Intramolecular hydroformylation of allyldiphenylphosphine using a monodentate phosphorus diamide based rhodium catalyst: An NMR study

Article abstract of DOI:10.1021/om000890r

The stepwise intramolecular hydroformylation reaction of allyldiphenylphosphine using a monodentate phosphorus diamide based rhodium catalyst has been studied using NMR spectroscopy. Reaction of the rhodium-hydride complex HRhL₃CO (L = triethylbiuretphenylphosphorus diamide) with allyldiphenylphosphine results in the formation of HRhL₂-(allylPPh₂)CO. This compound undergoes hydride migration to form a cyclic rhodium-alkyl complex. When carbon monoxide is added to this complex, ligand exchange and CO insertion occur, leading to the formation of the cyclic rhodium-acyl complex Rh(CO)CH₂CH₂ CH₂PPh₂L(CO)₂. Addition of hydrogen completes the cycle forming the coordinated hydroformylated allyldiphenylphosphine ligand. The aldehyde-functionalized phosphine ligand is hydrogenated to the corresponding alcohol.

Full text of DOI:10.1021/om000890r

Organometallics 2001, 20, 1079-1086
1079
In tr a m olecu la r Hyd r ofor m yla tion of
Allyld ip h en ylp h osp h in e Usin g a Mon od en ta te
P h osp h or u s Dia m id e Ba sed Rh od iu m Ca ta lyst: An NMR
Stu d y
Saskia C. van der Slot, Paul C. J . Kamer, and Piet W. N. M. van Leeuwen*
Institute of Molecular Chemistry, University of Amsterdam, Nieuwe Achtergracht 166,
1018 WV Amsterdam, The Netherlands
Received October 16, 2000
The stepwise intramolecular hydroformylation reaction of allyldiphenylphosphine using
a monodentate phosphorus diamide based rhodium catalyst has been studied using NMR
spectroscopy. Reaction of the rhodium-hydride complex HRhL3CO (L ) triethylbiuretphen-
ylphosphorus diamide) with allyldiphenylphosphine results in the formation of HRhL2-
(allylPPh2)CO. This compound undergoes hydride migration to form a cyclic rhodium-alkyl
complex. When carbon monoxide is added to this complex, ligand exchange and CO insertion
occur, leading to the formation of the cyclic rhodium-acyl complex Rh(CO)CH2CH2CH2P-
Ph2L(CO)2. Addition of hydrogen completes the cycle forming the coordinated hydroformylated
allyldiphenylphosphine ligand. The aldehyde-functionalized phosphine ligand is hydrogenated
to the corresponding alcohol.
Sch em e 1
In tr od u ction
The rhodium-catalyzed hydroformylation using phos-
phorus ligands is one of the most extensively studied
catalytic processes.1-6 Not only has the catalyst perfor-
mance been an important aspect of the studies during
the past decades but also the reaction mechanism and
the solution structure of the catalyst have been inves-
tigated in great detail.7,8 Although several intermediates
of the unmodified rhodium catalyst have been charac-
terized,9-11 only a small number of intermediates pro-
posed for the phosphorus-modified catalyst have been
identified,12-14 and very little is known about the
reactivity of the reaction intermediates and the revers-
ibility of the reaction steps proposed.15,16
Recently we presented a new group of phosphorus
diamide ligands for the rhodium-catalyzed hydrofor-
mylation reaction17 that is based on a biuret structure.
Although the ligands are very bulky, moderate selectiv-
ity for the linear aldehyde was found in the hydro-
formylation reaction of 1-octene;17 bulky diphosphite
ligands for instance often lead to high linear-to-
branched ratios.18 Here we report on a detailed study
of the stoichiometric hydroformylation reaction of al-
lyldiphenylphosphine using the monodentate phospho-
rus diamide based rhodium catalyst HRh(1)3CO. The
allyldiphenylphosphine rhodium complex 2 (Scheme 2)
is used as a model compound to investigate the hydro-
formylation reaction in a stepwise manner. The struc-
tures of the hydroformylation reaction intermediates as
proposed in Scheme 2 are investigated using NMR
spectroscopy. Not only is this type of intramolecular
* To whom correspondence should be addressed. E-mail: pwnm@
anorg.chem.uva.nl.
(1) Evans, D.; Osborn, J . A.; Wilkinson, G. J . Chem. Soc. A 1968,
3133.
(2) Brown, C. K.; Wilkinson, G. J . Chem. Soc. A 1970, 2753.
(3) Casey, C. P.; Whiteker, G. T.; Melville, M. G.; Petrovich, L. M.;
Gavney, J . A., J r.; Powell, D. R. J . Am. Chem. Soc. 1992, 114, 5535.
(4) Moasser, B.; Gladfelter, W. L.; Christopher Roe, D. Organome-
tallics 1995, 14, 3832.
(5) Casey, C. P.; Paulsen, E. L.; Beuttenmueller, E. W.; Proft, B.
R.; Malter, B. A.; Powell, D. R. J . Am. Chem. Soc. 1999, 121, 63
(6) van der Veen, L. A.; Boele, M. D. K.; Bregman, F. R.; Kamer, P.
C. J .; van Leeuwen, P. W. N. M.; Goubitz, K.; Fraanje, J .; Schenk, H.;
Bo, C. J . Am. Chem. Soc. 1998, 120, 11616.
(7) Moser,W. R.; Papile, C.; Brannon, D. A.; Duwell, R. A. J . Mol.
Catal. 1987, 271.
(8) Yagupsky, G.; Brown, C. K.; Wilkinson, G. J . Chem. Soc. (A)
1970, 1392.
(9) Feng, J .; Garland, M. Organometallics 1999, 18, 417.
(10) Liu, G.; Volken, R.; Garland, M. Organometallics 1999, 18, 3429.
(11) Die´guez, M.; Claver, C.; Masdeu-Bulto´, A. M.; Ruiz, A. Orga-
nometallics 1999, 18, 2107.
(15) Casey, P. C.; Petrovich, L. M. J . Am. Chem. Soc. 1995, 117,
6007.
(16) Lazzaroni, R.; Uccello-Barretta, G.; Benetti, M. Organometallics
1989, 8, 2323.
(17) van der Slot, S. C.; Kamer, P. C. J .; van Leeuwen, P. W. N. M.;
Fraanje, J .; Goubitz, K.; Lutz, M.; Spek, A. L. Organometallics 2000,
19, 2504.
(12) Bianchini, C.; Man Lee, H.; Meli, A.; Vizza, F. Organometallics
2000, 19, 849.
(13) Brown, J . B.; Kent, A. G. J . Chem. Soc. Perkin Trans. 2 1987,
1597.
(14) van der Slot, S. C.; Kamer, P. C. J .; van Leeuwen, P. W. N. M.;
Iggo, J . A.; Heaton, B. T. Organometallics 2001, 20, 430.
(18) Kamer, P. C.; Reek, J . N. H.; van Leeuwen, P. W. N. M. In
Rhodium Catalysed Hydroformylation; van Leeuwen, P. W. N. M.,
Claver, C., Eds.; Kluwer Academic Publishers: Dordrecht, 2000;
Chapter 3.
10.1021/om000890r CCC: $20.00 © 2001 American Chemical Society
Publication on Web 02/22/2001

1080 Organometallics, Vol. 20, No. 6, 2001
van der Slot et al.
Sch em e 2. P ossible Rea ction In ter m ed ia tes of th e
In tr a m olecu la r Hyd r ofor m yla tion of
Allyld ip h en ylp h osp h in e
phosphite-based rhodium complex, all subsequent steps
of the hydroformylation mechanism except the hydro-
genolysis were observed. The use of a monodentate
ligand instead of a bidentate ligand could enhance
ligand dissociation and thereby the hydrogenolysis,
leading to the hydroformylated allyldiphenylphosphine.
Resu lts a n d Discu ssion
The rhodium-hydride complex HRh(1)3CO was used
as starting material for this mechanistic study.17 This
complex has a trigonal bipyramidal structure with the
hydride ligand coordinated at an apical position. The
1H, 13C, and 31P NMR data for this complex are
summarized in Table 1 to facilitate comparison with the
data of the intermediates formed in this study. The 2J PH
coupling constant of 13 Hz is relatively large for a pure
cis-orientation of the phosphorus and hydride ligands,
indicating a small distortion of the trigonal bipyramidal
structure. The crystal structure confirmed that the
hydride and carbonyl ligands are coordinated at an
apical position.17 The rhodium atom is located slightly
below the equatorial plane defined by the phosphorus
atoms, displaced toward the carbonyl ligand. The 13C-
{1H} NMR spectrum showed a doublet of quartets in
the terminal CO region when the complex synthesis was
performed using 13CO. The large 2J CH coupling constant
of 36 Hz for trans CO-hydride coordination in a trigonal
bipyrimidal rhodium complex has also been reported by
Brown15 and van Leeuwen.23 The P-Rh-CO angle is
hydroformylation of interest from a mechanistic point
of view, but coupling of the substrate and ligand
function can lead to increased selectivity for either the
linear or branched product, as has been shown by
J ackson19 and Breit.20
Since this biuret-based rhodium catalyst shows mod-
erate selectivity for the linear aldehyde in the hydro-
formylation of 1-octene,17 special attention will be paid
to the regioselectivity of the hydride migration and
isomerization reaction. Van Leeuwen and co-workers21
investigated the hydride migration of several alkenyl-
phosphine-platinum complexes. They observed for vi-
nyldiphenylphosphine both the linear and branched
alkyl-platinum complex. The branched alkyl complex,
a three-membered ring system, is the initial product
formed by kinetic control; the linear alkyl complex, a
four-membered ring system, is the thermodynamically
more stable product formed after longer reaction times.
They observed only the linear (five-membered ring
system) product for allyldiphenylphosphine. Recently
van Leeuwen et al.22 studied the stepwise hydroformyl-
ation reaction of allyldiphenylphosphine using a rho-
dium catalyst containing a bidentate bulky phosphite
ligand. Similar to the platinum complexes, the linear
rhodium-alkyl complex was observed exclusively after
hydride migration, but after CO insertion both the
linear and branched rhodium-acyl complexes were
observed, indicating that the branched rhodium-alkyl
complex must have been formed as a transient inter-
mediate. Thus kinetically the formation of small rings
seems to be favored both in the rhodium and in the
platinum systems. In this study we will focus on the
reversibility of the hydride migration step and the effect
of using monodentate instead of chelating phosphorus
ligands in this reaction step. Using the bidentate
2
larger than 90°, leading to a relatively large J PC
coupling constant of 13 Hz.
Rea ction of HRh (1)3CO a n d Allyld ip h en ylp h os-
p h in e. One equivalent of allyldiphenylphosphine was
made to react with HRh(1)3CO at 233 K. The difference
in electronic ligand properties favors the exchange of a
π-acidic phosphorus ligand with an electron-donating
phosphine ligand. Free allyldiphenylphosphine was not
observed in the 31P NMR spectra (δ(31P) ) -13.6 ppm).
The major compound formed is the rhodium-hydride
complex 2 (Scheme 2), but in all cases approximately
10% of the disubstituted complex (HRh(1)(allylPPh2)2-
CO) and HRh(1)3CO were obtained in the 31P NMR
spectra (Figure 1, spectrum 1). When the reaction
mixture was warmed to room temperature, complete
conversion to complex 2 was obtained, but at this
temperature hydride migration occurs. To characterize
complex 2 completely, the solution was kept at 253 K
after addition of allyldiphenylphosphine to the rhodium-
hydride complex.
The structure of 2 (Scheme 2) was elucidated using
1
various NMR techniques (Table 1). The H NMR spec-
trum showed that the alkene moiety was still present.
All allyl protons were shifted upfield compared to the
resonances of noncoordinated allyldiphenylphosphine.
The hydride resonance was shifted downfield. The
magnitude of the phosphorus-hydride coupling con-
stant is approximately the same for both the phosphine
and the phosphorus diamide ligands and shows that the
trigonal bipyramidal structure is still slightly distorted.
The 2J CH coupling constant of 2 (48 Hz) is slightly larger
(19) J ackson, W. R.; Perlmutter, P.; Suh, G.-H.; Tasdelen, E. E.
Austr. J . Chem. 1991, 44, 951.
(20) Breit, B. Chem. Eur. J . 2000, 6, 1519.
(21) van Leeuwen, P. W. N. M; Roobeek, C. F.; Frijns, J . H. G.;
Orpen, A. G. Organometallics 1990, 9, 1211.
2
than the J CH coupling constant of HRh(1)3CO (36 Hz).
(22) van Rooy, A.; Kamer, P. C. J .; van Leeuwen, P. W. N. M. J .
Organomet. Chem. 1997, 535, 201.
(23) van Leeuwen, P. W. N. M.; Buisman, G. J . H.; van Rooy, A.;
Kamer, P. C. J . Recl. Chim. Trav. Pays-Bas 1994, 113, 61.

Hydroformylation of Allyldiphenylphosphine
Organometallics, Vol. 20, No. 6, 2001 1081
Ta ble 1. Sp ectr oscop ic Da ta of th e Differ en t In ter m ed ia tes Obser ved
δ(31P)a (ppm)
J RhP (Hz)
J P1P2 (Hz)
δ(13C) (ppm)
J RhC (Hz)
J PC (Hz)
δ(1H) (Hz)
J PH (Hz)
J CH (Hz)
J RhH (Hz)
a
a
complex
HRh(1)3CO
2
116 (P1)b
116 (P1)b
23 (P2)
180 (P1)
188 (P1)
143 (P2)
195 (P1)
147 (P2)
205 (P1)
147 (P2)
205 (P1)
156 (P2)
185
200c
202c
52
52
13
13
-11.1b
-10.8b
13
11
36
48
e3
e3
131
112
117
117
77
3b
5a
5b
6
119 (P1)d
60 (P2)
197d
47
16
106 (P1)c
16 (P2)
195 (broad)d
243 (broad)d
n.d.
n.d.
n.d.
63
n.d.
n.d.
106 (P1)c
56 (P2)
109 (P1)
27 (P2)
197
202
24 (P1)
<3 (P2)
13
-10.3
-10.7
10 (P1)
33 (P2)
12
8
5
122
188
141
7
117 (P1)e
25 (P2)
130
51
n.d.
e3
a
b
P1 belongs to the phosphorus diamide ligand resonance, P2 belongs to the allyldiphenylphosphine resonance. Determined at 253 K.
c Determined at 233 K. Determined at 203 K. e Determined at 268 K.
d
F igu r e 1. 31P{1H} NMR spectra of Complexes 2, 3b, and 5a : (1) 31P{1H} NMR of complex 2 obtained at 253 K, (2) 31P{1H}
NMR of complex 3b obtained at 203 K, (3) 31P{1H} NMR of complex 5a obtained at 233 K, (*) HRh(1)3CO, (**) 5b, (∆)
allyldiphenylphosphineoxide.
The phosphorus NMR spectrum (Figure 1, spectrum 1)
showed a doublet of doublets in the rhodium-phospho-
rus diamide region and a doublet of triplets in the
rhodium-phosphine region in a ratio of 2:1. The large
2J PP coupling constant of 131 Hz is consistent with the
coordination of all three phosphorus atoms in the
equatorial plane. The 13C{1H} NMR spectrum showed
a doublet of quartets at 202 ppm for 2 prepared with
approximately 40 ppm compared to the phosphine
chemical shift of complex 2 is characteristic of the
formation of a five-membered ring.21,24 The formation
of a four-membered ring will lead to an upfield shift of
approximately 40 ppm, as reported by Garrou.24 We
conclude from the downfield shift of the allyldiphen-
ylphosphine resonance that the linear rhodium-alkyl
complex 3b has been formed (Scheme 3). The spectro-
scopic data of complex 3b are presented in Table 1. The
2J PP coupling constant found for 3b (112 Hz) is much
13CO. The J RhC coupling constant (52 Hz) and J PC
coupling constant (13 Hz) are in the same range as the
data found for HRh(1)3CO, indicating that similar to
this complex, the phosphorus ligands are coordinated
in the equatorial plane and the carbonyl ligand and
hydride ligands are coordinated at an apical position of
the distorted trigonal bipyramid.
1
2
2
smaller than the J PP found for 2 (131 Hz). The forma-
tion of the five-membered chelate ring will lead to a
more distorted trigonal bipyramidal structure, resulting
in a smaller P-Rh-P angle. The 13C{1H} NMR spec-
trum showed a doublet of quartets at 197 ppm for the
1
2
13CO complex. Both J RhC (47 Hz) and J PC (16 Hz)
coupling constants are slightly different from 2, which
indicates a change in the structure of the complex. The
branched alkyl 3d (Scheme 3) was not observed in the
31P NMR spectra.
Hyd r id e Migr a tion . When a solution of 2 in CD2-
CL2 was warmed to room temperature, additional
(broad) resonances in the phosphorus spectrum ap-
peared. The intensity of the hydride and allyl reso-
1
nances in the H NMR spectrum decreased in intensity
and new resonances in the alkyl region appeared,
indicating that the hydride migration reaction has
started. When the solution of 2 was warmed for 10 min
at 313 K, complete conversion of the rhodium-hydride
complex was reached. At room temperature, the 31P
NMR spectrum showed a very broad resonance at the
position of the free phosphorus diamide ligand (64 ppm)
and a broad doublet around 60 ppm. The broad reso-
nances in the phosphorus NMR spectrum indicate that
fluxional processes occur at room temperature (vide
infra). When the temperature was decreased to 203 K,
these processes became slow on the NMR time scale,
and a doublet of doublets in the rhodium-phosphorus
diamide region (119 ppm) and a doublet of triplets at
60 ppm in a ratio 2:1 appeared (Figure 1, spectrum 2).
The downfield shift of the phosphine chemical shift of
Two different mechanisms have been proposed for the
hydroformylation reaction.2 These mechanisms differ in
the mode of attack of the alkene on the catalyst. In the
associative pathway the alkene coordinates to the five-
coordinated rhodium center; the six-coordinated 20-
electron complex undergoes hydride migration to form
a rhodium-alkyl complex. Immediately after coordina-
tion of the phosphine ligand, a very fast insertion would
be expected according to this mechanism, which does
not appear to be the case. In the dissociative pathway,
the commonly accepted mechanism,25 the five-coordi-
(24) Garrou, P. E. Chem. Rev. 1981, 81, 229.
(25) van Leeuwen, P. W. N. M.; Casey, C. P.; Whiteker, G. T. In
Rhodium Catalysed Hydroformylation; van Leeuwen, P. W. N. M.,
Claver, C., Eds.; Kluwer Academic Publishers: Dordrecht, 2000;
Chapter 4.

1082 Organometallics, Vol. 20, No. 6, 2001
van der Slot et al.
Sch em e 3. Over view of th e In ter m ed ia tes F or m ed in th e In tr a m olecu la r Hyd r ofor m yla tion of
Allyld ip h en ylp h osp h in e
nated rhodium-hydride complex first splits off one
ligand to form a 16-electron rhodium center. The alkene
coordinates to this four-coordinate rhodium center fol-
lowed by hydride migration.2 At room temperature, the
31P NMR spectrum of the rhodium-alkyl complex
showed very broad resonances, indicating that fluxional
processes occur at room temperature. The 31P NMR
spectrum showed a very broad resonance at the position
of the free ligand (64 ppm) and a broad doublet around
60 ppm. Similar to the phosphorus diamide ligands in
HRh(1)3CO, the phosphorus diamide ligands in complex
2 are in fast exchange with the free ligand.17 When we
consider the dissociative pathway for the hydride mi-
gration, this ligand coordination/dissociation process
creates a free coordination site for the alkene moiety at
the rhodium center (complex 2a , Scheme 3). After
alkene coordination (complex 2b/2c, Scheme 3), hydride
migration can occur to form the four-coordinate (cyclic)
alkyl complexes 3a and 3c (Scheme 3). Probably, these
coordinatively unsaturated complexes will coordinate a
phosphorus ligand to form the five-coordinate alkyl
complexes 3b and 3d . Comparison of the reaction times
to complete the hydride migration reaction using the
monodentate biuret-based rhodium complex (10 min at
313 K) with the bidentate phosphite-based rhodium
complex (3 h at 313 K)22 indicates that the hydride
migration in the former case is much faster. The use of
monodentate ligands instead of a bidentate ligand
causes the faster hydride migration observed in this
study. The monodentate phosphorusdiamide ligands are
in fast exchange on the NMR time scale with free ligand
in solution. Ligand dissociation creates a free coordina-
tion site for the alkene moiety at the rhodium center.
Formation of a free coordination site in case of the
bidentate phosphite ligand is hindered by the formation
of the rhodium-phosphite chelate ring. Since the ob-
served hydride migration is still orders of magnitude
slower than ligand exchange (NMR line broadening at
293 K), the alkene coordination/hydride migration may
involve dissociation of a ligand and hence a 16-electron
species.
In the hydroformylation reaction of 1-octene using
ligand 1, high isomerization rates and low selectivity
toward the linear product are observed,17 but using the
allyldiphenylphosphine as substrate the branched alkyl
complex was not observed in the NMR spectra. This in
combination with the observation of the branched
rhodium-acyl complex using the bidentate bulky phos-
phite system suggests that the branched rhodium-alkyl
complex is formed in the reaction mixture. Fast equi-
librium between the alkene complex 2b/2c and both
alkyl complexes 3a and 3c may occur at temperatures
higher than 203 K and the branched rhodium-alkyl
complex could be formed in the reaction mixture.
Similar to the vinyldiphenylphosphine platinum com-
plexes,21 the linear rhodium-alkyl complex (3b) could
be thermodynamically more stable than the branched
rhodium-alkyl complex (3d ) as a result of the larger
ring strain in the four-membered ring of complex 3d .
Therefore complex 3b is the only rhodium-alkyl com-
plex observed in the reaction mixture.
To visualize the formation of the branched alkyl
complex, we performed the hydride migration reaction
using the deuteriorhodium complex instead of the
hydridorhodium complex. When deuteride migration
leads exclusively to the linear alkyl complex, only a
deuterium label at the â-carbon atom will be observed
(complex 3b′, Scheme 4). When the branched alkyl
complex is formed but rearranges to the linear alkyl
complex, a complex with a deuterium label at the

Hydroformylation of Allyldiphenylphosphine
Organometallics, Vol. 20, No. 6, 2001 1083
Sch em e 4. P osition of th e Deu ter iu m La bel a fter Deu ter iu m Migr a tion a
a
Complex 2c′ has been omitted for brevity.
Ta ble 2. P r oton a n d Deu ter iu m Ch em ica l Sh ifts of th e Allyl- a n d Alk yl-Rh od iu m Com p lexes
rhodium-allyl complexes
hydride/deuteride (ppm)
δ(Hγ/Dγ) (ppm)
δ(Hâ/Dâ) (ppm)
δ(HR/DR) (ppm)
HRh(1)2(allylPPh2)CO (2)
DRh(1)2(allylPPh2)CO (2′)
HRh(1)2(CH2CDbCH2PPh2)CO
HRh(1)2(CHDcCHCH2PPh2)CO
rhodium-alkyl complexes
-10.8 (1H)
-10.6 (broad)
2.5 (m) (1H)
5.2 (m) (1H)
4.7 (m) (1H)
5.0 (m)
4.6 (m)
1.7 (broad, m)
RhCH2CHDCH2PPh2(1)2CO (3b′)
1.2 (broad)
RhCH2CHDCH2PPh2(1)2CO (3b′′)
R-position will be observed (complex 3b′′, Scheme 4). For
the deuteride migration using DRh(1)2(allylPPh2)CO (2′)
the same procedure was followed as for the hydride
complex. The complexes formed were investigated using
deuterium and phosphorus NMR spectroscopy. After
coordination of allyldiphenylphosphine to DRh(1)3(CO),
the solution was warmed to room temperature. The 31P
NMR spectra observed for the deuterated complexes are
identical to those of the hydride complexes. The deute-
rium resonances observed and the corresponding proton
chemical shifts are displayed in Table 2. The alkyl
resonances of the rhodium-alkyl complex are not
between the R- and â-position indicates that both the
four- and five-membered alkyl rings were formed.
Together with the allyl resonances, several broad reso-
nances in the alkyl region appeared. Similar to the
hydride migration experiments the reaction mixture
was warmed for 10 min at 313 K. Completion of the
deuteride migration was checked using 31P NMR spec-
troscopy. After cooling to room temperature, the two
allyl resonances had disappeared and two broad reso-
nances at 1.7 and 1.2 ppm were the only resonances
remaining in the 2H NMR spectrum. The position of
these two resonances is strongly indicative of a deute-
rium label at the R- or â-carbon atom of the linear
rhodium-alkyl complex, indicating that the four-
membered rhodium-alkyl complex was formed.
1
resolved in the H NMR spectrum because of overlap-
ping with the ethyl resonances of ligand 1. Directly after
2
warming to room temperature, the H NMR spectrum
clearly shows the (broad) deuteride resonance at -10.6
ppm, together with additional resonances at 5.0 and 4.6
ppm. These chemical shifts correspond to those observed
for the protons of the coordinated allyldiphenylphos-
phine (5.2 and 4.7 ppm) and belong to the deuterium
labels at the â- and R-carbon atoms. Appearance of
deuterium labels at these positions of the allyl moiety
shows that there is a fast equilibrium between the
rhodium-hydride/deuteride and the rhodium-alkyl
complexes (hydride migration followed by â-hydride
elimination). The scrambling of the deuterium label
Summarizing the results of the deuterium migration,
we conclude that initially both the linear and branched
rhodium-alkyl complexes are formed in the reaction
mixture. Similar to the vinyldiphenylphosphine plati-
num complexes, the larger ring system is the thermo-
dynamically more stable product. Probably, the four-
membered ring system is more sensitive toward â-hy-
dride elimination (as a result of the ring strain) than
the five-membered ring system. When the hydride
migration is complete, all complexes present are con-
verted to the stable linear alkyl complex.

1084 Organometallics, Vol. 20, No. 6, 2001
van der Slot et al.
CO In ser tion . The 31P NMR spectrum changed
tremendously when the solution of 3b was pressurized
with 5 bar CO at 253 K (Figure 1, spectrum 3). A sharp
doublet of doublets appeared in the rhodium-phospho-
rus diamide region, and a sharp doublet of doublets
appeared in the rhodium-phosphine region in a ratio
of 1:1. One of the phosphorus diamide ligands is
replaced by a CO ligand. The upfield shift of the
phosphine resonance compared to complex 3b indicated
that the phosphine ligand is now part of a six-membered
ring. The 13C{1H} NMR spectrum showed a very broad
singlet (W1/2 ≈ 60 Hz) at 243 ppm and two very broad
resonances around 195 ppm when 13CO was used. The
13C NMR resonance at 243 ppm is strongly indicative
of a rhodium-acyl carbon atom, proving that CO
insertion has occurred. The broad carbonyl resonances
in the 13C NMR spectrum indicate that the equatorial
and apical CO ligands are in slow exchange. A broad
multiplet appeared in the carbonyl ligand region when
the solution was cooled to 193 K, but the acyl carbon
resonance remained broad upon cooling to this temper-
ature. The J PC, J RhC, and J CC coupling constants re-
mained unresolved. On the basis of the upfield shift of
the phosphine resonance of approximately 40 ppm
compared to complex 3b, we conclude that the linear
rhodium-acyl complex 5a (Scheme 3) is the major
complex formed after CO insertion.
was kept at room temperature. Slowly, the CO insertion
started and the resonances of the linear acyl complex
5a appeared. An additional doublet of doublets appeared
2
at 56 ppm (1J RhP ) 156 Hz, J PP ) 117 Hz). An
additional resonance in the phosphorus diamide region
was not observed, but a 31P COSY spectrum showed
cross-peaks for the resonance at 56 ppm and the
phosphorus diamide resonance 106 ppm. The resonance
at 106 ppm also showed cross-peaks with the phosphine
resonance at 16 ppm, which belongs to the linear
rhodium-acyl complex. On the basis of the downfield
shift of this resonance, the correlation with the phos-
phorus diamide resonances that belong to a rhodium-
acyl complex, and the deuterium labeling experiments,
we conclude that the phosphorus resonance at 56 ppm
belongs to the branched rhodium-acyl complex 5b
(Scheme 3). Additional proof for the formation of the
branched alkyl/acyl complex is the presence of both the
resonances of the linear and branched aldehyde proton
1
in the H NMR spectra after addition of hydrogen. The
deuterium experiments showed that the branched alkyl
complex is formed in the reaction mixture, although this
complex was not observed in the 31P NMR spectra. The
presence of CO in the reaction mixture traps this
branched product by the formation of a five-membered
ring system after CO insertion.
Hyd r ogen olysis After the CO insertion we bubbled
H2 through the solution of 5a at room temperature.
Under atmospheric hydrogen pressure, hydrogenolysis
did not occur. Deinsertion of CO is observed, probably
because the insertion-deinsertion equilibrium is driven
to the rhodium-alkyl complex since CO is removed by
the H2 flow. A flow of 13CO through a solution of the
12CO rhodium-acyl complex 5a resulted in the appear-
ance of the rhodium-acyl resonance of complex 5a in
the 13C{1H} NMR spectrum showing the reversibility
of the CO insertion. The rhodium-acyl complex 5a and
the hydride complex HRh(1)2(CO)2 were the only com-
plexes observed when a 1:1 mixture of CO and H2 was
bubbled through the solution. These results indicate
that the hydrogenolysis does not occur at atmospheric
hydrogen pressure. This is not unlikely since kinetic
experiments with this ligand system showed that hy-
drogenolysis, depending on the conditions used, is one
of the rate-determining steps.14
When CO was added to the solution of 3b at room
temperature, approximately 10% of the rhodium-hy-
dride complex HRh(1)2(CO)2 was formed next to the
rhodium-acyl complex 5a. Reformation of the rhodium-
hydride complex (HRh(1)3CO) followed by exchange of
the phosphine ligand with CO can occur, since all
reaction steps are reversible. An additional pathway to
form the hydride complex HRh(1)2(CO)2 is â-hydride
elimination, one of the main side reactions in the
hydroformylation cycle. Dissociation of the phosphine
ligand induced by the ring strain will create a vacant
site for the coordination of the hydride (see Scheme 3).
After â-hydride elimination a carbonyl ligand coordi-
nates at the remaining vacant site. â-Hydride elimina-
tion is not unlikely in this system, as was shown in the
previous section.
For the hydride migration reaction, we proposed that
initially both the linear and branched rhodium-alkyl
complexes were formed in the reaction mixture. As a
result of ring strain, the four-membered ring system was
converted to the thermodynamically stable linear rho-
dium-alkyl complex. To detect the branched product
indirectly, we performed the hydride migration reaction
in the presence of CO. When the branched alkyl complex
is formed, it can immediately give CO insertion, and
the more stable five-membered acyl ring will be trapped.
This five-membered ring, if formed, will have the
characteristic downfield shift in the 31P NMR spectrum
as reported by Garrou.24 After the coordination of
allyldiphenylphosphine to HRh(1)3CO, the (high-pres-
sure) NMR tube was pressurized to 6 bar of CO at 253
K. The tube was warmed to room temperature, and the
reaction was monitored using NMR spectroscopy. Ini-
tially, the 31P NMR spectrum showed two complexes:
the hydride complex 2 and a small amount of the linear
alkyl complex 3b. Different from the hydride migration
experiments, the tube was not warmed to 313 K, but it
When the solution of the rhodium-acyl complex 5a
was pressurized to 15 bar of CO/H2 ) 1:2, the hydro-
genolysis occurred slowly overnight at room tempera-
ture. Directly after pressurizing with hydrogen, a weak
hydride resonance appeared in the hydride region of the
1H NMR spectrum and aldehyde resonances appeared
at approximately 10 ppm, proving that the hydrogenoly-
sis had started. After complete conversion of the rho-
dium-acyl complex has been reached, the aldehyde
resonance is present only in low concentration in the
proton NMR spectrum. The IR spectrum of this solution
showed a weak aldehyde absorption at 1751 cm-1 and
a medium absorption at 3340 cm-1 that probably
belongs to the hydroxyl group. J ackson and co-workers13
performed the hydroformylation of a range of alkenyl-
phosphines. Their results showed complete conversion
of the allyldiphenylphosphine to the alcohol instead of
the aldehyde. The presence of the aldehyde moiety in
close proximity of the rhodium center probably leads to

Hydroformylation of Allyldiphenylphosphine
Organometallics, Vol. 20, No. 6, 2001 1085
F igu r e 2. 31P{1H} NMR spectra of complexes 6 and 7: (1) 31P{1H} NMR of complex 6 obtained at 293 K, (2) 31P{1H} NMR
of complex 7 obtained at 268 K, (*) HRh(1)2(CO)2.
Sch em e 5. Com p lexes F or m ed a fter
In tr a m olecu la r Hyd r ofor m yla tion a n d
Hyd r ogen a tion of Com p lex 2
constant. The large J PH coupling constant suggests that
the phosphine ligand is now coordinated to some extent
at an apical position of the trigonal bipyramid (complex
6, Scheme 5). The phosphorus diamide ligand is still
coordinated at an equatorial position. The 13C NMR
F igu r e 3. Hydride resonance after selectively decoupling
of the phosphine or phosphorus diamide resonance: (1)
Hydride resonance of complex 6, (2) hydride resonance of
complex 6 after selective decoupling of the rhodium-
phosphine resonance, (3) hydride resonance of complex 6
after selective decoupling of the rhodium-phosphorus
diamide resonance.
spectrum confirms this hypothesis, showing only one
resonance for both carbonyl ligands having one large
2
2J CP coupling (24 Hz) and one small J CP coupling
constant (<3 Hz).
We depressurized the NMR tube after the hydro-
genolysis and hydrogenation reaction and removed the
hydrogen and CO gases by bubbling argon through the
solution. After removing the CO from the solution the
phosphorus NMR spectrum showed a doublet of dou-
blets in the phosphorus diamide region (117 ppm) and
a double triplet appears in the phosphine region (25
ppm) (spectrum 2, Figure 2). The multiplicity of the
rhodium-phosphine resonance shows that complex 6
exchanged a carbonyl ligand with a phosphorus diamide
further hydrogenation of the aldehyde to the corre-
sponding alcohol. (The hydrogenation of the aldehyde
is one of the side reactions observed in the hydrofor-
mylation reaction.)
After complete conversion of the rhodium-acyl com-
plex 5a , the 31P NMR spectrum showed a doublet of
doublets in both the phosphorus diamide (109 ppm) and
phosphine region (27 ppm) (Figure 2, spectrum 1).
Different from the previously described hydride complex
2
ligand. The J PH coupling constant (12 Hz) is again
2
(2), the J PH coupling constants for the phosphine and
similar for both the rhodium-phosphorus diamide and
the phosphine, indicating that the phosphine ligand has
moved to the equatorial plane (complex 7, Scheme 5).
The hydrogenolysis reaction is much slower than the
insertion reactions under the conditions studied. At
higher temperatures these differences are less pro-
nounced, and thus we conclude that the activation
enthalpy for the hydrogenolysis step is relatively high.
the phosphorus diamide ligand are not equal (33 and
10 Hz). Selectively decoupling of the phosphorus reso-
nance at 27 ppm or at 109 ppm showed that the large
2J PH coupling constant of 33 Hz results from the
phosphine atom, whereas the small coupling of 10 Hz
results from the phosphorus diamide atom (Figure 3).
2
The J PH coupling constant of 10 Hz is similar to the
2
previously obtained data, whereas the J PH coupling
constant of 33 Hz is too large for a cis coordination of
the hydride and phosphorus ligand. Two explanations
can be brought forward: an equilibrium mixture of ee-
ea coordination giving averaged signals in the NMR
spectra, provided that there is a fast exchange of the
ligand between the equatorial and apical positions;26
second, coordination of the phosphorus ligand in the
equatorial plane of a slightly distorted trigonal bipyra-
Con clu sion
The stepwise hydroformylation of allyldiphenylphos-
phine has been investigated using a monodentate bi-
uret-based rhodium catalyst. Coupling of the ligand and
substrate functions stabilized the intermediates of the
hydroformylation reaction and enabled the character-
ization of the complexes using NMR spectroscopy.
Hydride migration in the absence of CO led to
complete conversion of the rhodium-hydride complex
to the linear rhodium-alkyl complex (3b). The deute-
rium complex showed scrambling of deuterium at the
2
mid would lead to an increase of the J PH coupling
(26) van der Veen, L. A.; Boele, M. D. K.; Bregman, F. R.; Kamer,
P. C. J .; van Leeuwen, P. W. N. M.; Goubitz, K.; Fraanje, J .; Schenk,
H.; Bo, C. J . Am. Chem. Soc. 1998, 120, 11616.

1086 Organometallics, Vol. 20, No. 6, 2001
van der Slot et al.
niques. Dichloromethane-d2 was distilled from calcium hy-
dride. Other solvents were distilled from sodium/benzophe-
none. All glassware was dried by heating under vacuum. The
NMR spectra were recorded on a Bruker DRX-300 spectrom-
eter. Chemical shifts are given in ppm referenced to TMS or
H3PO4 (external). Ligand 1 and HRh(1)3CO were synthesized
according literature procedures.17 Because of the low yield of
HRh(1)3CO described in the literature,17 an optimized proce-
dure is described below.
R- and â-positions of the allyl/alkyl moiety in the hydride
migration reaction, indicating that both the linear and
the branched rhodium-alkyl complex had been formed.
The linear rhodium-alkyl complex was shown as the
thermodynamically stable product formed after longer
reaction times, similar to the results observed for
comparable platinum-alkyl complexes.21 Deuterium
scrambling at the R- and â-position of the allyl moiety
shows that the hydride migration is a reversible process
in the absence of carbon monoxide. Recently we studied
the reversibility of the hydride migration reaction in the
hydroformylation reaction of 1-octene using this biuret-
based rhodium catalyst.14 These experiments showed
that under hydroformylation conditions the hydride
migration is virtually irreversible. The results obtained
using allyldiphenylphosphine are in agreement with the
results observed for 1-octene. In the presence of CO the
hydride migration was directly followed by CO insertion,
and both the linear and branched rhodium-acyl com-
plexes were observed in the 31P NMR spectra, showing
that in the presence of CO the hydride migration is
irreversible. On the basis of these observations we
conclude that the selectivity in this system for either
the linear or branched aldehyde is determined by the
ratios of the rates of reaction that can take place
immediately after the hydride migration.
In a typical 5 mm tube NMR experiment 20 mg (0.020
mmol) of 2 was dissolved in 0.5 mL of CD2Cl2. A 0.2 mL (0.020
mmol) portion of a 0.1 M stock solution of allyldiphenylphos-
phine in CD2Cl2 was added dropwise to this solution at -40
°C. The yellow solution was stirred for 15 min at this
temperature before the NMR spectra were recorded. For the
10 mm high-pressure NMR tube, 80 mg (0.08 mmol) of HRh-
(1)3CO in 1.5 mL of CD2Cl2 was used.
HRh (1)3CO. A 20 mg (0.078 mmol) sample of Rh(acac)(CO)2
and 114 mg (0.39 mmol) of ligand 1 were dissolved in 20 mL
of cyclohexane. The solution was brought to an autoclave. The
autoclave was put under a pressure of 15 bar CO/H2 (1:1) and
stirred at 40 °C for 2 h. The solution was transferred to a
Schlenk flask. (The solution was saturated with 13CO in the
case of the 13CO-labeled complex.) The white complex precipi-
tated when the solution was concentrated to a volume of 5 mL.
The complex was washed three times with 10 mL of pentane
and dried under vacuum. Yield: 56 mg (72%), mp 106 °C (dec).
The precise NMR data of this compound are displayed in Table
1: IR (CH2Cl2) νNCO 1695 cm-1 (vs), 1660 cm-1 (vs), νRh-CO
2019 cm-1
.
Exp er im en ta l Section
Gen er a l In for m a tion . All preparations were carried out
under an atmosphere of argon using standard Schlenk tech-
OM000890R