Compromise between conjugation length and charge-transfer in nonlinear optical η5-monocyclopentadienyliron(II) complexes with substituted oligo-thiophene nitrile ligands: Synthesis, electrochemical studies and first hyperpolarizabilities

Article abstract of DOI:10.1016/j.jorganchem.2007.03.023

A systematic series of η⁵-monocyclopentadienyliron(II) complexes with substituted oligo-thiophene nitrile ligands of general formula [FeCp(P_P)(NC{SC₄H₂}_nNO₂)][P F₆] (P_P = dppe, (+)-diop;

Full text of DOI:10.1016/j.jorganchem.2007.03.023

Journal of Organometallic Chemistry 692 (2007) 3027–3041
www.elsevier.com/locate/jorganchem
Compromise between conjugation length and charge-transfer
in nonlinear optical g5-monocyclopentadienyliron(II) complexes
with substituted oligo-thiophene nitrile ligands:
Synthesis, electrochemical studies and first hyperpolarizabilities
a,b,
M. Helena Garcia
Jochen Campo , Wim Wenseleers *, Etienne Goovaerts
b
*, Paulo J. Mendes b,c, M. Paula Robalo b,d, A. Romao Dias ,
˜
e
e,
e
a
ˆ
Faculdade de Ciencias da Universidade de Lisboa, Ed.C8, Campo Grande, 1749-016 Lisboa, Portugal
Centro de Quı´mica Estrutural, Instituto Superior Te´cnico, Universidade Te´cnica de Lisboa, Av. Rovisco Pais, 1049-001 Lisboa, Portugal
b
c
´
´
Centro de Quı´mica de Evora, Universidade de Evora, Rua Roma˜o Ramalho 59, 7002-554 Evora, Portugal
´
d
Departamento de Engenharia Quı´mica, Instituto Superior de Engenharia de Lisboa, Rua Conselheiro Emı´dio Navarro, 1, 1959-007 Lisboa, Portugal
e
Department of Physics, University of Antwerp, Campus Drie Eiken, Universiteitsplein 1, B-2610 Wilrijk-Antwerpen, Belgium
Received 12 February 2007; received in revised form 17 March 2007; accepted 17 March 2007
Available online 23 March 2007
Abstract
A systematic series of g5-monocyclopentadienyliron(II) complexes with substituted oligo-thiophene nitrile ligands of general formula
[FeCp(P_P)(NC{SC4H2}nNO2)] [PF6] (P_P = dppe, (+)-diop; n = 1–3) has been synthesized and characterized. The electrochemical
behaviour of the new compounds was explored by cyclic voltammetry. Quadratic hyperpolarizabilities (b) of the complexes with dppe
coligands have been determined by hyper-Rayleigh scattering (HRS) measurements at two fundamental wavelengths of 1.064 and
1.550 lm, to uncover the two-photon resonance effect and to estimate static b values. The obtained overall results are found to be better
than for the related g5-monocyclopentadienyliron(II) complexes with p-benzonitrile derivatives. Although an increase of the resonant b
at 1.064 lm with increasing number of thiophene units in the conjugated ligand was found (up to 910 · 10ꢀ30 esu), the static values b0
remain practically unchanged, as shown by the 1.550 lm measurements. Combined with the electrochemical and spectroscopic data (IR,
NMR, UV–vis), this remarkable evolution of b shows that the increase of conjugation length is balanced by a decrease in charge-transfer
efficiency.
ꢀ 2007 Published by Elsevier B.V.
Keywords: Iron complexes; Thiophene nitrile ligands; Cyclic voltammetry; Nonlinear optics; Quadratic hyperpolarizabilities; Molecular first hyperpol-
arizabilities
1. Introduction
nonlinear optics (NLO) [1–6]. For second-order nonlinear
optics, strongly asymmetric systems are needed, which led
to the development of typical push–pull systems in which
the metal centre, bound to a highly polarizable conjugated
backbone, acts as an electron releasing or withdrawing
group. The strong charge-transfer (CT) transitions occur-
ring in organometallic compounds are expected to lead to
high molecular first hyperpolarizabilities b. In addition,
the position of the CT band, usually at visible wavelengths,
can be tuned by variation of the ligands and/or the metal
Organometallic compounds have given rise to a great
deal of interest owing to their application in the field of
*
Corresponding authors. Address: Faculdade de Cieˆncias da Univer-
sidade de Lisboa, Ed.C8, Campo Grande, 1749-016 Lisboa, Portugal
(M.H. Garcia).
E-mail addresses: lena.garcia@fc.ul.pt (M.H. Garcia), wim.wenseleers
@ua.ac.be (W. Wenseleers).
0022-328X/$ - see front matter ꢀ 2007 Published by Elsevier B.V.
doi:10.1016/j.jorganchem.2007.03.023

3028
M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
itself, to optimise the hyperpolarizability through (near)
resonant enhancement. After the initial interesting results
achieved with ferrocene systems [7,8], a further improve-
ment of the NLO response was obtained with half-sand-
wich complexes in which the metal centre is coplanar
with the p-conjugated backbone. Concerning this feature,
systematic studies were made on g5-monocyclopentadie-
nylmetal complexes presenting benzene-based conjugated
ligands bound to the metal centre through nitrile or acety-
lide linkages [9–13]. Ruthenium and iron organometallic
moieties were found to be much more efficient donor
groups for second-order NLO purposes than the tradi-
tional organic donor groups (NMe2, NH2, etc.), leading
to higher b values.
[FeCp(P_P)(NC{SC4H2}nNO2)][PF6] (P_P = dppe, (+)-
diop; n = 1–3). The chiral (+)-diop coligand is used to
favour the formation of a non-centrosymmetrical crystal
structure, required for macroscopic second-order NLO.
Quadratic hyperpolarizabilities (b) for the complexes with
dppe have been determined by hyper-Rayleigh scattering
(HRS) measurements at two different wavelengths of
1.064 and 1.550 lm, to reveal the effect of (near) resonant
enhancement. The least resonant b values (i.e. at 1.550
lm) can be used to obtain more dependable static values
b0, enabling the formulation of structure–NLO property
relationships. Spectroscopic data are used in order to eval-
uate the effect of the type of phosphine coligand and the
length of the conjugated ligand on the second-order NLO
response of the complexes, which is compared to the obser-
vations for the p-benzonitrile analogues. Cyclic voltamme-
try is also used as a tool for a better understanding of the
electronic factors responsible for the NLO behaviour of the
compounds studied in this work.
It is well known that the quadratic hyperpolarizability
of purely organic push–pull molecules increases strongly
with the length of the conjugated chain [14,15]. A previous
attempt to exploit this effect in nitrile metal complexes, by
replacing a phenyl by a biphenyl unit, was unsuccessful due
to the significant torsion angle in the latter ligand, breaking
the conjugation [9]. The much higher b value obtained
when the conjugation length is increased without affecting
planarity by insertion of a vinylene unit, further supported
this assumption. The effect of the phenylene-phenylene
dihedral angle on b has been examined theoretically for
comparable Ru complexes [16]. For oligo-thiophene
derived ligands, however, an improved planarity can be
expected from the limited sterical hindrance as compared
to oligo-phenyls. Although an ab initio calculation by Cio-
falo and La Manna [17] suggest that even for terthiophene
the gas-phase structure is not planar, they found a very low
energy barrier for conformational inter-conversion and as
a result the geometry is very sensitive to the chemical envi-
ronment. This is further illustrated by a semi-empirical cal-
culation by Porter et al. [18], which indicates that the
structure of bithiophene becomes planar in aqueous solu-
tion. Therefore, it seems fair to assume a negligible effect
of non-planarity in the present thiophene-based complexes,
especially because the oligo-thiophene is incorporated in a
push–pull system which should favour a more planar struc-
ture. Thiophene chains are also known to provide a higher
level of electronic coupling than benzenoid-based struc-
tures because of their more effective conjugation thus yield-
ing a good basis for high second-order NLO responses.
Therefore, we now study the conjugation-length depen-
dence of the first hyperpolarizability in a series of iron
cyclopentadienyl compounds incorporating thiophene
derived ligands. Although some reports concerning the
NLO properties of thiophene-based organometallic com-
plexes have been published, they are mainly about ferroce-
nyl and tricarbonyl chromium arene derivatives [19–25] in
which the metal is unfavourably placed outside of the con-
jugation plane [7], in contrast to the present complexes.
In this work, we report the synthesis, characterization
and electrochemical studies of a systematic series of
g5-monocyclopentadienyliron complexes with nitro-
substituted thiophene nitrile ligands of general formula
2. Results and discussion
The thiophene chromophores were synthesized accord-
ing to general procedures. 5-Nitrothiophene-2-carbonitrile
(L1) was prepared as described in the literature [26].
50-Nitro-2,20-bithiophene-5-carbonitrile (L2) and 500-nitro-
2,20:50,200-terthiophene-5-carbonitrile (L3) were prepared
by nitration of the previously synthesized 2,20-bithioph-
ene-5-carbonitrile and 2,20:50,200-terthiophene-5-carbonit-
rile, respectively, with a clay-supported cupric nitrate (claycop)
in Menke conditions (see Section 4).
The complexes (Fig. 1) were prepared by iodide abstraction
of [FeCp(P_P)I] (P_P = dppe or (+)-diop, dppe = 1,2-
bis(diphenylphosphino)ethane, diop = 2,3-O-isopropylid-
ene-2,3-dihydroxy-1,4-bis(diphenylphosphino)butane) with
TlPF6 in the presence of an excess of the appropriate thio-
phene derivative in dichloromethane at room temperature.
After workup, reddish microcrystalline products of [FeCp
(P_P)(NC{SC4H2}nNO2)][PF6] (P_P = dppe, n = 1 (1a),
n = 2 (2a), n = 3 (3a); P_P=(+)-diop, n = 1 (1b), n = 2
(2b), n = 3 (3b)) were obtained with yields in the range of
28–56%. The compounds are quite stable towards oxida-
tion in air and to moisture both in the solid state and in
solution. Formulation of the new compounds was sup-
1
ported by analytical data, IR and H, 13C and 31P NMR
spectra. The molar conductivities of ca. 10ꢀ3 M solutions
Fig. 1. Structural formulas of the complexes studied in this work.

M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
3029
of the complexes in nitromethane, in the range of
69–78 Xꢀ1 cm2 molꢀ1, are consistent with values reported
for 1:1 type electrolytes [27].
Typical IR bands confirm the presence of the cyclopen-
tadienyl coligand (ca. 3100–3040 cmꢀ1), the PF6ꢀ anion
(840 and 550 cmꢀ1) and the coordinated nitrile (2180–
2210 cmꢀ1) in all the complexes. No significant change
was observed on m(NC) upon coordination of L3 but in
the case of L1 and L2, depending on the coordinated phos-
phine coligand, negative shifts were found in the range of
5–45 cmꢀ1. These negative shifts, which indicate enhanced
p-backdonation from the metal d orbitals to the p* orbital
of the NC group, are higher than those found in the related
p-benzonitrile derivatives [10,28]. As the number of thio-
phene rings in the ligands increases, the shift of m(NC) upon
coordination was found to become less negative. So appar-
ently, the CT through the long conjugated ligand is ham-
pered, which has an unfavourable effect on the molecular
first hyperpolarizability b (see below).
1H and 13C NMR resonances for the cyclopentadienyl
ring are in the range usually observed for monocationic iro-
n(II) complexes and seem to be affected by the phosphine
coligand but are relatively insensitive to the nature of the
aromatic nitrile. For the nitrile ligands, the effect on
NMR resonances upon coordination is mainly found in
the deshielding observed at the carbon of the NC func-
tional group, which magnitude depends on the phosphine
coligand and the nature of the coordinated nitrile. The
observed deshielding is lower for the complexes with dppe
than for the ones with (+)-diop coligand and for the nitrile
ligands with less extended aromatic system. This behaviour
agrees with the proposed p-backdonation interaction
inferred from the IR spectra which indicates a decreased
p-backdonation with increasing the number of thiophene
rings in the ligands. The deshielding observed for the com-
plexes containing L1 and L2 derivatives are lower than
those found in the related p-benzonitrile complexes
[10,28], which can be a result of the enhanced p-backdona-
tion in the present compounds with thiophene ligands.
31P NMR data of the complexes showed the expected
deshielding of the phosphines upon coordination. The
dppe ligand was characterized by a singlet and (+)-diop
by two doublets due to its inequivalent phosphorus atoms.
The deshielding effect upon coordination was enhanced
with the better donor character of the phosphine, showing
an easier release of electron density towards the metal cen-
tre. No significant changes on the phosphorus electron den-
sity were observed by the presence of different coordinated
thiophenes.
Fig. 2. Absorption spectra of [FeCp(dppe)(NC{SC4H2}nNO2)][PF6]
(n = 1: 1a, n = 2: 2a, n = 3: 3a) recorded in CH2Cl2 (5.0 · 10ꢀ5 M).
(Top) Absorption spectra for the free thiophene ligands NC{SC4H2}nNO2
(n = 1: L1, n = 2: L2, n = 3: L3) and FeCp(dppe)I in the same
experimental conditions.
[FeCp(P_P)]+. For all the complexes 1a–3a the electronic
transition of the metallic moiety (k ꢁ 260 nm) and the
uncoordinated thiophene ligands (k ꢁ 290–440 nm) can
be clearly recognized. Furthermore, compounds 1a and
2a with one and two thiophene rings exhibit a separate
metal-to-ligand charge transfer (MLCT) band(s) in the vis-
ible region (k ꢁ 400–650 nm). This is, however, not the case
for the compound with three thiophenes 3a, where the
MLCT band is very probably overlapped with the intense
absorption band attributed to the p–p* transition occurring
in the coordinated thiophene ligand L3. The energies of the
MLCT bands are lower than those found in the related
p-benzonitrile complexes, which can lead to an increase
of the resonance enhancement of b, but also of the static
values b0. Clearly, the chain-lengthening leads to a batho-
chromic shift of the transitions of the conjugated ligand
and a hypsochromic shift of the MLCT band. The latter
effect was also observed for the related p-benzonitrile com-
pounds [10,28] and was explained on the basis of a break-
ing of the conjugation due to the presence of a significant
torsion angle between the aromatic rings. For the present
thiophene derivatives, however, the conjugated chain is
expected to be more planar because of the limited sterical
hindrance between the successive thiophene rings. So here
the hypsochromic shift indicates a lowering of the CT effi-
ciency with increasing length of the conjugated chain,
which has a detrimental effect on b.
The optical absorption spectra of all complexes were
recorded using 5.0 · 10ꢀ5 M solutions in dichloromethane.
The spectra for the dppe complexes 1a–3a typify the behav-
iour of the compounds studied in this work (Fig. 2) and the
optical data are summarized in Table 1. The spectra of all
compounds are characterized by the presence of intense
absorption bands which can be assigned based on the
spectra of the free ligands and the organometallic fragment

3030
M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
Table 1
Optical data for complexes [FeCp(P_P)(NC{SC4H2}nNO2)][PF6] in dichloromethane solution (ca. 5.0 · 10ꢀ5 mol dmꢀ3
)
Compound[FeCp(P_P)L][PF6]
k (nm)
e · 104 (Mꢀ1 cmꢀ1
)
P_P = dppe
263 (sh)
290
475
S
1.71
0.59
(1a)
NO2
L=
NC
S
570 (sh)
256 (sh)
358
423
S
1.70
1.24
NC
NO2
(2a)
530 (sh)
254 (sh)
437
S
S
S
2.40
(3a)
NC
NO2
P_P=(+)-diop
272 (sh)
292
500
S
2.48
0.75
L=
NC
S
NO2
(1b)
258 (sh)
360
450
S
1.49
0.97
NC
NO2
(2b)
255 (sh)
431
S
S
S
2.16
(3b)
NO2
NC
In order to study the solvatochromic behaviour of all the
compounds, absorption spectra were measured in different
solvents of increasing polarity (chloroform, methanol, ace-
tone, nitromethane and dimethylformamide). Although the
data must be carefully analyzed due to the broad and com-
posite structure of the bands, which lead to some uncer-
tainty in the attribution of kmax values, the studies show
a slight bathochromic shift on the MLCT bands upon
increasing solvent polarity. This positive solvatochromic
behaviour exhibited by these compounds is characteristic
of electronic transitions with an increase of the dipole
moment upon (photo-)excitation. The positive shift seems
to increase from compound 1 to compound 2 but much less
solvatochromic effect could be detected for the complexes
with three thiophene rings (3). In all studied solvents, with
dipolar moments in the range 1.1–3.8 D, a single structure-
less band was found for complexes 3a and 3b without evi-
dence of any shoulder. The reason for this might be the
overlapping of the MLCT and the ligand p–p* bands. In
fact, the MLCT band seems to shift to higher energies
and the intra-ligand band to lower energies with increasing
conjugation length and both are overlapped completely for
the compounds with three thiophene rings.
In order to obtain an insight on the electron richness of
the organometallic fragment, the electrochemical behav-
iour of all the compounds was studied by cyclic voltamme-
try in dichloromethane and acetonitrile between the limits
imposed by the solvents. As an example, the cyclic voltam-
metry response of 2a in dichloromethane is shown in Fig. 3,
and the most relevant data for the redox changes exhibited
by all the complexes in dichloromethane are summarized in
Table 2.
The electrochemical behaviour of the compounds in
dichloromethane is characterized by the presence of one
quasi-reversible redox wave attributed to Fe(II)/Fe(III)
oxidation, in the range 0.78–1.06 V, and two quasi-revers-
ible or irreversible waves attributed to reduction processes
occurring on the coordinated nitrile ligands, in the range
ꢀ0.58 to ꢀ0.91 V for redox process A and ꢀ1.19 to
ꢀ1.27 V for redox process B. Additionally, redox waves
with low intensity were found and attributed to the decom-
position products after both reductive processes. A first
anodic shoulder C was found at Epa ꢂ ꢀ0.8 V and origi-
nated from the first reduction on the coordinated ligand.
A second wave D at Ep/2 ꢂ ꢀ1.3 V and a third cathodic
wave E at Epc ꢂ ꢀ0.9 V appeared after the second reduc-

M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
3031
Table 2
Electrochemical data for the complexes [FeCp(P_P)(NC{SC4H2}nNO2)][PF6] and the corresponding thiophene ligands NC{SC4H2}nNO2 in CH2Cl2
solution at 20 ꢁC
Compound [FeCp(P_P)L][PF6]
P_P = dppe
Epc (V)
Epa (V)
Ep/2 (V)
Epa ꢀ Epc (mV)
Ic/Ia
0.84
ꢀ0.66
ꢀ1.29
0.91
ꢀ0.58
ꢀ1.05
0.88
ꢀ0.62
–
70
80
–
1.0
0.9
–
S
L=
NC
S
NO2
(1a)
0.77
ꢀ0.86
ꢀ1.28
0.85
ꢀ0.78
ꢀ1.20
0.81
ꢀ0.82
ꢀ1.24
80
80
80
1.0
1.0
0.8
S
NC
NO2
(2a)
0.74
ꢀ0.94
ꢀ1.31
0.81
ꢀ0.87
ꢀ1.23
0.78
ꢀ0.91
ꢀ1.27
70
70
80
1.0
0.9
0.6
S
S
S
NC
NO2
(3a)
P_P=(+)-diop
1.02
ꢀ0.61
ꢀ1.24
1.09
ꢀ0.54
ꢀ1.02
1.06
ꢀ0.58
–
70
70
–
0.9
0.9
–
S
L=
NC
S
NO2
(1b)
0.94
ꢀ0.82
ꢀ1.23
1.01
ꢀ0.74
ꢀ1.14
0.98
ꢀ0.78
ꢀ1.19
70
80
90
0.9
0.9
0.9
S
NC
NO2
(2b)
0.90
ꢀ0.94
ꢀ1.23
0.98
ꢀ0.86
ꢀ1.15
0.94
ꢀ0.90
ꢀ1.19
80
80
80
0.9
0.6
0.8
S
S
S
NC
NO2
(3b)
Ligand
ꢀ0.74
ꢀ1.59
ꢀ0.66
ꢀ1.23
ꢀ0.70
80
–
1.0
–
S
–
(L1)
NC
NO2
ꢀ0.88
ꢀ1.42
ꢀ0.80
ꢀ1.27
ꢀ0.84
80
–
0.9
–
S
S
–
NC
NO2
(L2)
ꢀ0.93
ꢀ1.43
ꢀ0.86
ꢀ1.30
ꢀ0.90
70
–
0.9
–
S
S
S
–
(L3)
NO2
NC
tive process. Moreover, a low anodic wave F at ꢀ0.15 V
arises only if the potential reaches the second reductive pro-
cess. This behaviour is illustrated by Fig. 3, which shows
that the waves E and F are not present if the potential is
reversed immediately after the first reductive process.
The Fe(II)/Fe(III) potential is affected both by the phos-
phine coligand and the nature of the coordinated nitrile
ligand. Replacing the dppe coligand by (+)-diop results
in an increase of the redox potential of the Fe(II)/Fe(III)
couple, which agrees with the relative donating ability of
the two phosphines to the metal centre. For the same phos-
phine, the Fe(II)/Fe(III) couple potential decreases with
the chain-lengthening of the thiophene ligand, which can
be due to a less effective release of electronic density from
the metal centre to the nitro-acceptor group with the
extension of the aromatic system. This indicates that
adding more thiophene units in the p-conjugated chain
will not necessarily lead to an increase of the molecular b
value, which is indeed confirmed by the NLO experiments
(see below). These results confirm the evidences of the

3032
M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
ing ability of the two studied phosphines and agree with the
spectroscopic data discussed above. The difference between
the two reduction potentials strongly depends on the length
of the conjugated ligand. This difference is a measure of the
interaction between the two negative charges. For the free
thiophene ligands, this difference becomes smaller as the
chain length increases. As was already recognized for oli-
gothiophenes reported in the literature, this trend can be
explained by a stronger coulombic interaction between
negative charges developed by shorter conjugated chains
[29]. Upon coordination, this trend remains unchanged
but the difference between the two reduction potentials
becomes smaller. This observation suggests that the orga-
nometallic fragment diminishes the coulombic interaction
between the two negative charges, as expected considering
the cationic character of the moieties.
The electrochemical studies in acetonitrile, at room tem-
perature, showed a significant difference concerning the
oxidative electrochemistry of the studied compounds. In
fact, the Fe(II)/Fe(III) redox couple of the complexes with
dppe (1a–3a) have no cathodic counterpart and a second
cathodic wave arises at 0.62 V which value is independent
of the coordinated thiophene. However, at ꢀ20 ꢁC, the
cathodic wave of the Fe(II)/Fe(III) redox couple arises
and the intensity of the cathodic wave at 0.62 V diminishes.
In addition, a small anodic wave at ꢂ0.70 V emerges. As
an example, the oxidative electrochemistry response of 1a
in acetonitrile is shown in Fig. 4, and the most relevant
data for the redox changes exhibited by all the complexes
in this solvent are summarized in Table 3. These results
suggest that the 17-electron species [FeCp(dppe)(NC
{SC4H2}n NO2)]2+, formed at the electrode surface at oxi-
dation potential, undergo fast decomposition at room tem-
perature (in an ECC process). The presence of a cathodic
wave which value is independent of the coordinated thi-
ophenes suggests that it originates from the decomposition
of [FeCp(dppe)(NC{SC4H2}nNO2)]2+ due to the decoordi-
nation of the thiophene ligand and its substitution by ace-
tonitrile solvent molecules, leading to the [FeCp(dppe)
(NCCH3)]2+ species before the reductive counterpart pro-
cess. These species are stable enough to undergo a reduc-
tion process with Epc = 0.62 V leading to a 18-electron
[FeCp(dppe)(NCCH3)]+ complex. In order to confirm this
Fig. 3. Cyclic voltammograms of 2a in CH2Cl2 containing 0.1 M
n-Bu4NPF6 (sweep rate = 200 mV/s). (Top) Oxidative and the first
reductive processes.
spectroscopic data discussed above, namely the relative
magnitude of the metal–ligand backdonation, which was
found to be smallest for the compounds with three thio-
phene rings. The Fe(II)/Fe(III) redox couple potentials of
this family of compounds are higher than those observed
for the related p-benzonitrile compounds [10,28] which is
in good agreement with the lower electronic density at
the metal centre in thiophene derivatives.
As mentioned above, the two main redox processes at
negative potentials are attributed to reductions occurring
at the coordinated thiophene ligands. The electrochemical
data for the free thiophene ligands in dichloromethane
are summarized in Table 2. The first wave is reversible or
quasi-reversible and corresponds to the formation of the
anion radical. Owing to the reversibility of the first wave,
the second reduction process can be attributed to a further
reduction of the anion radical to the corresponding dian-
ion. The irreversibility of this second redox process indi-
cates a poor stability of the dianion. Upon coordination,
the reversibility of the first reduction wave remains almost
unchanged when the potential is reversed at ꢀ1.1 V. Also,
it was found that the second reduction process at the thio-
phene ligands becomes less irreversible upon coordination.
This means that the iron organometallic fragment stabilizes
the dianionic thiophene ligand species. The electrochemical
data show also that the coordination of the thiophene
ligands lowers the corresponding reduction potentials
according to the overall electron-withdrawing effect of the
organometallic fragment. This effect is more pronounced
on the second reduction potentials and also depends on
the coordinated phosphine and the chain length of the thi-
ophene ligand. The higher reduction potentials found for
complexes with dppe are consistent with the relative donat-
Fig. 4. Cyclic voltammogram of 1a in NCCH3 containing 0.1 M
n-Bu4NPF6 (sweep rate=200 mV/s). (------) 20 ꢁC; (——) ꢀ20 ꢁC.

M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
3033
Table 3
Electrochemical data for complexes [FeCp(P_P)(NC{SC4H2}nNO2)][PF6] and for the thiophene ligands NC{SC4H2}nNO2 in NCCH3 solution at ꢀ20 ꢁC
Compound [FeCp(P_P)L][PF6]a
Epc (V)
Epa (V)
Ep/2 (V)
Epa ꢀ Epc (mV)
Ic/Ia
P_P = dppe
0.76
ꢀ0.64
ꢀ1.22
0.83
ꢀ0.56
ꢀ1.10
0.80
ꢀ0.60
ꢀ1.16
70
80
120
0.8
0.9
1.0
S
L=
NC
S
NO2
(1a)
0.70
ꢀ0.77
ꢀ1.18
0.76
ꢀ0.71
ꢀ1.12
0.73
ꢀ0.74
ꢀ1.15
60
60
60
0.8
1.0
1.0
S
NC
NO2
(2a)
0.67
ꢀ0.84
ꢀ1.22
0.75
ꢀ0.78
ꢀ1.16
0.71
ꢀ0.81
ꢀ1.19
80
60
60
0.8
1.0
1.0
S
S
S
NC
NO2
(3a)
NCCH3 (4a)
0.63
0.71
0.67
80
0.9
P_P=(+)-diop
0.93
ꢀ0.61
ꢀ1.31
1.01
ꢀ0.54
ꢀ1.01
0.97
ꢀ0.58
–
80
70
–
0.9
0.9
–
S
L=
NC
S
NO2
(1b)
0.84
ꢀ0.80
ꢀ1.16
0.92
ꢀ0.72
ꢀ1.07
0.88
ꢀ0.76
ꢀ1.12
80
80
90
0.9
0.9
0.9
S
NC
NO2
(2b)
0.82
ꢀ0.87
ꢀ1.25
0.90
ꢀ0.79
ꢀ1.17
0.86
ꢀ0.83
ꢀ1.21
80
80
80
0.8
0.9
0.9
S
S
S
S
NC
NO2
(3b)
NCCH3 (4b)
0.80
0.87
0.84
70
0.9
Ligand
ꢀ0.70
ꢀ1.77
ꢀ0.61
ꢀ1.40
ꢀ0.66
90
–
1.0
–
–
(L1)
NC
NO2
ꢀ0.82
ꢀ1.38
ꢀ0.74
ꢀ1.26
ꢀ0.78
80
–
1.0
S
S
–
NC
NO2
(L2)
ꢀ0.88
ꢀ1.36
ꢀ0.81
ꢀ1.27
ꢀ0.84
ꢀ1.32
70
90
1.0
1.0
S
S
S
NC
NO2 (L3)
a
For all the [FeCp(P_P)(NC{SC4H2}nNO2)]+ complexes an additional cathodic wave corresponding to reduction of [FeCp(P_P)(NCCH3)]2+ species
was observed (see text for details).
hypothesis, the complex [FeCp(dppe)(NCCH3)][PF6] was
synthesized and its electrochemistry studied in acetonitrile.
The results obtained for the Fe(II)/Fe(III) redox couple are
consistent with the formulated hypothesis (see Table 3).
The lability of thiophene ligands was also confirmed by
1H NMR spectroscopic studies. After the electrochemical
experiments of the complex 1a, both the solvent and the
supporting electrolyte were removed and the 1H NMR
spectrum of the sample showed the presence of resonances
attributed to both the [FeCp(dppe)(NCCH3)]+ complex
and the corresponding free thiophene ligand. The substitu-
tion of the thiophene ligands by the acetonitrile solvent in
electrochemical experiments during the oxidative process
was also observed for the (+)-diop complexes (1b–3b) but
in a smaller extent. In fact, a small cathodic wave at
ꢂ0.79 V was found, which is consistent with the data found

3034
M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
for the complex [FeCp((+)-diop)(NCCH3)][PF6] in a sepa-
rate experiment. This behaviour might be explained by a
less favourable kinetic of substitution due to steric crowd-
ing around the metal centre as a result of the presence of a
bulkier phosphine. The same general trends on oxidation
and reduction potentials by varying the phosphine coli-
gand and the nature of nitrile ligands were obtained in ace-
tonitrile. The electrochemical data for the free thiophene
ligands in acetonitrile are summarized in Table 3.
In view of the application in second-order nonlinear
optics of the compounds studied in this work, it is of inter-
est to consider the HOMO–LUMO gap, as expressed by
the difference between the first oxidation and reduction
ative magnitude of the HOMO energy. The results dis-
cussed above show that the HOMO, as expected, is
destabilized by the better electron donor phosphine dppe
and, to a lower extent, by the chain lengthening of the thi-
ophene ligand. On the other hand, the LUMO, the energy
of which can be related to the first reduction potential, is
destabilized as the chain-length of the thiophene ligand
increases and is only slightly affected by the nature of the
phosphine coligand. The difference between the Fe(II)/
(III) couple and the first reduction occurring on the thio-
phene ligand, calculated as an estimation of HOMO–
LUMO gap, is shown in Table 4.
The data show that the HOMO–LUMO gaps depend on
phosphine coligand, chain lengthening of the thiophene
ligand and the solvent used in the electrochemical experi-
ments. As expected, the lower HOMO–LUMO gaps
observed for dppe complexes are mainly due to the higher
energies of the HOMOs since the LUMOs, which are
mainly located on thiophene ligands, are only slightly
affected by the nature of the phosphine coligand. In fact,
as discussed above, replacing the (+)-diop coligand by
potentials. Extended Huckel MO calculations (EHMO)
¨
carried out on similar complexes containing p-benzonitrile
derivatives show that the HOMO of the complexes is essen-
tially localized in the metal fragment while the LUMO is
essentially a pure nitrile ligand orbital [10]. It is reasonable
to assume that the character of the HOMO and LUMO
remains unchanged for the complexes studied in this work.
Hence, the Fe(II)/Fe(III) potential can be related to the rel-
Table 4
Estimation of HOMO–LUMO gap based on electrochemical data for the complexes [FeCp(P_P)(NC{SC4H2}nNO2)][PF6]
Compound [FeCp(P_P)L][PF6]
Eox ꢀ Eared (V)
CH2Cl2
1.50
NCCH3
1.40
P_P = dppe
S
L=
NC
S
NO2
(1a)
1.63
1.69
1.47
1.52
S
NC
NO2
(2a)
S
S
S
NC
NO2
(3a)
P_P=(+)-diop
1.64
1.76
1.84
1.55
1.64
1.69
S
L=
NC
S
NO2
(1b)
S
NC
NO2
(2b)
S
S
S
NC
NO2
(3b)
a
Eox ꢀ Ered is the difference between the potential of the Fe(II)/Fe(III) redox couple and the first reduction potential occurring on the thiophene ligand.

M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
3035
the better donor dppe phosphine results in a decrease of the
redox potential of the Fe(II)/Fe(III) couple, and hence in
an increase of the HOMO energies. The chain lengthening
of the nitrile ligand leads to an increase in the HOMO–
LUMO gap, mainly as a result of the destabilizing effect
on the LUMO, as previously discussed. These results are
consistent with the electronic spectra discussed above,
where the relative energy of the MLCT bands within the
series of compounds 1a–3a and 1b–3b seem to follow the
HOMO–LUMO gap trend. In fact, the hypsochromic shift
observed for MLCT bands in the UV–Vis spectra with
lengthening of the thiophene ligand is in good agreement
with this observation. For all complexes, the Fe(II)/Fe(III)
potential is lower in acetonitrile which reflects a relative
destabilization of the HOMO when compared to that
observed in dichloromethane. The magnitude of this desta-
bilization slightly depends on the phosphine coligand and
the extension of the aromatic system. On the other hand,
acetonitrile stabilizes the LUMO in all the complexes but
the magnitude of this stabilization strongly depends on
the phosphine coligand and the extension of the aromatic
system. The relative stabilization of the LUMO increases
with the chain length of the thiophene ligands. For the
compounds with dppe coligand, the relative stabilization
corresponds to a difference in the first reduction potential
of 20 mV for 1a, 80 mV for 2a and 100 mV for 3a. For
the complexes with (+)-diop the relative stabilization of
the LUMO is lower: for 1b no net stabilization was
observed whereas for 2b and 3b the difference on the first
reduction potential was 20 mV and 70 mV, respectively.
As a result, the overall effect on the HOMO–LUMO gaps
gives lower values in acetonitrile (see Table 4). The differ-
ence between the HOMO–LUMO gaps in both solvents
and the relative contribution of HOMO (and consequently
of LUMO) to the net stabilization observed in acetonitrile
is shown in Fig. 5. For the compounds with one thiophene
ring (1a and 1b) the lower HOMO–LUMO gaps obtained
in acetonitrile are mainly due to the relative destabilization
of the HOMO in this solvent (Table 4). For compound 2a,
the net stabilization in acetonitrile has equal contributions
of the HOMO destabilization and the LUMO stabilization,
whereas for 2b the main contribution is due to the HOMO
destabilization. Finally, for 3a the net stabilization in ace-
tonitrile is mainly due to the LUMO stabilization whereas
for 3b it has almost equal contribution of HOMO destabi-
lization and LUMO stabilization. This means that as the
chain length of the thiophene ligands increases, the stabil-
ization of the LUMO becomes more important for the
decrease of the HOMO–LUMO gap in a more polar sol-
vent. Furthermore, this effect is more pronounced with a
better donor phosphine (dppe). The lower HOMO–LUMO
gaps obtained in acetonitrile compared to dichloromethane
agrees with the solvatochromic behaviour of the MLCT
bands of the compounds with one and two thiophene units
(see above), where a bathochromic shift upon increasing
the polarity of the solvent was found. The solvatochromic
behaviour of the MLCT band for 3a and 3b could not be
Fig. 5. Difference between HOMO–LUMO gaps in dichloromethane
and acetonitrile (taken from Table 4) and the relative contribution of
HOMO (black) and LUMO (white) on the net lower gaps observed in
acetonitrile
(HOMO–LUMO)acetonitrile).
(DGap(HOMO–LUMO) = (HOMO–LUMO)dichloromethane
ꢀ
clearly identified due to an overlapping of this transition
with the one attributed to the coordinated thiophene
ligand. However, the electrochemical results showed an
enhanced stabilization of the LUMO by acetonitrile with
the extension of the aromatic thiophene system, which indi-
cates a large dipole moment variation (Dl) upon (photo-)
excitation for the compounds with three thiophene rings,
and in particular for 3a.
The second-order NLO properties of the complexes with
dppe coligand (1a–3a) were studied by HRS measurements
at two different fundamental wavelengths of 1.064 and
1.550 lm. The obtained molecular first hyperpolarizabili-
ties b are summarized in Table 5, in which also the most
relevant related iron benzonitrile complexes are introduced
for comparison.
Because compounds 1a–3a exhibit an optical transition
in the visible range, the b values at 1.064 lm are strongly
affected by two-photon resonant enhancement. The well-
known two-level model (TLM) of Oudar and Chemla
[30], which is very often used in the literature to derive sta-
tic values, b0, is expected to become invalid close to
resonance as it ignores any kind of line-broadening mech-
anisms. This model, in which it is assumed that the lowest
energy CT transition yields the dominant contribution to b,
diverges whenever the (laser or) second-harmonic wave-
length approaches the transition wavelength. This can lead
to a strong overestimation of the resonance effect and
hence a significant underestimation of the static first hyper-
polarizability. Therefore, additional measurements were
performed at a longer fundamental wavelength of 1.550
lm, outside of the long wavelength absorption band of
each of the compounds, where reasonable results can be
expected from the TLM analysis. In this way, the effects
of resonant enhancement and increasing conjugation
length are disentangled, enabling the formulation of reli-
able relations between molecular structure and NLO prop-
erties. With this in mind, the static values b0 (see Table 5),
calculated using the TLM based on the long wavelength b

3036
M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
Table 5
Experimental first hyperpolarizabilities (b) and chemical structures of the compounds referred in the text
No.
Compound
bzzz at 1.064 lm
bzzz at 1.550 lm
keg (nm)a
bbzzz,0
1a
2a
3a
5a
6a
[FeCp(dppe)(NC{SC4H2}NO2)]+
[FeCp(dppe)(NC{SC4H2}2NO2)]+
[FeCp(dppe)(NC{SC4H2}3NO2)]+
[FeCp(dppe)(NC{C6H4}NO2)]+
[FeCp(dppe)(NC{C6H4}2NO2)]+
455
710
910
375c
240c
255
177
185
579
535
460
97
82
109
All measurements are performed in CHCl3 solution. The b values are expressed in 10ꢀ30 esu.
a
keg is the position of the longest wavelength absorption band obtained from a fit with multiple Gaussian bands.
b0 based on bzzz at 1.550 lm, using the two-level model with keg.
Ref. [9].
b
c
values, will be used in the upcoming discussion of the
the spectroscopic and electrochemical measurements,
which all indicated a decrease of p-backdonation for longer
conjugated ligands (see above). These observations demon-
strate very clearly that resonantly enhanced values at
1.064 lm can be extremely deceiving, as they seem to indi-
cate that compounds 1a–3a follow the traditional trend of
increasing b with longer conjugated chain. It is observed
that b increases much less upon going from 1.550 lm to
1.064 lm for 1a than for 2a and 3a. This behaviour cannot
be explained by the difference in resonance conditions (val-
ues of keg relative to 532 nm: 1a is closer to resonance than
3a), within the TLM, not even if homogeneous damping
[31] is taken into account. Clearly, near to resonance the
dispersion of the b values of compounds 1a–3a is not well
described by this simple model. This discrepancy cannot be
accounted for by the uncertainty in keg, since the longest
wavelength component in the spectrum was chosen, which
yields the most conservative estimate of the resonance
effect. Indeed, note that if a spectral component at shorter
keg value was selected, b values at 1.072 lm would be even
more resonant for 1a and less resonant for 2a and 3a. Pos-
sibly, the discrepancy is caused by a red shift between the b
dispersion and the linear absorption maximum, as reported
for an organic chromophore [32] and also observed in
wavelength dependent HRS for an iron nitrile complex
[33]. Because the behaviour of b in the resonance regime
is not well understood, measurements outside of this region
are required in order to formulate dependable structure–
NLO properties relationships.
To analyze the different factors contributing to b we can
make use of the above-mentioned TLM [30], according to
which b0 depends on the frequency and oscillator strength
of the electronic transition (usually approximated by (the
position of) the maximum of the lowest energy absorption
band) and the difference in dipole moment between the
ground and the CT-excited state Dl. Indeed, a decrease
of the first hyperpolarizability with increasing HOMO–
LUMO gaps (higher CT-transition frequency), was
observed for g5-monocyclopentadienylmetal complexes
with p-benzonitrile ligands [10,34]. A longer conjugated
chain favours larger oscillator strength (see Fig. 2) and
Dl, if the same amount of charge transfer can still be
achieved. Information on Dl can be obtained from the sol-
vatochromic studies, where a bathochromic shift of the
experimental results. For the calculation of the b0 values,
the position of the longest wavelength absorption band
obtained from a fit with multiple Gaussian bands is used
as keg. Of course, because of the composite structure of
the absorption bands, the choice of keg is somewhat arbi-
trary within the range of the low-energy electronic transi-
tions. This determination influences the exact b0 values,
but the general trends and conclusions discussed below
are quite insensitive to the precise value of keg.
From the experimental HRS data in Table 5, it appears
that the thiophene derivatives studied in this work gener-
ally yield higher resonant b values than the related p-substi-
tuted benzonitrile based complexes (compare 1a to 5a and
2a to 6a). This is attributed to a higher degree of p-backdo-
nation from the Fe(II) organometallic fragment to the thi-
ophene ligands leading to a more effective extension of
conjugation from the metal centre to the electron acceptor
NO2 group, as evidenced from the spectroscopic (IR,
NMR) and electrochemical data discussed above. The cor-
relation between the magnitude of b and the p-backdona-
tion process was previously addressed for systematic
variation of the metal ion in the related benzonitrile com-
plexes [9]. Table 5 also shows a steady increase in resonant
b values at 1.064 lm with increasing number of thiophene
units, leading to the large value of 910 · 10ꢀ30 esu for 3a.
The less resonant b values at 1.550 lm, however, are not
following this trend, and neither do their derived static val-
ues which are practically equal for all three thiophenes
under study. So, apparently the present complexes are
not exhibiting the usual evolution of increasing b with con-
jugation length, as reported for purely organic push–pull
molecules [14,15]. As discussed before, the oligo-thiophene
torsion angles are expected to be too small to be a signifi-
cant factor in the observed trend. Hence, the constancy
of the static b values with increasing length of the thio-
phene chain must be attributed to a competition between
the growing conjugation length, which tends to raise b,
and a decrease of the CT efficiency having a b lowering
effect. Apparently, the use of longer conjugated chains is
affecting the coupling between the organometallic donor
moiety and the nitro acceptor group, more than for typical
organic push–pull systems. The fact that the CT is ham-
pered in case of longer thiophene chains is also seen from

M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
3037
lower energy CT band was observed for the compounds 1a
and 2a (see above), indicating an increase of the dipole
moment upon excitation (Dl > 0). However, no clear trend
in Dl with increasing chain length can be concluded from
the solvatochromism, also because no significant solvato-
chromic effect was observed for 3a, due to an overlapping
of the MLCT band with an intra-ligand CT band. Anyway,
the electrochemical studies presented above seem to indi-
cate that a large Dl can be expected for this compound.
Hence, in terms of the TLM, the increase of b with larger
oscillator strength (see Fig. 2) and possibly Dl for longer
thiophene chains, is cancelled by the b lowering effect of
the higher CT-transition frequency [and higher HOMO–
LUMO gap (see Table 4), the lowest energy transition
shifts to higher energies with increasing conjugation length
(see Table 1)], and the lower amount of charge transferred.
So it seems that both oscillator strength and dipole
moment variation upon excitation Dl play an important
role in the second-order nonlinear optical properties of
the compounds studied in this work.
· 10ꢀ30 esu for the compound with three thiophene units.
The static values, however, are not following the usual evo-
lution of increasing b with conjugation length and are prac-
tically independent of the number of thiophene units in the
conjugated ligand. Based on the electrochemical and spec-
troscopic studies it is concluded that this constancy of b
upon chain-lengthening is caused by a compensation of
the favourable effects of increasing conjugation length by
a lowering of the CT efficiency. The present results also
demonstrate that resonantly enhanced b values can be
extremely deceiving in the evaluation of the molecular
NLO-behaviour, as the wavelength-dependence of b within
the resonance region is clearly different from the simple
two-level models. HRS measurements performed both
on- and off-resonance can lead to a better understanding
of the behaviour of b in the resonance region and hence
an improved modelling of the b dispersion.
4. Experimental
The spectroscopic and electrochemical data discussed
above also indicate that comparable, or somewhat lower,
b values can be expected by replacing the dppe coligand
by the chiral (+)-diop phosphine, as was also observed
for the related complexes with p-substituted benzonitrile
ligands [10,34]. Indeed, when comparing to the dppe com-
plexes, the (+)-diop compounds have higher HOMO–
LUMO gaps (Table 4). In addition, the electrochemical
data in dichloromethane and acetonitrile show that the
decreasing HOMO–LUMO gap observed in the more
polar solvent is to a larger extent due to the stabilization
of the LUMO for the complexes with dppe (Fig. 5), which
can lead to higher Dl. In spite of the somewhat lower b val-
ues expected for the (+)-diop complexes, the use of this
chiral coligand is important in order to obtain a non-cen-
trosymmetrical crystal structure, required for applications
in second-order NLO.
4.1. General procedures
All preparations and manipulations for the synthesis of
the complexes described in this work were carried out
under nitrogen or argon atmosphere using standard
Schlenck techniques. Solvents were purified according to
the usual methods [35]. Solid state IR spectra were taken
on a Perkin–Elmer 457 spectrophotometer with KBr pel-
lets; only significant bands are cited in the text. H, 13C
1
and 31P NMR spectra were recorded on a Varian Unity
300 spectrometer at probe temperature. The H and 13C
1
resonances are reported in parts per million (ppm) down-
field from internal Me4Si and the 31P NMR spectra are
reported in ppm downfield from external standard 85%
H3PO4. Coupling constants are reported in Hz. Spectral
assignments follow the numbering scheme shown below:
S
3. Conclusion
S
S
NC
NO2
13
2
5
6
9 10
A new class of compounds for second-order NLO mate-
rials was developed, combining the organometallic donor
fragment FeCp(P_P) (P_P = dppe, (+)-diop) with conju-
gated thiophene derived ligands. Spectroscopic and cyclic
voltammetry data suggest an improved electronic p-cou-
pling between the g5-cyclopentadienyliron fragment and
the p-system of the conjugated thiophene ligands, when
compared to the previously reported p-benzonitrile ana-
logues. Quadratic hyperpolarizabilities of the compounds
with dppe have been determined by HRS measurements
at two different fundamental wavelengths of 1.064 and
1.550 lm, revealing the two-photon resonance effect and
leading to more dependable static values b0. These mea-
surements show that the resonant first hyperpolarizabilities
at 1.064 lm are indeed higher than for the related benzonit-
rile compounds, and they can be scaled up by increasing
the conjugation length, leading to the high b = 910
7
8
3
4
11 12
Electronic spectra were recorded at room temperature
on a Shimadzu UV-1202 spectrometer. Melting points
were obtained using a Reichert Thermovar. The molar
conductivities of 1 mM solutions of the complexes in
nitromethane were recorded with a Schott CGB55 Kon-
duktometer at room temperature. Microanalyses were
performed using a Fisons Instruments EA1108 system.
Data acquisition, integration and handling were per-
formed using a PC with the software package Eager-
200 (Carbo Erba Instruments).
5-Nitro-2-thiophenecarboxaldehyde, 2,20-bithiophene
and 2,20:50,200-terthiophene were purchased from Aldrich.
2,20-Bithiophene-5-carboxaldehyde and 2,20:50,200-terthi-
ophene-5-carboxaldehyde were prepared by Vilsmeier reac-
tion from 2,20-bithiophene and 2,20:50,200-terthiophene,

3038
M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
respectively, as described in the literature [36]. 2,20-Bithi-
ophene-5-carbonitrile and 2,20:50,200-terthiophene-5-carbo-
nitrile were prepared as described below for L1 from
2,20-bithiophene-5-carboxaldehyde and 2,20:50,200-terthi-
ophene-5-carboxaldehyde, respectively, by formation of
the corresponding oximes and dehydration in situ. The
nitrating agent, the clay-supported cupric nitrate (claycop),
was prepared according to previously described [37].
[FeCp(dppe)I] and [FeCp((+)-diop)I] were prepared by
UV irradiation of a solution of [FeCp(CO)2I] in the pres-
ence of the appropriate phosphine in benzene. [FeCp(dp-
pe)NCMe][PF6] (4a) and [FeCp((+)-diop)NCMe][PF6]
(4b) were prepared by iodide abstraction of [FeCp(P_P)I]
(P_P = dppe or (+)-diop) with TlPF6 in acetonitrile at
room temperature.
(CDCl3): d 110.96 (C2), 113.20 (NC), 124.68 (C7), 126.06
(C4), 129.35 (C8), 138.37 (C3), 141.30 (C5), 141.39 (C6),
151.46 (C9); UV–Vis (CH2Cl2): kmax/nm (e/Mꢀ1 cmꢀ1
)
377 (22000).
4.4. Synthesis of 500-nitro-2,20:50,200-terthiophene-5-
carbonitrile (L3)
As described above for L2 from 2,20:50,200-terthiophene-
5-carbonitrile (1.0 g, 3.7 mmol), claycop (1.5 g) and acetic
anhydride (5 ml). The same workup of the reaction mixture
gives 500-nitro-2,20:50,200-terthiophene-5-carbonitrile (80%)
and 400-nitro-2,20:50,200-terthiophene-5-carbonitrile (20%)
isomers which were separated by fractioned recrystalliza-
tion from ethanol. The desired product (0.77 g, 65% yield)
was thus obtained as a red microcrystalline solid.
4.2. Synthesis of 5-nitrothiophene-2-carbonitrile (L1) [30]
IR (KBr) cmꢀ1: m(N„C) 2190; m(NO2) 1485 and 1325;
d(NO2) 750; 1H NMR (CD2Cl2): d 7.16 (d, 1H, H11,
3
5-Nitro-2-thiophenecarboxaldehyde (7.8 g, 0.05 mol)
was dissolved in pyridine (20 ml) and hydroxylamine
hydrochloride (6.9 g, 0.1 mol) in the same solvent (20 ml)
was added. Acetic anhydride (23 ml) was then added and
the mixture heated to reflux for 1 h. After cooling, the mix-
ture was poured onto ice and the precipitate formed was fil-
tered, washed with water, dissolved in diethyl ether and
dried over sodium sulfate. The solvent was removed under
vacuum and the solid was extracted in a Soxhlet apparatus
with n-pentane to give 6.3 g (82%) of the desired product as
yellow needles.
3JHH = 3.9), 7.23 (d, H, H3, JHH = 3.9), 7.29 (d, 1H,
3
3
H7, JHH = 3.9), 7.35 (d, 1H, H8, JHH = 3.9), 7.58 (d,
1H, H4, JHH = 3.9), 7.87 (d, 1H, H12, JHH = 4.5); 13C
NMR (CD2Cl2): d 109.02 (C2), 114.11 (NC), 123.62 (C8),
124.84 (C7), 127.44 (C4), 127.95 (C11), 130.19 (C12),
136.18 (C9), 137.24 (C6), 138.91 (C3), 143.36 (C5), 144.06
(C10), 150.29 (C13); UV–Vis (CH2Cl2): kmax/nm (e/
3
3
M
ꢀ1 cmꢀ1) 426 (37300).
4.5. Synthesis of [FeCp(P_P)(NC{SC4H2}nNO2)][PF6]
IR (KBr) cmꢀ1: m(N„C) 2225; m(NO2) 1515 and 1335;
d(NO2) 730; 1H NMR (CDCl3): d 7.59 (d, 1H, H3,
All the complexes (P_P = dppe or (+)-diop; n = 1, 2 or
3) were prepared as follows. TlPF6 (0.48 mmol) was
added to a solution of [FeCp(P_P)I] (0.40 mmol) and
the appropriate nitrile derivative (0.48 mmol) in dichloro-
methane (20 ml). The suspension was stirred at room
temperature for 48–96 h. A change was observed from
dark-violet to reddish with simultaneous precipitation of
TlI. After filtration, the solvent was evaporated under
vacuum and the solid residue washed several times with
diethyl ether/dichloromethane mixture to remove the
excess of the ligand. The obtained solid was recrystallized
from dichloromethane/n-pentane or n-hexane affording
the desired complexes almost as dark red microcrystalline
products.
3
3JHH = 4.5), 7.91 (d, 1H, H4, JHH = 4.2); 13C NMR
(CDCl3): d 111.93 (C2), 115.43 (NC), 127.54 (C4), 136.37
(C3), 155.62 (C5); UV–Vis (CH2Cl2):
M
kmax/nm (e/
ꢀ1 cmꢀ1) 301 (10200).
4.3. Synthesisof50-nitro-2,20-bithiophene-5-carbonitrile(L2)
2,20-Bithiophene-5-carbonitrile (5.7 g, 0.03 mol) was
added to a suspension of claycop (12.6 g) in acetic anhy-
dride (37 ml) below 5 ꢁC with vigorous stirring. Subse-
quently, the reaction mixture was allowed to achieve the
room temperature and stirred for 1 h. The mixture was
poured onto ice and the precipitate formed was filtered,
washed with dichloromethane and filtered again to sepa-
rate the clay. The filtrate was washed with a saturated solu-
tion of NaHCO3 and water, dried with Na2SO4 and the
solvent was removed under vacuum giving a yellow solid
as a mixture of 50-nitro-2,20-bithiophene-5-carbonitrile
(90%) and 40-nitro-2,20-bithiophene-5-carbonitrile (10%).
The two isomers were separated by fractioned recrystalliza-
tion from ethanol giving 5.5 g (78%) of the desired com-
pound as a yellow microcrystalline solid.
1H and 13C NMR data relative to the dppe and (+)-diop
coordinated phosphines are very similar in all compounds,
and are described below.
1
Compounds 1a and 2a. For dppe: H NMR (CDCl3): d
2.40–2.70 (m, 4H, CH2), 7.27–7.36 (m, 4H, C6H5),
7.48–7.64 (m, 12H, C6H5), 7.78–7.86 (m, 4H, C6H5); 13C
1
NMR (CDCl3): d 28.07 (t, CH2, JCP = 21.4), 129.31 and
3
129.66 (t, Cmeta, JCP = 4.9), 130.80 and 131.37 (s, Cpara),
2
131.08 and 132.67 (t, Corto, JCP = 4.8), 136.24 (t, Cipso
,
1JCP = 20.7).
IR (KBr) cmꢀ1: m(N„C) 2220; m(NO2) 1520 and 1335;
d(NO2) 730; 1H NMR (CDCl3): d 7.22 (d, 1H, H7,
Compound 3a. For dppe: 1H NMR (CD2Cl2): d
2.40–2.70 (m, 4H, CH2), 7.32–7.38 (m, 4H, C6H5), 7.48–
7.54 (m, 6H, C6H5); 7.58–7.66 (m, 6H, C6H5); 7.80–7.88
(m, 4H, C6H5); 13C NMR (CD2Cl2): d 28.10 (t, CH2,
3
3JHH = 4.2), 7.33 (d, 1H, H4, JHH = 3.9), 7.61 (d, 1H,
3
3
H3, JHH = 3.9), 7.89 (d, 1H, H8, JHH = 4.2); 13C NMR

M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
3039
3
3
1JCP = 21.4), 129.64 and 129.82 (t, Cmeta
,
3JCP = 4.9),
H3, JHH = 4.2), 6.96 (d, 1H, H4, JHH = 3.9), 7.15 (d,
3
3
131.66 and 131.66 (s, Cpara), 131.60 and 133.02 (t, Corto
,
1H, H11, JHH = 4.5), 7.17 (d, 1H, H7, JHH = 4.5), 7.32
1
3
3
2JCP = 4.9), 136.50 (t, Cipso, JCP = 22.4).
(d, 1H, H8, JHH = 4.2), 7.85 (d, 1H, H12, JHH = 4.5);
13C NMR (CD2Cl2): d 80.40 (g5-C5H5), 107.75 (C2),
123.82 (C8), 124.58 (C7), 127.76 (C11), 128.06 (C4),
129.32 (NC), 130.26 (C12), 136.54 (C9), 136.70 (C6),
139.59 (C3), 143.71 (C10), 143.88 (C5), 150.29 (C13); 31P
Compounds 1b and 2b. For (+)-diop: 1H NMR (CDCl3):
d 1.09 (s, 3H, CH3), 1.27 (s, 3H, CH3), 2.24–2.36 (m, 2H,
CH2), 3.03–3.18 (m, 2H, CH2), 3.20–3.28 (m, 1H, CH),
3.42–3.52 (m, 1H, CH), 7.20–7.27 (m, 4H, C6H5),
3
7.44–7.52 (m, 6H, C6H5), 7.63 (t, 4H, C6H5, JHH = 8.5),
NMR
(CD2Cl2):
d
97.63.
Anal.
Calc.
for
3
7.74 (t, 4H, C6H5, JHH = 7.5), 8.02 (t, 2H, C6H5,
C44H35F6N2O2P3S3Fe: C, 53.78; H, 3.59, N, 2.85; S, 9.79.
Found: C, 53.47; H, 3.39, N, 2.73; S, 9.54%.
3JHH = 8.6); 13C NMR (CDCl3): d 26.68 (s, CH3), 26.85
1
(s, CH3), 28.42 (d, CH2, JCP = 18.7), 30.92 (d, CH2,
2
1JCP = 27.0); 75.25 (d, CH, JCP = 11.0), 78.06 (d, CH,
4.5.4. [FeCp((+)-diop)(NC{SC4H2}NO2)][PF6] (1b)
Dark red; recrystallized from CH2Cl2/n-pentane; 37%
yield; m.p. 185 ꢁC (dec.); molar conductivity (Xꢀ1
cm2molꢀ1) 77.4; IR (KBr) cmꢀ1: m(N„C) 2180; 1H
NMR (CDCl3): d 4.23 (s, 5H, g5-C5H5), 7.80 (d, 1H, H3,
2JCP = 7.2), 109.14 (s, C(CH3)2), 128.86–134.21 (m,
C6H5), 138.22 (d, Cipso
,
1JCP = 42.9), 141.32 (d, Cipso
,
1JCP = 47.0).
1
Compound 3b. For (+)-diop: H NMR (CD2Cl2): d 1.00
(s, 3H, CH3), 1.20 (s, 3H, CH3), 2.22–2.37 (m, 2H, CH2),
3.01–3.17 (m, 2H, CH2), 3.21–3.32 (m, 1H, CH),
3.37–3.53 (m, 1H, CH), 7.32–7.80 (m, 18H, C6H5), 7.96–
8.07 (m, 2H, C6H5); 13C NMR (CD2Cl2): d 26.80 (s,
3JHH = 4.2), 7.85 (d, 1H, H4, JHH = 4.2); 13C NMR
3
(CDCl3): d 82.53 (g5-C5H5), 113.99 (C2), 127.76 (NC),
128.71 (C4), 139.22 (C3), 154.07 (C5); 31P NMR (CDCl3):
d
53.3 (2d, JPAPB = 47.4). Anal. Calc. for
C41H39F6N2O4P3SFe: C, 53.61; H, 4.28, N, 3.05; S, 3.49.
Found: C, 53.43; H, 4.35, N, 2.97; S, 3.37%.
1
CH3), 26.93 (s, CH3), 28.70 (d, CH2, JCP = 18.0), 31.32
1
2
(d, CH2, JCP = 20.7), 75.42 (d, CH, JCP = 11.0), 78.39
2
(d, CH, JCP = 7.2), 109.46 (s, C(CH3)2), 129.40–134.65
1
(m, C6H5), 138.80 (d, Cipso, JCP = 42.2), 141.27 (d, Cipso
1JCP = 42.0).
,
4.5.5. [FeCp((+)-diop)(NC{SC4H2}2NO2)][PF6] (2b)
Dark red; recrystallized from CH2Cl2/n-pentane; 32%
yield; m.p. 140–142 ꢁC; molar conductivity (Xꢀ1
cm2 molꢀ1) 72.6; IR (KBr) cmꢀ1: m(N„C) 2190; 1H
NMR (CDCl3): d 4.16 (s, 5H, g5-C5H5), 7.20 (d, 1H, H7,
4.5.1. [FeCp(dppe)(NC{SC4H2}NO2)][PF6] (1a)
Dark red; recrystallized from CH2Cl2/n-pentane; 56%
yield; m.p. 220 ꢁC (dec.); molar conductivity (Xꢀ1
cm2 molꢀ1) 75.3; IR (KBr) cmꢀ1: m(N„C) 2190; 1H
NMR (CDCl3): d 4.55 (s, 5H, g5-C5H5), 6.98 (d, 1H, H3
3JHH = 4.2), 7.31 (d, 1H, H3, JHH = 4.5), 7.54 (d, 1H,
3
3
3
H4, JHH = 4.2), 7.85 (d, 1H, H8, JHH = 4.2); 13CNMR
(CDCl3): d 81.76 (g5-C5H5), 109.35 (C2), 125.29 (C7),
126.99 (C4), 128.29 (C8), 129.48 (NC), 140.51 (C3),
141.12 (C5), 141.17 (C6), 151.30 (C9); 31P NMR (CDCl3):
3
3JHH = 4.2), 7.61 (d, 1H, H4, JHH = 4.2); 13C NMR
(CDCl3): d 81.03 (g5-C5H5), 114.00 (C2), 125.14 (NC),
128.22 (C4), 138.22 (C3), 153.54 (C5); 31P NMR (CDCl3):
d 97.22. Anal. Calc. for C36H31F6N2O2P3SFe: C, 52.83; H,
3.82, N, 3.42; S, 3.92. Found: C, 52.55; H, 3.80, N, 3.30; S,
4.01%.
d
53.7 (2d, JPAPB = 47.4). Anal. Calc. for
C45H41F6N2O4P3S2Fe: C, 54.01; H, 4.13, N, 2.80; S, 6.41.
Found: C, 53.93; H, 4.01, N, 2.88; S, 6.25%.
4.5.6. [FeCp((+)-diop)(NC{SC4H2}3NO2)][PF6] (3b)
Brownish-red; recrystallized from CH2Cl2/n-pentane;
28% yield; m.p. 134–136 ꢁC; molar conductivity (Xꢀ1
cm2 molꢀ1) 69.8; IR (KBr) cmꢀ1: m(N„C) 2190; 1H
NMR (CD2Cl2): d 4.08 (s, 5H, g5-C5H5), 7.16 (d, 1H,
4.5.2. [FeCp(dppe)(NC{SC4H2}2NO2)][PF6] (2a)
Dark red; recrystallized from CH2Cl2/n-hexane; 48%
yield; m.p. 230 ꢁC (dec.); molar conductivity (Xꢀ1
cm2 molꢀ1) 74.9; IR (KBr) cmꢀ1: m(N„C) 2200; 1H
NMR (CDCl3): d 4.50 (s, 5H, g5-C5H5), 6.76 (d, 1H, H3
3
3
H11, JHH = 4.5), 7.18 (d, 1H, H3, JHH = 4.2), 7.25 (d,
3
3
3
3JHH = 4.2), 7.07 (d, 1H, H4, JHH = 4.2), 7.11 (d, 1H,
1H, H7, JHH = 4.2), 7.26 (d, 1H, H4, JHH = 3.9), 7.33
3
3
3
3
H7, JHH = 4.2), 7.83 (d, 1H, H8, JHH = 4.5); 13C NMR
(CDCl3): d 80.24 (g5-C5H5), 110.14 (C2), 125.01 (C7),
126.42 (C4), 127.00 (NC), 128.04 (C8), 139.91 (C3),
140.51 (C5), 140.90 (C6), 151.07 (C9); 31P NMR (CDCl3):
d 97.49. Anal. Calc. for C40H33F6N2O2P3S2Fe: C, 53.35; H,
3.69, N, 3.11; S, 7.12. Found: C, 52.98; H, 3.74, N, 3.02; S,
6.98%.
(d, 1H, H8, JHH = 3.9), 7.86 (d, 1H, H12, JHH = 4.2),
13C NMR (CD2Cl2): d 81.78 (g5-C5H5), 107.48 (C2),
123.93 (C8), 125.17 (C7), 128.09 (C11), 128.11 (C4),
130.40 (C12), 132.02 (NC), 136.45 (C9), 136.94 (C6),
140.07 (C3), 143.71 (C10), 144.66 (C5), 150.51 (C13); 31P
NMR (CD2Cl2): d 53.4 (2d, JPAPB = 47.4). Anal. Calc.
for C49H43F6N2O4P3S3Fe: C, 54.35; H, 4.00, N, 2.59; S,
8.88. Found: C, 54.19; H, 3.91, N, 2.48; S, 8.74%.
4.5.3. [FeCp(dppe)(NC{SC4H2}3NO2)][PF6] (3a)
Brownish-red; recrystallized from CH2Cl2/n-hexane;
40% yield; m.p. 179–180 ꢁC; molar conductivity (Xꢀ1
cm2molꢀ1) 71.2; IR (KBr) cmꢀ1: m(N„C) 2210; 1H
NMR (CD2Cl2): d 4.49 (s, 5H, g5-C5H5), 6.50 (d, 1H,
4.6. Electrochemical experiments
The electrochemistry instrumentation consisted of a
EG&A Princeton Applied Research Model 273A

3040
M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
Potentiometer and experiments were monitored with a PC
computer loaded with Model 270 Electrochemical Analysis
Software 3.00 of EG&A from Princeton Applied Research.
Potentials were referred to a calomel electrode containing a
saturated solution of potassium chloride. The working
electrode was a 2-mm piece of platinum wire for voltamme-
try. The secondary electrode was a platinum wire coil. Cyc-
lic voltammetry experiments were performed at room
temperature and ꢀ20 ꢁC in a PAR polarographic cell.
Solutions studied were 1 mM in solute and 0.1 M in tetra-
butylammonium hexafluorophosphate as supporting elec-
trolyte. The electrochemical system was checked with a
1 mM solution of ferrocene in acetonitrile and dichloro-
methane for which the ferrocinium/ferrocene electrochem-
is a good approximation, within experimental accuracy, to
adopt the same value for the two different wavelengths.
Acknowledgements
Financial support from Fundac¸ao para a Cieˆncia e
˜
Tecnologia (FCT), project POCTI/QUI/48443/2002 and
from the Fund for Scientific Research of Flanders
(Belgium) (FWO–Vlaanderen) in the group Projects
G.0041.01 and G.0129.07 is gratefully acknowledged.
W.W. is a Postdoctoral Fellow of the FWO.
References
ical parameters (Ep/2 = 0.38 V in acetonitrile and Ep/2
=
[1] H.S. Nalwa, Appl. Organomet. Chem. 5 (1991) 349.
[2] N.J. Long, Angew. Chem., Int. Ed. Engl. 34 (1995) 21.
[3] S.R. Marder, in: D.W. Bruce, D. O’Hare (Eds.), Inorganic Materials,
second ed., Wiley, Chichester, 1996, p. 121.
[4] I.R. Whittall, A.M. McDonagh, M.G. Humphrey, M. Samoc, Adv.
Organomet. Chem. 42 (1998) 291.
[5] E. Goovaerts, W.E. Wenseleers, M.H. Garcia, G.H. Cross, in: H.S.
Nalwa (Ed.), Handbook of Advanced Electronic and Photonic
Materials, vol. 9, Academic Press, San Diego, 2001, p. 127 (Chapter 3).
[6] S. Di Bella, Chem. Soc. Rev. 30 (2001) 355.
[7] J.C. Calabrese, L.T. Cheng, J.C. Green, S.R. Marder, W. Tam, J.
Am. Chem. Soc. 113 (1991) 7227–7232.
0.41 V in dichloromethane; DE = 60–70 mV; Ia/Ic = 1)
were in good agreement with the literature [38,39].
The electrolyte was purchased from Aldrich Chemical
Co., recrystallized from ethanol, washed with diethyl ether,
and dried under vacuum at 110 ꢁC for 24 h. Reagent grade
acetonitrile and dichloromethane were dried over P2O5 and
CaH2, respectively, and distilled under argon atmosphere
before use. An argon atmosphere was also maintained over
the solution during the experiments.
[8] A. Krishnan, S.K. Pal, P. Nandakumar, A.G. Samuelson, P.K. Das,
Chem. Phys. 265 (2001) 313–322.
4.7. b Measurements
[9] W. Wenseleers, A.W. Gerbrandij, E. Goovaerts, M.H. Garcia, M.P.
Robalo, P.J. Mendes, J.C. Rodrigues, A.R. Dias, J. Mater. Chem. 8
(1998) 925.
HRS measurements at 1.064 lm were performed using
70 ps pulses from a 2 kHz repetition rate Nd:YAG regener-
ative amplifier seeded by a mode-locked Nd:YAG laser (as
described in Ref. [5]). The scattered second-harmonic light
is analyzed with a monochromator, in addition to elec-
tronic gating within a 4 ns time window in order to discrim-
inate efficiently between hyper-Rayleigh scattering and
(two-photon) luminescence. For the measurements at
1.550 lm a recently developed HRS-setup [33] is used,
based on an optical parametrical amplifier (OPA) pumped
by a Ti:Sapphire regenerative amplifier, and with single-
photon sensitive, parallel detection of a small spectral
range around the second-harmonic wavelength. With this
setup, measurements were also performed at 1.072 lm,
reproducing the 1.064 lm b values to within the experimen-
tal error of 5%. For the compounds studied here some
decomposition in the focused laser beam took place, which
was observed in the measurements at 1.064 lm by a slight
decrease in HRS intensity (typically in the order of 10%)
within a few seconds after the liquid cell was moved to
another position relative to the laser beam as well as
through a slight deviation from the quadratic power depen-
dence of the HRS signal. This problem was solved by hav-
ing a magnetic stirrer rotating in the liquid cell during the
HRS measurements in order to continuously refresh the
solution in the small volume of the laser focus. All mea-
surements are calibrated relative to pure chloroform [40],
for which the literature value of 0.49 · 10ꢀ30 esu is adopted
(EFISHG, from Ref. [41], using the usual assumptions [5]).
Because of the very low b dispersion of the pure solvent, it
[10] M.H. Garcia, M.P. Robalo, A.R. Dias, M.F.M. Piedade, A. Galvao,
˜
M.T. Duarte, W. Wenseleers, E. Goovaerts, J. Organomet. Chem 619
(2001) 252.
[11] M.H. Garcia, M.P. Robalo, A.R. Dias, M.T. Duarte, W. Wenseleers,
G. Aerts, E. Goovaerts, M.P. Cifuentes, S. Hurst, M.G. Humphrey,
M. Samoc, B. Luther-Davies, Organometallics 21 (2002) 2107.
[12] C.E. Powell, M.G. Humphrey, Coord. Chem. Rev. 248 (2004) 725.
[13] M.P. Cifuentes, M.G. Humphrey, J. Organomet. Chem. 689 (2004)
3968.
[14] J.L. Oudar, H. Leperson, Opt. Commun. 15 (1975) 258–262.
[15] A. Dulcic, C. Flytzanis, C.L. Tang, D. Pepin, M. Fetizon, Y.
Hoppilliard, J. Chem. Phys. 74 (1981) 1559–1563.
[16] A.M. McDonagh, I.R. Whittall, M.G. Humphrey, B.W. Skelton,
A.H. White, J. Organomet. Chem 519 (1996) 229.
[17] M. Ciofalo, G. La Manna, Chem. Phys. Lett. 263 (1996) 73.
[18] T.L. Porter, D. Minore, D. Zhang, J. Phys. Chem. 99 (1995) 13213.
[19] R.D.A. Hudson, A.R. Manning, D.F. Nolan, I. Asselberghs, R. Van
Boxel, A. Persoons, J.F. Gallagher, J. Organomet. Chem. 619 (2001)
141.
[20] R.D.A. Hudson, I. Asselberghs, K. Clays, L.P. Cuffe, J.F. Gallagher,
A.R. Manning, A. Persoons, K. Wostyn, J. Organomet. Chem. 637–
639 (2001) 435.
[21] M.H. Garcia, S. Royer, M.P. Robalo, A.R. Dias, J.-P. Tranchier, R.
Chavignon, D. Prim, A. Auffrant, F. Rose-Munch, E. Rose, J.
Vaissermann, A. Persoons, I. Asselberghs, Eur. J. Inorg. Chem.
(2003) 3895.
´
[22] B. Insuasty, C. Atienza, C. Seoane, N. Martın, J. Garın, J. Orduna,
R. Alcala
´, B. Villacampa, J. Org. Chem. 69 (2004) 6986.
[23] J.-L. Fillaut, J. Perruchon, P. Blanchard, J. Roncali, S. Golhen, M.
Allain, A. Migalsaka-Zalas, I.V. Kityk, B. Sahraoui, Organometallics
24 (2005) 687.
[24] Y. Liao, B.E. Eichinger, K.A. Firestone, M. Haller, J. Luo, W.
Kaminsky, J.B. Benedict, P.J. Reid, A.K.-Y. Jen, L.R. Dalton, B.H.
Robinson, J. Am. Chem. Soc. 127 (2005) 2758.

M.H. Garcia et al. / Journal of Organometallic Chemistry 692 (2007) 3027–3041
3041
´
[25] A.-L. Roy, M. Chavarot, E. Rose, F. Rose-Munch, A.-J. Attias,
[33] M.P. Robalo, A.P.S. Teixeira, M.H. Garcia, M.F.M. da Piedade,
M.T. Duarte, A.R. Dias, J. Campo, W. Wenseleers, E. Goovaerts,
Eur. J. Inorg. Chem. 11 (2006) 2175–2185.
´
D. Kreher, J.-L. Fave, C. Kamierszky, C.R. Chimie 8 (2005)
1256.
[26] P.M. Theus, W. Weuffen, H. Tiedt, Arch. Pharm. 301 (1968) 139.
[27] W.J. Geary, Coord. Chem. Rev. 7 (1977) 81.
[28] A.R. Dias, M.H. Garcia, M.P. Robalo, M.L.H. Green, K.K. Lai,
A.J. Pulham, S.M. Klueber, G. Balavoine, J. Organomet. Chem 453
(1993) 241.
[34] P.J. Mendes, J.P. Prates Ramalho, A.J.E. Candeias, M.P. Robalo,
M.H. Garcia, J. Mol. Struct. (Theochem) 729 (2005) 109.
[35] D.D. Perrin, W.L.F. Amarego, D.R. Perrin, Purification of Labora-
tory Chemicals, second ed., Pergamon Press, New York, 1980.
[36] Y. Wei, B. Wang, W. Wang, J. Tian, Tetrahedron Lett. 36 (1995)
665.
[29] P. Hapiot, F. Demanze, A. Yassar, F. Garnier, J. Phys. Chem. 100
(1996) 8397.
´
[37] P. Laszlo, A. Cornelis, Aldrichim. Acta 21 (1988) 97.
[30] J.L. Oudar, D.S. Chemla, J. Chem. Phys. 66 (1977) 2664.
[31] G. Berkovic, G. Meshulam, Z. Kotler, J. Chem. Phys. 112 (2000)
3997.
[38] I.V. Nelson, R.T. Iwamoto, Anal. Chem. 35 (1961) 867.
´
[39] R.R. Gagne, C.A. Koval, G.C. Lisensky, Inorg. Chem. 19 (1980)
2855.
[32] C.H. Wang, Y.C. Lin, O.Y. Tai, A.K.-Y. Jen, J. Chem. Phys. 119
(2003) 6237–6244.
[40] K. Clays, A. Persoons, Phys. Rev. Lett. 66 (1991) 2980.
[41] F. Kajzar, I. Ledoux, J. Zyss, Phys. Rev. A 36 (1987) 2210.