Synthesis, redox chemistry, molecular and electronic structure of some cyclopentadienylcobalt pentafulvene complexes

Article abstract of DOI:10.1016/S0022-328X(99)00033-9

The pentafulvene complexes [(C₅R₅)Co(C₆H₄R′R″)] 3d (R = H, R′ = Me, R″ = Ph), 3f (R = H, R′ = Ph, R″ = C₆H₄NO₂-4), 3g (R = H, R′ = R″ = SMe) and 9 (R = Me, R′ = R″ = Ph) were prepared from the respective fulvenes and [(C₅R₅)Co(C₂H₄)₂] 8a,b. Protonation of 3 occurs at the fulvene C-α to give substituted cobaltocenium cations. The crystal and molecular structures of 3g, 9 and of the protonation product [3e + H]⁺ (R = R′ = Ph) were determined. In the neutral complexes the non-planar fulvene ligands are essentially η⁴-coordinated with short uncoordinated exocyclic carbon-carbon double bonds. In contrast, the protonated species exhibit n⁵-coordinated planar five-membered rings. Using cyclic voltammetry, the complexes 3e,f and 9 were shown to reversibly undergo one-electron oxidation and reduction reactions to give the cations [3e]⁺, [3f]⁺, [9]⁺ and anions [3e]^-, [3f]^-, [9]^-. Complex 3g also forms an anion [3g]^-, but oxidation is irreversible. The X-band ESR spectra of [3e]^-, [3e]⁺ and of [9]^-, [9]⁺ were recorded. In marked contrast to the cations, the anion radicals exhibit considerable anisotropy of the g and A tensors. d-Electron spin densities ρ^d = 0.5 for the radical anions were derived from a detailed analysis and simulation of the ESR spectra. Using extended Hueckel and Fenske-Hall MO calculations for the model complex [(C₅H₅)Co(C₅H₄CH₂)] 3h, the ESR and electrochemical properties were explained by the different metal participation in the HOMO (= SOMO of the cation radicals) and LUMO (= SOMO of the anion radicals).

Full text of DOI:10.1016/S0022-328X(99)00033-9

Journal of Organometallic Chemistry 579 (1999) 391–403
Synthesis, redox chemistry, molecular and electronic structure of some
cyclopentadienylcobalt pentafulvene complexes
Hubert Wadepohl *, Franz-Josef Paffen, Hans Pritzkow
Anorganisch-chemisches Institut der Ruprecht-Karls-Uni6ersita¨t, Im Neuenheimer Feld 270, D-69120 Heidelberg, Germany
Received 19 November 1998
Abstract
The pentafulvene complexes [(C5R5)Co(C6H4R%R%%)] 3d (R=H, R%=Me, R%%=Ph), 3f (R=H, R%=Ph, R%%=C6H4NO2-4), 3g
(R=H, R%=R%%=SMe) and 9 (R=Me, R%=R%%=Ph) were prepared from the respective fulvenes and [(C5R5)Co(C2H4)2] 8a,b.
Protonation of 3 occurs at the fulvene C-a to give substituted cobaltocenium cations. The crystal and molecular structures of 3g,
9 and of the protonation product [3e+H]+ (R=R%=Ph) were determined. In the neutral complexes the non-planar fulvene
ligands are essentially h4-coordinated with short uncoordinated exocyclic carbon–carbon double bonds. In contrast, the
protonated species exhibit h5-coordinated planar five-membered rings. Using cyclic voltammetry, the complexes 3e,f and 9 were
shown to reversibly undergo one-electron oxidation and reduction reactions to give the cations [3e]+, [3f]+, [9]+ and anions
[3e]−, [3f]−, [9]−. Complex 3g also forms an anion [3g]−, but oxidation is irreversible. The X-band ESR spectra of [3e]−, [3e]+
and of [9]−, [9]+ were recorded. In marked contrast to the cations, the anion radicals exhibit considerable anisotropy of the g
and A tensors. d-Electron spin densities zd=0.5 for the radical anions were derived from a detailed analysis and simulation of
the ESR spectra. Using extended Hu¨ckel and Fenske–Hall MO calculations for the model complex [(C5H5)Co(C5H4CH2)] 3h, the
ESR and electrochemical properties were explained by the different metal participation in the HOMO (=SOMO of the cation
radicals) and LUMO (=SOMO of the anion radicals). © 1999 Elsevier Science S.A. All rights reserved.
Keywords: Cyclopentadienylcobalt pentafulvene complexes; X-ray crystal structure; Cyclic voltammetry; EPR; Fenske–Hall MO
calculations
1. Introduction
According to spectroscopic and theoretical evidence,
the ground state of pentafulvenes is adequately de-
scribed by the classical cross-conjugated structure 1(A)
Pentafulvenes have a rich organometallic chemistry,
[1]. However, the chemical reactivity of these poly-
and a wide variety of types of metal complexes are
olefins is more readily understood when a high polariz-
known [3]. With 14 valence electron (VE) organometal-
ability in the direction of 1(B) is assumed [2]. The
resonance structure 1(C) is unfavorable, due to the
anti-aromaticity of the five-membered ring system.
lic fragments, fulvene complexes with essentially h4-co-
ordination of the ring are formed. Structurally
characterized systems include the iron [4,5], cobalt [6–
9], rhodium [7] and nickel [5] complexes 2–5. There
appears to be no tendency of the fulvene ligands in
these complexes to attain an h5-coordination to the
* Corresponding author. Tel.: +49-6221-544827; fax: +49-6221-
544197.
E-mail address: bu9@ix.urz.uni-heidelberg.de (H. Wadepohl)
metal, which would imply a strong polarization of the
fulvene ligand, as in 1(B) or 1(C).
0022-328X/99/$ - see front matter © 1999 Elsevier Science S.A. All rights reserved.
PII: S0022-328X(99)00033-9

392
H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
2. Results and discussion
2.1. Syntheses and spectra
The complex 3e (R=R%=Ph) was first prepared in
situ from [(C5H5)Co(CO)2] and 6,6-diphenyl fulvene 1e
at 140°C [11]. This synthesis is much improved when 8a
is used as a source of the (C5H5)Co fragment [6,7]. In
the same way, the new complexes 3d (R=Me, R%=
Ph), 3f (R=C6H4NO2-4, R%=Ph) and 3g (R=R%=
SMe) were prepared in 10–60% yield from 8a and the
respective fulvenes. Likewise, the pentamethylcyclopen-
tadienyl complex 9 was obtained in 65% yield from
[(C5Me5)Co(C2H4)2] 8b and 1e.
Most of the fulvene complexes are very air sensitive
deep red solids or oils, except 3f, which has a deep blue
colour. The latter complex was only obtained in rather
low yield. Separation from decomposition products was
very tedious, due to the low solubility of 3f and its high
sensitivity. All the complexes 3 rapidly decompose
when deposited on chromatography columns (SiO2 or
Al2O3/5% H2O). Treatment with strong acids resulted
in the formation of the substituted cobaltocenium salts
In contrast, the dinuclear ‘cobaltafulvene’ complexes
6 were found to have structures more like the zwitterion
6(B) than the carbene complex 6(A) [10]. These com-
plexes are related to 3, via the isolobal relationship of
the fragments CR2 and Co(L)(C5R5).
[(C5R%%)Co{C5H4C(H)RR%}]+.
5
Attempts to prepare (C5H5)Co complexes with 6-
phenylfulvene, 6,6-bis(nitro)fulvene and hexaphenylful-
vene were not successful. In all cases, decomposition to
dark intractable residues took place when the ligands
were treated with 8a at room temperature or heated
with [(C5H5)Co(CO)2].
The NMR spectra of the complexes 3 and 9 show the
usual high field coordination shifts of the fulvene ring
protons, which resonate in the range 3.95l54.7.
Consistent with the asymmetry of the respective fulve-
nes, a separate resonance appears for each of the four
unisochroneous fulvene ring protons of 3d and 3f. In
the more symmetric 3g and 9 these protons form a
(AB)2 spin system, which appears as two pseudo-triplets
in 9. In 3g, both signals coincide and only one multiplet
of relative intensity 4 is observed. The coordination
shifts in 9 [Dl=2.3 (H-2, H-5), 1.7 (H-3, H-4)] are
somewhat larger than those in 3e (Dl=1.9, 1.19),
consistent with the stronger electron donor properties
of the (C5Me5)Co fragments compared with (C5H5)Co.
However, similar coordination shifts (Dl=2.6, 1.6)
were found for the fulvene ring protons in 2 (R=R%=
Ph) [5], despite the quite different donor/acceptor prop-
erties of the (CO)3Fe and (C5Me5)Co fragments.
High field shifts on coordination are also observed
for the resonances of the fulvene carbons C1 to C6.
Consistent with an essentially h4-coordination of the
five-membered rings, the shifts are much stronger for
C2 to C5 than for the non-coordinated C1 and C6.
Unfortunately, reliable carbon spectra of 3f could not
be obtained, due to the low solubility of this complex.
Recently, we reported on the unexpected formation of
(1,1,2,2-tetramethyl-1,2-ethanediyl)bis(cobaltocene)
7
when 6,6-dimethylfulvene 1a was treated with the Jonas
reagent [(C5H5)Co(C2H4)2] 8a [9].
Apparently, the likely intermediate 3a (R=R%=Me) is
unstable and dimerizes to give 7. In contrast, other
derivatives of 3, including [(C5H5)Co(6-methyl-6-ethyl-
fulvene)] 3b and [(C5H5)Co(6,6-diethylfulvene)] 3c,
could be prepared and did not show any tendency to
dimerize [9]. Here we describe further investigations of
the synthesis, redox chemistry and molecular and elec-
tronic structure of fulvene complexes of the type
[(C5R5)Co(pentafulvene)].

H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
393
Table 2
2.2. Crystal and molecular structure of
˚
Selected bond lengths (A) and angles (°) for the complex cation
[(C5H5)Co(C5H4C(SMe)2)] (3g), [(C5Me5)Co(C5H4-
CPh2)] (9) and [(C5H5)Co(C5H4C(H)Ph2)]+
[PF6]− ([3e+H]+)
[(C5H5)Co{C5H4C(H)Ph2}] [3e+H]+a
Co(1)ꢀC(1)
Co(1)ꢀC(3)
Co(1)ꢀC(5)
C(1)ꢀC(2)
C(1)ꢀC(6)
C(3)ꢀC(4)
C(6)ꢀC(7)
C(2)ꢀC(1)ꢀC(5) 107.0(7)
C(2)ꢀC(1)ꢀC(6) 127.3(7)
C(2)ꢀC(3)ꢀC(4) 109.6(8)
C(1)ꢀC(5)ꢀC(4) 109.7(8)
C(1)ꢀC(6)ꢀC(13) 110.2(5)
2.034(7) Co(1)ꢀC(2)
2.002(8) Co(1)ꢀC(4)
2.009(8) Co(1)ꢀC(19)···C(23) 2.001(8)–2.019(8)
1.416(9) C(1)ꢀC(5)
1.520(9) C(2)ꢀC(3)
1.402(12) C(4)ꢀC(5)
1.523(9) C(6)ꢀC(13)
2.026(7)
1.996(8)
Single crystal X-ray structure determinations were
carried out for the fulvene complexes 3g, 9 and for the
diorganylmethyl substituted cobaltocenium cation
[3e+H]+. Crystallographic details are described in the
Section 3. Important bond lengths and angles are given
in Tables 1 and 2.
1.416(9)
1.422(11)
1.415(10)
1.532(8)
C(5)ꢀC(1)ꢀC(6)
C(1)ꢀC(2)ꢀC(3)
C(3)ꢀC(4)ꢀC(5)
C(1)ꢀC(6)ꢀC(7)
125.6(7)
107.3(8)
106.3(8)
114.6(5)
The gross molecular structures of 3g and 9 (Figs. 1
and 2) are quite similar and closely resemble the previ-
ously published [8] structure of 3e. In all cases, the
fulvene ligand binds to the cobalt in essentially h4-fash-
ion via its endocyclic 1,3-diene system. This type of
coordination is invariably present in all structurally
characterized pentafulvene complexes with a 14 VE
metal ligand fragment (Table 3). As usual in metal
complexes of 1,3-dienes, the metal carbon bonds in-
volving the ‘inner’ diene carbon atoms C3 and C4 are
somewhat shorter than those involving the ‘outer’ diene
carbons C2 and C5. In the free pentafulvenes, the
observed pattern of endocyclic carbon–carbon bond
lengths are in accord with a polyene structure [15]. In
complexes 3e,g and 9, this bond length alternation is
much less pronounced within the five-membered ring.
In contrast, the exocyclic bond C1ꢀC6 is quite short
both in the free ligands and in the complexes.
a Estimated standard deviations in parentheses.
The fulvene ligands are folded along C2···C5, to
move C1 and the exocyclic carbon–carbon bond away
Fig. 1. Molecular structure of 3g.
Table 1
˚
Selected bond lengths (A) and angles (°) for the complexes
[(C5R5)Co(C6H4R%2] 3g (R=H, R%=SMe) and 9 (R=Me, R%=Ph)a
3g
9
Co(1)ꢀC(2/5)
Co(1)ꢀC(3/4)
Co(1)···C(1)
2.057(4)/2.058(4)
1.993(5)/1.992(5)
2.316(4)
2.081(3)/2.051(3)
1.993(3)/1.971(3)
2.342(3)
Co(1)ꢀC(9)···C(13)
Co(1)ꢀC(19)···C(23)
C(1)ꢀC(2/5)
C(2/4)ꢀC(3/5)
C(3)ꢀC(4)
C(1)ꢀC(6)
C(6)ꢀS(1/2)
C(6)ꢀC(7/13)
S(1/2)ꢀC(7/8)
C(6)ꢀC(1)ꢀC(2/5)
C(3/4)ꢀC(2/5)ꢀC(1)
2.033(5)–2.072(5)
2.028(3)–2.087(3)
1.460(4)/1.455(4)
1.410(5)/1.418(5)
1.409(5)
1.454(5)/1.458(6)
1.403(6)/1.418(6)
1.421(6)
1.373(6)
1.758(4)/1.760(4)
1.389(4)
1.458(4)/1.483(4)
1.789(5)/1.808(5)
128.6(4)/128.8(4)
110.7(4)/109.7(4)
129.3(3)/128.7(3)
110.3(3)/110.5(3)
107.3(3)/107.8(3)
C(2/3)ꢀC(3/4)ꢀC(4/5) 107.4(4)/107.6(4)
C(1)ꢀC(6)ꢀS(1/2)
C(1)ꢀC(6)ꢀC(7/13)
C(6)ꢀS(1/2)ꢀC(7/8)
S(1)ꢀC(6)ꢀS(2)
120.5(3)/120.6(3)
122.7(3)/119.8(3)
103.3(2)/103.7(2)
118.9(2)
a Estimated standard deviations in parentheses.
Fig. 2. Molecular structure of 9.

394
H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
Table 3
˚
Characteristic structural parameters (distances in A, angles in °) of some complexes of the type [LnM(C5R4% CR2)]
LnM
R, R%
MꢀC2/5, ꢀC3/4a
M···C1
C1ꢀC6
Fold anglesb
Ref.
(CO)3Fe
Ph, H
Ph, H
SMe, H
Ph, H
C6H4Me-4, H
H, Me
H, Me
C6H4NMe2-4, H
Ph, H
2.142(4), 2.060(4)
2.043(3), 1.989(3)
2.058(4), 1.993(5)
2.066(3), 1.982(3)
2.157(3), 2.021(3)
2.248(5), 2.200(5)
2.245(8), 2.188(8)
2.193(7), 2.135(8)
2.20(1), 2.07(1)
2.517(4)
2.341(3)
2.316(4)
2.342(3)
2.517(3)
2.610
2.635(7)
2.468(7)
2.43
1.357(5)
1.372(5)
1.373(6)
1.389(4)
1.382(4)
1.339(7)
1.37(1)
1.36(1)
1.37(1)
1.38(1)
22, −6
16, −5
14, −4
14, −4
20, 0
25, −1
25, −3
14, −5
8, −2
10, 0
[5]
[8]
(C5H5)Co
(C5H5)Co
(C5Me5)Co
(Me3P)2(Me)Co
[(Me3P)3Rh]+
(dppe)MeIr
(ind)Rh
This work
This work
[12]
[13]
[14]
[7]
[5]
(cod)Nic
2.20(1), 2.07(1)
2.41
a Averaged across the pseudo-mirror plane of the fulvene ligand.
b The first number gives the angle between the planes C2ꢀC3ꢀC4ꢀC5 and C2ꢀC1ꢀC5; the second number gives the angle between the latter plane
and the vector C1ꢀC6.
c Two independent molecules.
from the metal atom. In addition, the exocyclic double
bond C1ꢁC6 is at an angle to the plane C2ꢀC1ꢀC5,
inclined towards the metal atom (Table 3). In the cobalt
fulvene complexes, the distance CoꢀC1 is 13–15%
longer than the bonds CoꢀC2/C5. A similar geometry is
attained by the other examples in Table 3 (10–18%
difference). In the literature, there are a few complexes
of the alpha-carbon atom C(6) from the ring plane
(0.07 A, away from the metal atom).
˚
2.3. Redox chemistry
Cyclic voltammetry was carried out with solutions of
the complexes 3e,f,g and 9. Relevant data are summa-
rized in Table 4. The potential/current response for 3g
and 9 is shown in Fig. 4. The monoanions are gener-
ated by a quasi-reversible electrode process in all cases.
In contrast, while reversible oxidation of 3e,f and 9 is
observed, oxidation of 3g is irreversible. As the oxida-
tion product of 3g, a new species is formed at −0.2 V
(vs. SCE), which gives rise to a new, essentially irre-
versible cathodic response at −0.64 V (Fig. 4).
˚
where cobalt carbon distances as large as 2.4 A have
been described as bonding. A clear cut case is
[{P(OMe)3}3Co(h3-benzyl)] [16] where the distance
˚
CoꢀCortho of 2.408(3) A must be assumed to be bonding
because of the diamagnetism of the molecule. However,
for all the complexes in Table 3, a strong tendency of
the fulvene ligands towards the h4-coordination is obvi-
ous not only from the folding of the fulvene ligand, but
also from the above mentioned slip of the metal atom
in the h4-coordination plane (C2ꢀC3ꢀC4ꢀC5), which
places the cobalt atom closer to C3 and C4.
The radical cation [3g]+ is obviously unstable, and
decomposes probably through cleavage of a carbon
sulfur bond. When generated in CH2Cl2 solution from
In principle, at least part of the folding and slip of
the fulvene ligands could be due to unfavourable steric
interactions between the metal atom and the sub-
stituents on C6 of the fulvene. In order to get an idea of
the relevance of such repulsions the molecular struc-
tures of [3c+H]+ and [3e+H]+ were determined.
Unfortunately, the poor quality of the available crystals
of [3c+H]+[BPh4]− resulted in a very limited dataset.
Apparent but unresolvable disorder of the ethyl groups
precluded a detailed analysis of this structure. The
structural
parameters
of
the
(diphenyl-
methyl)cyclopentadienyl ligand in [3e+H]+ (Fig. 3)
are quite similar to those of the 1,1%-bis(benz-
hydryl)cobaltocenium cation [17].
In both [3c+H]+ and [3e+H]+, the C5H5 and the
substituted cyclopentadienyl rings are planar and h5-co-
ordinated. There is only a minor distortion involving a
barely significant lengthening of the bonds Co1ꢀC1 and
Co1ꢀC2 (Table 2). The bulky diphenylmethyl sub-
stituent in [3e+H]+ causes an only slight displacement
Fig. 3. Molecular structure of [3e+H]+
.

H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
395
[(C5H5)2Fe]+[BF4]−) in situ. Typical spectra, taken in
frozen solutions at liquid nitrogen temperature, are
depicted in Figs. 5–7. The spectra of the anions are
very much alike, as are those of the cations. However,
the latter are distinctly different from the former.
The resonances of the anions are indicative of a
rhombic g tensor with superimposed 59Co (I=7/2)
hyperfine structure. Part of the low field component is
well resolved, and the corresponding parameters are
easily obtained. The nearly equally spaced lines of this
feature are an indication that at least this hyperfine
tensor component shares its principal axis with the g
tensor. Computer simulation resulted in the complete
assignment of the spectra. The g tensor components
and hyperfine splitting parameters A are given in Table
5.
Table 4
Cyclic
voltammetry
data
for
some
complexes
[(C5R5)Co(C5H4CRR%)]a
−2/−1
−1/0
0/+1
+1/+2
3eb
9b
−1.88
(110)
−1.51
(140)
−0.92
(80)
−1.22
(150)
−1.16
(60)
−0.20
(70)
−0.32
(110)
+0.02
(50)
+0.71
c
+1.57
c
3fd
3gb
−1.85
+0.94
c
c
−1.52
(110)
−0.92
(96)
−0.20
c
a For each complex, E° [V] vs. SCE is given in the first line and the
peak separation DEp [mV] in the second line. Supporting electrolyte:
nBu4NPF6.
b In propylene carbonate.
c Irreversible.
For [3e]− an isotropic spectrum in liquid solution
could also be obtained. The eight equally spaced lines
of this resonance correspond to an isotropic cobalt
hyperfine splitting ŽA=76×10−4 cm−1, which is
close to the average of the three principle values of the
hyperfine tensor, taken from Table 5. Therefore, the
three components of A must have the same sign (as-
sumed to be negative, see below). The spectra of [3e]−
and [9]− are remarkably similar to those of the cy-
clopentadienone complexes [(C5H5)Co(R4C4CO)]−
d In 1,2-dimethoxyethane.
Fig. 4. Cyclic voltammogram of 3g (top) and 9. Solvent: propylene
carbonate; supporting electrolyte: nBu4NPF6. The direction of the
potential scan is indicated by arrows.
3e and AgBF4 in a preparative scale, [3e]+ also slowly
decomposes. By mass spectrometry, the cobaltocenium
derivatives [(C5H5)Co(C5H4CPh2H)]+ and [(C5H5)Co-
(C5H4CPh2Cl)]+ were detected in the decomposition
products. These could be formed by hydrogen or chlo-
rine abstraction from the solvent.
ESR resonances of [3e]−, [9]− and [3e]+, [9]+ could
be observed from samples which were partially reduced
Fig. 5. (a) Experimental X-band ESR spectra of [3e]− in frozen THF
at liquid nitrogen temperature. (b) Calculated spectrum using the
parameters in Table 5.
(with
a
potassium mirror) or oxidized (with

396
H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
described below. In the LCAO-MO framework, the
deviation Dgi=gi−ge from the free-electron value of
the g tensor components may be expressed in terms of
the spin-orbit coupling parameters n, the spin density
z=a2, the LCAO coefficients cjk and the energy differ-
ence E0−Ek of the orbitals k relative to the SOMO
with the energy E0. The relevant expressions and their
derivation are given in the literature [19]. For a primar-
ily metal dyz based SOMO, the principal g tensor
components are given by
c2x −y ,k+3c2z ,k
2
2
2
Dgxx=2nCoa2 %
(1)
(2)
(3)
E0−Ek
k
c2xy,k
Dgyy=2nCoa2 %
E0−Ek
k
c2xz,k
Dgzz=2nCoa2 %
E0−Ek
k
The hyperfine tensor is best expressed in terms of the
isotropic coupling ŽA, the Fermi contact contribution
As and the departure from axial symmetry
1
3
ꢀAꢁ=As+ P(Dgx+Dgy+Dgz)
(4)
Fig. 6. (a) Experimental X-band ESR spectra of [9]− in frozen THF
at liquid nitrogen temperature. (b) Calculated spectrum using the
parameters in Table 5.
where P=282×10−4 cm−1 is the dipolar coupling
constant for cobalt [20]. Combination of the expres-
sions for gi and Ai gives an equation to compute the 3d
spin density [21] zd=a2
4
7
2
3
5
42
ꢂ
n
Ax−ꢀAꢁ=P − a2+ Dgx− (Dgy+Dgz)
(5)
The assignment of the experimental g1, g2 and g3 to gx,
gy and gz is based on the following reasoning. Ax is very
likely to have the same sign as ŽA; thus Ax\Ay or Az.
This leads to g1=gx. The orbital closest to the SOMO
13a%% is the empty 21a% (dxz) (see Fig. 8). The next
closest MO, the filled 20a% (dxz) is much lower. All other
filled orbitals with cobalt d contribution are at even
lower energy. With this condition, Eq. (3) suggests
DgzB0, and therefore we can completely assign the
spectra (Table 5). d-Electron spin densities zd, calcu-
lated from the dipolar coupling tensor according to Eq.
(5), are given in Table 6.
Fig. 7. Experimental X-band ESR spectra of [3e]+ (a) and [9]+ (b) in
frozen CH2Cl2 at liquid nitrogen temperature. St denotes the reso-
nance of LiTCNQ (g=2.0025).
In our complexes, the Fermi contact coupling entirely
arises from spin-polarization (As is therefore expected
to be negative). This is because admixture of Co 4s into
the SOMO belonging to the a%% representation is sym-
metry forbidden. In this case, zd can be computed
independently from the isotropic coupling by
[10]− (R=Ph, C6F5) [18a]. For both types of complex,
the large value of A1 is consistent only with a large Co
3d contribution to the singly occupied molecular orbital
(SOMO). Similar statements were made in a number of
related cases [18b–e].
As=Qd · zd
(6)
A more detailed interpretation of the ESR parame-
ters is based on the results of the MO calculations
The spin polarization coupling constant Qd= −131×
10−4 cm−1 for Co 3d electrons was taken from the

H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
397
Table 5
ESR spectroscopic parameters (in frozen THF at 80 K) for the radical anions [(C5R5)Co(6,6-Ph2C6H4)]− [3e]− (R=H), [9]− (R=CH3) and
[(C5H5)Co(R4C5O)]− (R=Ph, C6F5) [10]−a
gx
gy
gz
Žg
Ax
Ay
Az
ŽA
[3e]−
[9]−
2.103
2.092
2.103
2.037
2.032
2.025
1.933
1.920
1.906
2.03b
(−)142
(−)143
(−)141
(−)43
(−)38
(−)39
(−)54
(−)60
(−)51
76b
(−)80c
(−)77c
[10]−d
a Hyperfine coupling constants A are given in units of 10−4 cm−1
b From the isotropic spectrum in liquid solution at 270 K.
c 1/3(Ax+Ay+Az).
.
d Reassigned data from ref. [18a].
literature [22]. Computed values for As/Qd are also
given in Table 6. The isotropic hyperfine coupling
constant ŽA=1/3(Ax+Ay+Az) can also be calcu-
lated from Eqs. (6) and (4); the values so obtained are
given in the last column of Table 6. They are in
reasonable agreement with the experimental values
(Table 5).
Based on the above reasoning we reassigned the
published [18a]. ESR spectra of the anion radicals of
the cyclopentadienone complexes [10]−. An extended
Hu¨ckel MO calculation for the neutral model complex
[(C5H5)Co(C5H4O)] gave an ordering of molecular or-
bitals with metal character comparable to that in the
fulvene complex 3h [23]. In particular, the LUMO
(which would be the SOMO in the anion radical) has a
high contribution of dyz. Assignment of the resonances
as given in ref. [18a] results in zdB0. With our new
ordering gxBgyBgz (Table 5), a more reasonable value
of 0.53 is obtained for zd, very close to that of the
fulvene complexes.
cross-conjugated p-system in the fulvene complex an-
ions seems to be less efficient in that respect.
Interestingly, no stable 19 VE radicals could be ob-
tained from a source of the 15 VE fragment (Me3P)3Co
and diarylfulvenes. Instead, upon metal coordination
reductive carbon–carbon coupling takes place at the
1-positions of the fulvene C5 rings, to give diamagnetic
dinuclear (h3-allyl)cobalt(I) derivatives (18 VE at Co)
[12].
The ESR spectra of the cations show a much smaller
g anisotropy (Fig. 7). Although the poorly resolved
‘Hypervalent’ formally 19 VE organometallic com-
plexes [24] have been discussed within a continuum of
the electronic structure spanning the extremes of purely
metal-centered and largely ligand centered ‘18+d’ radi-
cals [25]. For example, a high metal character of the
SOMO was found for the complex [(C5H5)Co(1,2,5,6-
h-cycloocta-1,5-diene)]− (zd=0.68) [18b,19]. In con-
trast,
[(C5H5)Co(1-4-h-cycloocta-1,3,5,7-tetraene)]−
(zd=0.40) is much more delocalized [26]. In this com-
plex, the majority of the ligand charge resides on the
four carbon atoms of the uncomplexed ‘butadiene-like’
p-system of the cyclooctatetraene ligand. Another ex-
ample of a largely ligand based 18+d radical appears
to be the neutral [(C5H5)Co(C7H9)] (C7H9=cyclohepta-
dienyl), as shown by its dimerization via formation of a
carbon carbon bond between the two C7 polyolefins
[27]. In relation to the cyclooctadiene and cyclooctate-
traene complexes, the fulvene (and cyclopentadienone)
complexes attain an intermediate position. Generally,
formation of an 18+d electronic structure is favoured
by the presence in one of the ligands of an uncom-
plexed p-system, which can delocalize the unpaired
electron. As indicated by the zd values of about 0.5, the
Fig. 8. Orbital interaction diagram for [(h-C5H5)Co(h5-pentafulvene)]
3h. The fulvene ligand is assumed to be planar.
Table 6
d-Electron spin densities for the radical anions [(C5R5)Co(6,6-
Ph2C6H4)]− [3e]− (R=H), [9]− (R=CH3) and [(C5H5)Co(R4C5O)]−
(R=Ph, C6F5) [10]−
a
zd
Asa
As/Qd
ŽA
[3e]−
[9]−
0.51
0.51
0.53
−73.8
−83.8
−79.5
0.56
0.64
0.61
−73.0
−69.6
−72.0
[10]−
a In units of 10−4 cm−1
.

398
H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
shoulders in the low temperature spectra cannot be
unambigously assigned, it is apparent that all of the
59Co hyperfine interaction tensor elements are much
smaller than in the anions. This is consistent with a
mainly ligand centered radical. The formation of substi-
tuted cobaltocenium derivatives from [3e]+ on standing
in solution (i.e. the uptake of a hydrogen or chlorine
radical at C-6 of the fulvene) lends chemical evidence
for this conclusion and also points to a SOMO local-
ized on the fulvene ligand.
orbital sets, which are hybridized out away from the
C5H5 ligand, and the p orbitals of fulvene, as shown in
Fig. 8.
The observed potentials for the first oxidations and
reductions also correlate with the idea of reduction
taking place mainly at the metal atom, and oxidation at
the fulvene ligands. For example, methyl substitution at
the C5H5 ligand, which pushes electron density onto the
metal [28], induces a marked cathodic shift of the first
reduction potential. The first oxidation potential is only
little affected (compare 3e with 9, Table 4). In contrast,
oxidation becomes more difficult in 3f, where the nitro
group pulls out electron density from the fulvene
ligand.
One member of the 2e set on (C5H5)Co (dyz) forms a
strong bonding combination with a2 on fulvene. Inter-
action of these orbitals in the antibonding fashion,
shown above, gives the LUMO 13a%% of the complex 3h.
This orbital has a high metal contribution (see below).
The bonding combination of (C5H5)Co 2e (dxz) with
the empty fulvene 3b1 orbital is destabilized by an
antibonding admixture of 2b1 on fulvene (see above).
The resulting MO, 20a%, is essentially non-bonding and
forms the HOMO of 3h. This orbital has a large
amplitude (essentially pz) on the exo methylene carbon
atom C-6 of the fulvene ligand. Both folding and slip of
the fulvene ligand decrease the energy of the HOMO,
by an increase of the bonding and a decrease of the
antibonding interactions. The energy of the LUMO is
also lowered by the slip distortion, but is increased by a
larger folding. Hence, we expect to find the smallest €
and d in the cation [3h]+, where 20a% is half populated.
Interestingly, within reasonable ranges for the
2.4. Theoretical calculations
Extended Hu¨ckel (EH) molecular orbital calculations
were carried out on the model complex
[(C5H5)Co(pentafulvene)] 3h. Geometrical and compu-
tational details are given in Section 3. As a convenient
starting point, a geometry was chosen for 3h which had
the fulvene with a planar regular five-membered ring
symmetrically h5-coordinated to the (C5H5)Co frag-
ment. An interaction diagram for the construction of
the orbitals of 3h is shown in Fig. 8. For neutral 3h as
well as the radical ions [3h]− and [3h]+ the effects were
studied when the fulvene ring was independently folded
along C1···C4 by an angle € and translated by a
distance d parallel to the plane of the C5H5 ring.
˚
parameters € (−10–20°) and d (−0.20–0.30 A) the
total one-electron energy of the complex varied little,
only by about 0.2 eV for 3h and [3h]+, and by 0.35 eV
−
˚
for [3h] . Optimized values were €=8°, d=0.22 A for
+
˚
the neutral 3h, €=4°, d=0.12 A for [3h] and €=6°,
−
˚
d=0.34 A for [3h] [31]. Mulliken overlap populations
for the optimized structures are given in Scheme 1. No
bond appears to be present between the cobalt atom
and C-1 of the fulvene ligand in 3h and [3h]−, corre-
sponding to an h4-coordinated ligand. In [3h]+, the
slightly positive overlap population for these two atoms
seems to indicate a weak bonding interaction. However,
due to the shallow energy minimum, this difference may
not be real [32].
The electronic structure of pentafulvene 1h has been
studied extensively by MO methods. A comparison of
several semi-empirical and ab initio calculations has
appeared [29]. All calculations give a consistent descrip-
tion of pentafulvene and its 6,6-dialkyl and 6,6-diaryl
derivatives as typical polyenes 1(A). The valence or-
bitals of the (C5H5)Co fragment have also been dis-
cussed in detail elsewhere [30]. The most important
orbital interactions between this fragment and the ful-
vene ligand are those between the (C5H5)Co 2e and 2a1
Mulliken atomic charges as calculated for the opti-
mized structures are shown in Scheme 2. Although the
actual numbers must be taken carefully, owing to the
approximate nature of the method used, the trend is
clear. Coordination to the (C5H5)Co fragment induces

H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
399
Scheme 1. Mulliken overlap populations [×103] in [(C5H5)Co(pentafulvene)]n (n=0, +1, −1). Values for the cobalt carbon interactions are
written in italics.
Scheme 2. Calculated charge distribution in [(C5H5)Co(pentafulvene)]n (n=0, +1, −1).
a strong polarization of the fulvene, the negative charge
building up mainly on the exocyclic carbon atom C-6.
This is in the direction opposite to that observed in free
fulvenes with strong electron donors (e. g. an amino
group) on C-6, where negative charge accumulates on the
five-membered ring, effectively generating a 6p-system
[33]. We note here that such ligands coordinate in a h5
fashion to 12 VE (CO)3(d6-M) fragments to give internal
iminium and amidinium salts [34].
One-electron reduction and oxidation of the complex
affect the net charges both on the metal and on the
fulvene ligand (Scheme 2). Compared with the neutral
complex, ca. 0.5 electrons are removed from the fulvene
ligand in [3h]+, resulting in a greatly reduced negative
charge on C-6 in the radical cation. Upon reduction of
3h, ca. 0.3 electrons are transferred essentially to the
carbon and hydrogen atoms of the five-membered ring
of the fulvene ligand in [3h]−, without changing signifi-
cantly the net charges on C-1 and C-6.
To get more reliable orbital energies Fenske–Hall
approximate SCF-MO calculations were carried out for
3h, [3h]+ and [3h]− in two different molecular ge-
ometries, with the fulvene ligand in the planar, h5-coor-
dinated and the approximately h4-coordinated
geometry, respectively. Compositions of the relevant
orbitals are given in Table 7. Regardless of the coordina-
tion geometry of the fulvene ligand, a high metal content
(about 60%) is calculated for the HOMO of the radical
anion [3h]−. An estimation of the g tensor for [3h]−,
using Eqs. (1)–(3) and the eigenvalues and eigenvectors
obtained from the Fenske–Hall calculations, gave the
odering Dgx\Dgy\Dgz and DgzB0. This is consistent
with the observed spectra although the experimental
values of these parameters in 3e and 9 were not repro-
duced particularly well. A negative contribution to any
Dgi can only be caused by empty orbitals with high metal
content energetically close to the SOMO. The only such
orbital in [3h]− is the SOMO+1 (i.e. the LUMO of the
anion), which is only about 0.5 eV above the SOMO. The
high dxz content of this orbital accounts for the large
negative value of Dgz.
The HOMO of the radical cation [3h]+ has a large
amplitude on the exocyclic fulvene carbon atom C-6 and
a much smaller metal content. For this orbital, the
relative contribution of the atomic orbitals on these
atoms are more dependent on the coordination geometry
of the fulvene ligand (Table 7).
Clearly, the ESR spectra of [3e]+, [3e]− and [9e]+,
[9e]− are well-accounted for by the theoretical calcula-
tions. In particular, the anions are essentially 19 VE
complexes with a large cobalt content of the SOMO. The
singly occupied molecular orbital of the cations is much
more ligand centered (localized to a large extent on
fulvene C-6); this is consistent with a much smaller
spin–orbit coupling and cobalt hyperfine interaction in
the ESR spectra.
3. Experimental
3.1. General procedures
All operations were carried out under an atmosphere
of purified nitrogen (BASF R3-11 catalyst) using

400
H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
3.2. [(p-Cyclopentadienyl)cobalt(6-methyl-6-phenyl-
pentaful6ene)] (3d)
Schlenk techniques. Solvents were dried by conven-
tional methods. The complexes [(C5H5)Co(6,6-R2C6H4)]
3e (R=Ph) [6], 3c (R=Et) [9], [(C5R5)Co(C2H4)2] 8a
(R=H) [35], 8b (R=Me) [36] and the fulvene ligands
6,6-R,R%-C6H4 1d (R=Me, R%=Ph) [15], 1e (R=R%=
Ph) [15], 1f (R=Ph, R%=p-NO2C6H4) [37], and 1g
(R=R%=SMe) [38] were prepared as described in the
literature. Petroleum ether refers to the fraction with
b.p. 30–60°C. NMR spectra were obtained on Bruker
AC 200 or AC 300 instruments (200.1 or 300.1 MHz
A 800 mg (4.77 mmol) sample of 6-methyl-6-phenyl-
fulvene 1d was added to a stirred solution of 860 mg
(4.77 mmol) of 8a in 50 ml of petroleum ether at room
temperature. A gas evolved and a the solution turned
red. After stirring for 15 h, a small amount of a dark
precipitate was removed by filtration. The product did
not crystallize from petroleum ether, toluene or diethyl
ether solutions at −25°C, and could only be obtained
as a very air sensitive deep red oil after complete
removal of solvent in vacuo.
for H, 50.3 MHz for 13C). H and 13C chemical shifts
are reported versus SiMe4 and were determined by
reference to internal SiMe4 or residual solvent peaks.
The multiplicities of the 13C resonances [odd (u) or even
(g)] were determined using the DEPT or the J-modu-
lated (JMOD) spin echo techniques. ESR spectra were
recorded on a Varian E6 spectrometer in the X-band
(9.3 GHz). The spectra were calibrated with a LiTCNQ
standard, Žg=2.0025. Simulation of the anisotropic
spectra was carried out with the program SimFonia [39]
using second-order perturbation theory. Mass spectra
were measured in the electron impact ionization mode
at 70 eV on a Finnigan MAT 8230 instrument. Electro-
chemical experiments were carried out using a EG&G
PARC model 173 potentiostat and a model 175 Univer-
sal Programmer; a platinum disk working electrode was
employed. Redox potentials are referenced to the stan-
dard calomel electrode (SCE). Elemental analyses were
performed locally by the microanalytical laboratory of
the organisch-chemisches Institut der Universita¨t
Heidelberg.
1
1
1H-NMR (200 MHz, in C6D6): l=1.86 (s, 3H, Me),
3.91 (br.s, 1H, CH), 4.420 (s, 5H, C5H5), 4.425 (s, 1H,
CH), 4.96 (br.s, 1H, CH), 5.04 (br.s, 1H, CH), 7.01 (m,
1H, Ph), 7.27 (m, 2H, Ph), 7.46 (m, 2H, Ph). 13C{1H}-
NMR (in C6D6): l=19.5 (Me), 55.8 (C-2 or C-5), 58.0
(C-5 or C-2), 77.0 (C-3 or C-4), 77.6 (C-4 or C-3), 78.9
(C5H5), 98.6 (C-1), 123.7 (g, Ph), 126.8 (g, Ph), 128.5 (g,
Ph), 145.5 (u, C-6). Reliable microanalyses could not be
obtained due to the high air sensitivity of the liquid
product.
3.2 [(p-Cyclopentadienyl)cobalt(6-phenyl-6-p-nitro-
phenylpentaful6ene)] (3f)
A mixture of 970 mg (4.82 mmol) of 8a and 1.09 g
(4.82 mmol) of 6-p-nitrophenyl-6-phenylfulvene 1f in 60
ml of toluene was stirred at room temperature. The
mixture quickly turned blue and evolution of a gas was
Table 7
Composition (%) of the frontier molecular orbitals for the model complexes [(C5H5)Co(C5H4CH2)]n 3h as derived from Fenske–Hall MO
calculations
n
Orbital
Energy (eV)
Co 3d
Co 4s
Co 4p
C-1
C-2
C-3
C-6
Ful6ene ligand planar
0
0
0
HOMO
LUMO
LUMO+1
−7.70
−4.27
−2.99
20.5a
51.7b
34.8a
0.5
0
0.02
2.0
0.6
3.4
2.3
0
0
2.3
8.9
5.7
0
2.2
7.4
60.0
0
14.3
−1
−1
SOMO
LUMO
2.54
3.56
60.5b
45.8a
0
0
0.6
2.9
0.3
3.8
7.6
5.1
1.7
6.9
0
8.5
+1
SOMO
−14.77
17.7a
0.6
3.6
0.2
3.7
0.4
57.6
Ful6ene ligand bent
0
0
0
HOMO
LUMO
LUMO+1
−9.03
−3.50
−2.64
30.0a
52.3b
33.6a
1.1
0
0
5.4
1.06
2.7
1.8
0.5
4.8
5.9
9.3
6.0
0.9
2.8
8.2
37.7
0.01
15.2
−1
−1
SOMO
LUMO
3.35
3.89
61.2b
44.5a
0
0
1.0
2.0
0.5
5.6
8.1
4.5
2.1
7.3
0.01
10.8
+1
SOMO
−15.50
26.7a
1.1
6.8
1.1
6.4
1.1
34.5
a Mainly 3dxz
b Mainly 3dyz
.
.

H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
401
observed. After 15 h, a large amount of a dark in-
tractable precipitate was removed by filtration. Solvent
was removed from the blue filtrate in vacuo, and the
residue redissolved in CH2Cl2. Petroleum ether was
added, and the solution was cooled to −25°C. A first
crop of the product 3f precipitated as a blue powder
after several days. Further product fractions were ob-
tained from the mother liquor by repeating the above
procedure. Total yield 380 mg (20%).
1H-NMR (200 MHz, in C6D6): l=1.51 (s, 15H,
Me), 3.96 (m, 2H, H-2/H-5), 4.36 (m, 2H, H-3/H-4),
7.00–7.57 (m, 10H, Ph). 13C{1H}-NMR (in C6D6):
l=10.2 (g, C5Me5), 63.0 (g, C-3/C-4 or C-2/C-5), 78.0
(g, C-2/C-5 or C-3/C-4), 90.2 (u, C5Me5), 99.7 (u, C-1),
123.7 (g, Ph), 127.2 (g, Ph), 130.1 (g, Ph), 146.0 (u,
C-6). MS: m/z=424 (100%, M+), 409 (24, [M−
CH3]+, 229 (5, [Ph2C6H3]+), 212 (20), 57 (5). Anal.
Calc. for C28H29Co (424.47): C, 79.23; H, 6.89; Found:
C, 79.09; H, 7.03.
1H-NMR (200 MHz, in C6D6): l=4.27 (m, 1H,
CH), 4.30 (s, 5H, C5H5), 4.46 (m, 1H, CH), 4.64 (m,
1H, CH), 4.72 (m, 1H, CH), 6.98–7.17 (m, 5H, Ph),
7.13 and 8.09 (m, AA%BB% pattern, 2×2H, C6H4).
13C{1H}-NMR (in C6D6): l=78.8 (g, CH), 80.3 (g,
C5H5), 81.4 (g, CH), 102.7 (u, C-1), 124.1 (g, CH),
125.9 (g, CH), 126.7 (g, CH), 129.0 (g, CH), 131.0 (g,
CH), 141.0 (u, C), 144.1 (u, C), 150.8 (u, C). MS:
m/z=399 (53%, M+), 370 (7), 351(8), 293 (11), 275
(100, [M−C5H5Co]+), 229 (58, [L−NO2]+), 228 (94),
226 (58), 215 (13), 202 (30), 189 (12, [Co(C5H5)2]+), 153
(8, [L−C6H4NO2]+), 152 (10), 150 (6), 124 (4,
[Co(C5H5)]+), 120 (7), 114 (7), 105 (9) (L=6-p-nitro-
phenyl-6-phenylfulvene). Anal. Calc. for C23H18CoNO2
(399.33): C, 69.18; H, 4.54; N, 3.51; Found: C, 68.18;
H, 4.76; N, 3.37.
3.6. [(p-Cyclopentadienyl){p-(pent-3-yl)cyclopentadien-
yl} cobalt ([3c+H]+)
A 260 mg (1.01 mmol) sample of 3c in 10 ml of
diethyl ether was treated with 0.5 ml HBF4 (80% in
diethyl ether). The mixture was stirred at ambient tem-
perature for 30 min. The colourless ether phase was
then decanted from a brown oil, which was repeatedly
washed with ether and then dissolved in acetone. A
solution of NaBPh4 (340 mg, 1 mmol) in 10 ml of
acetone was added. On cooling of the mixture to
−25°C, the product [3c+H]+ (350 mg, 60%) crystal-
lized as dark needles.
1H-NMR (200 MHz, in CD2Cl2): l=0.89 (t, 6H,
Me), 1.62 (m, 4H, CH2), 2.38 (m, 1H, H(C-a)), 5.53 (m,
2H, H-2/H-5 or H-3/H-4), 5.65 (s, 5H, C5H5), 5.68 (m,
2H, H-3/H-4 or H-2/H-5). 13C{1H}-NMR (in CD2Cl2):
l=10.0 (g, CH3), 25.7 (u, CH2), 39.1 (g, C-a), 82.6 (g,
C-2/C-5 or C-3/C-4), 82.9 (g, C-3/C-4 or C-2/C-5), 84.8
(g, C5H5), 85.1 (u, Cipso). FD-MS: m/z=259 (100%,
M+).
3.4. [(p-Cyclopentadienyl)cobalt(6,6-bis(methylmer-
capto)pentaful6ene)] (3g)
A 570 mg (3.3 mmol) sample of bis(methylmer-
capto)fulvene 1g was added to a stirred solution of 600
mg (3.3 mmol) of 8a in 50 ml of petroleum ether at
room temperature. A gas evolved and a red solid
precipitated. After standing for 36 h the mother liquor
was decanted from the precipitate, which was then
recrystallized from toluene. The product 3g (580 mg,
60%) crystallized as red platelets at −6°C, m.p. 88°C.
1H-NMR (300 MHz, C6D6): l=2.25 (s, 6H, CH3),
4.30 (s, 5H, C5H5), 4.71 (m, 4H, CH). 13C{1H}-NMR
(in C6D6): l=16.8 (g, CH3), 59.0 (g, CH), 77.2 (g, CH),
80.0 (g, C5H5), 139.7 (u, C-6). MS: m/z=294 (25%,
M+), 279 (100, [M−CH3]+), 246 (15), 232 (24, [M−
SMe2]+), 188 (9), 169 (14), 124 (11, [Co(C5H5)]+), 116
(14), 59 (9, Co+). Anal. Calc. for C13H15CoS2 (294.32):
C, 53.05; H, 5.14; Found: C, 52.79; H, 5.26.
3.7. X-ray crystal structure determinations of
[(C5H5)Co{6,6-bis(mercaptomethyl)ful6ene)] 3g,
[(C5Me5)Co(6,6-diphenylful6ene)] 9 and
[(C5H5)Co{C5H5C(H)Ph2}][PF6] [3e+H]+[PF6]−
Single crystals of 3g and 9 were obtained from hex-
ane solutions at −20°C. The salt [3e+H]+[PF6]− was
prepared from 3e and HPF6; crystallization from a
methylene chloride solution at 6°C gave single crystals.
The crystals were mounted in Lindemann capillary
tubes and transferred to a STOE-Siemens four circle
diffractometer. Intensity data were collected at ambient
temperature and corrected for Lorentz, polarization
and absorption effects (Table 8). The structures were
solved by the Patterson or direct methods and refined
on F2 with full-matrix least-squares using all measured
unique reflections. All non-hydrogen atoms in 3g, 9 and
[3e+H]+[PF6]− were given anisotropic displacement
parameters (ADPs). For complex 3g, all hydrogen
atoms were located from difference Fourier syntheses
and refined with isotropic ADPs. For the complexes 9
and [3e+H]+[PF6]− the hydrogen atoms were inserted
in calculated positions [40].
3.5. [(p-Pentamethylcyclopentadienyl)cobalt(6,6-di-
phenylpentaful6ene)] (9)
A petroleum ether solution (60 ml) of 740 mg (2.96
mmol) of 8b and 680 mg (2.96 mmol) of 6,6-diphenyl-
fulvene 1e was stirred at room temperature for 18 h.
After filtration, the volume of the deep red mixture was
reduced in vacuo to about half. On cooling to −25°C,
the product 9 (820 mg, 65%) crystallized as deep red
cubes, m.p. 108°C.

402
H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
Table 8
Details of the crystal structure determinations of [(h-C5R5)Co(6,6-R2% C6H4)] 3g (R=H, R%=SMe),
9
(R=Me, R%=Ph) and [(h-
[3e+H]+[PF6]−
C5H5)Co{C5H4C(H)Ph2}]+[PF6]− [3e+H]+[PF6]−
3g
9
Formula
C13H15CoS2
Orthorhombic
Pbca
7.502(4)
17.143(9)
20.264(10)
90
C28H29Co
Monoclinic
P21/n
8.960(6)
25.082(12)
10.590(8)
107.20(3)
2274(3)
4
C23H20CoF6P
Monoclinic
P21/n
9.062(7)
22.670(16)
10.757(8)
103.71(6)
2147(3)
4
Crystal system
Space group
˚
a (A)
˚
b (A)
˚
c (A)
i (°)
V (A )
3
˚
2606(2)
8
Z
Mr
294.32
1.500
1216
424.47
1.240
896
500.29
1.548
1016
0.93
dc (g cm−1
F000
)
v(Mo–Ka) (cm−1
X-radiation, u (A)
)
1.61
0.77
˚
Mo–Ka, graphite monochromated, 0.71069
Data collect. temperature
2r range [°]
Ambient
3–50
Ambient
3–51
Ambient
3–47
hkl-range
h
k
l
0–8
0–20
0–24
−10–10
0–29
0–12
−10–9
0–25
0–12
Reflections measured
Unique
2291
4142
3152
Observed [I52|(I)]
Absorption correction
Parameters refined
R-values
1514
Empirical
184
2895
Empirical
271
1676
Empirical
288
R (obs. reflections only)
wR2 (all reflections)
(w=1/[|2(F)+(A · P)2+B · P],
A, B
P
GoF
0.039
0.092
0.045
0.117
0.059
0.211
0.0307, 0.76
max(F2o, 0)+2F2c)/3)
1.016
0.0575, 0.54
0.0518, 2.81
1.036
0.33/−0.25
1.402
0.45/−0.31
−3
˚
Largest/smallest peak in difference-Fourier (e A
)
0.26/−0.45
The calculations were performed using the programs
SHELXS-86 and SHELXL-93 [41]. Graphical representa-
tions were drawn with the SCHAKAL-92 program [42].
group using the numerical Xa atomic orbital program
of Herman and Skilman [48] in conjunction with the
Xa-to-Slater basis program of Bursten and Fenske [49].
Exponents of 4s and 4p atomic orbitals for cobalt were
set to 2.0. These functions are less diffuse than the
atomic functions but lead to a better description of the
bonding in organometallic complexes [50]. A value of
1.20 was used for the hydrogen exponent. Idealized
models were used for the complexes studied. The un-
paired electron in the ‘open-shell’ systems was treated
as two half-electrons of opposite spin [51]. The geome-
try of 3h was set up with Cs molecular symmetry. The
3.8. Molecular orbital calculations
Extended Hu¨ckel calculations [43] were carried out
with CACAO [44] using the wavefunctions supplied with
the package. The basis set for the metal atom consisted
of ns, np, and (n−1) d orbitals. The s and p orbitals
were described by single Slater-type wave functions,
and the d orbitals were taken as contracted linear
combinations of two Slater-type wave functions. The
weighted Wolfsberg–Helmholtz formula [45] was em-
ployed. The iterative self-consistent field Fenske–Hall
procedure [46] is an approximation [47] of the Hartree–
Fock–Rothaan method and employs the atomic basis
functions and the molecular geometry as the only ad-
justable parameters. The STO basis functions used in
the FH calculations have been developed by the Fenske
˚
following distances (A) were used: CoꢀC(C5H5) 2.06,
Coꢀfulvene(normal) 1.68, CꢀC 1.40, CꢀH 1.08.
Acknowledgements
This work was supported by the Deutsche
Forschungsgemeinschaft under the auspices of the Son-

H. Wadepohl et al. / Journal of Organometallic Chemistry 579 (1999) 391–403
403
[27] W.E. Geiger, T. Gennett, G.A. Lane, A. Salzer, A.L. Rheingold,
Organometallics 5 (1986) 1352.
[28] D.C. Calabro, J.L. Hubbard, C.H. Blevins II, A.C. Campbell,
D.L. Lichtenberger, J. Am. Chem. Soc. 103 (1981) 6839.
[29] K.N. Houk, J.K. George, R.E. Duke, Tetrahedron 30 (1974)
523.
derforschungsbereich 247 of the Universita¨t Heidelberg
and by the Fonds der Chemischen Industrie. A Heisen-
berg fellowship to H.W. is gratefully acknowledged.
[30] M. Elian, M.M.L. Chen, D.M.P. Mingos, R. Hoffmann, Inorg.
Chem. 15 (1976) 1148.
References
[31] For comparison, EH calculations were also carried out for the
complex [(C5H5)Co(C5H6)] with the geometric parameters € and
d defined as for 3h. Not surprisingly, this structure optimized to
[1] D. Lloyd, Non-Benzenoid Conjugated Carbocyclic Compounds,
Elsevier, Amsterdam, 1984.
˚
[2] (a) P.A. Straub, D. Meuche, E. Heilbronner, Helv. Chim. Acta
49 (1966) 517. (b) E.D. Bergmann, Chem. Rev. 68 (1968) 41.
[3] R.C. Kerber, D.J. Ehntholt, Synthesis (1970) 449.
[4] E. Weiss, W. Hu¨bel, Chem. Ber. 95 (1962) 1186.
[5] F. Edelmann, B. Lubke, U. Behrens, Chem. Ber. 115 (1982)
1325.
[6] H. Wadepohl, Habilitationsschrift, Universita¨t Heidelberg, 1989.
[7] D. Rau, U. Behrens, J. Organomet. Chem. 387 (1990) 219.
[8] H. Wadepohl, H. Pritzkow, Acta Crystallogr. Sect. C 47 (1991)
2061.
a deeper minimum at larger € and d, 25° and 0.25 A, respec-
tively, than those obtained for 3h, corresponding to an h4-coor-
dinated cyclopentadiene ligand.
[32] In a structure of 3h where the coordination geometry of the
fulvene ligand was forced to h5 (i.e. symmetrical bonding of the
C5 ring of a planar fulvene to the cobalt atom), the overlap
population CoꢀC(1) came out only about half that of CoꢀC(2)
through to C(5). This quantity was only slightly dependent on
the molecular charge (increased by 0.005 in the cation, decreased
by 0.008 in the anion).
[9] H. Wadepohl, C.-W.v.d. Lieth, F.-J. Paffen, H. Pritzkow, Chem.
Ber. 128 (1995) 317.
[33] K. Hafner, K.H. Ha¨fner, M. Kreuder, G. Ploss, G. Schulz, E.
Sturm, K.H. Vo¨lpel, Angew. Chem. 75 (1963) 35.
[10] (a) H. Wadepohl, H. Pritzkow, Angew. Chem. 99 (1987) 132. (b)
H. Wadepohl, W. Galm, H. Pritzkow, A. Wolf, J. Chem. Soc.
Chem. Commun. (1993) 1459. (c) H. Wadepohl, W. Galm, H.
Pritzkow, A. Wolf, Chem. Eur. J. 2 (1996) 1453.
[34] P. Bickert, B. Hildebrandt, K. Hafner, Organometallics 3 (1984)
653 and references cited therein.
[35] K. Jonas, E. Deffense, D. Habermann, Angew. Chem. 95 (1983)
729; Angew. Chem. Suppl. (1983) 1005.
[11] J.E. Sheats, W. Miller, T. Kirsch, J. Organomet. Chem. 91
(1975) 97.
[12] H.F. Klein, E. Auer, T. Jung, C. Ro¨hr, Organometallics 14
(1995) 2731.
[13] D.W. Hoard, P.R. Sharp, Inorg. Chem. 32 (1993) 612.
[14] D.S. Glueck, R.G. Bergman, Organometallics 9 (1990) 2862.
[15] K.P. Zeller, in: H. Kropf (ed.), Methoden der Organischen
Chemie (Houben-Weyl), Vol. 5/2c, Thieme, Stuttgart 1985, and
references cited therein.
[36] S.A. Frith, J.L. Spencer, Inorg. Synth. 23 (1985) 15.
[37] G. Kresze, S. Rau, G. Satelus, H. Goetz, Liebigs Ann. Chem.
648 (1961) 51.
[38] R. Gompper, E. Kutter, Chem. Ber. 98 (1965) 2825.
[39] WINEPR SimFonia, version 1.25, Bruker Analytische
Meßtechnik, Karlsruhe, 1996.
[40] (a) SHELXS-86, G.M. Sheldrick, Acta Cryst. A46 (1990) 467.
(b) SHELXL-93, G.M. Sheldrick, Universita¨t Go¨ttingen 1993.
[41] Crystallographic data (excluding structure factors) for the struc-
tures reported in this paper have been deposited with the Cam-
bridge Crystallographic Data Centre as supplementary
publication Nos. CCDC 113629 (3g), 113630 (9) and 113631
([3e+H]+). Copies of the data can be obtained free of charge
on application to The Director, CCDC, 12 Union Road, Cam-
bridge CB2 1EZ, UK (Fax: +44-1223-336-033; e-mail:
teched@chemcrys.cam.ac.uk).
[42] SCHAKAL-92, E. Keller, Universita¨t Freiburg, 1992.
[43] (a) R. Hoffmann, J. Chem. Phys. 39 (1963) 1397. (b) R. Hoff-
mann, W.N. Lipscomb, J. Chem. Phys. 36 (1962) 2179, 3489.
[44] C. Mealli, D.M. Proserpio, J. Chem. Edu. 67 (1990) 39.
[45] J.H. Ammeter, H.-B. Bu¨rgi, J.C. Thibeault, R. Hoffmann, J.
Am. Chem. Soc. 100 (1978) 3686.
[16] J.R. Bleeke, R.R. Burch, C.L. Coulman, B.C. Schardt, Inorg.
Chem. 20 (1981) 1316.
[17] O. Koch, F. Edelmann, U. Behrens, Chem. Ber. 115 (1982) 1305.
[18] (a) H.v. Willigen, W.E. Geiger Jr., M.D. Rausch, Inorg. Chem.
16 (1977) 581. (b) T.A. Albright, W.E. Geiger Jr., J.
Moraczewski, B. Tulyathan, J. Am. Chem. Soc. 103 (1981) 4787.
(c) W.E. Geiger, T. Gennett, M. Grzeszczuk, G.A. Lane, J.
Moraczewski, A. Salzer, D.E. Smith, J. Am. Chem. Soc. 108
(1986) 7454. (d) L. Vasquez, H. Pritzkow, U. Zenneck, Angew.
Chem. 100 (1988) 705. (e) B.L. Ramakrishna, A.K. Salzer, U.
Ruppli, J.H. Ammeter, U. Ko¨lle, Inorg. Chem. 25 (1986) 1364.
[19] P.H. Rieger, in: W.C. Trogler (ed.), Organometallic Radical
Processes, Elsevier, Amsterdam, 1990, chapter 8.
[20] J.R. Morton, K.F. Preston, J. Magn. Reson. 30 (1978) 577.
[21] P. Coppens, T. Koritsanszky, P. Becker, Chem. Scr. 26 (1986)
463.
[22] B.M. Peake, P.H. Rieger, B.H. Robinson, J. Simpson, J. Am.
Chem. Soc. 102 (1980) 156.
[46] (a) M.B. Hall, R.F. Fenske, Inorg. Chem. 11 (1972) 768. (b)
R.F. Fenske, Pure Appl. Chem. 60 (1988) 1153.
[47] B.E. Bursten, Pure Appl. Chem. 63 (1991) 839.
[48] F. Herman, S. Skillman, Atomic Structure Calculations, Pren-
tice-Hall, Englewood Cliffs, NJ, 1963.
[23] J.H. Bieri, A.S. Dreiding, T.C.C. Gartenmann, E.R.F. Gesing,
R.W. Kunz, R. Prewo, J. Organomet. Chem. 306 (1986) 241.
[24] D.R. Tyler, Acc. Chem. Res. 24 (1991) 325.
[49] (a) B.E. Bursten, R.F. Fenske, J. Chem. Phys. 67 (1977) 3138.
(b) B.E. Bursten, R.J. Jensen, R.F. Fenske, J. Chem. Phys. 68
(1978) 3320.
[25] W.E. Geiger, Acc. Chem. Res. 28 (1995) 351.
[26] W.E. Geiger, P.H. Rieger, C. Corbato, J. Edwin, E. Fonseca,
G.A. Lane, J.M. Mevs, J. Am. Chem. Soc. 115 (1993) 2314.
[50] R.F. Fenske, D.D. Radtke, Inorg. Chem. 7 (1968) 479.
[51] T.F. Block, R.F. Fenske, C.P. Casey, J. Am. Chem. Soc. 98
(1976) 441.
.