reliable and efficient method for the analysis of all
possible isotopomers, which can evolve in a rather complex task
even for seemingly simple molecules. For methylsuccinic acid
5 there is a total of 32 isotopomers considering the three
positions in the carbon skeleton and including possible dia-
stereomeric substitutions. Up to 12 isotopomers are possible for
itaconic acid 1a. The numbering scheme and nomenclature
used to distinguish these isotopomers in the present paper is
given in Fig. 1.
{1H,2D}13C-NMR spectrum of a mixture of 1a and 5 obtained
from deuteration with D2 at 92% conversion. All signals are
clearly separated and quantitative analysis of the isotopomers
of 5 is possible by integration of the signals arising from C-1,
because there is no deuterium incorporation at this position
under the given conditions. Similar sets of comparable signals
with no significant distortion of signal intensities from differ-
ences in relaxation times can be identified also in other cases. To
substantiate the high accuracy of this technique, Table 2 shows
1
2
The concerted use of H, D, and 13C-NMR spectroscopy
together with two dimensional high resolution HSQC-experi-
J. Chem. Soc., Dalton Trans., 2002, 752–758
753

View Article Online
Table 2 Comparison of deuterium distribution as determined from
mass spectroscopy and {1H,2D}13C-NMR spectroscopy for represent-
ative transfer deuterations of 1c and 1d with DCOOD
present at low conversion where deuterium incorporation into
the substrate is still negligible (Fig. 3). The relative ratio of
these three main isotopomers remains fairly constant through-
out the reaction, whereas their contribution to the total number
of isotopomers decreases with conversion owing to deuterium
scrambling. This clearly indicates that two distinct processes are
operating: The intrinsic selectivity of the hydrogen addition
leads to a given distribution of hydrogen/deuterium at positions
C-1, C-2 and C-3, whereas scrambling leads to the formation of
more and less highly deuterated species through a separate
exchange processes. The relative distribution of the major iso-
topomers is largely determined by the intrinsic selectivity and
therefore allows conclusions about the hydrogen addition step
with different substrates.
1c
1d
Deuterium
content of 5
MS (%)
NMR (%)
MS (%)
NMR (%)
d0
d1
d2
d3
d4
13
43
33
9
14.5
40.2
33.8
9.6
6
38
39
14
3
6.7
37.9
37.4
14.8
3.2
2
1.9
the excellent agreement of the total deuterium incorporation
obtained from the sum of the individual isomers in the NMR
spectra with results from mass spectrometry for two typical
transfer hydrogenation experiments using 1c and 1d as sub-
strates. There are only minor deviations even for highly deuter-
ated isotopomers, where low concentrations and weak signal
intensities would exclude conventional 13C-NMR spectroscopic
analysis. Up to eighteen different isotopomers of 5 were
unambiguously identified and quantified with these techniques
in some of the samples resulting from transfer deuteration in
the present study.
The major isotopomer d2(2,3)-5 obtained during transfer
deuteration of 1a is formed via the expected vicinal addition of
two deuterium equivalents across the C᎐C double bond. The
᎐
d1(3) congener can result at least partly from incomplete deu-
teration of the protic positions in the substrate via the same
addition mode. From the beginning, there is also a noticeable
incorporation in position C-1, but it remains very small as
compared to the main vicinal 2,3-addition mode. This situation
changes completely when the corresponding dimethyl ester 1b is
used as substrate. In contrast to 1a, the diester 1b is reduced by
DCOOD/NEt3 almost exclusively via deuterium addition at C-1
and C-3, with very little incorporation at C-2 after full conver-
sion (Fig. 4). The isotopomer d2(1,3)-5 is formed with 57%
selectivity at 30% conversion, as compared to only 9% d2(2,3)-5
at this stage. Again, the ratio of about 8 : 1 remains constant up
to complete conversion, even though scrambling takes place
also in this case (total deuterium incorporation is 1.81). There
is no preference for the incorporation into either of the two
diastereomeric positions at C-1.
The reduction of itaconic acid 1a with DCOOD/NEt3 using
catalyst 4 in DMSO-d6 results in a complex mixture of at least
eleven isotopomers of 5 after full conversion (Fig. 3). There is a
This remarkable switch of the hydrogen addition mode can
be directly associated with the nature of the carboxylic group in
β-position (or allylic position). Substrate 1c contains a free
carboxylic acid group in this position and is reduced preferen-
tially via the 2,3-addition mode. Although the distinction of the
two sites is less pronounced for substrate 1d, the deuterium
pattern clearly indicates that the esterification of the β-position
significantly favours 1,3-incorporation. Fig. 5 summarises the
distribution of the d2 isotopomers for the four substrates 1a–d.
As most hydrogenation catalysts are effective also for double
bond isomerisation, it is important to assess the reduction and
deuterium addition for the carbon skeletons of mesaconic acid
(2) and citraconic acid (3), which are isomeric to those of the
itaconates 1. Neither 2 nor 3 were ever observed in the products
or mixtures during any of the experiments described here. This
rules out the formation of 3 as it would accumulate in the reac-
tion mixture owing to its extremely slow rate of reduction. The
reduction of 2 with DCOOD/NEt3 results in the formation of
only three significant isotopomers after full conversion. The
two monodeuterated species d1(1) and d1(2) sum to 20%, reflect-
ing mainly the incomplete deuteration of the protic positions of
DCOOD and 2. No significant amounts of products with more
than two deuteriums are detected. The isotopomer d2(1,2)-5,
which is expected from vicinal addition at the internal double
bond, is formed with a high selectivity of approximately 80%
throughout the course of the reaction. The same isotopomer is
also the major product from the transfer deuteration of 3, albeit
extreme scrambling and a very broad distribution of isotop-
omers is observed during the sluggish reduction of this com-
pound. Most notably, the d2(1,2) isotopomer of 5 is not formed
in appreciable amounts from any of the itaconates 1a–d
(Fig. 5). This demonstrates unambiguously that double bond
isomerisation to form the carbon skeletons of either 2 or 3
does not contribute to product formation during transfer hydro-
genation of 1a–d.
Fig. 3 Distribution of isotopomers of 5 obtained from transfer
deuteration of 1a with DCOOD at different conversions.
significant amount of isotopomers containing more than two
deuterium atoms. The overall deuterium incorporation corre-
sponds to 2.08 equivalents, which is a low estimate, however, as
the protic positions of commercial DCOOD and the deuterium
exchanged substrate are far from 100% deuterium content. The
deuterium scrambling may be associated with the reversibility
of the initial hydrogen transfer from the metal to the coordin-
ated substrate.6 If the substrate has time to leave the metal
center before it is reduced completely, a reversible first transfer
is expected to lead to detectable amounts of deuterated starting
material in the reaction mixture. Indeed, significant amounts of
deuterium incorporation in the starting material 1a can be
detected during the intermediate stages of transfer deuteration.
Under the conditions of Fig. 3, there is already deuterium
incorporation in 5.4% of the starting material 1a after 8.2%
conversion. After 58% conversion nearly 45% of 1a contains
deuterium mainly located at the C-3 position (35.7% of d1(3)-1a
and 7.9% of d2(3,3)-1a) with no significant preference for the
(E) or (Z) position.
The quantitative distribution of isotopomers of product 5
during the course of the transfer deuteration of 1a shows that
three isotopomers d1(3)-5, d2(1,3)-5, and d2(2,3)-5 are already
Most notably, the change of the deuterium addition mode upon
esterification is unique for the transfer hydrogenation method-
754
J. Chem. Soc., Dalton Trans., 2002, 752–758

View Article Online
ology and is not observed during deuteration of itaconic acid
derivatives 1a and 1b. As seen from Fig. 6, almost identical
distributions of isotopomers are obtained with both substrates
from deuteration with D2.13 It is also interesting to note that
polydeuteration is much more pronounced under hydrogenation
conditions than in transfer hydrogenation (compare Fig. 6 and
Fig. 4). The total amount of deuterium incorporation corres-
ponds to 4.63 and 2.85 equivalents for 1a and 1b, respectively.
Nevertheless, only traces of deuterated starting materials are
observed during the intermediate stages of the deuteration
experiments, indicating that scrambling and polydeuteration
occurs almost exclusively while the substrate is bound to the
rhodium center.
tinction of a hydridic hydrogen in the formyl position and a
protic hydrogen at the carboxylic position. Experiments using
HCOOD and DCOOH were carried out to examine a possible
preference for the incorporation from these to positions during
transfer hydrogenation. Earlier experiments with phenyl-
itaconic acid as the substrate have shown that there is a rapid
scrambling of the isotopes at the formyl and the protic position
during transfer hydrogenation with monodeuterated formic
acids.5,14,15 Fortunately, the much faster reduction rate of the
unsubstituted itaconates 1a and 1b prevented this scrambling
from being complete in the present case, revealing a clear pref-
erence for the hydrogen isotope from the formyl position to be
incorporated into the C-3 position. The d1(3) isotopomer of 5
was formed from DCOOH/NEt3 with 67% and 72% selectivity
for 1a and 1b, respectively. Most significantly, the experiments
The most significant and obvious difference between the
hydrogen transfer from H2 and HCOOH is the intrinsic dis-
Fig. 4 Isotopomer distribution of 5 obtained by transfer deuteration of 1a and 1b with DCOOD after full conversion.
Fig. 5 Distribution of the double deuterated isotopomers d2 of 5 obtained by transfer deuteration of 1a–1d with DCOOD after full conversion.
Fig. 6 Isotopomer distribution of 5 obtained by deuteration of 1a and 1b using D2 after full conversion.
J. Chem. Soc., Dalton Trans., 2002, 752–758
755

View Article Online
using HCOOD revealed that the change of the 2,3 vs. the 1,3
addition mode is directly associated with the delivery of the protic
hydrogen: the relative ratio of the two monodeuterated iso-
topomers d1(1)-5 and d1(2)-5 was 1 : 5 when 1a was reduced
with HCOOD/NEt3, whereas it changed to 7 : 1 for 1b.
expected to be the major pathway for isotope scrambling from
the formyl and protic position and for the incorporation of
more than one deuterium into substrates and product. In order
to account for the formation of polydeuterated compounds, the
insertion of the olefinic bond into the Rh–H bond to form the
chelated σ-alkyl intermediate IV must be reversible.6,13 This
must hold for both the hydrogenation and transfer hydrogenation
sequence. The insertion is of course equivalent to H migration
from rhodium to C-3, and the reverse β-hydrogen abstraction
from the same position readily explains why C-3 is the major
position for polydeuteration. The larger amount of poly-
deuteration with molecular deuterium reflects at least partly the
much higher D/H-ratio under 20 bar of D2 as compared to
transfer deuteration with five equivalents of DCOOD.
Irreversible liberation of the product from the rhodium
center through protolytic cleavage of the metal alkyl bond in
IV leads finally to the vicinal 2,3-addition product in transfer
deuteration. In hydrogenation, the Rh–C bond can be cleaved
by direct protolysis,17 via a classical oxidative addition/reductive
elimination pathway, or via base-assisted σ-bond metathesis
reaction with H2.18 The relative amount of polydeuter-
ation/scrambling to vicinal 2,3-addition is determined by the
relative ratios of the reversible insertion and the irreversible
product formation step. It is important to note that this is not
necessarily related to the amount of detectable deuterated
starting material, as this requires the additional step of de-
coordination of the substrate from the metal center in III (S2 =
substrate). This complexation/de-complexation equilibrium
depends strongly on the substrate, but is also largely influenced
by reaction conditions such as solvent, pH, etc. Indeed, a quali-
tative correlation of the amount of scrambling in the product
5 and the deuterium incorporation in the substrates 1–3 is
observed within the series of transfer deuteration experiments,
but not for the comparison of transfer hydrogenation and
hydrogenation.
Mechanistic conclusion
The enantioselective reduction of C᎐C double bonds in
᎐
unsaturated carboxylic acids and esters can be achieved using
either molecular hydrogen (hydrogenation) or a mixture of
formic acid and triethylamine (transfer hydrogenation). For a
wide range of substrates and catalysts, there are striking sim-
ilarities in reactivity and stereoselectivity for both techniques
when chiral rhodium phosphine complexes are used as cat-
alysts.3,16 Together with the results of a previous mechanistic
study,5 this strongly suggests that the major reaction paths and
the key intermediates during the hydrogen addition step are
largely similar in both processes. Scheme 2 depicts a mechanistic
An important contribution to the stability of IV results from
the formation of a five-membered chelating ring with coordin-
ation of C-3 and the carboxyl group in the β-position (allylic
position) of the substrate. It is important to note that an
isomeric σ-alkyl complex V exists that contains a similar five-
membered chelating ring with an almost identical coordin-
ation mode. In this structure, C-1 is bound directly to the metal
center and stabilisation occurs through the α-carboxyl group.
Protolytic cleavage of the Rh–C bond from intermediate V
corresponds to an overall 1,3-addition of two deuterium (or
hydrogen) atoms.
Scheme 2 Mechanistic rationale for the change from vicinal 2,3-
addition to 1,3-addition during transfer deuteration of itaconates 1a–d
(R, RЈ = D, Me, S2 = solvent or substrate).
scenario for the transfer deuteration of itaconates 1a–d that
allows the rationalisation of all the findings of the present study
on the basis of such a common catalytic pathway.
A possible pathway for the interconversion of IV and V must
involve the breaking of the sp3-hybridised C–H bond at C-1.
This process is equivalent to the β-hydride elimination that
leads to the polydeuteration at C-3. Although the latter is likely
to be preferred for statistical and stereo-electronic reasons, an
isomerisation pathway involving the scission of the C–H bond
at C-1 is clearly a viable process. The ratio of vicinal to 1,3-
addition depends in this scenario on the relative reaction rates
of Rh–C cleavage and interconversion as well as on the stability
difference between IV and V. Under the conditions of hydro-
genation, the product liberating step occurs always from inter-
mediate IV, regardless of the substitution pattern of itaconates
1a–d. Under transfer hydrogenation conditions, intermediate V
becomes kinetically accessible for the compounds 1b and 1d
bearing an ester group in the β-position. The stability of the
five membered ring in IV will be drastically reduced by ester-
ification of the carboxylic group in the β-position, thus favour-
ing both the rate of isomerisation and the relative stability of V.
The additional shift towards 1,3-addition by esterification of
both carboxylic groups is less obvious, but might be related to
possible assistance of the free ester group in the protolysis of
the Rh–C bond.
The transfer hydrogenation of 1a using rhodium bis(phos-
phine) catalysts exhibits a primary kinetic isotope effect kH/kD =
3.2 when the formyl hydrogen is replaced with deuterium.5 This
is in excellent agreement with other isotopic effects for the
decarboxylation of formic acid or formates to give CO2 and
metal hydrides. The scrambling of the isotopes of the formyl
position and the protic position during hydrogen transfer is also
best rationalised on the basis of the formation of rhodium
hydride intermediates. Although other hydrogen transfer mech-
anisms may be possible with heterogeneous catalysts,8 it seems
therefore most appropriate to identify the first step of the
catalytic cycle of the homogeneously rhodium phosphine cat-
alysed transfer hydrogenation with the decarboxylation of
coordinated formate to yield a rhodium hydride intermediate
such as III.
Just like III, all intermediates in Scheme 2 are drawn as
neutral species for simplicity. Obviously, the rhodium hydride
complexes can be subject to protonation under transfer hydro-
genation conditions. Similarly, they will exist in part as cationic
dihydrides under hydrogenation conditions. The exchange
of rhodium bound hydrides with protons in solution can be
In summary, the network of isomerisation and product form-
ing steps shown in Scheme 2 provides a plausible explanation
756
J. Chem. Soc., Dalton Trans., 2002, 752–758

View Article Online
for the key experimental facts of the present detailed study of
the hydrogen addition step in hydrogenation and transfer
hydrogenation. It is based on the assumption that the mech-
anistic pathways of both reactions involve closely related
intermediates.19 The results presented here provide therefore
important insight not only into the transfer hydrogenation
pathway, but also into asymmetric hydrogenation in general.
Recently, the competition of 2,3- vs. 1,3-addition was used as a
mechanistic probe to assess the hydrogen addition pathway dur-
ing hydrogenation in supercritical CO2 as solvent.20 The revers-
ibility of the olefin insertion and the potential influence of the
isomerisation on the stereoselectivity of the overall reduction in
both hydrogenation and transfer hydrogenation is another
intriguing aspect that emerges from this study.
(Mϩ), 763 (Mϩ Ϫ C5O2H8) high resolution MS: found: min.
863.117, max. 863.127; calc: 863.122.
General procedure for the H/D exchange of the protic positions in
substrates 1a, 1c and 1d
3 g of substrate were dissolved in 10 ml d1-methanol, the
solution stirred for 10 min and the solvent then removed
in vacuo. The procedure was repeated three times to achieve an
H/D-exchange of ≥90%
General procedure for transfer deuteration
The following procedure is representative for a reaction with
monitoring of conversion. Smaller scale reactions (1.5 mmol
substrate) were carried out in cases where only the final dis-
tribution was analysed. The appropriately O-deuterated sub-
strate 1–3 (22.5 mmol) and [(bppm)Rh(hfacac)] 4 (0.336 mmol,
1.5 mol% Rh) were dissolved in 15 ml d6-DMSO and triethyl-
amine (45.0 mmol) followed by the desired isotopomer of for-
mic acid (112.5 mmol) were added. The reaction was stirred at
30 ЊC and up to eight samples of 1.6 mL were withdrawn over a
period of 8 h and quenched with 10 ml 5% HCl. For substrates
1a, 2, and 3 the aqueous solution was extracted five times with a
total of approximately 40 ml diethyl ether. The combined
organic phases were washed with 5 ml 10% HCl, dried over
MgSO4, and evaporated to leave 5 as a white solid in >85%
yield. For substrates 1b, 1c and 1d, the aqueous solution
obtained after quenching was rendered basic by addition of
sodium hydroxide and stirred for two weeks at room temper-
ature to ensure complete hydrolysis of the ester groups.24 Com-
pound 5 was isolated from these solutions after acidification as
described for the other substrates.
Experimental
All catalytic reactions and manipulations of air-sensitive
materials were carried out under argon atmosphere. Solvents
were purified, dried and degassed according to standard pro-
cedures and stored under argon. The deuterium content in the
protic position of the formic acid isotopomers was found to
decrease upon prolonged storage in standard glassware even
under an argon atmosphere. Silylation of the glassware by
treatment with Me3SiCl and subsequent flame drying prior to
use ameliorated these problems to a certain extent. The follow-
ing commercially available compounds were used: (2S,4S)-1-
tert-butoxycarbonyl-4-(diphenylphosphinomethyl)pyrrolidine
(bppm) (>98%, Fluka), citraconic acid 3 (98%, Aldrich), D2
(MG: 2.7), dimethylitaconate 1b (>97%, Fluka), d2-formic acid
(CD: D > 99%, OD: D = 90%, Aldrich), d1-formic acid
(DCO2H, D = 98%, CIL), d1-formic acid (HCO2D, D =
98%, CIL), itaconic acid 1a (>99%, Fluka), mesaconic acid 2
(99%, Aldrich). α-Methylitaconate 1c and β-methylitaconate 1d
were synthesised according to literature procedures.21,22 NMR
spectra were recorded on Bruker AMX-300 for routine analysis
and on a DMX600 spectrometer for the analysis of isotop-
omeric mixtures. Chemical shifts δ are reported in ppm relative
to external H3PO4 for 31P and to SiMe4 for 1H and 13C, using the
solvent resonance as internal standard if possible. The enantio-
meric excess was determined by HPLC (instrument: MDLC-1;
stationary. phase: 250 mm Chiralcel OD-H, 4.6 mm i. d.;
mobile phase: n-heptane, 2-propanol, formic acid = 90 : 10 : 1;
T /p/F: 308 K/2.5 Mpa/0.5 ml minϪ1; detector: RI, E = 32) or
GC (HP 5890/531; column: 30 m G-TA gamma CD; G/228;
temperature: 180/60 0.8/min 87 5/min 180/300; gas: 0.9 bar
H2). Mass spectroscopic analyis of the deuterium content was
carried out on a Finnigan MAT95 in DE mode using CA
ionization (gas: NH3).
General procedure for deuteration reaction
A stainless steel pressure vessel was charged with substrate 1a
or 1b (2.0 mmol), catalyst 4 (30.0 µmol, 1.5 mol% Rh),
methanol-d1 (3 ml) and NEt3 (4.0 mmol). The solution was
pressurised with D2 (20 bar) at 30 ЊC and stirred for 2–20 h.
The reactor was cooled to room temperature, vented, and the
solution quenched and worked up as described above.
Acknowledgements
This work was supported by the Max-Planck-Gesellschaft and
the Fonds der Chemischen Industrie. We thank Dr Claude
de Bellefon and Dr Nathalie Tanchoux (Lyon) for com-
municating results prior to publication. The help of Dr R.
Mynott and C. Wirtz in the NMR spectroscopic analysis of the
isotopomeric mixtures is gratefully acknowledged.
Synthesis of [(bppm)Rh(hfcac)] 423
673 mg (1.62 mmol) [(cod)Rh(hfacac)] and 891mg (1.61 mmol)
bppm were each dissolved in 5 ml THF. Both solutions were
cooled to Ϫ78 ЊC and the bppm solution was slowly added to
the solution of the rhodium complex via a canulla. While warm-
ing to room temperature over night the orange color of the
solution changed to brown. The solvent was removed in vacuo
and the glassy residue dried at 50 ЊC under high vacuum for
four days to give 1.16 g (1.34 mmol, 83%) of 4 as a brown solid.
1H-NMR (300.1 MHz, CD2Cl2) δ 1.40 (s, 9H; (CH3)3C), 2.26–
3.76 (m, 6H; CH2), 4.57 (m, 1H; NCH ), 4.67 (m, 1H, PCH ),
5.95 (s, 1H, CF3CCH ), 7.70–8.23 (m, 20H; arom. CH ).
References and notes
1 (a) S. Gladiali and G. Mestroni, in Transition Metals for Organic
Synthesis, M. Beller and C. Bolm, Wiley-VCH, Weinheim, 1998,
vol. 2, p. 97; (b) R. B. Johnston, A. H. Wilby and I. D. Entwistle,
Chem. Rev., 1985, 85, 129.
2 (a) R. Noyori and S. Hashiguchi, Acc. Chem. Res., 1997, 30, 97;
(b) G. Zassinovich, G. Mestroni and S. Gladiali, Chem. Rev., 1992,
92, 1051.
3 (a) H. Brunner and W. Leitner, Angew. Chem., Int. Ed. Engl., 1988,
27, 1180; (b) H. Brunner, E. Graf, W. Leitner and K. Wutz,
Synthesis, 1989, 743; (c) H. Brunner and W. Leitner, J. Organomet.
Chem., 1990, 387, 209.
4 (a) J. M. Brown and P. A. Chaloner, J. Am. Chem. Soc., 1980, 102,
3040; (b) C. R. Landis and J. Halpern, J. Am. Chem. Soc., 1987, 109,
1746; (c) A. Harthun, R. Selke and J. Bargon, Angew. Chem., Int. Ed.
Engl., 1996, 35, 2505.
13C{1H}-NMR (50.3 MHz, CD2Cl2) δ 28.64 (s, (CH3)3C), 34.5
1
(s, NCHCH2), 37.9 (s, NCH2), 38.3 (s, NCH), 49.6 (d, JCP
=
62 Hz, PCH2), 55.0 (d, under CD2Cl2, PCH), 79.9 (s, (CH3)3C),
90.8 (s, CF3CCH), 127.4–135.6 (m, arom. CH), 157.7 (s, NCO),
171.6 (q, 2JCF = 33 Hz, CF3C), CF3 not detected.31P{1H}-NMR
5 W. Leitner, J. M. Brown and H. Brunner, J. Am. Chem. Soc., 1993,
115, 152.
6 D. J. Hardick, I. S. Blagbrough and B. V. L. Potter, J. Am. Chem.
Soc., 1996, 118, 5897.
7 N. Tanchoux and C. de Bellefon, Eur. J. Inorg. Chem., 2000, 1495.
1
2
(121.5 MHz, CD2Cl2) δ 39.5 (d,d, P1, JRhP1 = 185 Hz, JP1P2
=
2JP1P2Ј = 62 Hz), 65.8 (d,d, P2, 1JRhP2 = 191 Hz, 2JP1P2 = 62 Hz),
66.1 (d,d, P2Ј, 1JRhP2Ј = 193 Hz, 2JP1P2Ј = 62 Hz). MS: m/z = 863
J. Chem. Soc., Dalton Trans., 2002, 752–758
757

View Article Online
16 The major exception are catalysts with chiral ligands forming five-
8 For a related study on heterogeneously Pd-catalysed transfer
hydrogenation, see (a) J. Yu and J. B. Spencer, Chem. Commun.,
1998, 1935; (b) J. Yu and J. B. Spencer, Chem. Eur. J., 1999, 5, 2237.
9 B. T. Khai and A. Arcelli, J. Org. Chem., 1989, 54, 949.
10 Enantioselectivities of up to 94% can be achieved with in-situ-
formed Rh/bppm catalysts under optimised conditions.3
11 Details of the NMR spectroscopic assignment will be given
separately, S. Lange, W. Leitner, R. Mynott and C. Wirtz,
manuscript in preparation.
12 For related applications of simultaneously proton and deuterium
decoupled 13C-NMR spectroscopy in deuterium labelling studies,
see (a) J. M. Brown, A. E. Derome, G. D. Hughes and P. K.
Monaghan, Aust. J. Chem., 1992, 45, 143; (b) J. K. MacDougall,
M. C. Simpson and D. J. Cole-Hamilton, J. Chem. Soc., Dalton
Trans., 1994, 3061.
membered chelates, as the corresponding rhodium complexes are
not effective for the formyl C–H bond cleavage.3
17 C. Detellier, G. Gelbard and H. B. Kagan, J. Am. Chem. Soc., 1978,
100, 7556.
18 R. H. Crabtree, M. Lavin and L. Bonneviot, J. Am. Chem. Soc.,
1986, 108, 4032.
19 An alternative explanation involving rhodacylobutane intermediates
was put forward to rationalise the change of vicinal to 1,3-addition
during transfer hydrogenation of 1a or 1b in aqueous media.7
Although we cannot rule out this possibility, it does not provide a
satisfactory explanation for the decisive influence of the allylic
carboxyl group and the preferred incorporation of protons at C-1.
20 S. Lange, A. Brinkmann, P. Trautner, K. Woelk, J. Bargon and
W. Leitner, Chirality, 2000, 12, 450.
13 The distributions depicted in Fig. 6 are in line with those reported in
a previous study on deuteration of itaconates using D2: A. Buscai,
J. Bakos, M. Laghmari and D. Sinou, J. Mol. Catal. A: Chem., 1997,
116, 335.
14 This scrambling results mainly from exchange of rhodium hydrides
with the proton pool.5 The cleavage of the formyl C–H bond is
reversible, but the back reaction to give formic acid is thermo-
dynamically unfavourable in the absence of H2 and CO2 pressure.15
15 (a) W. Leitner, Angew. Chem., Int. Ed. Engl., 1995, 34, 2207; (b) P. G.
Jessop, T. Ikariya and R. Noyori, Chem. Rev., 1995, 95, 259.
21 K. Achiwa, P. A. Chaloner and D. Parker, J. Organomet. Chem.,
1981, 218, 249.
22 B. R. Baker, R. E. Schaub and J. H. Williams, J. Org. Chem., 1952,
17, 116.
23 K. Angermund, W. Baumann, E. Dinjus, R. Fornika, H. Görls,
M. Kessler, C. Krüger, W. Leitner and F. Lutz, Chem. Eur. J., 1997,
3, 755.
24 Acidic hydrolysis proved less suitable. Control experiments showed
that no measureable H/D exchange in product 5 occurred under the
basic conditions.
758
J. Chem. Soc., Dalton Trans., 2002, 752–758