Fluxionality of [(Ph3P)3Rh(X)]: The extreme case of X = CF3

Full text of DOI:10.1021/ja9005699

Published on Web 03/09/2009
Fluxionality of [(Ph3P)3Rh(X)]: The Extreme Case of X ) CF3
Jenni Goodman,‡ Vladimir V. Grushin,*,† Roman B. Larichev,† Stuart A. Macgregor,*,‡
William J. Marshall,† and D. Christopher Roe†
Central Research & DeVelopment, E. I. DuPont de Nemours & Co., Inc., Experimental Station, Wilmington,
Delaware 19880, and School of Engineering and Physical Sciences, William H. Perkin Building, Heriot-Watt
UniVersity, Edinburgh EH14 4AS, U.K.
Received January 23, 2009; E-mail: vlad.grushin-1@usa.dupont.com; s.a.macgregor@hw.ac.uk
Complexes of the type [(R3P)3Rh(X)] are ubiquitous in organo-
metallic chemistry and catalysis, the most renowned and widely
used member of the family being Wilkinson’s catalyst,
[(Ph3P)3Rh(Cl)].1 In the solid state, all of these species are square-
planar. While the vast majority retain the same rigid structure in
solution, a small number of [(R3P)3Rh(X)] complexes (X ) H, Me,
Ph) display equivalency of all three phosphine ligands.2-5 This
unusual phenomenon has received surprisingly little attention in
spite of its importance not only for basic science but also for the
development of new applications in synthesis and catalysis. In one
case, [(Ph3P)3Rh(H)], the contrast between a square-planar geometry
in the solid state and rapid intramolecular exchange in solution has
been clearly established.4 The mechanism of this exchange,
however, remains unknown, and it is still unclear how and why
certain anionic ligands X promote the rearrangement whereas others
confer stereochemical rigidity. Herein we report experimental and
computational studies on a novel, uniquely fluxional complex,
[(Ph3P)3Rh(CF3)], which maintains phosphine exchange at -100 °C.
Our studies on this and related systems not only clarify the
mechanism of the peculiar phosphine rearrangement in
[(R3P)3Rh(X)] but also provide new critical insights into the long-
puzzling nature of bonding in perfluoroalkyl metal complexes.
Treating [(Ph3P)3Rh(F)]6 with CF3SiMe3 in benzene resulted in a
dark-red solution, from which uniformly shaped red-orange crystals
were isolated upon addition of hexanes. X-ray analysis of the crystals
(twice) revealed the structure of the product as trans-
[(Ph3P)2Rh(CF2)(F)] (1; Figure 1), a difluorocarbene fluoride complex.
The formation of 1 is likely mediated by [(Ph3P)3Rh(CF3)] (2), which
undergoes R-F-elimination upon loss of one phosphine (eq 1).7-9
This allowed for high-yield (84%) isolation and full characterization
of 2, including X-ray analysis (Figure 1).
Figure 1. ORTEP drawings of 1 (left) and 2 (right). Selected bond distances
(Å) and angles (deg) for 1: Rh-C, 1.820(3); Rh-F1, 1.994(2); Rh-P2,
2.335(1); Rh-P1, 2.349(1); C-Rh-F1, 177.0(1); P2-Rh-P1, 169.0(1);
F3-C-F2, 100.0(2); F3-C-Rh, 130.8(2); F2-C-Rh, 129.1(2). For 2: Rh-C,
2.096(2); Rh-P2, 2.310(1); Rh-P1, 2.313(1); Rh-P3, 2.337(1); F1-C,
1.376(2); F2-C, 1.398(2); F3-C, 1.380(2); C-Rh-P1, 163.5(1); P2-Rh-P3,
156.5(1); F1-C-F3, 101.3(1); F1-C-F2, 101.7(1); F3-C-F2, 101.6(1).
Sharp doublets of quartets in the 19F and 31P NMR spectra11 of
2 over the temperature range from 25 to -60 °C indicated fast
intramolecular10 ligand exchange. Only at -100 °C (THF-d8) was
the exchange slow enough to observe two broad 31P doublets at
33.5 (2P, JRh-P ) 175 Hz) and 30.5 ppm (1P, JRh-P ) 120 Hz).
Although temperatures below -100 °C were not attainable, and
hence activation parameters could not be determined, magnetization-
transfer experiments allowed for an exchange rate measurement of
12.1 s-1 at -100 °C.
The methyl and phenyl analogues of 2, [(Ph3P)3Rh(X)] (X )
Me, Ph), were studied by VT 31P NMR for comparison. The
reported6b room temperature 31P NMR “unsymmetrical doublet”
from the Ph species became symmetrical at 40 °C (JP-Rh ) 163
Hz), and below the coalescence point (around -20 °C), a complex
second-order spectrum was observed, indicating apparent stereo-
chemical rigidity.12 For [(Ph3P)3Rh(Me)], the coalescence temper-
ature was ∼20 °C, and a well-resolved first-order A2BX spectrum
could be observed already at -10 °C. Measurements of the
exchange rate over the temperature range from -10 to -50 °C
allowed for the determination of activation parameters: Ea ) 16.5
( 0.6 kcal mol-1, ∆Gq ) 12.9 kcal mol-1 (calculated at -30 °C),
∆Hq ) 16.0 ( 0.6 kcal mol-1, and ∆Sq ) 12.8 ( 2.3 e.u. Using
this ∆Sq value with the exchange rate of 12.1 s-1 measured for 2
at -100 °C (see above) led to an estimate of ∆Hq ≈ 11.3 kcal
mol-1 for 2 under the careful assumption of similar entropies of
activation for X ) Me and CF3.
While evidence was obtained for the same composition of 1 in
the solid state in bulk,10 upon dissolution in benzene, toluene, or
THF, preisolated 1 quickly equilibrated with a number of species,
including 2. A 19F and 31P variable-temperature (VT) NMR study
indicated a complex system comprising several compounds in
equilibrium, with some of the equilibria being fast and some slow
on the NMR time scale at 25 °C. The presence of 2 pointed to
phosphine dissociation from 1. As 1 contains only two phosphines
per Rh, we propose that isomers of dinuclear complexes
[(Ph3P)2Rh2(CF2)2(F)2] could be produced.10 Remarkably, however,
addition of excess PPh3 to this multicomponent solution efficiently
shifted all of the equilibria to 2 as the only NMR-detectable species.
The above results point to the uniquely high fluxionality of 2
among known [(R3P)3Rh(X)] complexes. For R ) Ph and X ) Ph,
Me, and H, the exchange rates are similar (e.g., 180 s-1 at -10 °C
for X ) Me10 and 230 s-1 at -13 °C for X ) H4) and considerably
lower than for X ) CF3 (12.1 s-1 at -100 °C). To account for the
† E. I. DuPont de Nemours & Co., Inc.
‡ Heriot-Watt University.
9
4236 J. AM. CHEM. SOC. 2009, 131, 4236–4238
10.1021/ja9005699 CCC: $40.75
2009 American Chemical Society

C O M M U N I C A T I O N S
facile exchange in 2, we turned to density functional theory
calculations13 to elucidate the mechanism of this process, and initial
studies focused on the model system [(H3P)3Rh(CF3)] (2′). Com-
puted bond lengths for 2′ agree well with those for 2, although the
distortion from square-planar geometry seen experimentally is not
reproduced because of the use of PH3 ligands (see below).
[(H3P)3Rh(Cl)] is consistent with the rigidity of Wilkinson’s
catalyst1 at ambient temperature.14 Of the full systems, the TS for
[(Ph3P)3Rh(Cl)] unfortunately could not be located, but low barriers
were found for X ) CF3, Me, Ph, and H, mirroring the fluxionality
of these experimental systems. An intermediate barrier of 19.4 kcal
mol-1 was computed for
X ) CN, and experimentally,
[(Ph3P)3Rh(CN)] is rigid at room temperature.15,16
Table 1. ∆Hqcalc (MP2//BP86, kcal mol-1) for [(R3P)3Rh(X)]
X
CF3
Me
Ph
H
CN
Cl
R ) H
R ) Ph
17.6
8.7
19.9
14.8
21.2
12.5
12.7
9.4
28.2
19.4
48.5
N/A
The steric factors that facilitate rearrangement in 2 are again
evident for all of the [(Ph3P)3Rh(X)] species in Table 1, with ∆Hqcalc
always being less than those for the [(H3P)3Rh(X)] congeners.
Likewise, while [(Ph3P)3Rh(Ph)] is fluxional at room temperature,
sterically less encumbered [(Me3P)3Rh(Ph)]5 and cis-[(Ph3P)2(Ph2PF)-
Rh(Ph)] are not.6b,17 The size of X in [(Ph3P)3Rh(X)] also plays an
important role, with the reduction in barrier height upon introduction
of the PPh3 ligands being largest for X ) CF3 and Ph (∼9 kcal mol-1),
less for Me (5.1 kcal mol-1), and smallest for H (3.3 kcal mol-1).
Steric effects alone, however, do not account for all of the
observations: CF3 and Br are approximately the same size, as are
Me and Cl, and yet [(Ph3P)3Rh(Cl)] and [(Ph3P)3Rh(Br)] are rigid
in solution1 under conditions where [(Ph3P)3Rh(Me)] and 2 are
fluxional. Electronic factors are also important, and it is significant
that the previously reported fluxional [(Ph3P)3Rh(X)] systems (X
) Me, H, Ph) all feature ligands with strong trans influence.18 CF3
is also considered to be a strong trans-influence ligand,19 and this
is confirmed by the Rh-P1 distance of 2.31 Å in 2, which is as
long as that measured in [(Ph3P)3Rh(H)].4,10 Figure 2 shows that
as CF3 moves into an axial position trans to a developing vacant
site, the Rh-C bond shortens considerably, implying greater
donation to the metal center. Similar changes are seen for all X,
but one would expect the greater donor ability characteristic of
ligands with the strongest trans influence to offer the greatest
stabilization and thus produce lowest barriers. Weaker trans-
influence ligands such as X ) Cl are unable to stabilize the TS
sufficiently, so intramolecular phosphine exchange by this mech-
anism becomes inaccessible.
Validation of these ideas comes from computed natural atomic
charges for the reactant and TS structures (given in Table 2 for the
simplified [(H3P)3Rh(X)] models). TS formation entails increased
donation from the donor atom of X of ∼0.2e. Despite this, the
electron density on Rh actually diminishes in the TS, and it is the
P centers that ultimately receive the extra charge density. This can
be understood if the trigonal TS is considered to be a fragment of
a TBP. Such a structure has two occupied d orbitals with significant
σ-antibonding interactions with the phosphine ligands (consistent
with increased Rh-P distances in both TS(2′-I′)1 and I′ relative to
2′, Figure 2). These d orbitals are approximately σ-nonbonding in
the square-planar reactant, and thus TS formation entails a
delocalization of electron density onto the P centers, each typically
receiving ∼0.15e.20 The loss of electron density at Rh can be
mitigated by strong donation from X, and the highest charge density
at Rh is found when X ) CF3, Me, H, and Ph, consistent with the
strong trans influence of these ligands. The smaller charges on Rh
in the TS for X ) CN (despite its strong trans influence15,18) and
especially that for X ) Cl correlate with the higher barriers
computed in these cases. Judging only by the strongest negative
charge on Rh (Table 2), one would expect the hydride
[(Ph3P)3Rh(H)] to be the most fluxional species in the series. This
Figure 2. Computed reaction profile (kcal mol-1) for intramolecular
phosphine exchange in 2′ with selected distances (Å) and angles (deg).
Structures 2′1/2′2 and TS(2′-I′)1/TS(I′-2′)2 are structurally equivalent, with
the subscript indicating the P center trans to the CF3 group. Energies shown
in italics were computed at the MP2//BP86 level (see the text for details).
The reaction profile for phosphine exchange was investigated
by reducing the trans C-Rh-P1 angle in 2′1 (i.e., with CF3 initially
trans to P1). This led to transition state TS(2′-I′)1 (E ) +12.7 kcal
mol-1), in which CF3 lies above the coordination plane (P1-Rh-C
) 105°) and the cis phosphines move toward the vacant site
(P2-Rh-P3 ) 130°). TS(2′-I′)1 links to intermediate, I′ (E )
+11.5 kcal mol-1) which resembles a trigonal bipyramid (TBP)
with a vacant axial site trans to CF3. The near-C3V geometry of I′
means that three equivalent TS structures can be accessed by
increasing the relevant C-Rh-P angles: TS(2′-I′)1 returns CF3 trans
to P1, while TS(2′-I′)2 and TS(2′-I′)3 place CF3 trans to P2 (see
Figure 2) and P3, respectively. TS(2′-I′)1/2/3 are therefore high points
on an energy surface that equilibrates all three phosphine ligands
in 2′ with (for the BP86 functional) ∆Hqcalc ) 12.7 kcal mol-1
.
Calculations on the full [(Ph3P)3Rh(CF3)] system indicate a very
similar topology for the phosphine-exchange surface. However, a
considerable distortion away from square-planar is now computed,10
which is more in accord with the experimental data (Figure 1).
The reactant is thus distorted toward the TS geometry, and a much-
reduced barrier of only 4.0 kcal mol-1 is calculated for the full
system. The bulkier PPh3 ligands therefore facilitate exchange, but
∆Hqcalc is significantly less than the approximate experimental value
of 11.3 kcal mol-1. This result was independent of the choice of
functional and basis set, but an improved value of 8.7 kcal mol-1
was obtained when the energies of the BP86-optimized species were
recomputed at the MP2 level. The improved performance of the
MP2//BP86 method was confirmed for [(Ph3P)3Rh(Me)] (∆Hcqalc
)
14.8 kcal mol-1 vs 16.0 ( 0.6 kcal mol-1 from experiment), so
this approach was used subsequently.
Table 1 gives ∆Hqcalc for a range of [(R3P)3Rh(X)] species, and
these show good agreement with observed trends in fluxionality.
In particular, the very high value of 48.5 kcal mol-1 computed for
9
J. AM. CHEM. SOC. VOL. 131, NO. 12, 2009 4237

C O M M U N I C A T I O N S
is not the case, however, possibly because of the steric effect (see
above) that is counter-directing and particularly significant for the
small H ligand. Entropic effects may also differ in this case.16
Acknowledgment. This is DuPont CRD Contribution No. 8901.
J.G. thanks Heriot-Watt University for support.
Supporting Information Available: Experimental and computa-
tional details and X-ray analysis data (CIF) for 1 (two polymorphs), 2,
and [(Ph3P)2Rh(CO)(F)] and complete ref 13. This material is available
free of charge via the Internet at http://pubs.acs.org.
Table 2. Selected Computed Natural Atomic Charges (q) for
[(H3P)3Rh(X)] Reactants and Transition Statesa
reactants
transition states
X
q(Rh)
q(CR/X)
q(Pav)
q(Rh)
q(CR/X)
q(Pav)
References
CF3
Me
Ph
H
CN
Cl
-0.52
-0.48
-0.48
-0.63
-0.49
-0.45
+0.79
-0.96
-0.22
-0.09
-0.02
-0.52
+0.23
+0.23
+0.24
+0.22
+0.25
+0.25
-0.26
-0.24
-0.20
-0.45
-0.17
-0.08
+0.93
-0.73
+0.03
+0.14
+0.14
-0.33
+0.08
+0.10
+0.08
+0.10
+0.10
+0.10
(1) Jardine, F. H. Prog. Inorg. Chem. 1981, 28, 63.
(2) (a) Keim, W. J. Organomet. Chem. 1967, 8, P25. (b) Keim, W. J.
Organomet. Chem. 1968, 14, 179.
(3) Dewhirst, K. C.; Keim, W.; Reilly, C. A. Inorg. Chem. 1968, 7, 546.
(4) Strauss, S. H.; Diamond, S. E.; Mares, F.; Shriver, D. F. Inorg. Chem.
1978, 17, 3064.
(5) Price, R. T.; Andersen, R. A.; Muetterties, E. L. J. Organomet. Chem. 1989,
376, 407.
a Other centers display only minor changes in computed charge.10
(6) (a) Grushin, V. V.; Marshall, W. J. J. Am. Chem. Soc. 2004, 126, 3068.
(b) Macgregor, S. A.; Roe, D. C.; Marshall, W. J.; Bloch, K. M.;
Bakhmutov, V. I.; Grushin, V. V. J. Am. Chem. Soc. 2005, 127, 15304.
(7) Brothers, J. P.; Roper, W. R. Chem. ReV. 1988, 88, 1293.
(8) For recent well-defined examples, see: (a) Huang, D.; Koren, P. R.; Folting,
K.; Davidson, E. R.; Caulton, K. G. J. Am. Chem. Soc. 2000, 122, 8916.
(b) Huang, D.; Caulton, K. G. J. Am. Chem. Soc. 1997, 119, 3185.
(9) The formation of 1 and 2 upon treatment of [(Ph3P)3RhCl] with [M(CF3)2]
(M ) Cd, Hg) has been proposed, but neither complex was detected because
of facile hydrolysis leading to Rh carbonyl species. See: Burrell, A. K.;
Clark, G. R.; Jeffrey, J. G.; Rickard, C. E. F.; Roper, W. R. J. Organomet.
Chem. 1990, 388, 391.
Our computed data also allow us to address the factors underpin-
ning the strong trans influence of the CF3 ligand.19,21 Ligand trans
influence is thought to be predominantly or exclusively controlled
by field effects,22 so a strong trans influence implies strong electron
donation. In complete contrast, the CF3 group is widely recognized
in organic chemistry as a powerful electron acceptor.23
The computed charges in Table 2 confirm that the Rh atom in
2′ does indeed bear a large negative charge (-0.52). This actually
exceeds that in the CH3 analogue (-0.48), in spite of the opposite
strong charges on the carbon atoms of the CF3 (+0.79) and the
CH3 (-0.96) ligands (Figure 3). In fact, the charge distribution in
2′ parallels the long-known24 ꢀ-effect in fluorinated organic
molecules: substitution of F for H on a carbon atom increases the
negative charge on the next C (or H) atom. The explanation of the
ꢀ-effect in terms of π-donation from the F atoms24a,c accounts for
the apparent flow of electrons toward the metal center in 2′ (Figure
3). The charge data therefore shed light on the previously perplexing
strong trans influence of CF3 and other perfluoroalkyl ligands.19,21
Furthermore, an electrostatic attraction resulting from such
(10) See the Supporting Information for details.
(11) NMR data for 2 (C6D6, 20 °C, δ): 19F:-2.4 (dq, JF-P ) 32 Hz, JF-Rh
11.5 Hz). 31P: 29.4 (dq, JP-F ) 32 Hz, JP-Rh ) 158 Hz).
)
(12) Second-order 31P NMR spectra have been previously observed for similar
Rh(I) σ-aryls [(Me3P)3Rh(Ar)] (Ar ) Ph, Tol).5
(13) Frisch, M. J.; et al. Gaussian 03, revision D.01; Gaussian, Inc.: Pittsburgh,
PA, 2001. Calculations used the BP86 functional. Rh, P, and Cl centers
were described with SDD RECPs and basis sets, with extra polarization
functions on P and Cl; 6-31G** basis sets were used for all other atoms.
Quoted energies include a correction for zero-point energies.10
(14) Triplet [(H3P)3Rh(Cl)] (E ) +22.7 kcal mol-1) is more stable than the
singlet TS. This represents the lower limit for the barrier to phosphine
exchange via a spin-crossover mechanism but is still too high to be accessed
at room temperature. Triplet [(H3P)3Rh(CF3)] (E ) +27.4 kcal mol-1) is
much higher in energy than singlet TS(2′-I′) and I′.
(15) Fernandes, M. A.; Circu, V.; Weber, R.; Varnali, T.; Carlton, L. J. Chem.
Crystallogr. 2002, 32, 273.
(16) Relating ∆Hqcalc to fluxionality rates assumes similar entropic changes for
all X. Entropy changes are difficult to address accurately in the present
calculations because of the absence of solvent effects. Generally, ∆Scqalc
was a small positive value, possibly because of the greater space available
to the PPh3 ligands in the TS. One exception, however, was [(Ph3P)3Rh(H)],
where a small negative value of ∆Sqcalc was computed.10
Mδ- δ+CF3 polarization could contribute to the M-C bond
-
shortening in related polyfluoroalkyl complexes19,21 compared to
their nonfluorinated counterparts.25 The positive charge on the
carbon atom, however, is probably stabilized by the same pπ back-
donation mechanism from the fluorines.24,25b,c
(17) Macgregor, S. A.; Wondimagegn, T. Organometallics 2007, 26, 1143.
(18) Appleton, T. G.; Clark, H. C.; Manzer, L. E. Coord. Chem. ReV. 1973, 10,
335.
(19) For reviews, see: (a) Hughes, R. P. AdV. Organomet. Chem. 1990, 31, 183.
(b) Morrison, J. A. AdV. Organomet. Chem. 1993, 35, 211.
(20) Albright, T. A.; Burdett, J. K.; Whangbo, M.-H. Orbital Interactions in
Chemistry; John Wiley & Sons: New York, 1985.
(21) For discussions of the trans influence and trans effect of CF3 and other
polyfluoroalkyl ligands, see: (a) Bennett, M. A.; Chee, H.-K.; Robertson,
G. B. Inorg. Chem. 1979, 18, 1061. (b) Bennett, M. A.; Chee, H.-K.; Jeffery,
J. C.; Robertson, G. B. Inorg. Chem. 1979, 18, 1071. (c) Hughes, R. P.;
Meyer, M. A.; Tawa, M. D.; Ward, A. J.; Williamson, A.; Rheingold, A. L.;
Zakharov, L. N. Inorg. Chem. 2004, 43, 747. (d) Grushin, V. V.; Marshall,
W. J. J. Am. Chem. Soc. 2006, 128, 4632.
Figure 3. Natural atomic charges computed for [(H3P)3Rh(CH3)] and 2′10
and a resonance structure accounting for the ꢀ-effect.
(22) Landis, C. R.; Firman, T. K.; Root, D. M.; Cleveland, T. J. Am. Chem.
Soc. 1998, 120, 1842.
(23) See, for example: Uneyama, K. Organofluorine Chemistry; Blackwell:
Oxford, U.K., 2006.
In conclusion, the mechanism of the unusual intramolecular ligand
exchange in complexes of the type [(R3P)3Rh(X)] has been elucidated
by experimental and computational methods. The rearrangement occurs
via a distorted trigonal TS with the anionic ligand X in an axial position
trans to a vacant site. Our results explain why certain [(R3P)3Rh(X)]
complexes (X ) Cl,1 Br, I, F,6 CN,15 OR,26 NR227) are stereochemi-
cally rigid in solution, whereas others (X ) Alk,2,5 Ar,2,6b H3,4) are
fluxional under similar conditions. Exchange is governed by a
combination of steric and electronic factors and is facilitated by bulkier
ligands on the Rh as well as by strongly donating anionic ligands X
that can stabilize the TS.28 Our studies of the most fluxional new
complex 2 (X ) CF3) have also clarified the previously puzzling strong
trans influence and trans effect of Rf ligands as well as the nature of
the bonding in perfluoroalkyl complexes of late transition metals.
(24) (a) Pople, J. A.; Gordon, M. J. Am. Chem. Soc. 1967, 89, 4253. (b) Davis,
D. W.; Banna, M. S.; Shirley, D. A. J. Chem. Phys. 1974, 60, 237. (c)
Holmes, S. A.; Thomas, T. D. J. Am. Chem. Soc. 1975, 97, 2337.
(25) (a) Notably, the overall charge on the CF3 ligand (-0.30) is roughly the
same as on the CH3 group (-0.29; Figure 3). Hence, the CF3 ligand should
not be viewed as carbocationic.25b,c (b) Christe, K. O.; Hoge, B.; Boatz,
J. A.; Prakash, G. K. S.; Olah, G. A.; Sheehy, J. A. Inorg. Chem. 1999,
38, 3132. (c) Christe, K. O.; Zhang, X.; Bau, R.; Hegge, J.; Olah, G. A.;
Prakash, G. K. S.; Sheehy, J. A. J. Am. Chem. Soc. 2000, 122, 481.
(26) (a) Kuznetsov, V. F.; Yap, G. P. A.; Bensimon, C.; Alper, H. Inorg. Chim.
Acta 1998, 280, 172. (b) Osakada, K.; Ishii, H. Inorg. Chim. Acta 2004,
357, 3007. (c) Zhao, P.; Incarvito, C. D.; Hartwig, J. F. J. Am. Chem. Soc.
2006, 128, 3124.
(27) Zhao, P.; Krug, C.; Hartwig, J. F. J. Am. Chem. Soc. 2005, 127, 12066.
(28) No evidence has been obtained for substantial contributions of the proposed
H-bonding interactions of the Rh complexes with solvent molecules. See:
Carlton, L. Magn. Reson. Chem. 2004, 42, 760.
JA9005699
9
4238 J. AM. CHEM. SOC. VOL. 131, NO. 12, 2009