Rhodium(I) hydrogenation in water: Kinetic studies and the detection of an intermediate using 13C{1H} PHIP NMR spectroscopy

Article abstract of DOI:10.1016/j.ica.2006.08.042

The mechanism for hydrogenation of dimethylmaleate in water using cationic rhodium complexes with water-soluble bi-dentate phosphines has been investigated using kinetics and a novel method for the indirect detection of intermediates in catalytic hydrogenation reactions, whereby a late intermediate was detected. A mechanism is proposed involving fast, irreversible substrate binding followed by a rate-determining reaction with dihydrogen.

Full text of DOI:10.1016/j.ica.2006.08.042

Inorganica Chimica Acta 360 (2007) 1621–1627
www.elsevier.com/locate/ica
Rhodium(I) hydrogenation in water: Kinetic studies and the detection
13
of an intermediate using C{1H} PHIP NMR spectroscopy
a
a
b
b
˚
Marten Ahlquist , Mikaela Gustafsson , Magnus Karlsson , Mikkel Thaning ,
b
a,
*
Oskar Axelsson , Ola F. Wendt
a
Organic Chemistry, Department of Chemistry, Lund University, P.O. Box 124, 221 00 Lund, Sweden
b
GE Healthcare, Nycoveien 2, P.O. Box 4220 Nydalen, NO-0401 Oslo, Norway
Received 31 May 2006; accepted 28 August 2006
Available online 12 September 2006
Abstract
The mechanism for hydrogenation of dimethylmaleate in water using cationic rhodium complexes with water-soluble bi-dentate phos-
phines has been investigated using kinetics and a novel method for the indirect detection of intermediates in catalytic hydrogenation reac-
tions, whereby a late intermediate was detected. A mechanism is proposed involving fast, irreversible substrate binding followed by a
rate-determining reaction with dihydrogen.
ꢀ 2006 Elsevier B.V. All rights reserved.
Keywords: Hydrogenation; Rhodium; Water-soluble phosphines; Mechanism; para-Hydrogen; PHIP
1. Introduction
crucial dihydride intermediate proposed for these reactions
[7]. Still, the general problem with detecting intermediates,
also using para-hydrogen, is that there is no way to be sure
that an observed intermediate is on the pathway to prod-
ucts. On the contrary, it has often been shown that the
detection indicates high stability and a dead-end species.
Here, we present a mechanistic investigation of the hydro-
genation of dimethylmaleate with water-soluble cationic
rhodium catalysts. It includes a novel, para-hydrogen-
based technique for the detection of intermediates, that
can provide evidence of catalytic competency.
Asymmetric hydrogenation is an important way to
obtain chiral molecules, and using cationic rhodium com-
plexes has been one of the most successful approaches
[1]. One variety of these catalysts is based on water-soluble
phosphines, which allows for reactions to be performed in
water or under bi-phasic conditions. They often show a
high activity but in general the chiral induction is lower
than in organic solvents [2]. Mechanistic investigations of
aqueous systems are rare [3], but they point to a similar
mechanism for aqueous hydrogenations as was earlier
established in organic solvents [4]. Two of us have recently
been involved in efforts towards inducing a polarisation
enhancement in MRI contrast agents using rhodium catal-
ysed hydrogenation reactions in water with para-hydrogen
[5,6]. para-Hydrogen induced polarisation (PHIP) has also
been used extensively for mechanistic work and Bargon
and co-workers found some experimental evidence for the
2. Experimental
2.1. General procedures and materials
All experiments were carried out using standard Schlenk
techniques or in a glove box under nitrogen. Water was
degassed prior to use. All other solvents were dried and
degassed prior to use. All other reagents were of the best
quality available and used as received. [Rh(nbd)2]BF4 was
prepared according to the literature methods [8]. Tetrasulf-
*
Corresponding author. Tel.: +46 46 2228153; fax: +46 46 2228209.
E-mail address: ola.wendt@organic.lu.se (O.F. Wendt).
0020-1693/$ - see front matter ꢀ 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2006.08.042

1622
M. Ahlquist et al. / Inorganica Chimica Acta 360 (2007) 1621–1627
onated bis(diphenylphosphino)butane (DPPBTS), and
complexes 1 and 3 were prepared as described by Malm-
stro¨m et al. for the corresponding triflate complex [3]. Ele-
mental analyses were performed by H. Kolbe
cold for 80 min. The mixture was quenched on 20 g
degassed ice and a milky white suspension was obtained.
Triisooctylamine (1 ml) and toluene (10 ml) were added.
After stirring, another 10 ml of water was added and the
toluene phase was separated off. The resulting suspension
was centrifuged and the supernatant removed. The precip-
itate was dissolved in 15 ml of water and pH was adjusted
to 8.6 with sodium hydroxide. A white solid precipitated
and was collected. Based on carbon analysis and quantita-
tive NMR it contains approximately 20% Na2SO4 Æ xH2O,
Mikroanalytisches Laboratorium, Mulheim an der Ruhr,
¨
1
Germany. H, 31P and 13C NMR spectra were recorded
on a Varian Unity 300 or a Varian Unity Inova 500 spec-
trometer. Chemical shifts are given in ppm downfield from
TMS using residual solvent peaks (1H, 13C NMR) or
H3PO4 (31P NMR d 0) as external references. Assignments
are based on NOESY, HSQC and HMBC experiments.
Mass spectra were recorded on a Micromass Q-Tof Micro
MS instrument, using ESI+ and ESIꢀ ionisation and
water/acetonitrile (50/50) with 0.1% formic acid as solvent.
1
which could not be removed. H NMR (D2O): d 1.39 (b,
4H) 1.97 (b, 4H) 3.79 (s, 12H) 7.02 (bd, 4H) 7.40 (bt,
4H) 7.69 (bd, 4H). 31P NMR: (D2O) d ꢀ18.0 (s). Anal.
Calc. for C32H32Na4O16P2S4: C, 40.26; H, 3.38. Found:
C, 31.61–32.64; H, 3.42–3.66% (different batches contain
different amounts of sodium sulfate). m/z 977.0
(M+Na)+, 930.7 (M ꢀNa)ꢀ, 886.9 (MH2ꢀNa3)ꢀ, 97.0
ðHSO4ꢀÞ.
2.2. Preparations
2.2.1. Tetrasulfonated bis(dianisylphosphino)butane
(DAPBTS)
To a solution of dibutylphosphite (7.81 ml, 40 mmol) in
THF, sodium metal (0.92 g, 40 mmol) was added and the
mixture was stirred under reflux overnight. The solution
of sodiumdibutylphosphite was added dropwise to an ice-
cold grignard solution made from magnesium (2.19 g,
90 mmol) and 4-bromoanisole (10.7 ml, 85.5 mmol) in
ether/THF. The mixture was refluxed for 3 h and then left
stirring overnight. The reaction was quenched with 3 ml of
water. After stirring for 1 h, the mixture was diluted with
ethyl acetate and the organic phase was decanted into an
Erlenmeyer flask and the solid grey residue was treated
with 2 M HCl (aq) until it dissolved. The aqueous phase
was poured into the organic phase and the mixture was
stirred for 30 min. After phase separation the organic
phase was washed twice with water, dried and after evapo-
ration a white solid remained. This was recrystallised from
ethyl actetate giving 8.27 g (79%) of the dianisylphosphine
oxide. Fifty grams (0.19 mol) of this compound was
dissolved in THF (ca. 500 ml) and 76 ml n-BuLi (2.5 M
in hexanes) was added dropwise at RT. After 2 h, 1,4-
dibromobutane (10.71 ml, 94.9 mmol) was added dropwise
and after yet another 2 h, the reaction mixture was poured
onto 500 ml of 1 M NaH2PO4 (aq). After extraction with
dichloromethane (three times), the combined organic
phases were dried, the solvent was evaporated and the
product was purified by column chromatography on silica.
Elution with dichloromethane:methanol (10:0.7) gave
39.5 g (72%) of the bis(dianisylphosphineoxide)butane.
This was then reduced according to the literature [9] to give
2.2.2. [RhDAPBTS(D2O)2]BF4 (2-BF4)
A solution of 4 in degassed D2O was hydrogenated for
3 min followed by a nitrogen purge. The solution was used
1
immediately. H NMR (D2O): d 1.69 (bt, 4H) 2.12 (b, 4H)
3.82 (s, 12H) 7.1–7.9 (m, 12H). 31P NMR (D2O) d 51.5 (d,
1JRh–P 200 Hz).
2.2.3. [Rh(NBD)DAPBTS]BF4 (4-BF4)
In a typical procedure, a Schlenk flask was charged with
Rh(NBD)2(BF4) (2.1 mg, 5.7 lmol) and DAPBTS (6.4 mg,
5.3 lmol). One millilitre of D2O was added giving a yellow
solution that was used within two days. 1H NMR (D2O): d
1.45 (s, 2H) 1.56 (bt, 4H) 2.42 (b, 4H) 3.78 (s, 2H) 3.88 (s,
12H) 4.46 (s, 4H) 7.1–7.9 (m, 12H). 31P NMR (D2O) d 27.5
1
(d, JRh–P 154 Hz).
2.2.4. [Rh(DMM)DPPBTS]BF4 (5-BF4)
A solution of 1 was prepared as described in Ref. [2] and
an excess of dimethyl maleate (DMM) was added. 1H
NMR (D2O): d 3.56 (s, CH3O) 6.75 (s, HC@CH). 31P
1
1
NMR (D2O) d 18.6 (bd, JRh–P 129 Hz) 55.9 (bd, JRh–P
155 Hz). 13C NMR (D2O) d 133.5 (s, HC@CH) 177.0 (s,
C@O).
2.2.5. Rh(DMM)DAPBTS]BF4 (6-BF4)
A solution of 2 was prepared as described above and
approximately 1 equiv. of dimethyl maleate (DMM) was
added. 1H NMR (D2O): d 3.55 (s, CH3OCO) 3.65 (s,
CH3O) 6.74 (s, HC@CH). 31P NMR (D2O): d 17.3 (bd,
1
the
corresponding
phosphine,
bis(dianisylphosph-
1JRh–P 139 Hz) 53.5 (bd, JRh–P 154 Hz) 13C NMR
1
ino)butane, which was pure according to H NMR spec-
troscopy in CDCl3: d 1.54 (b, 4H) 2.02 (b, 4H) 3.79 (s,
12H) 6.85–6.90 (m, 8H) 7.30–7.40 (m, 8H).
(D2O): d 56.4 (s, CH3O), 133.5 (s, HC@CH) 175.7 (s,
C@O).
A flask was charged with bis(dianisylphosphino)butane
(277 mg), boric acid (200 mg) and 3 ml of conc. sulfuric
acid. The mixture was stirred until everything dissolved
and then was cooled on an ice-bath. Fuming H2SO4
(3.25 ml) was added dropwise and the mixture was stirred
2.3. Hydrogenation of DMM
A solution of DMM (13.7 ll, 0.105 mmol) in 3.0 ml
D2O was prepared in a round bottomed flask equipped
with a septum. The flask was flushed with N2(g). A

M. Ahlquist et al. / Inorganica Chimica Acta 360 (2007) 1621–1627
1623
evaluated as the slopes of these plots. Complete data are
reported in Table S2 and S3.
40
20
2.5. para-Hydrogen experiments
2.5.1. Sample preparation
Samples were prepared in 10 mm NMR tubes with a
drop-catching expansion volume by adding 5–10 ll of 7
and 1.5 ml degassed D2O. The tube was equipped with a
septum with a Teflon outlet and flushed with N2(g).
Through the septum, 0.25–0.50 ml of the catalyst stock
solution was added via syringe and the tube was flushed
with N2(g) again. The tube was sealed with parafilm to
keep oxygen out.
0
0
20
40
[DMM]/mM
Fig. 1. Observed pseudo first-order rate constants for substrate binding to
1 as a function of DMM concentration. T = 298.8 K; [Rh] = 0.13 mM.
2.5.2. Experimental setup
Pure parahydrogen was produced at 3 MPa and 15 K in
the presence of a hydrous ferric oxide catalyst purchased
0.1 ml portion of the stock solution of 3 or 4 was added
from C chem, P.O. Box 640, Lafayette, CO, USA, and
*
and H2(g) was bubbled through for 5–15 min. The resulting
stored (<5 days) in an aluminium cylinder equipped with
a regulator and teflon tubing. Before the experiment, the
p-H2 was flushed through the teflon tubing during ca.
10 min to saturate the material with p-H2 and to remove
all paramagnetic materials e.g. O2. The tubing was led
through the septum into the NMR-tube, until it reached
the bottom, and, in addition, the septum was equipped
with a teflon outlet. Some para-hydrogen was bubbled
through carefully, to activate the catalyst and to remove
other gases and the sample was placed inside the NMR
spectrometer. Before collecting a spectrum full relaxation
was allowed to take place.
1
solution contained only dimethylsuccinate as seen by H
NMR spectroscopy.
2.4. Kinetics
2.4.1. Substrate binding
The stopped-flow experiments were performed on an
Applied Photophysics Bio Sequential SX-17 MX,
stopped-flow spectrofluorimeter under anaerobic condi-
tions. The substitution of water in 1 by DMM was stud-
ied in water by observing the decrease in absorbance at
330 nm. The complex solution (1.3 · 10ꢀ4 M) was mixed
directly in the stopped-flow instrument with an equal
volume of DMM solution, resulting in a large excess
of the entering ligand (1.3–21 · 10ꢀ3 M), assuring pseudo
first-order conditions. The kinetic traces were fitted to
single exponentials using the software provided by
Applied Photophysics,1 resulting in observed rate con-
stants at different concentrations of the incoming ligand.
Rate constants are given as an average of at least five
runs. Complete data are reported in Table S1 and in
Fig. 1.
For ordinary PASADENA spectra, para-hydrogen was
bubbled through the sample inside the spectrometer, the
sample was allowed to settle, and the spectrum was col-
lected using a single 45ꢁ (1H) or 90ꢁ (13C) pulse.
Indirect detection experiments were performed in a sim-
ilar way with a few additional elements. The decoupler off-
set frequency (dof) and decoupler power (dpwr) were set to
an appropriate value. The decoupler was turned on and p-
H2 was bubbled through during 10 s. The decoupler was
kept on for another 10 s (20 s in total), after which it was
turned off to let the J-coupling evolve. This second delay
was 30 ms, and the fid was collected using a single 90ꢁ
pulse. The 13C signal of the product was detected and inte-
grated and after relaxation this procedure was repeated
using a different dof. Thus, the indirectly detected spectra
in Figs. 4, 5, S1 and S2 were constructed.
2.4.2. H2-consumption in the catalytic reaction
As described above, a solution of 1 or 2 (or 4) in H2O
was prepared and an appropriate amount of DMM was
added to a 30 ml aliquot of the rhodium solution. Using
a Buchi press-flow gascontroller 6002 apparatus, the H -
¨
2
consumption of these solutions was measured at constant
H2-pressure. In all experiments, the plots of H2-consump-
tion versus time were linear indicating a zero-order depen-
dence in the limiting reagent, i.e. the olefin. Rates were
3. Results and discussion
The rhodium complex [Rh(D2O)2DPPBTS]+ (1) (Chart
1) is a very active catalyst in the hydrogenation of a-ace-
tamidoacrylic acid, especially at low pH and [Rh(D2O)2-
DAPBTS]+ (2) is expected to display similar behaviour
[3].
1
Applied Photophysics Bio Sequential SX-17 MV Stopped Flow ASVD
They are readily formed via hydrogenation of the corre-
sponding norbornadiene complexes 3 and 4 (Scheme 1); 1
Spectrofluorimeter, software manual, Applied Photophysics Ltd. 203/205
Kingston Road, Leatherhead KT22 7PB, UK.

1624
M. Ahlquist et al. / Inorganica Chimica Acta 360 (2007) 1621–1627
MeO
HO3S
OMe
SO3H
O
O
MeO
OMe
OMe
HO3S
SO3H
OD2
P
P
P
P
OD2
OD2
DMM
Rh
Rh
OD2
O
O
C
MeO13
HO3S
SO3H
HO3S
MeO
SO3H
D
D
OMe
7
1
2
Chart 1.
DMM
H2/D2O
Rh(P-P)(nbd)+
Rh(P-P)(DMM)+
+
Rh(P-P)(D2O)2
3 (P-P = DPPBTS)
4 (P-P = DAPBTS)
1 (P-P = DPPBTS)
2 (P-P = DAPBTS)
5 (P-P = DPPBTS)
6 (P-P = DAPBTS)
Scheme 1.
has been characterised earlier [3] and 2 displays a similar
31P NMR spectrum with a doublet at 51.5 ppm (1JRh–P
Bubbling H2 for 5–15 min through an aqueous solution
of 1 or 2 and an excess of DMM, gave complete conversion
of the substrate to dimethyl succinate, showing that the
rhodium cations are effective catalysts.
=
200 Hz). Addition of a slight excess of dimethylmaleate
(DMM, Chart 1) to a D2O solution of 1 rapidly yielded a
31P NMR spectrum displaying a broadened doublet at
55.6 ppm (1JRh–P = 152 Hz) and another broadened dou-
blet at 18.2 ppm (1JRh–P = 128 Hz) in accordance with
the expected resonances for a [Rh(DMM)DPPBTS]+ (5)
complex. No resonances from 1 could be detected in solu-
tion. The coordination of DMM could be confirmed using
1H NOESY spectroscopy, where the olefinic protons of
DMM displayed cross-peaks with aliphatic and aromatic
protons of the phosphine ligand, showing that they are
close in space. The complex [Rh(DMM)DAPBTS]+ (6)
was obtained similarly and displayed similar spectral fea-
tures. For both 5 and 6, the coordination mode of DMM
is probably bidentate with one phosphorus trans to an oxy-
gen and one trans to the double bond. A monodentate
binding fashion through the olefin moiety, with one water
molecule also coordinated, cannot be ruled out. However,
the coupling constants of the resonances at 55.6 and 53.5
for 5 and 6, respectively, are substantially lower than those
found in 1 and 2, indicating a different trans influence of
the coordinated ligand, and this speaks against water coor-
dination. All reactions are depicted in Scheme 1.
Since the stability in solution of 1 is substantially higher
than that of 2, we chose the former to monitor the kinetics
for the substrate binding reaction using UV–Vis stopped-
flow spectroscopy. From the slope and intercept of a plot
of observed pseudo first-order rate constants versus
DMM concentrations, the forward and reverse rate con-
stants were computed to be 20.0 0.6 Mꢀ1 sꢀ1 and 0.025
0.006 sꢀ1, cf. Fig. 1. From the ratio of these rate con-
stants, a binding constant for DMM of (8.0 1.8) ·
102 Mꢀ1 was calculated.
The overall catalytic reaction with 4 as pre-catalyst
(forming 2 in situ) was studied by measuring the hydrogen
consumption at constant hydrogen pressure. The kinetic
traces are zero-order, showing the reaction to be zero-order
in olefin concentration. A plot of reaction rates at different
hydrogen pressures (Fig. 2) shows that the reaction is first-
order in P(H2). At constant hydrogen pressure the rhodium
concentration was also varied giving the plot in Fig. 3.
There seems to be some saturation but at low catalyst load-
ings a first-order dependence in catalyst concentration is
observed. This behaviour has been reported earlier [3].
Overall this gives the following rate law,
0.8
0.6
0.4
0.2
0
0
2
1
P/MPa
Fig. 2. Observed rate of hydrogenation of DMM as a function of
dihydrogen pressure using 2 as a catalyst. T = 298.8 K; [Rh] = 0.24 mM.

M. Ahlquist et al. / Inorganica Chimica Acta 360 (2007) 1621–1627
1625
trum with a strong signal at 3.65 ppm originating from the
methoxy-protons of the substrate, and one antiphase signal
at 2.51 ppm with a complicated coupling pattern, originat-
ing from the polarised product. No signals from any hyd-
ridic (or other possible) intermediates were detected,
implying that they are too short-lived under these condi-
0.8
0.6
0.4
0.2
0
1
tions. H spins usually relax faster than those of 13C and
1
indirect detection of a H NMR-spectrum via 13C could
therefore circumvent these problems. The amount of PHIP
observed in the product is dependent on how much of the
effect is lost due to relaxation in the intermediates. By irra-
diating potential intermediates with a radio frequency cor-
responding to their 1H NMR chemical shift during the
hydrogenation, the singlet state of the protons, originating
from the para-hydrogen, is prevented from evolving into
the triplet manifold under the influence of chemical shifts
and J-couplings. Little relaxation during catalysis should
give a strong PHIP when observing the 13C NMR signal
of the product and thus we can get a correlation between
the signal intensity from the product and the chemical shift
of the intermediate. In this way the decoupler radio fre-
quency was swept, in steps, over the whole region of possi-
ble 1H NMR frequencies and the integral of the anti-phase
13C signal of the product was measured. Hence, we could
obtain a frequency-product signal response diagram similar
to a classical continuous wave spectrum where absorption
is measured as a function of frequency. Using a fairly high
decoupler effect, the whole proton region from ꢀ20 to
10 ppm was swept giving spectra for both catalysts with
one or two (depending on resolution) broad peaks centred
around 1.6–2.0 ppm.
0
0.4
0.8
[Rh]/mM
Fig. 3. Observed rate of hydrogenation of DMM as a function of rhodium
concentration using 2 as a catalyst. T = 298.8 K; p(H2) = 1.0 MPa.
Rate ¼ kPðH2Þ½Rhꢁ
ð1Þ
which is compatible with the mechanism proposed ear-
lier for hydrogenation of dehydroamino acids involving
fast, reversible (vide infra) substrate binding prior to the
rate-determining reaction with H2 [3,4]. The second-order
rate constant, obtained from Fig. 2, was determined to be
1.2 0.2 MPaꢀ1 sꢀ1. The reaction using 3 as pre-catalyst
was investigated less thoroughly, but gave essentially the
same rate law, which is the same as reported for this cata-
lyst with acetamido acrylic acid; see Supplementary mate-
rial for all data. For 3, the average second-order rate
constant was 3.8 0.5 MPaꢀ1 sꢀ1. In this case, the zero-
order dependence in olefin concentration is in agreement
with the high binding constant found in the stopped-flow
experiments, giving a fast formation of the DMM adduct.
It is clear when comparing the first-order rate constants for
DMM displacement by water (0.025 sꢀ1) and the forward
reaction with dihydrogen (3.8 sꢀ1 at 1 MPa) that the
DMM binding is irreversible at practically all hydrogen
pressures and that the DMM complex is the resting state
of the catalyst.
Lately, a dihydride route (as opposed to an unsaturated
route), i.e. where the dihydrogen activation precedes the
substrate binding, has been shown to be a viable alternative
also for some cationic rhodium systems [10]. However, the
fact that substrate binding is faster than overall catalysis
and essentially irreversible makes this unlikely in the cur-
rent system.
Having established rate law and stoichiometric mecha-
nism we conducted a number of PHIP experiments under
PASADENA conditions [11]. Thus, a solution of 1 or 22
and isotopically labelled dimethyl maleate (7, see Chart
1) were reacted with para-H2, resulting in a 1H NMR spec-
Lowering the decoupler power gave a lower peak width
but the response always went through a null at around the
same frequency. Since we have no phase information in the
decoupling process we interpret this behaviour as a single
anti-phase peak. At a decoupler bandwidth of approxi-
mately 40 Hz the experiment was performed with smaller
0.6
0.4
0.2
0
-0.2
4
2
0
δ/ppm
Fig. 4. Indirect 1H NMR spectrum of the hydrogenation intermediate
2
Complex 4 was mostly used to give 2 in situ. The reason for this is the
using 3 as catalyst. [Rh] = 1.32 mM; [DMM]0 = 40.0 mM; and T = 298 K.
relative instability of 2.

1626
M. Ahlquist et al. / Inorganica Chimica Acta 360 (2007) 1621–1627
150 Hz wide, which corresponds to a rate for leaving
the site of ca. 3–400 sꢀ1 and hence a lifetime of the inter-
mediate of approximately 3 ms. This means that the over-
all rate is some two orders of magnitude slower than the
decay of 8. Thus, although this intermediate is long-lived
enough to give a large contribution to the spin relaxation,
its transformation into products cannot be the over-all
rate-determining step [13]. This would fit well with an
alkyl hydride, which according to the computational
results by Landis et al. is formed in the rate-determining
step. Obviously, any other intermediate after H2 addition
must contribute less to the relaxation. In the dihydrogen
complex the symmetry has not been broken and so this
gives no relaxation and presumably the dihydride also
has a short life-time possibly due to a reversible character
of the oxidative addition found in earlier computational
results [13]. The lifetime of 8 corresponds to a reaction
barrier of approximately 14 kcal/mol which is reasonable
for a reductive elimination step. It must be noted that
although we can link this intermediate directly to the cat-
alytic cycle the lack of coupling and other features of its
1H NMR spectrum makes any assignment far from
definite.
0.4
0
-0.4
3
2
1
0
δ/ppm
Fig. 5. Indirect 1H NMR spectrum of the hydrogenation intermediate
using as catalyst. [Rh] = 1.06 mM; [DMM]0 = 40.0 mM; and
T = 298 K.
4
1
frequency increments, giving the H NMR spectra shown
in Figs. 4 and 5 for complexes 3 and 4, respectively. No
polarisation of the 13C substrate signal was observed.
The fact that we see no polarisation of the substrate
peak as observed in some other systems, indicates that
there is no reversibility after the first C–H bond has been
formed. Furthermore, a dihydride mechanism (as opposed
to a mono-hydride mechanism) is operating also in water,
i.e. the two hydrogens are transferred to the same product
molecule and there is negligible proton exchange with the
solvent. The spectra in Figs. 4 and 5 could in principle arise
from J-coupling, but this would require a coupling con-
stant of ca. 240 Hz [12]. It seems more likely that the spec-
trum is due to a single, broad, anti-phase signal centred at
1.9 and 2.0 ppm when using 3 and 4 as pre-catalysts,
respectively; this cannot correspond to a dihydride interme-
diate. Instead we see a species where at least one hydrogen
has been transferred and one C–H bond is fully formed.
This leaves us with at least two possibilities: an alkyl
hydride species (shown below (8)) or a catalyst-succinate
complex.
In conclusion, we propose an unsaturated mechanism
with fast substrate binding and a rate-determining step
involving dihydrogen for the aqueous hydrogenation of
dimethylmaleate using water-soluble rhodium cations.
We also present a novel method for the indirect detec-
tion of intermediates in catalytic hydrogenation
reactions.
Acknowledgements
Financial support from the Swedish Research Council,
the Crafoord Foundation, the Knut and Alice Wallenberg
Foundation and the Royal Physiographic Society in Lund
is gratefully acknowledged.
Appendix A. Supplementary material
Tables and figures giving all primary kinetic and NMR
data. This material can be found, in the online version, at
doi:10.1016/j.ica.2006.08.042.
OMe
D
O
C
H
D
H
P
P
13C
O
References
Rh
OMe
[1] R. Noyori, Asymmetric Catalysis in Organic Synthesis, John Wiley &
Sons Inc., New York, NY, 1994.
8
[2] (a) Y. Amrani, L. Lecomte, D. Sinou, J. Bakos, I. Toth, B. Heil,
Organometallics 8 (1989) 542;
A succinate complex is presumably very short-lived3
and we therefore believe that what we see is the signal
from the hydrogen atom sitting next to a 13C carbonyl
in an alkyl hydride species. The peaks are about 100–
(b) I. Toth, B.E. Hanson, M.E. Davis, Tetrahedron Asymmetry 1
(1990) 895;
(c) L. Lecomte, D. Sinou, J. Bakos, I. Toth, B. Heil, J. Organomet.
Chem. 370 (1989) 277;
(d) D. Sinou, Adv. Synth. Catal. 344 (2002) 221.
[3] T. Malmstro¨m, O.F. Wendt, C. Andersson, J. Chem. Soc., Dalton
Trans. (1999) 2871.
3
A succinate complex in dynamic equilibrium with the polarised
[4] (a) J. Halpern, Science 217 (1982) 401;
product would affect the line shape of the free succinate. However, it is
observed to have sharp lines with resolved coupling.
(b) C.L. Landis, J. Halpern, J. Am. Chem. Soc. 109 (1987) 1746.

M. Ahlquist et al. / Inorganica Chimica Acta 360 (2007) 1621–1627
1627
´
˚
[5] K. Golman, O. Axelsson, H. Johannesson, S. Mansson, C. Olofsson,
[11] (a) C.R. Bowers, D.P. Weitekamp, J. Am. Chem. Soc. 109 (1987)
5541;
J.S. Petersson, Magn Reson Med 46 (2001) 1.
˚
´
[6] K. Golman, O. Axelsson, H. Johannesson, S. Mansson, C. Olofsson,
(b) For a recent review see: S.D. Duckett, D. Blazina, Eur. J. Inorg.
Chem (2003) 2901.
J.S. Petersson, WO 99/24080.
[7] R. Giernoth, H. Heinrich, N.J. Adams, R.J. Deeth, J. Bargon, J.M.
Brown, J. Am. Chem. Soc. 122 (2000) 12381.
[8] R. Uson, L.A. Oro, M.A. Garralda, M.C. Claver, P. Lahuerta,
Trans. Met. Chem. 4 (1979) 55.
[9] A. Bootle-Wilbraham, S. Head, J. Longstaff, P. Wyatt, Tetrahedron
Lett. 40 (1999) 5267.
[10] I.D. Gridnev, T. Imamoto, Acc. Chem. Res. 37 (2004) 633.
[12] An unsymmetric, end-on dihydrogen complex could in principle be
responsible for such a coupling constant and would in that case give
rise to considerable dipolar relaxation, but usually dihydrogen
complexes resonate at much higher field: G.J. Kubas, Acc. Chem.
Res. 21 (1988) 120.
[13] C.R. Landis, P. Hilfenhaus, S. Feldgus, J. Am. Chem. Soc. 121 (1999)
8741.