Time-resolved infrared spectroscopic study of reactive acyl intermediates relevant to cobalt-catalyzed carbonylations

Article abstract of DOI:10.1021/ic000378y

Time-resolved infrared spectroscopic studies have been used to characterize the reactive intermediate CH₃C(O)Co(CO)₂PPh₃ (I(Co)), which is relevant to the mechanism of the catalysis of alkene hydroformylation by the phosphine-modified cobalt carbonyls. Step-scan FTIR and (variable) single-frequency time-resolved infrared detection on the microsecond time scale were used to record the spectrum of I(Co) and to demonstrate that the principal photoproduct of the subsequent reaction of this species at P(CO) = 1 atm is the methyl cobalt complex CH₃Co(CO)₃PPh₃ (M(Co)). At higher P(CO) the trapping of I(Co) with CO to re-form CH₃C(O)Co(CO)₃PPh₃ (A(Co)) (rate = k(CO)[CO][I(Co)]) was shown to become competitive with the rate of acetyl-to-cobalt methyl migration to give M(Co) (rate = k(M)[I(Co)]). Activation parameters for the competing pathways in benzene were determined to be ΔH((+))(CO) = 5.7 ± 0.4 kJ mol^-1, ΔS((+))(CO) = -91 ± 12 J mol^-1 K^-1 and ΔH((+))(M) = 40 ± 2 kJ mol^-1, ΔS((+))(M)= -19 ± 5 J mol^-1 K^-1. The effects of varying the solvent on the competitive reactions of I(Co) were also explored, and the mechanistic implications of these results are discussed.

Full text of DOI:10.1021/ic000378y

3098
Inorg. Chem. 2000, 39, 3098-3106
Time-Resolved Infrared Spectroscopic Study of Reactive Acyl Intermediates Relevant to
Cobalt-Catalyzed Carbonylations1
Steven M. Massick,† Julienne G. Rabor,† Stefan Elbers,† Jon Marhenke,† Stefan Bernhard,‡
Jon R. Schoonover,‡ and Peter C. Ford*,†
Department of Chemistry, University of California, Santa Barbara, California 93106, and Bioscience/
Biotechnology Group (CST-4), Chemical Science and Technology Division, Los Alamos National
Laboratory, Los Alamos, New Mexico 87545
ReceiVed April 7, 2000
Time-resolved infrared spectroscopic studies have been used to characterize the reactive intermediate CH3C(O)Co(CO)2-
PPh3 (ICo), which is relevant to the mechanism of the catalysis of alkene hydroformylation by the phosphine-
modified cobalt carbonyls. Step-scan FTIR and (variable) single-frequency time-resolved infrared detection on
the microsecond time scale were used to record the spectrum of ICo and to demonstrate that the principal
photoproduct of the subsequent reaction of this species at PCO ) 1 atm is the methyl cobalt complex CH3Co-
(CO)3PPh3 (MCo). At higher PCO the trapping of ICo with CO to re-form CH3C(O)Co(CO)3PPh3 (ACo) (rate )
k
CO[CO][ICo]) was shown to become competitive with the rate of acetyl-to-cobalt methyl migration to give MCo
(rate ) kM[ICo]). Activation parameters for the competing pathways in benzene were determined to be ∆Hq
)
CO
5.7 ( 0.4 kJ mol-1, ∆S qCO ) -91 ( 12 J mol-1 K-1 and ∆H qM ) 40 ( 2 kJ mol-1, ∆S qM ) -19 ( 5 J mol-1
K-1. The effects of varying the solvent on the competitive reactions of ICo were also explored, and the mechanistic
implications of these results are discussed.
Introduction
Scheme 1 . Proposed Mechanism for Homogeneous Cobalt-
Catalyzed Hydroformylation
Carbonylation catalysis has been a major component of the
chemical industry since the discovery of homogeneous cobalt
carbonyl catalysts for alkene hydroformylation in 1938 by Otto
Roelen at Ruhrchemie AG.2,3 The original hydroformylation
catalysts were based on simple cobalt carbonyl precursors;
however, a phosphine-modified cobalt catalyst having a more
favorable linear-to-branched selectivity has been developed.2,3
Although rhodium catalysis is now predominant for propene
hydroformylation, cobalt-catalyzed processes continue to be
important for the hydroformylation of higher olefins owing to
the greater thermal stability of the cobalt catalysts at temper-
atures necessary for separating higher boiling products.3 In
addition there is continuing interest in new applications of cobalt
carbonyls as catalysts for syn-gas reactions.4-6
Pioneering studies by Heck and Breslow7 of alkene hydro-
formylation of alkenes led to the proposed catalytic cycle
illustrated by Scheme 1, and this model has served as the starting
point for quantitative reaction mechanism studies of various
aspects of unsubstituted and substituted cobalt carbonyl
catalysts.8-17 A key step in that scheme is the migratory insertion
of a CO into a Co-R bond (step 1d). This is the fundamental
carbon-carbon bond formation pathway in this and other
(8) Martin, J. T.; Baird, M. C. Organometallics 1983, 2, 1073-1078.
(9) Bartik, T.; Krummling, T.; Happ, B.; Sieker, A.; Marko, L.; Boese,
R.; Ugo, R.; Zucchi, C.; Palyi, G. Catal. Lett. 1993, 19, 383-389.
(10) Kovacs, I.; Ungvary, F.; Marko, L. Organometallics 1986, 5, 209-
215.
† University of California, Santa Barbara.
‡ Los Alamos National Laboratory.
(1) Reported in part at the 217th ACS National Meeting, Anaheim CA,
March 1999, INORG 168, and at the 213th ACS National Meeting,
San Francisco CA, April 1997, INORG 141.
(2) Parshall, G. W.; Ittel, S. D. Homogeneous Catalysis; John Wiley &
Sons: New York, 1992.
(3) Cornils, B.; Herrmann, W. A. Applied homogeneous catalysis with
organometallic compounds: a comprehensiVe handbook in two
Volumes; VCH: Weinheim, New York, 1996.
(11) Roe, D. C. Organometallics 1987, 6, 942-946.
(12) Pino, P.; Major, A.; Spindler, F.; Tannenbaum, R.; Bor, G.; Horvath,
I. T. J. Organomet. Chem. 1991, 417, 65-76.
(13) Borovikov, M. S.; Kovacs, I.; Ungvary, F.; Sisak, A.; Marko, L.
Organometallics 1992, 11, 1576-1579.
(14) Klingler, R. J.; Rathke, J. W. J. Am. Chem. Soc. 1994, 116, 4772-
4785.
(15) Rathke, J. W.; Klinger, R. J.; Kramarz, K. W.; Woelk, K. Prog. NMR
Spectrosc. 1997, 30, 209-253.
(4) Knifton, J. F.; Lin, J. J. J. Mol. Catal. 1993, 81, 27-36.
(5) Piotti, M. E.; Alper, H. J. Am. Chem. Soc. 1996, 118, 111-116.
(6) Rathke, J. W.; Klinger, R. J.; Krause, T. R. Organometallics 1991,
10, 1350-1355.
(16) Kramarz, K. W.; Klingler, R. J.; Fremgen, D. E.; Rathke, J. W. Catal.
Today 1999, 49, 339-352.
(17) Rosi, L.; Piacenti, F.; Bianchi, M.; Frediani, P.; Salvini, A. Eur. J.
Inorg. Chem. 1999, 67-68.
(7) Heck, R. F.; Breslow, D. S. J. Am. Chem. Soc. 1962, 84, 2499-2502.
10.1021/ic000378y CCC: $19.00 © 2000 American Chemical Society
Published on Web 06/03/2000

Acyl Intermediates
Inorganic Chemistry, Vol. 39, No. 14, 2000 3099
industrially important homogeneous catalysis mechanisms for
the carbonylation of organic feedstocks.
Migratory insertion has been extensively studied for model
compounds such as RMn(CO)5 and CpFe(CO)2R (Cp ) η5-
C5H5, R ) alky), e.g., eq 1.18-21 For example, kinetics and
system clearly has an analogy and relevance to the industrially
important, phosphine-modified cobalt carbonyl catalyst de-
scribed above. These studies have been carried out using a high-
pressure and variable temperature flow cell reactor27 in order
to examine the reactivity of presumed intermediates under a
wide range of CO partial pressures (PCO) and temperatures.
Experimental Section
Materials. The syntheses of CH3C(O)Co(CO)3PPh3 and CF3C(O)-
Co(CO)3PPh3 were carried out according to literature procedures.7,32
Solvents were reagent grade and subjected to standard drying and
distillation procedures.33 Research grade carbon monoxide (99.995%
purity from Spectra Gases) was used without further purification.
Solutions for TRIR Experiments. Solvents for flash photolysis
experiments were deoxygenated by entrainment with argon or carbon
monoxide or by repeated freeze-pump-thaw cycles. The concentra-
tions of the cobalt solutions were typically 2-3 mM. This resulted in
initial absorbances of 0.3-0.4 for the most intense of the terminal
carbonyl stretches of the parent complex for a 0.5 mm path length.
Solutions were prepared under various CO pressures (PCO) and, after
changes in PCO, were stirred and given a minimum of 45 min to
equilibrate. Concentrations of CO were calculated from published
solubility data.34,35 Both the step-scan and the single-frequency TRIR
experiments were carried out using flowing sample solutions in order
to minimize any complications arising from the accumulation and
secondary photolysis of photoreaction products. The temperature of
the solutions in the flow system and sample cell was regulated.
Step-Scan FTIR. The step-scan FTIR (BioRad FTS 60A/896)
instrument at LANL has been described in previous publications.36-38
In the configuration used here, a BioRad Fast TRS board controlled
both the motion of the mirrors and data collection at a repetition rate
of 10 Hz. The firing of the photolysis laser (355 nm output of a Nd:
YAG laser) was synchronized to the mirror movement by a digital delay
generator (Stanford Research Systems, model DG535) triggered by the
TRS board. The spectral window of the instrument was 2250-1250
cm-1, limited by a low-pass germanium optical filter at 2250 cm-1
and the CaF2 windows of the IR cell at the lower frequency. The
stereochemical studies of the reaction depicted in eq 1 concluded
that this first involves migration of the methyl group from the
metal to the cis CO ligand followed by trapping of the resulting
“unsaturated” acyl intermediate [CH3C(O)Mn(CO)4] (IMn) by
CO.
Owing to their low steady state concentrations, such inter-
mediates are difficult to observe directly under thermal reaction
conditions. Therefore, this laboratory has utilized flash pho-
tolysis to probe the quantitative reactivities of possible inter-
mediates in the migratory insertion reactions of model com-
pounds noted in eq 1.22-27 The method involves ligand
photolabilization from an acyl complex such as A to generate
non-steady-state concentrations of the coordinatively “unsatu-
rated” reactive intermediate I (eq 2). Time-resolved infrared
(TRIR) and time-resolved optical (TRO) detection28-31 are then
used to interrogate the spectra and reaction dynamics of I. The
key reactions observed are the trapping of I by various ligands,
including CO, to regenerate acyl complexes and the competing
migration of the alkyl group from the acyl group of I to the
metal center (the reverse of the insertion step). The dependence
of the rate constants kL and kM obtained, respectively, for these
two pathways with variables such as the nature of R and of the
solvent provides valuable insight regarding the transition states
for these pathways.
resolution of the transient difference spectra thus generated was 2 cm-1
.
The photolysis laser beam and the infrared output of the interfer-
ometer were overlapped and focused to a spot size of approximately 2
mm diameter onto a 0.5 mm path length CaF2 IR cell (International
Crystal Labs) through which the sample solutions of ACo were flowed.
The intensity of the transmitted IR beam was monitored by sampling
the output of a mercury cadmium telluride (MCT) detector (Graseby
1710117, rise time 250 ns) and amplifier (Graseby DP-8000-4 amplifier,
rise time of 1 µs) every 200 ns for 80 µs before and for 80 µs after the
photolysis laser pulse at each mirror step. One transient was recorded
for each interferometer step, so multiple scans were averaged to improve
the signal-to-noise (s/n) ratio. Thus, an array of interferograms with
respect to time was constructed during each step-scan experiment from
which were extracted transient difference spectra for different times.
Transient difference spectra were generated using the expression ∆A(t,ν)
) -log(I(t,ν)/I0(ν)) where I0(ν) is the average transmitted intensity
before the laser pulse at frequency ν.
Described here are analogous investigations using the cobalt
complex CH3C(O)Co(CO)3(PPh3) (ACo) as a precursor. This
(18) Atwood, J. D. Inorganic and organometallic reaction mechanisms;
Brooks/Cole Publishing Co.: Monterey, CA, 1985.
(19) Cavell, K. J. Coord. Chem. ReV. 1996, 155, 209-243.
(20) Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G. Principles
and applications of organotransition metal chemistry; University
Science Books: Mill Valley, CA, 1987.
(21) Jordon, R. B. Reaction Mechanisms of Inorganic and Organometallic
Systems, 2nd ed.; Oxford University Press: New York, 1998.
(22) Boese, W. T.; Ford, P. C. Organometallics 1994, 13, 3525-3531.
(23) Boese, W. T.; Ford, P. C. J. Am. Chem. Soc. 1995, 117, 8381-8391.
(24) McFarlane, K. L.; Ford, P. C. Organometallics 1998, 17, 1166-1168.
(25) McFarlane, K. L.; Lee, B.; Fu, W. F.; van Eldik, R.; Ford, P. C.
Organometallics 1998, 17, 1826-1834.
Unless otherwise noted, the solvent used in the step-scan FTIR
experiments was deuterated benzene (Aldrich). Carbon monoxide
(Matheson, scrubbed for both oxygen and water) was bubbled through
(32) Lindner, E.; Zipper, M. Chem. Ber. 1974, 107, 1444-1455.
(33) Riddick, J. A.; Bunger, W. B.; Sakano, T. K. Organic SolVents Physical
Properties and Methods of Purification, 4th ed.; John Wiley & Sons:
New York, 1986; Vol. II.
(34) IUPAC Solubility Data Series: Carbon Monoxide; Pergamon Press:
Oxford, 1990; Vol. 43.
(35) Payne, M. W.; Leussing, D. L.; Shore, S. G. J. Am. Chem. Soc. 1987,
87, 617-618.
(36) Schoonover, J. R.; Strouse, G. F. Chem. ReV. 1998, 98, 1335-1355.
(37) Schoonover, J. R.; Strouse, G. F.; Dyer, R. B.; Bates, W. D.; Chen,
P. Y.; Meyer, T. J. Inorg. Chem. 1996, 35, 273-274.
(38) Schoonover, J. R.; Strouse, G. F.; Omberg, K. M.; Dyer, R. B.
Comments Inorg. Chem. 1996, 18, 165-188.
(26) Belt, S. T.; Ryba, D. W.; Ford, P. C. J. Am. Chem. Soc. 1991, 113,
9524-9528.
(27) Massick, S. M.; Ford, P. C. Organometallics 1999, 18, 4362-4366.
(28) DiBenedetto, J. A.; Ryba, D. W.; Ford, P. C. Inorg. Chem. 1989, 28,
3503-3507.
(29) Ford, P. C.; DiBenedetto, J. A.; Ryba, D. W.; Belt, S. T. SPIE Proc.
1992, 1636, 9-16.
(30) Ford, P. C.; Bridgewater, J. S.; Lee, B. Photochem. Photobiol. 1997,
65, 57-64.
(31) Ford, P. C.; Bridgewater, J. S.; Massick, S.; Marhenke, J. Catal. Today
1999, 49, 419-430.

3100 Inorganic Chemistry, Vol. 39, No. 14, 2000
Massick et al.
Figure 1. Infrared spectrum of a 2.14 mM solution of the parent
complex CH3C(O)Co(CO)3PPh3 (ACo) in deuterated benzene equili-
brated with 1 atm of carbon monoxide. The peak positions are 1680
Figure 2. Infrared transient difference spectra of CH3C(O)Co(CO)3-
PPh3 (ACo) in C6D6 recorded after 355 nm flash photolysis. The overlay
of spectra indicates 5 µs intervals ranging from 0 to 40 µs following
the laser pulse.
cm-1 for the acyl νCO, 1959, 1979, 2048 cm-1 for the terminal νCO
,
and 2131 cm-1 for free CO in solution.
1979, and 1959 cm-1 with the degeneracy of the lower energy
E mode split because of a lowering of the symmetry by the
acyl group. The weaker acyl carbonyl stretch is also observable
the solutions for 20 min. The solutions were then transferred to a 50
mL gastight syringe (Hamilton) and attached to a flow system consisting
of the CaF2 IR cell followed by another 50 mL gastight syringe. Two
syringe pumps were alternately used to cycle the solution from syringe
to syringe at a rate of approximately 2 mL/min during the course of
the experiment. Typically, at least four scans were averaged to obtain
a satisfactory s/n ratio.
at νCO ) 1680 cm-1
.
Step-Scan FTIR Studies. These experiments at LANL
provided a clear insight into the events following the 355 nm
photolysis of the parent complex ACo. The spectra from
sequential 5 µs intervals following the photolysis pulse are
shown in Figure 2. This clearly depicts the prompt bleach of
the parent terminal νCO absorbances at 2048, 1979, and 1959
cm-1 characteristic of ACo and the prompt formation and decay
of a transient species ICo with a strong isolated absorbance at
1915 cm-1. The bleach of the acyl νCO band at 1680 cm-1 was
also seen. However, the interpretation of the temporal absorb-
ance changes between 1965 and 1940 cm-1 is less clear. The
bleach of the terminal νCO of ACo at 1959 cm-1 and an apparent
prompt absorbance at 1946 cm-1 attributable to ICo overlap with
other absorbance changes attributed to formation of a “perma-
nent” photoproduct.42
The temporal response of the bleach of ACo at 1979 cm-1
for the transient absorbance of the intermediate ICo at 1915 cm-1
and for the product formation at 1958 cm-1 are shown in curves
a, b, and c of Figure 3, respectively. The behavior of the product
absorbance at 1958 cm-1 (Figure 3c) merits some comment
because this region overlaps directly with absorbance changes
arising from ACo. The baseline reference for the temporal
transient absorption traces is normally defined by a selecting a
frequency interval in a spectral region where there are no
observed absorbance changes. However, in this instance a
correction for the overlap of the bleach of ACo at 1959 cm-1
with the product absorbance at 1958 cm-1 was approximated
by selecting a frequency interval for the baseline reference that
includes the bleach of ACo at 1979 cm-1 to account for both
the bleach and regeneration of ACo. The frequency intervals for
the product and baseline regions were 1960-1957 and 1980.5-
1977.5 cm-1, respectively. (Alternatively, a similar kinetic trace
with the same rate of exponential rise but much lower amplitude
could be obtained by selecting a narrow frequency interval
centered about 1953 cm-1.) The kinetic trace of product
Single-Frequency Time-Resolved Infrared Studies. The apparatus
for TRIR studies at UCSB has been described previously.28,30 In the
present configuration, the probe source consists of four lead-salt diode
lasers mounted on a Spectra Physics/Laser Analytics model SP5731
laser head and a CVI Digikrom model 240 monochromator. This
configuration provides a tunable infrared probe laser source with
frequencies between 2200 and 1500 cm-1. The probe laser beam was
focused to a 7 mm diameter spot and overlapped at the plane of the
sample cell with the 355 nm pump pulse from a Lumonics HY600
Nd:YAG laser. The transmitted intensity of the infrared probe beam
was focused to fill the 1 mm2 active area of a Fermionics model PV-
8-1 photovoltaic Hg/Cd/Te detector. A Fermionics model PVA 500-
50 preamplifier increased the signal from the detector 20-fold prior to
digitization by a LeCroy 9400 oscilloscope at a rate of 100 MHz.
A recently constructed high-pressure and variable-temperature
(HPVT) flow system27 was used in these studies. This system allows
for the equilibration of sample solutions in a modified Parr bomb with
gas pressures up to 100 atm and temperatures up to 150 °C. During
the photolysis experiments, sample solutions of a known CO concentra-
tion and temperature were allowed to flow from the Parr Bomb through
a high-pressure IR cell modified on the basis of modifications to a
published design39 of a low-pressure reservoir. The jacketed transfer
lines and HPVT IR cell were all maintained at the same temperature
and pressure as the Parr bomb reservoir.
Results
The infrared absorption spectrum of the parent complex
CH3C(O)Co(CO)3PPh3 (ACo) is shown in Figure 1. This is
consistent with a trigonal bipyramidal arrangement of the five
ligands with the three carbonyls in the equatorial plane and the
acyl and phosphine groups in the axial sites.40,4140,41 The A1
and E modes expected for the terminal carbonyl stretches in
local C3V symmetry are observed in benzene-d6 at νCO ) 2048,
(39) Noack, K. Spectrochim. Acta 1968, 43, 1024-1026.
(40) Kovacs, I.; Ungvary, F. Coord. Chem. ReV. 1997, 161, 1-32.
(41) Somlyai-Haasz, J.; Haasz, F.; Galamb, V.; Benedetti, A.; Zucchi, C.;
Palyi, G.; Krummling, T.; Happ, B.; Bartik, T. Organometallics 1991,
10, 205-217.
(42) Given that over a much longer time scale this photoproduct reacts
with CO to regenerate ACo, the term “permanent” refers only to spectral
changes on the time scale of the TRIR experiments.

Acyl Intermediates
Inorganic Chemistry, Vol. 39, No. 14, 2000 3101
Figure 3. Absorption changes following 355 nm flash photolysis of
CH3C(O)Co(CO)3PPh3 (ACo) in C6D6 monitored at (a) 1979 cm-1
corresponding to the bleach of the parent complex ACo, (b) 1915 cm-1
due to the absorption of the transient intermediate CH3C(O)Co(CO)2-
PPh3 (ICo), and (c) 1958 cm-1 showing the formation of the methyl
cobalt complex CH3Co(CO)3PPh3 (MCo) (generated by setting a baseline
region about 1979 cm-1 to correct for the parent bleach at 1959 cm-1).
Figure 5. (a) Transient difference spectrum extracted from the first
1.5 µs following 355 nm photolysis of CH3C(O)Co(CO)3PPh3 (ACo
)
in C6D6 (solid line) and the negative of the transient difference spectrum
for the time period from 40 to 80 µs (dashed line). (b) Resultant
spectrum from the addition of the spectra in (a) (solid line) and the
weighted product spectrum of CH3Co(CO)3PPh3 (MCo) as shown in
Figure 4b (dashed line). (c) Resultant spectrum of the intermediate
formed during the first 1.5 µs following the photolysis laser pulse.
Scheme 2
(MCo). However, there is a broad weak feature between 2030
and 1990 cm-1 possibly attributable to solvent heating because
benzene-d6 has some absorbance in this region. These results
are consistent with the scenario illustrated in Scheme 2;
photolysis of the acetyl complex ACo gives a reactive intermedi-
Figure 4. (a) Transient difference spectrum extracted from the time
interval 40-80 µs following the 355 nm photolysis of CH3C(O)Co(CO)3-
ate ICo that reacts further to give MCo or to re-form ACo
.
PPh3 (ACo) in C6D6, [CO] = 5mM, and the infrared spectrum of ACo
.
The spectrum of ICo can be generated by a similar analysis
of the difference spectrum extracted from the first 1.5 µs after
the pump laser pulse (Figure 5a). This (0-1.5 µs) difference
spectrum was corrected for “permanent” absorbance changes
such as the bleach of ACo or product formation after the first
1.5 µs by subtracting the difference spectrum for the final 40
µs (Figure 4a) weighted to compensate for the bleach of the
acetyl group at 1680 cm-1. However, as can be seen in Figure
5b, some of the MCo has already formed during the first 1.5 µs.
Thus, addition of a weighted spectrum of MCo (Figure 4b) from
Figure 5b generates a spectrum with absorbance changes at
1915, 1947, 2035, and possibly 1983 cm-1. The absorbance of
ICo at 1915 cm-1 provides an excellent region of the IR for
monitoring the kinetics of the intermediate ICo free from overlap
of absorbance changes from either parent ACo or photoproduct
(b) Sum of difference spectrum and enough of the parent spectrum to
equal the bleach at 1680 cm-1
.
formation at 1958 cm-1 fits well to an exponential rise with an
observed rate constant (kobs) of approximately 1.3 × 105 s-1
.
This value is in agreement with the kobs (1.2 × 105 s-1) obtained
from an exponential fit of the decay of ICo at 1915 cm-1 (Figure
3b), so it is reasonable to assume that ICo reacts directly to form
this photoproduct. Also of note, under these conditions (PCO
0.76 atm) the regeneration of ACo on this time scale is at most
=
quite modest (Figure 3a).
Since the decay of ICo is essentially complete by 40 µs, it is
possible to characterize the “permanent” product of the 355 nm
photolysis of ACo by extracting a difference spectra for the 40-
80 µs after photolysis (Figure 4a). The components of the (40-
80 µs) difference spectrum are the bleach of the parent spectrum
and the absorbance spectra associated with the photoproduct(s).
The latter can be generated by adding a weighted amount of
the ACo spectrum (Figure 4a) to the extracted difference
spectrum. Since the acetyl group νCO at 1680 cm-1 does not
overlap with other bands, the amplitude of the absorbance
change at this frequency provides a convenient measure of the
weighting factor. The resulting spectrum (Figure 4b) appears
to be that of the expected methyl complex43 CH3Co(CO)3PPh3
MCo
.
Single-Frequency TRIR Studies The single-frequency de-
tection methodology of the TRIR apparatus at UCSB is
especially well-suited for accurate kinetics measurements.
Furthermore, the HPVT flow system27 allows the laser pho-
tolysis experiments to be conducted on sample solutions of ACo
equilibrated under a wide range of PCO and temperature. In the
present case, the range of PCO employed was from 0 to 11 atm
(43) Hieber, W.; Lindner, E. Chem. Ber. 1961, 94, 1417-1425.

3102 Inorganic Chemistry, Vol. 39, No. 14, 2000
Massick et al.
Figure 8. Transient difference spectrum in the acyl stretching region
for the initial 0.5 ms following the photolysis of ACo in benzene at 25
°C and [CO] ) 0.0324 M. Transient absorptions were collected at single
frequencies in a point-by-point fashion. The inset plot depicts the change
in absorption at 1639 cm-1 (a) and 1679 cm-1 (b). The temporal trace
(a) is an average of three separate 80 shot acquisitions and that of (b)
is a single 80 shot acquisition.
Figure 6. Overlay of transient absorptions of CH3C(O)Co(CO)2PPh3
(ICo) at 1915 cm-1 following 355 nm flash photolysis of CH3C(O)Co-
(CO)3PPh3 (ACo) in benzene at 25 °C and [CO] ) 0, 20, 32, 45, 58,
71, and 84 mM from top to bottom, respectively. The initial absorbance
change of the individual traces were normalized to unity to better
illustrate the change in the decay rate. The inset graph is a plot of the
observed rate constant, kobs, obtained from single-exponential fits vs
[CO] for the same data.
Table 1. Rate Constants for Methyl Migration, kM, and for
Reaction with CO, kCO, Following 355 nm Photolysis of
CH3C(O)Co(CO)3PPh3 and CF3C(O)Co(CO)3PPh3 in Various
Solvents at 25 °C
solvent
kM (105 s-1
)
kCO (106 M-1 s-1
)
CH3C(O)Co(CO)3PPh3
12 ( 0.4
tetrahydrofuran
benzene
dichloromethane
dichloroethane
7.8 ( 0.5
11.4 ( 0.1
16 ( 1
0.62 ( 0.07
1.9 ( 0.2
3.5 ( 0.2
10 ( 0.8
CF3C(O)Co(CO)3PPh3
dichloroethane
0.46 ( 0.01
150 ( 6
function of [CO] can be modeled by a simple competition
between two available reaction paths: a unimolecular methyl
migration with the associated rate kM to form MCo; and a
bimolecular reaction with CO with the associated rate kCO to
re-form ACo (Scheme 2), with the observed rate constant kobs
) kCO[CO] + kM. Accordingly, a plot of kobs vs [CO] is linear
(Figure 6 inset) with the slope kCO ) 1.14 ((0.01) × 107 M-1
s-1 and the intercept kM ) 6.2 ((0.7) × 104 s-1
.
The reactivity of ICo generated by flash photolysis of ACo
was also investigated in 1,2-dichloroethane (DCE), dicholoro-
methane (CH2Cl2), and tetrahydrofuran (THF) solutions at 25
°C. In each case plots of kobs vs [CO] were linear with nonzero
intercepts (Figure 7). The kCO and kM values so obtained are
summarized in Table 1. Notably, there is little variation in kCO
with the nature of the solvent; the highest value of 1.6 ((0.1)
× 107 M-1 s-1 in CH2Cl2 is only a factor of 2 higher than the
lowest in THF. The transient spectra also display little change
Figure 7. Plot of the observed rate constant kobs vs the CO concentra-
tion for benzene, dichloroethane (DCE), tetrahydrofuran (THF) as
indicated for the decay of the transient intermediate CH3C(O)Co(CO)2-
PPh3 (ICo). Also included is the plot of kobs vs [CO] for the
trifluoroacetylcobalt complex (IF).
while T was varied from 25 to 45 °C. Variation of PCO over
this range was necessary to determine the reactivities of ICo
toward CO to re-form ACo, while variation of T provides the
opportunity to measure the activation parameters ∆H q and ∆S q
for both the kM and kCO pathways.
Shown in Figure 6 are the curves representing the decay of
ICo monitored at 1915 cm-1 in 25 °C benzene solutions
equilibrated with various PCO ranging from 0 to 11.3 atm. These
PCO values correspond to CO concentrations from 0 to 84
mM.33,34 The curves can be fit as single-exponential decays,
the rates of which increase linearly with [CO] (Figure 7). This
is consistent with the observation that on the microsecond to
millisecond time scale the bleaches of the parent complex at
1980 and 1679 cm-1 (Figure 8) both recover substantially in
the presence of high [CO] but very little for solutions equili-
brated under ∼0.76 atm CO (Figure 3). The decay of ICo as a
in νCO values found for ICo of 1915 ((2) cm-1, 1914 ((2) cm-1
,
and 1913 ((4) cm-1 in deuterated benzene, DCE, and THF,
respectively. These were determined both by the step-scan FTIR
technique (DCE and benzene) and from the point-by-point
single-frequency TRIR method (THF).
In contrast to the behavior of kCO, the rate constant for methyl
migration kM is more sensitive to the nature of the solvent (Table
1). For example, the value found in THF (kM ) 12 ((0.4) ×
105 s-1) is almost 20 times faster than that measured in benzene
(kM ) 6.2 ((0.7) × 104 s-1) at 25 °C. The values obtained for
kM in the less donating solvents benzene, CH2Cl2, and DCE
follow the order of the dielectric constants (ꢀr ) 2.3, 8.9, 10.4,

Acyl Intermediates
Inorganic Chemistry, Vol. 39, No. 14, 2000 3103
a transient species with a νCO at 1935 cm-1 corresponding to
the production of a transient intermediate presumably, CF3C(O)-
Co(CO)2PPh3 (IF). A transient absorbance for the acyl νCO was
also observed to lower energy for IF at 1650 cm-1 shifted only
16 ( 3 cm-1 from the acyl νCO of AF.
The temporal behavior of the transient absorbances attributed
to IF also fit well to single-exponential decays with kobs values
strongly dependent on [CO]. The plot of kobs vs [CO] for IF
proved to be linear with the slope kCO(25 °C) ) 1.5 ((0.06) ×
108 M-1 s-1 and a nonzero intercept kM(25 °C) ) 4.6 ((0.1)
× 104 s-1 (Table 1). Notably, kM is an order of magnitude
smaller and kCO is 15 times larger than the respective values
for ICo in DCE.
Table 2. Rate Constants and Activation Parameters for Methyl
Migration, kM, and for Reaction with CO, kCO, Following 355 nm
Photolysis of CH3C(O)Co(CO)3PPh3 in Benzene
temp (°C)
kM (104 s-1
)
kCO (107 M-1 s-1
)
25
30
35
40
45
6.2 ((0.7)
7.6 ((0.5)
10.5 ((0.9)
13.5 ((0.6)
18.0 ((0.8)
1.14 ((0.01)
1.19 ((0.01)
1.26 ((0.02)
1.34 ((0.01)
1.40 ((0.02)
∆H q (kJ mol-1
)
40 ((2)
-19 ((5)
5.7 ((0.4)
-91 ((12)
∆S q (J mol-1 K-1
)
respectively).33 Such an observation might suggest that the
solvent effect on the methyl migration pathway is due to
stabilization of a transition state for the kM pathway more polar
than ICo. However, the kM value measured in THF (ꢀr ) 7.6) is
anomalous and suggests that (at least in this case) the solvent
may be acting in a more intimate manner, perhaps as a donor
via the THF oxygen.
Discussion
The 355 nm flash photolysis of ACo leads to CO dissociation
and prompt formation of the “unsaturated” acetyl complex
CH3C(O)Co(CO)2PPh3, ICO. This reacts with CO (kCO) to re-
form the initial complex in competition with methyl group
migration from the acyl group to the cobalt (kM), in steps
analogous to steps d and e of Scheme 1, respectively. The
transient IR spectra are consistent with ICo being the primary
photoproduct, although manipulations of the step-scan spectral
data leave some doubt as to whether the minor product CH3C-
(O)Co(CO)3 (I′Co) is formed by phosphine photodissociation.
The latter should have higher frequency νCO bands consistent
with the observation of weaker bands at 2035 and possibly 1983
cm-1 in the step-scan spectra (Figure 5c). Earlier photolysis
studies by Sweany44 of CH3C(O)Co(CO)4 (A′Co) in argon
matrices demonstrated the formation of a species formulated
as I′Co, which displayed bands at 2081(m), 2008(s), 1977(s),
and 1686 (w) cm-1, so it is possible that the two stronger bands
of these correspond to the unassigned weak bands seen in Figure
5c. However, the absorbance at 1983 cm-1 is close to the ACo
band at 1979 cm-1 and its intensity varies depending on how
the bleach of the latter band is weighted in the difference
spectrum,45 while that at 2035 cm-1 overlaps directly with the
symmetric νCO of MCo. This peak appears to be invariant to
the manner by which the transient spectra are generated and
may indicate that some MCo formed promptly during the flash
photolysis of ACo. Indeed, regeneration of the ACo band at 1979
cm-1 was observed to be consistently somewhat less than
predicted by ratios of kM and kCO[CO]. Similar prompt (minor)
To determine the activation parameters ∆H q and ∆S q for
the two competitive pathways, kCO and kM, the kinetics for the
decay of ICo were examined by TRIR spectroscopy in benzene
over the temperature range 25-45 °C in 5 °C increments. These
data are summarized in Table 2. Plots of kobs vs [CO] are linear
at each temperature, and both kCO and kM increase with T.
However, the temperature sensitivity of kCO is slight (indicative
of a small enthalpic barrier), while that of kM is much larger.
From a linear least-squares fit of an Eyring plot (Figure S-2 of
Supporting Information) for kCO the enthalpy of activation
∆H q and entropy of activation ∆S q were determined to
CO
CO
be 5.7 ( 0.4 kJ mol-1 and -91 ( 12 J mol-1 K-1, respectively.
An Eyring plot for kM gave ∆H q and ∆S q values for the
M
M
methyl migration pathway of 40 ( 2 kJ mol-1 and -19 ( 5 J
mol-1 K-1, respectively.
Although the transient difference spectra obtained from the
step-scan FTIR experiments in the solvents C6D6 and DCE did
not show any absorbances in the acyl region that could be
convincingly related to ICo, the TRIR photolysis experiments
using the single-frequency probe source were successful in this
regard. Shown in Figure 8 is a transient difference spectrum
(collected point by point) of the acyl region averaged over a
0.5 µs interval immediately following the flash photolysis of
ACo in benzene. Prompt bleaching of the parent acyl νCO at 1680
cm-1 is seen as well as the formation of a new absorbance at
1639 cm-1. The inset plot in Figure 8 shows the time evolution
of the signals at 1680 and at 1639 cm-1 for TRIR experiments
carried out with PCO ) 4.4 atm. Despite the weakness of the
acyl signal, the regeneration of ACo (kobs ) 4.6 ((0.1) × 105
s-1) nearly matches the kobs value of 4.4 ((0.14) × 105 s-1
calculated for ICo in 25 °C benzene from the data in Table 1
for 32.4 mM CO. The even weaker decay of the absorbance
change at 1639 cm-1 (kobs ) 3.5 ((0.4) × 105 s-1) is also in
reasonable agreement. These observations are taken as evidence
that the absorbances at 1639, 1915, and 1947 cm-1 are all
characteristic of a single species ICo and that loss of a carbonyl
ligand from ACo shifts the acyl carbonyl stretching frequency
formation of the methyl complex was observed in the flash
22
photolysis of CH3C(O)Mn(CO)5, AMn
.
Regardless of these
possible minor pathways, the dynamics of ICo decay, ACo re-
formation, and MCo formation are all shown to be closely
coupled and unaffected by the possible formation of a minor
secondary product, so the ensuing discussion will not be further
concerned with this issue.
What is the Structure of ICo? The step-scan FTIR spectrum
recorded during the first 1.5 µs following photolysis of ACo in
(44) Sweany, R. L. Organometallics 1988, 8, 175-179.
(45) Mixing in a weighted spectrum of ACo instead of subtracting the
difference spectrum from the time interval 40-80 µs results in a
somewhat stronger absorbance at 1983 cm-1. (In either case, an
additional correction is required to compensate for the amount of the
final product MCo formed during the first 1.5 µs, and this spectrum of
41 ((2) cm-1 lower in energy for ICo
.
CF3C(O)Co(CO)3PPh3. To investigate the role that the acetyl
group plays in the stabilization of ICo, photolysis experiments
on the trifluoroacetylcobalt derivative CF3C(O)Co(CO)3PPh3
(AF) were performed. The infrared spectrum of AF displays νCO
at 2067(w), 2012(s), 1986(s), and 1666(m) cm-1, similar to the
spectrum of ACo but with νCO bands generally shifted to higher
energy. Flash photolysis of AF (355 nm) in DCE at 25 °C leads
to prompt bleaching of the parent νCO absorption and generates
M
Co is itself generated by addition of the spectrum of ACo (only 3-4%)
to the 40-80 µs difference spectrum.) Notably, the absorbance of the
peak of ACo at 1979 cm-1 is 0.3 (50% ∆ transmittance) in the spectrum
used in these subtractions and any nonlinear detector response could
affect line shapes to the extent that would result in poor subtractions
with the weak transient spectra showing absorbance changes of only
0.003 (0.7% ∆ transmittance) at 1983 cm-1. This latter consideration
gives greater confidence to the transient spectra mixed with the least
amount of the spectrum of ACo
.

3104 Inorganic Chemistry, Vol. 39, No. 14, 2000
Massick et al.
Scheme 3
However, the above data do not differentiate between the
structures C and B for ICo, since both involve intramolecular
stabilization of the vacant coordination site by the acetyl ligand.
The absorbance changes in the acyl region clearly show a νCO
band at 1639 cm-1 attributed to ICo shifted 41 cm-1 to lower
frequency and with a relative intensity no more than a third
that of the acyl band for ACo at 1680 cm-1. In Sweany’s study44
of CH3C(O)Co(CO)4 (A′Co) in argon matrices, 310 nm pho-
tolysis produced the unsaturated complex I′Co with an acyl band
benzene-d6 (Figure 5c) displays the νCO peaks at 1947 and 1915
cm-1 with the expected shift to lower energy of the νCO of the
remaining terminal CO ligands upon loss of one CO. Interpreta-
tion of the spectra and kinetics of the “unsaturated” species ICo
can be discussed in terms of three potential structures, the cyclic
η2-acyl complex C, the solvent-stabilized species S, or the
bidentate â-agostic complex B. In each structure the vacant
coordination site is stabilized via interaction with a Lewis base,
i.e., interaction with the lone pairs of the carbonyl oxygen or
with a C-H bonding electron pair of the acyl methyl or with a
C-H or lone pair from the solvent.
It is evident from Table 1 that the rate constant kCO for the
reaction of ICO with CO is relatively insensitive to the solvent
medium; kCO in THF is only a factor of 2 less than in CH2Cl2
and about 30% less than in benzene. Since the rates of CO
reaction with various solventometal carbonyl species such as
Cr(CO)5(Sol)46 and CH3Mn(CO)4(Sol)22 are markedly dependent
on the nature of Sol, the behavior of ICo argues against this
species having the S configuration in the four solvents discussed
here. Instead it is likely that the site left open by CO
photodissociation is stabilized by coordination to the acetyl
group. Such coordination may be via η2-binding of the acyl
carbonyl as in C or by an agostic interaction with a methyl
C-H as in B. This conclusion is reinforced by the insensitivity
of the principal νCO band of ICo at ∼1915 cm-1 to the same
solvents.
at 1686 cm-1, a comparable 38 cm-1 shift from that for A′Co
.
Under these conditions I′Co was remarkably stable even in the
presence of CO and H2, whereas the methyl and hydride
analogues CH3Co(CO)3 and HCo(CO)3 (from photolysis of the
methyl and hydridotetracarbonylcobalt complexes,48 respec-
tively) both react readily with CO and H2. Furthermore, the
relative intensity of the acyl νCO observed for I′Co was only
30% that of the parent complex A′Co. Clearly, the reactivity
patterns and spectral properties for ICo and I′Co are sufficiently
similar that it is likely that their structures are also similar.
Sweany has attributed the diminished intensity of the acyl
ν
CO band to a cyclic η2-conformation for the acyl group of I′Co
,
resulting in weakening of the dipole moment derivative of the
C-O stretch due to opposing changes in the dipoles of the
Co-O and Co-C bonds in C′. This assignment draws support
from density functional calculations on reactive intermediates
of the unmodified cobalt-catalyzed hydroformylation cycle by
Ziegler et al.,49-51 which predict the η2-acyl structure (C′) for
CH3C(O)Co(CO)3 as the most stable conformation for I′Co. Two
η2-conformations were predicted, one with the acyl group in
the equatorial site. The other with the acyl carbon in the axial
site and the oxygen pointing toward the equatorial site as shown
in C is predicted to be the more stable.50 The C′ conformations
were predicted to be lower in energy than the corresponding
agostic B′ conformations by a minimum of 16 kJ mol-1 in each
case.
In accord with the interpretation that ICO is not the solvento
complex S, it is notable that the entropies of activation for the
“unimolecular” acetyl-to-cobalt methyl migration (∆S qM ) -19
J mol-1 K-1) and the bimolecular CO addition (∆S qCO ) -91
J mol-1 K-1) determined in benzene are both negative. This
would further indicate that, in this medium, neither pathway is
dominated kinetically by a step involving solvent dissociation.
These observations and conclusions can be compared to an
earlier TRIR study here of the reaction dynamics and spectra
There are now numerous examples of stable η2-acyl com-
plexes that have been structurally characterized.52,53 These
generally show lower acyl νCO frequencies than displayed by
either ICo or I′Co. However, a more important comparison would
be the shift in the acyl band when a terminal CO and the
resulting vacancy is occupied by an η1-acyl rearranging to η2-
coordination. For example,52 the iron(II) complex Fe(η1-
C(O)CH2CMe3)Br(CO)2dippe (dippe is 1,2-bis(diisopropylphos-
phino)ethane) displays terminal νCO bands at 2004 and 1959
cm-1 and an acyl νCO absorption at 1634 cm-1 while the
structurally characterized Fe(η2-C(O)CH2CMe3)Br(CO)dippe
has analogous bands at 1908 and 1591 cm-1. The qualitative
pattern of the shift in the terminal νCO frequencies and the 43
cm-1 difference in acyl νCO frequencies between η1- and η2-
acyl complexes are comparable to those observed for ACo and
of the analogous manganese intermediate CH3C(O)Mn(CO)4,
22,23
IMn
.
That system also showed little variation of kCO from
various weak donor solvents ranging from perfluoromethyl-
cyclohexane to CH2Cl2, and it was argued that IMn has the η2-
acyl configuration C in these media. Consistent with this
suggestion was the shift of the acyl νCO from 1661 cm-1 for
A
Mn to 1607 cm-1 for IMn in alkane solutions, although it should
ICo. Such spectral observations are consistent with, but do not
be noted that replacing CO with nearly any ligand would be
expected to shift this band to lower frequency. However, in THF,
kCO was more than an order of magnitude smaller47 and the
terminal νCO bands for IMn exhibited a clear shift to lower
energy, leading to the interpretation that, unlike ICo, IMn has
the solvento configuration in THF.
prove, the assignment of ICo (and I′Co) as having the C
conformation.
Notably, the terminal νCO shifts between the IR spectrum of
AF and the TRIR spectrum of IF in DCE are similar, but the
shift in the acyl νCO in that case (16 ( 3 cm-1) is much smaller.
This may imply that the electron-withdrawing nature of the CF3
decreases the basicity of the acyl carbonyl so that η2-coordina-
(46) Kelly, J. M.; Long, C.; Bonneau, R. J. Phys. Chem. 1983, 87, 3344-
3349.
(47) Owing to the limitations of the TRIR apparatus, the upper limit for
kCO in THF was determined to be <5 × 102 M-1 s-1, which is an
order of magnitude slower than the kCO values measured in CH2Cl2
or cyclohexane. However, for the faster reaction with P(OMe)3 (IMn
+ L f Mn(CO)3L(C(O)CH3)) the kL value in THF (1.7 × 103 M-1
s-1) was 3 orders of magnitude slower than that in cyclohexane (1.4
× 106 M-1 s-1).23
(48) Sweany, R. L.; Russell, F. N. Organometallics 1988, 7, 719-727.
(49) Versluis, L.; Ziegler, T.; Baerends, E. J.; Ravenek, W. J. Am. Chem.
Soc. 1989, 111, 2018-2025.
(50) Versluis, L.; Ziegler, T. Organometallics 1990, 9, 2985-2992.
(51) Sola, M.; Zeigler, T. Organometallics 1996, 15, 2611-2618.
(52) Hermes, A. R.; Girolami, G. S. Organometallics 1987, 7, 394-401.
(53) Durfee, L. D.; Rothwell, I. P. Chem. ReV. 1988, 88, 1059-1079.

Acyl Intermediates
Inorganic Chemistry, Vol. 39, No. 14, 2000 3105
tion is no longer found. Since it is unlikely that the CF3 group
of IF participates in agostic bonding as in B, this leaves S as
the remaining option. The 15-fold greater reactivity of IF with
CO would be consistent with this intermediate being less
Scheme 4
stabilized than ICo
.
Dynamics of Methyl Migration and CO Trapping Reac-
tions of ICo. The kinetics data described above clearly emphasize
a major difference between ICo and the manganese analogue
IMn. The reactivity of the former both toward CH3 migration
and toward trapping by CO is much higher, which may be one
reason the cobalt carbonyls are an effective industrial car-
bonylation catalyst. For example, the kM for ICo in 25 °C benzene
is 6.2 × 104 s-1 and that for kCO is 1.1 × 107 M-1 s-1 while
the respective values for IMn are 6.7 s-1 and 3.3 × 103 M-1
and for PFMC, and kM values (25 °C) in these solvents are
nearly identical, although the latter solvent is unlikely to form
a stable metal-solvent complex. Temperature-effect studies in
cyclohexane show ∆H q and ∆S q values of 44 kJ mol-1 and
-5 J K-1 mol-1 and 34 kJ mol-1 and -28 J K-1 mol-1 for kM
s-1
.
Although the absolute reactivities of the two intermediates
are considerably different, the patterns show some important
analogies. For most of the solvents the kCO values for ICo and
and kCO, respectively. The near-zero ∆S q value for IFe is
M
IMn do not vary significantly with the donor nature of the solvent
suggestive of a concerted CH3 migration pathway perhaps from
an agostic intermediate. Thus, an η2-acyl-stabilized species
analogous to C does not appear to play a role in this case.
When one considers the systems that have been subjected to
time-resolved spectroscopic techniques to probe the reactive
unsaturated intermediates in migratory insertion, the absence
of a single dominant pattern is frustrating. With regard to
structures, it appears that the model IMn and IFe complexes form
the solvento species in stronger donors such as THF but are
stabilized by the acetyl group in weakly donating solvents such
as alkanes. For IMn such stabilization appears to take the form
of η2-coordination of the acyl carbonyl and this pattern appears
to be paralleled by the catalytically important I′Co and ICo. Acyl
stabilization of IFe under such conditions may be different,
namely, agostic coordination of the CH3. The data are not
unambiguous regarding the relative energies of these acyl-
stabilized intermediates. Perhaps this should not be surprising
given the NMR studies by Carmona et al., which showed the
acyl complexes Mo(C(O)CH3)(S2CX)(CO)(PMe3)2 to be present
in a labile equilibrium between several such forms.54 Thus, one
is drawn to conclude that under certain conditions there may
be a fine balance among the C, S, and B type configurations of
these unsaturated intermediates and that the specific energetics
favoring one or another may reflect subtle shifts in experimental
conditions (Scheme 4).
(the exception being THF with the latter). It is this observation
that was viewed above as indicating an internally stabilized
structure for ICo (B or C). However, for both ICo and IMn, the
solvent appears to play a role in the methyl migration pathway.
For example, with IMn, kM is 50-fold higher in DCE than in
perfluoromethylcyclohexane (PFMC) and this could be inter-
preted either in terms of the transition state being more polar
than IMn (thus, the rate is accelerated by the more polar medium)
or in terms of a more specific solvent metal interaction. The
latter argument, but not the former, is consistent with the kM
being at least 10-fold larger in cyclohexane than in PFMC for
IMn. The pattern for ICo is similar; kM follows an order (THF >
DCE > CH2Cl2 > benzene) more or less consistent with the
polarity and/or donor ability of the solvent. If both IMn or ICo
are primarily present as η2-acyl complexes, the solvent effects
on kM may reflect the need for the acetyl group to rearrange to
a transition-state configuration for acyl-to-methyl migration that
is likely to resemble B. This might be assisted by direct
interaction of the solvent with the metal center (the interpretation
we have favored)22 or by stabilization owing to changes in
molecular polarity during this rearrangement.
The activation parameters reported here for CO trapping of
ICo in benzene indicate a very small activation enthapy (+6 kJ
mol-1) but a substantially negative activation entropy (-91 J
mol-1 K-1) for kCO. This suggests that addition of CO to this
intermediate has considerable associative character, perhaps a
concerted displacement of the acyl oxygen with formation of
the Co-CO bond. The “unimolecular” kM pathway displays a
higher ∆H q (+40 kJ mol-1) but a less negative ∆S q (-19
Thermal kinetics studies of RC(O)Co(CO)3L (R ) alkyl)
using in situ IR10,12 and NMR6,11 techniques have reported an
inverse dependence on [CO] for the rates of CO exchange,
reaction with H2, and isomerization of R. The unsaturated
intermediate RC(O)Co(CO)2L has been invoked in mechanism
discussions (e.g., eqs 3 and 4), but these techniques could not
M
M
J mol-1 K-1). The negative ∆S q may reflect the more organized
nature of the solvation required to effect rearrangement of C to
an acyl configuration more compatible with CH3 migration to
the metal.
k′-CO
CH3C(O)Co(CO)4 {
} CH3C(O)Co(CO)3 + CO (3)
k′-CO[CO]
Another model for migratory insertion that we and others
have studied in some detail is concerned with the mechanism
of CpFe(CO)(L)(C(O)CH3) formation (L ) CO or PR3) from
the methyl analogue CpFe(CO)2CH3 (MFe). The response of
the νCO values and of kL and kCO to solvent donor properties
first led to the conclusion the IFe has the solvento structure S
for the donor solvents THF and acetonitrile and possibly in the
chloroalkane and alkane solutions.26 Methyl migration is also
suppressed in THF and CH3CN, unlike the patterns for IMn and
k′M
CH3C(O)Co(CO)3 {
} CH3Co(CO)4
(4)
k′-M[CO]
probe the spectra or dynamics of this species directly. For
example, Roe used high-pressure NMR magnetization exchange
experiments11 to determine activation parameters for CO loss
from A′Co in methylcyclohexane-d14 as ∆H q ) 92 kJ mol-1
,
∆S q ) 33 J mol-1 K-1 and suggested that the transition state
for CO dissociation involved concerted coordination of the acyl
ICo where more polar solvents tend to increase kM. The situation
in alkane solutions is more ambiguous. The shift in the acyl
(54) Carmona, E.; Contreras, L.; Poveda, M. L.; Sanchez, L. J. J. Am. Chem.
Soc. 1991, 113, 4322-4324.
νCO from AFe to IFe is the same (-70 cm-1) for cyclohexane

3106 Inorganic Chemistry, Vol. 39, No. 14, 2000
Massick et al.
oxygen. Although I′Co was not detected in these experiments,
the data provide an estimate for the k′CO/k′M ratio of 21 for I′Co
at T ) 70 °C. Extrapolation of the data from laser flash
photolysis of ACo in benzene to the same conditions predicts a
similar branching kCO/kM ratio (31) for the PPh3-modified
reactive intermediate ICo. The similarity in these ratios of I′Co
and ICo supports the interpretation that the flash photolysis of
in benzene solution. The solvent dependence of the kM pathway
may simply reflect the necessity of solvent association to
facilitate the isomerization of C to B. The former has the CH3-
in a stereochemical location unfavorable for migration, while
the latter appears to be well on its way toward the formation of
M. Alternatively, M may be formed directly from S by a
concerted step kSM. In these contexts, the behavior of ICo
23
ACo is accessing the same reactive intermediate as that generated
parallels that of the manganese analogue IMn
.
However, the
thermally.
reaction rates of the former are orders of magnitude faster,
consistent with the role of phosphine-modified cobalt carbonyls
as carbonylation catalysts. Although in Scheme 4 arrows are
drawn unidirectionally from I to A and from I to M to denote
the decay of intermediates generated via a photoreaction, we
believe it likely that the thermal alkyl carbonylation pathways
relevant to the catalysis mechanisms are sampling analogous
intermediates such that microscopic reversibility may be as-
sumed.
Summary
Scheme 4 outlines key aspects of the above discussion
regarding the behavior of reactive intermediate(s) prepared by
photodissociation of CO from ACo. This can also serve as a
more generalized model for discussing the intermediates gener-
ated similarly from the iron and manganese complexes AFe and
AMn. Intermediate I (Scheme 2) can be considered to be an
equilibrium ensemble of the three species C, S, and B from
which methyl migration to form M or trapping with CO to form
A occurs. For the solvents probed in the present system, the
TRIR spectra are consistent with the conclusion that C is the
most prevalent configuration of ICo. That ICo is not predomi-
nantly S is substantiated by the insensitivity of kCO to solvent
donor properties. In contrast, it is clear that kM is solvent-
dependent in a manner that suggests a more intimate role of
solvent in the methyl migration pathway, and this observation
argues against ICo being B.
Thus, the formation of ACo from ICo appears to occur via the
concerted displacement of the oxygen of the η2-coordinated
carbonyl via an associative reaction of CO with C (kCCO), and
this view is substantiated by the large negative ∆S q of that step
Acknowledgment. This research was sponsored by a grant
(DE-FG03-85ER13317) to P.C.F. from the Division of Chemical
Sciences, Office of Basic Energy Sciences, U.S. Department
of Energy and by a Collaborative UC/Los Alamos Research
(CULAR) Initiative grant from Los Alamos National Labora-
tory. Preliminary studies of this system were aided by Dr. W.
T. Boese.
Supporting Information Available: Figure S-1 showing step-scan
FTIR spectra after flash photolysis of ACo and Figure S-2 Eyring plots
for kM and kCO pathways. This material is available free of charge via
the Internet at http://pubs.acs.org.
IC000378Y