Bridging Fluorides and Hard/Soft Mismatch in d6 and d 8 Complexes: The Case of [TI(μ-F)3Ru(PPh 3)3]

Article abstract of DOI:10.1021/ic034888e

[RuCl₂(PPh₃)₃] reacts with thallium(I) fluoride to give either [Tl(μ-F)₃Ru(PPh₃)₃] (1) or [Tl(μ₃-F)(μ₂-Cl)₂RU ₂(μ₂-Cl)-(μ₂-F)(PPh₃) ₄] (2) depending on the excess of TIF used. Both 1 and 2 were fully characterized, including X-ray structure determinations. Complex 1 reacts with dihydrogen to form the known ruthenium hydride complex [Ru-(H) ₂(H₂)(PPh₃)₃] upon hydrogenolysis of the Ru-F bond. The reaction of 1 with activated alkyl bromides (R-Br) gives the corresponding alkyl fluorides and the trinuclear complex [Tl(μ ₃-F)(μ₂-F)(μ₂-X)Ru₂(μ ₂-Br)(μ₂-F)(PPh₃)₄] (X = Br, F) (3), whose structure closely resembles that of 2. However, 1 is not active as catalyst for the nucleophilic fluorination of R-Br in the presence of thallium fluoride. The effect of the bridging coordination mode of fluoride on the Ru-F bond is discussed in terms of the HSAB principle, which suggests a more general model for predicting the stability of d⁶ and d⁸ complexes containing hard ligands (such as fluoro, oxo, and amido).

Full text of DOI:10.1021/ic034888e

Inorg. Chem. 2003, 42, 8417−8429
Bridging Fluorides and Hard/Soft Mismatch in d6 and d8 Complexes:
The Case of [Tl(µ-F)3Ru(PPh3)3]
Claus Becker, Iris Kieltsch, Diego Broggini,† and Antonio Mezzetti*
Department of Chemistry and Applied Biosciences, Swiss Federal Institute of Technology,
ETH Ho¨nggerberg, CH-8093 Zu¨rich, Switzerland
Received July 28, 2003
[RuCl2(PPh3)3] reacts with thallium(I) fluoride to give either [Tl(µ-F)3Ru(PPh3)3] (1) or [Tl(µ3-F)(µ2-Cl)2Ru2(µ2-Cl)-
(µ2-F)(PPh3)4] (2) depending on the excess of TlF used. Both 1 and 2 were fully characterized, including X-ray
structure determinations. Complex 1 reacts with dihydrogen to form the known ruthenium hydride complex [Ru-
(H)2(H )(PPh3)3] upon hydrogenolysis of the Ru−F bond. The reaction of 1 with activated alkyl bromides (R−Br)
2
gives the corresponding alkyl fluorides and the trinuclear complex [Tl(µ3-F)(µ2-F)(µ2-X)Ru2(µ2-Br)(µ2-F)(PPh3)4] (X
) Br, F) (3), whose structure closely resembles that of 2. However, 1 is not active as catalyst for the nucleophilic
fluorination of R−Br in the presence of thallium fluoride. The effect of the bridging coordination mode of fluoride on
the Ru−F bond is discussed in terms of the HSAB principle, which suggests a more general model for predicting
the stability of d6 and d8 complexes containing hard ligands (such as fluoro, oxo, and amido).
Introduction
of low-valent fluoro complexes of late transition metals is
dominated by the ionic character of the M-F bond, which
is an effect of the high electronegativity of fluorine. In fact,
d6 and d8 fluoro complexes are generally labile and highly
reactive toward nucleophiles, unless some kind of stabilizing
interaction is operative.
Our tentative explanation was that all factors that enhance
the electronegativity of the metal in the complex decrease
the electronegativity difference between M and F and
stabilize the M-F bond. This, admittedly very simplistic,
concept is based on Pauling’s electronegativity considerations
and accounts for the observation that stable d6 and d8 fluoro
complexes can be obtained by using strong π-accepting
coligands that deplete the electron density at the metal.
Examples are [MF2(CO)2(PPh3)2] (M ) Ru or Os)9 and
[RuH(F)(CO)(PR3)2].4b,10 Along the same lines, we have
suggested8 that the electron deficient nature of the metal
stabilizes the M-F bond of 16-electron, π-stabilized11
complexes such as [IrH2F(PBut2Ph)2]12 and [RuF(dppp)2]+
In contrast to the large number of known ruthenium chloro
complexes, there are relatively few literature reports concern-
ing fluoro complexes of ruthenium(II).1,2 In general, fluoro
complexes of late transition metals are interesting both from
a conceptual viewpoint (as examples of a combination of a
“soft” late transition metal with a “hard” ligand) and as
potential catalysts in organic transformations.3 In the dis-
cussions of the properties of d6 and d8 fluoro complexes,
there has been increasing consensus that fluoride acts as a
stronger π-donor toward the metal than the heavier halides.1,4-6
Also, as all orbitals with π-symmetry in d6 and d8 metals
are filled, the 4-electron filled/filled pπ/dπ repulsion dis-
cussed by Mayer7 and Caulton4 has been considered to play
a pivotal role in these complexes. In an alternative approach,
we have recently suggested8 that the structure and reactivity
* To whom correspondence should be addressed. E-mail: mezzetti@
inorg.chem.ethz.ch.
† X-ray structures.
(1) Doherty, N. M.; Hoffman, N. W. Chem. ReV. 1991, 91, 553.
(2) Murphy, E. F.; Murugavel, R.; Roesky, W. W. Chem. ReV. 1997, 97,
3425.
(3) Pagenkopf, B. L.; Carreira, E. M. Chem. Eur. J. 1999, 5, 3437.
(4) (a) Caulton, K. G. New J. Chem. 1994, 18, 25. (b) Poulton, J. T.;
Sigalas, M. P.; Folting, K.; Streib, W. E.; Eisenstein, O.; Caulton, K.
G. Inorg. Chem. 1994, 33, 1476.
(9) (a) Brewer, S. A.; Brisdon, A. K.; Holloway, J. H.; Hope, E. G.; Peck,
L. A.; Watson, P. G. J. Chem. Soc., Dalton Trans. 1995, 1073. See
also: (b) Smith, G.; Cole-Hamilton, D. J.; Gregory, A. C.; Goodman,
N. G. Polyhedron 1982, 1, 97.
(10) Poulton, J. T.; Sigalas, M. P.; Eisenstein, O.; Caulton, K. G. Inorg.
Chem. 1993, 32, 5490.
(5) Grushin, V. V. Chem. Eur. J. 2002, 8, 1006.
(6) Fagnou, K.; Lautens, M. Angew. Chem., Int. Ed. 2002, 41, 26.
(7) Mayer, J. M. Comments Inorg. Chem. 1988, 8, 125.
(8) Mezzetti, A.; Becker, C. HelV. Chim. Acta 2002, 85, 2686.
(11) (a) Rachidi, I. E.; Eisenstein, O.; Jean, Y. New. J. Chem. 1990, 14,
671. (b) Riehl, J. F.; Jean, Y.; Eisenstein, O.; Pe´lissier, M. Organo-
metallics 1992, 11, 729.
(12) Cooper, A. C.; Caulton, K. G: Inorg. Chim. Acta 1996, 251, 41.
10.1021/ic034888e CCC: $25.00 © 2003 American Chemical Society
Published on Web 10/30/2003
Inorganic Chemistry, Vol. 42, No. 25, 2003 8417

Becker et al.
(dppp ) 1,3-bis(diphenylphosphino)propane).13 The latter
species is Lewis acidic because of the overall positive charge
and the coordinative unsaturation of ruthenium.
Owing to its Lewis acidity, [RuF(dppp)2]+, which was
prepared by reaction of [RuCl(dppp)2]+ with TlF,13 is an
active fluorinating agent of alkyl halides R-X (R ) Cl, Br,
or I) both as a stoichiometric reagent and as catalyst in
combination with thallium(I) fluoride.14 With the less bulky
ligand 1,2-bis(diphenylphosphino)ethane (dppe), the reaction
of [RuCl(dppe)2]+ with TlF yields the Ru-F-Tl bridged
species [Tl(µ-F)2Ru(dppe)2]+ instead of the five-coordinate
species [RuF(dppe)2]+.14
In our screening of five-coordinate, 16-electron fluoro
complexes that might catalyze fluoride-transfer reactions, we
targeted the preparation of the yet unknown difluoro complex
[RuF2(PPh3)3]. The corresponding complexes [RuX2(PPh3)n]
(X ) H, Cl, Br; n ) 3, 4) catalyze a plethora of reactions15
and act as precursors for the synthesis of other ruthenium
complexes. As described below, the well-established strategy
based on the fluorination of chloro complexes of ruthenium-
(II) with thallium(I) fluoride yielded, instead, the title
compound [Tl(µ-F)3Ru(PPh3)3] (1), which is a formal TlF
adduct of the target complex [RuF2(PPh3)3].
The isolation of 1, as well as Perutz’s report of the fluoro-
bridged binuclear cation [Ru2(µ-F)3(PEt3)6]+,16 prompted us
to reassess our ideas concerning the M-F bonding in fluoro
complexes of late transition metals and to develop a general
model that could explain both metal-centered (push-pull
interactions and π-stabilization) and ligand-based factors, and
in particular the role that the coordination mode of the fluoro
ligand (terminal vs bridging) plays in the stabilization of d6
and d8 fluoro complexes. The ultimate goal is to develop a
general understanding of the factors that determine the
stability of d6 and d8 complexes containing hard ligands, such
as fluoro, oxo, and amido.17
Figure 1. Experimental and simulated 19F (upper) and 31P (lower) NMR
spectra of 1.
Reaction of [RuCl2(PPh3)3] with TlF. [RuCl2(PPh3)3] (500 mg,
521 µmol) and thallium fluoride (350 mg, 1.6 mmol, 3 equiv) were
suspended in dichloromethane (8 mL) in a Teflon reaction vessel
in a glovebox and stirred at room temperature for 28 h. Then, the
reaction mixture was filtered through a microfilter to remove the
precipitated thallium salts, and the solvent was evaporated affording
a dark red-brown solid. Crystals of [Tl(µ-F)3Ru(PPh3)3] (1) and
[Tl(µ3-F)(µ2-Cl)2Ru2(µ2-Cl)(µ2-F)(PPh3)4] (2) were isolated from
a mixed batch of yellow (1) and red (2) crystals obtained by slow
diffusion of pentane into a CDCl3 solution of the dark red-brown
solid.
Experimental Section
General Methods. All operations were carried out under purified
nitrogen in a glovebox by using Teflon reaction vessels and distilled
solvents. All reagents and solvents were Fluka puriss. grade or had
comparable purity. Dichlorotris(triphenylphosphine)ruthenium(II)
was prepared according to literature procedures.18 NMR spectra
were recorded on Bruker Avance 250 and 300 spectrometers.
Chemical shifts δ are in ppm relative to internal SiMe4 (1H), to
external 85% H3PO4 (31P), and to external CFCl3 (19F). NMR spectra
were run at room temperature (unless otherwise stated) and were
simulated with Swan-MR.19 Elemental analyses were measured by
the analytical service of the chemistry department of the ETH
Zu¨rich.
[Tl(µ-F)3Ru(PPh3)3] (1). [RuCl2(PPh3)3] (2.00 g, 2.07 mmol),
thallium fluoride (2.23 g, 10 mmol), and triphenylphosphine (2.63
g, 10 mmol) were suspended in dichloromethane (11 mL) in a
Teflon reaction vessel in a glovebox and stirred at room temperature
for 20 h. Then, the reaction mixture was filtered through a
microfilter to remove the precipitated thallium salts, and the solvent
was evaporated. The orange residue was recrystallized from
dichloromethane/pentane and dried in a vacuum to give 2 as a
yellow powder (1.45 g, 1.25 mmol, 60%). 31P{1H} NMR (CD2Cl2,
121 MHz), T ) 25 °C: δ 49.5 (br s, 3 P). At -20 °C (Figure 1):
δ 49.5 (MM′M′′ part of AA′A′′MM′M′′X spin system, JP,P′
)
-26.6, JP,F ) +201.4, JP,F′ ) -11.0, JP,F′′ ) -10.4, JTl,P ) +3.0,
3 P). 19F NMR (CD2Cl2, 282 MHz), T ) 25 °C: δ -251 (br). At
T ) -60 °C: δ -251 (AA′A′′ part of an AA′A′′MM′M′′X spin
system, JF,F′ ) -100.0, JTl,F ) +890.3, 3 F). Anal. Calcd for
C54H45F3P3RuTl: C, 56.43; H, 3.95, P, 8.08. Found: C, 56.42; H,
4.23; P, 8.05.
[Tl(µ3-F)(µ2-Cl)2Ru2(µ2-Cl)(µ2-F)(PPh3)4] (2). [RuCl2(PPh3)3]
(250 mg, 260 µmol) and thallium fluoride (29 mg, 130 µmol, 0.5
equiv) were suspended in dichloromethane (5 mL) in a Teflon
reaction vessel in a glovebox and stirred at room temperature for
20 h. Then, the reaction mixture was filtered through a microfilter
to remove the precipitated thallium salts yielding a dark red solution,
(13) (a) Barthazy, P.; Hintermann, L.; Stoop, R. M.; Woerle, M.; Mezzetti,
A.; Togni, A. HelV. Chim. Acta 1999, 82, 2448. (b) Barthazy, P.; Stoop,
R. M.; Woerle, M.; Togni, A.; Mezzetti, A. Organometallics 2000,
19, 2844.
(14) Barthazy, P.; Togni, A.; Mezzetti, A. Organometallics 2001, 20, 3472.
(15) Naota, T.; Takaya, H.; Murahashi, S. I. Chem. ReV. 1998, 98, 2599.
(16) Jasim, N. A.; Perutz, R. N.; Archibald, S. J. J. Chem. Soc., Dalton
Trans. 2003, 2184.
(17) Holland, P. L.; Andersen, R. A.; Bergman, R. G. Comments Inorg.
Chem. 1999, 21, 115.
(18) Hallman, P. S.; Stephenson, T. A.; Wilkinson, G. Inorg. Synth. 1970,
12, 237.
(19) Balacco, G. J. Chem. Inf. Comput. Sci. 1994, 34, 1235.
8418 Inorganic Chemistry, Vol. 42, No. 25, 2003

The Case of [Tl(µ-F)3Ru(PPh3)3]
Table 1. Crystal Data and Structure Refinement for 1-3
1
2
3
formula
C55H46Cl3F3-
P3RuTl
1268.62
0.74 × 0.60
× 0.14
hexagonal
P63
12.7484(18)
12.7484(18)
17.362(4)
90
90
120
2443.6(7)
2
1.724
C73H62Cl5F2-
P4Ru2Tl
1684.87
0.15 × 0.05
× 0.02
C73H61BrCl3F4-
P4Ru2Tl
1730.87
0.14 × 0.13
× 0.12
fw
cryst size (mm3)
cryst syst
space group
a (Å)
b (Å)
c (Å)
monoclinic
P21/c
monoclinic
P21/n
19.725(7)
14.541(6)
26.270(10)
90
110.084(12)
90
7077(5)
4
1.581
13.1385(5)
25.1451(9)
22.4359(7)
90
93.053(1)
90
7401.6(5)
4
1.553
R (deg)
â (deg)
γ (deg)
V (Å3)
Z
F
calcd (g/cm-3
)
abs coeff (mm-1
θ range (deg)
)
3.914
1.84-31.06
19984
3.018
2.12-26.41
44840
3.358
2.24-28.31
55113
no. reflns collcd
no. indep reflns
4808 [Rint
0.2066]
4808
)
14469 [Rint
0.0393]
10769
)
18386 [Rint
0.0833]
7416
)
no. obsd reflns
GOF on F2
R1 indices
[I > 2σ(I)]
wR2 indices
(all data)
0.988
0.0536
1.065
0.0507
0.842
0.0568
0.1420
0.1728
0.1653
largest peak/hole
5.037/-1.863
2.022/-1.347
1.281/-1.528
(e Å-3
)
(PPh3)4] (Figure S1) was assumed for the simulation of the 31P and
19F NMR spectra of 3 with an AA′BB′(MM′)2X spin system. The
broad 19F spectrum yielded only approximate values of JF,F′ (-150
Hz, (µ2-F)-Ru-(µ3-F)) and JTl,F (+2100 Hz, (µ2-F)-Tl) that were
not refined. 31P NMR (CDCl3, 121 MHz): δ 59.84 ((MM′)2 part,
JP,P′ ) -37.0, JP,F(trans) ) +169.6, JP,F(cis-P-Ru-(µ3-F)) )
+4.9, JP,F(P-Ru-(µ2-F)) ) -4.3, JF,F′((µ3-F)-Ru-(µ3-F)) )
-133.7, 4 P). 19F NMR (CDCl3, 282 MHz): δ -341.7 (AA′ part,
Figure 2. Experimental and simulated 31P NMR spectrum of 2.
whose 31P NMR spectrum showed the signals of 2 (34%), of
unreacted [RuCl2(PPh3)3] (36%), and of other unidentified products
(30%). Red crystals of 2‚CH2Cl2 were obtained by addition of
pentane to the dichloromethane solution. On the basis of the
equivalence of the bridging fluorides observed in the 19F NMR
spectrum (see Results and Discussion sections), the spin system
was assumed as AA′(MM′)2X (A ) F, M ) P, X ) Tl) assuming
a pseudo-C2V symmetry for the molecule (Figure 2). As the 19F
NMR spectrum is broad and featureless, only the 31P NMR
subspectrum was simulated. 31P{1H} NMR (CD2Cl2, 121 MHz):
δ 55.70 ((MM′)2 part of an AA′(MM′)2X spin system, JP,P′ ) -31.9,
JP,F ) +160.8, JP,F′ ) -11.5, 4 P, JF,F′ ) -115.8). 19F NMR
(CD2Cl2, 282 MHz): δ -346 (br, 2F). Anal. Calcd for C72H60-
Cl3F2P4Ru2Tl‚CH2Cl2: C, 52.04; H, 3.71. Found: C, 52.20; H, 4.13.
[Tl(µ3-F)(µ2-F)(µ2-X)Ru2(µ2-Br)(µ2-F)(PPh3)4] (3). Complex
1 (80 mg, 69.6 µmol) and bromodiphenylmethane (18.8 mg, 76.1
mmol) were dissolved in CH2Cl2 (3 mL) in a Teflon vessel in a
glovebox. After stirring at room temperature for 22 h, the reaction
mixture was filtered through a microfilter to remove the thallium
salts. CD2Cl2 was added to the red-brown solution, which was then
J
J
F,F′((µ2-F)-Ru-(µ3-F)) ) ca. -150, 2 µ3-F), -422.6 (BB′ part,
F,F′((µ2-F)-Ru-(µ2-F)) ) assumed as +100, no effect on spectral
appearance, 2 µ2-F). Anal. Calcd for C72H60Br1.39F3.61P4Ru2Tl: C,
52.91; H, 3.70. Found: C, 52.68; H, 4.24.
X-ray Studies. X-ray studies were performed at room temper-
ature on a Siemens SMART platform with CCD detector, nor-
mal focus molybdenum-target X-ray tube (λ ) 0.71073 Å), and
graphite monochromator by using ω-scans. Unit cell dimensions
determination and data reduction were performed by standard
procedures, and an empirical absorption correction (SADABS) was
applied for 1, but not for 2 and 3. The structures were solved
with SHELXS-96 using direct methods, and refined by full-
matrix least-squares on F2 with anisotropic displacement param-
eters for all non-H atoms. Hydrogen atoms were introduced at
calculated positions (except for disordered solvent molecules) and
refined with the riding model and individual isotropic thermal
parameters. Table 1 contains some crystallographic data; atomic
coordinates, anisotropic displacement coefficients, and an extended
list of interatomic distances and angles of 1-3, as well as the results
of a preliminary X-ray structure determination of a crystal from
the mixed sample of 1 and 2, are available as Supporting
Information.
1
analyzed by H, 19F, and 31P NMR spectroscopy. The conversion
of 1 was 67% (as determined by 31P NMR spectroscopy). Com-
plex [Tl(µ3-F)(µ2-F)(µ2-X)Ru2(µ2-Br)(µ2-F)(PPh3)4] (3) was ob-
tained as the only metal-containing product. Fluorodiphenylmethane
(Ph2CHF) was detected by 19F NMR spectroscopy (CH2Cl2/
CDCl3): δ -167.2 (lit. -169,20 d, 1F, JF,H ) 46.5 Hz, lit. 48,20
Ph2C(H)-F). 1H NMR (CDCl3): δ 6.34 (s, 1H, CBr-H, Ph2CHBr,
63%), 6.50 (d, 1H, JF,H ) 47.0 Hz, Ph2C(F)-H (Ph2CHF), 37%).
Red crystals of 3 formed upon diffusion of pentane into the
CH2Cl2/CDCl3 solution. The C2V-symmetric structure [TlRu2BrF4-
Yellow prisms of 1 were obtained (together with red crystals of
2, see below) by diffusion of pentane into a CDCl3 solution of 1
and 2. Octants collected: h, -13 to 18, k, -18 to 17, l, -24 to 24.
Max and min calculated transmission factors were 0.6103 and
0.1598. The Flack x parameter was -0.024(5). The crystal contains
one CDCl3 molecule as solvent of crystallization. Selected bond
(20) Lai, C.; Kim, Y. I.; Wang, C. M.; Mallouk, T. E. J. Org. Chem. 1993,
58, 1393.
Inorganic Chemistry, Vol. 42, No. 25, 2003 8419

Becker et al.
Table 2. Selected Bond Distances (Å) and Angles (deg) for
[Tl(µ-F)3Ru(PPh3)3] (1)
Table 4. Selected Bond Distances (Å) and Angles (deg) for
[Tl(µ3-F)(µ2-F)(µ2-X)Ru2(µ2-Br)(µ2-F)(PPh3)4] (3)
Tl-F
2.389(4)
2.3239(14)
2.925(5)
Ru-F
2.133(3)
3.3446(9)
2.09
Tl-F(4)
2.503(4)
Tl-F/Br(5)
2.804(9)/
2.989(3)
Ru-P
F‚‚‚C(2)
Ru‚‚‚Tl
F‚‚‚C(2)
Tl-F(1)
2.694(4)
2.153(4)
2.150(4)
2.476(12)
2.067(4)
Tl‚‚‚F(2)
2.843(4)
2.161(4)
2.168(4)
2.4754(12)
2.247(9)/
2.461(3)
2.254(2)
2.248(2)
3.7217(8)
2.15
Ru(1)-F(1)
Ru(1)-F(2)
Ru(1)-Br(3)
Ru(1)-F(4)
Ru(2)-F(1)
Ru(2)-F(2)
Ru(2)-Br(3)
Ru(2)-F/Br(5)
F-Tl-F′
P-Ru-P′
P-Ru-F
P-Ru-F′′
66.73(13)
97.79(5)
99.80(10)
85.47(10)
F-Ru-F′
Ru-F-Tl
P-Ru-F′
76.03(15)
95.25(12)
161.50(10)
Ru(1)-P(1)
2.244(2)
2.2711(19)
3.5316(7)
3.1396(9)
2.24
Ru(2)-P(3)
Table 3. Selected Bond Distances (Å) and Angles (deg) for
[Tl(µ3-F)Ru2Cl2(µ2-Cl)(µ2-F)(PPh3)4] (2)
Ru(1)-P(2)
Ru(2)-P(4)
Tl‚‚‚Ru(1)
Tl‚‚‚Ru(2)
Ru(1)‚‚‚Ru(2)
F(1)‚‚‚H-C(30)
F(1)‚‚‚H-C(62)
F(4)‚‚‚H-C(24)
F(2)‚‚‚H-C(8)
F(2)‚‚‚H-C(50)
F/Br(5)‚‚‚H-C(38)
F/Br(5)‚‚‚H-C(62)
Tl-Cl(4)
Tl-F(1)
3.0195(19)
2.668(3)
2.195(3)
2.211(3)
2.3785(16)
2.4130(16)
2.2677(16)
2.2766(15)
3.7835(12)
3.1909(11)
2.14
Tl-Cl(5)
Tl‚‚‚F(2)
2.9719(17)
3.101(3)
2.209(3)
2.242(4)
2.3709(16)
2.4038(16)
2.2848(17)
2.2686(17)
3.7783(11)
2.26
2.34
2.28/1.88
2.21/2.83
2.44
2.16
Ru(1)-F(1)
Ru(1)-F(2)
Ru(1)-Cl(3)
Ru(1)-Cl(4)
Ru(1)-P(1)
Ru(1)-P(2)
Tl‚‚‚Ru(1)
Ru(2)-F(1)
Ru(2)-F(2)
Ru(2)-Cl(3)
Ru(2)-Cl(5)
Ru(2)-P(3)
Ru(2)-P(4)
Tl‚‚‚Ru(2)
F(1)-Tl-F(4)
67.20(12)
F(1)-Tl-F/Br(5)
Ru(1)-Br(3)-Ru(2)
Tl-F(1)-Ru(2)
62.8(2)/
71.6(1)
78.73(4)
F(4)-Tl-F/Br(5)
123.2(2)/
125.3(1)
92.84(14)
Tl-F(1)-Ru(1)
Tl-F(4)-Ru(1)
99.49(14)
94.3(3)/
85.6(1)
93.28(15)
70.30(15)
80.47(11)
Ru(1)‚‚‚Ru(2)
F(1)‚‚‚H-C(62)
Cl(4)‚‚‚H-C(20)
F(2)‚‚‚H-C(8)
F(2)‚‚‚H-C(50)
Cl(5)‚‚‚H-C(38)
100.78(16) Tl-F/Br(5)-Ru(2)
2.14
2.50
2.67
Ru(1)-F(1)-Ru(2) 93.39(16)
F(1)-Ru(1)-F(2) 70.77(16)
Ru(1)-F(2)-Ru(2)
F(1)-Ru(2)-F(2)
F(1)-Ru(2)-Br(3)
F(1)-Tl-Cl(4)
Cl(4)-Tl-Cl(5)
Tl-F(1)-Ru(1)
68.76(7)
127.33(4)
101.71(12)
87.54(5)
92.84(11)
71.03(12)
77.78(9)
88.68(9)
162.45(9)
96.36(9)
F(1)-Tl-Cl(5)
Ru(1)-Cl(3)-Ru(2)
Tl-F(1)-Ru(2)
69.92(7)
84.42(5)
101.13(11)
88.66(5)
91.56(13)
70.19(12)
77.67(9)
88.41(9)
166.21(9)
92.46(9)
78.94(9)
89.17(9)
99.86(9)
162.61(9)
164.17(5)
91.26(6)
99.03(5)
100.34(6)
90.18(6)
97.44(6)
F(1)-Ru(1)-Br(3) 80.64(11)
F(1)-Ru(1)-F(4)
86.07(16)
F(2)-Ru(2)-F/Br(5) 87.6(3)/
82.3(1)
169.24(13)
91.75(12)
79.14(12)
Tl-Cl(4)-Ru(1)
Ru(1)-F(1)-Ru(2)
F(1)-Ru(1)-F(2)
F(1)-Ru(1)-Cl(3)
F(1)-Ru(1)-Cl(4)
F(1)-Ru(1)-P(1)
F(1)-Ru(1)-P(2)
F(2)-Ru(1)-Cl(3)
F(2)-Ru(1)-Cl(4)
F(2)-Ru(1)-P(1)
F(2)-Ru(1)-P(2)
Tl-Cl(5)-Ru(2)
Ru(1)-F(2)-Ru(2)
F(1)-Ru(2)-F(2)
F(1)-Ru(2)-Cl(3)
F(2)-Ru(2)-Cl(5)
F(1)-Ru(2)-P(3)
F(1)-Ru(2)-P(4)
F(2)-Ru(2)-Cl(3)
F(1)-Ru(2)-Cl(5)
F(2)-Ru(2)-P(3)
F(2)-Ru(2)-P(4)
Cl(3)-Ru(2)-Cl(5)
Cl(3)-Ru(2)-P(3)
Cl(3)-Ru(2)-P(4)
Cl(5)-Ru(2)-P(3)
Cl(5)-Ru(2)-P(4)
P(3)-Ru(2)-P(4)
F(1)-Ru(1)-P(1)
F(1)-Ru(1)-P(2)
165.67(12) F(1)-Ru(2)-P(3)
96.78(12)
F(1)-Ru(2)-P(4)
F(2)-Ru(2)-Br(3)
F(1)-Ru(2)-F/Br(5) 81.1(3)/
F(2)-Ru(1)-Br(3) 79.48(11)
F(2)-Ru(1)-F(4)
84.70(15)
94.91(13)
92.0(1)
99.36(13)
161.67(12)
F(2)-Ru(1)-P(1)
F(2)-Ru(1)-P(2)
Br(3)-Ru(1)-F(4) 161.98(12) Br(3)-Ru(2)-F/Br(5) 160.1(3)/
F(2)-Ru(2)-P(3)
167.12(13) F(2)-Ru(2)-P(4)
79.39(9)
90.13(9)
91.96(10)
166.92(10)
161.43(9)
94.89(9)
94.69(6)
Br(3)-Ru(1)-P(1) 96.55(6)
Br(3)-Ru(1)-P(2) 95.54(6)
F(4)-Ru(1)-P(1)
F(4)-Ru(1)-P(2)
P(1)-Ru(1)-P(2)
Br(3)-Ru(2)-P(3)
Br(3)-Ru(2)-P(4)
F/Br(5)-Ru(2)-P(3) 101.9(3)/
Cl(3)-Ru(1)-Cl(4) 164.96(5)
93.32(13)
98.08(12)
97.48(7)
Cl(3)-Ru(1)-P(1)
Cl(3)-Ru(1)-P(2)
Cl(4)-Ru(1)-P(1)
Cl(4)-Ru(1)-P(2)
P(1)-Ru(1)-P(2)
95.28(6)
94.84(5)
95.83(6)
93.11(6)
100.30(6)
89.4(1)
F/Br(5)-Ru(2)-P(4) 93.3(3)/
102.6(1)
P(3)-Ru(2)-P(4)
98.34(9)
(PPh3)3].21 1H NMR (CDCl3, 25 °C): δ ) -17.4 (q, JP,H ) 26 Hz,
1 H, Ru-H), 6.9-7.4 (m, 45 H, aromatic H). 31P{1H} NMR
(CDCl3, 25 °C): δ 58.5 (br, 3 P). H NMR (CDCl3, -60 °C): δ
-18.5 (dt, JP(eq),H ) 34 Hz, JP(ax),H ) 22 Hz, 1 H, Ru-H), 6.9-7.4
(m, 45 H, aromatic H). 31P{1H} NMR (CDCl3, -60 °C): δ 39.7
(d, JP,P′ ) 29 Hz, 1 P, equatorial P), 95.3 (t, JPP ) 29 Hz, 2 P,
axial P).
lengths and angles are given in Table 2. Complex 2 was identified
in a preliminary X-ray study carried out on a red crystal from the
reaction of [RuCl2(PPh3)3] with TlF (3 equiv), which yielded the
mixture of 1 and 2 mentioned above. The following X-ray analysis
of a red prism (obtained by diffusion of pentane into CH2Cl2)
obtained from the synthesis with 0.5 equiv of TlF confirmed the
identity of 2. All reported data refer to the latter crystal. Octants
collected: h, -21 to 24, k, -17 to 18, l, -32 to 27. Max and
min calculated transmission factors were 0.9477 and 0.6603. A
CH2Cl2 molecule was detected in three different sites with oc-
cupancies refined to 0.79(1), 0.43(1), and 0.36(1). Selected bond
lengths and angles are given in Table 3. Red prisms of 3 were
obtained by diffusion of pentane into a CH2Cl2/CDCl3 solution at
room temperature. Octants collected: h, -17 to 15, k, -27 to 33,
l, -29 to 19. Max and min calculated transmission factors were
0.6887 and 0.6507. The crystal contained one CDCl3 molecule as
solvent of crystallization. Selected bond lengths and angles are given
in Table 4.
1
Reaction of 1 with H2 in THF-d8. Compound 1 (20 mg, 17.4
µmol) was dissolved in THF-d8 (1 mL) in an NMR tube, and H2
gas was bubbled through the solution for 10 min, during which the
solution color changed from yellow to light brown. Small amounts
of a colorless solid (TlF) precipitated. The 1H and 31P NMR spec-
tra indicated the formation of [Ru(H)2(H2)(PPh3)3].22 1H NMR
(THF-d8, 25 °C): δ -7.57 (br s, 4 H, Ru-H), 6.91-7.22 (m,
45 H, aromatic H). 31P{1H} NMR (THF-d8, 25 °C): δ 56.8 (s, 3
1
P). H NMR (THF-d8, -100 °C): δ -7.57 (br s, 4 H, Ru-H),
(21) (a) Hallman, P. S.; Evans, D.; Osborn, J. A.; Wilkinson, G. J. Chem.
Soc., Chem. Commun. 1967, 305. (b) Hallman, P. S.; McGarvey, B.
R.; Wilkinson, G. J. Chem. Soc. A 1968, 3143.
(22) (a) Harris, R. O.; Hota, N. K.; Sadavoy, L.; Yuen, J. M. C. J.
Organomet. Chem. 1973, 54, 259. (b) Ashworth, T. V.; Singleton,
E. J. Chem. Soc., Chem. Commun. 1976, 706. (c) Hamilton, D.
G.; Crabtree, R. H. J. Am. Chem. Soc. 1988, 110, 4126. (d) Van Der
Sluys, L. S.; Kubas, G. J.; Caulton, K. G. Organometallics 1991, 10,
1033.
Reaction of 1 with H2 in CDCl3. Compound 1 (20 mg, 17.4
µmol) was dissolved in CDCl3 (1 mL) in an NMR tube, and H2
gas was bubbled through the solution for 5 min. Within seconds,
the color turned from yellow to dark red, and small amounts of a
colorless solid precipitated. The 1H and 19F NMR spectra of
the reaction solution indicated the formation of [Ru(H)Cl-
8420 Inorganic Chemistry, Vol. 42, No. 25, 2003

The Case of [Tl(µ-F)3Ru(PPh3)3]
Scheme 1. Synthesis of [Tl(µ-F)3Ru(PPh3)3] (1) and
[RuCl3F2(PPh3)4Tl] (2)
Table 5. Hardness Parameters of F‚ and FHF from Calculated I and A
E(total)
(hartree)
Aa
(eV)
Ia
(eV)
ηb
F‚‚‚F
(Å)
species
F-
(eV) modelc
d
-99.418586 1.28
-99.623847 3.40
-99.620258 3.02
(3.40)
HF
MP2
CCSD spe
spe
spe
F‚
-99.371652
7.24 HF
6.84 MP2
7.00 CCSD spe
(7.01)
spe
spe
-99.498820
-99.509397
F+ e
-98.792661
-98.871069
-98.884217
15.76
17.08
17.01
HF
MP2
CCSD spe
spe
spe
(presumably polynuclear complexes), which could neither
be isolated nor identified.
(17.42)
FHF- f -199.488606 3.15
-199.885603 6.16
HF
MP2
eqg 2.268
eqg 2.329
As thallium(I) fluoride is available as anhydrous salt and
forms insoluble chlorides, it is an ideal candidate for
chloride-fluoride metathesis. The reaction of [RuCl2(PPh3)3]
with TlF (3 equiv) in dichloromethane at room temperature
yielded a mixture of products and uncoordinated PPh3, as
indicated by the 31P NMR spectrum of the reaction solution.
The two major products were identified as [Tl(µ-F)3Ru-
(PPh3)3] (1) and [Tl(µ3-F)(µ2-Cl)2Ru2(µ2-Cl)(µ2-F)(PPh3)4] (2)
(see below). Pure 1 was obtained by adding an excess of
PPh3 (5 equiv) to the reaction solution (Scheme 1), which
suppresses PPh3 dissociation and the formation of 2.
The formulation of [Tl(µ-F)3Ru(PPh3)3] (1) is supported
by 31P and 19F NMR spectroscopy, elemental analysis, and
an X-ray study. As already observed for other fluoro com-
plexes,5 traces of water in the solvents (CDCl3 or CD2Cl2)
lead to the loss of P,F-coupling in the 31P NMR subspectrum
and to the disappearance of the 19F NMR signal of 1.
Furthermore, the 31P and 19F NMR spectra of 1 in CD2Cl2
or CDCl3 are temperature-dependent, which hints to the
occurrence of a yet unidentified dynamic process. As 1 is a
nonelectrolyte in CH2Cl2, this process probably does not
involve ionization. Resolved NMR spectra were obtained at
-20 °C (31P) and at -60 °C (19F) in CD2Cl2 dried over
molecular sieves. The 31P NMR spectrum of [Tl(µ-F)3Ru-
(PPh3)3] (1) at -20 °C was simulated as the MM′M′′ part
of the total AA′A′′MM′M′′X spin system (Figure 1 and
Experimental Section). The room temperature19F NMR
spectrum of [Tl(µ-F)3Ru(PPh3)3] (AA′A′′ part) shows a broad
doublet with a splitting of about 1060 Hz. The resolved
spectrum obtained at -60 °C is a doublet of doublets whose
simulation gave 1JTl,F ) 890 Hz besides the same trans-JP,F
coupling constant (201 Hz) found for the 31P (MM′M′′)
subspectrum.
-199.883831 4.93
CCSD spe 2.329
FHFf
-199.372931
-199.659182
-199.702798
5.20 HF
4.04 MP2
eqg 3.121g
eqg 2.980h
5.00 CCSD spe 2.980h
FHF+ e,f -198.874702
-199.135370
13.56
14.25
14.93
HF
MP2
eqg 2.383
eqg 2.429
-199.154050
CCSD spe 2.429
a Experimental values (from ref 44) in parentheses. b Defined as η ) (I
- A)/2. c HF ) HF/6-31+G*//HF/6-31+G*, MP2 ) MP2/6-31+G*//MP2/
6-31+G*, CC ) CCSD/6-31+G*//MP2/6-31+G*. d Single point energy
(spe) or equilibrium geometry (eqg). e S ) 1. f Corrected for thermal
energy (T ) 298.15 K, P ) 1 atm). g H-F distances are 0.914 and
2.207 Å. h H-F distances are 0.934 and 2.046 Å.
6.91-7.22 (m, 45 H, aromatic H). 31P{1H} NMR (THF-d8, 25
°C): δ 60.1 (br d, JPP ) ∼20 Hz, 2 P), 48.8 (br t, JPP ) ∼20 Hz,
1 P).
Calculations. Calculations were performed by using Spartan’02
(Wavefunction, Inc., Irvine, CA). Single point energies of F‚, F-,
and F+ were determined at the HF, MP2, and CCSD levels with
the 6-31+G* basis set. The geometries of FHF-, FHF‚, and FHF+
were optimized at the HF/6-31+G* and MP2/6-31+G* level.
Additionally, single point energies were calculated at the CCSD
level by using the corresponding MP2 optimized geometries.
Distances and energies (corrected for thermal energy, 298.15 K)
are given in Table 5. All species are linear. The FHF- anion and
the FHF+ cation are centrosymmetric (D∞h), whereas FHF‚ has C∞V
symmetry and can be described as F′-H‚‚‚F‚. Frequency calcula-
tions (zero imaginary) indicated that all structures correspond to
energy minima. The MP2/6-31+G* frequencies of FHF- are in
reasonable agreement with experimental values: 604 (sym stretch-
ing, expt 600-615), 1231 cm-1 (asym stretching, expt 1284-1563),
and 1288 (2-fold degenerate bending, expt 1199-1274).23 In
F′-H‚‚‚F‚, the spin density is concentrated on F‚ (1.00), and it is
0.00 on H and F′. The Mulliken charges are -0.51 (F′), +0.49
(H), and +0.02 (F‚). In the FHF+ biradical, the Mulliken charges
are F (+0.14), H (+0.72).
Complex 2 was obtained as the main product (34% yield)
by reacting [RuCl2(PPh3)3] with TlF (0.5 equiv) in CH2Cl2.
Unreacted [RuCl2(PPh3)3] (36%) and other unidentified
products (30%) were detected by 31P NMR spectroscopy.
Crystallization from CH2Cl2/pentane yielded pure 2, which
was studied by X-ray crystallography (see below). The 19F
NMR spectrum is a broad and featureless signal at δ -346
with a line width w1/2 of 360 Hz, which is much less than
the expected splitting caused by Tl-F coupling. In fact, the
1JTl,F coupling constant is expected to be not less than 800
Hz on the basis of the NMR spectra of 1 and of [Tl(µ-F)2Ru-
(dppe)2]+.14 As this suggests that fast chemical exchange
between the Tl-bound and the Tl-unbound fluorides occurs,
Results
Reaction of [RuCl2(PPh3)3] with TlF. Preliminary at-
tempts of halide metathesis between [RuCl2(PPh3)3] and
silver or cesium fluoride in dichloromethane solution were
unsuccessful. The reaction with AgF in CH2Cl2 led to
extensive decomposition to unidentified products, probably
owing to the traces of moisture invariably present in the salt.
Although CsF can be dried easily, its reaction with [RuCl2-
(PPh3)3] in CH2Cl2 led to a mixture of several products
(23) Panich, A. M. Chem. Phys. 1995, 196, 511.
Inorganic Chemistry, Vol. 42, No. 25, 2003 8421

Becker et al.
Figure 4. ORTEP view of [Tl(µ3-F)(µ2-Cl)2Ru2(µ2-Cl)(µ2-F)(PPh3)4] (2)
(30% probability ellipsoids).
Figure 3. ORTEP view of [Tl(µ-F)3Ru(PPh3)3] (1) (30% probability
ellipsoids).
The Ru-P distance (2.324(1) Å) is unexceptional. A
H‚‚‚F nonbonded contact between the fluoro ligand and the
ortho H atom on C(2) of a phenyl ring is significantly shorter
than the sum of the van der Waals radii of F and H (ca. 2.7
Å). On the basis of the Ru‚‚‚C(2) distance of 2.925(5) Å, a
H‚‚‚F distance of 2.09 Å is calculated by assuming d(C-H)
) 0.93 Å. Similar contacts have been considered as weak
C-H‚‚‚F hydrogen bonds.13b,24
Literature data about Tl-F bond lengths are very rare.
The Tl-F distance in 1 (2.389(4) Å) is shorter than in [Tl-
(µ-F)2Ru(dppe)2] (2.419(7) and 2.419(8) Å) and in a Tl(III)
fluoro porphyrin (2.441(6) Å),25 which can be ascribed to
the effect of the triple bridge. These Tl-F distances are
considerably shorter than in the (µ3-F)-Tl units of fluoro
aluminates (in the range 2.528(6)-2.598(7) Å)26 or in
C-F‚‚‚Tl linkages between thallium(I) and nonionic fluorine
(2.952(5)-3.048(5) Å).27 The closest nonbonded contact of
the Tl atom in 1 is to a C atom of a neighboring molecule
(Tl‚‚‚C(11) ) 3.958(3) Å). The closest intramolecular contact
is Tl‚‚‚C(14) (4.174(4) Å).
X-ray Structure of [Tl(µ3-F)(µ2-Cl)2Ru2(µ2-Cl)(µ2-F)-
(PPh3)4] (2). Complex 2 crystallized as red prisms from
dichloromethane/pentane. The core of the trimetallic complex
is a confacial Ru2F2Cl3P4 bioctahedron (Figures 4 and 5).
The Tl atom lies on the shared face of the octahedra and
binds to one of the Ru,Ru-bridging F atoms and to two Cl
atoms. The F(1) atom is µ3-bridging between Ru(1), Ru(2),
and Tl. The Tl-F(2) distance is much longer than the
Tl-(F(1) one (3.101(3) vs 2.668(3) Å) and is best considered
as nonbonding (Table 3). Owing to the µ3-bridging mode of
pseudo-C2V symmetry was assumed for complex 2 in
solution. Thus, the 31P NMR subspectrum was simulated as
the (MM′)2 part of an AA′(MM′)2X spin system (see Figure
2 and Experimental Section).
X-ray Structure of [Tl(µ-F)3Ru(PPh3)3] (1). Yellow
crystals of 1 belonging to the hexagonal P63 space group
were grown from CDCl3/pentane. The Tl and Ru atoms lie
on a crystallographic 63 axis. The F, P, and C atoms in
general positions generate the rest of the molecule (Figure
3). The complex can be described as being composed by a
pseudo-octahedral [RuF3(PPh3)3]- anion and a Tl+ cation
capping the F3 face. The Tl-F-Ru bridge is slightly
stretched at the fluoride pivots (the Tl-F-Ru angle is
95.2(1) °), which allows for a nonbonded Ru‚‚‚Tl distance
of 3.3446(9) Å. This is slightly shorter than in the doubly
F-bridged [Tl(µ-F)2Ru(dppe)2)]+, in which the Ru‚‚‚Tl
distance is 3.6640(8) Å.14
The triple (µ-F) bridge causes a distortion of the pseudo-
octahedral geometry at ruthenium. The trans-P-Ru-F angle
is 161.5(1)°, and the F-Ru-F′ angles are closed down to
76.03(15)° (Table 2). Thus, the F-Ru-F angle in 1 is
smaller than in [Tl(µ-F)2Ru(dppe)2]+ (77.9(3)°)14 and in cis-
[RuF2(dppp)2]) (78.2(1)°).13b The small F-Ru-F′ angles in
1 release the steric strain around ruthenium and allow for
relatively large P-Ru-F and P-Ru-P′ angles of 99.8(1)°
and 97.79(5)°, respectively. Analogous effects have been
observed in [Tl(µ-F)2Ru(dppe)2]+ and cis-[RuF2(dppp)2].14,13b
The Ru-F bond length in 1 (2.133(3) Å) is comparable to
those of the closely related [Tl(µ-F2)Ru(dppe)2]+ (2.112(7)
and 2.119(7) Å).14 Both values are significantly longer than
in the related terminal fluoro complexes cis-[RuF2(dppp)2]
(2.069(3) and 2.056(3) Å)13b and cis,cis,trans-[RuF2(CO)2-
(PPh3)2] (2.011(4) Å).9a The most interesting comparison is
between 1 and the recently reported confacial bioctahedral
cation [Ru2(µ-F)3(PEt3)6]+,16 which features similar Ru-F
distances (in the range 2.132(2)-2.170(2) Å vs 2.133(3) Å
in 1) and an average F-Ru-F angle of 73.7° (vs 76.0(2)°
in 1).
(24) (a) Coleman, K. S.; Fawcett, J.; Holloway, J. H.; Hope, E. G.; Russell,
D. R. J. Chem. Soc., Dalton Trans. 1997, 3557. (b) Cockman, R. W.;
Ebsworth, E. A. V.; Holloway, J. H.; Murdoch, H.; Robertson, N.;
Watson, P. G. In Inorganic Fluorine Chemistry; Thrasher, J. S.,
Strauss, S. H., Eds.; ACS Symposium Series 555; Americal Chemical
Society: Washington, DC, 1994.
(25) Coutsolelos, A. G.; Orfanopoulos, M.; Ward, D. L. Polyhedron 1991,
10, 885.
(26) Gonsior, M.; Krossing, I.; Mitzel, N. Z. Anorg. Allg. Chem. 2002,
628, 1821.
(27) Takemura, H.; Nakashima, S.; Kon, N.; Yasutake, M.; Shinmyozu,
T.; Inazu, T. J. Am. Chem. Soc. 2001, 123, 9293.
8422 Inorganic Chemistry, Vol. 42, No. 25, 2003

The Case of [Tl(µ-F)3Ru(PPh3)3]
Scheme 2. Reactivity of [Tl(µ-F)3Ru(PPh3)3] (1) with Dihydrogen
There are some short F‚‚‚H-C nonbonded contacts to H
atoms of the phenyl rings, but also Cl‚‚‚H-C interactions
(2.50, 2.67 Å) that are considerably shorter than the sum of
the van der Waals radii of Cl and H, which exceeds 2.9 Å.
Selected X‚‚‚C distances (X ) F or Cl) are given in Table
3. We have reported short intramolecular F‚‚‚H-C contacts
to the diphosphine ligand in [RuF(dppp)2]+ and cis-[RuF2-
(dppp)2],13b but also X‚‚‚H-C interactions, that can be
possibly described as weak hydrogen bonds, in the [RuX-
(dppp)2]+ series (X ) Cl, Br, or I).38 In a recent report,
Grushin has suggested that the fluoro ligands in [(PR3)(Ph)-
Pd(µ-F)2Pd(Ph)(PR3)] (R ) iPr or c-C6H11) retain a consider-
able basicity (as indicated by the short F‚‚‚H-C contacts
with the PiPr3 ligand or a CH2Cl2 molecule) despite their
bridging coordination mode.39
Reactivity of 1. Complex 1 readily decomposes in wet
CDCl3 to give a mixture of products, as shown by the 31P
NMR spectrum of the reaction solution. The 19F NMR
spectrum indicates that free F- and FHF- are formed by
protonolysis of the Ru-F bond. Similarly, the d6 fluoro
complex [IrF(Ph)(C5Me5)(PMe3)] is hygroscopic and dis-
sociates F- in wet solvents.40 However, 1 is stable in the
presence of moisture in the solid state and is therefore less
reactive than cis-[RuF2(dppp)2], which readily decomposes
in moist air even in the solid state and must be handled under
an inert atmosphere.13b
The Ru-F bond of 1 undergoes hydrogenolysis readily.
When [Tl(µ-F)3Ru(PPh3)3] (1) was dissolved under nitrogen
in THF-d8 in an NMR tube and dihydrogen was bubbled
through the reaction mixture, the yellow color of the solution
turned to pale brown within minutes, and a white solid
(thallium fluoride) precipitated. The 31P and 1H NMR spectra
of the reaction solution (a singlet at δ 56.8 and a broad signal
at δ -7.57, respectively) indicated that 1 was quantitatively
converted into the ruthenium hydride [Ru(H)2(H2)(PPh3)3]
as the only metal-containing product (Scheme 2).22 When
CD2Cl2 or CDCl3 was used as solvent, [Ru(H)Cl(PPh3)3]21
was eventually formed by reaction of [Ru(H)2(H2)(PPh3)3]
with the chlorinated solvent.41
Figure 5. ORTEP view of the coordination core of [Tl(µ3-F)(µ2-Cl)2Ru2-
(µ2-Cl)(µ2-F)(PPh3)4] (2) (30% probability ellipsoids).
fluoride, the Tl-F(1) bond in 2 is longer than in 1 (2.668-
(3) vs 2.389(4) Å). Similar (µ3-F)-Tl distances have been
found in Tl(I)-bound fluoro aluminates (2.528(6)-2.598(7)
Å).26 The Tl-Cl distances (3.0195(19) and 2.9719(17) Å)
in 2 can be compared to those of [Tl(µ-Cl)RuH(P(CH2CH2-
PPh2)3)] (2.815(4) Å)28 and [TlCl2Ru(PPh3)(S3)]2+ (average
Tl-Cl distance is 3.03 Å; S3 ) 1,4,7-trithiacyclononane).29
The octahedral coordination around Ru(1) and Ru(2) in 2
is similar to that of 1 and results in a Ru(1)‚‚‚Ru(2)
nonbonded distance of 3.1909(11) Å (Table 3). This is in
line with the values found for confacial bioctahedral com-
plexes of ruthenium(II), in which there are no bonding
interactions between the two d6 metals.30 The Ru-F bond
lengths in 2 are in the range 2.195(3)-2.242(4) Å (average
2.21 Å), and hence are longer than in 1 (2.133(3) Å). In
particular, the Ru-F bonds to F(2) are longer than those to
F(1), which is probably an effect of the different bridging
modes (µ2 vs µ3). The Ru-F distances to the µ2-bridging
F(2) (2.211(3) and 2.242(4) Å) are longer than in [Ru2(µ-
F)3(PEt3)6]+ (2.15 Å)16 and in [Ru2H2(µ-F)2(PiPr3)4] (2.12
Å)31 probably owing to the effect of the bridging chloride.
Although (µ-X)3 bridges are common in ruthenium chemis-
try,32-36 a Ru2(µ2-X)2(µ3-X)M structural motif (X ) halo
ligand, M ) metal ion) analogous to that of 2 has only been
found as part of the Ru3(µ2-Cl)3(µ3-Cl)2 core.37
(28) Bianchini, C.; Masi, D.; Linn, K.; Mealli, C.; Peruzzini, M.; Zanobini,
F. Inorg. Chem. 1992, 31, 4036.
(29) Blake, A. J.; Christie, R. M.; Roberts, Y. V.; Sullivan, M. J.; Schro¨der,
M.; Yellowlees, L. J. J. Chem. Soc., Chem. Commun. 1992, 848.
(30) Knottenbelt, S. Z.; McGrady, J. E.; Heath, G. A. J. Chem. Soc., Dalton
Trans. 2003, 227.
The reactivity of 1 toward H2 parallels that of [RuCl2-
(PPh3)3], which is known to activate H2 heterolytically in
the presence of NEt3 to give [RuH(Cl)(PPh3)] or [RuH2(H2)-
(31) Coalter, J. N.; Huffman, J. C.; Streib, W. E.; Caulton, K. G. Inorg.
Chem. 2000, 39, 3757.
(32) Clucas, W. A.; Armstrong, R. S.; Buys, I. E.; Hambley, T. W.; Nugent,
K. W. Inorg. Chem. 1996, 35, 6789.
(33) (a) La Placa, S. J.; Ibers, J. A. Inorg. Chem. 1965, 4, 778. (b)
MacFarlane, K. S.; Joshi, A. M.; Rettig, S. J.; James, B. R. Inorg.
Chem. 1996, 35, 7304.
(34) (a) Raspin, K. A. J. Chem. Soc. A 1969, 461. (b) Statler, J. A.;
Wilkinson, G.; Thornton-Pett, M.; Hursthouse, M. B. J. Chem. Soc.,
Dalton Trans. 1984, 1731.
(37) (a) Mashima, K.; Hino, T.; Takaya, H. Tetrahedron Lett. 1991, 32,
3101. (b) Mashima, K.; Hino, T.; Takaya, H. J. Chem. Soc., Dalton
Trans. 1992, 2099. (c) Mashima, K.; Komura, N.; Yamagata, T.; Tani,
K.; Haga, M. Inorg. Chem. 1997, 36, 2908. (d) Liu, S. H.; Yang, S.
Y.; Lo, S. T.; Xu, Z. T.; Ng, W. S.; Wen, T. B.; Zhou, Z. Y.; Lin, Z.
Y.; Lau, C. P., Jia, G. C. Organometallics 2001, 20, 4161.
(38) Barthazy, P.; Broggini, D.; Mezzetti, A. Can. J. Chem. 2001, 79, 904.
(39) Grushin, V. V.; Marshall, W. J. Angew. Chem., Int. Ed. 2002, 41,
4476.
(35) Rhodes, L. F.; Sorato, C.; Venanzi, L. M.; Bachechi, F. Inorg. Chem.
1988, 27, 604.
(36) Holmes, N. J.; Levason, W. L.; Webster, M. J. Chem. Soc., Dalton
Trans. 1997, 4223.
(40) Veltheer, J. E.; Burger, P.; Bergman, R. G. J. Am. Chem. Soc. 1995,
117, 12478.
(41) For a related reaction, see: Christ, M. L.; Sabo-Etienne, S.; Chaudret,
B. Organometallics 1994, 13, 3800.
Inorganic Chemistry, Vol. 42, No. 25, 2003 8423

Becker et al.
Scheme 3. Reactivity of [Tl(µ-F)3Ru(PPh3)3] (1) with Ph2CHBr
(PPh3)3] when a stronger base (such as NaOH) is used.42 In
an initial step, dihydrogen probably induces the dissociation
of thallium fluoride from 1 to form an elusive [RuF2(PPh3)3]
intermediate analogous to [RuCl2(PPh3)3], which can add
dihydrogen to form an acidic (η2-H2) complex43 that under-
goes hydrogenolysis. However, neither of these species was
detected by NMR spectroscopy. In fact, contrary to what is
observed with [RuCl2(PPh3)3], the hydrogenolysis of the
Ru-F bond occurs without the presence of a base. This
reaction is a common feature of fluoro complexes of ruthen-
ium,10,13 as it is driven by the thermodynamically favored
formation of HF.
Figure 6. ORTEP view of [Tl(µ3-F)(µ2-X)2Ru2(µ2-Br)(µ2-F)(PPh3)4] (3)
(30% probability ellipsoids).
As a formal adduct of [RuF2(PPh3)3] with TlF, [Tl(µ-
F)3Ru(PPh3)3] (1) is a shuttle of thallium(I) fluoride in
solvents of low polarity. Therefore, we investigated the
reaction of 1 with alkyl bromides (tert-butyl bromide,
1-bromo-1-phenylethane, and bromodiphenylmethane) with
the dual goal of preparing [RuF2(PPh3)3] and developing a
new reagent for the nucleophilic fluorination of alkyl
bromides.13,14 Complex 1 and the alkyl bromide were
dissolved in CH2Cl2 under argon, and the reaction solution
was sampled and analyzed by 31P and 19F NMR spectroscopy.
Bromodiphenylmethane (1.1 equiv vs 1) reacted with 1 to
form Ph2CHF as the only product (33% yield), as indicated
Figure 7. ORTEP view of the coordination core of [Tl(µ3-F)(µ2-X)2Ru2-
(µ2-Br)(µ2-F)(PPh3)4] (3) (30% probability ellipsoids).
1
by the H and 19F NMR spectra of the reaction solution
(Scheme 3). After 22 h of reaction time, the 31P NMR
spectrum showed the presence of the mixed-halide trimetallic
complex [Tl(µ3-F)(µ2-F)(µ2-X)Ru2(µ2-Br)(µ2-F)(PPh3)4] (X
) F or Br) (3) (67% yield) along with unreacted 1 (33% of
starting). The yields observed for the above reaction suggest
the stoichiometry shown in Scheme 3,44 in which Ph2CHBr
reacts with two equiv of 1 to give 3 (1 equiv) and Ph2CHF
(1 equiv). Thallium(I) fluoride precipitates from the solution
and is thus lost in a nonproductive fashion. Complex 3
crystallized upon diffusion of pentane into the CH2Cl2/CDCl3
reaction solution and was characterized by X-ray (see below).
Less activated substrates led to sluggish reactions with 1
and to the formation of mixtures of products that could not
be analyzed. Thus, 1 reacted with tert-butyl bromide (5
equiv) to produce some (CH3)3CF (detected by 19F NMR
spectroscopy), but the conversion of 1 was only ca. 34%
after 3 days of reaction time. The 31P NMR spectrum of the
reaction indicated that a mixture of several unidentified
complexes having 31P and 19F spectral patterns similar to
those of 2 was formed (see below). Using a large excess of
tert-butyl bromide (100 equiv vs 1) did not significantly
improve the total conversion of 1 (37%). In this case, 3 was
formed as the major product (more than 90% selectivity).
Complex 1 reacted quantitatively with a large excess (200
equiv) of 1-bromo-1-phenylethane within 22 h to give several
metal complexes that were not identified.
X-ray Structure and 31P and 19F NMR Spectra of 3.
Crystals of the trimetallic complex [Tl(µ3-F)(µ2-F)(µ2-X)-
Ru2(µ2-Br)(µ2-F)(PPh3)4] (3) (X ) F or Br) were obtained
as red prisms from a CH2Cl2/CDCl3/pentane solution. The
overall structure (Figure 6) is similar to that of 2 (Figure 4).
The coordination core of 3 (Figure 7) is analogous to that
of complex 2 (Figure 5). The terminal halide ligand X(5)
on Ru(2) exhibits F/Br disorder. The refinement of X with
F and Br gave occupancies of 61% and 39%, respectively,
and reasonable Ru-X distances (Table 4). The triple bridge
linking the Ru atoms is formed by two µ2-bridging halides,
Br(3) and F(2), and by F(1), which acts as a µ3-ligand
between Tl, Ru(1), and Ru(2). The Ru-F bond lengths of
the Ru(µ2-F)Ru bridge are slightly shorter in 3 (2.150(4)-
(42) Grushin, V. V.; Vymenits, A. B.; Vol’pin, M. E. J. Organomet. Chem.
1990, 382, 185.
(43) For a discussion of the effect of F on the acidity of coordinated
dihydrogen in [RuX(η2-H2)(diphosphine)2], see: Xu, Z. T.; Bytheway,
I.; Jia, G. C.; Lin, Z. Y. Organometallics 1999, 18, 1761.
(44) This stoichiometry is approximate, however, as 3 is, at least in the
crystal analyzed by X-ray, a Br/F-mixed species with the compostition
C72H60Br1.39F3.61P4Ru2Tl.
8424 Inorganic Chemistry, Vol. 42, No. 25, 2003

The Case of [Tl(µ-F)3Ru(PPh3)3]
2.168(4) Å) than in 2 (2.164(4)-2.194(4) Å). Interestingly,
the Ru(1)-F(4) distance of 2.067(4) Å in one Ru(µ2-F)Tl
bridge is closer to those found for the terminal Ru-F bonds,
such as in cis-[RuF2(dppp)2] (2.069(3) and 2.056(3) Å),13b
than to the Ru-F bond length in 1 (2.133(3) Å). This
suggests that the (µ2-F)-Tl interaction is weaker in 3 than
in 1. The Ru(2)-X(5), Tl-X(5), and Tl-F(4) distances are
not significant owing to the F/Br disorder.
The Tl-F(1) and Tl‚‚‚F(2) distances are 2.694(4) and
2.843(4) Å vs 2.635(4) and 2.973(5) Å in 2. Although the
Ru(1)-Br(3) and Ru(2)-Br(3) distances are obviously
longer than the Ru(1)-Cl(3) and Ru(2)-Cl(3) ones by about
0.11 Å, the confacial RuX3Ru bioctahedron is slightly less
stretched in 3 than in 2, as indicated by the Ru(1)‚‚‚Ru(2)
nonbonded distances of 3.1396(9) and 3.1503(9) Å, respec-
tively. The latter feature suggests that the interaction between
Tl and the terminal halide X effectively “holds together” the
Ru‚‚‚Ru core. Accordingly, the µ-Br-bridge is distal to
thallium, which minimizes the Tl-F(4) and Tl-X(5) dis-
tances.
even at -60 °C, the simulation of the 19F NMR spectrum
yielded a qualitative agreement with the experimental one
but no exact spectral parameters. The 31P NMR spectrum of
3 (Figure S2, Supporting Information) is analogous to that
of 2 (Figure 2) and was simulated as the (MM′)2 part of an
AA′BB′(MM′)2X spin system by assuming the C2V-sym-
metric structure [Tl(µ3-F)2(µ2-F)2Ru2(µ2-Br)P4] (see Experi-
mental Section). Finally, on the basis of the similarity of
their spectra, we assign the 31P NMR signals of the
unidentified byproducts of the reactions of 1 with R-Br to
structural analogues of 3 containing a different halide set.
Attempted Catalytic Fluorination. Along the lines previ-
ously developed for [RuF(P-P)2]+ (P-P ) dppp or dppe),13,14
we tried to turn the stoichiometric fluorination of alkyl
bromides by 1 into a catalytic reaction. However, 1-bromo-
1-phenylethane did not react with thallium(I) fluoride (1.1
equiv) in the presence of 1 (10 mol %) as catalyst in
CH2Cl2 solution. No fluorinated product was observed even
after 48 h. This suggests that the dissociation of TlF from 1
is required for catalytic activity. With an excess of thallium
fluoride present in the reaction mixture, the equilibrium is
shifted toward the stable TlF adduct 1, and no fluoride
transfer takes place. Thus, the additional TlF inhibits the
catalytic reaction, and only the stoichiometric reaction of
Scheme 3 is possible.
Like in 1 and 2, the halide ligands form a network of
hydrogen bonds to some H atoms of the phenyl rings. The
shortest F‚‚‚H-C contacts are F(2)‚‚‚H-C(8) (2.15 Å) and
F(4)‚‚‚H-C(24) (2.16 Å), whereas the short Br(5)‚‚‚H-
C(38) distance is physically meaningless and clearly an
artifact of disorder.
Discussion
The similarity of 2 and 3 and the F/Br disorder observed
in 3 indicate that the trinuclear framework of 2 and 3
maintains its stability within a range of different halide
combinations. The comparison between 1, 2, and 3 indicates
that the Ru(µ2-X)Ru linkage (X ) F, Cl, or Br) is stable
provided that one halide X is chloride or bromide. The failure
to observe complexes containing a Ru(µ2-F)Ru core among
the products of 1 hints to the possibility that the bulky PPh3
ligands would cause severe steric crowding in the hypotheti-
cal binuclear complex [Ru2(µ-F)3(PPh3)6]+. In the reaction
of [RuCl2(PPh3)3] with TlF, the first halide exchange
probably produces the hypothetical intermediate [RuClF-
(PPh3)3], which aggregates to 2 by dissociation of PPh3 and
addition of the Tl(I)Cl formed in the reaction. The tendency
of the mixed-halide species to aggregation explains why the
stoichiometric reaction of 1 with alkyl bromides is not
quantitative: after the first halide metathesis between 1 and
R-Br, the resulting mixed-halide species aggregates to 3.
The room-temperature 19F NMR spectrum of 3 shows
signals at δ ) -341.7 and -422.6. The latter, a doublet of
poorly resolved multiplets with an exceptionally large JTl,F
coupling constant (2100 Hz), is attributed to the fluorides
of the Ru-F-Tl bridges. The signal of the fluoro ligands
in the Ru-F-Ru bridge (δ -341.7, w1/2 ) 375 Hz) is a
broad, tripletlike signal owing to the coupling with the two
F atoms of the Ru-(µ2-F)-Tl bridges (JF,F′ ca. 150 Hz)
(Figure S1, Supporting Information). The w1/2 of 375 Hz is
smaller than the expected Tl-F coupling constant (which
should be around 800 Hz), which is attributed to the fast
flipping of the Tl(I) ion between the fluoro ligands of the
two nonequivalent Ru-F-Tl bridges as already assumed for
2. As the dynamic process is fast on the NMR time scale
The Stabilization of 1: Steric Factors. There are both
steric and electronic reasons that can explain why the reaction
of [RuCl2(PPh3)3] with Tl(I)F gives [Tl(µ-F)3Ru(PPh3)3]
instead of [RuF2(PPh3)3] or [Ru2(µ-F)3(PPh3)6]+. The small
size of the fluoro ligand is obviously important to stabilize
[Tl(µ-F)3Ru(PPh3)3] as compared to [RuF2(PPh3)3]. In the
mixed-valence confacial bioctahedral complexes [Ru2(µ-X)3-
(tacn)2] (X ) Cl, Br, I; tacn ) 1,4,7-triazacyclononane), the
repulsive interactions between the halides in the bridge
increase with increasing size of the halide (Cl < Br < I).
However, the concomitant lengthening of the Ru-X bond
limits the effect, and the X-Ru-X angle opens only slightly
on going from Cl (88.8(14)°) to Br (90.0(9)°) and to I (90.9-
(2)°).32
With bulky phosphine ligands such as PPh3 (or PCy3), five-
coordinate [RuX2L3] complexes are formed when X is Cl33
(or Br). With smaller phosphines, the triply halide-bridged
binuclear cations [L3Ru(µ-X)3RuL3]+ can be obtained
instead.34-36 As the structural effects of the halide in halo-
bridged bioctahedral phosphine complexes of ruthenium(II)
have not been systematically investigated, a homogeneous
assessment is not possible. The F-Ru-F and P-Ru-P
angles in 1 (76.0(3)° and 97.79(5)°, respectively) compare
with the corresponding average values of [Ru2(µ-F)3(PEt3)6]+
(73.7° and 95.8°).16 In the chloro series, the Cl-Ru-Cl and
P-Ru-P average values are 77.2° and 96.8° for [Ru2(µ-
Cl)3(PEt2Ph)6]+,34a 80.7° and 95.4° for [Ru2(µ-Cl)3(PMe)6]+,34b
and 77.2° and 88.3° for [Ru2(µ-Cl)3((Ph2PCH2)3CCH3)2]+.35
The iodo derivative [Ru2(µ-I)3(PMe2Ph)6]+ has an average
I-Ru-I angle of 80.0° and P-Ru-P angles in the range
92.2(4)-98.9(2)°.36 These values confirm that the lengthen-
Inorganic Chemistry, Vol. 42, No. 25, 2003 8425

Becker et al.
ing of the Ru-X bond down the halide series partially
compensates the increase of the steric crowding with halide
size on going from the (µ-F)3 to the (µ-I)3 bridge, so that
the X-Ru-X angles remain nearly constant. Eventually, the
phosphine-halide and phosphine-phosphine interactions are
more important than those within the RuX3Ru unit in
determining the overall steric crowing in the complex.30 Thus,
the aggregation of two [RuF2(PPh3)3] units to form [Ru2(µ-
F)3(PPh3)6]+, by analogy with [Ru2(µ-F)3(PEt3)6]+,16 is prob-
ably disfavored by the interaction between the bulky PPh3
ligands.
With bulkier donor sets than F3P3, even the coordination
of the relatively small TlF to a five-coordinate fragment can
be unfavorable. In fact, when a F2P4 donor set is present on
ruthenium, the reaction of [RuCl(P-P)2]+ (P-P ) diphos-
phine) with TlF yields the thallium adduct [Tl(µ-F)2Ru-
(dppe)2]+ with 1,2-bis(diphenylphosphino)ethane (dppe)14 but
the five-coordinate [RuF(dppp)2]+ in the case of the more
bulky dppp ligand.13 Thus, the formation of 1 instead of
[RuF2(PPh3)3] is in line with the F3P3 donor set being overall
less crowded than the F2P4 one.
covalent bonds. Its validity decreases as the ionicity of the
A-B bond increases, as Pearson46 and Komorowski47 have
pointed out.
We suggest here a revision of our analysis of the M-F
bond that overcomes the contradiction mentioned above by
focusing on chemical hardness and the hard/soft acid/base
(HSAB) principle46,48 rather than on electronegativity. Parr
and Pearson have defined chemical hardness of Lewis acids
and bases as the average of the ionization potential I and
electron affinity A48
η ) (I - A)/2
and applied it to the HSAB principle.48 The M-X bond
formed in the reaction of the Lewis acid M+ with the
base X-
M+ + X- f M-X
has both a covalent and an ionic component. Thus, the
(heterolytic) bond dissociation energy of the M-X bond can
be decomposed into an ionic and a covalent component:
The Stabilization of 1: Electronic Factors. Some halo
complexes of late transition metals show surprising trends
in a series of parameters (CO stretching frequencies, oxida-
tion potentials, etc.) along the series F, Cl, Br, I, which
suggest that the fluoro analogues are the most electron rich,
although fluorine has the highest electronegativity.1 This
“inverse halide order” has been explained by postulating an
exceptionally strong π-donation from the fluoro ligand,6
which acts in combination with the 4-electron filled/filled
pπ/dπ repulsions discussed by Mayer7 and Caulton.4 We
have pointed out that there is consistent experimental
evidence that the “inverse halide order” observed for d6 and
d8 fluoro complexes is caused by the M-F bond having a
higher ionic component than M-X bonds (X ) Cl, Br, or
I),8 which is an effect of the high electronegativity of fluorine.
Also, an analysis of the literature indicates that many d6 and
d8 complexes with a terminal fluoro ligand either contain
strong π-acidic ligands or are 16-electron, coordinatively
unsaturated complexes, such as [IrH2F(PBut2Ph)2]12 and
[RuF(dppp)2]+.13 Thus, we have tentatively suggested that
all factors that reduce the electronegativity difference
between M and F favor the formation of fluoro complexes
of d6 and d8 metals and lower the nucleophilic reactivity of
the fluoro ligand.8 We have argued, in terms of Pauling’s
electronegativity ø, that reducing the electronegativity dif-
ference ∆ø between M and F decreases the ionicity and
enhances the covalent character of the M-F bond.
E ) E(ionic) + E(covalent)
The combination of a soft acid with a soft base gives a
strong M-X bond in which the covalent component is the
major one, and accounts for the bond strength. In hard/hard
combinations, the electrostatic interactions result in a strong
bond whose major component is the ionic one. A hard-soft
(mismatched) adduct has a weak bond, as it is stabilized
neither by the ionic component (weak charge transfer) nor
by the covalent one (small orbital interactions).48
In the hypothesis that the hard/soft mismatch between F
and M is responsible for the weakness of the M-F bond,
and thus for the high nucleophilic reactivity of fluoro ligands
in d6 and d8 complexes, then all factors that reduce the hard/
soft mismatch between M and F are expected to strengthen
the M-F bond. This can be achieved either by hardening
M or by softening F. Before illustrating this line of reasoning
with some examples, we point out the difference between
this approach and our previous proposal.8 Thus, the hardening
of M strengthens the M-F bond because the ionic compo-
nent of the bond becomes larger than in the hard/soft
mismatched M-F, and not necessarily because the M-F
bond becomes “more covalent” as we previously sug-
gested.8,49 However, one should appreciate that the hard/soft
mismatch and the electronegativity difference between M
and F are closely related, as electronegativity and hardness
are roughly proportional.48,50 Thus, any increase of elec-
tronegativity at the metal goes along with the “hardening”
of M and the decrease of the M/F hard/soft mismatch.
However, we now notice that this conclusion contradicts
the Pauling’s central idea that the increasing polarity of a
covalent bond reinforces the bond itself. In fact, the con-
tradiction arises from the limited validity of Pauling’s bond
energy equation45
(45) Pauling, L. The Nature of the Chemical Bond; Cornell University
Press: Ithaca, NY, 1960; p 92.
(46) Pearson, R. G. Chemical Hardness; Wiley-VCH: Weinheim, 1997;
p 24.
(47) Komorowski, L. Struct. Bonding 1993, 80, 45.
(48) (a) Parr, R. G.; Pearson, R. G. J. Am. Chem. Soc. 1983, 105, 7512.
(b) Pearson, R. G. Proc. Natl. Acad. Sci. U.S.A. 1986, 83, 8440.
(49) We are grateful to Dr. Vladimir Grushin for bringing this point to our
attention and for the enlightening discussions on this topic.
1
DoAB ) (DAo A + DoBB) + 23(øA - øB)2
2
which holds for bonds where (øA - øΒ) is small, that is, for
8426 Inorganic Chemistry, Vol. 42, No. 25, 2003

The Case of [Tl(µ-F)3Ru(PPh3)3]
In the fluoro carbonyl complexes [MF2(CO)2(PR3)2] (M
) Ru or Os),9a the strong π-accepting CO ligands stabilize
the dπ-orbitals of M, which increases the HOMO-LUMO
gap in the [M(CO)2(PR3)2]2+ fragment. As the HOMO-
LUMO gap is a rough estimate of the hardness of a
molecule,48b its increase corresponds to the hardening of M,
which strengthens the ionic component of the M-F bonds.
Analogously, phosphine complexes of ruthenium(II) are soft
and, in general, do not form complexes with neutral oxygen
donors. However, phosphine complexes of ruthenium are
oxophilic when containing hard coligands, such as imines
or CO. The hardened ruthenium atom then forms aqua and
triflato complexes, as in [RuCl(OH2)(PNNP)]+ (PNNP )
N,N′-bis{(o-diphenylphosphino)benzylidene}-(1S,2S)-[di-
aminocyclohexane]),51 [RuCl2(CO)(OH2)(PEt3)2],52 and [Ru(O3-
SCF3)2(OH2)(CO)(dppe)].53 The latter complex forms, in the
presence of an excess of water, the extremely interesting
species [Ru(CO)(OH2)3(dppe)]2+, in which the strong π-ac-
cepting ligand and the double positive charge cooperate in
enhancing the hardness of the metal, and hence its oxophi-
licity.
The hard/soft mismatch argument works equally well for
the π-stabilized, 16-electron cations [MFL4]+ (M ) d6 ion;
L ) phosphine ligand). Let us consider [MF2L4] and the
corresponding fluoride dissociation product, the 16-electron
species [MFL4]+. In fact, [MFL4]+ and [MF2L4] derive from
the acid-base reaction between fluoride and [ML4]2+ or
[ML4]+, respectively. As any Lewis base (in this case
fluoride) reduces the hardness of an ionic acid,48a the 16-
electron complex [MFL4]+ is softer than the 14-electron
[ML4]2+ fragment. Thus, the hard/soft mismatch between M
and F is more severe in the neutral 18-electron complex
[MF2L4] than in the corresponding 16-electron cation [MFL4]+.
Besides the metal-based effects discussed above, the “hard/
soft-mismatch” argument also explains ligand-based effects,
and in particular the stability of the binuclear fluoro-bridged
species of the type M-F-M′, such as [Ru2(µ-F)3(PEt3)6]+
(where M′ ) [Ru(PEt3)3]2+)16 and 1, 2, and 3 (where M′ )
Tl+). Fluoride bridging between Ru and a soft acid M′
(Ru-F-M′) is a softer ligand than the terminal fluoro ligand
(Ru-F) because of the symbiotic relationship of the former
with M′ (a soft acid).54 As both Ru(II) and Tl(I) are soft (η
) 5.86 and 7.16 eV, respectively),55a this applies both to
[Ru2(µ-F)3(PEt3)6]+, where M′ is [Ru(PEt3)3]2+, and to 1,
where M′ is Tl+.
(η ) 5.86 eV) and F is hard (η ) 7.01 eV),55 the Ru-F
bond in [RuF2(PPh3)3] will be weak. In the hypothetical
reaction
[RuF2(PPh3)3] + TlF f [Tl(µ-F)3Ru(PPh3)3]
the soft thallium atom interacts with the hard F atoms of
[RuF2(PPh3)3], and the soft ruthenium center with the hard
F atom of TlF. On the basis of the principle of symbiosis,54
these interactions harden ruthenium and soften fluoride in
[RuF2(PPh3)3]. As a result, the above reaction causes a
decrease of the hard/soft mismatch between Ru and F in
[RuF2(PPh3)3], and the Ru-F bond gets stronger.
A special case of bridging fluoride is given by the
bifluoride anion FHF-. Stable d6 and d8 bifluoride complexes
have general formulas [RuH2-n(FHF)nP4] (n ) 1 or 2),56
[PtH(FHF)P2],57 and [Pd(FHF)(Ph)P2].58 These complexes,
which exhibit Ru-F‚‚‚F linkages with F‚‚‚F distances in the
range 2.28-2.40 Å and different degrees of bending,58a seem
to contradict the principle of symbiosis, as both H+ and HF
are very hard.55 However, there is an inverse relationship
between hardness and molecular size59 that allows one to
predict that bifluoride FHF- is softer than F-. To put that
on a quantitative basis, we estimated the hardness η of
fluoride and bifluoride by using Pearson’s operational
definition of chemical hardness of anionic bases B- as the
average of the ionization potential I and electron affinity A
of the corresponding radicals B‚.60,61 Thus, we calculated η
of F‚ and FHF‚ from the corresponding ionizaton potentials
(I) and electron affinities (A). The I and A values of the
F‚ radical were calculated from the energies of F‚, F+, and
F- at the HF, MP2, and CCSD levels with the 6-31+G*
basis set. The calculated I of F‚ (MP2, 17.08; CCSD, 17.01
eV) is underestimated but fairly close to the experimental
value (17.42 eV) (Table 5).48a The calculated electron affinity
is in excellent agreement with the experimental value (3.40
eV)48a for the MP2 model (3.40 eV), whereas the CCSD
(55) (a) Pearson, R. G. Inorg. Chem. 1988, 27, 734. (b) The fact that Tl(I)
is soft with η ) 7.16 eV, yet F is hard with η ) 7.01 eV, sounds
confusing. However, it should be noted that, as the hardness scales
for cations and anions span different ranges, the numerical values of
η(anion) and η(cation) cannot be compared. Only comparison between
anions or cations are meaningful: Tl(I) (η ) 7.16 eV) is soft in
comparison to Li+ (η ) 35.12 eV), and F- (η ) 7.16 eV) is hard in
comparison to I- (η ) 3.70 eV).
(56) (a) Whittlesey, M. K.; Perutz, R. N.; Greener, B.; Moore, M. H. Chem.
Commun. 1997, 187. (b) Jasim, N. A.; Perutz, R. N.; Foxon, S. P.;
Walton, P. H. J. Chem. Soc., Dalton Trans. 2001, 1676. (c) Kirkham,
M. S.; Mahon, M. F.; Whittlesey, M. K. Chem. Commun. 2001,
813.
(57) (a) Jasim, N. A.; Perutz, R. N. J. Am. Chem. Soc. 2000, 122, 8685.
(b) Hintermann, S.; Pregosin, P. S.; Ru¨egger, H.; Clark, H. C. J.
Organomet. Chem. 1992, 435, 225.
(58) (a) Roe, D. C.; Marshall, W. J.; Davidson, F.; Soper, P. D.; Grushin,
V. V. Organometallics 2000, 19, 4575. (b) Pilon, M. C.; Grushin, V.
V. Organometallics 1998, 17, 1774. (c) Pilon, M. C.; Grushin, V. V.
J. Am. Chem. Soc. 1998, 119, 44769.
The above considerations allow us to explain why 1 is
formed instead of [RuF2(PPh3)3]. As ruthenium(II) is soft
(50) According to Mulliken and Jaffe´, the electronegativity ø is defined as
ø ) (I + A)/2. Both ø and η are dominated by the ionization potential
I, which is the largest contribution to both of them. Thus, large η
values parallel large ø values. See also: Huheey, J. E.; Keiter, E. A.;
Keiter, R. L. Inorganic Chemistry: Principles of Structure and
ReactiVity; HarperCollins: New York, 1993.
(51) Stoop, R. M.; Bachmann, S.; Valentini, M.; Mezzetti, A. Organome-
tallics 2000, 19, 4117.
(52) Sun, Y.; Taylor, N. J.; Carty, A. J. Inorg. Chem. 1993, 32, 4457.
(53) Mahon, M. F.; Whittlesey, M. K.; Wood, P. T. Organometallics 1999,
18, 4068.
(59) A general inverse relationship has been found between the global
hardness of a molecule and molecular size (roughly estimated by the
number of atoms in the molecule). See: Baekelandt, B. G.; Mortier,
W. J.; Schoonheydt, R. A. Struct. Bonding 1993, 80, 211.
(60) Pearson, R. G. In The Concept of the Chemical Bond; Maksic, Z. B.,
Ed.; Springer: Berlin, 1990; p 46.
(54) (a) Jørgensen, C. K. Inorg. Chem. 1964, 3, 1201. (b) For a
quantomechanical interpretation, see: Nalewajski, R. F. Struct. Bond-
ing 1993, 80, 117.
(61) Analogous calculations for open-shell radicals (including F‚) have been
reported: (a) Kar, T.; Scheiner, S.; Sannigrahi, A. B. THEOCHEM
1988, 427, 79. (b) Roy, R. K.; Pal, S. J. Phys. Chem. 1995, 99, 17822.
Inorganic Chemistry, Vol. 42, No. 25, 2003 8427

Becker et al.
value (3.02 eV) is slightly underestimated. In sum, the
experimental hardness of fluoride (7.01 eV)48a is fairly well
reproduced by calculation (η ) 6.84 (MP2) and 7.00 eV
(CCSD)).
For the FHF‚ radical, the adiabatic ionization potential (I)
and electron affinity (A) were calculated from the energies
of FHF-, FHF‚, and FHF+ after geometry optimization at
the HF/6-31+G* and MP2/6-31+G* levels.62 Vibrational
analysis indicated that all the optimized structures correspond
to energy minima. Both models (HF and MP2) gave
analogous structures (Table 5). Bifluoride FHF- has a linear,
centrosymmetric structure (D∞h symmetry), with a F‚‚‚F
distance of 2.268 (HF) and 2.329 Å (MP2), which is in good
agreement with the experimental F‚‚‚F distances in the range
2.26-2.40 Å.63 The FHF‚ radical has C∞V symmetry and can
be described as F′-H‚‚‚F‚, with a large F‚‚‚F separation and
one short and one long H-F distance (Table 5). The FHF+
cation is centrosymmetric (D∞h) with a F‚‚‚F distance of
2.383 Å.
Figure 8. Schematic orbital description of the Ru-F-Tl bond in 1.
Energy calculations at the HF, MP2, and CCSD levels
gave the energy values reported in Table 5. Although the
calculated η values of FHF‚ vary depending on the model
used, the qualitative trends of I and A show that the
interaction with an HF molecule significantly increases the
electron affinity of the F‚ radical and reduces its ioniza-
tion potential. The sum of these two effects is a decrease
of the hardness of fluoride upon interaction with HF. The
η values indicate that bifluoride (FHF-) is less hard (η )
4.04-5.20 eV) than fluoride (ηexp ) 7.01 eV). This is in
agreement with qualitative arguments based on molecular
size59 and, more importantly, with chemical evidence, that
is, the formation of bifluoride complexes instead of the
corresponding species containing terminal fluoro ligands.56-58
Thus, the bridging coordination mode (M-F-M′) softens
fluoride even when a hard partner (M′) such as HF is
involved.
Interestingly, there is a parallel between fluoride and other
hard, π-donor ligands, such as oxo, hydroxo, and amido,
which tend to prefer the bridging coordination mode when
bound to soft late transition metals in low oxidation states.
Mayer has attributed this preference to the redistribution of
the 4-electron filled/filled pπ/dπ repulsion over several
centers.7 Holland, Andersen, and Bergman have criticized
this approach and argued that, as the stabilization energy
from π-bonding between M and N in alkylamide complexes
of early transition metals is estimated to be less than 10 kcal/
mol, the corresponding 4-electron pπ/dπ repulsion energy
would be small (probably not more than 1-2 kcal/mol).17
They suggested an alternative analysis of alkoxo and amido
cyclopentadienyl Ni complexes that focuses on the ionic
character of the M-X bond rather than on π-effects, which
is analogous to our analysis of the M-F bond. Application
of Drago’s E-C theory64 to thermochemical data of cyclo-
pentadienyl Ni complexes indicates that the ionic component
of the M-X bond dominates over the covalent one and
determines the thermodynamic preferences of M toward N
or O donors.17
A final remark is that the explanation of the electronic
factors across the Ru-F-M bridge in terms of π-effects is
inconsistent with experiment, which indirectly reinforces the
hard/soft-mismatch argument exposed above. As sketched
in Figure 8, the doubly occupied Ru and F orbitals with
π-symmetry (Ru/F) give rise to a 4e-interaction (Ru-F). The
Tl atom involves the p(F)-orbitals with Ru-F π-character
(πRu-F) in a partially covalent dative Tl-F σ-bond (σTl-F
)
and lowers their energy (F-Tl). This is expected to relieve
the filled/filled interaction between F and Ru, that is, to lower
the energy of the π*Ru-F orbital (Ru-F-Tl). Clearly, the
same consideration applies to [Ru2(µ-F)3(PEt3)6]+, which can
be formally considered as formed by a “[P3RuF3]-” fragment
interacting with a “[RuP3]2+” dication. However, the Ru-F
distances in the latter species (in the range 2.132(2)-2.170-
(2) Å) and in 1 (2.133(3) Å) are very similar, which suggests
that Tl+ and “[RuP3]2+” cause similar effects upon coordi-
nation onto “[P3RuF3]-”. This observation suggests that
there is no significant destabilization that can be ascribed to
pπ/dπ filled/filled interactions, which should be operative
in [Ru2(µ-F)3(PEt3)6]+ but not in 1.
Final Remarks
We have recently discussed the effect of the ionic nature
of the M-F bond in d6 and d8 complexes and described some
metal-based features that stabilize low-valent fluoro com-
plexes of late transition metals.8 These features are the
presence of strong π-acceptor ligand that gives a push-pull-
interaction4 with F or the electron deficiency of the π-sta-
bilized11 cationic complexes [MXL4]+. We agree with
Perutz’s statement that the bridging mode favors the coor-
dination of fluoride to late transition metal ions.16 Therefore,
we suggested here a unified vision of the coordination modes
of fluoride (terminal or bridging) in terms of the HSAB
(62) MP2 calculations have been reported to account satisfactorily for
electron correlation in bifluoride: Kawahara, S.; Uchimaru, T.; Taira,
K. Chem. Phys. 2001, 273, 207.
(63) See refs 6-25 in ref 23.
(64) Drago, R. S. Applications of Electrostatic-CoValent Models in
Chemistry; Surfside: Gainesville, FL, 1994.
8428 Inorganic Chemistry, Vol. 42, No. 25, 2003

The Case of [Tl(µ-F)3Ru(PPh3)3]
model. As for the bridging coordination mode, we notice
that additional covalent interactions “soften” the fluoro ligand
and reduce thereby the hard/soft mismatch between Ru and
F. The latter effect strengthens the covalent component of
the Ru-F bond in the Ru-F-M′ unit and stabilizes d6
complexes containing bridging fluoro ligands. This explana-
tion can be reversed to explain the metal-based effects
mentioned above. In fact, push-pull interactions and elec-
tron deficiency “harden” the metal and strengthen the M-F
bond by means of its ionic component in complexes that
either contain strong π-accepting ligands or are 16-electron
cations.
Supporting Information Available: Listings of X-ray data of
1, 2, and 3 (atomic coordinates, anisotropic thermal parameters,
bond distances and angles, and calculated positions of H atoms)
(CIF format), as well as the 31P and 19F NMR spectra of 1 (S1)
and the 31P NMR spectrum of 3 (S2). This material is available
free of charge via the Internet at http://pubs.acs.org.
IC034888E
Inorganic Chemistry, Vol. 42, No. 25, 2003 8429