Metallocene-terminated allylium salts: The effect of end group on localization in polymethines

Article abstract of DOI:10.1021/ja012472z

A range of 1,3-di(metallocenyl)allylium salts [Mc(CH)₃Mc′]⁺[X]^- {Mc, Mc′ = ferrocenyl (Fc), 2,3,4,5,1′,2′,3′,4′-octamethylferrocen-1-yl (Fc″), ruthenocenyl (Rc); X = BF₄, PF₆} was synthesized by reaction of (2-lithiovinyl)metallocenes with formylmetallocenes, followed by treatment of the resulting alcohols with HX. Two salts with X = BAr′₄ {Ar′ = 3,5-(CF₃)₂C₆H₃} were synthesized by anion metathesis from the corresponding PF₆ salts. The crystal structure of [Fc″(CH)₃Fc″]⁺[PF₆]^- contains symmetrical termethine cations, while the same appears to be true in the disordered structure of [Fc(CH)₃Fc]⁺[PF₆]^-. The formally unsymmetrical cation in [Fc(CH)₃Fc″]⁺[BF₄]^- is only slightly unsymmetrical with little bond-length alternation in the allylium bridge. In contrast, the crystal structures of [Rc(CH)₃Rc]⁺[PF₆]^- and [Rc(CH)₃Rc]⁺[BAr′₄]^- both contain a bond-alternated Peierls-distorted cation, which can be considered as a ruthenocene bridged to a [(η⁶-fulvene)(η⁵-cyclopentadienyl)ruthenium] cation by a vinylene moiety. The strong similarity between solid-state and solution infrared and Raman spectra of [BF₄]^-, [PF₆]^-, and [BAr′₄]^- salts of [Rc(CH)₃Rc]⁺ indicates that the C-C stretching constant in the allylium chain and, therefore, the structure, of this ion are largely independent of the local environment, suggesting that the unsymmetrical structures observed in the crystal structures are not simply an artifact of packing. Differences in the solvatochromism of [Rc(CH)₃Rc]⁺ and [Fc(CH)₃Fc]⁺ also suggest a localized structure for the former cation in solution. Electrochemistry, UV-visible-NIR spectroscopy, and DF calculations give insight into the electronic structure of the metallocene-terminated allylium cations. Using an analogy between polymethines and mixed-valence compounds, the difference between the behaviors of [Fc(CH)₃Fc]⁺ and [Rc(CH)₃Rc]⁺ is attributed to larger reorganization energy associated with the geometry differences between metallocene and [(η⁶-fulvene)(η⁵-cyclopentadienyl)metal] structures in the ruthenium case.

Full text of DOI:10.1021/ja012472z

Published on Web 05/08/2002
Metallocene-Terminated Allylium Salts: The Effect of End
Group on Localization in Polymethines
Stephen Barlow,*,†,‡,§ Lawrence M. Henling,† Michael W. Day,†
William P. Schaefer,† Jennifer C. Green,‡ Tony Hascall,‡ and Seth R. Marder†,§,|
Contribution from the Beckman Institute, 139-74, California Institute of Technology,
Pasadena, California 91125, Inorganic Chemistry Laboratory, UniVersity of Oxford,
South Parks Road, Oxford, OX1 3QR, U.K., Jet Propulsion Laboratory, California Institute of
Technology, Pasadena, California 91109, and Department of Chemistry, UniVersity of Arizona,
Tucson, Arizona 85721
Received November 5, 2001
Abstract: A range of 1,3-di(metallocenyl)allylium salts [Mc(CH)3Mc′]+[X]- {Mc, Mc′ ) ferrocenyl (Fc),
2,3,4,5,1′,2′,3′,4′-octamethylferrocen-1-yl (Fc′′), ruthenocenyl (Rc); X ) BF4, PF6} was synthesized by
reaction of (2-lithiovinyl)metallocenes with formylmetallocenes, followed by treatment of the resulting alcohols
with HX. Two salts with X ) BAr′4 {Ar′ ) 3,5-(CF3)2C6H3} were synthesized by anion metathesis from the
corresponding PF6 salts. The crystal structure of [Fc′′(CH)3Fc′′]+[PF6]- contains symmetrical termethine
cations, while the same appears to be true in the disordered structure of [Fc(CH)3Fc]+[PF6]-. The formally
unsymmetrical cation in [Fc(CH)3Fc′′]+[BF4]- is only slightly unsymmetrical with little bond-length alternation
in the allylium bridge. In contrast, the crystal structures of [Rc(CH)3Rc]+[PF6]- and [Rc(CH)3Rc]+[BAr′4]-
both contain a bond-alternated “Peierls-distorted” cation, which can be considered as a ruthenocene bridged
to a [(η6-fulvene)(η5-cyclopentadienyl)ruthenium] cation by a vinylene moiety. The strong similarity between
solid-state and solution infrared and Raman spectra of [BF4]-, [PF6]-, and [BAr′4]- salts of [Rc(CH)3Rc]+
indicates that the C-C stretching constant in the allylium chain and, therefore, the structure, of this ion are
largely independent of the local environment, suggesting that the unsymmetrical structures observed in
the crystal structures are not simply an artifact of packing. Differences in the solvatochromism of [Rc-
(CH)3Rc]+ and [Fc(CH)3Fc]+ also suggest a localized structure for the former cation in solution.
Electrochemistry, UV-visible-NIR spectroscopy, and DF calculations give insight into the electronic structure
of the metallocene-terminated allylium cations. Using an analogy between polymethines and mixed-valence
compounds, the difference between the behaviors of [Fc(CH)3Fc]+ and [Rc(CH)3Rc]+ is attributed to larger
reorganization energy associated with the geometry differences between metallocene and [(η6-fulvene)-
(η5-cyclopentadienyl)metal] structures in the ruthenium case.
Introduction
Typical polymethine cations or anions (cyanines) have C2V-
symmetric delocalized structures in which the C-C bond lengths
are all approximately equal; the structure of a simple poly-
methine is well described as shown in Figure 1a; i.e., with equal
contributions from two limiting resonance forms. However, it
is predicted1-3 that at sufficiently long chain lengths polyme-
thines are subject to the Peierls distortion.4 Where the end group
can stabilize the overall charge, the structure of a Peierls-
distorted polymethine is unsymmetrical with a polyene-like
bond-length alternation (BLA) between formally single and
* Author to whom correspondence should be addressed (University of
Arizona). E-mail: sbarlow@email.arizona.edu.
† Beckman Institute, California Institute of Technology.
‡ University of Oxford.
Figure 1. Representations of the structures of (a) a typical polymethine,
(b) a Peierls-distorted polymethine, and (c) the Peierls-distorted polymethine
reported by Tolbert and Zhao.5
§ University of Arizona.
| Jet Propulsion Laboratory, California Institute of Technology.
(1) Kuhn, C. Synth. Met. 1991, 41-43, 3681-3688.
(2) Craw, J. S.; Reimers, J. R.; Bacskay, G. B.; Wong, A. T.; Hush, N. S.
Chem. Phys. 1992, 167, 77-99.
(3) Tolbert, L. M.; Ogle, M. E. Synth. Met. 1992, 391-396.
(4) Peierls, R. E. Quantum Theory of Solids; OUP: Oxford, 1955.
double bonds; it is now dominated by one of the two extreme
resonance forms (Figure 1b). Tolbert and Zhao reported the first
experimental evidence for a localized polymethine; the com-
9
10.1021/ja012472z CCC: $22.00 © 2002 American Chemical Society
J. AM. CHEM. SOC. 2002, 124, 6285-6296
6285

A R T I C L E S
Barlow et al.
pound shown in Figure 1c was shown to be unsymmetrical by
comparison of its IR and UV-visible-NIR spectra with those
of symmetrical lower homologues.5 The localization to an
unsymmetrical structure can be understood by reference to
mixed-valence chemistry;5,6 symmetrical and Peierls-distorted
polymethines are analogous to Robin and Day7 class III and
class II systems, respectively.
This paper is concerned with the effect of the end group on
localization in polymethines. Work by Tolbert3,8,9 has distin-
guished between organic polymethines where the end groups
can stabilize the overall charge and those where they cannot.
In the former case, the distortion localizes the charge on one
end, as shown in Figure 1b. In the latter case, for example, in
phenyl-terminated polymethine anions, the charge is centered
in the middle of the chain, extending over approximately 15
carbon atoms in both directions. Similar differences in behavior
between amino-terminated polyene/polymethine cations and
phenyl-terminated polyene anions were predicted by Hush and
co-workers.2,10 We were interested in whether the choice of
charge-stabilizing end group could affect the chain length at
which localization occurred, as might be anticipated from the
previously mentioned analogy with mixed-valence chemistry.
The classification of a mixed-valence species in the class I/II/
III spectrum is determined by the interplay of electronic coupling
between the two alternative localized structures, V, and the
vertical reorganization energy of Marcus theory, λ.11,12 Thus,
increased λ should cause a polymethine to localize at shorter
chain length. One contribution to λ is the energy required to
invert the sense of the BLA of the localized polymethine chain,
λbla; another is the energy required to interconvert the geometries
of the charged and uncharged (e.g., H2N+ ) vs H2N-) forms
of the end groups, λend. In the case of various organic donors
or acceptors, and for many organometallic donors, this end group
effect is limited to changing one or two bond lengths. The good
conjugation between the end group and the polymethine chain
leads to V sufficiently large to dominate until rather large chain
length. The group 8 metallocenes might be expected to have
somewhat different reorganization-energy characteristics. The
resonance structure important in stabilizing carbocations R to
metallocenes is considered to be the [(η6-fulvene)(η5-cyclopen-
tadienyl)metal] cation (Figure 2a).13 These structures differ
significantly in geometry from the neutral metallocenes, the exo
C-C bond being distorted from the plane of the fulvene to allow
interaction of the R-carbon with the metal atom.14-18 We were
Figure 2. (a) Stabilization of a metallocenyl carbocation by resonance with
a [(η6-fulvene)(η5-cyclopentadienyl)metal] cation, (b) a [1.1]metallo-
cenophane carbocation, and (c) a crystallographically characterized trime-
tallic termethine reported by Schottenberger, Bildstein, and co-workers.27
Chart 1
interested to see whether these large structural differences would
lead to a large λend; moreover, differences between iron and
ruthenium systems offered the possibility of tuning λend. Previous
work on metallocene-terminated polymethines has been largely
limited to ferrocene compounds, [Fc(CH)nFc]+. The n ) 1
member, the bis(ferrocenyl)methylium ion, has been the subject
of several studies,19-22 including a crystal structure determina-
tion showing an essentially symmetrical cation in the tetrafluo-
roborate.23 The cations formed by hydride abstraction from the
bridges of [1.1]ferrocenophanes (Figure 2b)24 can be regarded
as derivatives of [FcCHFc]+. The [Fc(CH)3Fc]+, 1+, cation
(Chart 1) has also been reported on several occasions,22,25-27
and the closely related cation shown in Figure 2c has been
shown to be essentially symmetrical in the crystal structure of
its tetrafluoroborate.27 Tolbert and co-workers synthesized a
series of [Fc(CH)nFc]+[BF4]- with n ) 1, 3, 5, 9, 13 and
reported electrochemical data and the low-energy UV-visible-
NIR absorption maxima.22 We recently reported that the [Rc-
(5) Tolbert, L. M.; Zhao, X. J. Am. Chem. Soc. 1997, 119, 3253-3258.
(6) Nelsen, S. F.; Tran, H. Q.; Nagy, M. A. J. Am. Chem. Soc. 1998, 120,
298-304.
(7) Robin, M. B.; Day, P. AdV. Inorg. Chem. Radiochem. 1967, 10, 247-422.
(8) Tolbert, L. M.; Ogle, M. E. J. Am. Chem. Soc. 1990, 112, 9519-9527.
(9) Tolbert, L. M. Acc. Chem. Res. 1992, 25, 561-568 and references therein.
(10) Craw, J. S.; Reimers, J. R.; Bacskay, G. B.; Wong, A. T.; Hush, N. S.
Chem. Phys. 1992, 167, 101-109.
(11) Hush, N. S. Chem. Phys. 1975, 10, 361-362.
(12) It should also be noted that class II-type behavior is also favored by
introduction of formal asymmetry into the system; i.e., by making the two
extreme structures unequal in energy.
(19) Boev, V. I.; Dombrovskii, A. V. Zh. Obshch. Khim. 1980, 50, 2520-2525.
(20) Jutz, C. Tetrahedron Lett. 1959, 1-4.
(21) Bildstein, B.; Denifl, P.; Wurst, K. J. Organomet. Chem. 1995, 496, 175-
186.
(13) Watts, W. E. In ComprehensiVe Organometallic Chemistry; Wilkinson, G.,
Stone, F. G. A., Abel, E. W., Eds.; Pergamon: London, 1988; Vol. 8.
(14) Behrens, U. J. Organomet. Chem. 1979, 182, 89-98.
(15) Yanovsky, A. I.; Struchkov, Y. T.; Kreindlin, A. Z.; Rybinskaya, M. I. J.
Organomet. Chem. 1989, 369, 125-130.
(22) Tolbert, L. M.; Zhao, X.; Ding, Y.; Bottomley, L. A. J. Am. Chem. Soc.
1995, 117, 12891-12892.
(23) Cais, M.; Dani, S.; Herbstein, F. H.; Kapon, M. J. Am. Chem. Soc. 1978,
100, 5554-5558.
(24) Mueller-Westerhoff, U. T.; Nazzal, A.; Pro¨ssdorf, W.; Meyerle, J. J.; Collins,
R. L. Angew. Chem., Int. Ed. Eng. 1982, 21, 293-294.
(25) Bunton, C. A.; Carrasco, N.; Watts, W. E. J. Chem. Soc., Perkin Trans. 2
1979, 1267-1273.
(26) Boev, V. I.; Dombrovskii, A. V. Zh. Obshch. Khim. 1985, 55, 1173-1177.
(27) Lukasser, J.; Angleitner, H.; Schottenberger, H.; Kopacka, H.; Schweiger,
M.; Bildstein, B.; Ongania, K.-H.; Wurst, K. Organometallics 1995, 14,
5566-5578.
(16) Rybinskaya, M. I.; Kreindlin, A. Z.; Struchkov, Y. T.; Yanovsky, A. I. J.
Organomet. Chem. 1989, 359, 233-243.
(17) Kreindlin, A. Z.; Dolgushin, F. M.; Yanovsky, A. I.; Kerzina, Z. A.;
Petrovskii, P. V.; Rybinskaya, M. I. J. Organomet. Chem. 2000, 616, 106-
111.
(18) Barlow, S.; Cowley, A. R.; Green, J. C.; Brunker, T. J.; Hascall, T.
Organometallics 2001, 20, 5351-5359.
9
6286 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002

Metallocene-Terminated Allylium Salts
A R T I C L E S
Scheme 1
Figure 3. Schematic showing the definitions of some of the parameters
given in Table 1 and used in discussing the crystal structures of the 1,3-
di(metallocenyl)allylium cations.
(CH)3Rc]+, 3+, cation is markedly unsymmetrical in the crystal
structure of its hexafluorophosphate, 3b.28 In this paper, we
present evidence supporting an unsymmetrical structure for the
3+ cation in other salts and in solution and report a comparative
structural, electrochemical, and spectroscopic study of sym-
metrical and unsymmetrical allylium (termethine) cations with
ferrocenyl (Fc), 2,3,4,5,1′,2′,3′,4′-octamethylferrocen-1-yl (Fc′′)
and ruthenocenyl (Rc) capping groups. The cations studied (1+-
5+), and their salts (1a-5a, 1b-3b, 5b, 1c, 3c), are defined in
Chart 1.
Figure 4. View of the 2+ cation, [Fc′′(CH)3Fc′′]+, in the crystal structure
of its hexafluorophosphate, 2b. Non-hydrogen atoms are represented by
50% probability ellipsoids, while hydrogens are shown as spheres of
arbitrary radius.
Results and Discussion
unsymmetrical species 4a and 5b. Some important bond lengths
and angles, defined in Figure 3, are compared in Table 1. Details
of the structure determinations are given in Table 4, the
Experimental Section, and the Supporting Information.
The apparent bond lengths in the allylium bridge of 1b are
affected by disorder. However, both ferrocenyl end groups
appear to show comparable distortions from ideal ferrocene
geometry; both the magnitude and slight asymmetry of these
distortions are comparable with those found in the structures
of [FcCHFc]+[BF4]-23 and of the tetrafluoroborate salt of the
cation shown in Figure 2c.27 In both of these previous structures,
and in that of a [1.1]ferrocenophane-based cation24 (Figure 2b)
the C-C bond lengths indicate symmetrical bridging groups.
Thus, it seems likely that 1b is a typical, class III-like,
polymethine. The 1+ cation adopts a syn conformation in the
crystal of 1b (the cation shown in Figure 2c is also syn), with
both iron atoms on the same face of the bridging ligand, giving
the cation approximate Cs symmetry.
The structure of the 2+ cation in 2b (Figure 4) is clearly
delocalized (class III-like); the bond lengths in the allylium
bridge are not affected by disorder and clearly indicate an
absence of BLA. The two octamethylferrocenyl groups show
similar distortions, less extensive than those in the recently
reported nonamethylferrocenylmethylium cation, where values
of 2.567(12) Å and 23° were found for the Fe-CR bond length
and â, respectively, in the crystal structure of its [BAr′4]- salt.17
The distortions of the end groups are less marked than those in
the structures of 1b and of previously reported di(ferrocenyl)
cations,23,24,27 presumably reflecting the role played by the
electron-donating methyl groups in stabilizing the allylium
cation, reducing the need for donation from the metal. The 2+
cation has approximate C2 (rather than Cs) symmetry in the
structure of 2b; i.e., it adopts an anti conformation in which
the two metallocene moieties are twisted away from one another
(as might be expected from steric considerations); in all the other
structures, the two metallocenes are syn.
Synthesis. A range of metallocene-terminated allylium
[BF4]-, [PF6]-, and [BAr′4]- {Ar′ ) 3,5-bis(trifluoromethyl)-
phenyl} salts (see Chart 1) was synthesized as shown in Scheme
1. The known compound 1-ferrocenyl-2-bromoethene29 and its
new octamethylferrocenyl and ruthenocenyl analogues were
readily obtained from the Wittig reaction between the corre-
sponding aldehydes and bromomethyltriphenylphosphonium
bromide,30 some elimination to the alkyne31,32 being observed
in the ruthenium case. Bromine-lithium exchange, followed
by reaction with metallocenyl aldehydes, afforded 1,3-dimet-
allocenylprop-1-ene-3-ols. (E)-FcCHdCHCH(OH)Fc was previ-
ously synthesized from (E)-FcCHdCHC(O)Fc;25-27 since it has
been found to be unstable with respect to reoxidation to the
ketone,27 and since pure products are obtained from the crude
alcohols in the next step, we used our alcohols without full
purification and characterization. Pure tetrafluoroborate and
hexafluophosphate salts of the 1,3-di(metallocenyl)allylium
cations were obtained after treatment with the appropriate acid,
followed by crystallization from dichloromethane layered with
diethyl ether.33 Two [BAr′4]- {Ar′ ) 3,5-(CF3)2C6H3} salts, 1c
and 3c, were synthesized using the metathesis reaction between
the corresponding hexafluorophosphates and [Na]+[BAr′4]-. The
salts were studied by X-ray crystallography, cyclic voltammetry,
and by electronic, vibrational, and NMR spectroscopy
Solid-State Structures. We have determined the crystal
structures of 1b, 2b, 3b, and 3c, which contain symmetrically
substituted allylium cations, as well as those of the formally
(28) Barlow, S.; Henling, L. M.; Day, M. W.; Marder, S. R. Chem. Commun.
1999, 1567-1568.
(29) FcCHdCHBr has also been obtained from the reaction of FcCHdCHCO2H
and N-bromosuccinimide (Naskar, D.; Das, S. K.; Giribabu, L.; Maiya, B.
G.; Roy, S. Organometallics 2000, 19, 1464-1469), and from the reaction
of FcCHdCHSnPh3 and bromine (Nesmeyanov, A.; Borisov, A. E.;
Novikova, N. V. IzV. Akad. Nauk. SSSR, Ser. Khim. 1972, 1372-1375).
(30) Driscoll, J. S.; Grisely, D. W.; Pustinger, J. V.; Harris, J. E.; Matthews, C.
N. J. Org. Chem. 1964, 29, 2427-2431.
(31) Hofer, O.; Schlo¨gl, K. J. Organomet. Chem. 1968, 13, 443-456.
(32) Rausch, M. D.; Siegel, A. J. Org. Chem. 1969, 34, 1974-1976.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6287

A R T I C L E S
Barlow et al.
ruthenocenyl carbocation structures (ranges 2.270(3)-2.571(4)
Å, 29.1-42.6°, and 7.1-12.9°, respectively,15,18,37-39 excluding
values for the severely disordered structure of [Cp*Ru(η6-C5-
Me4CH2)]+[PF6]- 40) and with parameters obtained in a DFT
optimization of [RcCH2]+ (2.025-2.254 Å, 2.263 Å, 5.3°, and
41.2° for Ru-CCp, Ru-CR, R, and â, respectively).18,41 In
contrast, the Ru2 ruthenocene has “normal” ruthenocene
geometry; the Ru-CCp distances fall in ranges similar to those
for ruthenocene itself (2.181(2)-2.188(2) Å)42 and for its
derivatives such as (E)-1,2-dimethyl-1,2-diruthenocenylethene
(2.171(3)-2.216(3) Å).43 The R-carbon, C13, is very slightly
bent away from Ru2 with a very long and, therefore, nonbonded
Ru-CR distance. The allylium bridge shows significant BLA;
the formally double and single CH-CH bonds differ in length
by 0.100(6) Å. The structure can, therefore, be described as
ruthenocene and [(η6-fulvene)(η5-cyclopentadienyl)ruthenium]+
moieties linked by an vinylene bridge.
Figure 5. View of the 4+ cation, [Fc(CH)3Fc′′]+, in the crystal structure
of its tetrafluoroborate, 4a. Non-hydrogen atoms are represented by 50%
probability ellipsoids, while hydrogens are shown as spheres of arbitrary
radius.
The structure of 5b is isomorphous with that of 3b; the 5+
cation is composed of a “normal” ferrocene with a vinylene
bridge to a [(η6-fulvene)(η5-cyclopentadienyl)ruthenium] unit.
Thus, ruthenocene is clearly more effective than ferrocene in
stabilizing the positive charge in these systems, in accord with
what is known about the stability of simple metallocenyl
carbocations,44 with electrochemical and spectroscopic data
(vide infra) and with the previously reported structures of
[FcCHRc]+[PF6]-37 and of salts of a mixed iron-ruthenium
[1.1]metallocenophane cationic derivative.37,38 The main dif-
ferences from the structure of 3b, apart from the expected
differences in the metal-carbon bond lengths in the “normal”
metallocene unit, are in the disorder of the hexafluorophosphate
counterions. The bond lengths in the allylium bridge of 5b are
identical, within experimental error, to the corresponding values
for 3b, as is the Ru1-CR bond length.
An important question that arises is whether the unsym-
metrical structure of the cation in 3b arises from the influence
of crystal-packing effects. The [(η6-fulvene)(η5-cyclopentadi-
enyl)ruthenium]+ moiety of 3b is approached more closely by
hexafluorophosphate anions than the “normal” ruthenocene. Is
this due to the effect of the unsymmetrical cation on the crystal
packing, or does this packing determine the cation symmetry?
Evidence for the former hypothesis is provided by comparison
of the packing diagrams for 1b and 3b, which are quite similar
(Figure 7). Moreover, such severe effects of counterion upon
polymethines have not previously been reported,36 although
effects of counterion and packing motif upon the intramolecular
electron-transfer rate in class II mixed-valence compounds are
Figure 6. View of the 3+ cation, [Rc(CH)3Rc]+, in the crystal structure of
its hexafluorophosphate, 3b. Non-hydrogen atoms are represented by 50%
probability ellipsoids, while hydrogens are shown as spheres of arbitrary
radius.
The unsymmetrical species 4a also adopts a delocalized
structure (Figure 5); the BLA of 0.023(9) Å is insignificant at
the 3σ level. However, the data do show the octamethylferrocene
moiety to be more distorted than the ferrocene, reflecting a
greater contribution to stabilization of the charge from its higher
lying HOMO. Thus, the 4+ cation seems best described as a
hybrid of the Fc′′+)CHsCHdCHsFc and Fc′′sCHdCHs
CHdFc+ resonance structures, but with the former structure
being somewhat more important due to its lower energy. The
appropriate mixed-valence analogy is an inherently unsym-
metrical class III species (the appropriate potential well diagram
is shown in Figure 4c of ref 34).
The remaining structures are localized. 3b (Figure 6) is the
first crystallographically characterized Peierls-distorted poly-
methine.35,36 One of the two ruthenocenes (that containing Ru1)
is best described as a [(η6-fulvene)(η5-cyclopentadienyl)ruth-
enium]+ moiety; the values of M1-CR1, R1, and â1 in Table 1
are similar to the corresponding values of previously reported
(36) Although one crystal structure of an organic cyanine has been claimed as
evidence for unsymmetrical localization (Borowiak, T. E.; Bokii, N. G.;
Struchkov, Y. T. Zh. Strukt. Khim. 1972, 13, 480), Smith has pointed out
that the apparent bond-length alternation is insignificant, given the large
standard deviations for the relevant interatomic distances (Smith, D. L.
Photogr. Sci. Eng. 1974, 18, 309-322).
(33) Attempts to synthesize the tetrafluoroborate of the mixed octamethylfer-
rocenyl/ruthenocenyl species led to a mixture of [Fc′′(CH)3Rc]+[BF4]- and
[Fc′′(CH)3Rc]2+([BF4]-)2, as shown by elemental analysis and by a crystal
structure determination for the dicationic species (Barlow, S.; Day, M. W.;
Marder, S. R. Acta Crystallogr. 2000, C56, 303-304). The electrochemical
data for the reaction product do indeed indicate a more easily oxidized
iron center than any of the other species studied.
(37) Watanabe, M.; Motoyama, I.; Takayama, T. Bull. Chem. Soc. Jpn. 1996,
69, 2877-2884.
(38) Watanabe, M.; Moyoyama, I.; Takayama, T. J. Organomet. Chem 1996,
524, 9-18.
(34) Creutz, C. Prog. Inorg. Chem. 1983, 30, 1-73.
(39) Sato, M.; Kawata, Y.; Kudo, A.; Iwai, A.; Saitoh, H.; Ochiai, S. J. Chem.
Soc., Dalton Trans. 1998, 2215-2224.
(35) Da¨hne and Reck have reported a more subtle asymmetry (bond-length
alternation of ∼0.01 Å) in the crystal structure of [Me2N(CH)7NMe2]+[BPh4]-
(Da¨hne, L.; Reck, G. Angew. Chem., Int. Ed. Engl. 1995, 34, 690-692).
An even more subtle asymmetry is reported in the structure of the
corresponding picrate salt (Da¨hne, L.; Reck, G. Z. Kristallogr. 1995, 210,
948-951). This effect is absent with inorganic counterions, even when, as
in the [BF4]- salt (Da¨hne, L.; Grahn, W.; Jones, P. G.; Chrapkowski, A. Z.
Kristallogr. 1994, 209, 54-516), they are unsymmetrically arranged relative
to the cations and has, therefore, been attributed to polarization of the
polymethine π-system by interactions with the aromatic rings of the anion.
(40) Kreindlin, A. Z.; Petrovskii, P. V.; Rybinskaya, M. I.; Yanovskii, A. I.;
Struchkov, Y. T. J. Organomet. Chem. 1987, 319, 229-237.
(41) A similar geometry was found in a MNDO/AM1 study: Gal’pern, E. G.;
Gambaryan, N. P.; Kreindlin, A. Z.; Rybinskaya, M. I.; Stankevich, I. V.;
Chistyakov, A. L. Metalloorg. Khim. 1992, 5, 831-838.
(42) Seiler, P.; Dunitz, J. D. Acta Crystallogr. 1980, B36, 2946-2950.
(43) Chiu, C.-F.; Song, M.; Chen, B.-H.; Kwan, K. S. Inorg. Chim. Acta 1997,
266, 73-79.
(44) Hill, E. A.; Richards, J. H. J. Am. Chem. Soc. 1961, 83, 3840-3846.
9
6288 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002

Metallocene-Terminated Allylium Salts
A R T I C L E S
Table 1. Selected Crystallographically Determined Geometric Parameters (Defined in Figure 3) for 1,3-Di(metallocenyl)allylium Salts
1b
2b
3b
3c
4a
5b
(M1 ) Fe1,
M2 ) Fe2)
(M1 ) Fe1,
M2 ) Fe2)
(M1 ) Ru1,
M2 ) Ru2)
(M1 ) Ru1,
M2 ) Ru2)
(M1 ) Fe1 {Fc},
M2 ) Fe2 {Fc′′})
(M1 ) Ru,
M2 ) Fe)
M1-CCp1a/Å
M2-CCp2a/Å
M1-CR1b/Å
M2-CR2b/Å
2.001(6)-2.064(6)
2.009(6)-2.058(7)
2.834(7)
2.032(3)-2.079(3)
2.048(3)-2.076(3)
3.024(3)
3.031(3)
1.379(4)
1.380(4)
3.8(2)
6.2(2)
4.5
4.7
2.089(2)-2.234(3)
2.167(2)-2.194(3)
2.381(3)
3.237(3)
1.443(4)
1.343(4)
9.0(2)
4.0(2)
38.2
-1.0
2.100(7)-2.221(7)
2.139(7)-2.199(7)
2.604(8)
2.010(5)-2.057(5)
2.028(4)-2.093(4)
3.017(4)
2.837(4)
1.380(6)
1.403(6)
1.4(2)
8.2(2)
5.0
14.5
2.080(3)-2.217(4)
2.038(4)-2.065(4)
2.385(4)
3.151(4)
1.444(5)
1.344(5)
9.1(2)
4.2(3)
37.9
-0.4
2.933(7)
3.104(8)
c
CR1-Câ /Å
1.469(10)f
1.461(10)f
4.8(4)
1.452(10)
1.372(10)
6.6(7)
1.8(7)
28.2
c
CR2-Câ /Å
d
R1 /deg
d
R2 /deg
13.4(4)
13.2
9.1
e
â1 /deg
e
â2 /deg
4.5
a Range of bond lengths within the metallocene. b Distance from metal atom to the allylium carbon R to the metallocene. c Bond length from the allylium
carbon R to the metallocene to the â-carbon (the central atom of the bridge). d Ring tilt; angle between the planes of the five-membered rings of the metallocene.
e Angle between the Cipso-CR bond and the plane of the five-membered ring to which the R-carbon is bound. f Values severely affected by unmodeled
disorder (see Experimental Section).
Figure 8. View of the 3+ cation, [Rc(CH)3Rc]+, in the crystal structure of
its [BAr′4]- salt, 3c. Non-hydrogen atoms are represented by 50% probability
ellipsoids, while hydrogens are shown as spheres of arbitrary radius.
3+ cation in both 3b and 3c suggests that the localization is not
significantly dependent upon the nature of the counterions.
UV-Visible-NIR Spectroscopy and Orbital Structure.
UV-visible-NIR spectra were recorded for the 1,3-di(metal-
locenyl)allylium salts in a variety of solvents, the use of the
[BAr′4]- ion enabling spectra of the 1+ and 3+ cations to be
acquired in some solvents of low polarity. For a given cation,
essentially identical spectra were obtained for [BF4]-, [PF6]-,
and [BAr′4]- salts. Our data for 1+ salts were similar to the
published data.27
The UV-visible-NIR spectra of the metallocene-terminated
allylium cations (Figure 9) resemble the spectra of metallocene-
(π-bridge)-acceptor second-order nonlinear optical (NLO)
dyes;45 both classes of compounds show two prominent absorp-
tions, hereafter referred to as the low-energy (LE) and high-
energy (HE) absorptions. The orbital structures obtained for 1+
Figure 7. Packing diagrams for (a) 1b and (b) 3b.
and 3+ cations using DF calculations based upon the crystal-
well known. Nevertheless, we further investigated the influence
lographically determined geometries (details of which are given
in the Experimental Section) support this analogy. As in the
NLO dyes,46 the highest occupied orbitals are metal d-based
orbitals, below which is an orbital (π) composed of an out-of-
phase combination of the HOMOs of the cyclopentadienyl and
π-bridge fragments (π is the HOMO-6 for a symmetrical
structure, where the filled d-orbitals of both metallocenes are
higher in energy, and the HOMO-3 for unsymmetrical cations,
where the three highest occupied orbitals are all localized on
the “normal” metallocene). The LUMO has π* character.
of the counterion by crystallizing the 3+ cation with the
weaklycoordinating, delocalized [BAr′4]- {Ar′ ) 3,5-(CF3)2C6H3}
anion. The structure of the 3+ cation in 3c (Figure 8) is similar
to that in 3b. First, there is a clear BLA of 0.080(14) Å in the
allylium bridge (indistinguishable at the 3σ level from that in
3b). Second, only one of the end groups shows distortion toward
the η6-fulvene structure; although this distortion is a little less
severe than that in 3b, with somewhat longer Ru-CR distance
and a smaller â, these parameters are very close to the range
for previously reported ruthenocenyl carbocation structures and
â and Ru-CR have previously been shown to vary with crystal
packing in salts of [RcCH2]+18 and of a mixed iron-ruthenium
[1.1]metallocenophane cation.37,38 The localized structure of the
(45) Barlow, S.; Marder, S. R. Chem. Commun. 2000, 1555-1562.
(46) Barlow, S.; Bunting, H. E.; Ringham, C.; Green, J. C.; Bublitz, G. U.;
Boxer, S. G.; Perry, J. W.; Marder, S. R. J. Am. Chem. Soc. 1999, 121,
3715-3723.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6289

A R T I C L E S
Barlow et al.
Figure 9. UV-visible-NIR spectra of bis(metallocenyl)allylium cations
in dichloromethane.
Isosurfaces for π and π* of the two ions are shown in Figure
10; these are clearly similar to the HOMO-3 and LUMO of the
NLO dyes (Figure 5 of ref 46). In the case of the unsymmetrical
cation, the LUMO is more localized toward the formally cationic
end of the molecule, and the highest π orbital is localized toward
the “normal” metallocene end. Nonetheless, Figure 10 shows
that the important orbitals do not change very dramatically in
form or extent on breaking the symmetry. In analogy with the
assignment of the spectra of the NLO dyes,46 and consistent
with our orbital scheme, we assign the LE band of the
metallocene allylium species to a metal-to-π* transition and the
HE band to a π-to-π* transition.
The assignment is further supported by the differences in the
spectra of different metallocene polymethines. Comparing the
two absorption maxima of [FcCHFc]+,19 1+, and [Fc(CH)5Fc]+47
cations shows the energy of the HE band to be more sensitive
to chain length,45 consistent with our assignment, which predicts
that the energy of the HE band will depend on the energies of
both π (which will be raised by increasing chain length) and
π* (which will be lowered), while the LE band should depend
on the energies of metal-based d-orbitals (more-or-less un-
changed with chain length) and on π*. Moreover, the variation
of the energy of the HE band with effective conjugation length
closely parallels that for the [Ph(CH)nPh]+ series, as shown in
Figure 7 of ref 45. In 2+, methylation should raise both the
filled metal d-orbitals and the highest filled π orbital, consistent
with the observed red shifts of both bands. Replacement of Fe
by Ru leads to blue shifts of 0.35 and 1.07 eV for the HE and
LE bands, respectively. A metal-to-π* band should be more
sensitive to metal identity than a π-to-π* band, consistent with
our assignment (ruthenocene has a lower energy HOMO than
ferrocene48). The shift in the HE band partly reflects some metal
contribution to π and π* but will also be affected if, as proposed,
3+ has an unsymmetrical structure.
Figure 10. DF molecular orbitals for the 1+s(a) the π* LUMO and (b)
the highest lying π orbitalsand 3+ cationss(c) π* and (d) π.
Indeed the π-to-π* transition of the compound shown in Figure
1c showed a much stronger variation of νjmax with 1/n2 - 1/D,
where n and D are the solvent refractive index and dielectric
constant, respectively, than those of shorter homologues.5 The
solvent dependence indicates a decrease or a reversal in dipole
moment between ground and excited state; this is due to the
greater charge-transfer character of the transition due to
localization of the π HOMO toward the quinoidal end of the
molecule and of the π* LUMO toward the formally cationic
end as shown in Figure 8 of ref 5 (in this respect, the Peierls-
distorted polymethine resembles a donor-acceptor polyene).
Similarly, our calculations show that the donor orbitalssboth
metal-based and π-basedsand the π* acceptor orbital of an
unsymmetrical di(metallocenyl)allylium cation are more local-
ized toward “normal” and formally cationic ends, respectively,
than in a symmetrical species and, therefore, that both the metal-
to-π* and π-to-π* transitions should have more charge-transfer
Since the intervalence charge-transfer spectra of class II and
class III mixed-valence compounds show different patterns of
solvatochromism,34 similar differences are anticipated between
some absorptions of localized and delocalized polymethines.
(47) Barlow, S.; Perrins, L. Unpublished results.
(48) Cauletti, C.; Green, J. C.; Kelly, M. R.; Robbins, J.; Smart, J. C. J. Electron
Spectrosc. Relat. Phenom. 1980, 19, 327-353.
9
6290 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002

Metallocene-Terminated Allylium Salts
A R T I C L E S
Table 2. Selected IR Data for 1,3-Di(metallocenyl)allylium Salts in
the Solid State and in Dichloromethane Solution
ν¯allylium/cm-1
compd
KBr
CH Cl2
2
1a
1b
1c
2a
2b
3a
3b
3c
4a
5a
5b
1548
1558
1561
1548
1538
1603
1609
1584
1553
1598
1598
1560
1558
1555
1538
1542
1601
1598
1600
1556
1596
1599
in solution and in its crystalline [BF4]- salt is the same as it is
in crystals of its [PF6]- and [BAr′4]- salts (where the cation is
known to be unsymmetrical by X-ray crystallography, vide
supra). Moreover, 3+ salts show bands (1584-1603 cm-1) at
similar frequency to those of the 5+ cation (1596-1599 cm-1),
which is markedly unsymmetrical in both the crystal structure
Figure 11. Solvatochromism of the 1+ and 3+ cations. The scatter in the
data for the high-energy band of the 3 cation reflects the difficulty of
accurately determining the peak position, due to overlap with the low-energy
band. The solvents used were, in order of increasing 1/n2 - 1/D: benzene,
1,4-dioxane, Et2O, EtOAc, THF, PhCN, DMSO, DMF, acetone, MeCN,
and MeOH.
1
of its hexafluorophosphate and in solution (according to H
NMR spectroscopy, vide infra). The symmetrical 1+ (1548-
1560 cm-1) and 2+ (1538-1561 cm-1) cations show allylium
bands significantly lower in frequency than the 3+ and 5+
character. The solvent dependence of the spectra of 3+ was,
therefore, compared with that of 1+. Figure 11 compares the
variation of both LE and HE maxima for 1+ and 3+ with the
polarity of a range of nonhalogenated solvents.49 The stronger
dependence of νjmax with 1/n2 - 1/D clearly supports an
unsymmetrical solution structure for 3+. In addition, it can be
seen from Figure 9 that the absorption bands for 3+ and 5+ are
somewhat broader than those for 1+, 2+, and 4+, also consistent
with class II-like behavior for the first two cations; the
unsymmetrical cation in Figure 1c was also reported to show a
broader π-to-π* transition than its lower symmetrical homo-
logues.5
IR Spectroscopy. Infrared spectroscopy of salts of the 1,3-
di(metallocenyl)allylium salts reveals a strong band in the range
∼1550-1600 cm-1 (Table 2), which we assign to a stretching
mode of the multiply bonded allylium bridge (neither ru-
thenocene, the counterions used, nor the [(η6-fulvene)(η5-
cyclopentadienyl)ruthenium] cation18 shows a band in this
region). Despite complications arising from absorption of the
laser fundamental, we were also able to observe this band in
the Raman spectra of a number of compounds (see Experimental
Section). Calculations by Reimers and Hush predict that
symmetrical and unsymmetrical (Peierls-distorted) forms of
[NH2(CH)nNH2]+ (n ) 19, 25) have different vibrational
frequencies.50 The insensitivity of the allylium band of the 3+
cation to counterion and to medium suggests that its structure
cations. The formally unsymmetrical 4+ cation (1553-6 cm-1
)
shows a band at similar frequency to the delocalized 1+ and 2+
cations. This is consistent with both crystallographic and NMR
data, both of which show considerably less BLA in the allylium
bridge of the 4+ cation than in that of the 5+ cation.
NMR Spectroscopy. The three nominally symmetrical al-
1
lylium cations, 1+, 2+, and 3+, all show room-temperature H
and 13C NMR spectra consistent with symmetrical structures.
The observation of a symmetrical solution structure for 3+ is
not particularly surprising since the cation is likely to be
fluxional on the NMR time scale (which is, of course, much
longer than the time scales for electronic and vibrational
spectroscopies); the Peierls-distorted polymethine (Figure 1c)
previously reported by Tolbert and Zhao is also symmetrical
on the NMR time scale.5 We attempted to “freeze out” the
interconversion between the two alternative unsymmetrical
structures of 3+ by recording spectra for 3a in dichloromethane-
1
d2 solution at temperatures down to -85 °C (500 MHz for H
and 125 MHz for 13C). Although some chemical shifts changed
considerably and some of the resonances corresponding to the
ruthenocenyl moieties were very considerably broadened in both
1H and 13C spectra, there was relatively little broadening of
bridge 1H and 13C resonances and the 1H-1H coupling constant
associated with the allylium bridge was unchanged. This
suggests that the process being affected is not the interconversion
of the two extreme structures; an alternative possibility is that
we are observing slowing of rotation of the ruthenocenyl end
groups about the Rc-CR bonds. The failure to observe localiza-
tion by NMR allows the rate of interconversion between the
two extreme structures to be estimated as in excess of 4000 s-1
down to -85 °C, corresponding to a barrier of less than 32.3
kJ mol-1 (≡ 0.33 eV ≡ 2700 cm-1).51
(49) If halogenated solvents are included, considerably more scatter is introduced
to the plots for both 1+ and 3+ cations, although the best-fit lines are still
steeper for the diruthenium species. In particular, absorption maxima for
all our metallocenyl allylium cations are considerably more red-shifted in
dichloromethane than might be expected from simple solvent polarity or
polarizability parameters. Similar behavior has been observed for charge-
transfer-type bands in other metallocene systems (for example, see data
and graph in the Supporting Information for ref 46, and Figure 4 in: Barlow,
S. Inorg. Chem. 2001, 40, 7047-7053) and may indicate some sort of
specific interaction between metallocenes and chlorinated solvents. Indeed,
it has been found that νmax for the intervalence charge-transfer transitions
of [FcNdNFc]+ do not correlate well with 1/n2 - 1/D for chlorinated
solvents but that better correlation is found if νmax is plotted against 150-
(1/n2 - 1/D) - (AN - DN), where AN and DN are the solvent acceptor
and donor numbers, respectively: Kurosawa, M.; Nankawa, T.; Matsuda,
T.; Kubo, K.; Kurihara, M.; Nishihara, H. Inorg. Chem. 1999, 38, 5113-
5123.
The mixed ferrocenyl/ruthenocenyl species, 5+, appears to
1
have an unsymmetrical bridge according to H-1H coupling
constants (12.2, 14.5 Hz), consistent with the BLA seen in the
crystal structure of 5b (vide supra). In the octamethylferrocenyl/
(50) Reimers, J. R.; Hush, N. S. Chem. Phys. 1993, 176, 407-420.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6291

A R T I C L E S
Barlow et al.
Table 3. Electrochemical Half-Wave Potentials (mV, vs Ferrocenium/Ferrocene) for Symmetrical and Unsymmetrical
1,3-Di(metallocenyl)allylium Cations in THF and Dichloromethane (∼10-4 M 1a-5a, 0.1 M [Bu4N]+[PF6]-, 50 mV s-1
)
THF
CH Cl2
2
+
+
a
2+ +b
3+ 2+c
+
a
2+ +b
3+ 2+c
cation, C
C /C
C /C
C /C
C /C
C /C
C /C
1+
2+
3+
4+
5+
-670
-930
-715
-785
-705
+145
+645
-635
-965
-760
-805
-730
+210
+820
-140
+385
+500
-120
+580
+740
Eox ) ∼+455d
Eox ) ∼+570d
-55
-10
+70
Eox ) ∼+455d
+115
Eox ) ∼+540d
a-c +
/C, C2+/C+, and C3+/C2+ are half-wave potentials vs ferrocenium/ferrocene for the [Mc(CH)3Mc′]+/[Mc(CH)3Mc′]•, [Mc(CH)3Mc′]2+/[Mc(CH)3Mc′]+,
and [Mc(CH)3Mc′]3+/[Mc(CH)3Mc′]2+ couples, respectively. d Irreversible; potential of oxidation peak given.
ferrocenyl species, 4+, there is less variation in the bridge 1H-
1H coupling constants (13.1, 14.2 Hz), consistent with the
reduced BLA seen in the structure of 4a and with the infrared
data.
thenocene portions of 3+ and 5+ undergo irreversible oxidation,
as does ruthenocene itself,56 while the C+/C features in these
compounds resembled those of the diiron cations. The C+/C
data allow comparison of the abilities of the various metal-
locenes to stabilize positive charge.57 The diruthenocene cation
3+ is 45 mV (in THF) harder to reduce than its diferrocene
analogue 1+, indicating greater stabilization of the positive
charge in the former case. There is only a small difference (10
mV in THF) in the ease of reduction of 3+ and 5+, suggesting
that only one ruthenocene can effect most of the charge
stabilization, consistent with a localized structure for 3+. The
octamethylferrocene compounds are harder still to reduce,
presumably largely due to the inductive effect of the methyl
groups.
The 13C shift of Câ, the central C of the allylium bridge, δâ,
is related to the π-charge density, Fâ, at that atom by Fâ ) (δâ
- δo)/c,8,52 where δo and c are constants (the relationship
between δR and FR will be less straightforward due to the effects
of direct interaction with the metal). Comparison of the NMR
data for 1+ (δâ ) 125.3 ppm) and 3+ (δâ ) 119.6 ppm) cations
clearly shows less positive charge density in the diruthenium
compound, indicating greater stabilization of the positive charge
by ruthenocene end groups, consistent with electrochemical data
(vide infra). Moreover, the shift for 5+ (120.8 ppm) indicates
that most of the extra stabilization effected by ruthenocene can
be effected by just one ruthenocene, also consistent with
electrochemical data (vide infra) and with a localized structure
for the 3+ cation.
Electrochemistry. Cyclic voltammetric data for the bis-
(metallocenyl)allylium tetrafluoroborates (1a-5a) are sum-
marized in Table 3.53 For each of the diiron cations, 1+, 2+,
and 4+, three one-electron redox processes were observed
(voltammograms for 2+ are shown in the Supporting Informa-
tion). The C2+/C+ couple was observed at positive potential
relative to oxidation of the parent metallocene54 and showed
reversibility comparable to that of ferrocenium/ferrocene under
the same conditions. Processes at higher and lower potentials
are attributed to C3+/C2+ and C+/C couples, respectively; both
show characteristic features of EC processes (i.e., those where
the electron transfer itself is reversible but where the oxidized
species undergoes a chemical reaction on a time scale compa-
rable with that of the electron transfer) and cycling to potentials
approaching those for these couples leads to the buildup of
features we attribute to decomposition products.55 The ru-
Summary
We have synthesized and characterized a range of 1,3-di-
(metallocenyl)allylium cations. The metallocenes stabilize the
carbocationic charge through contributions from [(η6-fulvene)-
(η5-cyclopentadienyl)metal] cation structures. The [Fc(CH)3Fc]+,
1+, and [Fc′′(CH)3Fc′′]+, 2+, cations are symmetrical species.
The terminal ferrocenes are less distorted toward a [(η6-fulvene)-
(η5-cyclopentadienyl)iron] structure in 2+ due to the stabilizing
effect of the methyl groups upon the positive charge. The
formally unsymmetrical [Fc(CH)3Fc′′]+, 4+, cation is electroni-
cally somewhat unsymmetrical according to crystallographic and
NMR data. However, this asymmetry is insufficient to result in
crystallographically significant BLA in the allylium bridge. The
vibrational spectrum shows a C-C stretch attributable to the
allylium bridge at a frequency similar to the symmetrical 1+
and 2+ cations. This cation can, therefore, be regarded delo-
calized, despite its inherent asymmetry.
Replacement of ferrocenyl with ruthenocenyl leads to rather
different properties. The crystal structures of [Rc(CH)3Rc]+[PF6]-,
(51) The rate and the barrier were estimated as follows. We assumed the 13C
frequency of the C5H5 of the cationic end of the localized molecule (ν1)
will be approximately the same as that of the 5+ cation, while the frequency
of the other C5H5 group (ν2) can be estimated by assuming that the value
observed for the 3+ cation (νobs) is an average of ν1 and ν2. The rate constant,
k, obeys the inequality, k > π(ν1 - ν2)/x2, while the barrier height, ∆G‡,
is related to k by, k ) (kBT/h) exp(-∆G‡/RT). Similar results are obtained
by assuming ν1 and ν2 to be equal to the C5H5 13C frequencies of the
ruthenocenylmethylium cation (ref 18) and ruthenocene, respectively.
(52) Spiesecke, H.; Schneider, W. G. Tetrahedron Lett. 1961, 468-72.
(53) It has previously been reported that the polymethines [Fc(CH)nFc]+[BF4]-
(n ) 1, 3, 5, 9, 13) undergo two oxidations in dichloromethane (ref 22);
the values reported for 1a correspond to potentials of -40 and +140 mV
vs ferrocenium/ferrocene, clearly at variance with our values. We suggest
that it is possible that the previous measurements may have been
complicated by the presence of the decomposition products (see ref 55) of
1+, which we observed at E1/2 ) ∼+10 mV in dichloromethane.
(54) 1,2,3,4,1′,2′,3′,4-Octamethylferrocene is oxidized at -360 and -445 mV
vs ferrocenium/ferrocene in THF and CH2Cl2, respectively: Mendiratta,
A.; Barlow, S.; Day, M. W.; Marder, S. R. Organometallics 1999, 18, 454-
456.
(55) When potentials approaching C3+/C2+ were employed, an additional feature
was found to grow in at a potential similar to the parent metallocene, its
intensity increasing with successive cycles. Similar features are seen when
the potential is cycled around C+/C; these features are presumably due to
the products of the side reactions associated with C3+/C2+ and C+/C. In
the case of 1+ in THF, at least some of the decomposed material was found
to be adhering to the working electrode. When the electrodes were removed
from the analyte solution, rinsed with fresh solvent, and inserted into fresh
THF electrolyte solution, a feature was observed with peak potentials Eox
) +20 mV and Ered ) -20 mV (Eox ) +5 mV and Ered ) -15 mV in
CH2Cl2). It was previously shown that reduction of [FcCHFc]+ to the radical
is followed by dimerization (Tirouflet, J.; Laviron, E.; Moise, C.; Mugnier,
Y. J. Organomet. Chem. 1973, 50, 241-6) so some of the new signals
observed may be due to dimerized species.
(56) Diaz, A. F.; Mueller-Westerhoff, U. T.; Nazzar, A.; Tanner, M. J.
Organomet. Chem. 1982, 236, C45-C48 and references therein.
(57) All the 1,3-di(metallocenyl)allylium cations are considerably more readily
reduced than the ruthenocenylmethylium cation, presumably due to more
delocalized lower-energy π*-type LUMOs in the present compounds; the
LUMO of the monomeric species is much more localized on ruthenium
(ref 18).
9
6292 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002

Metallocene-Terminated Allylium Salts
A R T I C L E S
3b, [Rc(CH)3Rc]+[BAr′4]-, 3c, and [Fc(CH)3Rc]+[PF6]-, 5b,
are all markedly unsymmetrical and comprise a [(η6-fulvene)-
(η5-cyclopentadienyl)ruthenium] cation linked by a vinylene
bridge to an undistorted metallocene. The structures of 3b and
3c are the first of Peierls-distorted polymethines; i.e., sym-
metrically substituted species that adopt an unsymmetrical
structure with polyene-like bond-length alternation and in which
the overall charge is localized on only one of the two end groups.
Localization in 3c, despite the noncoordinating delocalized
counterion, suggests the asymmetry is due to an inherent
property of the cation rather than to crystal-packing effects. Fast
time-scale spectroscopic techniques are also consistent with an
instantaneously unsymmetrical structure for 3+ in solution; the
allylium C-C stretching frequencies are similar for 3+ and 5+
cations and significantly different from those for 1+, 2+, and
4+, and, moreover, the frequencies for 3+ salts are similar in
solution and in the solid. The electronic spectrum of 3+ shows
broader absorptions, more strongly dependent upon solvent
polarity, than that of 1+, also consistent with a localized solution
structure for 3+. On the (slower) NMR time scale, a symmetrized
structure is observed for 3+.
The behavior of 3+ parallels that of many class II mixed-
valence compounds where the apparent delocalization depends
on the time scale of the technique used and often varies between
solution and the solid state. The difference in behavior between
1+ and 3+ cations can also be rationalized using an analogy
with mixed-valence chemistry; localization in the diruthenium
compound may be due to a larger reorganization energy in the
metallocene/[(η6-fulvene)(η5-cyclopentadienyl)metal] system for
ruthenium than for iron, presumably connected with the greater
tendency toward η6 coordination found for the larger metal.
Control of the reorganization energy in this way may be a useful
design principle for the synthesis of other charge-localized
polymethines. Furthermore, tuning the reorganization energy
could be used as a method of controlling the barrier height
between the two extreme localized forms of the molecule. In
principle, the sense of the bond-length alternation could be
reversed by application of an external electric field, leading to
a large change in polarization, which could be exploited in
optical and electronic applications.
Computational Methods. Single-point calculations were carried out
using the Amsterdam Density Functional package (ADF99).58 The
electronic configurations were described by an uncontracted triple-ú
basis set of Slater orbitals, with a single polarization functional added
to the main group atoms. The cores were frozen up to 2p for Fe, 3p
for Ru, and 1s for C. Relativistic corrections were included using the
ZORA formalism. The structures of the cations were taken from those
in the crystal structures of 1b and 3b, the bond lengths in the bridge of
1+ being modified from the apparent values from the disordered crystal
to chemically reasonable values (CCp-CR, 1.43 Å; CR-Câ, 1.38 Å, as
in the structure of 2b). C-H bond distances were set at 1.09 Å. Energies
were calculated using Vosko, Wilk, and Nusair’s local exchange
correction,59 with nonlocal exchange corrections by Becke,60 and
nonlocal exchange correlation corrections by Perdew.61,62
General Procedure for 1-Metallocenyl-2-bromoethenes, McCHd
CHBr. A slurry of KOtBu (12.5 mmol) in dry THF (100 mL) was
added dropwise under nitrogen to a slurry of [Ph3PCH2Br]+[Br]- 30 (12.5
mol) in dry THF (100 mL) at -78 °C. After 1 h, a solution of the
appropriate aldehyde (10 mmol) in dry THF (30 mL) was added
dropwise. The reaction mixture was allowed to warm to room
temperature and stirred at room temperature until the reaction was
complete by TLC. Water and diethyl ether (100 mL each) were added,
and the layers were separated. The aqueous layer was extracted with
more ether (2 × 100 mL); the combined ether extracts were washed
with saturated aqueous sodium chloride (2 × 100 mL), dried over
magnesium sulfate, filtered, and evaporated under reduced pressure.
Purification procedures are given below. The reactions were also carried
out on scales of between 3 and 26 mmol.
FcCHdCHBr. The general procedure above was used to synthesize
FcCHdCHBr29 (6.33 g, 21.8 mmol, 84%) from FcCHO (5.56 g, 26.0
mmol); the reaction time was 13 h, and the crude product was purified
by dissolving in hexane and passing through a silica plug. FcCHd
CHBr (∼1:2 E/Z isomer mixture) was obtained as an orange-red oil,
crystallizing after prolonged drying in vacuo: 1H NMR (300 MHz,
benzene-d6) (E-isomer) δ 6.66 (d, J ) 14.0 Hz, 1H), 6.02 (d, J ) 14.0
Hz), 3.92 (s, 5H), 3.87 (apparent s, 4H); 1H NMR (300 MHz, benzene-
d6) (Z-isomer) δ 6.34 (d, J ) 7.7 Hz, 1H), 5.85 (d, J ) 7.7 Hz, 1H),
4.61 (t, J ) 1.7 Hz, 2H), 4.03 (t, J ) 1.7 Hz, 2H), 3.96 (s, 5H); 13C
NMR (75 MHz, benzene-d6) (isomeric mixture) δ 135.2, 131.3, 102.8,
101.5, 81.9, 79.2, 70.1, 69.6, 69.5, 69.3, 69.2, 66.7; high-resolution
EIMS calcd for C12H11FeBr M+ 289.9394, found 289.9396.
Fc′′CHdCHBr. The general procedure above was used to synthesize
Fc′′CHdCHBr (2.53 g, 6.28 mmol, 57%) from Fc′′CHO63 (3.60 g, 11.0
mmol); the reaction time was 72 h and the crude product was purified
by dissolving in hexane and passing through a silica plug. Fc′′CHd
CHBr (∼2:3 E/Z isomer mixture) was obtained as an orange-red oil:
1H NMR (300 MHz, benzene-d6) (E-isomer) δ 6.96 (d, J ) 14.0 Hz,
1H), 6.18 (d, J ) 14.0 Hz, 1H), 3.10 (s, 1H), 1.66 (s, 6H), 1.62 (s,
Experimental Details
General Considerations. Elemental analyses were performed by
Atlantic Microlab Inc. (Norcross, GA) or by the Inorganic Chemistry
Laboratory’s Analytical Service. Most NMR spectra were recorded
using a General Electric QE-300 spectrometer. Variable-temperature
data were acquired using a Varian Unity Plus 500 spectrometer; data
are reported for room temperature (∼25 °C), unless stated otherwise.
UV-visible-NIR spectra were acquired using Hewlett-Packard 8452A
spectrometers, a Varian Cary 5E spectrometer, or a GBC Instruments
Cintra 10 spectrometer. IR spectra were acquired with a Perkin-Elmer
Spectrum 1000 spectrometer; Raman spectra were acquired with a
Nicolet Raman 950 instrument, using the 633-nm line of a He/Ne laser
at a power of 1 mW. Solid samples were run as either neat powders or
mixed with KBr. Cyclic voltammetry was performed under argon on
dry tetrahydrofuran or dichloromethane solutions ∼10-4 M in sample
and 0.1 M in [nBu4N]+[PF6]- using a glassy carbon working electrode,
a platinum auxiliary electrode, a AgCl/Ag pseudoreference electrode,
and a BAS 100B potentiostat. Potentials were referenced by the addition
of ferrocene to the cell and are quoted relative to the nearest 5 mV
relative to the ferrocenium/ferrocene couple at 0 V. Where necessary,
solvents were distilled from calcium hydride (dichloromethane) or
sodium benzophenone ketyl (THF, diethyl ether, pentane).
1
6H), 1.56 (s, 6H), 1.52 (s, 6H); H NMR (300 MHz, benzene-d6) (Z-
isomer) δ 6.78 (d, J ) 7.4 Hz, 1H), 6.21 (d, J ) 7.4 Hz, 1H), 3.15 (s,
1H), 1.65 (s, 6H), 1.60 (s, 6H), 1.59 (s, 6H), 1.56 (s, 6H); 13C NMR
(75 MHz, benzene-d6) (isomeric mixture) δ 135.8, 131.9, 109.8, 101.0,
81.7, 80.9, 80.8, 80.7, 80.5, 80.1, 79.7, 79.0, 78.9, 77.4, 72.1, 72.0,
11.9, 11.5, 11.2, 10.2, 10.1, 9.6 (some methyl resonances presumably
(58) Baerends, E. J.; Berces, A.; Bo, C.; Boerringter, P. M.; Cavallo, L.; Deng,
L.; Dickson, R. M.; Ellis, D. E.; Fan, L.; Fischer, T. H.; Fonseca Guerra,
C.; van Gisbergen, S. J.; Groeneveld, J. A.; Gritsenko, O. V.; Harris, F.
E.; van den Hoek, P.; Jacobsen, H.; van Kessel, G.; Kootstra, F.; van Lenthe,
E.; Osinga, V. P.; Philipsen, P. H. T.; Post, D.; Pye, C. C.; Ravenek, W.;
Ros, P.; Schipper, P. R. T.; Schreckenbach, G.; Snijiders, J. G.; Sola, M.;
Swerhone, D.; te Velde, G.; Vernooijis, P.; Versluis, L.; Visser, O.; van
Wezenbeek, E.; Wiesenekker, G.; Wolff, S. K.; Woo, T. K.; Ziegler, T.
ADF Program System Release 1999; Department of Theoretical Chemistry,
Vrije Universiteit: Amsterdam, 1999.
(59) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1990, 58, 1200.
(60) Becke, A. D. Phys. ReV. 1988, A38, 3098-3100.
(61) Perdew, J. P. Phys. ReV. 1986, B34, 7406.
(62) Perdew, J. P. Phys. ReV. 1986, B34, 8822-8824.
(63) Zou, C.; Wrighton, M. S. J. Am. Chem. Soc. 1990, 112, 7578-7584.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6293

A R T I C L E S
Barlow et al.
coincident); high-resolution LSIMS calcd for C20H27FeBr M+ 402.0645,
found 402.0647.
3H), 4.71 (m. 1H), 4.66 (m, 2H), 4.59 (m, 1H), 4.49 (s, 5H), 4.48 (s,
5H), 4.44 (m, 2H), 4.42 (s, 5H), 4.41 (s, 5H), 4.37 (2 overlapping m,
3H), 1.58 (d, J ) 6.8 Hz, 1H), 1.51 (d, J ) 5.2 Hz, 1H); LSIMS m/z
518 (8%, M+), 501 (10%, M+ - OH); high-resolution LSIMS calcd
for C23H22ORu2 M+ 517.9758, found 517.9760.
FcCHdCHCH(OH)Rc. Crude FcCHdCHCH(OH)Rc (∼1:2 E:Z
mixture) was obtained as a dull orange powder (419 mg, 0.89 mmol,
58%) from FcCHdCHBr (440 mg, 1.52 mmol) and RcCHO (393 mg,
1.52 mmol) in a fashion similar to the previous two compounds: 1H
NMR (300 MHz, benzene-d6) (isomeric mixture) δ 6.45 (d, J ) 15.7
Hz, 1H), 6.20 (d, J ) 11.3 Hz, 2H), 6.05 (dd, J ) 15.7, 6.2 Hz, 1H),
5.66 (dd, J ) 11.3, 8.5, 2H), 5.43 (t, J ) ∼7.5 Hz, 2H), 4.75-3.90
(many overlapping peaks), 1.85 (d, J ) 8.0 Hz), 1.62 (d, J ) 7.0 Hz,
2H); high-resolution LSIMS calcd for C23H22OFeRu M+ 472.0064,
found 472.0077.
Fc′′CHdCHCH(OH)Fc′′. This compound was synthesized by the
general method from Fc′′CHdCHBr (990 mg, 2.46 mmol) and Fc′′CHO
(800 mg, 2.45 mmol); it was obtained in crude form as a brown oil
(1.36 g, 2.09 mmol, 85%), which decomposed on standing, after flash
chromatography under nitrogen on basic alumina, eluting with ethyl
acetate/hexanes (1:7). The desired fractions were identified by testing
for a darkening (allylium cation formation) with concentrated HCl. Due
to its instability, this compound was used directly without further
purification or characterization.
FcCHdCHCH(OH)Fc′′. Crude FcCHdCHCH(OH)Fc′′ was ob-
tained as a brown oil (0.96 g, 1.48 mmol, 41%) from FcCHdCHBr
(1.06 g, 3.66 mmol) and Fc′′CHO (1.19 g, 3.65 mmol) in analogy to
the previous compound.
RcCHdCHBr and RcCtCH. The general procedure above was
used to synthesize RcCHdCHBr (700 mg, 2.09 mmol, 55%) from
RcCHO64 (984 mg, 3.80 mmol); the reaction time was 90 h, and the
crude product was purified by flash chromatography on silica, eluting
with hexane. Evaporation of the first fraction afforded RcCHdCHBr
(∼1:1 E/Z isomer mixture) as a yellowish oil, solidifying after prolonged
drying in vacuo: 1H NMR (300 MHz, benzene-d6) (isomeric mixture)
δ 6.59 (d, J ) 13.8 Hz, 1H), 6.30 (d, J ) 7.8 Hz, 1H), 6.00 (d, J )
13.8 Hz, 1H), 5.71 (d, J ) 7.8 Hz, 1H), 5.04 (m, 2H), 4.44 (m, 2H),
4.39 (s, 5H), 4.38 (m, 2H), 4.33 (m, 2H), 4.31 (s, 5H); 13C NMR (75
MHz, benzene-d6) (isomeric mixture) δ 134.3, 130.4, 102.8, 102.1, 85.9,
83.2, 72.2, 71.6, 71.5, 71.2, 71.0, 69.0; high-resolution LSIMS calcd
for C12H11RuBr M+ 337.9067, found 337.9078; Evaporation of the
second fraction gave an off-white solid found to be RcCtCH31,32 (220
mg, 0.98 mmol, 26%): 1H NMR (300 MHz, benzene-d6) δ 4.82 (m,
2H), 4.42 (s, 5H), 4.26 (m, 2H), 2.41 (s, 1H); 13C NMR (75 MHz,
benzene-d6) δ 81.8, 74.5, 74.4, 72.3, 71.1, 68.4; high-resolution LSIMS
calcd for C12H10Ru M+ 255.9826, found 255.9823; IR (KBr) 3265 (vs),
3115, 3095, 2100 (s), 1439, 1404, 1260, 1220, 1096, 1027 (s), 9992,
902, 813 (s), 654 (s), 614, 555, 520 cm-1
.
General Procedure for Synthesis of 1,3-Dimetallocenylprop-1-
ene-3-ols, McCHdCHCH(OH)Mc′. A 1-metallocenyl-2-bromoethene
(2 mmol) was dissolved in a 1:1:4 mixture of pentane, diethyl ether,
and THF (15 mL). The solution was cooled to -98 °C (MeOH/liquid
N2) and treated with a 1.7 M pentane solution of tBuLi (4 mmol). After
30 min, the reaction was allowed to warm to -78 °C over 30 min.
The reaction mixture was then treated with a solution of the appropriate
metallocenyl aldehyde (2 mmol) in THF (10 mL). After 15 min at -78
°C, the reaction mixture was allowed to warm to room temperature
over 1 h and treated with water (20 mL) and diethyl ether (20 mL).
The layers were separated, and the aqueous layer was extracted with
three additional portions of ether. The combined ether extracts were
dried over potassium carbonate, filtered, and evaporated under reduced
pressure. Since the instability of FcCHdCHCH(OH)Fc with respect
to oxidation to FcCHdCHC(O)Fc has previously been noted,27 and
since our octamethylferrocenyl alcohols appear even more unstable,
we did not attempt to fully purify and characterize the alcohols. The
crude materials were isolated as described below and used without
purification in the synthesis of the allylium salts.
FcCHdCHCH(OH)Fc.25-27 Crude FcCHdCHCH(OH)Fc26,27 (912
mg, 2.14 mmol, 46%) was prepared from FcCHdCHBr (1.36 g, 4.68
mmol) andFcCHO (1.00 g, 4.67 mmol) by the general method and
obtained as an orange powder in an ∼1:2 E:Z isomer ratio, after
recrystallization of the evaporation residue from hot heptane. Its identity
was checked using NMR and mass spectroscopy before conversion to
salts of 1+: 1H NMR (300 MHz, benzene-d6) (isomeric mixture) δ
6.39 (d, J ) 15.7 Hz, 1H), 6.19 (d, J ) 11.3 Hz, 2H), 6.12 (dd, J )
15.7, 6.3 Hz, 1H), 5.80 (dd, J ) 10.9, 8.9 Hz, 2H), 5.49 (br, dd, 2H),
4.89 (br, apparent t, 1H), 4.91-3.91 (m), 1.79 (br d, 2H), 1.70 (br d,
1H); 13C NMR (300 MHz, benzene-d6) δ 131.5, 130.0, 93.3, 93.1, 83.3,
81.4, 71.9, 71.0-66.1 (numerous overlapping peaks but δ 69.5 and
68.8 are very strong), 61.7; LSIMS m/z 426 (57%, M+), 409 (100%,
M+ - OH); high-resolution LSIMS calcd for C23H22OFe2 M+ 426.0369,
found 426.0378.
General Procedure for Tetrafluoroborate Salts of Metallocene-
Terminated Allylium Cations. HBF4 (1-2 mL, 85% solution in
diethyl ether) was freeze-pump-thaw degassed and added dropwise
to a deoxygenated solution of the appropriate alcohol precursor (100-
200 mg) in a minimum quantity of dry diethyl ether (25-120 mL).
The mixture instantly darkened and precipitation occurred. The
supernatant was decanted by cannula, and the precipitate was washed
with diethyl ether (2 × 25 mL) before drying in vacuo. The salts were
crystallized by extraction into dichloromethane (∼20 mL), filtration,
and layering with diethyl ether (∼200 mL).
General Procedure for Hexafluorophosphate Salts of Metal-
locene-Terminated Allylium Cations. Deoxygenated 60% aqueous
hexafluorophosphoric acid (1 mL) was added dropwise to a stirred
deoxygenated solution of the appropriate alcohol (0.5 mmol) in THF;
the mixture instantly darkened. After 15 min, deoxygenated water (75
mL) was added, affording a dark precipitate. The precipitate was
allowed to settle, and the supernatant was removed by cannula. The
solids were washed with water (2 × 75 mL) and dried in vacuo, before
extraction into dichloromethane (∼20 mL), filtration, and crystallization
by layering with diethyl ether (∼200 mL).
[Fc(CH)3Fc]+[BF4]-27 (1a). Pure 1a (80 mg, 0.16 mmol, 70%) was
obtained by the general procedure from the corresponding crude alcohol
(100 mg, 0.23 mmol): 1H NMR (300 MHz, dichloromethane-d2) δ
8.39 (d, J ) 13.4 Hz, 2H), 6.49 (t, J ) 13.4 Hz, 1H), 5.59 (apparent
s, 4H), 5.05 (apparent s, 4H), 4.51 (s, 10H); 13C NMR (75 MHz,
dichloromethane-d2) δ 157.3, 125.3, 87.5, 83.5, 76.3, 74.0; high-
resolution LSIMS calcd for C23H21Fe2 M+ - BF4 409.0342, found
409.0342; UV (CH2Cl2) λmax 431 (ꢀ431 23 000), 787 (ꢀ787 14000) nm;
UV (CHCl3) λmax 428, 772 nm; UV (THF) λmax 422, 746 nm; UV
(EtOAc) λmax 418, 734 nm; UV (MeCN) λmax 420, 745 nm; UV (MeOH)
λmax 422, 750 nm; UV (DMSO) λmax 422, 748 nm; UV (DMF) λmax
424, 744 nm; UV (acetone) λmax 423, 751 nm; UV (PhCN) λmax 428,
766 nm; UV (H2O) λmax 414, 730 nm; IR (KBr) 3095, 1548 (s), 1444,
1414, 1374, 1280, 1225, 1156, 1066 (s, br), 917, 833, 634 cm-1; IR
(CH2Cl2) 1560, 1093, 1025, 835 cm-1. Anal. Calcd for C23H21BF4Fe2:
C, 55.71; H, 4.27. Found: C, 55.61; H, 4.30.
RcCHdCHCH(OH)Rc. Crude RcCHdCHCH(OH)Rc (∼1:1 iso-
mer mixture) was obtained in a manner similar to its diiron analogue
(off-white powder, 604 mg, 1.05 mmol 50%) from RcCHdCHBr (700
mg, 2.08 mmol) and RcCHO (540 mg, 2.08 mmol): 1H NMR (300
MHz, benzene-d6) (isomeric mixture) δ 6.38 (d, J ) 15.6 Hz, 1H),
6.17 (d, J ) 10.0 Hz, 1H), 6.00 (dd, J ) 15.6, 6.2 Hz, 1H), 5.57-5.48
(overlapping dd and m, 3H), 4.84 (m, 2H), 4.77 (m, 2H), 4.73 (m,
[Fc(CH)3Fc]+[PF6]- (1b). Red crystals of pure 1b (170 mg, 0.31
mmol, 58%) were obtained from the corresponding crude alcohol (225
(64) Sanders, R.; Mueller-Westerhoff, U. T. J. Organomet. Chem. 1996, 512,
219-224.
9
6294 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002

Metallocene-Terminated Allylium Salts
A R T I C L E S
mg, 0.53 mmol) using the general method: IR (KBr) 3115, 1558 (vs),
1449 (s), 1409, 1374, 1284, 1225 (s), 1161, 1106, 1056, 982, 828 (vs),
634, 555 cm-1; IR (CH2Cl2) 1558 (s), 1450, 1225, 846 (s) cm-1; Raman
(CH2Cl2) 1542 cm-1. Anal. Calcd for C23H21PF6Fe2: C, 49.85; H, 3.82.
Found: C, 49.66; H, 3.82.
555 cm-1; IR (CH2Cl2) 1598 (s), 848 (s); Raman (KBr) 1609 (s), 1449,
1374, 1305, 1283, 1251, 1209, 1140, 1103, 638 cm-1; Raman (CH2-
Cl2) 1601 (s), 1449, 1280, 1210, 1144 cm-1. Anal. Calcd for C23H21-
PF6Ru2: C, 42.86; H, 3.28. Found: C, 42.87; H, 3.31.
[Rc(CH)3Rc]+[BAr′4]- (3c). Pure 3c (80 mg, 0.059 mmol, 80%)
was obtained from 3b (53 mg, 0.080 mmol) and [Na]+[BAr′4]- (65
mg, 0.073 mmol) using the same method as for the preparation of 1c:
1H NMR (300 MHz, dichloromethane-d2, 25 °C) δ 7.72 (br s, 8H),
7.56 (s, 4H), 7.29 (d, J ) 13.5 Hz, 2H), 6.14 (t, J ) 13.5 Hz, 1H),
5.44 (t, J ) 1.8 Hz, 4H), 5.12 (t, J ) 1.8 Hz, 4H), 4.80 (s, 10H); UV
(Et2O) λmax 358, 442; UV (EtOAc) λmax 358, 442; UV (1,4-dioxane)
λmax 376, 458; UV (benzene) λmax 388, 479. IR (KBr) 1584 (s), 1357
[Fc(CH)3Fc]+[BAr′4]- {Ar′ ) 3,5-Bis(trifluoromethyl)phenyl}
(1c). Pure 1c (120 mg, 0.094 mmol, 76%) was obtained from 1b (120
mg, 0.14 mmol) and [Na]+[BAr′4]-65 (110 mg, 0.12 mmol) using the
method described by Manr´ıquez and co-workers for the synthesis of
[FeCp2]+[BAr′4]-:66 UV (Et2O) λmax 432, 787 nm; UV (1,4-dioxane)
λmax 432, 782 nm; UV (benzene) λmax 437, 793 nm; IR (KBr) 1561
(vs), 1453, 1357 (s), 1278 (vs), 1235, 1128 (vs, br), 890, 840, 832,
713, 683, 669. 642 cm-1; IR (CH2Cl2) 1558 (s), 1451, 1355 (s), 1284
(vs), 1229, 1127 (vs, br), 897 (s), 839, 672, 635 cm-1. Anal. Calcd for
C55H33BF24Fe2: C, 51.92; H, 2.61. Found: C, 51.74; H, 3.31.
[Fc′′(CH)3Fc′′]+[BF4]- (2a). Dark green microcrystals of pure 2a
(180 mg, 0.25 mmol, 58%) were obtained from the corresponding crude
alcohol (282 mg, 0.43 mmol) using the general method: 1H NMR (300
MHz, dichloromethane-d2) δ 8.08 (d, J ) 13.7 Hz, 2H), 6.85 (t, J )
13.7 Hz, 1H), 3.58 (s, 1H), 1.99 (s, 12H), 1.80 (s, 12H), 1.50 (s, 12H),
1.46 (s, 12H); 13C NMR (75 MHz, dichloromethane-d2) δ 153.4, 127.1,
97.9, 88.1, 87.6, 86.3, 85.6, 76.3, 11.2, 11.0, 10.7, 9.2; high-resolution
LSIMS calcd for C39H53Fe2 M+ - BF4 633.2846, found 633.2868; UV
(dichloromethane) λmax 492 (ꢀ492 30 000), 880 (ꢀ880 16 000) nm; UV
(THF) λmax 477, 847 nm; UV (EtOAc) λmax 478, 830 nm; UV (MeCN)
λmax 484, 850 nm; UV (DMSO) λmax 480, 847 nm; IR (KBr) 2956,
2896, 2856, 1548 (s), 1478, 1450, 1404, 1379, 1240, 1175, 1056 (s,
br), 1026 (s, br), 897, 674, 629, 609, 530 cm-1; IR (CH2Cl2); 2951,
2910, 2868, 1538 (s), 1383, 1179, 1088 (s), 1025 (s) cm-1; Anal. Calcd
for C39H53BF4Fe2: C, 65.03; H, 7.43. Found: C, 64.78; H, 7.43.
[Fc′′(CH)3Fc′′]+[PF6]- (2b). Green needles of pure 2b (260 mg,
0.33 mmol, 58%) were obtained from the corresponding crude alcohol
(375 mg, 0.58 mmol) using the general method. IR (KBr) 2946, 2916,
2856, 1538 (s), 1483, 1449, 1399, 1379, 1240, 1175, 1021, 840 (s),
634, 555 cm-1; IR (CH2Cl2) 2951, 2910, 2868, 1542 (s), 1382, 1236,
1180, 1020, 846 (s); Anal. Calcd for C39H53PF6Fe2: C, 60.17; H, 6.86;
Found: C, 60.40; H, 6.49.
(s), 1278 (vs), 1130 (vs, br), 901, 888, 839, 830, 713, 683, 669 cm-1
;
IR (CH2Cl2) 1600 (s), 1355 (s), 1284 (vs), 1128 (vs, br), 898 (s), 672
cm-1. Anal. Calcd for C55H33BF24Ru2: C, 48.47; H, 2.44. Found: C,
48.11; H, 2.93.
[Fc(CH)3Fc′′]+[BF4]- (4a). The general method was used to obtain
dark platelets of 4a (210 mg, 0.35 mmol, 59%) from crude FcCHd
CHCH(OH)Fc′′ (316, 0.59 mg): 1H NMR (300 MHz, dichloromethane-
d2) δ 8.36 (d, J ) 14.2 Hz, 1H), 7.83 (d, J ) 13.1 Hz, 1H), 6.58 (dd,
J ) 14.2, 13.1 Hz, 1H), 5.01 (apparent t, J ) ca.1.8 Hz, 2H), 4.86
(apparent t, J ) 1.8 Hz, 2H), 4.37 (s, 5H), 3.68 (s, 1H), 1.99 (s, 6H),
1.67 (s, 6H), 1.49 (s, 6H), 1.48 (s, 6H); 13C NMR (75 MHz,
dichloromethane-d2) δ 152.4 (CH), 145.5 (CH), 126.1 (CH), 104.1
(quat), 94.8 (quat), 91.7 (quat), 91.3 (quat), 83.5 (quat; remaining quat
not observed), 80.2 (CH), 76.5 (CH), 72.0 (CH), 70.9 (CH), 10.7 (CH3),
10.6 (CH3), 9.0 (CH3; remaining methyl resonance presumably overlap-
ping); high-resolution LSIMS calcd for C31H37Fe2 M+ - BF4 521.1594,
found 521.1594; UV (CH2Cl2) λmax 401 (ꢀ401 15 000), 438 (ꢀ438 14 000),
748 (ꢀ748 9100) nm; UV (THF) λmax 413, 438, 747 nm; UV (MeCN)
λmax 395, 425, 723 nm; UV (DMSO) λmax 397, 430, 722 nm; IR (KBr)
3095, 2956, 2906, 2856, 1553 (s), 1439, 1379, 1320, 1300, 1235, 1156,
1056 (s, br), 927, 902, 822, 694, 629 cm-1; KBr (CH2Cl2) 2951, 2917,
1556 (s), 1382, 1236, 1083 (s), 1020 (s), 826 cm-1. Anal. Calcd for
C31H37BF4Fe2: C, 61.23; H, 6.13. Found: C, 60.83; H, 6.07.
[Fc(CH)3Rc]+[BF4]- (5a). Brown microcrystals of 5a (80 mg, 0.15
mmol, 52%) were obtained from the corresponding crude alcohol (150
mg, 0.29 mmol) by the general procedure: 1H NMR (300 MHz,
dichloromethane-d2) δ 7.58 (d, J ) 14.5 Hz, 1H), 7.41 (d, J ) 12.2
Hz, 1H), 6.25 (dd, J ) 14.5, 12.2 Hz, 1H), 5.91 (t, J ) 1.9 Hz, 4H),
5.37 (t, J ) 1.9 Hz, 4H), 5.05 (s, 5H), 4.86 (apparent s, 2H), 4.71
(apparent s, 2H), 4.27 (s, 5H); 13C NMR (75 MHz, dichloromethane-
d2) δ 151.8, 124.4, 120.8, 95.0, 87.7, 86.0, 82.9, 79.4, 75.9, 72.6, 70.9;
high-resolution LSIMS calcd for C23H21FeRu M+ - BF4 455.0036,
found 455.0053; UV (CH2Cl2) λmax 404 (ꢀ404 11 000), 648 (ꢀ648 5000)
nm; UV (THF) λmax 373, 547 nm; UV (MeCN) λmax 367, 553 nm. IR
(KBr) 3095, 3025, 2916, 2846, 1598 (s), 1449, 1409, 1374, 1285, 1210,
[Rc(CH)3Rc]+[BF4]- (3a). Pure 3a (180 mg, 0.31, 46%) was
obtained from the corresponding crude alcohol (345 mg, 0.67 mmol)
using the general method: 1H NMR (300 MHz, dichloromethane-d2,
25 °C) δ 7.35 (d, J ) 13.5 Hz, 2H), 6.17 (t, J ) 13.5 Hz, 1H), 5.43 (t,
1
J ) 1.9 Hz, 4H), 5.18 (t, J ) 1.9 Hz, 4H), 4.83 (s, 10H); H NMR
(500 MHz, dichloromethane-d2, -85 °C) δ 7.06 (d, J ) 13.3 Hz, 2H),
6.02 (t, J ) 13.3 Hz, 1H), 5.33 (s, br, 4H), 5.08 (s, very br, 4H), 4.72
(apparently two overlapping s, 10H); 13C NMR (75 MHz, dichlo-
romethane-d2, 25 °C) δ 133.2, 119.6, 90.6, 81.7, 78.7, 75.9; 13C NMR
(125 MHz, dichloromethane-d2, -85 °C) δ 128, 119, 90.0. 80.9, 78.1,
75.5 (v br); high-resolution LSIMS calcd for C23H21Ru2 M+ - BF4
500.9730, found 500.9716; UV (CD2Cl2) λmax 384 (ꢀ384 9000), 468 (ꢀ468
9700) nm; UV (CHCl3) λmax 360, 450 nm; UV (THF) λmax 329, 429
nm; UV (MeCN) λmax 346, 431 nm; UV (MeOH) λmax 335, 430 nm;
UV (DMF) λmax 339, 423 nm; UV (DMSO) λmax 329, 427 nm; UV
(PhCN) λmax 349, 441 nm; IR (KBr) 3085, 2916, 1603 (s), 1498, 1449,
1404, 1205, 1076 (vs, br), 813, 698, 530 cm-1; IR (CH2Cl2) 1603 (s),
1508, 1448, 1404, 1210, 1046 (vs, br), 837, 818 cm-1; Raman (KBr)
1605 (s), 1449, 1144, 1103, 634 cm-1. Anal. Calcd for C23H21BF4Ru2:
C, 47.11; H, 3.60. Found: C, 46.98; H, 3.59; N, 0.17.
1076 (vs), 922, 827, 693, 669 cm-1; IR (CH2Cl2) 1596, 1075 cm-1
.
Anal. Calcd for C23H21BF4FeRu: C, 51.05; H, 3.91. Found: C, 50.92;
H, 3.92; N, 0.14.
[Fc(CH)3Rc]+[PF6]- (5b). Red-brown platelike crystals of 5b (90
mg, 0.15 mmol, 60%) were obtained from crude FcCHdCHCH(OH)-
Rc (140 mg, 0.25 mmol) by the general procedure: IR (KBr) 3115,
2926, 1598 (s), 1449, 1409, 1374, 1280, 1215, 1141, 1106, 833 (s),
698, 634, 555 cm-1; IR (CH2Cl2) 1599, 840 cm-1. Anal. Calcd for
C23H21PF6FeRu: C, 46.09; H, 3.53. Found: C, 46.33; H, 3.50; N, 0.13.
X-ray Crystallography. Crystals of 1b, 2b, 3b, 4a, and 5b were
grown by layering dichloromethane solutions with diethyl ether, while
3c crystals were grown by layering a THF solution with pentane.
Crystallographic data were acquired using Mo KR radiation (λ )
0.710 69 Å) using Enraf-Nonius CAD-4 or Kappa CCD diffractometers
(CCD intensity data were processed using the programs DENZO and
SCALEPACK67). Solution and refinement was carried out using
[Rc(CH)3Rc]+[PF6]- (3b). Golden-brown plates of pure 3b (106
mg, 0.16 mmol, 71%) were obtained from the corresponding crude
alcohol (120 mg, 0.23 mmol) using the general method: IR (KBr) 3115,
2950, 1609 (s), 1503, 1449, 1409, 1210, 1136, 1036, 922, 837 (vs),
(65) Brookhart, M.; Grant, B.; Volpe, A. F. Organometallics 1992, 11, 3920-
3922.
(66) Cha´vez, I.; Alvarez-Carena, A.; Molins, E.; Roig, A.; Maniukiewicz, W.;
Aranacibia, A.; Aranacibia, V.; Brand, H.; Manr´ıquez, J. M. J. Organomet.
Chem. 2000, 601, 126-132.
(67) Otwinowski, Z.; Minor, W. In Processing of X-ray Diffraction Data
Collected in Oscillation Mode, Methods Enzymol.; Carter, C. W., Sweet,
R. M., Eds.; Academic Press: New York, 1997; Vol. 276.
9
J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002 6295

A R T I C L E S
Barlow et al.
Table 4. Details of Crystal Structure Determinations
1b
2b
3b
3c
4a
5b
formula
formula wt
cryst appearance
cryst size/mm3
cryst system
space group
diffractometer
T/K
C
23H21F6Fe2P
C39H53F6Fe2P
778.48
C23H21F6PRu2
644.51
dark golden plate
0.39 × 0.39 × 0.09
monoclinic
C2/c
C
55H33BF24Ru2
C
31H37BF4Fe2
C
23H21F6FePRu
554.07
1362.76
608.13
599.29
red fragment
0.44 × 0.37 × 0.30
orthorhombic
Pbcn
CAD-4
85
green blade
0.41 × 0.19 × 0.15
orthorhombic
P212121
CAD-4
dark red block
0.2 × 0.2 × 0.2
monoclinic
P21/c
Kappa CCD
150
greenish fragment
not measured
monoclinic
P21/n
CAD-4
85
dark brown plate
0.39 × 0.37 × 0.09
monoclinic
C2/c
CAD-4
85
CAD-4
85
85
a/Å
b/Å
c/Å
14.420(3)
12.267(2)
24.062(5)
90
4256.3(14)
8
1.729
1.499
2240
9.446(3)
17.746(5)
21.634(7)
90
3626.5(19)
4
1.426
0.902
1632
26.409(6)
15.004(3)
11.970(3)
113.83(2)
4338.6(17)
8
1.973
1.526
2528
16.5037(6)
18.7047(3)
18.3623(5)
111.946(2)
5257.6(3)
4
1.722
0.698
2688
8.702(3)
20.419(4)
15.446(3)
93.24(2)
2740(12)
4
1.474
1.103
1264
26.411(6)
14.753(3)
11.900(3)
112.92(2)
4270.7(17)
8
1.864
1.522
2384
â/deg
V/Å3
Z
D
calc/g cm-3
µ/mm-1
F(000)
θmax
25.0
9150
3752
0.034
25.0
11 341
6354
0.019
25.0
10 256
3818
0.0144
27.5
40 156
11 887
0.0568
25.0
11 091
4830
0.026
25.0
10 102
3753
0.0172
no. of reflctns
indept reflctns
Rint
absorpn corrctn
soln method
refinmnt
program
refinmnt
method
data/restr/
param
R1, wR2
(all data)
final GOF
(on F2)
none
SHELXS-97
SHELXL-97
none
SHELXS-97
SHELXL-97
ψ scan
SHELXS-97
SHELXL-97
from equiv refs
SHELXS-97
SHELXL-97
ψ scan
SHELXS-86
CRYM programs
ψ scan
isomorph. with 3b
SHELXL-97
full matrix F2
3752/0/310
0.101, 0.133
2.66
full matrix F2
6354/0/645
0.034, 0.060
1.51
full matrix F2
3818/0/392
0.024, 0.049
1.76
full matrix F2
11887/60/736
0.1269, 0.1817
1.021
full matrix F2
4830/0/343
0.059, 0.113
2.73
full matrix F2
3753/0/382
0.040, 0.076
2.32
∆Fmax
,
1.50, -0.59
0.37, -0.38
0.46, -0.45
1.24, -1.24
1.21, -0.80
1.21, -0.89
∆Fmin/e Å-3
SHELXS68 or CRYM69 programs (as specified in Table 4). Details of
the crystal data, data collection, and structure solutions and refinements
are given in Table 4; additional details available as Supporting
Information in the case of 2b, 3c, 4a, and 5b, while the Supporting
Information for ref 28 gives full details for the structures of 1b and
3b. The two largest difference peaks in the refined structure of 1b
indicate disorder, whereby the cations can adopt two alternative
orientations that are related to one another by an approximate reflection
in a plane perpendicular to the plane of the allylium bridge and passing
through both Fe atoms. This type of disorder was previously seen in a
number of other polymethine structures, for example, in that of
[Me2N(CH)7NMe2]+[Cl]-‚4H2O.70 However, in our case, one of the
orientations is clearly much less important than the other (∼1:9); we
were not able to adequately model the disorder, and so no disorder is
included in the resulting structure. However, a number of anomalous
bond lengths and angles in the structure reflect the presence of
unmodeled disorder. In the structure of 3c, two of the largest difference
peaks possibly relate to an alternative orientation of the allylium bridge
similar to that in 1b; again the disorder was not modeled, but here
there are no unusual bond lengths arising from the unmodeled disorder.
Some of the CF3 groups of the anion in 3c were modeled, using
restraints, as rotationally disordered. In the structures of both 3b and
5b, the hexafluophosphate anions occupy two different Wyckoff sites;
at both sites there is orientational disorder, which was successfully
modeled as two distinct orientations on each site. For each site, the
ratios between the two orientations are slightly different in the two
structures. Difference maps, as well as the final goodness of fit, for
the structure of 4a indicate some disorder associated with the tetrafluo-
roborate anion; however, we could not develop a chemically reasonable
model for this disorder and, therefore, treated it as an ordered group in
the refinement. Further details of the structures are given in the
Supporting information.
Acknowledgment. Support from the National Science Foun-
dation, Air Force Office of Scientific Research (AFOSR) at
Caltech is gratefully acknowledged. The research described in
this paper was performed in part by the Jet Propulsion
Laboratory (JPL), California Institute of Technology, as part
of its Center for Space Microelectronics Technology and was
supported by the Ballistic Missile Defense Initiative Organiza-
tion, Innovative Science and Technology Office through an
agreement with the National Aeronautics and Space Administra-
tion (NASA). We thank Kenneth Calgren for his assistance with
Raman measurements, Dr. Paul Roussel for his assistance with
low-temperature NMR experiments, Dr. Andrew Cowley for
his efforts with a highly disordered crystal of 3c‚xCH2Cl2, and
Johnson Matthey for a loan of ruthenium precursors. We also
thank the reviewers of the manuscript for helpful comments.
Supporting Information Available: Details of the crystal
structure determinations of 2b, 3c, 4a, and 5b in CIF format
(details for the structures of 1b and 3b have previously appeared
as Supporting Information for ref 28). Figure showing cyclic
voltammograms for 2a; 1H and 13C NMR spectra for McCHd
CHBr {Mc ) Fc, Fc′′, Rc} (PDF). This material is available
free of charge via the Internet at http://pubs.acs.org.
(68) Sheldrick, G. M. SHELXS-86, SHELXS-97, and SHELXL-97, Programs
for Crystallography; University of Go¨ttingen: Go¨ttingen, Germany, 1986
and 1997.
(69) Duchamp, D. J. Am. Crystallogr. Assoc. Meet., Bozeman, MT, 1964; Paper
B14.
(70) Groth, P. Acta Chem. Scand. 1987, B41, 547-550.
JA012472Z
9
6296 J. AM. CHEM. SOC. VOL. 124, NO. 22, 2002