An unusual solvent isotope effect in the reaction of W(CO)5(solv) (solv = cyclohexane or Cyclohexane-d12) with THF

Article abstract of DOI:10.1021/om990828y

Time- and temperature-resolved infrared absorption spectroscopy is used to probe the reaction of THF with W(CO)₅(solv) (solv = cyclohexane, cyclohexane-d₁₂) to form W(CO)₅(THF). In cyclohexane-do, δH? for the reaction is d

Full text of DOI:10.1021/om990828y

1682
Organometallics 2000, 19, 1682-1691
An Un u su a l Solven t Isotop e Effect in th e Rea ction of
W(CO)5(solv) (solv ) Cycloh exa n e or Cycloh exa n e-d 12)
w ith THF
Riki Paur-Afshari, J . Lin, and Richard H. Schultz*
Department of Chemistry, Bar-Ilan University, Ramat-Gan 52900, Israel
Received October 18, 1999
Time- and temperature-resolved infrared absorption spectroscopy is used to probe the
reaction of THF with W(CO)5(solv) (solv ) cyclohexane, cyclohexane-d12) to form W(CO)5-
(THF). In cyclohexane-d0, ∆Hq for the reaction is determined to be 3.6 ( 0.2 kcal mol-1 (15.2
( 0.9 kJ mol-1) and ∆Sq ) -13.7 ( 2.5 eu (-57.3 ( 10.5 J mol-1 K-1). In cyclohexane-d12,
∆Hq ) 2.4 ( 0.6 kcal mol-1 (10.2 ( 2.5 kJ mol-1) and ∆Sq ) -18.3 ( 3.5 eu (-76.6 ( 14.6
J mol-1 K-1). The low activation enthalpy in comparison to the (CO)5W-cyclohexane bond
dissociation energy, the negative entropy of activation, and comparison with other
spectroscopic experiments involving reactions of W(CO)5(cyclohexane) indicate that the
reaction proceeds by an associative interchange reaction mechanism. An unusual isotope
effect, in which reaction proceeds more slowly in cyclohexane-d12 even though ∆Hq is lower,
is explained in terms of relative populations of low-frequency vibrations and the relative
C-H and C-D bond strengths in the solvent molecule released upon uptake of THF. These
experimental results are supported by the results of an ab initio DFT calculation of the
structure and vibrational spectrum of W(CO)5(cyclohexane).
In tr od u ction
One aspect of the chemistry of such species that has
been of particular recent interest has been that of “σ-
bound” TM-alkane complexes.5 Such species are known
to be intermediates in the photolytic activation of alkane
C-H bonds by TM complexes,6,7 and recently, a low-
temperature NMR study provided direct evidence for
the existence of a σ-bound [CpRe(CO)2]-cyclopentane
complex.8 In this context, Bergman and Moore and co-
workers have performed extensive kinetic studies on the
alkane C-H activation reactions of “Cp*Rh(CO)” gener-
ated photolytically in liquid Kr or Xe solution in order
to try to elucidate the details of the C-H bond activation
process.9,10 One particularly intriguing result arising
from those studies was the observation that the equi-
librium between the solvated Cp*Rh(CO)(Kr) interme-
diate and the σ-bound Cp*Rh(CO)(alkane) complex
appears to favor the alkane complex more strongly upon
deuteriation of the alkane (i.e., shows an inverse equi-
librium isotope effect (EIE)).9 Because the complex goes
on to activate a C-H bond, direct measurement of this
effect was not possible. The question of possible isotope
The reactions and physical and chemical properties
of transition-metal (TM) intermediates have been of
considerable interest over the last two decades.1
A
particular area of interest has been trying to understand
the properties of coordinatively unsaturated TM inter-
mediates in which a coordination site normally occupied
by a strongly bound ligand is “vacant”; it is generally
assumed that on the time scale of bimolecular reactions
the “vacant” site is actually occupied by a weakly bound
solvent molecule. Because such species are so reactive,
detecting them and measuring their properties can only
be accomplished through sophisticated fast kinetic
techniques such as time-resolved infrared spectroscopy
(TRIR), time-resolved UV/vis spectroscopy (TR-UV/vis),
or photoacoustic calorimetry (PAC). These techniques
have succeeded in identifying and characterizing a
number of TM intermediates and transient electronic
states with lifetimes down to the picosecond time
scale.2-4
* To whom all correspondence should be sent. E-mail: schultr@
mail.biu.ac.il.
(1) Boese, W.; McFarlane, K.; Lee, B.; Rabor, J .; Ford, P. C. Coord.
Chem. Rev. 1997, 159, 135.
(5) Hall, C.; Perutz, R. N. Chem. Rev. 1996, 96, 3125.
(6) (a) J anowicz, A. H.; Bergman, R. G. J . Am. Chem. Soc. 1982,
104, 352. (b) Hoyano, J . K.; Graham, W. A. G. J . Am. Chem. Soc. 1982,
104, 3723. (c) Shilov, A. E.; Shul’pin, G. B. Chem. Rev. 1997, 97, 2879.
(7) (a) Buchanan, J . M.; Stryker, J . M.; Bergman, R. G. J . Am. Chem.
Soc. 1986, 108, 1537. (b) Periana, R. A.; Bergman, R. G. J . Am. Chem.
Soc. 1986, 108, 7332. (c) Bromberg, S. E.; Yang, H.; Asplund, M. C.;
Lian, T.; McNamara, B. K.; Kotz, K. T.; Yeston, J . S.; Wilkens, M.;
Frei, H.; Bergman, R. G.; Harris, C. B. Science 1997, 278, 260.
(8) Geftakis, S.; Ball, G. E. J . Am. Chem. Soc. 1998, 120, 9953.
(9) (a) Schultz, R. H.; Bengali, A. A.; Tauber, M. J .; Weiller, B. H.;
Wasserman, E. P.; Kyle, K. R.; Moore, C. B.; Bergman, R. G. J . Am.
Chem. Soc. 1994, 116, 7369. (b) Bengali, A. A.; Schultz, R. H.; Moore,
C. B.; Bergman, R. G. J . Am. Chem. Soc. 1994, 116, 9585.
(2) Kelly, J . M.; Long, C.; Bonneau, R. J . Phys. Chem. 1983, 87, 3344.
(3) (a) Yang, G. K.; Vaida, V.; Peters, K. S. Polyhedron 1988, 7, 1619.
(b) Morse, J . M., J r.; Parker, G. H.; Burkey, T. J . Organometallics 1989,
8, 2471. (c) Braslavsky, S. E.; Heibel, G. E. Chem. Rev. 1992, 92, 1381.
(d) Hester, D. M.; Sun, J .; Harper, A. W.; Yang, G. K. J . Am. Chem.
Soc. 1992, 114, 5234. (e) Leu, G.-L.; Burkey, T. J . J . Coord. Chem.
1995, 34, 87.
(4) (a) Welch, J . A.; Peters, K. S.; Vaida, V. J . Phys. Chem. 1982,
86, 1941. (b) Simon, J .; Peters, K. S. Chem. Phys. Lett. 1983, 98, 53.
(c) Wang, L.; Zhu, X.; Spears, K. G. J . Am. Chem. Soc. 1988, 110, 8695.
(d) Lee, M.; Harris, C. B. J . Am. Chem. Soc. 1989, 111, 8963. (e) Xie,
X.; Simon, J . D. J . Phys. Chem. 1989, 93, 4401. (f) J oly, A. G. Nelson,
K. A. Chem. Phys. 1991, 152, 69.
(10) (a) Weiller, B. H.; Wasserman, E. P.; Bergman, R. G.; Moore,
C. B. J . Am. Chem. Soc. 1989, 111, 8288. (b) McNamara, B. K.; Yeston,
J . S.; Bergman, R. B.; Moore, C. B. J . Am. Chem. Soc. 1999, 121, 6437.
10.1021/om990828y CCC: $19.00 © 2000 American Chemical Society
Publication on Web 03/31/2000

Reaction of W(CO)5(solv) with THF
Organometallics, Vol. 19, No. 9, 2000 1683
(Lambda Physik model Compex 102). The photolysis pulse
produces the transient W(CO)5(solv) complex. Typically, the
reaction solution flows through the cell (flow is maintained
by a Fluid Metering Systems PiP pipet pump) to ensure that
each laser pulse irradiates fresh solution. Reaction progress
is monitored by measuring the time-dependent absorption of
the IR light emanating from one of two sources described in
detail below. Reaction conditions were chosen to keep ∆A0 <
0.1 for the intermediate in order to ensure a linear relation
between the concentration of the intermediate and the ob-
served absorbance changes. As a preliminary test of the
instrument, we performed measurements of the reaction of
W(CO)5(CyH) with hex-1-ene and obtained rate constants that
were within experimental error of those of Dobson and
co-workers in their TR-UV/vis study of reaction 1.16
2. Step -Sca n F TIR (S2F TIR). For determination of time-
dependent IR spectra, we use an S2FTIR spectrometer13
(Bruker model Equinox 55 running OPUS 3.0 software) as the
IR source. The IR output of the spectrometer exits the
instrument, is collimated by a parabolic mirror, passes through
the cell, is collected by an elliptical mirror, and impinges on
an MCT detector (Kolmar model PV11-1-J 2, active area 1 mm2,
rise time 18 ns) with separate ac and dc preamplifier outputs.
For this experiment, signal digitization is done by a 250 MHz
PAD 82a fast A/D converter; output of the ac channel of the
detector preamplifier is used to monitor the time-dependent
signals, and phase correction is done by simultaneous monitor-
ing of the dc output of the detector preamplifier. To maximize
the resulting signal-to-noise ratio, preliminary experiments are
done to obtain peak dc and ac signals as close as possible to
(1 V, the maximum input to the PAD 82a board. Signal
maximization of the dc channel is performed by using an
oscilloscope to measure the interferogram obtained on a static
reaction solution. The source aperture is opened until the
detector signal is maximized without clipping; one of a series
of home-built high-frequency signal attenuators is then used
to bring the signal to approximately (0.9 V. The ac signal is
then measured under nominal photolysis conditions and
amplified (Stanford Research Systems model SR445 300 MHz
amplifier) to bring the peak ac signal as close as possible to 1
V.
S2FTIR spectra are normally obtained at 4 cm-1 resolution.
To minimize the number of data points needed for the
interferogram, as well as to protect the detector from stray
UV light, IR filters are placed in front of the detector and
between the IR source and cell. For these experiments, OCLI
model W04944-8 filters (10% cutoffs at approximately 2280
and 1810 cm-1) were used, and spectra were collected over the
range 1760-2340 cm-1. Typically, 5-10 laser shots are aver-
aged at each mirror position, and a stabilization delay of 150-
200 ms is used at each mirror position prior to photolysis.
3. Diod e La ser Sou r ce. For collection of kinetic data, a
diode laser source is used. In this apparatus, a liquid N2 cooled
F igu r e 1. Instrumental schematic.
effects in the σ-binding of small molecules (e.g., dihy-
drogen or alkanes) to transition-metal centers has been
of some theoretical interest as well.11,12
We wanted to try to see if we could observe a similar
effect in a system that does not activate the C-H bonds
in order to see if a significant isotope effect is a
fundamental property of TM-alkane interactions. To
this end, we chose to study the temperature and
reactant concentration dependences of a reaction of the
σ-bound complex of tungsten pentacarbonyl with cyclo-
hexane (CyH) or its perdeuteriated analogue (CyD). In
the present study, we obtain activation parameters for
the simplest possible reaction of such a species, ligand
exchange to form a stable complex, reaction 1.
W(CO)5(solv) + L f W(CO)5L
(1)
In these experiments, “solv” ) CyH or CyD and “L” )
THF.
The primary goal of this experiment was to determine
the nature of the isotope effect on the σ-binding of an
alkane to a transition-metal fragment by observing what
effect deuteriation of the alkane has on the course of
reaction 1. A second goal was to compare the activation
parameters for reaction 1 in CyH and CyD in order to
see what insight such measurements could give about
the dynamics and mechanism of this ligand substitution
reaction itself. Finally, an ab initio density-functional
theory (DFT) calculation of the structure of the W(CO)5-
(CyH) complex was performed in order to see what
additional insight into the dynamics of reaction 1 we
could obtain from a more detailed understanding of the
structure of the reactant.
(13) (a) Palmer, R. A.; Manning, C. J .; Rzepiela, J . A.; Widder, J .
M.; Chao, J . L. Appl. Spectrosc. 1989, 43, 193. (b) Uhmann, W.; Becker,
A.; Taran, C.; Siebert, F. Appl. Spectrosc. 1991, 45, 390. (c) Manning,
C. J .; Palmer, R. A.; Chao, J . L. Rev. Sci. Instrum. 1991, 62, 1219. (d)
Manning, C. J .; Griffiths, P. R. Appl. Spectrosc. 1993, 47, 1345. (e)
Palmer, R. A.; Chao, J . L.; Dittmar, R. M.; Gregoriou, V. G.; Plunkett,
S. E. Appl. Spectrosc. 1993, 47, 1297. (f) J ohnson, T. J .; Simon, A.;
Weil, J . M.; Harris, G. W. Appl. Spectrosc. 1993, 47, 1376. (g) Powell,
J . R.; Crocombe, R. A. Proc. SPIE-Int. Soc. Opt. Eng. 1993, 2089, 244.
(14) Hermann, H.; Grevels, F.-W.; Henne, A.; Schaffner, K. J . Phys.
Chem. 1982, 86, 5151.
(15) Although the code was written independently, the following
sources were consulted for development of the algorithms: (a) Natrella,
M. G. Experimental Statistics; National Bureau of Standards: Wash-
ington, D.C., 1963. (b) Bevington, P. R.; Robinson, D. K. Data Reduction
and Error Analysis for the Physical Sciences, 2nd ed.; McGraw-Hill:
New York, 1992. (c) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.;
Flannery, B. P. Numerical Recipes in C, 2nd ed.; reprinted with
corrections; Cambridge University Press: New York, 1994.
(16) Dobson, G. R.; Asali, K. J .; Cate, C. D.; Cate, C. W. Inorg. Chem.
1991, 30, 4471.
Exp er im en ta l Section
1. Gen er a l Com m en ts. The experiments described in this
paper were performed on our recently constructed transient
infrared (TRIR) spectrometer. An instrumental schematic is
shown in Figure 1. Reaction takes place in a temperature-
controllable ((1 °C in the experiments described here), 0.5 mm
path length CaF2 IR cell (International Crystal Laboratories
CryoTherm with model 2405-4761 temperature controller). A
reaction solution containing typically (2.5-5) × 10-4 mol L-1
W(CO)6 in CyH or CyD and a large excess of THF undergoes
photolysis by the 308 nm pulsed output (ca. 20 ns pulse width,
typically 50-80 mJ /pulse at the laser) of a XeCl excimer laser
(11) Bender, B. J . Am. Chem. Soc. 1995, 117, 11239.
(12) (a) Abu-Hasanayn, F.; Krogh-J esperson, K.; Goldman, A. S. J .
Am. Chem. Soc. 1993, 115, 8019. (b) Schaller, C. P.; Bonanno, J . B.;
Wolczanski, P. J . Am. Chem. Soc. 1994, 116, 4133.

1684 Organometallics, Vol. 19, No. 9, 2000
Paur-Afshari et al.
continuously tunable lead-salt diode laser (Muetek model DH4,
range 1830-2050 cm-1) is used as a source of CW IR light.
The output of the laser is collimated to a 4-5 mm diameter
beam, after which it passes through a monochromator (SPEX
model 270M, 150 grooves/mm grating blazed at 4 µm, cali-
brated from successive orders of a He-Ne laser and normally
run at ∼1 cm-1 resolution), then through the cell collinear with
the UV beam, and on to the detector. For the experiments
described here, an InSb detector (Cincinnati Electronics model
SDD-32E0-S1-05M, 2 mm2 active area, rise time 50 ns) was
used. The detector output is sent to a digital oscilloscope
(LeCroy model 9310A, 100 MS/s) for collection and averaging.
Typically, each kinetic trace is the average of the signal
following 10-100 laser shots. A background trace is then taken
and subtracted from the transient signal, and the combined
signal passed via a GPIB interface to a PC for storage.
4. Rea gen ts. For the experiments described here, W(CO)6
(99%) was obtained from Strem and used without further
purification. Cyclohexane was obtained in spectrophotometric
or HPLC grade (>99.8% purity) and distilled from Na/
benzophenone immediately prior to use. Cyclohexane-d12
(99.6% D) was obtained from Aldrich and used without further
purification; isotopic purity and freedom from contamination
by H2O were confirmed by 1H NMR spectroscopy. THF (99.9%,
anhydrous grade, inhibitor-free, used and stored under Ar) was
obtained from Aldrich and used without further purification.
After preparation, reaction solutions were degassed with Ar
and kept under an Ar atmosphere for the duration of the
experiment. No evidence for interference from residual water
in the solvent14 or in the THF was observed.
5. Da ta An a lysis. For each kinetic run, the pseudo-first-
order rate constant kobs is determined either by a weighted
linear least-squares fit to ln|(∆A∞ - ∆A0)| or by a nonlinear
least-squares single-exponential fit to ∆A itself; the two fitting
methods yield values for kobs for a single kinetic run that do
not differ significantly. Reported values for kobs normally
represent the average of at least two independent data sets.
For low concentrations, the reported values include determi-
nations based on both the rate of decay of the intermediate
and the rate of growth of the product. Because our diode laser
is significantly less powerful at the absorbance peak of the
product than it is at that of the intermediate, and because the
product peak absorption is less intense than the intermediate
peak absorption, we found that at higher rates the S/N ratio
for the product growth was not always sufficiently high to
produce reliable values for kobs; in these cases, the reported
rate was determined from the intermediate decay alone. In
all cases, the rate of growth of the product was consistent with
the rate of decay of the intermediate. Pseudo-first-order rate
constants are reported with 1σ uncertainties in the precision.
We estimate the absolute uncertainties in kobs to be ap-
proximately (10%, based on uncertainties in the concentration
and temperature and on long-term run-to-run reproducibility;
these absolute uncertainties are not included in the rate
constants reported in Tables S-1 and S-2.
F igu r e 2. S2FTIR difference spectra (room temperature)
for the reaction of W(CO)5(CyH) with 0.015 mol L-1 THF.
Shown are ∆A (absorbance change relative to absorbance
prior to photolysis) as a function of energy (1900-2000
cm-1) at the photolysis flash and at time intervals of 1, 5,
and 20 µs following photolysis. The two peaks at 1954 and
1928 cm-1 that decrease with time are attributed to
W(CO)5(CyH); the two peaks at 1933 and 1912 cm-1 that
increase with time are attributed to W(CO)5(THF); and the
bleach at 1981 cm-1 corresponds to loss of W(CO)6.
with two new peaks at 1954 and 1928 cm-1, which we
assign to the E and A1 C-O stretches of W(CO)5(CyH).14
These two peaks disappear with time, and simulta-
neously, two new peaks, at 1933 and 1912 cm-1
,
corresponding to the E and A1 C-O stretches of W(CO)5-
(THF), grow in with the same time dependence. Sym-
metry arguments require that there be a third C-O
stretch of A1 symmetry. This very weak peak (on the
order of 1-5% of the intensity of the E stretch), not
shown in Figure 2, occurs at 2087 cm-1 for W(CO)5(CyH)
and at 2074 cm-1 for W(CO)5(THF) in CyH solution.
The diode laser source was used for the actual kinetic
measurements. At each temperature and THF concen-
tration, kinetic traces were obtained by setting the diode
laser to a frequency corresponding to an absorption of
the W(CO)5(CyH) transient and measuring the time-
dependent signal at that frequency, and then making
the same measurement at a frequency corresponding
to an absorbance of the W(CO)5(THF) product. While
the S2FTIR experiment shows that the peak absorbance
of the W(CO)5(CyH) transient is at 1954 cm-1, due to
the output characteristics of the diode laser (and pos-
sibly to slight uncertainties in the mutual calibrations
of the two instruments), we found that the best signal-
to-noise ratio was obtained when we set the diode laser
to 1955 or 1956 cm-1. Kinetic measurements made at a
variety of frequencies near the product peak showed no
dependence of the reaction rate constant on the fre-
quency being probed (i.e., the absorbance line shape is
independent of time), however. Typical kinetic traces
are shown in Figure 3. Kinetic traces were obtained at
five temperatures from 20 to 60 °C and over a THF
concentration range of 0.004-0.24 mol L-1, and results
are shown in Figure 4. The observed pseudo-first-order
rate constant is linear with increasing [THF]; linear fits
to kobs as a function of [THF] are also shown in the
figure.
Second-order rate constants and activation parameters were
determined by linear least-squares fits to the concentration
dependence of kobs and are reported with 1σ uncertainties.
Reported uncertainties for the Eyring activation parameters
are 90% confidence limits to least-squares linear fits weighted
by the variances of the rate constants. The linear fits were
performed using “C” computer programs written in our labora-
tory;15 exponential fits were performed with commercially
available software.
Resu lts
W(CO)5(CyH) + THF . Time-dependent S2FTIR spec-
tra of the reaction of W(CO)5(CyH) with THF (1900-
2000 cm-1) are shown in Figure 2. At the flash, a bleach
corresponding to loss of parent W(CO)6 appears, along
W(CO)5(CyD) + THF . We also measured the kinetics
of reaction 1 with solv ) CyD over the concentration
range [THF] ) 0.005-0.1 mol L-1. Because of the

Reaction of W(CO)5(solv) with THF
Organometallics, Vol. 19, No. 9, 2000 1685
F igu r e 3. Typical diode laser transient traces. Shown are
time-dependent absorbances at 1956 cm-1, due to absorp-
tion by W(CO)5(CyH), and at 1933 cm-1 (×2), due to
absorption by W(CO)5(THF), for reaction of W(CO)5(CyH)
with 0.059 mol L-1 THF at T ) 20 °C. The solid lines are
single-exponential fits to the data with kobs ) 7.34 × 105
s-1
.
F igu r e 5. Experimental results for reaction of W(CO)6-
(CyD) with THF. The observed pseudo-first-order rate
constant kobs/106 is plotted as a function of [THF]. Part a
shows results for reaction at 20 °C (b), 30 °C (O), and 40
°C (9); part b shows results for reaction at 50 °C (b) and
60 °C (O). The solid lines are linear fits to kobs as a function
of [THF].
similar conclusion was reached by Dobson and co-
workers regarding 351 nm photolysis of W(CO)5(hex-1-
ene).16 Therefore, for the experiments in CyD, rather
than flowing the solution through the cell into a “dump”
flask, at each concentration of THF, a single solution
was made, and that solution circulated through the cell.
Furthermore, to minimize errors of measurement of
small quantities of THF, for the experiments in CyD, a
slightly different procedure was used for sample prepa-
ration. A stock solution of 0.1 M THF in CyD was
prepared, and aliquots of this solution were diluted to
produce the solutions used in the various kinetic experi-
ments. We also note that for reactions in CyD the
majority of the reported rate constants were obtained
from measurements of the rate of decay of the interme-
diate alone; each reported value of kobs is the average
of several independent kinetic runs.
F igu r e 4. Experimental results for reaction of W(CO)6-
(CyH) with THF. The observed pseudo-first-order rate
constant kobs/106 is plotted as a function of [THF]. Part a
shows results for reaction at 20 °C (b), 30 °C (O), and 40
°C (9); part b shows results for reaction at 50 °C (b) and
60 °C (O). The lines are linear fits to kobs as a function of
[THF].
Results for kobs as a function of [THF] are shown in
Figure 5. Once again, kobs is a linear function of [THF],
and the magnitudes of the pseudo-first-order rate
constants are similar to, although somewhat smaller
than, those observed in CyH.
expense of the solvent, we hoped not to need a continu-
ous flow of solution through the cell. To this end, we
performed a preliminary experiment with CyH as the
solvent in which successive kinetic runs were made with
the solution flowing through the cell and then with no
flow at all. We observed that the time dependence of
the changes in the IR absorptions was independent of
whether the solution was flowing through the cell and
of the number of laser shots previously fired upon a
particular sample. These observations imply that 308
nm photolysis of W(CO)5(THF) regenerates the same
intermediate produced by photolysis of W(CO)6; a
Discu ssion
Activa tion P a r a m eter s a n d th e Mech a n ism of
Rea ction 1. Second-order rate constants “ka” are de-
rived from the pseudo-first-order rate constants kobs
shown in Figures 4 and 5 at each temperature as the
slopes of the linear fits to kobs as a function of [THF]
and are summarized in Table 1.

1686 Organometallics, Vol. 19, No. 9, 2000
Paur-Afshari et al.
Ta ble 1. Secon d -Or d er Associa tive Ra te Con sta n ts
enthalpy for the reaction as a whole to be at least 10-
15 kcal mol-1, since the reaction of “naked” W(CO)5
should be nearly barrierless4 and the ∆Hq determined
as described above should thus be near the (CO)5W-
alkane bond dissociation energy (BDE), which generally
are found to be of that order.3,18 Thus, a much lower
activation enthalpy (<5 kcal mol-1) indicates the reac-
tion is probably not dissociative. Furthermore, the
strongly negative activation entropy indicates that the
transition state for the reaction is significantly more
ordered than the separated reactants, again evidence
for an associative type of mechanism in which there is
more bond forming than bond breaking at the transition
state. This conclusion is also consistent with our S2FTIR
kinetics study19 in which we discovered that the room-
temperature reaction rate constant of reaction 1 varies
directly with the ability of L to donate electrons as
measured by the CO spectrum of the product complex.
We explained this correlation in terms of the better
ability of a strongly electron-donating ligand to stabilize
an associative transition state by continuing to provide
electron density at the nominally 16-electron metal
center as the solvent molecule recedes.
To our knowledge, activation parameters for reaction
1 in CyH have only been reported three times previ-
ously. Dobson and Spradling20 performed a TR-UV/vis
study of reaction 1 with L ) 4-acetylpyridine (4-acpy)
and determined that ∆Hq ) 3.4 ( 0.2 kcal mol-1 and
∆Sq ) -13.0 ( 0.6 eu, and concluded from these results
that the reaction proceeds by an associative mechanism.
The close quantitative agreement between the activa-
tion parameters for THF and 4-acpy appears to be
fortuitous, however.
(ka ) for Rea ction 1 in CyH a n d CyD
temperature,
°C
10-7 ka(CyH),
10-7 ka(CyD),
L mol-1 s-1
L mol-1 s-1
kCyH/kCyD
20
30
40
50
60
1.21 ( 0.03
1.56 ( 0.09
1.98 ( 0.05
2.43 ( 0.11
2.85 ( 0.12
0.93 ( 0.06
1.12 ( 0.06
1.24 ( 0.06
1.49 ( 0.10
1.82 ( 0.14
1.30 ( 0.07
1.40 ( 0.08
1.59 ( 0.06
1.63 ( 0.08
1.56 ( 0.09
F igu r e 6. Eyring analyses for reaction 1 in CyH (b) and
CyD (9). The error bars are 1σ uncertainties, and the solid
lines are least-squares fits weighted by these uncertainties.
The linear fits shown in the figures have nonzero
intercepts with values on the order of 1% of the values
of the slopes. While the existence of these nonzero
intercepts may indicate a small amount of dissociative
(i.e., multistep reaction) behavior, we feel that a more
likely interpretation is that in practice even at [THF]
) 0 the rate of disappearance of the intermediate will
not be zero, due to reaction with unphotolyzed W(CO)6,
recombination with CO, or reaction with impurities such
as residual H2O. The W(CO)5(CyH) intermediate pro-
duced by photolysis of a solution of W(CO)6 in CyH with
no other reactant added has been observed to disappear
with a rate constant of on the order of 104 s-1 (the exact
reaction rate will obviously depend on the initial con-
centration of W(CO)6 and the energy of the photolysis
pulse).14 Because the intercepts are so small relative to
the slopes, they have not been further considered in our
analyses.
From the temperature dependences of the second-
order rate constants, we obtain Eyring activation pa-
rameters for reaction 1, Figure 6. For reaction of
W(CO)5(CyH), we obtain ∆Hq ) 3.6 ( 0.2 kcal mol-1
and ∆Sq ) -13.7 ( 3.5 eu; for reaction of W(CO)5(CyD),
we obtain ∆Hq ) 2.4 ( 0.6 kcal mol-1 and ∆Sq ) -18.3
( 3.5 eu.17
Evidence for the likely mechanism of reaction 1 can
be obtained from the activation parameters. If reaction
1 were dissociative, we would expect the activation
In a second TR-UV/vis study, Dobson et al.16 meas-
ured activation parameters for reaction 1 with L ) hex-
1-ene. The authors interpreted their results according
to a purely dissociative mechanism, obtaining activation
parameters for the solvent dissocation step of ∆H1q
)
8.2 kcal mol-1 and ∆S1q ) -3.7 eu and attributing the
difference between ∆Hq and the (CO)5W-CyH BDE to
1
residual solvation of the “naked” W(CO)5 intermediate.
Careful reexamination of their raw data reveals, how-
ever, that their results are equally consistent with an
associative mechanism for the reaction of W(CO)5(CyH)
with hex-1-ene, with ∆Hq ) 7.1 kcal mol-1 and ∆Sq )
-8.0 eu. Furthermore, Weiller21 has deduced a dissocia-
tive mechanism for reaction 1 in liquid Xe (L ) CO) from
his observation that ∆Hq for reaction 1 in that system
is the same to within experimental error as the gas-
phase (CO)5W-Xe bond dissociation energy,22 which is
perhaps further evidence that the reaction of W(CO)5-
(alkane) with hex-1-ene is associative after all.
A third study of reaction 1 in CyH was performed by
Breheny et al.,23 who investigated the recombination of
several solvated group 6 intermediates including W(CO)5-
(CyH) with CO. They reported that for this reaction ∆Hq
(17) These second-order rate constants “ka” assume an associative,
single-step mechanism. A reviewer has pointed out that our kinetic
results are also consistent with simultaneous associative and dissocia-
tive reaction. In fact, under our experimental conditions, it is possible
that even a purely dissociative reaction would be kinetically indistin-
guishable from a purely associative reaction (Lugovskoy, S.; Schultz,
R. H. Work in progress)! For a dissociative mechanism in which the
W-alkane bond is broken before the THF molecule arrives, the
(18) Burkey, T. J . Personal communication.
(19) Schultz, R. H.; Krav-Ami, S. J . Chem. Soc., Dalton Trans. 1999,
115.
(20) Dobson, G. R.; Spradling, M. D. Inorg. Chem. 1990, 29, 880.
(21) (a) Weiller, B. H. J . Am. Chem. Soc. 1992, 114, 10910. (b)
Weiller, B. H.; Wasserman, E. P.; Moore, C. B.; Bergman, R. G. J . Am.
Chem. Soc. 1993, 115, 4326.
(22) Wells, J . R.; Weitz, E. J . Am. Chem. Soc. 1992, 114, 2783.
(23) Breheny, C. J .; Kelly, J . M.; Long, C.; O’Keeffe, S.; Pryce, M.
T.; Russell, G.; Walsh, M. M. Organometallics 1998, 17, 3690.
reported ∆Hq will be ∆H1q - ∆H-q 1 + ∆Hq2, where ∆H1q and ∆H-q 1 refer
to the dissociation and formation of (CO)5W-alkane and ∆Hq refers
to the formation of the product. As discussed below, we feel2 that a
dissociative mechanism is extremely unlikely for this reaction, but it
cannot be ruled out on kinetic grounds alone.

Reaction of W(CO)5(solv) with THF
Organometallics, Vol. 19, No. 9, 2000 1687
Ta ble 2. Geom etr ica l P a r a m eter s for W(CO)5(CyH)
As Deter m in ed by DF T
atom paira
bond length, Å
atoms ABCa
ABC, deg
W-CR
2.998
2.116
2.993
2.065
1.978
3.231
3.152
1.135
1.101
1.105
1.551
Ccis-W-Ccis
Ctrans-W-Ccis
O-C-W
179.9
89.9
179.0
132.2
79.6
168.8
153.8
110.7
106.9
W-HR1
W-HR2
W-Ccis
W-Ctrans
Ccis-O
W-HR1-CR
W-HR2-CR
Ctrans-W-HR1
Ctrans-W-HR2
HR1-CR-HR2
H-C-H (av)
Ctrans-O
CR-HR1
CR-HR2
C-H (av)
C-C (av)
F igu r e 7. Equilibrium structure of W(CO)5(CyH) as
calculated by DFT.
a
Ccis ) carbonyl C cis to CyH; Ctrans ) carbonyl C trans to CyH;
CR ) CyH C nearest to W.
) 5.5 kcal mol-1 and ∆Sq ) -14 eu. While it might seem
surprising that the enthalpy of activation for reaction
with CO is higher than that for reaction with THF,
these results are in fact consistent with an associative
mechanism in which reaction of an electron-withdraw-
ing ligand will tend to have a higher ∆Hq than reaction
of an electron-donating one.
In the strictest sense, however, an “associative” ligand
exchange reaction (type A in the Langford-Gray nota-
tion24) requires detection of an intermediate of increased
coordination number. We have not detected such an
intermediate in our studies, and to our knowledge,
“W(CO)5(solv)(L)” has never been detected as a species
with an independent existence. In the context of the
experimental results presented here and in our previous
paper, it seems most reasonable to consider reaction 1
with L ) THF as occurring via an “associative inter-
change” (Ia) process.
Ab In itio Ca lcu la tion of W(CO)5(CyH). To our
knolwedge, except for measurements of IR C-O stretch-
ing frequencies and of UV/vis spectra, no W(CO)5-
(alkane) complex has been characterized experimen-
tally, and the only theoretical characterization of such
complexes has been an ab initio MP2 calculation of
W(CO)5 complexation with small (C1-C3) alkanes and
fluoroalkanes performed by Zaric´ and Hall.25 To better
understand the dynamics of reaction 1 and the observed
isotope effect in the present system, we therefore
performed an ab initio DFT calculation26 at the B31LYP/
CEP-31G* level with full optimization in order to
determine the equilibrium structure and vibrational
frequencies for W(CO)5(CyH) and W(CO)5(CyD). The
optimized geometry for the W(CO)5(CyH) complex is
shown pictorially in Figure 7, and some geometrical
parameters of interest are given in Table 2.
CyH molecule. Rather, the W(CO)5 fragment interacts
primarily with a single equatorial C-H bond of CyH in
a manner qualitatively similar to the optimized geom-
etries that Zaric´ and Hall25 found for the complexes with
the smaller alkanes. This C-H bond lengthens slightly,
from 1.10 to 1.14 Å (for smaller alkanes, Zaric´ and Hall
calculated that this bond lengthens to about 1.16 Å).
The CyH ring lies at an angle of approximately 45°
relative to adjacent cis carbonyls. As expected for such
a weakly bound complex, the geometry of the W(CO)5
fragment is essentially unchanged from its geometry in
W(CO)6,27 except that the W-C bond trans to the CyH
molecule contracts by approximately 0.1 Å. Bond lengths
and angles of the other bonds of CyH are not signifi-
cantly altered from their values in the free molecule.
To obtain more accurate estimates of the vibrational
spectra of the complexes, we also performed a calcula-
tion at the same level of theory on the CyH and CyD
molecules. We found that the calculated vibrational
frequencies of CyH and CyD required a scaling factor
of ∼0.94 in order to match the experimental values;28
the use of such a correction factor is typical for ab initio
calculations of vibrational structure.29 By using the
scaling factor determined from CyH and CyD for the
calculated vibrational frequencies of the W(CO)5(CyH)
and W(CO)5(CyD) complexes, we were able to reproduce
the experimentally observed C-O stretching frequencies
to within 1%. Representative vibrational frequencies of
interest are given in Table 3.
Of the 81 vibrational modes of W(CO)5(C6H12), 48 can
be attributed to the C6H12 molecule, 27 to the W(CO)5
fragment, and the remaining 6 to the (CO)5W-C6H12
stretch, the two (CO)5W-C6H12 bends, and the three
hindered rotations of the C6H12 molecule. In Table 3,
we list frequencies for those modes that contain the
largest components of the two (CO)5W-C6H12 bending
motions and of the three hindered rotations, but it
should be noted that many low-frequency modes nomi-
nally of the W(CO)5 skeleton alone also show a depend-
ence on the C6X12 isotopomer present in the complex.
Since motions of the CyH ring are necessary to com-
We find that in the complex the W atom does not sit
symmetrically relative to the nearest methylene of the
(24) Langford, C. H.; Gray, H. B. Ligand Substitution Processes; W.
A. Benjamin: New York, 1966.
(25) Zaric´, S.; Hall, M. B. J . Phys. Chem. A 1997, 101, 4646.
(26) Frisch, M. J .; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J . R.; Zakrzewski, V. G.; Montgomery, J . A.;
Stratmann, R. E.; Burant, J . C.; Dapprich, S.; Milam, J . M.; Daniels,
A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J .; Barone, V.;
Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford,
S.; Ochterski, J .; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma,
K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J . B.;
Cioslowski, J .; Ortiz, J . V.; Stefanov, B. B.; Liu, G.; Liashenko, A.;
Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J .; Keith,
T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.;
Challacombe, M.; Gill, P. M. W.; J ohnson, B. G.; Chen, W.; Wong, M.
W.; Andres, J . L.; Head-Gordon, M.; Replogle, E. S.; Pople, J . A.
Gaussian 98 (Revision A.7); Gaussian, Inc.: Pittsburgh, 1998.
(27) Arnesen, S. P.; Seip, H. M. Acta Chem. Scand. 1966, 20, 2711.
(28) (a) Shimanouchi, T. Tables of Molecular Vibrational Frequen-
cies, Consolidated Volume I; National Bureau of Standards: Wash-
ington, D.C., 1972. (b) Sverdlov, L. M.; Kovner, M. A.; Krainov, E. P.
Vibrational Spectra of Polyatomic Molecules; translated from the
Russian by the IPST staff; Wiley: New York, 1973. (c) Dorofeeva, O.
V.; Gurvich, L. V.; J orish, V. S. J . Phys. Chem. Ref. Data 1986, 15,
437.
(29) Hehre, W. J .; Radom, L.; Schleyer, P. v. R.; Pople, J . A. Ab Initio
Molecular Orbital Theory; Wiley: New York, 1986; Chapter 6.

1688 Organometallics, Vol. 19, No. 9, 2000
Paur-Afshari et al.
Ta ble 3. Selected Ca lcu la ted a Vibr a tion a l
F r equ en cies (cm -1) for W(CO)5(CyH) a n d
W(CO)5(CyD)
an associative reaction. Since ∆Hq is lower in CyD than
in CyH, yet the reaction rate constant is lower, ∆Sq must
be more negative in CyD than in CyH (∆∆Sq ) 4.6 (
4.3 eu). A plausible explanation for this difference relies
on the effect of low-frequency bending and hindered
rotational motions on the overall entropy of the complex.
Since these motions are of lower frequency in the CyD
complex than in the CyH complex (Table 2), they will
be more highly populated in the former. As the system
approaches the transition state, these degrees of free-
dom will be lost as the solvent molecule leaves, causing
∆Sq to be more negative when the reaction takes place
in CyD.32
vibration
W(CO)5(CyH)
W(CO)5(CyD)
71, 114, 127
W-CyH (CyD) 73, 116, 144
hindered
rotationb
W-CyH (CyD) 55, 84
58, 82
122
bendb
W-CyH (CyD) 124
stretch
C-O stretch
CR-HR stretch
C-H stretch
1933, 1949, 1977, 2061c 1936, 1949, 1977, 2061
2578 1893
2892, 2893, 2894, 2898, 2119, 2120, 2124, 2125,
2901, 2935, 2937, 2938, 2130, 2181, 2188, 2190,
Similar considerations have been invoked in other
studies of isotope effects in σ-binding of C-H(D) bonds
to transition metal centers. While Calvert and Shapley33
reported that in the compound HOs3(CO)10CHnD3-n the
agostic Os-H(D)-C interaction favors Os-H over Os-D
binding, due to zero-point energy (ZPE) effects that will
preferentially weaken a C-H bond relative to a C-D
bond, theoretical studies indicate that low-frequency
vibrations and hindered rotations of a σ-bound molecule
not involving the three-center bond itself can contribute
significantly to the overall energy of the complex,
leading to an inverse EIE in the binding of a small
molecule to a TM center.11,12 Similarly, Parkin and co-
workers34 have explained the inverse EIE observed in
H2 or D2 addition to M(PMe3)4X2 (M ) Mo or W, X )
halide) in terms of contributions of isotope-sensitive
vibrational frequencies other than that of the M-H(D)
stretch.
Two experimental studies of isotope effects in reac-
tions of metal-alkane complexes have been performed
previously. Zhang and Dobson35 observed that at room
temperature ligand substitution by hex-1-ene at Cr-
(CO)5(solv) is faster for solv ) octane-d0 than it is for
solv ) octane-d18. They reported a ratio kH/kD ) 1.4 in
the room-temperature reaction rate constants, similar
in magnitude to that observed in the present study with
L ) THF and solv ) CyH or CyD. They explained this
“normal” isotope effect as being due to more bond
breaking than bond forming in the transition state, since
an M-D bond is expected to have a lower ZPE than an
M-H bond. Our observation of ∆Sq < 0 for reaction in
CyD calls this interpretation into question, however.
As mentioned in the Introduction, Bergman and
Moore and co-workers investigated the alkane C-H
bond activation reactions of photolytically produced
Cp*Rh(CO)(Kr) with CyH and neopentane and their
deuteriated analogues in liquid Kr solution.9 Although
they could not measure the properties of the
Cp*Rh(CO)-alkane complex directly, their kinetic re-
2946, 2948, 2962
2197, 2200, 2207
a
b
Corrected values; see text. Contain some amount of W(CO)5
skeletal motion; see text. c Experimental values for the three IR-
active vibrations are 1928, 1954, and 2087 cm-1
.
pensate for any motion of the W(CO)5 in order to
maintain the center of mass, there will necessarily be
mixing of “W-C6H12 bending” into the low-frequency
carbonyl modes, and CyH “bending” and “hindered
rotational” motions are actually combinations of low-
frequency carbonyl bends with bending of the CyH ring
against the W(CO)5 skeleton.
We calculate that a C-H (C-D) stretching vibration
associated with the H(D) atom closest to the W atom
drops in frequency by some 10% relative to its frequency
in the free alkane. Except for the lowering of this
frequency and loss of degeneracy of the Eg and Eu
vibrational modes of the free alkane due to the lowering
of symmetry, the remaining vibrational frequencies
associated with motions of the CyH molecule itself do
not alter very much upon complexation. We also obtain
a (CO)5W-CyH bond dissociation enthalpy (after cor-
rection to 298 K) of ∆H ) 13.6 kcal/mol, which is in good
(although arguably fortutious30) agreement with the
experimental gas-phase value of 11.6 ( 3 kcal/mol.31 The
specific applications of the results of our calculation to
our experimental results will be discussed in the suc-
ceeding sections.
Solven t Isotop e Effects on ∆Sq in Rea ction 1. As
shown in Figures 4-6 and Table 1, at all temperatures
studied, reaction 1 shows a “normal” kinetic isotope
effect in which the second-order rate constant is larger
in CyH than in CyD. The relative rate of reaction in the
two solvents appears to be temperature-dependent,
however. Since the second-order rate constant is a more
quickly rising function of temperature in CyH than in
CyD, it appears that the enthalpy of activation must
be higher in CyH even though kobs itself is higher in
CyH at all temperatures measured. The Eyring analysis
presented in Figure 6 indicates that ∆Hq is indeed lower
in CyD and that the differences in second-order rate
constants are due primarily to entropic effects.
(32) An alternative explanation for the differences in activation
parameters for reaction in the two solvents has been suggested to us,
according to which the “bulkier” CyH molecule must move further away
from the metal center than CyD in order to give the THF molecule
access to the metal center. While we cannot absolutely rule out this
explanation, we feel that it is unlikely, as the bond lengths of CyH
and CyD are virtually identical, and the mean vibrational amplitudes
differ by an amount only on the order of 1% of the size of the molecule
(Ewbank, J . D.; Kirsch, G.; Schaefer, L. J . Mol. Struct. 1976, 31, 39).
(33) Calvert, R. B.; Shapley, J . R. J . Am. Chem. Soc. 1978, 100, 7726.
(34) (a) Rabinovitch, D.; Parkin, G. J . Am. Chem. Soc. 1993, 115,
353. (b) Hascall, T.; Rabinovitch, D.; Murphy, V. J .; Beachy, M. D.;
Friesner, R. A.; Parkin, G. J . Am. Chem. Soc. 1999, 121, 11402.
(35) Dobson, G. R.; Zhang, S. Unpublished results cited in: Zhang,
S.; Dobson, G. R.; Zang, V.; Bajaj, H. C.; van Eldik, R. Inorg. Chem.
1990, 29, 9, 3477.
As we saw above, ∆Sq of reaction 1 is negative,
indicating an overall loss of the number of degrees of
freedom as the reaction system approaches the transi-
tion state, consistent with the late transition state of
(30) Zaric´ and Hall25 reported that their calculated dissociation
energies for (CO)5W-alkane complexes did not converge monotonically
at increasingly high levels of theory, but rather oscillated around the
true value.
(31) Brown, C. E.; Ishikawa, Y.-I.; Hackett, P. A.; Rayner, D. M. J .
Am. Chem. Soc. 1990, 112, 2530.

Reaction of W(CO)5(solv) with THF
Organometallics, Vol. 19, No. 9, 2000 1689
sults strongly implied an inverse EIE for the binding
of an alkane to Cp*Rh(CO) and that, moreover, this EIE
is driven almost entirely by differences in ∆S (much as
is observed here for binding to W(CO)5). They attributed
this entropic effect to differences in the populations of
low-frequency vibrational and hindered rotational modes.
While Breheny et al.23 did not measure isotope effects,
they reached a similar conclusion from their observation
of a dependence of ∆Sq on the solvent size for recombi-
nation of a given TM fragment with CO in a series of
alkane solvents.
that is, reaction proceeds more rapidly in CyH in the
temperature range studied here.
Rea ction 1 a n d Sta tistica l Mech a n ica l Mod els
for Isotop e Effects. In the past few years, a number
of workers have considered isotope effects in reactions
of organometallic species in terms of equilibrium sta-
tistical mechanics.5,12,37 The equilibrium constant (Keq)
for reaction 2,
A + B* a A* + B
(2)
where asterisks indicate isotopically labeled species, is
given by eq 3,
Thus, we can see that the inverse isotope effect on
∆Sq observed in reaction 1 is not unreasonable. In the
present case (and presumably in that of Zhang and
Dobson35 as well), however, the inverse isotope in the
entropy of activation is insufficient to overcome the
difference in the enthalpic barrier, and the overall
kinetic isotope effect (KIE) is normal.
Solven t Isotop e Effects on ∆Hq in Rea ction 1.
Next, we consider the effect of solvent deuteriation on
the activation enthalpy of reaction 1. We observe that
∆Hq is significantly lower in CyD even though the
reaction proceeds more slowly. In this section, we
discuss qualitatively what effects might lead to a lower
activation enthalpy in CyD than in CyH. Inverse KIEs
have been seen previously36 in reductive elimination
reactions of X-M-H complexes in which a metal-
hydrogen bond is broken and a new bond with a higher
X-H stretching frequency is formed. In these reactions,
an inverse KIE is seen because the reaction has a late
transition state in which the C-H bond is already fully
formed or nearly so. In such a case, the ZPE change will
be less for C-D bond formation than for C-H bond
formation, so the activation enthalpy will be lower for
the deuteriated complex.
A* A* A*
tr
Q
Qrot vib
Q
QBtrQBrotQBvib
Keq
)
e-∆∆ZPE/RT (3)
A
A
A
B* B* B*
(
)(
)
QtrQrotQvib Qtr
Q Q
rot vib
where Qtr, Qrot, and Qvib are the translational, rotational,
and vibrational partitition functions of the various
species, and ∆∆ZPE is the overall change in ZPE, i.e.,
(ZPEB* + ZPEA) - (ZPEB + ZPEA*). The expression on
the right side of eq 3 is normally written as the product
of four terms: SYM, a “symmetry” term arising from
the rotational partition functions; MMI, a “mass and
moment of inertia” term from the translational and
rotational partition functions; EXC, an “excitation” term
arising from population of excited vibrational states
reflected in the vibrational partition functions; and ZPE,
the exponential that includes the zero-point energies.
This model has been used successfully to explain EIEs
in the binding of dihydrogen38 and ethylene11 to transi-
tion-metal complexes and isotope effects in oxidative
addition of H-H (D-D) and C-H (C-D) bonds12,34 of
the type shown as reaction 4:
While in the present experiments, the C-H(D) bond
of the alkane is not broken, it is reasonable to assume
that similar effects are influencing the course of reaction
1. For an Ia reaction 1, the W-solv σ-interaction will
be greatly diminished at the transition state relative
to that of the W(CO)5(solv) complex. J ust as with
reductive elimination, the “formation” (or, perhaps more
accurately, reconstruction) of a stronger bond engenders
a smaller ZPE change and therefore a lower enthalpic
barrier to reaction in CyD. Indeed, when we use the
vibrational frequencies obtained from our ab initio
calculation to calculate the relative ZPE change upon
dissociation of the W(CO)5(alkane) complex into W(CO)5
+ CyH or CyD, we find that the overall ZPE change is
approximately 20 cm-1 less for dissociation of the CyD
complex than for the CyH complex.
It thus appears that the “normal” KIE observed in
reaction 1 is in fact due to a combination of two “inverse”
isotope effects. The additional vibrational density of
states available to the W(CO)5(CyD) complex means
that reaction 1 has a more negative ∆Sq in W(CO)5(CyD)
than in W(CO)5(CyH). On the other hand, the late
transition state typical of an associative reaction leads
to a lower ∆Hq for loss of CyD. In the case of reaction 1,
the entropic effect is more significant, so ∆Gq is higher
for reaction in CyD, and we see a “normal” isotope effect;
D2
9
[M]-H2 7H [M]
8 [M]-D2
(4)
2
To gain a better understanding of the dynamics of
reaction 1 and to support the conclusions described
above, we used the results of the ab initio calculation
of the structure of W(CO)5(CyH) in order to model the
EIE for the exchange of CyD and CyH at the W(CO)5
center, reaction 5; note that for this reaction the SYM
terms cancel.
(CO)5W-CyD + CyH a (CO)5W-CyH + CyD (5)
For reaction 5, the two terms in eq 3 that are related to
the vibrational partitition function, EXC and ZPE, both
favor formation of (CO)5W-CyD (i.e., predict an inverse
EIE): we find that at 298 K, EXC ) 0.924 and ZPE )
0.918.39 That is, as discussed above, the additional low-
(37) (a) Bigeleisen, J .; Goeppert-Mayer, M. J . Chem. Phys. 1947,
15, 261. (b) McLennan, D. J . Isotopes in Organic Chemistry; Buncel,
E., Lee, E., Eds.; Elsevier: New York, 1987; Vol. 7, Chapter 6, pp 393-
480.
(38) Bender, B. R.; Kubas, G. J .; J ones, L. H.; Swanson, B. I.; Eckert,
J .; Capps, K. B.; Hoff, C. D. J . Am. Chem. Soc. 1997, 119, 9179.
(39) While only “isotope-sensitive modes” will contribute to the
overall isotope effect, as discussed above, many nominally insensitive
modes do have some isotope dependence. Therefore, we calculated the
complete partition functions using the calculated geometries and
corrected vibrational frequencies for all four species. Values for
fundamental constants were taken from: Cohen, E. R.; Taylor, B. N.
Phys. Today 1999, 52, BG5.
(36) (a) J ones, W. D.; Feher, F. J . Acc. Chem. Res. 1989, 22, 91. (b)
Rosini, G. P.; Soubra, S.; Vixamar, M.; Wang, S.; Goldman, A. S. J .
Organomet. Chem. 1998, 554, 41.

1690 Organometallics, Vol. 19, No. 9, 2000
Paur-Afshari et al.
frequency vibrational modes available to the CyD
complex cause its formation to be slightly favored.40
For the MMI term, we calculate MMI ) 1.541, i.e.,
tending toward a normal isotope effect; a similar
tendency was reported by Bender11 for complexation of
C2H4 and C2D4 to Os2(CO)8. The normal isotope effect
for the MMI term arises because the ratios of the masses
and of IAIBIC (product of the moments of inertia) are
larger for the CyD/CyH pair than for the W(CO)5(CyD)/
W(CO)5(CyH) pair. Thus, the additional translational
and rotational degrees of freedom available to the
W(CO)5(CyD) complex relative to W(CO)5(CyH) do not
make up for the degrees of freedom lost by the CyD
molecule upon complexation. It is not clear, however, if
the MMI term calculated from the partition functions
accurately reflects the true situation for reaction 5 in
solution. The MMI term reported above is calculated
from the complete rotational and translational partition
functions and thus assumes unhindered rotation and
translation for all of the species involved in the equi-
librium. While such assumptions are probably good for
H and D2,41 the rotation of CyH in liquid cyclohexane
itself is somewhat hindered;42 clearly, the assumption
of free rotation and translation for the complex is almost
certainly not justified. We also note that use of the
Teller-Redlich product rule,11,34,38,43 which relates the
translational and rotational partition functions to the
molecular vibrations, yields MMI ) 1.199, in reasonable
agreement with the MMI calculated directly from Qtr
F igu r e 8. Free energy diagram for the reaction of W(CO)5-
(CyH) and W(CO)5(CyD) with THF. The ordinate is ∆G in
kcal/mol at 298 K for reaction 1 for the two complexes.
∆∆G° for the starting complexes was derived from the DFT
calculation as described in the text, and ∆Gq from the
experimental results.
and Qrot
.
The three factors (EXC × ZPE × MMI) calculated
from the partition functions combine to predict a normal
overall EIE for reaction 5 of 1.30. If we use the MMI
predicted from the Teller-Redlich product rule, we
obtain an EIE of 1.01. In the limit that MMI ) 1 (i.e.,
rotational and translational densities of states are
identical for the two sides of reaction 5), we obtain an
inverse overall EIE of 0.846. Thus, while we cannot say
absolutely which side of reaction 5 is favored, it is clear
that the vibrational degrees of freedom favor formation
of W(CO)5(CyD) and that it is the loss of these degrees
of freedom at the transition state that causes ∆Sq to be
more negative for reaction of W(CO)5(CyD) than for
W(CO)5(CyH). In any event, ∆G for reaction 5 is <0.2
kcal/mol, indicating that ∆∆G° for the two complexes
is essentially the inverse of the ∆∆G° for CyH and CyD,
∼1.7 kcal/mol;44 the uncertainties in the free energies
of formation of C6H12 and C6D12 and of ∆Gq for reaction
1 are larger than the ∆G of reaction 5. The calculation
of the EIE for reaction 5 enables us to construct the free
energy diagram for reaction 1 shown in Figure 8, which
makes it clear that even though ∆Hq is lower for
W(CO)5(CyD) than it is for W(CO)5(CyH), the overall
KIE is normal; that is, ∆∆G at the transition state is
smaller than it is in the ground state.
Con clu sion s
In this study, we use TRIR spectroscopy to measure
the kinetics and thermodynamics of the ligand reaction
of W(CO)5(solv) with THF for solv ) cyclohexane (CyH)
and cyclohexane-d12 (CyD). We have determined the
activation parameters for this reaction to be ∆Hq ) 3.6
( 0.2 kcal mol-1, ∆Sq ) -13.7 ( 2.5 eu for solv ) CyH,
and ∆Hq ) 2.4 ( 0.6 kcal mol-1, ∆Sq ) -18.3 ( 3.5 eu
for solv ) CyD. The low enthalpies of activation and
negative entropies of activation indicate that the reac-
tion proceeds through an associative interchange (Ia)
mechanism. In addition, the observed isotope effects are
consistent with an Ia mechanism as well. “Freezing out”
of low-frequency vibrational modes at the transition
state leads to a more negative ∆Sq, while the breakdown
of the M-(C-H) σ-interaction at the transition state
leads to a lower ∆Hq for reaction in CyD. Over the
temperature range studied here, the overall ∆Gq is
higher in CyD, leading to a “normal” kinetic isotope
effect. Finally, an ab initio calculation of the structures
and vibrational spectra of W(CO)5(C6H12) and W(CO)5-
(C6D12) reveals that the tungsten atom interacts with
the alkane primarily through a single C-H bond and
(40) Bergman and co-workers9 reported that for Rh(CO)2 at 165 K
formation of the CyD complex is favored over formation of the CyH
complex by a factor of approximately 10. Our calculations indicate that
for W(CO)5 the equilibrium does move more toward the CyD complex,
but by a considerably more modest amount (about 5% difference
between 298 and 165 K).
(41) (a) Taylor, D. G., III; Strauss, H. L. J . Chem. Phys. 1989, 90,
768. (b) Hunter, J . E., III; Taylor, D. G. III.; Strauss, H. L. J . Chem.
Phys. 1992, 97, 50.
(42) (a) Farman, H.; Dore, J . C.; Bellisent-Funel, M.-C. Mol. Phys.
1987, 61, 583. (b) Farman, H.; O’Mard, L.; Dore, J . C.; Bellisent-Funel,
M.-C. Mol. Phys. 1991, 73, 855.
(43) McQuarrie, D. A. Statistical Mechanics; Harper & Row: New
York, 1976; p 148.
(44) Calculated from ∆H° and ∆S° data cited in: Afeefy, H. Y.;
Liebman, J . R.; Stein, S. E. Neutral Thermochemical Data. In NIST
Chemistry Webbook, NIST Standard Reference Database Number 69;
Mallard, W. G., Lindstrom, P. J ., Eds.; National Institutes of Standards
and Technology: Gaithersburg, MD, 1998 (http://webbook.nist.gov).

Reaction of W(CO)5(solv) with THF
Organometallics, Vol. 19, No. 9, 2000 1691
that the proposed explanation for the observed kinetic
isotope effect in the reaction of the complex with THF
is consistent with the structures and vibrational spectra
obtained from the ab initio calculation.
Aped of the Bar-Ilan Department of Chemistry. The
technical assistance of the members of the Bar-Ilan
University Electronics Shop is also gratefully acknowl-
edged. Finally, R.H.S. would like to acknowledge Dr.
H. Schultz for bringing ref 15a to his attention and for
useful discussions regarding the statistical analyses.
Ack n ow led gm en t. This research was supported in
part by a grant from the Israel Foundation, adminis-
tered by the Israel Academy of Arts and Sciences; by
the U.S.-Israel Binational Science Foundation; and by
the Bar-Ilan University Research Authority. In addition,
partial funding for the diode laser spectrometer was
provided by the Rashi Foundation as part of a Guastalla
Fellowship awarded to R.H.S. from 1994 to 1997. The
ab initio calculations were performed by Dr. Pinchas
Su p p or tin g In for m a tion Ava ila ble: Tables of experi-
mentally obtained pseudo-first-order reaction rate constants,
calculated atomic coordinates for W(CO)5(CyH), and calculated
vibrational frequencies for W(CO)5(CyH) and W(CO)5(CyD).
This material is available free of charge via the Internet at
http://pubs.acs.org.
OM990828Y