Investigating the kinetics of homogeneous hydrogenation reactions using PHIP NMR spectroscopy

Article abstract of DOI:10.1021/ja984353y

The combination of parahydrogen induced polarization (PHIP), kinetics and NMR spectroscopy yields a powerful analytical tool: quantitative in situ NMR spectroscopy. Two versions of PHIP NMR experiments are presented to investigate the kinetics of homogeneously catalyzed hydrogenations. The first method, an experimental variation of the ROCHESTER experiment (ROCHESTER = rates of catalytic hydrogenation estimated spectroscopically through enhanced resonances), allows one to determine the hydrogenation rate independently of relaxation and other sources of decay, e.g., subsequent chemical reaction steps. The second method named DYPAS (dynamic PASADENA spectroscopy) uses a variable delay between the end of the hydrogen-addition period and the detection pulse. In principle, all processes during this delay can be described by a set of coupled differential equations. Their solutions can be fitted to the experimental data by a least-squares optimization of the involved kinetic parameters. The DYPAS method can be used to determine the rates of formation as well as the rates of decomposition of stable intermediates and has been applied to the case of freshly hydrogenated and still catalyst-attached product molecules. We provide kinetic data for the formation and decomposition of these unusual product-catalyst complexes during the hydrogenation of different styrene derivatives with a cationic Rh¹ catalyst containing a chelating diphosphine ligand. The kinetic measurements indicate that the rate of formation of the catalyst-attached product increases whereas the rate constant of its decomposition diminishes if the para position of the arene ring of styrene carries an electron- donating substituent. In the case of p-aminostyrene as the substrate, the detachment step turned out to be rate limiting for the catalytic cycle. With certain substituted styrenes and cationic Rh¹ complexes containing chiral chelating diphosphine ligands, two geometrically different (diastereomeric) product-catalyst adducts can be discriminated via PHIP NMR spectroscopy. The associated alternative reaction pathways have been analyzed by applying the DYPAS method, which can also be used to investigate the mechanism of an asymmetric hydrogenation.

Full text of DOI:10.1021/ja984353y

J. Am. Chem. Soc. 1999, 121, 5311-5318
5311
Investigating the Kinetics of Homogeneous Hydrogenation Reactions
Using PHIP NMR Spectroscopy
Patrick Hu1bler, Ralf Giernoth, Gu1nther Ku1mmerle, and Joachim Bargon*
Contribution from the Institute of Physical and Theoretical Chemistry, UniVersity of Bonn,
Wegelerstrasse 12, D-53115 Bonn, Germany
ReceiVed December 16, 1998. ReVised Manuscript ReceiVed April 12, 1999
Abstract: The combination of parahydrogen induced polarization (PHIP), kinetics and NMR spectroscopy
yields a powerful analytical tool: quantitatiVe in situ NMR spectroscopy. Two versions of PHIP NMR
experiments are presented to investigate the kinetics of homogeneously catalyzed hydrogenations. The first
method, an experimental variation of the ROCHESTER experiment (ROCHESTER ) rates of catalytic
hydrogenation estimated spectroscopically through enhanced resonances), allows one to determine the
hydrogenation rate independently of relaxation and other sources of decay, e.g., subsequent chemical reaction
steps. The second method named DYPAS (dynamic PASADENA spectroscopy) uses a variable delay between
the end of the hydrogen-addition period and the detection pulse. In principle, all processes during this delay
can be described by a set of coupled differential equations. Their solutions can be fitted to the experimental
data by a least-squares optimization of the involved kinetic parameters. The DYPAS method can be used to
determine the rates of formation as well as the rates of decomposition of stable intermediates and has been
applied to the case of freshly hydrogenated and still catalyst-attached product molecules. We provide kinetic
data for the formation and decomposition of these unusual product-catalyst complexes during the hydrogenation
of different styrene derivatives with a cationic RhI catalyst containing a chelating diphosphine ligand. The
kinetic measurements indicate that the rate of formation of the catalyst-attached product increases whereas the
rate constant of its decomposition diminishes if the para position of the arene ring of styrene carries an electron-
donating substituent. In the case of p-aminostyrene as the substrate, the detachment step turned out to be rate
limiting for the catalytic cycle. With certain substituted styrenes and cationic RhI complexes containing chiral
chelating diphosphine ligands, two geometrically different (diastereomeric) product-catalyst adducts can be
discriminated via PHIP NMR spectroscopy. The associated alternative reaction pathways have been analyzed
by applying the DYPAS method, which can also be used to investigate the mechanism of an asymmetric
hydrogenation.
Introduction
The main advantage of the method is the associated strong signal
enhancement (in the order of 103) that qualifies it as a powerful
and versatile in situ method. As a prerequisite for the PHIP
effect to occur, both former parahydrogen nuclei must be
J-coupled in the resulting molecule of interest. The PHIP effect
allows one to monitor the fate of the two former parahydrogen
nuclei during the reaction via spin dynamics.11,12 Using this in
situ method, the hydrogenation of styrene derivatives with
cationic RhI catalysts has been shown to be partially reversible.13
In other cases, hydrogen-containing intermediates were de-
tected14 and reaction products of parahydrogen and organome-
tallic compounds were characterized.15,16
In consequence of the importance for the synthesis of
pharmacologically relevant substances, catalytic homogeneous
hydrogenations using transition metal complexes are still a field
of considerable interest.1,2 For example, cationic RhI complexes
containing bidentate phosphine ligands have developed into an
important class of catalysts that enable regio- as well as
stereoselective hydrogenations.3-6 Looking at the methodical
progress achieved during the past decade, PHIP NMR spec-
troscopy (PHIP ) parahydrogen induced polarization) has
developed into a powerful tool to investigate these reactions.7-10
(1) Brunner, H. In Applied Homogeneous Catalysis with Organometallic
Compounds; Cornils, B., Herrmann, W. A., Eds.; VCH: Weinheim,
Germany, 1996; pp 201-219.
In previous applications of PHIP NMR spectroscopy, the main
emphasis has been on qualitatiVe investigations of homoge-
(2) Parshall, G. W.; Ittel, S. D. Homogeneous Catalysis; Wiley: New
York, 1992; pp 25-39.
(10) Natterer, J.; Bargon, J. Prog. Nucl. Magn. Reson. Spectrosc. 1997,
31, 293-315.
(3) Schrock, R. R.; Osborn J. A. J. Am. Chem. Soc. 1976, 98, 2134-
2147.
(4) Schrock, R. R.; Osborn J. A. J. Am. Chem. Soc. 1976, 98, 4450-
4455.
(5) Brown, J. M. Angew. Chem., Int. Ed. Engl. 1987, 26, 190-203.
(6) Halpern, J. In Asymmetric Synthesis (Chiral Catalysis); Morrison, J.
D., Ed.; Academic Press: London, 1985; Vol. 5, pp 41-69.
(7) Eisenberg, R. Acc. Chem. Res. 1991, 24, 110-116.
(8) Eisenberg, R. J. Chinese Chem. Soc. 1995, 42, 471-481.
(9) Bargon, J. In Applied Homogeneous Catalysis with Organometallic
Compounds; Cornils, B., Herrmann, W. A., Eds.; VCH: Weinheim,
Germany, 1996; pp 672-683.
(11) Hu¨bler, P.; Natterer, J.; Bargon J. Ber. Bunsen-Ges. Phys. Chem.
1998, 102, 364-369.
(12) Buntkowsky, G.; Bargon, J.; Limbach, H. H. J. Am. Chem. Soc.
1996, 118, 8677-8683.
(13) Harthun, A.; Giernoth, R.; Elsevier, C. J.; Bargon, J. J. Chem. Soc.,
Chem. Commun. 1996, 2483-2484.
(14) Harthun, A.; Kardyrov, R.; Selke, R.; Bargon, J. Angew. Chem.,
Int. Ed. Engl. 1997, 36, 1103-1105.
(15) Duckett, S. B.; Newell, C. L.; Eisenberg, R. J. Am. Chem. Soc.
1994, 116, 10548-10556.
(16) Jang, M.; Duckett, S. B.; Eisenberg, R. Organomatallics 1996, 15,
2863-2865.
10.1021/ja984353y CCC: $18.00 © 1999 American Chemical Society
Published on Web 05/22/1999

5312 J. Am. Chem. Soc., Vol. 121, No. 22, 1999
Hu¨bler et al.
and solvents were used as obtained from Aldrich. The hydrogenation
reactions conducted inside of the NMR magnet were performed at 298
K using a conventional 1H-multinuclear probehead. The hydrogen
addition took place during a time interval of 3 s controlled by the
spectrometer console via pulse programming. During this time interval,
a glass capillary was lowered into the probehead and para-enriched
hydrogen gas was bubbled through the reactive solution. The spectra
were recorded in a nonspinning mode after a delay of at least 1 s
between the end of the hydrogen addition and the 45° detection pulse
(to avoid any inhomogeneity of the solution due to bubbles). All spectra
were recorded after any polarization stemming from the previous scan
had ceased. The enrichment of parahydrogen is achieved via passing
H2 through activated charcoal at 77 K and has been described
elsewhere.18,24 By this means, a steady stream of a nearly 50:50 mixture
of ortho- and parahydrogen is obtained. Under these conditions, one-
third of the parahydrogen does not contribute to the PHIP intensity as
this fraction is compensated by the residual orthohydrogen. This
corresponds to a net surplus of 33% parahydrogen.
To determine the intensities of the PHIP signals, the NMR software
Nuts was used. The antiphase signals were transformed into inphase
signals using the magnitude calculation mode. By this means, the
integrals of the transformed signals represent the intensities of the PHIP
patterns. To determine values (and their accuracy) for the kinetic
parameters involved in the different reaction pathways, the latter were
transformed into systems of coupled differential equations that were
solved analytically using the software package Mathematica. The
experimental data were fitted to these analytical solutions by a least-
squares optimization performing the Marquardt-Levenberg algorithm
implemented in Sigmaplot. The errors represent the asymptotic standard
errors.
neously catalyzed hydrogenation reactions. For a more detailed
understanding of catalytic cycles, however, it is important to
know not only the structure of possible intermediates but also
the rates of their formation and decomposition. Kinetic methods
may suggest indirectly the occurrence of certain reaction steps,
whereas the short lifetime of many intermediates frequently
prevents their direct detection.17 Because kinetic methods have
already proved to be essential for the understanding of homo-
geneous catalysis, it is highly attractive to combine them with
the PHIP effect for quantitatiVe in situ investigations. So far,
only two experiments have been proposed to determine kinetic
parameters via PHIP NMR spectroscopy: One approach
combines a continuous-flow NMR technique and the PHIP effect
to measure the intensity of polarization under steady-state
conditions.18,19 Accordingly, during the hydrogenation of 1,4-
diphenylbuta-1,3-diyne with [Rh(PPh3)2(nbd)]PF620 as the cata-
lyst precursor, the rates of several reaction steps were deter-
mined. This technique provides detailed kinetic information, but
it requires a special continuous-flow probehead and the pos-
sibility of varying the pressure of H2 over a wide range.
According to the second approach, the rate of product formation
can be measured via monitoring the decay of the PHIP intensity
that frequently follows first-order kinetics if certain mechanistic
and experimental conditions are met. As an example, the
hydrogenation of ethyl (Z)-R-acetamidocinnamate with [Rh-
(chiraphos)(nbd)]BF4 in CD3OD has been investigated in this
fashion.21
We propose an analogous experiment that is also based upon
satisfying pseudo-first-order conditions during hydrogenation
reactions. However, our experimental setup uses a spectrometer-
controlled hydrogen addition to the reactive solution, thereby
bypassing the problem of a relayed phase transfer of hydrogen
from the gas phase into the solution which otherwise prevents
the determination of an accurate hydrogenation rate. In addition,
the use of conventional NMR equipment provides the possibility
of controlling the temperature. Furthermore, the theoretical
framework of this experiment provided by R. Eisenberg et al.21
has been generalized.
Recently, we described the observation of still catalyst-
attached product molecules during the hydrogenation of styrene
derivatives with cationic RhI catalysts containing chelating
diphosphine ligands.22 The detection of this special type of
intermediate with the help of in situ PHIP NMR spectroscopy
revealed that the rate of detachment is surprisingly low.
Therefore, a kinetic method is attractive to draw further
conclusions. We show that, in certain cases, the detachment step
is rate limiting for the catalytic cycle. We also present a suitable
reaction scheme to describe this effect, which is “translated”
into a set of coupled differential equations and validated
experimentally. In the analogous case of iridium complexes,
the corresponding product-catalyst adducts are known to be
stable such that they can even be isolated.23
Results and Discussion
Hydrogenation Rates Determined by the ROCHESTER
Experiment. PHIP NMR spectroscopy detects polarized product
molecules (or intermediates) during homogeneously catalyzed
hydrogenations. The PHIP effect produces strongly enhanced
antiphase signal patterns in the NMR spectrum. The polarized
hydrogenation product is represented by the density operator
σPHIP ) IZ1IZ2 (in the case of weak coupling25), whereby the
numbers indicate the positions of the two former parahydrogen
nuclei.10 The density operator σPHIP can be transformed into
observable magnetization by a 45° detection pulse. The polar-
ization, represented by σPHIP ) IZ1IZ2, is the subject of
longitudinal relaxation. The PHIP intensity is determined both
by the chemical formation (according to the rate kHYD) and
relaxation. A detailed description of the various relaxation
processes is currently in preparation but not the subject of this
study. To determine kHYD, it turned out to be sufficient to use
a simplified description using only one resulting rate of
longitudinal autorelaxation (RZZ).
It is possible to determine the rate of hydrogenation, kHYD
,
independently of the relaxation rate and other subsequent
reaction steps. The experiment suited to reach this goal is based
upon the fact that polarization can be observed during a time
interval of several tens of seconds after the end of the hydrogen
addition. This fact is a consequence of a small hydrogenation
rate (kHYD) being mostly in the order of kHYD e 1 s-1 under
typical catalytic conditions. Obviously, this method is limited
to hydrogenations with kHYD e 1 s-1; therewith polarization is
observed during a reasonable period of time. Figure 1 outlines
the experimental procedure to determine kHYD independently
of the relaxation rate (RZZ) with a ROCHESTER-type experi-
Experimental Section
1
The H-PHIP NMR spectra were recorded on a Bruker DRX 200
spectrometer at a proton resonance frequency of 200 MHz. Reagents
(17) Halpern, J. Science 1982, 217, 401-407.
(18) Bargon, J.; Kandels, J.; Woelk, K. Z. Phys. Chem. 1993, 180, 65-
93.
(19) Woelk, K.; Bargon, J. Z. Phys. Chem. 1993, 182, 155-165.
(20) nbd ) norbornadiene.
(21) Chinn, M. S.; Eisenberg, R. J. Am. Chem. Soc. 1992, 114, 1908-
1909.
(24) Woelk, K.; Bargon, J. J. ReV. Sci. Instrum. 1992, 63, 3307-3310.
(25) This density matrix is the result of a hydrogenation conducted inside
of the NMR magnet, i.e., under PASADENA conditions (PASADENA )
parahydrogen and synthesis allow dramatically enhanced nuclear alignment);
see ref 10.
(22) Giernoth, R.; Hu¨bler, P.; Bargon, J. Angew. Chem., Int. Ed. Engl.
1998, 37, 2473-2475.
(23) Crabtree, R. H.; Mellea, M. F.; Quirk, J. M. J. Chem. Soc., Chem.
Commun. 1981, 1217-1218.

Kinetics of Homogeneously Catalyzed Hydrogenations
J. Am. Chem. Soc., Vol. 121, No. 22, 1999 5313
Figure 1. Modified ROCHESTER experiment to determine kHYD
independently of the relaxation rate (and independently of subsequent
reaction steps). The arrows indicate the up and down motion of the
capillary providing a stream of parahydrogen. The number n is typically
on the order of 10, and tREP should be at least 2T1.
Scheme 1a
Figure 2. Intensity of polarization (i.e., IZ1IZ2 ) can be determined
by integration of the signal group at the chemical shift of the nucleus
1 (or 2), δ1 (or δ2), in the magnitude calculation mode. As long as the
expectation values of other longitudinal two-spin order terms IZ1IZn
(n > 2) do not exceed IZ1IZ2 , they do not contribute to the integral
that, therefore, is proportional to IZ1IZ2 . The nuclei 1 and 2 are
the two former parahydrogen nuclei, whereas nucleus 3 is an addi-
tional proton that is involved in relaxation processes leading, e.g., to
IZ1IZ2 f -IZ1IZ3.
a S: substrate; Cat: active catalyst; S-HH: hydrogenation product.
mental setup (ROCHESTER21 ) rates of catalytic hydrogenation
estimated spectroscopically through enhanced resonances). To
avoid confusion, we use this acronym throughout this study
whenever the experiment as depicted in Figure 1 is used. As
described in the Experimental Section, hydrogen is added during
the time interval symbolized by the arrows in Figure 1, which
indicates the up and down motion of the capillary. During this
period of time, para-enriched hydrogen gas is bubbled through
the reactive solution at atmospheric pressure. The main advan-
tage of this experimental approach is that no hydrogen atmo-
sphere remains above the sample at the end of the H2 addition.
Therefore, the (most likely rate-determining) step of a relayed
hydrogen transfer from the gas phase into the solution is avoided
and excluded from consideration. Otherwise, this phase-transfer
step complicates the description of the kinetics which then is
not governed by a first-order decay of [H2] anymore.
Scheme 2b
b S: substrate; Cat: active catalyst; S-HH*: polarized product
molecule.
δ1 is not altered by two-spin order terms IZ1IZn with n * 1, 2.
In the example depicted in Figure 2, the decay of IZ1IZ2 is
dominated by the rate of longitudinal autorelaxation, i.e.,
ZZ
d IZ1IZ2 /dt ≈ R1 IZ1IZ2 . All processes leading to IZ1IZ2
f
IZ1IZ3, i.e., cross-relaxation and cross-correlated interactions,
have been neglected, since the corresponding terms are expected
ZZ
As already outlined,21 the starting point for the theoretical
framework of this experiment is an approach of J. Halpern et
al.26,27 assuming the mechanism depicted in Scheme 1 for a
homogeneously catalyzed hydrogenation, i.e., the product
formation follows a pseudo-first-order kinetics if the substrate
S is present in large excess. This type of mechanism frequently
applies if the hydrogenation is catalyzed by a cationic RhI
complex containing a chelating diphosphine ligand.26 According
to this mechanism, the kinetics of the product (S-HH) formation
can be described by eq 1,
to be small relative to R1 IZ1IZ2 . Hence, the PHIP intensity
as a function of the time can be described by a two-step reaction
as indicated in Scheme 2. The time dependence of [S-HH*]
(that is proportional to IZ1IZ2 ) is given by eq 2, where [H2*]0
is the concentration of parahydrogen at t ) 0.
k
HYD[H2*]0
ZZ - kHYD
[S-HH *](t) )
(exp(-kHYDt) - exp(-RZZt))
R
(2)
In a related study, it was pointed out that eq 2 yields a first-
k1K
d[S-HH]
dt
21
order decay for [S-HH*] if RZZ . kHYD
.
However, this
)
[S][Cat]TOT[H2]
K[S] + 1
relation is not a necessary requirement for the experiment and
the theory since RZZ is possibly of the same order of magnitude
as kHYD (e.g., 0.5 s-1). Nevertheless, eq 2 can be applied to the
ROCHESTER experiment if the main feature of the PHIP
method is taken into account. During each FID, only the
polarization is detected that has been formed just a few seconds
before. Therefore, each detection pulse reads out the current
“compromise” between formation and relaxation. Simulta-
neously, further polarization is generated independently during
the nth cycle and is transformed into observable magnetization
by the (n+1)th detection pulse. The “repetition time” tREP
(typically in the order of 10 s) is constant, whereas the
concentration [H2*] is not because parahydrogen is consumed
by the chemical reaction independently of the NMR experiment.
As explained above, the concentration [H2*] follows a first-
order exponential decay. Therefore, the expression [H2*]0 in
eq 2 is now a function of the time and has to be replaced by
the expression [H2*]00 exp(-kHYDt), whereas t is now repre-
sented by the constant tREP. Thus, the time dependence of the
) k*HYD[Cat]TOT[H2]
) kHYD[H2]
(1)
where [Cat]TOT ) [Cat] + [S-Cat]. To obtain a pseudo-first-
order kinetics, an excess of the substrate S is required. This
condition has been met for all experiments throughout this study.
To describe the time dependence of the PHIP intensity,
another “reaction” must be considered: the relaxation of σPHIP
) IZ1IZ2. Neglecting cross-relaxation as well as cross-correlated
effects between two dipolar interactions, the decay of IZ1IZ2 is
mainly determined by the rate of longitudinal two-spin order
relaxation, RZZ. As outlined in Figure 2, IZ1IZ2 can be
determined by integration of the associated signals in the
magnitude calculation mode. The intensity of a signal group at
(26) Halpern, J.; Riley, D. P.; Chan, A. S. C.; Pluth, J. J. J. Am. Chem.
Soc. 1977, 99, 8055-8057.
(27) Landis, C. R.; Halpern, J. Organometallics 1983, 2, 840-842.

5314 J. Am. Chem. Soc., Vol. 121, No. 22, 1999
Hu¨bler et al.
Table 1. Rates for the Hydrogenation of Different Acetylene
Derivatives Determined Using the ROCHESTER Experiment:
Phenylacetylene (1), 3,3-Dimethylbut-1-yne (2),
kinetics. The polarization signals of interest can be caused either
by the hydrogenation product or by an intermediate with a
lifetime sufficient to observe NMR resonances. An example for
the latter case is presented in the subsequent section, where the
generation of catalyst-attached product molecules is shown to
follow a pseudo-first-order kinetics. The use of cationic RhI
complexes throughout this study does not mean that all systems
must follow the “unsaturated route” 1,2 of olefin hydrogenation.
The experimental conditions can, in principle, be varied so that
complementary pseudo-first-order cases are reached. We want
to mention briefly a few cases. (1) Systems that follow the
“unsaturated route”, whereby the substrate occurs in excess:
This leads to eq 1 and a first-order decay of polarization, as
discussed above. (2) Systems following the “unsaturated route”,
whereby hydrogen occurs in excess, and K[S],1: In this case,
the rate law can be approximated by
(Trimethylsilyl)acetylene (3), and (Triphenylsilyl)acetylene (4)a
substrate
R-
Ph-
(CH3)3C-
(CH3)Si-
(Ph)3Si-
kHYD (s-1
)
1
2
3
4
0.140 ( 0.005
0.085 ( 0.002
0.064 ( 0.004
0.085 ( 0.003
a All measurements were performed using a total catalyst concentra-
tion ([Cat]TOT ) [Cat] + [S-Cat], see Scheme 1) of [Cat]TOT ) 14.16
mM (catalyst precursor [Rh(dppb)(cod)]BF4 in acetone-d6). The values
of kHYD were determined by linear regression.
PHIP intensity (that is proportional to [S-HH*]) is obtained
as
k1K
d[S-HH]
dt
)
[S][Cat]TOT[H2]excess
K[S] + 1
[S-HH*](t) )
≈ k1K[Cat]TOT[H2]excess[S]
) kHH2Y-Dexc[S]
k
HYD[H2*]00
) - exp(-RZZtREP))
(5)
exp(-kHYDt)(exp(-kHYD REP
t
R
ZZ - kHYD
We are currently working on implementing a method experi-
mentally in which the polarization decay is substrate-controlled.
If both complementary pseudo-first-order conditions can be
achieved, the relation
(3)
This expression is equivalent to
[S-HH*](t) exp(-kHYDt)
(4)
S-exc
k
HYD
1
To investigate whether the experimental intensities as a
function of t ) ntREP can be described by eq 4, the following
substrates were “tested”: phenylacetylene (1), 3,3-dimethylbut-
1-yne (2), (trimethylsilyl)acetylene (3), and (triphenylsilyl)-
acetylene (4). [Rh(dppb)(cod)]BF428 in acetone-d6 was used as
the catalyst precursor. The experimental results are summarized
in Table 1. The small standard deviations indicate that the
mechanistic model as well as the experiment are suited to
describe and investigate the above reactions.
As an in situ method, PHIP NMR spectroscopy always yields
the “true” rates associated with the formation of the product
molecules that have actually been formed by a mechanism
transferring the parahydrogen nuclei in pairs. Other competing
reaction pathways are possible, but if those are “PHIP-silent”,
no perturbation results from these processes. To investigate
whether the rate of product formation as determined by the PHIP
method is actually associated with the dominant reaction
pathway, the results can be compared to the turnover rate as
measured by conventional NMR spectroscopy. On the other
hand, the PHIP method yields in situ values, and therefore, any
changes of the rates can be monitored as a function of the
turnover.
We want to note that “conventional” in-phase NMR signals
cannot be used to calibrate the integrals of the PHIP patterns,
since the former are possibly subject to a saturation effect if
tREP is chosen so that they do not have enough time to recover
completely before the next cycle begins. The PHIP patterns, of
course, do not have to recover but are created freshly by the
hydrogenation.
Application of the ROCHESTER Method to Other
Mechanistic and Experimental Conditions. The method
outlined in the preceding section, more precisely the way of
data analysis used, is limited to cases in which the rate of
production of the PHIP signals follows a pseudo-first-order
)
(6)
H2-exc
HYD
K[H2]excess
k
holds (if, of course, the same mechanism applies in both cases).
This would provide for an interesting method to determine K
in situ which, otherwise, is difficult to achieve using conven-
tional kinetic methods. (3) Systems following the “hydride
route”: As a typical system, the catalytic properties of Wilkin-
sons’s catalyst were investigated29-31 and the decay of the
observable dihydride concentration, [RhH2ClL3] (L ) PPh3),
was measured by a kinetic study in the presence of a large excess
of cyclohexene. These experimental conditions lead to a pseudo-
first-order rate law.29 This example shows that it is possible to
“implement” simple kinetic laws by a careful adjustment of the
experimental conditions. Current work focuses on the imple-
mentation of accurate conditions to use PHIP for the determi-
nation of hydrogenation rates more universally by considering
several different mechanisms.
The ortho/para equilibration in the static magnetic field during
homogeneous hydrogenations can be regarded by a density
matrix description of an intermediate state involving a Hamil-
tonian that does not commute with the density matrix of
parahydrogen, i.e., causing a symmetry breakdown in the
intermediate state.11 If the hydrogenation yields a weakly
coupled product spin system, all information about intermediate
states gets lost and only the operator IZ1IZ2 remains. In contrast
to the isotropic density matrix of parahydrogen, the expectation
value of this operator decreases as a function of time due to
dipolar relaxation. The relaxation of IZ1IZ2 in the product
molecule is dominated by RZZ, whereas relaxation during the
lifetime of the intermediate state leads to a competing “reaction
(29) Halpern, J.; Okamoto, T.; Zakhariev, A. J. Mol. Catalysis 1976, 2,
65-68.
(30) de Croon, M. H. J. M.; van Nisselrooij, P. F. M. T.; Kuipers, H. J.
A. M.; Coenen, J. W. E. J. Mol. Catalysis 1978, 4, 325-335.
(31) Halpern, J. Inorg. Chim. Acta 1981, 50, 11-19.
(28) dppb ) diphenylphosphinobutane, cod ) cycloocta-1,5-diene.

Kinetics of Homogeneously Catalyzed Hydrogenations
J. Am. Chem. Soc., Vol. 121, No. 22, 1999 5315
Figure 3. Operators produced by a 45° detection pulse applied to
σPHIP ) IZ1IZ2. ZQ: zero quantum coherence, SQ: single quantum
coherence, DQ: double quantum coherence.
pathway” and thereby to an effective hydrogenation rate that is
diminished depending on the lifetime of the intermediate, or, if
the intermediate state is passed through due to a fast equilibrium
(nonproductive parahydrogen addition and elimination), depend-
ing on the fraction of the intermediate state. The effect thereof
is expected to be negligible in most cases, since the lifetime
and hence the population of an intermediate is usually small.
Accordingly, the small conversion rates as determined by Brown
et al.32 (on the order of 10-4 s-1 in solution) are small relative
to the duration of the experiment.
To finish the description of the ROCHESTER experiment,
we want to discuss qualitatively the influence of relaxation. A
necessary condition for this experiment is that all coherences
produced by the nth detection pulse have actually decayed to
zero such that they do not contribute to the magnetization
produced by the (n + 1)th detection pulse. Figure 3 summarizes
the different operators produced by a single 45° detection pulse.
The 45° pulse transforms σPHIP into the coherences ZQ, SQ,
and DQ corresponding to the coherence order p ) 0, 1, and 2,
respectively. These coherences are affected by relaxation during
the nth FID and during a subsequent delay. The transversal
relaxation of the coherences SQ and DQ is most likely
dominated by the inhomogeneity of the external magnetic field.
Figure 4. Reaction sequence for the formation and decomposition of
catalyst-attached ethylbenzene derivatives. The asterisks mark the
positions of the two former parahydrogen nuclei. Solvent molecules in
the coordination sphere have been neglected.
investigated by R. H. Crabtree et al. to apply to hydrogenation
reactions catalyzed by Ir complexes, where the analogous
product-catalyst adducts turned out to be stable such that they
can be isolated.23
When using a cationic RhI complex containing a chiral
chelating diphosphine ligand, ethylbenzene derivatives without
a C2 axis in the molecular plane (the same holds for the
corresponding styrene derivatives) form two diastereomeric
product-catalyst complexes. Accordingly, two geometrically
different adducts can be discriminated via separated resonances
in their PHIP NMR spectrum. A measurement of the rates of
their formation and decomposition should allow to monitor the
effectiveness of the chiral catalyst and, therefore, yield informa-
tion about the origin of enantioselection. Such studies are
currently in progress.
The operators 0.25ZQX ) 0.25(IX1 X2 + IY1IY2) and 0.5IZ1IZ2
I
possibly relax more slowly and may affect the (n+1)th FID if
the subsequent delay is not long enough. The sum of both is
isotropic except for a term R(t)0.25IZ1IZ2, where R(t) defines
the fraction that has not decayed to zero at the beginning of the
(n + 1)th cycle. Only this fraction is transformed into further
observable magnetization. Assuming that all operators decay
with the same rates, the two subsequent 45° pulses can be
combined to one 90° pulse which does not lead to any detectable
magnetization. The influence of incomplete relaxation is
As already mentioned, the distinct observation of these
catalyst-attached product molecules is a consequence of a small
detachment rate. To be visible as individual and nonbroadened
signals in the NMR spectrum, the intermediate must be stable
with respect to the time scale as defined by the amount of the
shift of the corresponding resonances (typically on the order of
20-60 Hz at a proton resonance frequency of 200 MHz). The
intermediate can hardly be detected by conventional NMR
spectroscopy in the thermodynamic equilibrium because the
corresponding concentration is very low. Accordingly, we take
advantage of the unique ability of the PHIP NMR method: the
accumulation of intermediate molecules during the transient
chemical reaction and, simultaneously, the generation of a
magnetic nonequilibrium state that yields a large signal en-
hancement. A few seconds after the hydrogen addition has
stopped, the chemical reaction has accumulated “sufficient”
intermediate molecules in the solution for detection by PHIP if
their decomposition rate is not too high. Nevertheless, this
concentration is frequently still much too low for the detection
by conventional NMR methods. Therefore, it is possible to
measure rates associated with the formation and decomposition
of such intermediates only as a consequence of the PHIP signal
enhancement.
expected to be negligible, however, if tREP is on the order of
ZZ
2-3 T1
.
Determining the Rate of Product Detachment from the
Catalyst Using the DYPAS Method. Recently, we reported
the detection of a special type of intermediate that occurs during
homogeneously catalyzed hydrogenations: freshly hydrogenated
product molecules still being attached to a cationic RhI catalyst
via an arene ring.22 The rate of the detachment step turned out
to be relatively low, and as a consequence thereof, these
intermediates become visible in the PHIP NMR spectrum as
resonances shifted to higher fields. This type of intermediate
can be observed with a styrene derivative as the substrate and
a cationic RhI complex containing a chelating diphosphine
ligand. The rate of product formation may be strongly influenced
by this “additional” reaction step that can be the rate limiting
and, hence, the decisive one. This case has already been
Figure 4 explains the occurrence of the PHIP signals
corresponding to free and attached ethylbenzene derivatives, and
it defines the model used to quantitatively describe the expected
and observed time dependence of the PHIP intensities. As shown
(32) Brown, J. M.; Canning, L. C. J. Organomet. Chem. 1983, 255, 103-
111.

5316 J. Am. Chem. Soc., Vol. 121, No. 22, 1999
Hu¨bler et al.
Figure 5. “Pulse sequence” of the DYPAS experiment to determine
the rates of all processes leading to the formation and the decay of a
PHIP NMR signal. The delay time τ is varied between 1 and
approximately 30 s, depending on the rates of the involved processes.
Figure 7. DYPAS intensities of attached (squares) and free (triangles)
ethylbenzene during the hydrogenation with the catalyst precursor [Rh-
(dppb)(cod)]BF4 in acetone-d6. The symbols represent the experimental
values, and the lines are the optimized fits according to the eqs 9 and
10. The best fit has been achieved with kHYD ) 0.111 ( 0.009 s-1
,
kOFF ) 0.351 ( 0.003 s-1, RF ) 0.20 ( 0.02 s-1, and RB ) 1.23
ZZ
ZZ
( 0.25 s-1. The high value for RB (relative to RFZZ) is in good
ZZ
agreement with the expectation of an increased correlation time in the
bound state.
Figure 6. Theoretical course of the DYPAS intensities during the
hydrogenation of styrene derivatives as a function of the delay time τ
(according to the eqs 9 and 10). The parameters kOFF and RZZ are fixed
to kOFF ) 0.15 s-1 and RZZ ) 0.1 s-1. Solid lines: attached product,
dashed lines: free product. The numbers represent different values for
Table 2. Experimental Values for the Rates kHYD and kOFF
Obtained during the Hydrogenation of Different Styrene Derivatives
Performing the DYPAS Experiment Depicted in Figure 5a
kHYD
.
in Figure 5, DYPAS is proposed to determine the detachment
rate (kOFF) as well as the hydrogenation rate (kHYD). According
to Figure 4, the processes that take place during the delay τ
can be described by the following system of coupled differential
equations:
X-
-H
-OMe
-OEt
-NH2
kHYD (s-1
)
kOFF (s-1
)
0.111 ( 0.009
0.138 ( 0.010
(0.40)
0.35 ( 0.03
0.45 ( 0.04
0.273 ( 0.024
0.113 ( 0.036
(0.84)
-kHYD
0
0
0
[H2*]
InB
[H2*]
InB
-kOFF + RZBZ
d
dt
a The total catalyst concentration was 5.52 mM in acetone-d6. The
errors are the standard errors of the least-squares optimization. The
numbers in brackets indicate experimental values with a high uncer-
tainty, even though the fits are very good.
kHYD
)
(7)
(
) (
)(
)
-RZFZ
InF
InF
kOFF
0
In eq 7, InB and InF are the PHIP intensities of bound and free
ethylbenzene, respectively, and RBZZ and RFZZ are the associated
rates of longitudinal relaxation. For simplicity, any constant
factor relating the concentration and the intensity has been
disregarded. The solutions of eq 7 under the boundary conditions
[H2](0) ) [H2]0 and InB(0) ) InF(0) ) 0 are
diminishes if the rate of relaxation increases. In the case of a
low production rate, the overall intensity is low whereas the
polarization can be observed during a longer period of time.
To show that the model defined in Figure 4 is well-suited to
describe the PHIP intensity as a function of time, several para-
substituted styrenes have been investigated using the catalyst
precursor [Rh(cod)(dppb)]BF4 in acetone-d6. As a characteristic
example, the experimental intensities observed during the
hydrogenation of p-methoxystyrene are shown in Figure 7
(together with the least-squares fit according to the eqs 9 and
10). The results of the least-squares optimizations are sum-
marized in Table 2. Substitution of the para position of styrene
by electron-density donating groups roughly causes two opposite
effects: (i) the hydrogenation rate increases whereas (ii) the
rate of product detachment decreases. We propose that both
effects are a consequence of the increasing bonding strength
between the arene ring (of the substrate and the product) and
the cationic RhI catalyst. This effect possibly causes an increase
of K (and, therefore, kHYD; see eq 1) and most likely causes a
decrease of kOFF. The experimental data included in Table 2
suggest that the detachment process is also influenced by sterical
effects. Possibly, the rate of detachment is diminished in the
case of X ) H because this substituent is sterically less
pretentious than the other ones. To draw precise conclusions, it
is necessary to determine more experimental data including
activation parameters. It is important to realize that, in the case
[H2*](t) ) [H2*]0 exp(-kHYDt)
(8)
InB(t) )
kHYD[H2*]0
RZBZ + kOFF - kHYD
(exp(-kHYDt) - exp(-(RZBZ + kOFF)t))
(9)
InF(t) )
kOFF
RZFZ - RZBZ - kOFF
kHYD
kHYD - RZFZ
kHYD
kHYD - RZFZ
InB(t) +
[H2*](t) -
(
[H2*]0
exp(-RZFZt) (10)
)
The theoretical time dependence of the PHIP intensities is shown
in Figure 6. The higher the value of kOFF (or the smaller kHYD),
the later the intensity of the free product equals the intensity
associated with the bound molecules. The overall intensity

Kinetics of Homogeneously Catalyzed Hydrogenations
J. Am. Chem. Soc., Vol. 121, No. 22, 1999 5317
while hydrogenating 3-methylstyrene with [Rh((R)-(S)-JOSI-
PHOS)(cod)]BF433 as the catalyst precursor.22 We propose the
mechanism depicted in Figure 8 to quantitatively account for
the occurrence of the two product-catalyst resonances. The time
dependence of the PHIP intensity during the delay τ (according
to the DYPAS experiment depicted in Figure 5) can be described
by the following system of coupled differential equations:
-kΣ
0
kHYD,a
0
0
0
-RZFZ
kOFF,a
-kOFF,a - RZBZ
kOFF,b
0
0
0
(
)
0
-kOFF,b - RZBZ
kHYD,b
[H2*]
InF
InB,a
InB,b
[H2*]
InF
d
dt
)
(11)
InB,a
InB,b
(
)
(
)
where kΣ ) kHYD,a + kHYD,b. InB,a/b is the PHIP intensity
associated with the bound product molecules (i.e., the two
diastereomeric catalyst-product complexes a and b). InF is the
PHIP intensity associated with the polarized free product
ZZ
molecules, and RB/F represents the rate of longitudinal
relaxation. The solutions of this system of coupled differential
equations were derived by Mathematica under the boundary
conditions [H2](0) ) [H2]0 and InF(0) ) InB,a(0) ) InB,b(0) )
0:
InF ) [H2*]0
kHYD,akOFF,a
(kHYD,a + kHYD,b - kOFF,a - RBZZ)(kHYD,a + kHYD,b - RZFZ)
(
(exp(-(kHYD,a + kHYD,b)t) - exp(-RZFZt)) +
k
HYD,bkOFF,b
(kHYD,a + kHYD,b - kOFF,b - RBZZ)(kHYD,a + kHYD,b - RZFZ)
(exp(-(kHYD,a + kHYD,b)t) - exp(-RZFZt)) -
k
HYD,akOFF,a
(kHYD,a + kHYD,b - kOFF,a - RBZZ)(kOFF,a + RBZZ - RZFZ)
(exp(-(kOFF,a + RZBZ)t) - exp(-RFZZt)) -
Figure 8. Reaction scheme to explain the occurrence of two diaster-
eomeric product-catalyst complexes during the hydrogenation of
certain styrene derivatives with chiral catalysts. Cat$ symbolizes the
active chiral catalyst species, and the asterisks mark the positions of
the two former parahydrogen nuclei. The indices a and b indicate the
two different reaction pathways, whereas the indices B and F represent
rates associated with bound and free product molecules, respectively.
Solvent molecules in the coordination sphere have been neglected.
k
HYD,bkOFF,b
(kHYD,a + kHYD,b - kOFF,b - RBZZ)(kOFF,b + RBZZ - RZFZ)
(exp(-(kOFF,b + RZBZ)t) - exp(-RFZZt)) (12)
)
InB,a/b
)
of strongly electron-donating substituents (e.g., NH2), the
detachment step becomes rate limiting for the catalytic cycle.
To establish a link between the ROCHESTER and the
DYPAS experiment, kHYD can also be determined independently
of RZZ and kOFF according to the ROCHESTER experiment
depicted in Figure 1. In this case, [S-HH]*attached (polarization
stemming from attached product molecules) follows a first-order
decay according to the rate kHYD which does not change the
kHYD,a/b
[H2*]0
(kHYD,a + kHYD,b - kOFF,a/b - RZBZ)
(
(-exp(-(kHYD,a + kHYD,b)t) + exp(-(RZBZ + kOFF,a/b)t))
)
(13)
(14)
validity of eqs 3 and 4 if RZZ is replaced by k′ ) kOFF + RZZ
.
[H2*](t) ) [H2*]0 exp(-(kHYD,a + kHYD,b)t)
This has indeed been verified experimentally during the
hydrogenation of styrene with [Rh(cod)(dppb)]BF4 (c ) 3 mM).
Under these conditions, a value of kHYD ) (0.058 ( 0.001) s-1
was determined with the ROCHESTER experiment.
To verify the suitability of the mechanistic model as defined
by Figure 8, we used [Rh((R)-(S)-JOSIPHOS)(cod)]BF4 as the
catalyst precursor that was generated in situ by stirring an
As pointed out before, two diastereomeric product-catalyst
complexes are generated in the case of a chiral catalyst. They
can easily be discriminated via PHIP NMR spectroscopy, e.g.,
(33) (R)-(S)-JOSIPHOS ) (R)-1-[(1S)-2(diphenylphosphino)ferrocenyl]-
ethyldicyclohexylphosphine.

5318 J. Am. Chem. Soc., Vol. 121, No. 22, 1999
Hu¨bler et al.
ing detailed information about the background of the observed
hydrogenation rate. The second method, DYPAS, provides a
possibility to monitor all processes occurring during a variable
delay time. It can be used to determine the rates of various
relaxation processes as well as to measure the rates of the
formation and the decomposition of all visible intermediates
and product molecules. With this method, the occurrence of a
special type of intermediate, product-catalyst adducts, in the
PHIP NMR spectra recorded during the hydrogenation of styrene
derivatives with cationic RhI catalysts can be investigated. The
associated processes, namely the formation and decomposition
of these intermediates, have been described by sets of coupled
differential equations that have been solved and fitted to the
experimental data. The production rates of the product-catalyst
complexes as well as the kinetics of their detachment turned
out to be strongly influenced by the electronic properties of
substituents in the arene ring of ethylbenzene (the hydrogenation
product): Electron-donating substitution in the para position
increases the hydrogenation rate but decreases the detachment
rate. In the case of p-aminoethylbenzene, the detachment step
proved to be rate limiting for the turnover. We used DYPAS to
visualize two diastereomerically different product-catalyst
complexes in the case of a chiral catalyst and 3-methylstyrene
as the substrate. These two complexes are formed by two
“diastereomeric reaction pathways” that both yield the same
product molecule. Due to its ability to yield kinetic parameters,
this method should provide a possibility to screen and optimize
chiral catalysts: Different chiral catalyst systems can be
expected to cause more or less distinct differences between the
corresponding reaction pathways. These differences reflect and
indicate the potential and suitability of a catalyst complex for
an asymmetric induction. In summary, the results demonstrate
that PHIP NMR spectroscopy can be utilized as a powerful and
extraordinary kinetic in situ method to investigate homogeneous
hydrogenation reactions.
Figure 9. Experimental DYPAS intensities (symbols) vs delay time τ
during the hydrogenation of 3-methylstyrene with [Rh((R)-(S)-JOSI-
PHOS)(cod)]BF4 (c ) 0.0133 M). The best fit has been achieved with
the following parameters: kHYD,a ) (0.034 ( 0.002) s-1, kHYD,b ) (0.044
( 0.002) s-1, RB ) (1.03 ( 0.15) s-1, RF ) (0.88 ( 0.06) s-1
,
ZZ
ZZ
kOFF,a ) (0.48 ( 0.08) s-1, and kOFF,b ) (0.44 ( 0.06) s-1. Circles
represent the free product; triangles and squares represent the two
different product-catalyst complexes.
equimolare mixture of [Rh(cod)2]BF4 and (R)-(S)-JOSIPHOS
in acetone-d6 without the presence of any substrate. Thereafter,
3-methylstyrene was added and hydrogenated. As a result of a
DYPAS experiment, the PHIP intensities obtained as a function
of τ and the optimized multiparameter fits are shown in Figure
9. The good agreement reveals that the model as defined by
Figure 8 and eq 11 successfully fits the experimental PHIP
intensities. This strongly indicates that Figure 8 is well-suited
to describe the reactions governing the delay τ. The method
thus allows to differentiate quantitatively between two possible
diastereomeric reaction pathways. Further measurements with
other catalysts are currently under way to investigate the
differences between the rates corresponding to the two diaster-
eomeric alternatives. The results might be significant for
optimizing catalysts for the stereoselective hydrogenation of a
prochiral substrate.
Acknowledgment. We thank the Deutsche Forschungsge-
meinschaft (DFG; SFB 334), the Federal Ministry of Science
and Technology (BMBF), the Katalyseverbund NRW, and the
Fonds der Chemischen Industrie for financial support. P.H.
thanks the Fonds der Chemischen Industrie for a graduate
scholarship, and R.G. thanks the Konrad Adenauer-Foundation
accordingly. We especially thank Prof. A. Togni (ETH Zurich)
for providing the ferrocenyl ligands.
Conclusions
We propose two PHIP NMR methods to quantitatively
investigate homogeneous hydrogenations. The first method,
ROCHESTER, focuses on the rate of hydrogen transfer and,
depending on the experimental conditions, is capable of provid-
JA984353Y