Time-resolved infrared dynamics of C-F bond activation by a tungsten metal-carbonyl

Article abstract of DOI:10.1021/jp055465w

Chemical reactions that break, or activate, C-H and C-F are of tremendous synthetic interest. The intramolecular C-F bond activation of a tungsten carbonyl system has been studied by time-resolved infrared spectroscopy. The formation of solvent complexes and the final product are monitored using time-resolved infrared spectroscopy of the CO stretches. The rate of the reaction is shown to be limited by the formation of an intermolecular complex between the tungsten metal center and a solvent molecule. Comparison with DFT calculations shows that in the absence of solvent molecules the intramolecular complex with the tethered perfluorobenzene ring is energetically favorable, but is not the primary kinetic product because of the initial geometry of the complex.

Full text of DOI:10.1021/jp055465w

20
2006, 110, 20-24
Published on Web 12/15/2005
Time-Resolved Infrared Dynamics of C-F Bond Activation by a Tungsten Metal-Carbonyl
Matthew C. Asplund,* Aaron M. Johnson, and Jared A. Jakeman
Department of Chemistry and Biochemistry, Brigham Young UniVersity, ProVo, Utah 84602
ReceiVed: September 26, 2005; In Final Form: NoVember 23, 2005
Chemical reactions that break, or activate, C-H and C-F are of tremendous synthetic interest. The
intramolecular C-F bond activation of a tungsten carbonyl system has been studied by time-resolved infrared
spectroscopy. The formation of solvent complexes and the final product are monitored using time-resolved
infrared spectroscopy of the CO stretches. The rate of the reaction is shown to be limited by the formation
of an intermolecular complex between the tungsten metal center and a solvent molecule. Comparison with
DFT calculations shows that in the absence of solvent molecules the intramolecular complex with the tethered
perfluorobenzene ring is energetically favorable, but is not the primary kinetic product because of the initial
geometry of the complex.
Introduction
photon causes dissociation of the CO, and begins the C-F bond
activation process.
C-F bonds are among the strongest in all of chemistry, and
so chemical reactions that can break these bonds are of particular
interest. Since the first demonstration of activating C-H bonds
using organometallic molecules1,2 there has been intense interest
in understanding the reaction mechanisms involved. Studies of
C-H, Si-H, and C-halide activation reactions have been made
with nanosecond and microsecond time resolution in liquefied
noble gases3-5 and low-temperature matrices6-8 and with
femtosecond time resolution6-15 in room-temperature liquids.
Infrared spectroscopy of these reactions has focused on the CO
stretching modes of the metal carbonyls, which have large
absorption cross-sections and are very sensitive to the oxidation
state of the metal atom.
The first organometallic system for activating C-F bonds at
room temperature was reported by Richmond and co-work-
ers.16,17 This original C-F bond activation system was based
on a W(CO)3(PrCN)3 which is coupled to a dinitrogen ligand 1
which has a tethered perfluorobenzene ring. A reaction mech-
anism was proposed18 in which the first step was to couple the
ligand onto the metal center, displacing two of the nitrile ligands.
This is followed by dissociation of the third nitrile ligand and
addition of the tethered perfluorobenzene ring across the
tungsten metal center to form the final product 2. Of particular
interest, C-F activation occurs only with the tethered ring, and
not with perfluorobenzene in the solution. This reaction proceeds
by thermal dissociation of the nitrile ligand, and so cannot be
studied by pump-probe time-resolved spectroscopy to get
information on a sub-millisececond time scale. By coupling the
dinitrogen ligand on a W(CO)4 fragment, we can create a model
compound of a key intermediate of the reaction. For this work
a modified version of the dinitrogen ligand was used. This
structure of the ligand is shown in molecule 3. The CO ligand
is more strongly bound than the PrCN, and the reaction does
not proceed thermally at room temperature. Absorption of a UV
Experimental Section
Transient infrared measurements were made using a Bruker
IFS-66 FTIR spectrometer, with the necessary modifications for
step-scan and rapid-scan experiments. Step-scan measurements
used a fast MCT detector with a 10 ns rise time, and signals
were digitized with a 12-bit, 100 MS/s digitizer. Laser pulses
for UV excitation were generated as the frequency tripled (355
nm) or quadrupled (266 nm) output of a Nd:YAG laser
(Coherent Infinity). All excitations on compound 3 were done
at 355 nm. Excitations on compound 6 were done at 266 nm
because its absorbance is shifted to shorter wavelengths, and
so it did not show any reaction with 355 nm excitation. The
pump and probe beams were 8 mm in diameter and were
completely overlapped in the sample, with an angle of ap-
proximately 20 degrees between the two beams. Excitation
energy was kept at 3-5 mJ per pulse to minimize sample
degradation during the experiment.
The dinitrogen ligand in compound 3 was synthesized
according to a previously published procedure.19 The ligand was
coupled to W(CO)6 under an inert atmosphere to form 3. Purity
was confirmed by NMR and IR spectroscopy. Sample concen-
trations were 5 mM in the metal carbonyl and were made with
dry, oxygen-free solvents and kept under a positive pressure of
argon during experiments to minimize sample degradation.
Ab initio calculations were performed using the NWChem
computational chemistry package.20 DFT calculations were
performed using the B3LYP hybrid functional, using the
LANL2DZ effective core potential basis set on W atoms and
6-31G basis sets on all other atoms. Geometry calculations were
minimized using the standard tight convergence criteria, and
correct minimization was tracked by vibrational analysis to
eliminate negative frequencies.
* Corresponding author. E-mail: Asplund@chem.byu.edu.
10.1021/jp055465w CCC: $33.50 © 2006 American Chemical Society

Letters
J. Phys. Chem. B, Vol. 110, No. 1, 2006 21
SCHEME 1
Results and Discussion
The C-F activation reaction is initiated by irradiation of the
tungsten carbonyl 3 with 355 nm light. This causes the
dissociation of a CO ligand and formation of the C-F bond-
activated product 4. The progress of the reaction can be
measured in the infrared spectrum by looking at the metal-
CO stretches. Previous femtosecond infrared studies of the
photodynamics of metal carbonyls21 provide information on the
early dynamics of similar reactions. In solution, W(CO)6 absorbs
a UV photon, leading to rapid dissociation of one of the CO
ligands. This coordinatively unsaturated W(CO)5 species forms
a complex with a solvent molecule in the first few hundred
femtoseconds following dissociation, and then undergoes vi-
brational cooling, which takes ∼10 ps. Complexes of metal
carbonyls with THF are known to be much more stable than
complexes of alkanes or haloalkanes, with THF ligands displac-
ing the weaker ligands on a microsecond time scale.22 When
kept in a dry, oxygen-free environment, we have measured the
stability of W(CO)5THF to be several hours. It is expected that
the short-time dynamics of this reaction will be similar, with
the initial dissociation of a CO ligand, followed by formation
of a complex to stabilize the metal center.
Figure 1. Difference spectra of the photochemical reaction of
compound 3 following photolysis. In CH2Cl2 (dashed line) the reaction
produces a C-F activated product, while in THF (solid) the reaction
stops at the tungsten-THF solvent complex.
Figure 1 shows a static FTIR difference spectra after
photolysis by a single 3 mJ UV laser pulse in both THF (solid
line) and methylene chloride (dashed line) solvents. FTIR spectra
of the original C-F activation product 2 by Richmond et al.16

22 J. Phys. Chem. B, Vol. 110, No. 1, 2006
Letters
TABLE 1: Results from DFT Calculations for Chemical Species Described in This Paper, Including Absolute Energies, and CO
Vibrational Frequencies
molecule
calculated CO frequencies (cm-1
)
calculated energy (Hartree)
energy difference (kcal/mol)a
parent molecule 3
intramolecular complex
THF solvent complex
CH2Cl2 solvent complex-DMC
CH3CN solvent complex
C-F activated product 4
C-F activated product 216
1974, 1875, 1862, 1852
1923, 1839, 1829
1903, 1819, 1801
1903, 1796, 1790
1903, 1812, 1811
1991, 1937, 1870
2004, 1952, 1879
-1059.252108
-1441.91786
-1671.880473
-2401.569626
-1574.674603
-1441.9819
0
-7.089831173
-3.082489817
-26.59552806
-40.19
a Column 4 gives the difference in energy between the intermolecular solvent complex and the energy of the intramolecular complex plus the
energy of the solvent ligand.
show CO stretches at 2020, 1937, and 1906 cm-1, with the 1937
peak being the strongest. DFT calculations on this compound
give predicted frequencies of 2004, 1952, and 1879 cm-1
.
Calculations performed using a B3LYP DFT combined func-
tional usually show frequencies that are 3-4% too high
compared to experiment, but this does not appear to be the case
in this system. The highest calculated frequency (2004 cm-1
)
is actually lower than the measured frequency (2020 cm-1). For
compound 3, both of the difference spectra show negative
absorbances at 2011, 1878, and 1831 cm-1 resulting from
bleaching of the parent molecule. When THF is used as a
solvent, we observe new absorbances at 1876 and 1840 cm-1
,
which represent a THF solvent complex. On the time scale of
the difference spectra, we observe no formation of C-F
activated product in THF, suggesting that either the intramo-
lecular complex necessary for activation never forms, or that it
forms, but the perfluorobenzene ring is displaced by the THF
solvent ligand. This species is stable for several hours until it
reacts thermally to form the C-F bond activated product. When
CH2Cl2 is used as a solvent, we observe similar bleach peaks,
but we observe a new absorbance peak at 1946 cm-1, close to
the previously measured 1937 cm-1, and in reasonably good
agreement with DFT calculations of the product CO stretches
at 1991, 1937, and 1879 cm-1 shown in Table 1. This shows
that we have formed the C-F activated product in methylene
chloride solution.
Using nanosecond step-scan FTIR spectroscopy, we have
measured the spectra of the intermediates formed during the
first 200 µs following excitation in three different solvents. In
each case, the spectrum does not evolve on the time scale from
100 ns to 200 µs, so only averaged spectra are shown. Figure
2 shows the transient spectrum in three different solvents: THF,
CH2Cl2, and in CH3CN. In the THF solvent, the nanosecond
transient spectrum matches the static difference spectrum in
Figure 1, showing that a complex is formed within the first 100
ns, and that it does not change over the time of the experiment.
The transient spectrum in acetonitrile shows a similar spectral
signature, with peaks at 1878 and 1831 cm-1. In methylene
chloride, the intermediate spectrum shows peaks at 1866 and
1821 cm-1. The transient spectra show no formation of C-F
bond activated product at 1947 cm-1 during the first 200 µs of
the reaction. The transient spectrum in methylene chloride shows
what may be a small peak at 1947 cm-1, but it is small enough
that it cannot be shown to be statistically significant. Calculated
CO frequencies for the intramolecular and intermolecular
complexes are shown in Table 1. For the intermolecular
complex, they each have a CO stretch at 1903 cm-1 that
corresponds to the symmetric stretch of the three CO ligands.
The next CO stretch varies in frequency for each of the three
solvents: 1842 in THF, 1830 in acetonitrile, and 1820 in
methylene chloride. The relative frequencies agree well with
the calculated frequencies of 1819 with THF, 1812 with
acetonitrile, and 1796 with methylene chloride. The fact that
Figure 2. Nanosecond time-resolved spectra of the reaction intermedi-
ates in THF, methylene chloride, and acetnitrile. All intermediates show
a peak at ∼1903 cm-1, corresponding to the symmetric CO stretch of
a solvent complex, and peaks at 1819 (THF), 1812 (acetonitrile), and
1796 (methylene chloride), which correspond to the higher frequency
asymmetric CO stretch.
the frequency of this CO stretch is solvent dependent suggests
that this is a solvent complex, and not an intermolecular
complex, which would be expected to have a similar spectrum
in the different solvents. The calculated frequencies of the
intramolecular complex are about 20 cm-1 higher in frequency
than the solvent complexes, and there is no evidence of peaks
in that region of the transient spectrum. Studies of a similar
chemical system,12 which undergoes activation of a pendant
ligand, shows competition between intermolecular (with a
solvent molecule) and intramolecular (with the pendant ligand)
complex formation. In that system approximately 50% of the
molecules formed an intramolecular complex, and the other 50%
forming an intermolecular complex with a solvent molecule.
On a time scale of 34 ns, the solvent ligand is displaced by the
pendant ligand to form the intramolecular complex. In com-
parison, we only see formation of the intermolecular complex
in this system.
The spectral shifts provide strong evidence that the short-
time reaction intermediate in the C-F bond activation reaction
is a solvent complex and not an intramolecular complex with
the perfluorobenzene. This is surprising since the proximity of
the pendant ligand would seem to favor formation of the
intramolecular complex. DFT calculations of the structure of
the compound 3 are shown in Figure 3. They show that the
pendant ligand is oriented away from the metal atom, allowing
solvent molecules easy access to the unsaturated metal center,
but making formation of the intramolecular complex slow. If
the kinetic product is the intermolecular complex, then in order

Letters
J. Phys. Chem. B, Vol. 110, No. 1, 2006 23
to either the lack of formation of the intramolecular complexed
intermediate, or to the raising of the energy of the final product,
making it thermodynamically unstable. DFT calculations on the
intramolecular complex show that changing this one double
bond to a single bond changes the energy of the complex
significantly. For the double bonded complex, the complex is
40 kcal/mol lower in energy than the naked tricarbonyl (i.e.,
the parent molecule with one CO ligand removed), but for the
single bonded complex, the complex is 23 kcal/mol higher in
energy. Thus there is no thermodynamic driving force for the
pendant ligand to displace the solvent and form the intramo-
lecular complex, so no reaction occurs.
Over the first 200 µs of the reaction, there is no decrease in
the absorbance of the intermediate or any absorbance due to
formation of product. Our current step scan electronics can
measure only up to 200 µs. Rapid-scan FTIR allows for
measurements of the spectrum on a millisecond time scale.
Measurements of the reaction on the millisecond time scale show
that the product is fully formed by 14 ms, the shortest time
measurable with this technique. This is true even when the
reaction mixture is cooled to 0 °C. While we have not been
able to directly observe the C-F activation reaction, we can
use this information to estimate an interval for the value of this
time constant. If we assume that the reaction is at least three
times longer than the 200 µs longest time from the step-scan
measurement, and three times shorter than the 14 ms time from
the rapid scan measurement, then the reaction must be occurring
on a time scale of 0.6-5.0 milliseconds. Further, if we assume
that the reaction follows standard Arrhenius behavior, then the
reaction rate should be roughly 5 times faster at room temper-
ature than at 0 °C. Thus we can assume that the reaction rate is
between 0.6 and 1.0 ms.
The reactivity of compound 3 can be compared to a model
compound 6, W(CO)4(tmeda), where tmeda is (CH3)2NCH2-
CH2N(CH3)2. This has the same basic structure, but because it
has no perfluorobenzene side chain it cannot form an intramo-
lecular complex. After photolysis, the initial product has a CO
vibration at 1970 cm-1. On a time scale of 30 s, this frequency
shifts to 1981 cm-1. This 11 cm-1 shift is similar to the 14
cm-1 shift measured as the difference between σ and π
complexes in a previous study of C-H bond activation.8 This
shift from a σ complex, where the metal is interacting with
electrons on a fluorine atom, to a η2 complex, where the metal
is interacting with the bond between two carbon atoms on the
ring, would change the geometry away from the transition
geometry. Measurements of W(CO)6 and W(CO)4(tmeda) in
C6F6 have shown that, while a solvent complex is formed after
photolysis, the C-F bond activation reaction never occurs. The
formation of an η2 complex with benzene provides a possible
explanation for this important difference between the metal
center in the C-F activating complex 3 and the metal center in
W(CO)4(tmeda) or W(CO)6. DFT calculations show that for the
C-F activating compound 3, the energy of the π complex is 7
kcal/mol higher in energy than the σ complex. For W(CO)6 this
is inverted, and the energy of the π complex is 10 kcal/mol
lower in energy than the σ complex. If formation of the σ
complex is necessary for the C-F reaction, then the pendant
ligand facilitates the reaction by acting as scaffolding, holding
the ring in the correct orientation for reaction, and eliminating
formation of the η2 complex.
Figure 3. DFT-calculated 3-dimenstional structure from DFT calcula-
tions of (a) parent after losing CO ligand, (b) intermediate σ complex,
and (c) final C-F activated complex.
for the final product to form, the pendant perfluorobenzene
ligand must displace the solvent molecule and form a complex
between the fluorine atom and the metal atom. This complex
can then go on to form the final C-F activated product. The
energy difference between the intra- and intermolecular systems
can be approximated by calculating the difference between the
energy of the intermolecular complex and the sum of the
energies of the intramolecular complex and the solvent ligand.
These energy values are shown in Table 1 in the energy
difference column. This shows that all three intermolecular
complexes are lower in energy than the intramolecular com-
plexes, with the methylene chloride complex as the weakest,
and acetonitrile the strongest. Reaction of this intramolecular
complex to form the C-F activation product lowers the energy
by 40 kcal/mol, providing the driving force for the final C-F
activation step.
An interesting variation of this molecular system was also
studied by Richmond and co-workers.23 Compound 5 is almost
identical to compound 1, but with a single bond attaching the
ring to the nitrogen rather than a double bond. Even though it
is very similar in structure to the original compound it was found
to have no C-F bond activation reactivity. This could be due
Summary
We have investigated the photochemical C-F bond activation
reaction of W(CO)4 with a tethered perfluorobenzene ligand.

24 J. Phys. Chem. B, Vol. 110, No. 1, 2006
Letters
(9) 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.
Following excitation and CO ligand loss, the coordinatively
unsaturated metal complex forms a σ complex with a solvent
molecule. Final reaction depends on displacement of the solvent
molecule by the tethered ring and formation of a σ complex.
This then reacts to form the final C-F activated product. This
work suggests that the C-F bond process could be made more
efficient if the molecule was altered to force the ring to be in
closer proximity to the metal center.
(10) Yang, H.; Asplund, M. C.; Kotz, K. T.; Wilkens, M. J.; Frei, H.;
Harris, C. B. J. Am. Chem. Soc. 1998, 120, 10154.
(11) Yang, H.; Snee, P. T.; Kotz, K. T.; Payne, C. K.; Frei, H.; Harris,
C. B. J. Am. Chem. Soc. 1999, 121, 9227.
(12) Yeston, J. S.; To, T. T.; Burkey, T. J.; Heilweil, E. J. J. Phys. Chem.
B 2004, 108, 4582.
(13) Jiao, T. J.; Pang, Z.; Burkey, T. J.; Johnston, R. F.; Heimer, T. A.;
Kleiman, V. D.; Heilweil, E. J. J. Am. Chem. Soc. 1999, 121, 4618.
Acknowledgment. The author gratefully acknowledges Dr.
Thomas Richmond for synthesis and characterization of the
metal complexes studied, and for assistance in interpreting the
data, and the Ira and Marylou Fulton Supercomputer center for
computational resources.
(14) George, M. W.; Dougherty, T. P.; Heilweil, E. J. J. Phys. Chem.
1996, 100, 201.
(15) Dougherty, T. P.; Grubbs, W. T.; Heilweil, E. J. J. Phys. Chem.
1994, 98, 9396.
(16) Richmond, T. G.; Osterberg, C. E.; Arif, A. M. J. Am. Chem. Soc.
1987, 109, 8091.
References and Notes
(17) Kiplinger, J. L.; Richmond, T. G.; Osterberg, C. E. Chem. ReV.
1994, 94, 373.
(1) Hoyano, J. K.; Graham, W. A. G. J. Am. Chem. Soc. 1982, 104,
3723.
(2) Janowicz, A. H.; Bergman, R. G. J. Am. Chem. Soc. 1982, 104,
352.
(3) Arndtsen, B. A.; Bergman, R. G.; Mobley, T. A.; Peterson, T. H.
Acc. Chem. Res. 1995, 28, 154.
(4) Bengali, A. A.; Bergman, R. G.; Moore, C. B. J. Am. Chem. Soc.
1995, 117, 3879.
(18) Richmond, T. G. Coord. Chem. ReV. 1990, 105, 221.
(19) Osterberg, C. E.; King, M. A.; Arif, A. M.; Richmond, T. G. Angew.
Chem., Int. Ed. Engl. 1990, 29, 888.
(20) Kendall, R. A.; Apra, E.; Bernholdt, D. E.; Bylaska, E. J.; Dupuis,
M.; Fann, G. I.; Harrison, R. J.; Ju, J. L.; Nichols, J. A.; Nieplocha, J.;
Straatsma, T. P.; Windus, T. L.; Wong, A. T. Comput. Phys. Commun.
2000, 128, 260.
(5) Bergman, R. G. Science 1984, 223, 902.
(6) Asbury, J. B.; Ghosh, H. N.; Yeston, J. S.; Bergman, R. G.; Lian,
T. Q. Organometallics 1998, 17, 3417.
(7) Asbury, J. B.; Hang, K.; Yeston, J. S.; Cordaro, J. G.; Bergman,
R. G.; Lian, T. Q. J. Am. Chem. Soc. 2000, 122, 12870.
(8) Asplund, M. C.; Snee, P. T.; Yeston, J. S.; Wilkens, M. J.; Payne,
C. K.; Yang, H.; Kotz, K. T.; Frei, H.; Bergman, R. G.; Harris, C. B. J.
Am. Chem. Soc. 2002, 124, 10605.
(21) Lian, T.; Bromberg, S. E.; Yang, H.; Proulx, G.; Bergman, R. G.;
Harris, C. B. J. Am. Chem. Soc. 1996, 118, 3769.
(22) Schultz, R. H.; Krav-Ami, S. J. Chem. Soc., Dalton Trans. 1999,
115.
(23) Buffin, B. P.; Poss, M. J.; Arif, A. M.; Richmond, T. G. Inorg.
Chem. 1993, 32, 3805.