Bond energy M-C/H-C correlations: Dual theoretical and experimental approach to the sensitivity of M-C bond strength to substituents

Article abstract of DOI:10.1039/b210036n

DFT methods are used to quantify the relationship between M-C and H-C bond energies; for ML_n = Re(η⁵-C₅H₅)(CO)₂H and fluorinated aryl ligands, theoretical and experimental investigations of ortho-fluorine substitution indicate a much larger increase in the M-C than in the H-C bond energy, so stabilising C-H activation products.

Full text of DOI:10.1039/b210036n

Bond energy M–C/H–C correlations: dual theoretical and experimental
approach to the sensitivity of M–C bond strength to substituents†
Eric Clot,a Maria Besora,b Feliu Maseras,b Claire Mégret,a Odile Eisenstein,*a Beatriz Oelckersc and
Robin N. Perutz*d
a LSDSMS (UMR 5636), Université Montpellier 2, 34095 Montpellier, France.
E-mail: eisenst@univ-montp2.fr
b Unitat de Química Física, Universitat Autònoma de Barcelona, 08193 Bellaterra, Catalonia, Spain
c
Instituto de Química, Universidad Católica de Valparaíso, Casilla 4059, Valparaíso, Chile
d Department of Chemistry, University of York, York, UK YO10 5DD. E-mail: rnp1@york.ac.uk
Received (in Cambridge, UK) 11th October 2002, Accepted 23rd December 2002
First published as an Advance Article on the web 24th January 2003
5
DFT methods are used to quantify the relationship between
M–C and H–C bond energies; for MLn
Re(h5-
C5H5)(CO)2H and fluorinated aryl ligands, theoretical and
experimental investigations of ortho-fluorine substitution
indicate a much larger increase in the M–C than in the H–C
bond energy, so stabilising C–H activation products.
Experiments on formation of Rh(h -C5R5)(PMe3)(C6H3F2)H
with R = H and Me by C–H activation, demonstrated a
thermodynamic preference for fluorine to be ortho to rhodium.6
Recently, we have studied the C–H activation reaction of
=
5
several fluoroarenes by {Re(h -C5Me5)(CO)2} generated pho-
5
5
tochemically from Re(h -C5Me5)(CO)3 and Re(h -C5Me5)-
(CO)2(N2) [eqn. (2)].7 The observed product is either the
Functionalisation of alkanes or arenes through an oxidative
addition/reductive elimination pathway requires the cleavage of
the strong C–H bond to give a hydrido–alkyl or a hydrido–aryl
intermediate. C–H activation reactions are often very selective
for the strongest C–H bonds because of the formation of even
stronger metal–carbon bonds.1 Here we demonstrate the great
sensitivity of M–C(aryl) bond dissociation energies (BDE) to
fluorine substituents on the bound aryl ring. Moreover, we show
that DFT methods offer a predictive way of probing bond
energy correlations that is especially valuable considering the
dearth of experimental data. BDEs in transition metal com-
plexes are hard to determine by experiment.2 Often, relative
values are obtained for a group of similar complexes through
competitive kinetic and/or equilibrium studies [eqn. (1)].
(2)
2
hydrido-aryl complex or the h -arene adduct, depending on the
number of F atoms on the phenyl ring. As the number of fluorine
substituents increases, the C–H activation product appears to
become more favoured.
To gain a better understanding of the reactivity of the
fluoroarenes, we have undertaken a systematic study of the
hydrido–aryl derivatives for all the possible substitution
patterns. The complexes Re(Cp)(CO)2(H)(C6H52nFn) (n =
5
0–5, 20 complexes in total, Cp = h -C5H5, see ESI Table S2†)
and associated C6H62nFn were optimised by DFT methods
(B3PW91 functional, see ESI†). To estimate the Re–C BDE
[D(Re–C)] and H–C BDE [D(H–C)]), the fragments Re-
(Cp)(CO)2(H)· and C6H52nFn· were optimised as spin doublets
with unrestricted methods.
(1)
Bryndza, Bercaw et al.3 introduced correlations of relative
BDE values for LnM–X [DD(M–X)] with absolute thermo-
chemical BDE values for H–X [D(H–X)]. A correlation with
5
slope of 1.0 was obtained for MLn = Ru(h -C5Me5)(PMe3)2
In Fig. 1, calculated DD(Re–C) values (relative to the case n
= 0) are plotted against the corresponding calculated D(H–C)
values. A very good correlation is obtained (r = 0.987) with a
slope of 2.25, larger than any observed previously. The method
of study has been validated by the agreement between the trends
in experimental and calculated BDE (H–C) for a wide range of
hydrocarbyl groups and for pentafluorophenyl (see ESI†). For
and Pt(Ph2PCH2CH2PPh2)Me with X = C-, N-, and O-based
ligands. The equilibrium constant for eqn.(1) is close to 1.0 and
the energies of bonds broken and formed are similar. Bergman,
5
Andersen et al. examined NHAr ligands bound to Ni(h -
C5Me5)(PEt3) and found a slope of 1.9.4 The greater sensitivity
of Ni–N bond energies to the aryl substituent was associated
with the ionic component of the Ni–N bond.
In the field of C–H bond activation by transition metal
complexes, a slope of the DD(M–C) vs. D(H–C) line greater
than 1.0 has been observed. A good correlation for DD(Ti–C)
vs. D(H–C) with slope of 1.36 was established for reaction of
hydrocarbons
with
(tBu3SiO)2Ti(NNSitBu3).5
For
TpARh(NCtBu) [TpA = hydridotris(3,5-dimethylpyrazolyl)bor-
ate] a plot of DD(Rh–C) vs. D(H–C) for R = hydrocarbyl
ligands revealed a slope of 1.22.1 Significant selectivity among
these ligands is achieved in bond formation with a strong
thermodynamic preference for M–aryl bonds.‡ The trends in
metal–carbon bond energies are not well understood and there is
great potential for contributions from theory.
† Electronic supplementary information (ESI) available: methods of
calculation; Fig. S1: Comparison of calculated and experimental C–H bond
dissociation energies for organic molecules; Table S1, comparison of
calculated and experimental CO-stretching frequencies; Table S2, total
energies, BDE for Re–C and H–C; Table S3, NPA charges q(C) and q(aryl)
for the organic fragments C6H62nFn and the rhenium complexes; prepara-
Fig. 1 Plot of bond dissociation energies showing DD(Re–C) vs. D(H–C)
(kJ mol21) for Re(Cp)(CO)2(H)(C6H52nFn) and H–C6H52nFn The value of
DD(Re–C) is taken as zero for n = 0. The slope is 2.25 and the correlation
is 0.987.
5
tion and spectroscopic data for Re(h -C5Me5)(CO)2(2,6-C6H3F2)H. See
http://www.rsc.org/suppdata/cc/b2/b210036n/
490
CHEM. COMMUN., 2003, 490–491
This journal is © The Royal Society of Chemistry 2003

the organometallic system, the calculations were checked for
consistency with crystallographic bond lengths and CO-
stretching frequencies (see ESI, Table S1†). We believe that this
is the first demonstration of such a correlation of calculated
BDE (M–X) with calculated BDE (H–X), a type of correlation
that is well established by experiment.
Fig. 1 demonstrates that the position of the fluorine
substituents, but not their total number, has a critical influence.
Substitution at the ortho position increases the energy of the
Re–C bond most, resulting in a step pattern. There are three
groups of molecules differentiated by the number (0 to 2) of
fluorine atoms at the ortho position.
Two opposing effects of ortho-fluorine substitution are found
to be responsible for the slope. The C–H bond increases in
strength with F-substitution as a consequence of the destabilis-
ing influence of fluorine on the radical C6H52nFn·. Thus the
ortho F-substituted C6H4F· is less stable by 10.3 kJ mol21 than
the meta-substituted isomer. The para-substituted species lies
only 3.7 kJ mol21 above the meta-substituted one (Fig. 2). The
F-substituent has a different effect on the Re complexes. The
ortho F-substituted Re(Cp)(CO)2(H)(C6H4F) is now the most
stable system whereas the meta and para substituted systems
follow the energy order of C6H4F·.
Fig. 3 Relative bond dissociation energies DD(Re–C) (kJ mol21) vs. total
NPA charge on the bound aryl fragment, 2q(aryl) (see ESI, Table S3†).
5
to the metal [cf.
d
283.8 and 2122.1 for Re(h -
C5Me5)(CO)2(2,5-C6H3F2)H]. The IR and mass spectra of the
product were also consistent with the identification. It was
2
striking that the h -coordination product was not observed.
A natural population analysis (NPA)8 of the total charge
distributions, q(aryl), on the aryl fragments (sum of all the
individual NPA charges for C6H52nFn) in the organometallic
and organic systems was carried out in order to probe the origin
of these results. Like the bond energy, the negative charge on
the aryl changes more rapidly in the organometallic than in the
organic system.§ A plot (Fig. 3) of DD(Re–C) as a function of
2q(aryl) shows a step pattern similar to Fig. 1 indicating that
the changes in bond energy are associated with changes in the
charge distribution. Furthermore, ortho fluorine substitution
dominates both the change in bond energy and the change in
q(aryl). These effects are reminiscent of the increase in bond
energy with Pauling electronegativity difference. The im-
portance of the electronegativity difference in influencing the
strength of the bond between the two partners has been
addressed by Labinger and Bercaw for metal–hydride and
metal–alkyl bonds.9 The increased participation of ionic
character in stabilising the M–C bonds agrees with the analysis
of their data by Bergman, Andersen and coworkers.4 Like these
authors, we cannot exclude a contribution from p bonding.
Fluorine substituents have been shown to have a much
greater influence on a M–aryl bond energy than on the
corresponding H–aryl bond energy. The total charge on the aryl
fragment appears to be an important factor suggesting that the
nature of the metal–ligand fragment, MLn, may influence the
slope. Preliminary calculations with other MLn fragments in
place of d6 {ReCp(CO)2} reveal similar correlations with slopes
varying between 2 and 3.
We have shown that DFT calculations are effective for
probing bond energy correlations BDE (M–X) vs. BDE (H–X)
for X = aryl. When F-substituents are added, H–C bond
energies increase, but not nearly as much as the corresponding
M–C bond energies. We associate this trend with the observa-
tion of C–H bond activation at {ReCp*(CO)2} with poly-
fluorinated benzenes but not with benzene.7,10 The maximum
effect calculated for F-substitution is found for two ortho-
fluorine substituents, leading to the expectation of selective C–
H activation that we have confirmed by experiment. The
methodology has potential for predictive applications to a wide
range of other BDE (M–X) vs. BDE (H–X) correlations.
We are grateful to EPSRC, The Royal Society, DURSI
(Catalonia, Spain) and CONICYT (Chile) for support.
5
In our previous work, we had shown that photolysis of Re(h -
C5Me5)(CO)2(N2) in liquid 1,4-difluorobenzene yielded a
5
2
mixture of Re(h -C5Me5)(CO)2(2,3-h -C6H4F2) and the ther-
5
modynamically less stable C–H activation product Re(h -
C5Me5)(CO)2(2,5-C6H3F2)H.7b In contrast, the reaction with
2,3,5,6-tetrafluorobenzene yielded only the C–H activation
product.7a The theoretical results described above led us to
predict that the reaction with 1,3-difluorobenzene should yield
the 2,6-C–H activation product since this species would benefit
from the stabilisation by two ortho-fluorines. Accordingly, we
Notes and references
‡ The contributions of kinetics and thermodynamics to product selectivity
are analysed in ref. 1.
§ q(aryl) in the organometallic system correlates linearly with q(aryl) in the
organic molecule (slope = 2.42, r = 0.991).
5
irradiated a sample of Re(h -C5Me5)(CO)2(N2) in liquid
1 (a) W. D. Jones and E. T. Hessel, J. Am. Chem. Soc., 1993, 115, 554; (b)
D. D. Wick and W. D. Jones, Organometallics, 1999, 18, 495.
2 J. A. Martinho-Simões and J.-L. Beauchamp, Chem. Rev., 1990, 90,
629.
3 H. E. Bryndza, L. K. Fong, R. A. Paciello, W. Tam and J. E. Bercaw, J.
Am. Chem. Soc., 1987, 109, 1444.
1,3-C6H4F2. The H and 19F NMR spectra at 253 K showed
1
85% conversion to a product readily identified as the expected
Re(h -C5Me5)(CO)2(2,6-C6H3F2)H (see ESI†). The 19F reso-
5
nance at low field, d 274.1, is characteristic of fluorines ortho
4 P. L. Holland, R. A. Andersen, R. G. Bergman, J. Huang and S. P.
Nolan, J. Am. Chem. Soc., 1997, 119, 12800.
5 J. L. Bennett and P. T. Wolczanski, J. Am. Chem. Soc., 1997, 119,
10696.
6 A. D. Selmeczy, W. D. Jones, M. G. Partridge and R. N. Perutz,
Organometallics, 1994, 13, 522.
7 (a) F. Godoy, C. L. Higgitt, A. H. Klahn, B. Oelckers, S. Parsons and R.
N. Perutz, J. Chem. Soc., Dalton Trans., 1999, 2039; (b) J. J. Carbó, O.
Eisenstein, C. L. Higgitt, A. H. Klahn, F. Maseras, B. Oelckers and R.
N. Perutz, J. Chem. Soc., Dalton Trans., 2001, 1452.
8 A. E. Reed, L. A. Curtis and F. Weinhold, Chem. Rev., 1988, 88,
899.
9 J. A. Labinger and J. E. Bercaw, Organometallics, 1988, 7, 926.
10 H. van der Heijden, A. G. Orpen and P. Pasman, J. Chem. Soc., Chem.
Commun., 1985, 1576.
Fig. 2 Influence of site of F substitution on the energies (kJ mol21) of
Re(Cp)(CO)2(H)(C6H4F) (left) and C6H4F· (right). The energies are based
on MLn + C6H5F as zero. The scales are drawn qualitatively.
CHEM. COMMUN., 2003, 490–491
491