CF3-Ph reductive elimination from [(Xantphos)Pd(CF 3)(Ph)]

Article abstract of DOI:10.1021/om200985g

CF₃-Ph reductive elimination from [(Xantphos)Pd(Ph)(CF ₃)] (1) and [(i-Pr-Xantphos)Pd(Ph)(CF₃)] (2) has been studied by experimental and computational methods. Complex 1 is cis in the solid state and predominantly cis in solution, undergoing degenerate cis-cis isomerization (ΔG^*_exp = 13.4 kcal mol ^-1; ΔG^*_calc = 12.8 kcal mol ^-1 in toluene) and slower cis-trans isomerization (ΔG _calc = +0.9 kcal mol^-1; ΔG^* _calc = 21.9 kcal mol^-1). In contrast, 2 is only trans in both solution and the solid state with trans-2 computed to be 10.2 kcal mol ^-1 lower in energy than cis-2. Kinetic and computational studies of the previously communicated (J. Am. Chem. Soc. 2006, 128, 12644), remarkably facile CF₃-Ph reductive elimination from 1 suggest that the process does not require P-Pd bond dissociation but rather occurs directly from cis-1. The experimentally determined activation parameters (ΔH^* = 25.9 ± 2.6 kcal mol^-1; ΔS^* = 6.4 ± 7.8 e.u.) are in excellent agreement with the computed data (ΔH^*_calc = 24.8 kcal mol^-1; ΔG^*_calc = 25.0 kcal mol^-1). ΔG^*_calc for CF₃-Ph reductive elimination from cis-2 is only 24.0 kcal mol^-1; however, the overall barrier relative to trans-2 is much higher (ΔG^* _calc = 34.2 kcal mol^-1) due to the need to include the energetic cost of trans-cis isomerization. This is consistent with the higher thermal stability of 2 that decomposes to PhCF₃ only at 100 °C and even then only in a sluggish and less selective manner. The presence of excess Xantphos has a minor decelerating effect on the decomposition of 1. A steady slight decrease in k_obs in the presence of 1 and 2 equiv of Xantphos then plateaus at [Xantphos]:1 = 5, 10, and 20. Specific molecular interactions between 1 and Xantphos are not involved in this kinetic effect (NMR, T₁ measurements). A deduced kinetic scheme accounting for the influence of extra Xantphos involves the formation of cis-[(η¹- Xantphos)₂Pd(Ph)(CF₃)] that, by computation, is predicted to access reductive elimination of CF₃-Ph with ΔG ^*_calc = 22.8 kcal mol^-1.

Full text of DOI:10.1021/om200985g

Article
pubs.acs.org/Organometallics
CF3−Ph Reductive Elimination from [(Xantphos)Pd(CF3)(Ph)]
Vladimir I. Bakhmutov,*,§ Fernando Bozoglian,† Kerman Gomez,† Gabriel Gonzalez,†
́
́
Vladimir V. Grushin,*,† Stuart A. Macgregor,*,‡ Eddy Martin,† Fedor M. Miloserdov,†
Maxim A. Novikov,† Julien A. Panetier,‡ and Leonid V. Romashov†
†The Institute of Chemical Research of Catalonia (ICIQ), Tarragona 43007, Spain
‡School of Engineering and Physical Sciences, William H. Perkin Building, Heriot-Watt University, Edinburgh EH14 4AS, U.K.
§Department of Chemistry, Texas A&M University, College Station, Texas 77842, United States
S
* Supporting Information
ABSTRACT: CF3−Ph reductive elimination from [(Xantphos)Pd-
(Ph)(CF3)] (1) and [(i-Pr-Xantphos)Pd(Ph)(CF3)] (2) has been
studied by experimental and computational methods. Complex 1 is
cis in the solid state and predominantly cis in solution, undergoing
degenerate cis−cis isomerization (ΔG≠exp = 13.4 kcal mol−1; ΔG≠
=
calc
12.8 kcal mol−1 in toluene) and slower cis−trans isomerization
(ΔGcalc = +0.9 kcal mol−1; ΔG≠calc = 21.9 kcal mol−1). In contrast, 2 is
only trans in both solution and the solid state with trans-2 computed
to be 10.2 kcal mol−1 lower in energy than cis-2. Kinetic and computational studies of the previously communicated (J. Am.
Chem. Soc. 2006, 128, 12644), remarkably facile CF3−Ph reductive elimination from 1 suggest that the process does not require
P−Pd bond dissociation but rather occurs directly from cis-1. The experimentally determined activation parameters (ΔH≠ = 25.9
2.6 kcal mol−1; ΔS≠ = 6.4 7.8 e.u.) are in excellent agreement with the computed data (ΔH≠calc = 24.8 kcal mol−1; ΔG≠
=
calc
25.0 kcal mol−1). ΔG≠calc for CF3−Ph reductive elimination from cis-2 is only 24.0 kcal mol−1; however, the overall barrier relative
to trans-2 is much higher (ΔG≠ = 34.2 kcal mol−1) due to the need to include the energetic cost of trans−cis isomerization.
calc
This is consistent with the higher thermal stability of 2 that decomposes to PhCF3 only at 100 °C and even then only in a
sluggish and less selective manner. The presence of excess Xantphos has a minor decelerating effect on the decomposition of 1. A
steady slight decrease in kobs in the presence of 1 and 2 equiv of Xantphos then plateaus at [Xantphos]:1 = 5, 10, and 20. Specific
molecular interactions between 1 and Xantphos are not involved in this kinetic effect (NMR, T1 measurements). A deduced
kinetic scheme accounting for the influence of extra Xantphos involves the formation of cis-[(η1-Xantphos)2Pd(Ph)(CF3)] that,
by computation, is predicted to access reductive elimination of CF3−Ph with ΔG≠ = 22.8 kcal mol−1.
calc
substrates, and so the extension of these methods to Pd-
catalyzed aromatic trifluoromethylation has been highly sought
after.
INTRODUCTION
■
Selectively fluorinated aromatic compounds are widely used as
building blocks and intermediates in the synthesis of
pharmaceuticals, agrochemicals, and specialty materials.1 Of
particular interest are derivatives bearing trifluoromethyl groups
on the aromatic ring.1−7 These important compounds are
manufactured via a Swarts-type process involving exhaustive
chlorination of a methyl group on the ring, followed by Cl/F
exchange on the trichloromethyl product with HF.8 In this
process, 3 equiv of Cl2 are consumed and 6 equiv of HCl are
formed for each equivalent of the desired products that is
produced. An alternative to this eco-unfriendly method would
be highly efficient metal-catalyzed cross-coupling of aromatic
electrophiles, such as haloarenes with an inexpensive
nucleophilic CF3 source. Such reactions employing stoichio-
metric amounts of copper are well-known,2−7 and more
recently, catalytic versions of this transformation were
developed.9 However, the copper-promoted/catalyzed reac-
tions only readily occur with costly iodoarenes, whereas
cheaper aryl bromides and chlorides are much less reactive
toward Cu-CF3 reagents.7 In contrast, palladium is able to
mediate cross-coupling reactions with Ar-Br and Ar-Cl
The intrinsic problem with nucleophilic aromatic trifluor-
omethylation, catalyzed by palladium, is the exceptional
strength and inertness of the Pd−CF3 bond: even at high
temperatures (135−145 °C), the CF3−Ar reductive elimination
step either does not occur at all10 or is sluggish and poorly
selective.11 In sharp contrast, analogous nonfluorinated
complexes reductively eliminate toluene at as low as 15−40
°C.12 In 2006, an important result for the area was
reported,13,14 showing that [(Xantphos)Pd(Ph)(CF3)] under-
goes clean and facile CF3−Ph reductive elimination at as low as
50−80 °C (eq 1). This finding demonstrated, for the first time,
the very possibility of Pd-catalyzed trifluoromethylation of aryl
halides. Indeed, in 2010, the first trifluoromethylation of
chloroarenes, catalytic in Pd-BrettPhos (or RuPhos), was
Special Issue: Fluorine in Organometallic Chemistry
Received: October 14, 2011
Published: December 28, 2011
© 2011 American Chemical Society
1315
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
populated trans isomer. Despite that, broadened 19F and 31P
NMR signals are always observed from the major, cis isomer of
1 at room temperature (Figures S1−S6, Supporting Informa-
tion). On cooling a solution of 1 in CD2Cl2 to 0 °C, the 31P
resonances from cis-1 appeared as 1:1 well-resolved doublets of
quartets at 8.4 ppm (P trans to CF3; JP−F = 48 Hz, JP−P = 22
Hz) and 12.6 ppm (P cis to CF3; JP−F = 8 Hz, JP−P = 22 Hz). In
THF, however, slightly less well-resolved 31P NMR multiplets
for cis-1 could only be observed at a much lower temperature,
−40 °C. Analysis of the VT NMR spectra of 1 in THF, toluene,
and CD2Cl2 (Figures S1−S6, Supporting Information) showed
that the temperature evolution of the 19F and 31P signals is
typical of P−P exchange leading to averaging of the P sites on
the NMR time scale. The observed temperature dependence of
the 31P line shapes originates from a multicenter exchange that
is difficult to treat quantitatively. Therefore, line-shape analysis
of the spectra was performed in approximation for a two-site
exchange with the splitting of the signals from the 31P−31P and
31P−19F spin−spin coupling replaced by the correspondingly
increased resonance widths. 31P line-shape simulations by a
conventional computer program produced exchange rate
constants and free energy values for degenerate cis-1 to cis-1
isomerization (Table 1). As can be seen, these parameters show
reported by Buchwald’s group.15 They also performed a
computational and kinetic study on [(BrettPhos)Pd(Ar)(CF3)]
to show that Pd−CF3 bond cleavage is the main contributor to
the activation barrier to reductive elimination.
Only three ligands are currently known that promote CF3−
Ar reductive elimination from Pd(II):16 Xantphos, BrettPhos,
and RuPhos. To find other, less expensive ligands for Pd-
catalyzed aromatic trifluoromethylation, a detailed mechanistic
understanding of CF3−Ar bond formation at Pd(II) is
desirable. Herein, we report a combined experimental and
computational study of the CF3−Ph reductive elimination from
[(Xantphos)Pd(Ph)(CF3)] (1), the first reported complex13 to
cleanly and readily undergo this transformation.
EXPERIMENTAL STUDIES
■
Synthesis and Characterization of [(Xantphos)Pd(Ph)-
(CF3)] and [(i-Pr-Xantphos)Pd(Ph)(CF3)]. The previously
developed procedure13 was used to synthesize 1 via treatment
of [(Xantphos)Pd(Ph)(F)] (prepared in situ) with Ruppert’s
reagent, CF3SiMe3. The fluoride complex was made by our
originally developed method involving ultrasound-promoted
Pd-I/AgF exchange.17 When a newly purchased batch of AgF
(Aldrich, 99.9% trace metals basis, Lot no. MKBB6837) was
used to react with [(Xantphos)Pd(Ph)(I)],13 the primary
product [(Xantphos)Pd(Ph)(F)] and eventually 1 prepared
from it were contaminated with a dark material. This
suggested17b that the new AgF reagent contained more
water18 than the one used in the original work.13 It has been
previously shown17b that the H2O-induced Pd(II)/P(III) to
Pd(0)/P(V) redox process readily occurring under the reaction
conditions produces deeply colored Pd(0) species that stain the
palladium fluoride product. Indeed, after the AgF reagent (5
mg) was sonicated in rigorously dry C6D6 (99% D, 0.6 mL) for
Table 1. Rate Constants and ΔG≠ Values (25 °C) for
Degenerate cis-1 to cis-1 Isomerization in Different Solvents
As Derived from 31P NMR Line-Shape Analysis
solvent
kexch, s−1
ΔG≠, kcal mol−1
THF
909
417
86
13.0
13.4
14.3
toluene
CD2Cl2
only a slight solvent dependence. In the absence of fast (on the
NMR time scale) Pd−P dissociation, a much slower process,
the degenerate isomerization of cis-1 likely proceeds via a
trigonal transition state (see below). The nondegenerate cis−
trans isomerization of 1 occurs with a higher energy barrier, as
follows from the NMR data.
We have also synthesized an analogue of 1 bearing isopropyl
groups on the phosphorus centers, [(i-Pr-Xantphos)Pd(Ph)-
(CF3)] (2; i-Pr-Xantphos = 4,5-bis(diisopropylphosphino)-9,9-
dimethylxanthene). The desired precursor for the synthesis of
2, [(i-Pr-Xantphos)Pd(Ph)(Br)], was not produced upon
treatment of Pd2dba3 with PhBr in the presence of i-Pr-
Xantphos in benzene or toluene.19 Ligand exchange between
[(Ph3P)2Pd(Ph)(Br)] and i-Pr-Xantphos in THF or benzene
gave 2 in high yield, but even when i-Pr-Xantphos was used in
30% excess, the starting complex [(Ph3P)2Pd(Ph)(Br)] was still
present (6−10%) in the reaction mixture at equilibrium (31P
NMR). Although attempts to isolate pure [(i-Pr-Xantphos)Pd-
(Ph)(Br)] in the bulk failed, we were able to characterize this
species by single-crystal X-ray diffraction (Figure 1) and also in
1
3 h, a peak from water at 0.5 ppm appeared in the H NMR
spectrum of the sample. Integration of the water signal against
the residual solvent peak allowed the lower limit of the water
content in the AgF reagent to be estimated at ca. 0.3 mol %.
The staining problem emerging from the presence of
excessive water in some AgF samples can be solved by our
previously developed17b technique. Indeed, when the reaction
of [(Xantphos)Pd(Ph)(I)] with the new AgF reagent was
repeated in the presence of 5 mol % PhI and 5 mol %
Xantphos, the solution of [(Xantphos)Pd(Ph)(F)] produced
was not nearly as dark after filtration. On treatment of this
solution with CF3SiMe3, 1 was formed and isolated pure as
well-shaped, pale yellow crystals in 60−85% yield, as previously
communicated.13
Although cis in the crystal, complex 1 exists as a mixture of
cis and trans isomers in solution.13 As expected, the cis-to-trans
ratio is solvent-dependent, being ca. 10:1 in benzene or toluene,
>20:1 in THF, and >50:1 in CH2Cl2. Although the equilibrium
between cis-1 and trans-1 establishes within the time of
dissolution, there is no fast exchange on the NMR time scale
between the two isomers at 25 °C. This is clearly indicated by
sharp, well-resolved 19F and 31P resonances from the less-
20
1
solution by H and 31P NMR. It was found, however, that 2
can be easily made by reacting [(Ph3P)2Pd(Ph)(Br)] with i-Pr-
Xantphos in THF, followed by treatment of the equilibrated
mixture with CF3SiMe3/CsF (eq 2). Note that 1 could not be
prepared this way from [(Xantphos)Pd(Ph)(I)] because of the
facile and irreversible displacement of Xantphos by “CF3−” that
is uncontrollably released from CF3SiMe3/F−.13 This is a clear
indication that, as expected, more electron-rich i-Pr-Xantphos
binds more tightly to Pd than Xantphos.
1316
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
phenyl ligand are nonequivalent as well, appearing as separate
resonances and suggesting hindered rotation around the Pd−
Ph bond. On the other hand, the 19F and 31P NMR signals from
2 appear as well-resolved multiplets, a triplet (−13.1 ppm) and
a quartet (20.0 ppm), respectively, with JP−F = 13.8 Hz,
pointing to free rotation around the Pd−CF3 bond.
Complex 2 appeared to be less prone to CF3−Ph reductive
elimination than 1. No reaction was observed after 2 was
heated in toluene at 80 °C overnight, whereas 1 undergoes full
decomposition to PhCF3 at this temperature within a few
hours.13 After 3 days at 100 °C, 2 was decomposed to 80−90%
conversion with only ca. 50−60% selectivity to PhCF3 (19F
NMR). Therefore, the CF3−Ph bond formation was studied in
detail only for complex 1.
Kinetic Studies of the Formation of PhCF3 from
[(Xantphos)Pd(Ph)(CF3)]. Two general pathways could be
considered for the formation of PhCF3 from 1 (Scheme 1). In
Figure 1. ORTEP drawing of [(i-Pr-Xantphos)Pd(Ph)(Br)] with
thermal ellipsoids drawn to the 50% probability level and all H atoms
omitted for clarity.
Scheme 1
Complex 2 is air- and moisture-stable. Unlike 1, which is cis
in the solid state and predominantly cis in solution,13 2 bearing
bulkier isopropyl groups on the P atoms is exclusively trans
both in the crystal (Figure 2) and in solution. The Pd atom in 2
addition to direct CF3−Ph reductive elimination (Pathway A),
one could imagine another, indirect route leading to the same
product (Pathway B). This other proposed pathway involves α-
fluorine elimination, followed by migratory insertion of the
resultant difluorocarbene into the Pd−Ph bond and subsequent
PhCF2−F reductive elimination. Indeed, α-fluorine elimination
is well known for late-transition-metal complexes bearing CF3
ligands,21−23 palladium included.11 Furthermore, migratory
insertion within a CH3PtCF2 moiety to give the correspond-
ing PtCF2CH3 species has been observed.24 Finally, closely
related reductive elimination of PhCOF from [(Ph3P)2Pd-
(PhCO)(F)] has been reported17a,14,25 to easily occur even
below room temperature.
The involvement of Pathway B could be examined by
carrying out the decomposition of 1 in the presence of water
because dihalocarbene complexes are instantaneously hydro-
lyzed to carbonyl species.21a,23 If the latter carries a σ-aryl ligand
on Pd, CO migratory insertion should easily occur, followed by
a series of reductive elimination processes leading to aromatic
carbonyl compounds, such as (ArCO)2O, ArCOF, and
ArCOOH.11 The decomposition of 1 in toluene−water at 60
°C, in the absence or in the presence of extra Xantphos (2
equiv), produced PhCF3 in high yield (19F NMR). Benzoic acid
derivatives PhCOX (X = F, OH, PhCOO), however, were not
Figure 2. ORTEP drawing of 2 with thermal ellipsoids drawn to the
50% probability level and all H atoms omitted for clarity.
1
is very crowded, as follows from the appearance of four H
NMR signals from the nonequivalent methyl groups of the
isopropyl substituents. Furthermore, all five protons of the σ-
1317
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
detected (19F NMR, GC-MS), which prompted us to exclude
Pathway B from further considerations.
Scheme 3
To gain insight into the mechanism of reaction 1, the kinetics
of the thermal decomposition of 1 in toluene were studied by
19F NMR. In the absence of extra Xantphos, the reaction
produced PhCF3 along with palladium black and [(Xant-
phos)2Pd] emerging from disproportionation of unstable
[(Xantphos)Pd] that is co-formed in the reductive elimination
process (Scheme 2).
potentially affect the rate of reductive elimination from the
metal center. Whatever the reason, the observed effect of an
extra ligand contrasts the reported26 lack of influence of added
Xantphos on the rate of reductive elimination of N-aryl
amidates from Xantphos-stabilized Pd(II).
Scheme 2
Possible interactions between 1 and Xantphos were probed
by 19F and 31P NMR analysis of 1 in the presence of a large
excess of Xantphos. This experiment was carried out in CD2Cl2
because of the insufficient solubility of 1 in toluene and
benzene and of Xantphos in THF. The addition of 20 equiv of
Xantphos to a solution of 1 in CD2Cl2 ([1] = 2 × 10−2 M)
resulted in the appearance of new, very low intensity signals in
the NMR spectra. The signals were registered approximately 10
min (19F) and 30 min (31P) after the NMR sample was
prepared. In the 19F NMR spectrum, a triplet emerged at −19.7
ppm with JP−F = 24 Hz. In the 31P NMR spectrum, a quartet at
16.1 ppm with the same coupling constant (JP−F = 24 Hz) and a
singlet of the same integral intensity at −19.5 ppm were
detected. These resonances were consistent with the structure
of trans-[(η1-Xantphos)2Pd(Ph)(CF3)], the cis isomer being
not detected. We, therefore, reasoned that [(η1-Xantphos)2Pd-
(Ph)(CF3)] was produced as a result of an equilibrium between
1 and Xantphos, as shown in eq 3. Integration of the new
The reaction proceeded with high selectivity toward the
formation of PhCF3 and displayed first-order kinetics.
Measuring the rate constants with experimental error of ca.
15% in the temperature range from +50 to +80 °C allowed for
the determination of the activation parameters: ΔH≠ = 25.9
2.6 kcal mol−1 and ΔS≠ = 6.4 7.8 e.u. Despite the modest
accuracy in the determination of the activation entropy, the
value obtained points to participation of one molecule of 1 in
the kinetically controlled step of the process. The reaction was
found to be twice as slow in THF as in toluene at 60 °C.
However, activation parameters in THF could not be
determined because of the narrow temperature window
available for this solvent (bp = 66 °C). In the kinetic runs in
toluene, the selectivity of the reaction was either >90% (at 50,
60, and 70 °C) or near to 90% (80 °C), indicating that the
simultaneous decomposition of the byproduct [(Xantphos)Pd],
an unstable zerovalent palladium complex, did not interfere
significantly with the CF3−Ph bond-forming process. It is
worth noting, however, that, in accord with the previously
reported data,13 the reaction was nearly 100% selective in the
entire temperature range when performed in the presence of
extra Xantphos to stabilize the [(Xantphos)Pd] in the form of
[(Xantphos)2Pd]. Therefore, a kinetic study of the reaction in
the presence of extra Xantphos was also carried out.
The reaction of 1 in the presence of Xantphos that is also
consumed (eq 1) exhibited first-order kinetics. A series of rate
measurement experiments revealed that Xantphos had a small,
yet reproducible, effect on the rate of the thermal
decomposition of 1. At 70 °C, kobs decreased from 6.4 ×
10−3 min−1 to 5.5 × 10−3 and 4.6 × 10−3 min−1 in the presence
of 1 and 2 equiv of added Xantphos, respectively. The effect
then plateaued at kobs = ca. 4.3 × 10−3 min−1 in the presence of
5, 10, and 20 equiv of Xantphos. The positive, nonzero y
intercept on the plot of kobs vs [Xantphos]−1 (Figure S10,
Supporting Information) suggested that reaction 1 might
proceed via two mechanistic pathways with similar barriers,
one of which is suppressed by an extra ligand. In principle, the
extra Xantphos effect might also be caused by (i) catalysis of
the reaction with some species that are sequestered by
Xantphos or (ii) changes in properties of the medium in the
vicinity of 1 in the presence of Xantphos and/or by specific
interactions between 1 and Xantphos. In addition, wide bite
angle diphosphines can also change their coordination mode
from η2 to η1, as shown in Scheme 3,27,28 and this may also
signals against those from 1 allowed for the determination of
the molar ratio of trans-[(η1-Xantphos)2Pd(Ph)(CF3)] to 1 at
ca. 1:50. On the basis of these data, Keq for reaction 3 was
estimated at ca. 5 × 10−2 L mol−1. Importantly, the triplet at
−19.7 ppm with JP−F = 24 Hz was also observed in the kinetic
runs in the presence of extra Xantphos. Reliable quantification
of the appearance and disappearance of this resonance in the
kinetic runs was precluded by its low intensity.
Specific molecular interactions between 1 and Xantphos
might change the nature of the medium in the vicinity of the
complex, leading to a decrease in conformational mobility of 1
and thereby affecting the rate of the reductive elimination. To
explore this hypothesis, we measured spin−lattice relaxation
times29 for the P nuclei in Xantphos alone, in 1 alone, and in a
2:1 mixture of Xantphos and 1 in toluene-d8 at ambient
temperature (Table 2). In all these measurements, the
concentrations were maintained at the levels used for the
kinetic studies. The data indicate that the T1 time of the P
nuclei in the free ligand was not affected by the presence of 1
1318
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
exchange in [(R3P)3RhX] complexes (R = H, Ph; X = H, Me,
CF3, Ph), and the accessibility of such processes is promoted by
the high trans influence of the anionic ligands X.23 In the
Table 2. Spin−Lattice Relaxation Times (T1) of Xantphos, 1,
and Their Mixtures in Toluene-d8 at 25 °C
T1, sec
present case, a second transition state, TS(cis-1 - cis-1′)CF , was
3
sample
Xantphos
9.3
cis-1
trans-1
also located with now CF3 axial and Ph equatorial. TS(cis-1 -
Xantphos (0.14 M)
cis-1′)CF is only slightly higher in energy than TS(cis-1 - cis-
3
1 (0.07 M)
4.0
3.2
6.3
1′)Ph (G = +13.6 kcal mol−1; see the Supporting Information),
reflecting the similar trans influences of the Ph and CF3 ligands.
The CF3 ligand has long been known to exhibit a very strong
trans influence.7
Xantphos (0.14 M) + 1 (0.07 M)
9.1
10.7
(9.3 vs 9.1 s)30 and points to a lack of interaction between 1
and Xantphos occurring rapidly on the NMR time scale. We
can, therefore, rule out any influence of such interactions on the
rate of reductive elimination from 1.
The cis−trans isomerization transition state, TS(cis-1 -
trans-1), exhibits near Cs symmetry and is formed via
simultaneous widening of both the C1−Pd−C2 and the P1−
Pd−P2 angles, to 116.3° and 133.8°, respectively. During the
cis-1 to trans-1 interconversion, the bowed structure of the
Xantphos ligand backbone flips in order to accommodate the
CF3 ligand in trans-1, and in TS(cis-1 - trans-1) this moiety is
nearly exactly planar. The free energy barrier to cis−trans
isomerization is 21.9 kcal mol−1, significantly higher than that
for cis−cis isomerization and consistent with cis−trans
isomerization being slow on the NMR time scale at 25 °C.
trans-1 itself is computed to be 0.9 kcal mol−1 less stable than
cis-1, which is, therefore, predicted to be the dominant species
at equilibrium in toluene, as is indeed observed. This difference
increases to 1.2 kcal mol−1 in THF and 1.3 kcal mol−1 in
CH2Cl2, mirroring the experimentally observed decrease in the
equilibrium concentration of trans-1 in these more polar
solvents.
Taken together, our observations on the experimentally
detected extra Xantphos effect indicate that this should be
incorporated into kinetic considerations. For that reason, the
VT kinetic study was repeated in the presence of 2 equiv of
Xantphos (approaching the plateau region; see above) and a
new set of activation parameters was obtained: ΔH≠ = 29.3
2.8 kcal mol−1 and ΔS≠ = 15.9 8.4 e.u. These new parameters
are only slightly distinguishable, within the determination
errors, from the ones measured in the absence of extra
Xantphos (ΔH≠ = 25.9 2.6 kcal mol−1 and ΔS≠ = 6.4 7.8
e.u.; see above). Nonetheless, the kinetic data (Figures S9 and
S10 and Tables S1 and S3, Supporting Information) clearly
indicate the trend for an increase in the effective energy barrier
in the presence of Xantphos.
COMPUTATIONAL STUDIES
■
CF3−Ph Reductive Elimination from cis-1. Experimental
evidence suggests that cis-1 undergoes a direct C−C reductive
elimination process, and calculations based on this mechanism
do reproduce the experimental behavior well. Thus, in toluene,
Density functional theory (DFT) calculations based on the
dispersion-corrected functional, B97-D, have been performed in
order to assess various possible mechanisms for CF3−Ph
reductive elimination from cis-1. As part of this study, we also
modeled the dynamic behavior of cis-1, both to compare with
the experimental NMR data and to elucidate the transition
states involved in these processes.
ΔH≠
= 24.8 kcal mol−1 (cf. 25.9
2.6 kcal mol−1
calc
experimentally). Moreover, ΔG≠
= 25.0 kcal mol−1,
calc
confirming a small entropic effect, and increases to 25.9 kcal
mol−1 when recomputed in THF, consistent with the slower
reaction observed experimentally in that medium. The
computed reductive elimination transition state, TS(cis-1 -
3), is accessed via narrowing the C1−Pd−C2 angle to 53.9°
and increasing the P1−Pd−P2 angle to 110.5°. The C1−Pd−
C2 plane also twists further away from square-planar
coordination (τ = 46.7°); similar twisting of an originally
planar reactant structure has also been computed for C(sp2)−
C(sp3) reductive elimination from [Pd(PH3)2(CH3)(R′)]
systems (R′ = Ph, vinyl).33 In the present case, the precursor
cis-1 also exhibits significant out-of-plane distortion, suggestive
of a ground-state destabilization effect. However, during
conformational searches for alternative reductive elimination
transition states, we located a range of structures, the most
extreme of which had τ = 14.2 and 86.5° (see the Supporting
Information). As both these structures remained close in energy
to TS(cis-1 - 3), it appears that a twisted transition-state
geometry is not a necessary condition for CF3−Ph reductive
elimination in this system. In all these transition states, CF3−Ph
reductive elimination is characterized by a significant elongation
of the Pd−CF3 bond, for example, from 2.08 Å in cis-1 to 2.38
Å in TS(cis-1 - 3). In contrast, the Pd−Ph distance in TS(cis-1
- 3) is 2.08 Å, only 0.02 Å longer than that in cis-1. This has
been noted before by Buchwald and co-workers,15 who ascribed
a major component of the CF3−Ar reductive elimination
barrier to Pd−CF3 bond lengthening. The initial intermediate
formed via C−C coupling from cis-1 is 3 (G = −20.7 kcal
Structure and Dynamic Behavior of cis-1. The
computed structure of cis-1 is shown in Figure 3 and compares
well with the reported experimental data.13 The metal−ligand
bond lengths are reproduced to within 0.04 Å, the major
discrepancy being an underestimation of the Pd−P2 bond trans
to Ph (calculated, 2.43 Å, cf. experimental, 2.466 Å). Key P1−
Pd−P2 and C1−Pd−C2 angles are reproduced to within 1°,
and a distortion from square-planar coordination (quantified by
τ, the angle between the C1−Pd−C2 and P1−Pd−P2 planes)
is also reproduced, although slightly exaggerated (τcalc = 34° cf.
26° experimentally).
We have characterized pathways for both the degenerate cis−
cis isomerization and the cis−trans isomerization of cis-1. cis−
cis isomerization proceeds via TS(cis-1 - cis-1′)Ph with ΔG≠
calc
= 12.8 kcal mol−1 in toluene, in excellent agreement with the
experimental value of 13.4 kcal mol−1. ΔG≠
is almost
calc
unchanged (12.9 kcal mol−1) when recomputed in either THF
or CH2Cl2, consistent with the lack of significant solvent
dependence for this process. TS(cis-1 - cis-1′)Ph exhibits a
truncated trigonal bipyramidal structure with an axial Ph ligand
and a vacant trans axial site. As the Ph ligand drops into this
axial position, the CF3 group moves to a central equatorial site
(C1−Pd−P1 = 128°; C1−Pd−P2 = 127°). These movements
continue after the transition state to give cis-1′, an equivalent
structure to cis-1, but now with CF3 trans to P2 and Ph trans to
P1. Similar trigonal structures have been located as
intermediates or transition states for intramolecular phosphine
1319
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
Figure 3. Computed stationary points for (i) cis−cis isomerization, (ii) cis−trans isomerization, and (iii) CF3−Ph reductive elimination in cis-1. Free
energies (kcal mol−1) are computed in toluene, and selected distances are given in angstroms. H atoms are removed for clarity, and for cis-1,
experimental data11 are included in italics for comparison.
mol−1), in which the PhCF3 product is bound in an η2-fashion
to the {(Xantphos)Pd} moiety.
indicate that this is 10.2 kcal mol−1 more stable than the cis
form. ΔG≠ for reductive elimination from cis-2 is 24.0 kcal
calc
Overall, our B97-D calculations reproduce the behavior of
cis-1 very well, with cis−cis isomerization being the most
accessible process (ΔG≠calc = 12.8 kcal mol−1), followed by cis−
trans isomerization (ΔG≠calc = 21.9 kcal mol−1), and finally C−
mol−1, actually slightly lower than that from cis-1. However, the
overall barrier relative to trans-2 is 34.2 kcal mol−1, and so the
lack of facile CF3−Ph reductive elimination is this case reflects
the relative inaccessibility of the cis isomer. Similarly, we
compute barriers in excess of 33 kcal mol−1 for CF3−Ph
reductive elimination from [(dppe)Pd(Ph)(CF3)] and
[(dppp)Pd(Ph)(CF3)], consistent with the need for prolonged
heating at high temperatures to observe this process
experimentally in these systems.11 Similar trends have been
reported in a recent computational study of CF3−Ph reductive
elimination from [Pd(L-L)Ph(CF3)] species that appeared
during the final drafting of this work.35
C reductive elimination (ΔG≠
= 25.0 kcal mol−1). The
calc
barriers for cis−cis isomerization and reductive elimination are
in excellent quantitative agreement with the experimental data,
and observed solvation effects are also qualitatively reproduced
through PCM calculations.34
For completeness, we have also modeled CF3−Ph reductive
elimination from [(i-Pr-Xantphos)Pd(Ph)(CF3)], 2. This
species is only observed as the trans isomer, and calculations
1320
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
Figure 4. Schematic forms of intermediate 4, with associated free energies computed in toluene (kcal mol−1) for each species and their associated
CF3−Ph reductive elimination transition states.
CF3−Ph Reductive Elimination from cis-1 in the
Presence of Added Xantphos. Calculations have also been
employed to shed light on the role of excess Xantphos in
retarding the rate of reductive elimination from cis-1.
Alternative processes may derive from an intermediate 4, in
which one phosphine arm has dissociated from cis-1, or from
the bis-Xantphos complex [(η1-Xantphos)2Pd(Ph)(CF3)] (5)
that was apparently observed experimentally in low equilibrium
concentration in CD2Cl2 (eq 3). We, therefore, need to assess
the ease of phosphine dissociation from, or Xantphos addition
to, cis-1.
allows for stronger secondary interactions to develop, through
either the oxygen or the phenyl substituents. These species are,
therefore, not truly 3-coordinate and so are more accessible
than 4Ph(1) or 4CF . The lowest-energy transition state derived
3
The computation of phosphine binding energies in organo-
metallic complexes, LnM-PR3, has received considerable
attention recently as one means to test the new generation of
dispersion-corrected functionals.36 Dispersion is not well
treated by standard functionals, which tend to underestimate
phosphine binding energies as a result, often dramatically so,
when bulky ligands are involved.23b,37 A dispersion-corrected
functional treats the intramolecular dispersion between the
LnM and PR3 fragments within the LnM−PR3 complex;
however, upon dissociation, the dispersion between the LnM
and PR3 species with their environment (e.g., solvent) is either
neglected or may not be well-captured by standard continuum
solvation models. Thus, ligand bond dissociation energies may
be severely overestimated, even when calculated with a
dispersion-corrected functional. Note that these problems do
not reflect the accuracy or otherwise of the DFT-D approach,
but rather an incomplete modeling of the chemical system
under consideration. These difficulties, and related issues
associated with estimation of entropic and solvation effects,
have been discussed very recently by Fey and Harvey.38
Against this background, we located a range of different
structures for intermediate 4 with the B97-D functional, and
these are shown schematically in Figure 4. The 3-coordinate T-
Figure 5. Computed structure of TS-4Ph(3) with associated free
energy computed in toluene in kcal mol−1 and selected distances given
in angstroms. H atoms are removed for clarity.
from these structures is TS-4Ph(3) (Figure 5) and features
significant interaction of one Xantphos Ph substituent with the
metal center. However, at +33.9 kcal mol−1, TS-4Ph(3) is almost
9 kcal mol−1 above TS(cis-1 - 3) and so would not be expected
to contribute significantly to the overall reductive elimination
process.
Given the difficulties described above in the computation of
phosphine dissociation energies, we have reoptimized the
structures in Figure 4 with the BP86 functional, where no
specific treatment of dispersion is included (see Figure S9 in
the Supporting Information). In most cases, the relative
energies decrease by between 2 and 8 kcal mol−1, the only
exception in fact being TS-4Ph(3) (G = +35.4 kcal mol−1). This
probably reflects the retention of a significant Pd···Ph
interaction in TS-4Ph(3), which, therefore, does not involve
complete ligand dissociation. This effect also means that the
energies of intermediates 4 (and their associated transition
states) are much less sensitive to dispersion effects than those
of species 5 (see below). With the BP86 functional, the most
accessible transition state is TS-4Ph(2) (G = +30.7 kcal mol−1).
However, this remains 9.4 kcal mol−1 above TS(cis-1 - 3) (G =
shaped structures 4Ph(1) (G = +15.8 kcal mol−1) and 4CF (G =
3
+18.5 kcal mol−1) are formed by phosphine arm loss (trans to
Ph and CF3, respectively), followed by the rotation of the
Xantphos moiety away from the Pd center. The barriers to C−
C reductive elimination from these species, (ΔG≠calc = 25.1 and
22.9 kcal mol−1, respectively) are similar or even slightly
reduced compared to the direct process from cis-1. However, in
both cases, the high energy of the 3-coordinate intermediate
means that the overall barriers to PhCF3 formation via these
species are in excess of 40 kcal mol−1. We also located
alternative structures 4Ph(2) (G = +8.6 kcal mol−1) and 4Ph(3)
(G = +10.0 kcal mol−1) in which the Pd−P bond trans to Ph in
cis-1 is cleaved but the orientation of the Xantphos ligand
1321
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
Figure 6. Computed structures of cis- and trans-5 and the associated CF3−Ph reductive elimination transition state, TS(cis-5 - 6). Selected distances
are given in angstroms, and H atoms are removed for clarity. Free energies in toluene are quoted relative to cis-1 plus free Xantphos, where the free
energy of trans-5 is estimated to be +2.0 kcal mol−1 from its equilibrium concentration in CH2Cl2 at room temperature (see text for details).
+21.0 kcal mol−1) when the latter is also recomputed with the
BP86 functional. Thus, with the both BP86 and BP97-D
functionals, the pathway via phosphine arm loss and
intermediate 4 remains distinctly less accessible than the direct
reductive elimination from cis-1.
The computed structures of the lowest-energy conformation
of both cis-5 and trans-5, as well as the associated CF3−Ph
reductive elimination transition state, TS(cis-5 - 6), are shown
in Figure 6. Both trans-5 and cis-5 exhibit regular square-planar
coordination geometries, and cis-5 is 5.1 kcal mol−1 less stable
than trans-5, reflecting the usual preference for the trans isomer
in [(PR3)2Pd(X)(Ph)] species. The major change in accessing
TS(cis-5 - 6) is again the lengthening of the Pd−CF3 distance,
in this case by ca. 0.3 Å. Much less distortion of the CF3−Pd−
Ph unit is seen in both cis-5 (τ = 12.8°) and TS(cis-5 - 6) (τ =
19.7°), suggesting that the more flexible bis-Xantphos
coordination mode imposes less steric pressure on the system
than the chelate ligand in cis-1 and TS(cis-1 - 3). Despite this,
the barrier to reductive elimination relative to cis-5 is only 22.8
kcal mol−1, the lowest value for this step computed in this
study. Relative to trans-5, this equates to a barrier of 27.9 kcal
mol−1, rising to 29.9 kcal mol−1 relative to cis-1 plus free
Xantphos. CF3−Ph reductive elimination through bis-Xantphos
cis-5 is, therefore, computed to be more accessible than
through the phosphine-dissociated intermediate 4, and the
calculations suggest that cis-1 and cis-5 may both contribute to
the reductive elimination process.
In assessing the accessibility of [(η1-Xantphos)2Pd(Ph)-
(CF3)], 5, we made use of the observation that addition of
excess Xantphos to cis-1 in CH2Cl2 leads to an equilibrium
containing a species that is formulated as trans-[(η1-
Xantphos)2Pd(Ph)(CF3)], trans-5, in a ca. 1:50 ratio (eq 3).
This implies a free energy change of around 2 kcal mol−1, and
we have used this value to benchmark a range of functionals for
the Xantphos addition step. We now found dramatic variation
in the computed Xantphos binding free energies, consistent
with the very large size of the cis-1/Xantphos system, with
GGA and hybrid-GGA functionals severely underestimating the
binding energy (BP86, BLYP, B3LYP, B3PW91: ΔG = +16 to
+20 kcal mol−1), whereas this was overestimated with
dispersion-corrected functionals (BLYP-D, B3PW91-D,39
B97-D, ΔG = −14 to −17 kcal mol−1). With the standard
B97-D (toluene) approach a value of −15.1 kcal mol−1 is
obtained, where the solvent correction amounts to only 1.5 kcal
mol−1. Given these difficulties, we decided to estimate the free
energy of trans-5 based on the equilibrium concentration of
this species as per eq 3; that is, trans-5 should lie approximately
2 kcal mol−1 above cis-1 plus free Xantphos. As the subsequent
reductive elimination from cis-[(η1-Xantphos)2Pd(Ph)(CF3)]
then involves no change in molecularity, the barrier to this
process should be less sensitive to the methodology employed
and, therefore, bear comparison to the earlier results on cis-1
and intermediate 4.
DISCUSSION
■
The above-described experimental and computational studies
suggest that the CF3−Ph reductive elimination from 1 occurs
from its cis isomer and does not require Pd−P bond
dissociation. However, a key factor strongly influencing the
interpretation of the first-order kinetic data obtained
experimentally for the formation of PhCF3 from 1 is the
unexpected effect of the extra Xantphos ligand on the measured
reaction rate. While being slower in the presence of an extra
1322
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
Scheme 4
ligand, the reaction still occurs even in the presence of a 20-fold
excess of Xantphos. As mentioned above, the extra Xantphos
effect could, a priori, be rationalized in terms of specific or
nonspecific interactions between molecules of 1 and Xantphos,
which reduce the ground-state energy of 1 and act rapidly in a
prekinetic step of the process. Such intermolecular interactions
would lead to a larger inertia moment of short-living
1·Xantphos, which, in turn, should reduce the correlation
times of molecular reorientations. The reduced correlation
times would then be well detected as a considerable decrease in
T1 times measured for Xantphos in the presence of 1.29
However, no such effect was observed as the 31P T1 time of
Xantphos did not change in the presence of 1 (Table 2).
Therefore, the extra Xantphos effect should be included in the
kinetic considerations.
As eq 4 includes [XP], kobs may depend on the Xantphos
concentration in the system. However, if k5 = 0 (C exists in fast
equilibrium with B but does not decompose to PhCF3), or if k4
= 0 (no conversion of B to C), the reaction should be
independent of [XP]. In other words, only the formation of C
and its transformation to PhCF3 can account for the extra
Xantphos effect within this model. Figure 7 displays the fitting
Kinetic Scheme 4 displays possible mechanistic pathways to
PhCF3 from 1. To follow the tradition of using letter characters
for reactive species in kinetic equations, we depict, in Scheme 4,
cis-1 as A, its open form 4 as B, 5 (a product of addition of
Xantphos to 4) as C, and Xantphos as XP. Cis−trans
isomerization of 1 is not included in the kinetic scheme
because this process is faster than the PhCF3 formation. As
shown above, the equilibrium between cis-1 and trans-1
establishes within the time of dissolution even at room
temperature, and calculations confirm the lower barrier
associated with cis−trans isomerization compared with
reductive elimination. Although only trans-5 could be observed
by 19F and 31P NMR (eq 3), cis-5 is more likely to undergo C−
C reductive elimination. Despite the fact that cis-5 is computed
to be 5.1 kcal mol−1 less stable than trans-5 (see above), both
isomers should be easily accessible from cis-1 via 4. Two
conclusions follow from Scheme 4. First, if B is not involved (k2
= 0), then the CF3−Ph reductive elimination proceeds directly
from A and should not be influenced by extra Xantphos.
Therefore, the Xantphos effect suggests that B should be
involved as a kinetically significant species. Second, if the final
product (PhCF3) is formed only from B, Xantphos in
sufficiently large excess should suppress the reaction altogether.
As this is inconsistent with the experimental results, the
formation of PhCF3 in the absence of Xantphos occurs from A
and, possibly, B. In the presence of Xantphos, C is formed,
which may also decompose to PhCF3 (Scheme 4). Two general
cases can be considered.
Figure 7. Dependence of kobs on Xantphos concentration at 70 °C,
fitted to y = (ab + cdx)/(af + dx), where y = kobs, c = (k1 + k2), and x =
[XP].
of the experimental rate constants described above to the
equation y = (ab + cdx)/(af + dx), where y = kobs, c = (k1 + k2),
and x = [XP], as performed with a standard program package.
Because of the large number of unknowns, these fitting
procedures are obviously not single-valued. The y(x) function,
however, is critical to the magnitude of constant c = k1 + k2, i.e.
the experimentally observed dependence of kobs on [XP] can be
reproduced only for k1 + k2 = 0.004 min−1 or for k2 = 0.004
min−1 if k1 = 0 (PhCF3 formation occurs only from B and/or
C). It is noteworthy that the kobs determination error of ca. 15%
(see above) is large for the magnitude of the Xantphos effect
observed, that is, the decrease of kobs by only ca. 50% when
going from [XP] = 0 to [XP] = 2−20 × [1].40 Being
reproducible, however, the effect ought to be discussed using
the available results, yet the experimental errors should be kept
Case 1. All three species, A, B, and C, produce PhCF3
(Scheme 4). The rate law for this case (eq 4) is derived on the
basis of formal kinetics, with intermediates B and C being
present in quasi-stationary concentrations.
1323
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
in mind and care exercised when drawing conclusions from the
data shown in Figure 7.
It is worth reemphasizing that two major assumptions have
been made in this work regarding C. First, the 19F and 31P
NMR signals observed from the minor species emerging on
addition of excess Xantphos to 1 are fully consistent with the
formulation of trans-[(η1-Xantphos)2Pd(Ph)(CF3)] (trans-C),
as shown in eq 3. However, no unambiguous evidence for this
formulation (e.g., an X-ray structure) could be obtained
because of the low concentration of this species and its lability
toward ligand loss leading to 1. Also, as described above, the
limited solubility of 1 in toluene at room temperature
precluded determination of Keq for reaction 3 in this solvent
that was used in the kinetic studies. Therefore, the Keq was
determined in CH2Cl2, a solvent that would likely affect the
rate of decomposition of 1 and that could not be used in the
kinetic study because of its low boiling point of 40 °C. Second,
the computed figure for the overall barrier to CF3−Ph reductive
elimination from cis-C (29.9 kcal mol−1) heavily relies on the
first assumption and suffers from the incomplete modeling of
the ligand addition process, in particular, solute−solvent
dispersive interactions, as discussed in detail above.
Importantly, however, these assumption-based studies have
produced a more reliable computed number of only 22.8 kcal
mol−1 for the barrier to CF3−Ph reductive elimination from cis-
C. We believe that this result may be viewed as a guideline for
the development of other ligands for CF3−Ar bond formation
at Pd(II). Apparently, a wide bite angle phosphine is not a
critical requirement for facile CF3−Ph reductive elimination
from Pd(II), and it is conceivable that bulky enough ligands
that nonetheless can adopt cis configuration on Pd may appear
suitable for the desired transformation.
If the equilibrium between A and B (Scheme 4) was fast on
the NMR time scale, it would contribute to the observed
temperature dependence of the 19F and 31P NMR spectra of
cis-1 (A). However, the obtained value of c = k1 + k2 = 0.004
min−1 shows that the equilibrium between A and B is slow, in
accord with the VT NMR analysis of the degenerate
isomerization of cis-1 (13.4 kcal mol−1 in toluene; see Table
1; ΔG≠
= 12.8 kcal mol−1) and in agreement with the
calc
computed free energy cost (+15.9 kcal mol−1) for full
dissociation of one of the Pd−P bonds in 1 (see above).
This kinetic model suggests that, in the absence of extra
Xantphos ([XP] = 0), CF3−Ph reductive elimination can
proceed directly from A and/or from its open form B. In the
presence of extra Xantphos, however, the formation of PhCF3
from B (k3 in eq 4) is, to some extent, suppressed, yet an
additional, slower pathway to PhCF3 via C becomes opera-
tional.
Case 2. Only A and C give rise to PhCF3, whereas B forms
in the reaction but does not undergo CF3−Ph reductive
elimination. In this case, k3 = 0 and, therefore, eq 4 transforms
to eq 5.41 Like in Case 1, in this case the fitting (Figure 7)
shows that the corresponding y(x) function is critical to the
magnitude of c = (k1 + k2) = 0.004 min−1.
Although Cases 1 and 2 are kinetically indistinguishable,42
our computational results suggest that Case 1 is improbable
because of the prohibitively high (>40 kcal mol−1) barrier to
reductive elimination from B. Case 2, however, is consistent
with both experimental and computational data pointing to
realistic free energy values for PhCF3 reductive elimination
from A (25.0 kcal mol−1) and from cis-C (29.9 kcal mol−1).
These numbers are in good agreement with the experimentally
determined activation parameters ΔH≠ = 25.9 2.6 kcal mol−1
and ΔS≠ = 6.4 7.8 e.u. (in the absence of extra Xantphos),
and ΔH≠ = 29.3 2.8 kcal mol−1 and ΔS≠ = 15.9 8.4 e.u. (in
the presence of 2 equiv of Xantphos).
A comment is also due on the suitability of Xantphos for Pd-
catalyzed trifluoromethylation of aromatic electrophiles, such as
haloarenes (Scheme 5). We think that this catalytic trans-
Scheme 5
As mentioned above, three events could, in principle, account
for the Xantphos effect, i.e. the CF3−Ph bond formation from 1
is (i) governed by two mechanistic pathways with similar
barriers, of which one is altered by an extra ligand; (ii) catalyzed
by some species that are sequestered by Xantphos; and (iii)
influenced by changes in properties of the medium in the
vicinity of 1 in the presence of Xantphos and/or by specific
interactions between 1 and Xantphos. While we have
demonstrated plausibility of item (i) and ruled out item (iii)
by the T1 studies, a comment is due on item (ii) that is catalysis
of reaction 1 by a small quantity of some species that may be
present in the sample of 1 and/or produced during the
reaction, and that are deactivated by extra Xantphos.
Apparently this scenario can be ruled out, too. The sample of
1 used in this work was spectroscopically pure and, therefore,
could only contain impurities in the amount of <1%. Therefore,
the molar ratio of Xantphos used in the studies (1−20 equiv to
1) to this impurity would be in the range of 100−2000, i.e. the
scavenging of the catalytically active species would be pseudo-
zero-order in Xantphos in all of the runs and no dependence of
kobs for the formation of PhCF3 from 1 on [Xantphos] would
be observed.
formation is unlikely to be developed with Xantphos, despite
the facile CF3−Ph reductive elimination from 1. One of the
problems emerges at the nucleophilic trifluoromethylation of
the Pd−X bond in [(Xantphos)Pd(Ar)(X)] that is formed in
the Ar-X oxidative addition step. The “CF3−” equivalents
1324
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
conventionally used for the Pd−CF3 bond formation, e.g.,
CF3SiMe3/F−, easily and irreversibly displace not only halides
on the metal but also the stabilizing tertiary phosphines11,13,43
to give catalytically inactive poly(trifluoromethyl) complexes
(shown as [LmPd(CF3)x(Ar)] in Scheme 5). This trans-
formation is avoided only for Pd complexes of tightly binding
chelating ligands, such as dppe, dppp, and tmeda,11 but not
Xantphos.13 A less widely recognized, yet perhaps more
general, problem is the enhanced ability of trifluoromethyl Pd
aryls to undergo transmetalation with other aryl Pd complexes,
including [(Xantphos)Pd(Ar)(X)], a key intermediate that is
formed in the oxidative addition step. As a result, [(Xantphos)-
Pd(Ar)(X)] and [(Xantphos)Pd(CF3)(Ar)] react to give
catalytically inactive [(Xantphos)Pd(CF3)(X)] and
[(Xantphos)Pd(Ar)2], the latter quickly undergoing Ar−Ar
reductive elimination (Scheme 5). Like numerous other
transmetalation reactions, this documented11,13 catalyst deacti-
vation pathway should be facilitated by “nucleophilic aid”, that
is, coordination of the halide ligand X to the Pd atom bearing
the CF3 ligand. As the CF3 ligand stabilizes7 orbitals on the
metal, LUMO included, this halide ligand coordination
(bridging) should be favored, contributing to the trans-
metalation process. We propose that this might be a reason
for aryl chlorides being much better substrates than aryl
bromides for Buchwald’s Pd-catalyzed trifluoromethylation15
because LPd(Ar)X transmetalates LPd(Ar)CF3 more easily for
X = Br that is more electron-rich than Cl. Ways to minimize the
transmetalation leading to catalyst deactivation should be to use
ArX substrates that are derivatives of less nucleophilic X− and
run the process at lower concentrations of Pd to slow down
transmetalation that is a bimolecular reaction.
An interesting effect of extra Xantphos has been observed on
the rate of decomposition of 1 at 70 °C. In the presence of 1
and 2 equiv of Xantphos, a steady slight decrease in kobs is
detected, which then plateaus at Xantphos:1 = 5, 10, and 20,
the overall drop in reaction rate being only ca. 50%. A series of
T1 measurements indicated that this minor effect is not caused
by changes in either the properties of the medium or weak
specific molecular interactions between 1 and Xantphos. A
kinetic scheme has been proposed that accounts for the
Xantphos effect and involves the formation of [(η1-
Xantphos)2Pd(Ph)(CF3)] (5) in small quantities from 1 and
the excess ligand. A computational study prompted by the
kinetic scheme gave ΔG≠ = 22.8 kcal mol−1 for CF3−Ph
calc
reductive elimination from cis-5, the lowest barrier for this step
computed in the present work. The energy of cis-5 relative to
cis-1 and free Xantphos could not be accurately assessed
because of the difficulty in capturing all environmental effects
with the current chemical model system. Importantly, however,
the low barrier to CF3−Ph bond formation from cis-5 implies
that a wide bite angle diphosphine is not a critical requirement
for facile CF3−Ph bond formation at Pd(II) and that bulky
enough ligands able to adopt a cis configuration on Pd may
appear suitable for the desired transformation.
EXPERIMENTAL SECTION
■
All chemicals were purchased from Aldrich, Alfa Aesar, and Apollo
Scientific. All manipulations were performed under argon in a
glovebox. Benzene, toluene, and THF were distilled from Na/
OCPh2 and stored over 4 Å molecular sieves in a glovebox. i-Pr-
Xantphos44 and [(Ph3P)2Pd(Ph)(Br)]45 were synthesized by the
literature procedures. Routine NMR spectra were recorded on a
Bruker Avance 400 Ultrashield spectrometer equipped with a BBFO
plus SmartProbe. A Bruker Avance 500 Ultrashield NMR instrument
equipped with a BBI probe was used for VT NMR and kinetic studies.
31P T1 time measurements were performed using a Bruker Avance III
500 Ultrashield NMR spectrometer equipped with a QNP CryoProbe.
The standard inversion−recovery method was applied to collect the
data that were treated to calculate T1. All NMR data were processed
with the Topspin 2.1 Bruker software. A Bruker-Nonius diffractometer
equipped with an Apex II 4K CCD area detector was used for single-
crystal diffraction studies. An Agilent Technologies 7890A chromato-
graph equipped with a 5975C MSD unit was used for GC-MS analysis.
Elemental analyses were performed by the Microanalysis Center at the
Complutense University of Madrid.
CONCLUSIONS
■
Two trifluoromethyl Pd(II) phenyl complexes, [(Xantphos)-
Pd(Ph)(CF3)] (1) and [(i-Pr-Xantphos)Pd(Ph)(CF3)] (2),
have been studied by experimental and computational methods.
Complex 1 is exclusively cis in the solid state and
predominantly cis in solution, undergoing degenerate cis−cis
isomerization (ΔG≠ = 13.4 kcal mol−1; ΔG≠ = 12.8 kcal
exp
calc
mol−1 in toluene) that is much less energy demanding than its
cis−trans isomerization (ΔG≠ = 21.9 kcal mol−1). In accord
calc
with the experimental data, cis-1 is computed to be ca. 1 kcal
mol−1 more stable than trans-1. In contrast, 2 is only observed
as the trans isomer in both solution and the solid state (NMR,
X-ray), also in full accord with the much higher free energy
(+10.2 kcal mol−1) for the cis isomer in this case.
Preparation of [(Xantphos)Pd(Ph)(CF3)] (1). This complex was
prepared using our previously developed13 procedure, except
[(Xantphos)Pd(Ph)(F)] was generated from [(Xantphos)Pd(Ph)(I)]
and AgF in the presence of 5 mol % PhI and 5 mol % Xantphos (see
above).
As previously communicated,13 1 undergoes remarkably
facile CF3−Ph reductive elimination from Pd(II) under mild
conditions to give PhCF3 in high yield. Kinetic and computa-
tional studies of PhCF3 formation from 1 indicated that the
reductive elimination does not require P−Pd bond dissociation
but rather occurs directly from the cis isomer. The
experimentally determined activation parameters (ΔH≠ = 25.9
Preparation of [(i-Pr-Xantphos)Pd(Ph)(CF3)] (2). A mixture of i-
Pr-Xantphos (0.20 g, 0.45 mmol) and [(Ph3P)2Pd(Ph)(Br)] (0.27 g,
0.34 mmol) in THF (5 mL) was gently warmed to 50 °C at stirring
until all solids had dissolved to produce a yellow solution. After 36 h at
room temperature, the equilibrated mixture contained [(i-Pr-
Xantphos)Pd(Ph)(Br)] and [(Ph3P)2Pd(Ph)(Br)] in a 92:8 molar
ratio (31P NMR). This solution was vigorously stirred with CsF (0.10
g, 0.66 mmol) and CF3SiMe3 (0.10 mL, 0.68 mmol) for 4.5 h. Extra
CF3SiMe3 (0.05 mL, 0.34 mmol) was then added, and after stirring for
2 more hours, full conversion was observed by 31P NMR. The solution
was evaporated in air and the residue extracted with CH2Cl2 (5 mL).
The extract was filtered, evaporated to ca. 0.5 mL, and treated with
MeOH (15 mL in portions). After 1 h, the pale tan, well-shaped
crystals were separated, washed with MeOH, and dried under vacuum.
The yield of 2 was 0.13 g (55%). X-ray quality crystals were grown
from benzene−MeOH. Calcd for C34H45F3OP2Pd: C, 58.75; H, 6.52.
2.6 kcal mol−1; ΔS≠ = 6.4
7.8 e.u.) are in excellent
agreement with the computed data (ΔH≠calc = 24.8 kcal mol−1;
ΔG≠ = 25.0 kcal mol−1). Interestingly, ΔG≠ for CF3−Ph
calc
calc
reductive elimination from cis-2 is 24.0 kcal mol−1, slightly
lower than that from cis-1. Nonetheless, the overall barrier
relative to trans-2 is much higher (ΔG≠calc = 34.2 kcal mol−1) as
it includes the high energy cost of 10.2 kcal mol−1 to isomerize
trans-2 to cis-2. This is consistent with the higher thermal
stability of 2 that decomposes to PhCF3 only at 100 °C, and
even then only in a sluggish and less-selective manner.
1
Found: C, 58.70; H, 6.45. H NMR (C6D6, 25 °C), δ: 0.3 (q, J = 7.1
Hz, 6H, CH3CH), 0.85 (q, J = 6.3 Hz, 6H, CH3CH), 1.3 (s, 3H, arom-
1325
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
CH3), 1.4 (q, J = 7.6 Hz, 6H, CH3CH), 1.5 (m, 6H, CH3CH), 1.65 (s,
3H, arom-CH3), 2.3 (m, 2H, CHCH3), 3.15 (m, 2H, CHCH3), 5.6 (d,
J = 7.5 Hz, 1H, σ-C6H5), 6.35 (t, J = 7.5 Hz, 1H, σ-C6H5), 6.65 (t, J =
7.2 Hz, 1H, σ-C6H5), 6.85 (m, 2H, arom H), 6.95 (t, J = 7.4 Hz, 1H, σ-
C6H5), 7.05 (t, J = 7.7 Hz, 2H, arom H), 7.35 (dd, J = 7.7 and 1.5 Hz,
2H, arom H), 8.4 (d, J = 7.5 Hz, 1H, σ-C6H5). 19F NMR (C6D6, 25
°C), δ: −13.1 (t, JF−P = 13.8 Hz). 31P NMR (C6D6, 25 °C), δ: 20.0 (q,
JF−P = 13.8 Hz).
Determination of Keq for the Reaction of 1 with Xantphos.
Addition of recrystallized Xantphos (140 mg, 0.24 mmol) to a solution
of 1 (10 mg, 1.2 × 10−2 mmol) in CD2Cl2 (0.6 mL) in a 5 mm NMR
tube resulted in the appearance of new signals in the NMR spectra. In
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: vgrushin@iciq.es (V.V.G.), s.a.macgregor@hw.ac.uk
(S.A.M.), bakhmoutov@mail.chem.tamu.edu (V.I.B.).
ACKNOWLEDGMENTS
■
We thank Prof. Jose
́
Ramon
́
Galan
́
-Mascaros
́
for a discussion.
The ICIQ Foundation and Consolider Ingenio 2010 (Grant
CSD2006-0003) are thankfully acknowledged for support, as
are the EPSRC and Heriot-Watt University for a doctoral
training award to J.A.P.
the 19F NMR spectrum, a new triplet emerged at −19.7 ppm (JP−F
=
24 Hz). In the 31P NMR spectrum, a quartet at 16.1 ppm with the
same coupling constant (JP−F = 24 Hz) and a singlet of the same
integral intensity at −19.5 ppm were observed. These data are
consistent with the structure of trans-[(η1-Xantphos)2Pd(Ph)(CF3)].
Integration of the new signals against those from 1 allowed for the
determination of the molar ratio of trans-[(η1-Xantphos)2Pd(Ph)-
(CF3)] to 1 at ca. 1:50. On the basis of these data, Keq between 1 +
Xantphos and trans-[(η1-Xantphos)2Pd(Ph)(CF3)] in CD2Cl2 at 25
°C was estimated at ca. 5 × 10−2 L mol−1.
REFERENCES
■
(1) For selected monographs, see: (a) Clark, J. H.; Wails, D.;
Bastock, T. W. Aromatic Fluorination; CRC Press: Boca Raton, FL,
1996. (b) Kirsch, P. Modern Fluoroorganic Chemistry; Wiley-VCH:
Weinheim, Germany, 2004. (c) Uneyama, K. Organofluorine Chemistry;
Blackwell: Oxford, U. K., 2006. (d) Ojima, I. Fluorine in Medicinal
Chemistry and Chemical Biology; Wiley-Blackwell: Chichester, U. K.,
2009.
Kinetic Measurements. In a glovebox, a 5 mm NMR tube was
charged with 1 (4 mg), Xantphos (in certain experiments), and
toluene (0.7 mL). The sample was quickly sonicated at room
temperature to facilitate dissolution of 1 and placed in the preheated
probe of an NMR spectrometer. The reactions were followed by 19F
NMR spectroscopy (NS = 64 − 256; D1 = 5 s; SW = 200 ppm) and
kobs derived from plots of integral intensities of the signals from PhCF3
and 1 versus time. The spectra were automatically measured every 15
min for the runs at 80 and 70 °C, and every 30 min and 1 h for the
runs at 60 and 50 °C, respectively. The concentration of 1 used (6.9
mM) was close to the solubility of the complex in toluene at room
temperature and to the lower limit for accurate rate measurements by
integration of the signals from 1 and from the product, PhCF3. Higher
concentrations of 1 could not be used because of its limited solubility
in toluene. Exponential fittings were performed using the SPECFIT46
software. All traces were fitted to a single exponential model. No
significant difference in the calculated constants was observed between
the [(Xantphos)Pd(Ph)(CF3)] and PhCF3 plots. In all cases, the final
values reported in the Supporting Information correspond to average
of all calculated constants for experiments under the same conditions.
Computational Details. Calculations were run with Gaussian
0947 with geometry optimizations using the B97-D dispersion
corrected functional.48 Pd and P centers were described with the
Stuttgart RECPs and associated basis sets,49 with added d-orbital
polarization on P (ζ = 0.387).50 6-31G** basis sets were used for all
other atoms.51 All stationary points were fully characterized via
analytical frequency calculations as either minima (all positive
eigenvalues) or transition states (one negative eigenvalue), and IRC
calculations and subsequent geometry optimizations were used to
confirm the minima linked by each transition state. Extensive
functional tests were carried out on the barrier to reductive elimination
from cis-1, and the inclusion of dispersion effects was found to
increase the barrier for this process and so produce better agreement
with experiment. See the Supporting Information for details. All
stationary points were subject to extensive conformational searching
using our recently published protocol.52 Free energies are computed at
298.15 K and 1 atm pressure, and solvation effects are included
through the PCM approach, using UFF atomic radii.
(2) Burton, D. J.; Yang, Z. Y. Tetrahedron 1992, 48, 189.
(3) McClinton, M. A.; McClinton, D. A. Tetrahedron 1992, 48, 6555.
(4) Burton, D. J.; Lu, L. Top. Curr. Chem. 1997, 193, 45.
(5) Schlosser, M. Angew. Chem., Int. Ed. 2006, 45, 5432.
(6) Roy, S.; Gregg, B. T.; Gribble, G. W.; Le, V.-D.; Roy, S.
Tetrahedron 2011, 67, 2161.
(7) Tomashenko, O. A.; Grushin, V. V. Chem. Rev. 2011, 111, 4475.
(8) Filler, R. Adv. Fluorine Chem. 1970, 6, 1.
(9) (a) Inoue, M.; Araki, K. Jpn. Patent JP 2009-234921, 2009.
(b) Oishi, M.; Kondo, H.; Amii, H. Chem. Commun. 2009, 1909.
(c) Knauber, T.; Arikan, F.; Roschenthaler, G.-V.; Gooßen, L. J.
̈
Chem.Eur. J. 2011, 17, 2689. (d) Popov, I.; Lindeman, S.; Daugulis,
O. J. Am. Chem. Soc. 2011, 133, 9286. (e) Weng, Z.; Lee, R.; Jia, W.;
Yuan, Y.; Wang, W.; Feng, X.; Huang, K.-W. Organometallics 2011, 30,
3229. (f) Li, Y.; Chen, T.; Wang, H.; Zhang, R.; Jin, K.; Wang, X.;
Duan, C. Synlett 2011, 1713.
(10) Culkin, D. A.; Hartwig, J. F. Organometallics 2004, 23, 3398.
(11) Grushin, V. V.; Marshall, W. J. J. Am. Chem. Soc. 2006, 128,
4632.
(12) (a) Ozawa, F.; Kurihara, K.; Fujimori, M.; Hidaka, T.;
Toyoshima, T.; Yamamoto, A. Organometallics 1989, 8, 180.
(b) Brown, J. M.; Guiry, P. J. Inorg. Chim. Acta 1994, 220, 249.
(13) Grushin, V. V.; Marshall, W. J. J. Am. Chem. Soc. 2006, 128,
12644.
(14) Grushin, V. V. Acc. Chem. Res. 2010, 43, 160.
(15) Cho, E. J.; Senecal, T. D.; Kinzel, T.; Zhang, Y.; Watson, D. A.;
Buchwald, S. L. Science 2010, 328, 1679.
(16) CF3−Ar bond formation at Pd(IV) has been recently reported:
(a) Ball, N. D.; Kampf, J. W.; Sanford, M. S. J. Am. Chem. Soc. 2010,
132, 2878. (b) Ye, Y.; Ball, N. D.; Kampf, J. W.; Sanford, M. S. J. Am.
Chem. Soc. 2010, 132, 14682. (c) Ball, N. D.; Gary, J. B.; Ye, Y.;
Sanford, M. S. J. Am. Chem. Soc. 2011, 133, 7577.
(17) (a) Fraser, S. L.; Antipin, M. Yu.; Khroustalyov, V. N.; Grushin,
V. V. J. Am. Chem. Soc. 1997, 119, 4769. (b) Pilon, M. C.; Grushin, V.
V. Organometallics 1998, 17, 1774.
(18) Complete removal ofwater from AgF originating from aqueous
solutions is virtually impossible: Horn, E.; Snow, M. R. Aust. J. Chem.
1980, 33, 2369.
(19) Instead, the reaction produced a stable zerovalent Pd complex
that is formulated as [(i-Pr-Xantphos)2Pd2(dba)] on the basis of the
1H NMR data. Numerous attempts to obtain X-ray quality crystals of
this complex failed, mostly because the compound crystallizes as
extremely thin, fiber-like needles that are not suitable for single-crystal
X-ray analysis. Heating of the complex in the presence of PhBr did not
produce selectively [(i-Pr-Xantphos)Pd(Ph)(Br)] for the synthesis of
2, but rather a mixture of complexes, including a product of dba
migratory insertion into a Pd−Ph bond (1H NMR).
ASSOCIATED CONTENT
■
S
* Supporting Information
VT NMR spectra, rate constants, Eyring plots, alternative
kinetic models, and details of computational (PDF, including
full ref 47) and crystallographic studies (CIF). This material is
available free of charge via the Internet at http://pubs.acs.org.
1326
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
(20) 1H NMR for [(i-Pr-Xantphos)Pd(Ph)(Br)] (CD2Cl2, 25 °C), δ:
1.05 (m, 12H, i-Pr-CH3), 1.3 (m, 12H, i-Pr-CH3), 1.8 (s, 6H,
xanthene-CH3), 2.6 (m, 4H, i-Pr-CH), 7.1 (t, 1H, 4-PdPh), 7.2 (t, 2H,
3-PdPh), 7.4 (dm, 2H, 2-PdPh), 7.5 (tt, J = 7.5 and 0.7 Hz, 2H,
xanthene-CH), 7.55 (m, 2H, xanthene-CH), 7.85 (dd, J = 7.5 and 2.0
Hz, 2H, xanthene-CH). The structure of [(i-Pr-Xantphos)Pd(Ph)
(Br)] in solution strongly depends on the polarity of the medium.
CD2Cl2 solutions of this complex are colorless, displaying a singlet in
the 31P NMR spectra at 34.4 ppm, whereas in benzene and THF, the
complex is bright yellow, resonating at 15.2 and 14.9 ppm,
respectively. The significant difference in the 31P NMR chemical
shifts as well as in color indicates that, in less-polar benzene and THF,
the complex exists as a neutral species [(i-Pr-Xantphos)Pd(Ph)(Br)],
whereas in more polar CD2Cl2, it is ionic, [(i-Pr-Xantphos)
Pd(Ph)]+Br−, with the i-Pr-Xantphos ligand being in a P,O,P-
coordination mode. For neutral and ionic forms of closely related
Xantphos-type complexes, see, for example: (a) Kamer, P. C. J.; Van
Leeuwen, P. W. N. M.; Reek, J. N. H. Acc. Chem. Res. 2001, 34, 895.
(b) Zuideveld, M. A.; Swennenhuis, B. H. G.; Boele, M. D. K.; Guari,
Y.; van Strijdonck, G. P. F; Reek, J. N. H.; Kamer, P. C. J.; Goubitz, K.;
Fraanje, J.; Lutz, M.; Spek, A. L.; van Leeuwen, P. W. N. M. Dalton
2002, 2308. (c) Koblenz, T. S.; Dekker, H. L.; De Koster, C. G.; Van
Leeuwen, P. W. N. M.; Reek, J. N. H. Chem. Commun. 2006, 1700.
(21) For reviews, see: (a) Brothers, P. J.; Roper, W. R. Chem. Rev.
1988, 88, 1293. (b) Torrens, H. Coord. Chem. Rev. 2005, 249, 1957.
(c) Hughes, R. P. Eur. J. Inorg. Chem. 2009, 4591.
Crespo, P.; Hernando, A.; Litran, R.; Sanchez Lopez, J. C.; Lopez
Cartes, C.; Fernandez, A.; Ramirez, J.; Gonzalez Calbet, J.; Vallet, M.
Phys. Rev. Lett. 2003, 91, 237203. (e) Parolin, T. J.; Salman, Z.;
Chakhalian, J.; Song, Q.; Chow, K. H.; Hossain, M. D.; Keeler, T. A.;
Kiefl, R. F.; Kreitzman, S. R.; Levy, C. D. P.; Miller, R. I.; Morris, G.
D.; Pearson, M. R.; Saadaoui, H.; Wang, D.; MacFarlane, W. A. Phys.
Rev. Lett. 2007, 98, 047601. (f) Oba, Y.; Shinohara, T.; Oku, T.;
Suzuki, J.-i.; Ohnuma, M.; Sato, T. J. Phys. Soc. Jpn. 2009, 78, 044711.
(g) Oba, Y.; Okamoto, H.; Sato, T.; Shinohara, T.; Suzuki, J.;
Nakamura, T.; Muro, T.; Osawa, H. J. Phys. D 2008, 41, 134024.
(h) Teng, X.; Han, W.-Q.; Ku, W.; Hucker, M. Angew. Chem., Int. Ed.
2008, 47, 2055. (i) Xiao, C.; Ding, H.; Shen, C.; Yang, T.; Hui, C.;
Gao, H.-J. J. Phys. Chem. C 2009, 113, 13466.
(32) (a) Evans, D. F. J. J. Chem. Soc. 1959, 2003. (b) Live, D. H.;
Chan, S. I. Anal. Chem. 1970, 42, 791. (c) Ostfeld, D.; Cohen, I. A. J.
Chem. Educ. 1972, 49, 829. (d) Schubert, E. M. J. Chem. Educ. 1992,
69, 62.
(33) (a) Ananikov, V. P.; Musaev, D. G.; Morokuma, K.
Organometallics 2005, 24, 715. See also (b) Sakaki, S.; Mizoe, N.;
Musashi, Y.; Biswas, B.; Sugimoto, M. J. Phys. Chem. A 1998, 102,
8027.
(34) Calculations also probed the energetics of Pathway B, based on
a sequence of α-F migration, Ph transfer onto the CF2 ligand, and F−
CF2Ph coupling (see Scheme 1). An overall barrier of 26.1 kcal mol−1
was computed, with the highest-lying transition state corresponding to
the Ph transfer step. Thus, calculations suggest that this mechanism
may be competitive with direct reductive elimination; however, we rule
out this process on the basis of the experimental studies carried out in
excess water, which provided no evidence for a difluorocarbene
intermediate. See the Supporting Information for full details.
(35) Anstaett, P.; Schoenebeck, F. Chem. Eur. J. 2011, 17, 12340.
The transition state for CF3−Ph reductive elimination from cis-1
reported in this study exhibits less out-of-plane distortion than in
TS(cis-1 - 3), but was only slightly higher in energy when recomputed
with the B97-D approach used here. In general, we found that the
energy of the transition states and the precise ordering of the different
transition-state structures varied subtly depending on whether
dispersion effects were included in the calculation or not.
(22) (a) Huang, D.; Caulton, K. G. J. Am. Chem. Soc. 1997, 119,
3185. (b) Huang, D.; Koren, P. R.; Folting, K.; Davidson, E. R.;
Caulton, K. G. J. Am. Chem. Soc. 2000, 122, 8916.
(23) (a) Goodman, J.; Grushin, V. V.; Larichev, R. B.; Macgregor, S.
A.; Marshall, W. J.; Roe, D. C. J. Am. Chem. Soc. 2009, 131, 4236.
(b) Goodman, J.; Grushin, V. V.; Larichev, R. B.; Macgregor, S. A.;
Marshall, W. J.; Roe, D. C. J. Am. Chem. Soc. 2010, 132, 12013.
(24) Appleton, T. G.; Hall, J. R.; Mathieson, M. T.; Neale, D. W. J.
Organomet. Chem. 1993, 453, 307.
(25) Grushin, V. V. Chem.Eur. J. 2002, 8, 1006.
(26) Fujita, K.; Yamashita, M.; Puschmann, F.; Martinez Alvarez-
Falcon, M.; Incarvito, C. D.; Hartwig, J. F. J. Am. Chem. Soc. 2006, 128,
9044.
(36) Grimme, S. J. Comput. Chem. 2004, 25, 1463.
(27) Hamann, B. C.; Hartwig, J. F. J. Am. Chem. Soc. 1998, 120, 3694.
(28) Birkholz, M.-N.; Freixa, Z.; van Leeuwen, P. W. N. M. Chem.
Soc. Rev. 2009, 38, 1099.
(37) (a) Sieffert, N.; Buehl, M. Inorg. Chem. 2009, 48, 4622.
(b) Minenkov, Y.; Occhipinti, G.; Jensen, V. R. J. Phys. Chem. A 2009,
113, 11833. (c) McMullin, C. L.; Jover, J.; Harvey, J. N.; Fey, N.
Dalton Trans. 2010, 39, 10833.
(29) Bakhmutov, V. I. Practical NMR Relaxation for Chemists; Wiley:
Chichester, U.K., 2004.
(38) Fey, N.; Ridgway, B. M.; Jover, J.; McMullin, C. L.; Harvey, J. N.
(30) Unexpectedly, the T1 times for both the cis and the trans
isomers of 1 were longer in the presence of Xantphos: 10.7 vs 4.0 s for
cis-1 and 6.3 vs 3.2 s for trans-1. It is hardly conceivable that Xantphos,
especially in such low amounts, could change dramatically the nature
of the medium, lowering its viscosity, which would result in faster
molecular reorientations of 1. The T1 effect observed could be more
plausibly rationalized by paramagnetic species that are generated from
1 in the absence of Xantphos, but not in its presence. The CF3−Ph
reductive elimination from 1 occurs at as low as room temperature,
albeit slowly (see above). In the presence of extra Xantphos, the
unstable Pd(0) byproduct of this reaction, [(Xantphos)Pd], is
efficiently stabilized in the form of [(Xantphos)2Pd] (eq 1). As a
result, only diamagnetic species are produced in the entire process. In
the absence of extra Xantphos, however, [(Xantphos)Pd] decomposes
(Scheme 2) to give rise to palladium colloids, which, unlike the bulk
metal, exhibit ferromagnetic or near-ferromagnetic behavior.31
Although these ferromagnetic species could not be detected by ESR,
nor (considering their low concentration) by the Evans method,32
their presence in the extra Xantphos-free solutions of 1 provides a
rationale for the shortened relaxation times for both cis-1 and trans-1
(Table 2).
Dalton Trans. 2011, 40, 11184.
(39) Dispersion effectscomputed with BLYP and B3PW91 employed
Grimme’s DFT-D3 setof parameters. See Grimme, S.; Antony, J.;
Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104.
(40) Considering the small magnitude of the Xantphos effect and the
kobs determination error, the difference between the current (eq 4) and
initial (Figure 7) concentration of Xantphos is ignored. This difference
is even less important for the experiments using Xantphos in 5-, 10-,
and 20-fold excess.
(41) Strictly speaking, our computational results indicate that k3 ≠ 0
(see Figure 5 and text above) and might be comparable to k1 and k5.
The calculations also show, however, that B (4) is 15.8−18.5 kcal
mol−1 higher in energy than A (cis-1) (Figure 4), suggesting that k−2
and k4 are larger than k1, k3, and k5. Therefore, the general kinetic
equation (eq 4) may be simplified for two cases, as shown in the
Supporting Information: (i) k2 and k−4 are comparable to k1, k3, and k5
(eq S1, Supporting Information) and (ii) k2 and k−4 are smaller than
k1, k3, and k5 (eq S2, Supporting Information). Note that eq S2 is
identical with eq 5 (Case 2).
(42) (a) There is another case that is kinetically indistinguishable
from Cases 1 and 2: C does not produce PhCF3 but decomposes via a
different pathway during the reaction (Scheme S3, Supporting
Information). The rate law for this scenario is presented in eq S5
(Supporting Information). This case is implausible, however, because
of the high selectivity of the reaction that translates into the rate
(31) (a) Taniyama, T.; Ohta, E.; Sato, T. Europhys. Lett. 1997, 38,
195. (b) Hori, H.; Teranishi, T.; Nakae, Y.; Seino, Y.; Miyake, M.;
Yamada, S. Phys. Lett. A 1999, 263, 406. (c) Shinohara, T.; Sato, T.;
Taniyama, T. Phys. Rev. Lett. 2003, 91, 197201. (d) Sampedro, B.;
1327
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328

Organometallics
Article
constants kobs derived from the appearance of PhCF3 and from the
disappearance of 1 being identical within the determination error. (b)
Models omitting B, such as direct associative conversion of A to C in
the presence of Xantphos (Scheme S1, Supporting Information) and,
possibly, cis−trans isomerization of C (Scheme S2, Supporting
Information), are not considered. The rate laws for these models
(eqs S3 and S4, Supporting Information) include [XP] only in the
numerators and not in the denominators, thereby failing to agree with
the experimentally observed Xantphos effect.
(43) Naumann, D.; Kirij, N. V.; Maggiarosa, N.; Tyrra, W.;
Yagupolskii, Y. L.; Wickleder, M. S. Z. Anorg. Allg. Chem. 2004, 630,
746.
(44) Asensio, G.; Cuenca, A. B.; Esteruelas, M. A.; Medio-Simon, M.;
Olivan, M.; Valencia, M. Inorg. Chem. 2010, 49, 8665.
(45) Grushin, V. V. Organometallics 2000, 19, 1888.
(46) Binstead, R. A.; Zuberbuhler, A.; Jung, B. SPECFIT, [3.0.34];
̈
2003.
(47) Frisch, M. J.; et al. Gaussian 09, revision A.02; Gaussian Inc.:
Wallingford, CT, 2009.
(48) Grimme, S. J. Comput. Chem. 2006, 27, 1787.
(49) Andrae, D.; Haussermann, U.; Dolg, M.; Stoll, H.; Preuss, H.
Theor. Chim. Acta 1990, 77, 123.
(50) Hollwarth, A.; Bohme, M.; Dapprich, S.; Ehlers, A. W.; Gobbi,
̈
̈
A.; Jonas, V.; Kohler, K. F.; Stegmann, R.; Veldkamp, A.; Frenking, G.
̈
Chem. Phys. Lett. 1993, 208, 237.
(51) (a) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972,
56, 2257. (b) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta 1973, 28,
213.
(52) Haller, L. J. L.; Page, M. J.; Erhardt, S.; Macgregor, S. A.;
̈
Mahon, M. F.; Naser, M. A.; Velez, A.; Whittlesey, M. K. J. Am. Chem.
Soc. 2010, 132, 18408.
1328
dx.doi.org/10.1021/om200985g | Organometallics 2012, 31, 1315−1328