Electron transfer in a trinuclear oxo-centred mixed-valence iron complex, in solid and solution states

Article abstract of DOI:10.1039/b100780g

Complexes [Fe^III₂M^IIOL₃] (M = Fe, Co, Ni, Cu; 2, 3, 4, 5) have been synthesised in which L^2- is a pentadentate ligand designed to coordinate all three metal atoms in the central cluster and to inhibit dissociation and solvent exchange processes. Crystal structures for 2, 4 and 5 show threefold symmetry, attributed to rotational disorder. Magnetisation data for 2 indicate strong superexchange between basis oxidation states Fe(3+, 3+, 2+). Comparisons of IR spectra across the series of complexes confirm the non-threefold symmetry of the mixed-valence cluster on the vibrational time scale, both in the solid state and in solution. Proton NMR spectra in solution at room temperature do not distinguish the three iron sites, suggesting that pseudo-rotation by thermal electron transfer also operates. Cyclic voltammetry and spectroelectrochemical measurements show that the mixed-valence iron complex 2 can be oxidised reversibly to give the tri-iron(III) complex [Fe₃OL₃]⁺ and reduced reversibly and quasireversibly to give respectively [Fe₃OL₃]^- and tri-iron(II) [Fe₃OL₃]^2-, E⁰ = 85,-635, -1230 mV (versus Fc^+/0) in dichloromethane (T = 298 K, 0.1 M [n-Bu₄N][PF₆]). Moessbauer spectra of 2 indicate significant valence delocalisation even at low temperature (4.2 K) with estimated valences Fe(2.9+, 2.9+, 2.2+) in the solid state. At higher temperatures no lifetime broadening is observed but additional Moessbauer absorptions are consistent with increasing proportions of trimer molecules with greater delocalisation, i.e. Fe(2.75+, 2.75+, 2.5+). In frozen solution (THF) the spectra indicate increasing proportions of molecules fully valence-delocalised on the Moessbauer time scale. The data are accounted for with a model which places the complex at the Robin-Day class III/II borderline. It combines strong superexchange with significant double exchange even at the lowest temperatures, while at higher temperatures in solution complete valence delocalisation occurs through intramolecular electron transfer at rates intermediate between the IR and NMR time scales.

Full text of DOI:10.1039/b100780g

View Article Online / Journal Homepage / Table of Contents for this issue
Electron transfer in a trinuclear oxo-centred mixed-valence iron
complex, in solid and solution states†
Christian Stadler,*‡a Jörg Daub,*a Jürgen Köhler,b Rolf W. Saalfrank,*b
Veaceslav Coropceanu,c Volker Schünemann,c Claudia Ober,c Alfred X. Trautwein,*c
Stewart F. Parker,d Mehmet Poyraz,e Tomohiko Inomatae and Roderick D. Cannon*e
a Institut für Organische Chemie der Universität Regensburg, Universitätsstraße 31,
D-93053 Regensburg, Germany
b Institut für Organische Chemie der Universität Erlangen-Nürnberg, Henkestraße 42,
D-91054 Erlangen, Germany
c Institut für Physik der Medizinischen Universität Lübeck, Ratzeburger Allee 160,
D-23538 Lübeck, Germany
d Rutherford Appleton Laboratory, Chilton, Didcot, UK OX11 0QX
e School of Chemical Sciences, University of East Anglia, Norwich, UK NR4 7TJ.
E-mail: r.cannon@uea.ac.uk
Received 23rd January 2001, Accepted 28th August 2001
First published as an Advance Article on the web 31st October 2001
Complexes [FeIII2MIIOL3] (M = Fe, Co, Ni, Cu; 2, 3, 4, 5) have been synthesised in which L2Ϫ is a pentadentate ligand
designed to coordinate all three metal atoms in the central cluster and to inhibit dissociation and solvent exchange
processes. Crystal structures for 2, 4 and 5 show threefold symmetry, attributed to rotational disorder. Magnetisation
data for 2 indicate strong superexchange between basis oxidation states Fe(3ϩ, 3ϩ, 2ϩ). Comparisons of IR spectra
across the series of complexes confirm the non-threefold symmetry of the mixed-valence cluster on the vibrational
time scale, both in the solid state and in solution. Proton NMR spectra in solution at room temperature do not
distinguish the three iron sites, suggesting that pseudo-rotation by thermal electron transfer also operates. Cyclic
voltammetry and spectroelectrochemical measurements show that the mixed-valence iron complex 2 can be oxidised
reversibly to give the tri-iron() complex [Fe3OL3]ϩ and reduced reversibly and quasireversibly to give respectively
[Fe3OL3]Ϫ and tri-iron() [Fe3OL3]2Ϫ, E 0 = 85, Ϫ635, Ϫ1230 mV (versus Fcϩ/0) in dichloromethane (T = 298 K, 0.1 M
[n-Bu4N][PF6]). Mössbauer spectra of 2 indicate significant valence delocalisation even at low temperature (4.2 K)
with estimated valences Fe(2.9ϩ, 2.9ϩ, 2.2ϩ) in the solid state. At higher temperatures no lifetime broadening is
observed but additional Mössbauer absorptions are consistent with increasing proportions of trimer molecules
with greater delocalisation, i.e. Fe(2.75ϩ, 2.75ϩ, 2.5ϩ). In frozen solution (THF) the spectra indicate increasing
proportions of molecules fully valence-delocalised on the Mössbauer time scale. The data are accounted for with
a model which places the complex at the Robin–Day class III/II borderline. It combines strong superexchange with
significant double exchange even at the lowest temperatures, while at higher temperatures in solution complete
valence delocalisation occurs through intramolecular electron transfer at rates intermediate between the IR and
NMR time scales.
carboxylate ligands, [M3O(OOCR)6L3], and is structurally
Introduction
related to the active sites of numerous iron–oxo proteins.5 In
Electron transfer (ET) processes continue to be at the forefront
this class, the transition between valence-localised and delocal-
of kinetic and mechanistic studies, with applications to funda-
ised phases has been studied by crystallographic, thermo-
mental processes in chemistry, physics and biology.1 Especially
dynamic and spectroscopic methods.6 The dynamics of electron
valuable are model systems in which the electron moves
transfer and of coupled molecular motions have been measured
between chemically equivalent sites where it is only weakly
in the solid state by NMR,7 neutron scattering,8 and infrared9
trapped, i.e. mixed-valence molecules at the borderline between
spectroscopies. Solution studies however are complicated by
localisation and delocalisation. Two-site exchange has been
rapid ligand-exchange reactions10 and low stability of the
studied intensively,2 and recent work has confirmed the role of
reduced forms.11 We have now been able to circumvent these
solvent organisation, framework vibrations and magnetic
limitations using the novel system [M3OL3] (Scheme 1) in which
exchange in controlling electron transfer rates.3 Three-site
exchange offers further possibilities. With suitable ligands,
a wide variety of metal combinations can be assembled in a
common framework; also the interactions can be studied
through their direct effect on the molecular symmetry. One such
class of compounds4 has the central trimetallic unit bridged by
† Dedicated to Professor Dieter Sellmann on the occasion of his 60th
birthday.
‡ Present address: Instituto de Catálisis y Petroleoquímica (CSIC),
Campus Universidad Autónoma, Cantoblanco, 28049, Madrid, Spain.
Scheme 1
DOI: 10.1039/b100780g
J. Chem. Soc., Dalton Trans., 2001, 3373–3383
3373
This journal is © The Royal Society of Chemistry 2001

View Article Online
1
mp 226 ЊC; H NMR ([D6]-DMSO, 400 MHz) δ 3.71 (s, 3 H,
CH3), 6.66 (s, 2 H, NH2); 13C NMR ([D6]-DMSO, 100.5 MHz)
δ 31.50 (CH ), 155.84 (C᎐N); EI-MS m/z 99 (20%) [C N H ϩ].
᎐
3
2
5
5
1-Trideuteromethyl-5-aminotetrazole. Prepared as above
using trideuteromethyl iodide (0.1 mol, 14.5 g). Yield 3.47 g
(34%); mp 207 ЊC; 1H NMR ([D6]-DMSO, 400 MHz) δ 6.63 (s, 2
H, NH2); 13C NMR ([D6]-DMSO, 100.5 MHz) δ 30.86 (CD3),
155.84 (C᎐N); EI-MS m/z 102 (100%) [C N H D ϩ].
᎐
2
5
2
3
H2L, 1.15 A solution of 1-methyl-5-aminotetrazole (20 mmol,
1.98 g) in 25 ml of 1-methylpyrrolidin-2-one containing lithium
chloride (2.0 g) was cooled to 0 ЊC, and 2,6-dichloroformyl
pyridine (10 mmol, 2.03 g) was added at once. The mixture was
stirred for 72 h at 20 ЊC and was poured into 100 ml of ice water.
After stirring for 2 h the white precipitate was filtered off and
washed with 100 ml of water. The products were used without
further purification. Yield 2.63 g (80%); mp >250 ЊC (decomp.);
1H NMR ([D6]-DMSO, 400 MHz) δ 4.02 (s, 6 H, CH3), 8.39
(t, J = 7.81 Hz, 1 H, Ar–H{4}), 8.47 (d, J = 7.81 Hz, 2 H,
Ar–H{3,5}), 12.08 (s, 2 H, NH); 13C NMR ([D6]-DMSO,
100.5 MHz) δ 34.64 (2 CH3), 126.90 (2 Ar–CH{3,5}), 140.64
Fig. 1 Molecular structure of complexes [Fe3OL3]nϩ and [Fe2MOL3].
each pentadentate ligand L simultaneously binds all three metal
atoms. The molecular structure is shown in Fig. 1. As we show
here, the complexes are robust in solution, and in case of the
homonuclear complex [Fe3OL3], 2, all four redox states from
[MIII3OL3]ϩ to [MII3OL3]2Ϫ are electrochemically accessible and
can be characterised spectroscopically. The focus of this paper
is the localisation or delocalisation of the ‘extra’ electron in the
nominally mixed-valence cluster iron(,,). It is perhaps
worth stating that from our point of view, delocalisation can
mean either of two things: mixing of wavefunctions so that the
electronic configuration provides electron density distributed
over more than one centre, or reversible variation of bond
lengths so the electron in effect hops between centres at a
definite frequency. We call these ‘static’ and ‘dynamic’ delocal-
isation, respectively. In a preliminary study12 of 2, Mössbauer
spectra indicated that the iron valencies are localised, at least at
4.2 K, whereas the electrochemical data at first sight suggested
delocalisation. Meanwhile, temperature-dependent Mössbauer
spectra have been analysed13 to show the existence of partially
delocalised states in the solid state throughout the whole tem-
perature range. To provide further insight into the electronic
structure of this system we have now performed a range of
further measurements, crystallographic, spectroscopic, electro-
chemical, and spectroelectrochemical, and we have compared
the mixed-valence cluster with the oxidised tri-iron() and
analogous mixed-metal complexes. We also outline a theoretical
model consistent with the data and show finally that the
mixed-valence molecule [Fe3OL3], 2, is close to the localised–
delocalised limit, and very different from the analogous basic
carboxylato-complexes.
(Ar–CH{4}), 146.76 (2 Ar–C{2,6}), 149.80 (2 C᎐N), 162.88
᎐
(2 C᎐O); EI-MS m/z 329 (2%) [H Lϩ].
᎐
2
H2L/CD3, 1/CD3. The same ligand with both methyl groups
fully deuterated was synthesised in the same way, using
1-trideuteromethyl-5-aminotetrazole (20 mmol, 2.04 g). Yield
1.88 g (57%); mp >250 ЊC (decomp.); 1H NMR (CDCl3–
CF3COOH, 400 MHz) δ 8.33 (t, J = 7.81 Hz, 1 H, Ar–H{4}),
8.60 (d, J = 7.81 Hz, 2 H, Ar–H{3,5}), due to the solvent
the NH-protons cannot be determined; 13C NMR (CDCl3–
CF3COOH, 100.5 MHz) δ 35.50 (2 CD3), 127.94 (2 Ar–
CH{3,5}), 140.62 (Ar–CH{4}), 146.52 (2 Ar–C{2,6}), 149.36 (2
C᎐N), 162.09 (2 C᎐O); EI-MS m/z 336 (2%) [H (L/CD )ϩ ϩ H].
᎐
᎐
2
3
[Fe3OL3], 2.12 To a solution of 1 (1.5 mmol, 494 mg) in 150
ml THF was added triethylamine (4.5 mmol, 455 mg, 0.63 ml)
and FeCl3ؒ6H2O (1.5 mmol, 405 mg). The mixture was heated
for 5 days under reflux, cooled to 20 ЊC, and filtered. The
solvent was removed under vacuum and the solid residue was
dissolved in 200 ml trichloromethane. To remove (HNEt3)-
[FeL2], the organic phase was washed eight times with 200 ml
portions of warm water (60 ЊC). The organic layer was separ-
ated, dried over sodium sulfate, and concentrated to 100 ml.
While 200 ml of hexane were added, a crystalline product
precipitated, which was filtered off and dried under vacuum (oil
pump). The final product was 2ؒ6CHCl3 — see syntheses of 3–5
below. Yield 245 mg (42%); mp >250 ЊC (decomp.); FAB-MS
m/z 1166 (100%) [Fe3OL3ϩ ϩ H].
[Fe3O(L/CD3)3], 2/CD3. The deuteromethyl analogue of 2
was prepared in the same way using 1/CD3 (1.5 mmol, 503 mg).
Yield 249 mg (41%); mp >250 ЊC (decomp.); FAB-MS m/z 1185
(100%) [Fe3O(L/CD3)3ϩ ϩ H].
Experimental
Materials
[Fe2CoOL3], 3; [Fe2NiOL3], 4; [Fe2CuOL3], 5. General pro-
cedure: To a solution of 1 (1.5 mmol, 494 mg) in 150 ml THF
was added triethylamine (4.5 mmol, 455 mg, 0.63 ml), the
appropriate metal chloride (1 mmol, 238 mg CoCl2ؒ6H2O,
238 mg NiCl2ؒ6H2O or 171 mg CuCl2ؒ2H2O) and FeCl3ؒ6H2O
(1 mmol, 270 mg). The mixture was heated for 9 days under
reflux, cooled to 20 ЊC, and filtered. The solvent was removed
under vacuum, the solid residue was dissolved in 200 ml tri-
chloromethane, and the organic phase was washed eight times
with 200 ml portions of warm water (60 ЊC). The organic layer
was separated, dried over sodium sulfate, and concentrated to
100 ml. On adding 200 ml of pentane or hexane, a crystalline
product precipitated, which was filtered off and dried under
vacuum (oil pump). 3: Yield 146 mg (25%); mp > 250 ЊC
All reagents and solvents were purchased from commercial
sources and used as received.
Synthesis
1-Methyl-5-aminotetrazole.14 Methyl iodide (0.1 mol, 14.2 g)
in 160 ml of acetone was added to a solution of 5-amino-
tetrazole monohydrate (0.1 mol, 10.3 g) and sodium hydroxide
(0.1 mol, 4.0 g) in 40 ml of water. The mixture was heated for 3
h under reflux, evaporated to dryness and the residue extracted
with five 50 ml portions of hot toluene. The toluene-insoluble
material was dissolved in 25 ml of hot water and cooled at 5 ЊC.
The solid material was removed by filtration, washed with 2 ml
of cold water and crystallised from water. Yield 3.27 g (33%);
3374
J. Chem. Soc., Dalton Trans., 2001, 3373–3383

View Article Online
ϩ
(decomp.); FAB-MS m/z 1169 (100%) [Fe2CoOL3 ϩ H]. 4:
perature under nitrogen using a conventional undivided elec-
Yield 269 mg (46%); mp >250 ЊC (decomp.); FAB-MS m/z 1168
trochemical cell with a three-electrode arrangement and a
computer-controlled Amel 5000 system. Redox potentials were
internally referenced against ferrocene/ferrocenium (Fc/Fcϩ).
CH2Cl2 for electrochemical experiments was refluxed and dis-
tilled over P2O5, stirred over K2CO3, deoxygenated with dry
nitrogen and finally filtered over activated Al2O3 (Woelm Super
I neutral) at Ϫ20 ЊC under an atmosphere of nitrogen. Tetra-n-
butylammonium hexafluorophosphate was prepared and puri-
fied according to a previously described procedure.16
UV/VIS/NIR spectroelectrochemistry was performed at 290
K under nitrogen on solutions previously analysed by cyclic
voltammetry. An electrochemical quartz cell with a three-
electrode arrangement (gold minigrid as transparent working
electrode), a Perkin-Elmer Lambda 9 spectrophotometer and
an Amel 2053 potentiostat were used.17
For FTIR spectroelectrochemistry a BioRad FTS 155 spec-
trometer and Amel 2053 potentiostat were used. The novel
spectroelectrochemical cell has been described recently.18 For a
spectroelectrochemical experiment, the analyte solution from
cyclic voltammetry experiments (0.6 ml) was introduced with a
syringe using Schlenk techniques. All spectra were recorded at
room temperature under nitrogen at a resolution of 8 cmϪ1. The
argon-flushed empty sample compartment was used as back-
ground and the spectrum of analyte-free electrolyte solution
was used to obtain corrected spectra.
(100%) [Fe2NiOL3ϩ ϩ H]. 5: Yield: 352 mg (60%); mp > 250 ЊC
ϩ
(decomp.); FAB-MS m/z 1173 (100%) [Fe2CuOL3 ϩ H]. The
solvent molecule content of the crystalline products was
deduced from X-ray crystallography and found to depend on
the choice of solvent used in the final precipitation step. For
2 and 5 this was hexane and the products were 2ؒ6CHCl3
and 5ؒ6CHCl3; for 4 it was pentane and the product was
4ؒ1.2CHCl3ؒ1.8C5H12. For 3 it was also hexane: the solvation
was not determined but CHCl3 was detected in the IR
spectrum.
[FeIII3OL3](NO3), 6. To a solution of 2 (0.05 mmol, 58 mg) in
60 ml of dichloromethane was added dropwise a solution of
(NH4)2[Ce(NO3)6] (0.05 mmol, 28 mg) in 40 ml of methanol.
The solution was stirred for 1 h at 20 ЊC and the solvent was
removed under vacuum. The solid residue was extracted four
times with 80 ml portions of dichloromethane. The organic
phases were combined and concentrated to 40 ml. While 100 ml
of hexane were added, a crystalline product precipitated, which
was filtered off and dried under vacuum (oil pump). Yield 35
mg (57%); mp >250 ЊC (decomp.); FAB-MS m/z 1166 (100%)
[Fe3OL3ϩ ϩ H].
(HNEt3)[FeIIIL2], 7. To a solution of FeCl3ؒ6H2O (1 mmol,
270 mg) in 60 ml of dichloromethane was added a solution of 1
(2 mmol, 659 mg) in 100 ml of dichloromethane. After dropwise
addition of triethylamine (4 mmol, 405 mg, 0.56 ml) the solu-
tion was stirred for 12 h at 20 ЊC. The solvent was removed
under vacuum and the solid residue was dissolved in 20 ml
trichloromethane. While 10 ml of diethyl ether were added, a
crystalline product precipitated which was filtered off and dried
under vacuum (oil pump). Yield: 710 mg (87%); mp >250 ЊC
(decomp.); FAB-MS m/z 713 (100%) [FeL2Ϫ ϩ 2Hϩ]; 813 (30%)
[HNEt3ϩ ϩ FeL2].
Density functional calculations
DFT calculations were performed on a model of the mixed-
valence iron complex 2, at the B3LYP level19 using the
Gaussian-98 package.20 Although the molecule is extremely
large for calculations at this level it was decided not to simplify
it drastically because the electronic structure of the ligand
system was expected to influence the properties of the Fe3O
core significantly. Therefore, coordinates were taken from the
X-ray analysis,12 but with hydrogens substituted for the methyl
groups and one of the iron atoms slightly displaced (by 0.01 Å
along the O–Fe vector) in order to lower the molecular sym-
metry from D3 to C2 and allow the density matrix to reproduce
the properties of a not fully valence-delocalised molecule. Since
the calculations are very time-consuming we did not attempt
to optimise the geometries, and we used the rather small
LANL2DZ basis set with a non-relativistic effective core
potential (ECP) at the iron atoms.21 Spin states S = 0, S = 1 and
Spectroscopic methods
UV/VIS/NIR spectra were recorded on a Perkin-Elmer Lambda
9 spectrophotometer at room temperature in CH2Cl2 and
CDCl3 (Merck, Uvasol).
IR spectra of the complexes were recorded on a Mattson
Galaxy series 4020 FTIR instrument, using KBr discs, at room
temperature, and at 80 K for complexes 2, 4 and 5. Generally at
low temperature, all peaks were shifted up by about 2 cmϪ1 and
resolution of close peaks was improved, but the only additional
peaks resolved were shoulders on one band in 2 and 4, as noted
below.
S = 2 were calculated and converged to 10Ϫ6
.
Crystal structure analysis
Crystal data for 4ؒ1.2CHCl3ؒ1.8C5H12. C43.2H49.8Cl3.6Fe2N33-
NiO7, M = 1441.39. Crystal dimensions: 0.8 × 0.6 × 0.5 mm,
trigonal, space group R3c (no. 161 of ref. 22): a = 2.3184(3), c =
22.634(3) Å, V = 10536(2) Å3, Z = 6. Diffractometer: Stoe-
Siemens-Huber, fitted with a graphite monitor and Siemens
SMART CCD area detector. Radiation: Mo-Kα, λ = 0.71073
Å, T = 133(2) K. Measurement details: φ and ω scan, range 35Њ
≤ 2θ ≤ 56.5Њ, section of the reciprocal lattice h = Ϫ26 to 0, k = 0
to 30, l = 0 to 30. A total of 42727 reflections were measured,
of which 2902 were considered independent, and 2452 were
considered observed as defined by I > 2(σ)I. Linear absorption
coefficient: 0.877 mmϪ1, absorption correction: semi-empirical
The inelastic incoherent neutron scattering (IINS) spectrum
of complex 2 was run on TOSCA at the Rutherford Appleton
Laboratory. The powdered sample, 0.23 g, was enclosed in
aluminium foil, folded to fill the beam area, 5 × 2 cm, as evenly
as possible. Data were treated using the standard in-house
program GENIE.
Mössbauer spectra were recorded using a conventional spec-
trometer in the constant-acceleration mode. Isomer shifts are
given relative to α-Fe at room temperature. The spectra
obtained at 20 mT were measured in a He bath cryostat (Oxford
MD 306) equipped with a pair of permanent magnets. The
spectra were analysed by least-squares fits using Lorentzian line
shapes.
from equivalents. The largest residual density was 0.388 e ÅϪ1
.
Refined parameters: 245, R1 = 0.0371, wR2 = 0.1047 (all data).
The structure was solved by direct methods (ref. 23) and refined
with all data by full-matrix-least-squares on F 2 using
SHELXL-97 (ref. 24c). Non-hydrogen atoms were refined
anisotropically. For the hydrogen atoms the riding model was
used. The solvent molecules are severely disordered. They were
refined with distance restraints and restraints for the aniso-
tropic displacement parameters to give the formula indicated
here.
NMR spectra were run on a Jeol JNM-GX-400 400 MHz
spectrometer, at room temperature in CDCl3. EI mass spectra
were run on a Varian MAT 311A (70 eV), and FAB mass spec-
tra on Micromass ZabSpec (Csϩ).
Electrochemistry and spectroelectrochemistry
Cyclic voltammetry at a platinum electrode was performed in
CH2Cl2–TBAH (0.1 M, TBAH = [n-Bu4N][PF6]) at room tem-
J. Chem. Soc., Dalton Trans., 2001, 3373–3383
3375

View Article Online
Table 1 Metal–metal distances (Å) in MIII2MIIO clusters
Mean
Ref.
[Fe3OL3], 2
3.267
3.282
3.287
3.308
3.306
12
a
[Fe2NiOL3], 4
a
[Fe2CuOL3], 5
[FeIII2FeIIO (O2CMe)6(py)3](py)
[FeIII2CoIIO(O2CMe)6(py)3](py)
26
26
a This work—see crystallographic data.
Crystal data for 5ؒ6CHCl3. C39H33Cl18CuFe2N33O7, M =
1889.32. Crystal dimensions: 0.3 × 0.3 × 0.35 mm, trigonal,
space group R3c (no. 161 of ref. 22): a = 23.314(3), c = 22.811(3)
Å, V = 10808(2) Å3, Z = 6. Diffractometer: Nonius Kappa CCD
area detector. Radiation: Mo-Kα, λ = 0.71073 Å, T = 173(2) K.
Measurement details: ω scan, range 1.75Њ ≤ θ ≤ 22.51Њ, section
of the reciprocal lattice h = Ϫ30 to 30, k = Ϫ30 to 26, l = Ϫ29 to
28. A total of 6804 reflections were measured, of which 2337
were considered independent, and 1356 were considered signifi-
cant as defined by I > 2(σ)I. Linear absorption coefficient: 1.436
mmϪ1, absorption correction: Scalepeak. Refined parameters:
132, R1 = 0.0629, wR2 = 0.1347 (all data). The structure was
solved by direct methods (ref. 23) and refined with all data by
full-matrix-least-squares on F 2 using SHELXL-97 (ref. 24c).
Non-hydrogen atoms were refined anisotropically. For the
hydrogen atoms the riding model was used. The solvent
molecules are severely disordered. They were refined with dis-
tance restraints and restraints for the anisotropic displacement
parameters to give the formula indicated here.
Fig. 2 IR spectra of 2–5 and (HNEt3)(FeIIIL2), 1000–400 cmϪ1
.
T = 298 K, KBr disc. (a) [Fe3OL3]ؒ6CHCl3; (b) [Fe2CoOL3]ؒxCHCl3;
(c) [Fe2NiOL3]ؒ1.2CHCl3ؒ1.8C5H12 “corrected” by subtraction of
a
spectrum of [Fe3OL3]ؒ6CHCl3; (d) [Fe2CuOL3]ؒ6CHCl3; (e)
(HNEt3)[FeIIIL2].
CCDC reference numbers 165605 and 153559.
See http://www.rsc.org/suppdata/dt/bl/b100780g/ for crystal-
lographic data in CIF or other electronic format.
Results
Structures
The molecular structure shown in Fig. 1 was established by
crystallography for the mixed-valence iron complex 212 and
mixed-metal iron–nickel and iron–copper complexes 4, 5;25 and
using the solid state IR spectra for the iron–cobalt complex 3
and the oxidised, tri-iron() complex 6. In solution, the
electrochemical data indicate facile electron transfer involving
the couples FeIII3/FeIII2FeII and FeIII2CoIII/FeIII2CoII, and also
FeIII2MII/FeIIIFeIIMII, (M = Fe, Co, Ni, Cu), and hence imply
similar structures for the oxidised and reduced forms of these
complexes, with little if any ligand rearrangement accompany-
ing the electron transfers.
The four neutral complexes crystallise with solvent molecules
depending on the conditions of preparation as noted above.
The molecular point group is D3; thus in each complex the three
metal ions are indistinguishable, and the central metal oxygen
units are planar by imposed symmetry. The complex molecules,
which are stacked on threefold axes in the crystal, are strongly
chiral, the central pyridine unit of each ligand being inclined to
the metal triangle plane (at 33.3Њ in 4), so that the tetrazolyl
donor units of each ligand are directed from opposite sides
of the plane. The overall size of the metal atom triangle as
measured by the mean metal–metal distance in the Fe2M units
does not vary significantly with different metals M, but it is
apparently affected by the ligand, being slightly less than in the
carboxylate-bridged complexes with the same central metal–
oxygen clusters (Table 1).
Fig. 3 IR spectra, 1000–600 cmϪ1. T = 298 K, KBr disc. (a) [Fe3OL3]ؒ
6CHCl3; (b) [Fe3O(L/CD3)3]ؒ6CHCl3; (c) H2L; (d) H2(L/CD3). T:
features assigned to the tetrazole units (including substituents).
P: features assigned to the pyridine rings and H atoms of the
dipicolinate residue.
Here we concentrate on the identification of metal–oxygen
stretch frequencies. Criteria for assignment include deuteration
shifts, IINS intensities, and comparisons with well-analysed
spectra of related molecules.27–31
In the chloroform solvate of 2, the strong band at 759 cmϪ1
contains intensity due to νas(CCl3) but quantitative compar-
isons with the CDCl3 analogue showed that the predominant
absorption is due to features other than chloroform. Also at low
temperature the two bands in the CHCl3 solvate were partly
Vibrational spectra—solid state
IR spectra of the mixed-valence complex 2 are compared with
spectra of other complexes in the range 1000–400 cmϪ1 in Fig. 2
and 3. Selected frequencies and assignments are listed in Table 2.
3376
J. Chem. Soc., Dalton Trans., 2001, 3373–3383

View Article Online
Table 2 Summary of vibrational frequencies (600–850 and 1350–1660 cmϪ1) and assignments for complexes [Fe2MOL3]
[Fe3OL3]
2
IINS
[Fe3OL3]
2
IR
[Fe2CoOL3]
3
IR
[Fe2NiOL3]
4
IR
[Fe2CuOL3]
5
IR
Assignmentsa
sh
sh
sh
692s
646br
672
680
νas(MO)
673
682
695
672
681
673
682
tetrazole ring
tetrazole ring
py ring ν26
νas(MO)
tetrazole ring
py ring γ(CH)
νas(MO)
697sh
703
698sh
703
742
740she
752
740sh
750s
747sh
752
≈760
785
805
837
859
≈755b
759c, d
784
805
839
≈760
759
783
805
838
861
759
783
805
840
863
νas(CCl3)
780m
811w
tetrazole ringf
850br m
857
872
py ring γ(CH)
CH3 rock?f, g
1363w
1450vs
1484
≈1350
1348
1446
≈1350
≈1445
≈1358
amide II
1445
1454??
CH3 asymmetric bendg
1602
1662
1602
≈1602
≈1603
py ring ν8a
1630br
1662
≈1655
1660
amide I
a py = pyridine ring of ligand H2L; bands numbered ν8a, ν26 are correlated with those of pyridine itself, in the scheme of ref. 28; amide = –CO–NϪ–
unit; bands numbered I, II, in the scheme of ref. 29. b Overlaid by νas(CCl3) but also strong when νas(CCl3) is shifted by deuteration—see discussion.
Presumably also coincides with γ(CH). c Shifts to ca. 735 cmϪ1 in the CDCl3 solvate. d Shoulder ca. 770 cmϪ1 at T = 80 K. e More pronounced at 742
cmϪ1, at T = 80 K. f Present in 1-methyl-5-aminotetrazole, not in dipicolinic acid. Assignment consistent with ref. 23. g Assigned on basis of CH3/CD3
shifts.
resolved. As noted above, the Fe2Ni complex 4 was contamin-
ated with 2. With appropriate subtraction it was possible to
eliminate features of 2 without introducing any obvious over-
subtractions, at least in the range 850–400 cmϪ1. The corrected
spectrum is shown in Fig. 2. The main effect of subtraction is to
resolve more clearly the pyridine mode at 752 cmϪ1, and the
broad shoulder at 760 cmϪ1. The Fe2Co and Fe2Cu complexes
were more pure, and the spectra shown in Fig. 2 are uncor-
rected. The sample of the oxidised complex [FeIII3OL3](NO3), 6,
was found to be more significantly mixed with 2, and it was not
possible to separate the pure spectrum on the basis of any
unique peaks.
In the range 500 to 700 cmϪ1, and likewise in the IINS
spectrum of 2, the strongest peaks were mostly assignable to
well-defined ligand vibrations. All complexes show a strong
peak at ca. 760 cmϪ1 which we assign as νas(Fe2MO). The
expected second component of this vibration has been assigned
in the mixed-metal complexes at 703 cmϪ1 (Fe2CoO, Fe2NiO)
and 646 cmϪ1 (Fe2CuO). In the mixed-valence complex 2 it
is not seen and the only likely region where it could be obscured
is 650 to 700 cmϪ1
.
IR of 2 in solution: spectroelectrochemistry
IR spectra for progressive oxidation and reduction of 2 in
solution are shown in Fig. 4. Both electrochemical processes
could be driven in either direction and yielded consistent
spectra, the cation [Fe3O]ϩ in particular being stable in solution,
even though the spectra confirmed that the solid prepar-
ations of 6 contained a significant fraction of 2. Metal–oxygen
stretches could not be monitored in solution owing to strong
solvent absorptions but the effect of redox state on the amide-I
band is clear. This band is broad in 2, but on oxidation it
Fig. 4 IR spectral changes during oxidation and reduction of
[Fe3OL3] in CH2Cl2, T = 298 K. Successive spectra were collected at
constant time intervals (15 s) and constant electrochemical potential:
(a) E = 250 mV vs. Fc/Fcϩ, (b) E = Ϫ850 mV vs. Fc/Fcϩ. The arrows
indicate the direction of the spectral changes.
sharpens and shifts to higher energy by almost 30 cmϪ1
,
Electrochemistry
whereas on reduction it broadens and shifts to lower energy by
about 20 cmϪ1. Further reduction produced changes which were
qualitatively similar, finally yielding two broad and poorly
All electrochemical data that can be ascribed to well-defined
electrode processes are summarised in Table 3. The expected
impurity signals in the voltammograms of 4 and 5 were easily
distinguished. The tri-iron complex 2 shows three distinct one-
electron processes (Fig. 6): a reversible oxidation at ϩ85 mV, a
resolved bands for the homovalent dianion [FeII3OL3]2Ϫ
.
Spectra of the four oxidation states from [FeIII3OL3]ϩ to
[FeII3OL3]2Ϫ are compared in Fig. 5.
J. Chem. Soc., Dalton Trans., 2001, 3373–3383
3377

View Article Online
Fig. 5 IR spectra of the four redox states of [Fe3OL3] recorded during
electrochemical oxidation/reduction in CH2Cl2, T = 298 K. [Fe3OL3]ϩ
(dashed line), [Fe3OL3] (solid line), [Fe3OL3]Ϫ (dotted line), [Fe3OL3]2Ϫ
(dotted-dashed line).
Fig. 7 1H NMR spectra (270 MHz, in CDCl3, T = 298 K). (a)
[Fe3OL3], 2; (b) [Fe2CuOL3], 5.
at Ϫ1425 and Ϫ1220 mV, respectively. These complexes are not
electroactive at moderate anodic potentials.
NMR spectra
Proton NMR spectra of the mixed-valence complex 2 and the
Fe2Cu complex 5 are compared in Fig. 7. Chemical shifts and
assignments are listed in Table 4. Apart from solvent peaks, the
spectrum of 2 shows the three sharp and well separated peaks
expected for protons of one set of equivalent CH3 groups and
one set of equivalent pyridine rings (H in meta and para
positions). Shifts and linewidths are similar to those reported
for complexes [Fe3O(OOCR)6(py)3]nϩ (n = 0, 1) in solution at
room temperature.10,39 The spectrum of 5 shows traces of 2 as
impurity, but otherwise, the proton peaks of 5 appear in slightly
different positions from 2, and each is composite. The band
profile for the meta protons of the pyridine residue clearly
shows an overlap of three peaks of similar intensity, two of
similar width and one narrower, though all three are consider-
ably broader than in 2. The profile for the methyl groups
is equally clear, again with one sharper and two broader
bands. These are the profiles expected for C2 symmetry in the
heterometallic complex.
Fig.
6
Cyclic voltammograms of [Fe3OL3], 2, (solid line) and
[Fe2CoOL3], 3, (dashed line) in CH2Cl2. T = 298 K, scan rate 50 mV sϪ1
.
Table 3 Half-wave potentials (E/mV, vs. Fc/Fcϩ) for complexes
[Fe2MOL3]a
M
Fe
Co
Ni
Cu
Notes
[Fe2MOL3]ϩ/0
[Fe2MOL3]0/Ϫ1
[Fe2MOL3]0/Ϫ2
85
Ϫ635
Ϫ1230
35
Ϫ660
Ϫ1310
b
b
c
c
d
Ϫ725
Ϫ1425
Ϫ550
Ϫ1220
a Cyclic voltammetry in CH2Cl2—0.1 M TBAH; scan rate 50 mV sϪ1
.
b Also very weak signal at 70 mV assigned to impurity [Fe3OL3].
c Reversible, ∆Ep = 60 10 mV. d Quasi-reversible, ∆Ep = 100 10 mV.
Electronic spectra
Complex 2 shows absorption bands over all the measured wave-
length region up to 2300 nm (Fig. 8). Additional weak and
broad, and thus unresolved, bands are present, some of them at
longer wavelengths, but the main feature, unique to the mixed-
valence complex 2, is the band at 1680 nm (6000 cmϪ1), clearly
demonstrated by the differences between successive spectra
during electrochemical oxidation, as well as by comparison
with the mixed-metal complexes 3–5 (Table 4). None of the
bands of 2 seems to be significantly solvatochromic when going
from CH2Cl2 to CH3CN. As regards the oxidised complex 6,
again the contamination of the prepared sample with 2 pre-
vented direct observation of the spectrum, but again it was
successfully measured by spectroelectrochemistry. Progressive
oxidation of 2 gave progressive changes in the spectrum, and
the reverse changes were seen on re-reduction. During the
oxidation the near-IR bands disappear completely, finally
yielding a spectrum with maxima at 258 and 480 nm (Fig. 8).
Reversible reduction to the iron(,,) mixed-valence anion
reversible reduction at Ϫ635 mV and a quasi-reversible reduc-
tion at Ϫ1230 mV with peak potential separation of 100 mV at
a scan rate of 50 mV sϪ1. The Fe2Co complex 3 has three similar
processes: reversible oxidation at ϩ35 mV, reversible and quasi-
reversible reduction at Ϫ660 and Ϫ1310 mV (Fig. 6). Assign-
ment of the first of these three as an electron transfer at the
cobalt centre is supported by comparisons of CoIII/II and FeIII/II
potentials for a series of mononuclear complexes. The differ-
ence varies strongly with ligand environment, in a sequence
from weak-field to strong-field ligands.32 As applied to com-
plexes 2 and 3 this difference measures the equilibrium:
FeIII2CoIII ϩ Fe3
FeIII2CoII ϩ FeIII
(1)
3
where only the unambiguous oxidation states are shown. The
Fe2Ni and Fe2Cu complexes 4 and 5 are reduced at Ϫ725 and
Ϫ550 mV, respectively, with a second quasi-reversible reduction
3378
J. Chem. Soc., Dalton Trans., 2001, 3373–3383

View Article Online
Table 4 Proton NMR chemical shifts (ppm) in oxo-centred trinuclear complexesa
Pyridine-H
ortho
Alkyl-H
CH3
meta
para
[Fe3OL3]
[Fe2CuOL3]
—
—
—
—
63.9
—
34.2
30.3
29.3
29.0
27.8
17
13.3
4
9.0
10.3
8.3
8.0
8.1
[Fe3O(OOCCMe3)6(py)3]ϩ b
[Fe3O(OOCCMe3)6(py)3]b
Obscured
8.8
4.4
a In CDCl3, T = 298 K. b Ref. 10.
Table 5 Absorption maxima (λ/nm) of complexes [Fe2MOL3]nϩ a
[Fe3OL3]
2
[Fe3OL3]ϩ
[Fe3OL3]Ϫ
[Fe2CoOL3]
3
[Fe2NiOL3]
4
[Fe2CuOL3]
5
2ϩ b
2Ϫ b
268, 425, 742,
900sh, 1680, 1900sh
258, 460sh, 480
269, 375sh, 585,
267, 428, 500sh,
580sh, 735sh
257sh, 266, 424,
560sh, 820, 1680d
260, 290sh, 330sh,
438, 675sh, 1680d
730sh, ≈1300c
a In CH2Cl2, T = 298 K. b In CH2Cl2–0.1 M TBAH, T = 290 K. c Weak and not well characterised. d Very weak signal assigned to impurity 2.
Fig. 8 Electronic spectra of [Fe3OL3] and [Fe3OL3]ϩ, in CH2Cl2. T =
290 K. The arrows indicate the direction of spectral changes during
the electrochemical oxidation of [Fe3OL3]. Spectra were collected at
the following potentials: Ϫ60, Ϫ20, 20, 60, 100, 140 mV vs. Fc/Fcϩ.
Inset: A more detailed view of the NIR bands of [Fe3OL3], in CDCl3,
T = 298 K.
Fig. 9 Mössbauer spectra of [Fe3OL3], 2, in the solid state. (a) T = 4.2
K, (b) T = 332 K. Field 20 mT perpendicular to the γ-beam. Broken
lines are Lorentzians fitted for four quadrupole doublets.
[Fe3OL3]Ϫ is also accompanied by almost complete disappear-
ance of NIR bands, as well as a strong shift of UV/VIS bands.
The final spectrum consists of narrow bands at 269 and 585 nm
along with broad signals in the same region and probably a very
weak absorption at 1300 nm. Further reduction, to [Fe3OL3]2Ϫ
,
led to a further decrease of absorption in the NIR region, but
the limited stability of the fully reduced species did not allow
complete spectral characterisation. Table 5 shows the spectra of
all the observed redox states of complex 2.
Mössbauer spectra
Observed and fitted Mössbauer spectra of 2 are shown in Fig. 9
and 10; Fig. 9 giving representative spectra of the solid at three
temperatures, Fig. 10 giving spectra of frozen solutions in THF.
Isomer shifts and quadrupole splittings of all complexes are
listed in Table 6. The complexes expected to contain only
trivalent iron, i.e. mixed-metal 3, 4 and 5 and fully oxidised 6,
all have similar values of δ, in the range 0.44 to 0.47 mm
s
Ϫ1. Analyses of numbers and widths of peaks in all the
other spectra were based on extensive trials of alternative
schemes, subject only to the usual constraint that all spectra are
expected to consist of overlays of doublets due to quadrupole
splitting. The relatively broad line widths of [FeIII2CoIIOL3],
[FeIII2CuIIOL3] and [FeIII3OL3] are probably due to the onset of
magnetic broadening at low temperature. In the spectra of the
Fig. 10 Mössbauer spectra of [Fe3OL3], 2, in frozen THF. (a) T = 4.2
K and (b) T = 160 K with a field of 20 mT perpendicular to the γ-beam.
Broken lines are Lorentzians fitted for three quadrupole doublets.
J. Chem. Soc., Dalton Trans., 2001, 3373–3383
3379

View Article Online
Table 6 Mössbauer parameters of complexes [Fe2MOL3], 2–5, and [Fe3OL3](NO3), 6
T /K
δ/mm sϪ1 a
∆EQ/mm sϪ1 a
Γ/mm sϪ1 b
Relative contribution (%)
Iron() complexes
[Fe2CoOL3], 3
4.2
0.46
0.46
0.47
0.44
1.79
1.75
1.53
1.10
0.60
0.38
0.90
0.66
100
100
100
100
[Fe2NiOL3], 4
4.2
4.2
4.2
[Fe2CuOL3], 5
[FeIII3OL3](NO3), 6
[Fe3OL3], 2, solid state
I
4.2
0.52
0.99
0.58
0.80
1.85
2.65
2.00
2.27
0.34
0.40
0.40
0.40
60
30
7
II
III
IV
3
I
II
III
IV
332
0.37
0.79
0.47
0.59
1.72
1.88
1.69
1.62
0.24
0.34
0.34
0.34
30
15
37
18
[Fe3OL3], 2, THF solution
I
4.2
0.55
0.91
0.66
1.83
2.50
2.11
0.36
0.47
0.28
60
30
10
II
M
I
II
M
160
0.50
0.84
0.62
1.73
2.39
1.87
0.27
0.36
0.40
18
9
73
a Standard deviation 0.01 mm sϪ1 b Standard deviation 0.01–0.04 mm sϪ1, depending on value of Γ.
.
Hiex = Ϫ2 Jab SaؒSb Ϫ 2 J (SaؒSc ϩ SbؒSc)
(2)
mixed-valence complex 2, generally four doublets were required
as discussed below. In the frozen solution of 2 in THF, only
three doublets were required, one of which, increasing in
strength with increasing temperature, has δ = 0.66 mm sϪ1 at 4.2
K and δ = 0.62 mm sϪ1 at 160 K. The change in δ with temper-
ature is attributed to the second-order Doppler effect; estimates
where Sa = Sb = 5/2 and Sc = 2 are the spins at the Fe3ϩ and
Fe2ϩ sites, at temperatures 4.2 to 295 K gave a good fit with Jab
Ϫ70 cmϪ1, J = Ϫ24.5 cmϪ1. For the complex [Fe3O(OOCCMe3)-
=
(py)3](py) neutron scattering spectra have been fitted with Jab
=
of δSOD are in the range 0.05 to 0.1 mm sϪ1 40
and a value of
,
Ϫ60 cmϪ1, J = Ϫ18 cmϪ1, though some evidence of mixing of
energy levels was also presented.43 The more negative values for
the present complex 2 are consistent with the shorter inter-
atomic distances; and the difference (Jab Ϫ J), being similar in
both cases, implies that the deviations from threefold symmetry,
though concealed, are similar in both clusters.
0.04 mm sϪ1, gives our experimental value δ = 0.62 mm sϪ1 at
160 K.
Discussion
As noted above, the crystal structures of at least two of
the mixed-metal complexes show threefold symmetry, one of
them being the FeIII2CuII complex, in which presumably the
copper() centres are subject to Jahn–Teller distortion. This
emphasises the ability of the ligand to conceal asymmetry by
rotational disorder. It is not surprising therefore that the mixed-
valence compound 2 also shows apparent threefold symmetry,
regardless of whether this is due to disorder or valence
delocalisation.
The IR spectra show conclusively that the central metal ion
triangle in the mixed-valence complex [Fe3OL3], 2 does not have
threefold symmetry on the vibrational time scale, either in the
solid state (from 80 K upwards) or in solution at room temper-
ature. This is most clearly demonstrated by the spectroelectro-
chemistry of 2 in solution, showing the contrast of the sharp
amide-I band in the oxidised complex [Fe3OL3]ϩ with that of
the mixed-valence complex. These observations indicate
inequivalence of the iron sites, i.e. at least some degree of
valence localisation, but they do not indicate the degree of
localisation, nor do they exclude dynamic delocalisation, except
to say that this must be slower than the time scale of vibrational
spectroscopy. From the solution NMR spectra however it is
equally clear that dynamic delocalisation, i.e. intramolecular
electron transfer, does take place, in solution, at a rate faster
than the NMR time scale, at room temperature.
Returning to the IR spectra, it may be significant that only
one component of the asymmetric metal cluster vibration
νas(Fe3O) is seen in the solid state spectra, but the complexity of
the spectra in the relevant region leaves open the possibility that
the lower-frequency component is present. What can be said,
from the data summarised in Table 7, is that any splitting
between the two components of νas(Fe3O) must be considerably
less than in the carboxylate-bridged complexes. The same
is true for the mixed-metal Fe2M complexes and it may be a
consequence of the different ligand environments rather than
the electronic character of the mixed-valence cluster.
It is not possible to estimate distances in [FeIII2FeIIOL3],
i.e. the hypothetical valence-localised form of complex 2, but
it is possible to make the corresponding estimate in two
complexes of the carboxylate series. The mixed-valence com-
plex [FeIII2FeII(OOCMe)6(3-Etpy)3]ؒ0.5C6H5CH3 apparently
shows complete valence localisation,41 and in [CrIII2CoIIO-
(OOCCMe3)6(py)3]ؒ(py), the Cr–O and Co–O distances were
distinguished by EXAFS, though rotationally disordered in
the crystal.42 The distortions of the metal-ion triangle from
equilateral to isosceles are similar, though differently expressed.
In the CrIII2CoII complex, by describing the triangle as
approximately equilateral, the displacement of the oxygen atom
from the centre could be estimated as 0.05 Å; in the FeIII2FeII
complex the metal–metal distances could be measured
independently and the corresponding displacement is about
0.10 Å.
Temperature-dependent magnetisation measurements sup-
port the assignment of the component oxidation states in 2 with
antiferromagnetic coupling.13 Using the HDVV spin-only
model with the Hamiltonian:
From the electrochemical properties of the complexes two
comproportionation constants can be obtained as follows:
3380
J. Chem. Soc., Dalton Trans., 2001, 3373–3383

View Article Online
Table 7 Frequencies (cmϪ1) of components of νas(Fe2MO) in oxo-centred trinuclear complexes
νas/cmϪ1
Ratio
Ref.
[Fe3OL3]ؒ6CHCl3, 2
[Fe3O(OOCMe)6(py)3](py)
[Fe3O(OOCCMe3)6(py)3]
760
695
705
700/670
570
580
1.09/1.13
1.22
1.22
This work
44
9, 39
[Fe2NiOL3]ؒ6CHCl3, 4
[Fe2NiO(OOCMe)6(py)3](py)
[Fe2NiO(OOCCMe3)6(py)3]
760
722
721
705
565
595
1.08
1.28
1.21
This work
44
45
Fe
[FeIII3OL3]ϩ ϩ [Fe3OL3]Ϫ
[FeIII2CoIIIOL3]ϩ ϩ [Fe2CoIIOL3]Ϫ
2 [Fe3OL3], Kcom
(3)
Co
2 [FeIII2CoIIOL3], Kcom
(4)
Fe
Co
with Kcom = 2 × 1012, Kcom = 7 × 1011. Here only the
unambiguous oxidation states of iron are indicated. Both
constants are large but in view of the different ionic charges
they are likely to be dominated by environmental effects,
especially solvation.46 Iron(,) mixed-valence species are
involved in both reactions. If iron(/) delocalisation is
Fe
Co
important, it will tend to increase Kcom and decrease Kcom
,
relative to the values expected for full localisation.
Intervalence charge-transfer is a well-known test of valence
delocalisation. In the present series of complexes a band at 6000
cmϪ1 is unique to the mixed-valence complex 2. We interpret
this cautiously since it is partly overlapped, and also because of
the other weaker absorptions observed at lower energies. Never-
theless it is clear that the 6000 cmϪ1 band is much narrower than
calculated for a localised intervalence transition (the Hush
formula47 gives the full width at half height, 3700 cmϪ1, com-
pared with ca. 2000 cmϪ1, observed) and no solvatochromism
was detected. This indicates significant static valence delocalis-
ation, though not necessarily the full extent of delocalisation
defined as Robin–Day class III:48 recent arguments by Lambert
and Nöll indicate that a class III/II borderline state is consistent
with the spectrum.3g (The fact that the further reduced complex
[Fe3OL3]Ϫ, assigned as iron(,,), shows no low-energy IT
band probably indicates some structural modification).
The most direct indications of the mixed-valence character
of the tri-iron cluster are the Mössbauer spectra, and here the
results are in complete contrast with earlier work on the
carboxylate-bridged series.13 Over the temperature range 4.2 to
332 K, all solid state spectra of 2 were adequately fitted with
two pairs of Lorentzian lines, consistent with two sets of mixed-
valence molecules coexisting. A linear interpolation of isomer
shifts between those determined for pure iron() and iron() in
comparable ligand environments49 gave fractional oxidation
states for the two components, i.e. Fe(2.9ϩ, 2.9ϩ, 2.2ϩ) and
Fe(2.75ϩ, 2.75ϩ, 2.5ϩ) the proportion of the latter increasing
with increasing temperature. Band shapes gave no sign of
temperature-dependent lifetime broadening. For each of the
two components, this is consistent with either of the two limit-
ing situations described above, i.e. static or dynamic, where
dynamic in this case means faster than the Mössbauer time
scale. The static option was preferred for both components,
leading to the paradox of a system which shows some valence
delocalisation at 4.2 K, yet does not achieve complete delocal-
isation at room temperature. It was argued that by a suitable
blend of three known mechanisms of electronic interaction, the
paradox could be resolved. Starting as above from the assign-
ment of high-spin Fe(3ϩ, 3ϩ, 2ϩ) for the basis electronic state,
the spin-only model (eqn. (2)) gives a ground state (S, Sab) =
(1,1) for the trinuclear cluster, and states (2,0), (2,1), (0,2)
located at energies 7, 100 and 130 cmϪ1 above the ground state
(Fig. 11). Double exchange leads to a splitting of each of the
spin-only states into A and E components, and it was shown
that for the second and third excited states, the splitting is both
Fig. 11 Resonance-exchange levels of [Fe3OL3], 2, for the case of weak
double-exchange interaction.
stronger and less sensitive to symmetry distortions, than it is
for the ground and first excited states. It is this difference that
justifies the use of the fully localised valence model to treat
the magnetisation data, while double-exchange is still strong
enough to cause significant partial valence delocalisation,
affecting the spectra.
The almost complete localisation of the extra electron in the
ground state is of course explained by vibronic interaction,
leading to inequality of bond lengths around the iron sites. The
resulting dynamic problem has been solved for trimeric systems,
but so far only in the d1–d1–d0 and d1–d1–d2 cases.51 A full
treatment for the present case of d5–d5–d6 will be given else-
where. Here it is sufficient to mention the results obtained in the
limit of strong vibronic coupling, in which the electron transfer
is introduced as a perturbation.52 In the vibrational ground
state (n = 0) the situation is the same as in Fig. 11. In the excited
vibrational states, however, the resonance splitting is consider-
ably larger by a factor of [(λ2/4πhν3) Ϫ 1] where λ is the vibronic
coupling parameter, so that λ2/4πhν3 is the number of vibra-
tional quanta necessary for intervalence charge transfer. The
relevant vibrational level is presumably around 760 cmϪ1—our
rough estimate of νas(Fe3O)—or less if other framework vibra-
tions dominate the reorganisation process. Consequently, the
electron delocalisation in the first vibrational state (n = 1) is
even more pronounced than that described above for the
exchange states at 100 and 130 cmϪ1. These states become popu-
lated at higher temperatures, and on this basis the Mössbauer
data are interpreted in terms of increasing contributions from
static delocalisation, as distinct from lifetime broadening due to
electron hopping.
The Mössbauer results for complex 2 in frozen THF
solution, reported in this work, are different again. Only three
J. Chem. Soc., Dalton Trans., 2001, 3373–3383
3381

View Article Online
Table 8 DFT calculated energies and selected atomic charges of
[Fe3OL3], 2
of vibrational and Mössbauer spectroscopies. In terms of the
Robin–Day classification, the mixed-valence iron complex is
evidently at the borderline between classes II and III.
Atomic charges in the Fe3O coreb, c
Spin state Total energya Fe1
Fe2
Fe3
O
Acknowledgements
This work was supported by Deutsche Forschungsgemeinschaft
and the Volkswagen-Stiftung. V. C. acknowledges the support
of the Alexander von Humboldt Foundation, and T. I. the
support of the Japan Society for the Promotion of Science.
R. D. C. holds an Emeritus Leverhulme Fellowship.
S = 0
S = 1
S = 2
Ϫ3740.522
Ϫ3740.629
Ϫ3740.643
ϩ0.682
ϩ0.759
ϩ0.777
ϩ0.686
ϩ0.769
ϩ0.760
ϩ0.686
ϩ0.769
ϩ0.760
Ϫ0.609
Ϫ0.779
Ϫ0.767
a E(UB ϩ HF Ϫ LYP) in Hartree units. b Mulliken atomic charges.
c Compare an average value of 0.767, obtained for the iron centres in
the monovalent cation (S = 1/2) with formal oxidation state ϩ3.
References and notes
doublets were required to fit the spectra, two of them consistent
with partially localised valence states similar to those in the
low-temperature component of the solid, i.e. Fe(2.85ϩ, 2.85ϩ,
2.3ϩ), the other consistent with full delocalisation Fe(2.67ϩ,
2.67ϩ, 2.67ϩ): δ = 0.66 mm sϪ1 at 4.2 K and δ = 0.62 mm sϪ1 at
160 K. The apparently delocalised component accounts for
70% of the molecules at 160 K. Again the data do not dis-
tinguish between static and dynamic situations.
1 Recent reviews include: (a) C. C. Moser, C. C. Page, R. Farid and
P. L. Dutton, J. Bioenerg. Biomembr., 1995, 27, 263; (b) P. F. Barbara,
T. J. Meyer and M. A. Ratner, J. Phys. Chem., 1996, 100, 13148; (c)
H. B. Gray and J. R. Winkler, Annu. Rev. Biochem., 1996, 65, 537;
(d ) H. B. Gray and J. R. Winkler, Advances in Chemical Physics,
Volume 106, Electron Transfer–From Isolated Molecules to
Biomolecules, Part One, ed. M. Bixon and J. Jortner, Wiley, New
York, 1999, pp. 35–202.
2 (a) R. D. Cannon, Electron Transfer Reactions, Butterworth,
London, 1980; (b) R. D. Cannon, Electron Transfer in Inorganic,
Organic and Biological Systems: (Adv. Chem. Ser., 228), ed. J. R.
Bolton, N. Mataga and G. McLendon, ACS, Washington, DC,
1991; (c) A. M. Kuznetsov, Charge Transfer Processes in Physics,
Chemistry and Biology, Gordon & Breach, New York, 1995.
3 (a) G. A. Neyhart, C. J. Timpson, W. D. Bates and T. J. Meyer, J. Am.
Chem. Soc., 1996, 118, 3730; (b) T. Ito, T. Hamaguchi, H. Nagino,
T. Yamaguchi, J. Washington and C. Kubiak, Science, 1997, 277,
660; (c) K. D. Demadis, G. A. Neyhart, E. M. Kober and T. J. Meyer,
J. Am. Chem. Soc., 1998, 120, 7121; (d ) C. E. B. Evans, M. L.
Naklicki, A. R. Rezvani, C. A. White, V. V. Kondratiev and R. J.
Crutchley, J. Am. Chem. Soc., 1998, 120, 13096; (e) K. D. Demadis,
E.-S. El-Samanody, G. M. Coia and T. J. Meyer, J. Am. Chem. Soc.,
1999, 121, 535; ( f ) T. Ito, T. Hamaguchi, H. Nagino, T. Yamaguchi,
H. Kido, I. S. Zavarine, T. Richmond, J. Washington and C. Kubiak,
J. Am. Chem. Soc., 1999, 121, 4625; (g) C. Lambert and G. Nöll,
J. Am. Chem. Soc., 1999, 121, 8434; (h) M. Ketterle, W. Kaim, J. A.
Olabe, A. R. Parise and J. Fiedler, Inorg. Chim. Acta, 1999, 291, 66.
4 R. D. Cannon and R. P. White, Prog. Inorg. Chem., 1988, 36,
195.
In the oxo-centred carboxylate series, temperature-dependent
Mössbauer spectra have been reported for many of the mixed-
valence iron(/) complexes.53,54 Only rarely have gradual
transitions been observed. For example in the solid complex
[Fe3O(OOCCH3)6(3-Etpy)3]ؒCH3CCl3, δ increases from 0.54
mm sϪ1 at 43 K to 0.60 mm sϪ1 at 100 K, and to 0.65 mm sϪ1 at
293 K, at which temperature the iron atoms are spectro-
scopically indistinguishable.55 In most cases however, sudden
transitions are observed from full valence localisation at low
temperature, to spectra consistent with full delocalisation over
three centres (or sometimes over only two centres, Fe(3ϩ, 2.5ϩ,
2.5ϩ)) at higher temperature. The changes are then interpreted
in terms of valence de-trapping, and absence of line broadening
is explained by assuming that the internal electron transfer rate
passes through the Mössbauer time scale rather rapidly, as the
temperature is increased, in phase transitions controlled by
cooperative effects in the solid state. In the present systems the
outstanding feature of the Mössbauer data, in both solid and
frozen solution states, is the evidence for some degree of valence
delocalisation at the lowest temperatures.
Our discussion of ground and excited states is in reasonable
agreement with the results of the DFT calculations. As shown
in Table 8, the triplet and quintet states are significantly lower
than the singlet, and this pattern qualitatively reflects the
temperature-enhanced electron delocalisation discussed above.
It is also of interest that the valence states appear more different
in the triplet and quintet states than in the singlet state. The
actual differences are small but the calculations were based on
an almost symmetrical molecular model.
5 (a) S. J. Lippard, Angew. Chem., 1988, 100, 353; (b) S. J. Lippard,
Angew. Chem., Int. Ed. Engl., 1988, 27, 344.
6 (a) T. Nakamoto, M. Hanaya, M. Katada, K. Endo, S. Kitagawa
and H. Sano, Inorg. Chem., 1997, 36, 4347; (b) C.-C. Wu, S. A. Hunt,
P. K. Gantzel, P. Guetlich and D. N. Hendrickson, Inorg. Chem.,
1997, 36, 4717; (c) R. Wu, M. Poyraz, F. E. Sowrey, C. E. Anson,
S. Wocadlo, A. K. Powell, U. A. Jayasooriya, R. D. Cannon,
T. Nakamoto, M. Katada and H. Sano, Inorg. Chem., 1998, 37,
1913.
7 Y. Kaneko, M. Nakano, M. Sorai, H. G. Jang and D. N.
Hendrickson, Inorg. Chem., 1989, 28, 1067 and references therein.
8 R. D. Cannon, U. A. Jayasooriya, S. K. Arap Koske, R. P. White and
J. H. Williams, J. Am. Chem. Soc., 1991, 113, 4158.
9 R. Wu, S. K. Arap Koske, R. P. White, C. E. Anson, U. A.
Jayasooriya and R. D. Cannon, J. Chem. Soc., Chem. Commun.,
1994, 1657.
10 (a) F. E. Sowrey, C. J. MacDonald and R. D. Cannon, J. Chem. Soc.,
Faraday Trans., 1998, 94, 1571; (b) F. E. Sowrey, Ph.D. Thesis,
University of East Anglia, Norwich, 1997.
11 (a) R. Manchanda, Inorg. Chim. Acta, 1996, 245, 91; (b) B. P.
Straughan, O. M. Lam and A. Earnshaw, J. Chem. Soc., Dalton
Trans., 1987, 97; (c) G. Losada, M. A. Mendiola and M. T. Sevilla,
Inorg. Chim. Acta, 1997, 255, 125; (d ) A. M. Bond, R. J. H. Clark,
D. G. Humphrey, P. Panayiotopolous, B. W. Skelton and A. H.
White, J. Chem. Soc., Dalton Trans., 1998, 1845; (e)
E. Constantinescu, H. Storch, C. Turta and V. Meriacre, Rev. Roum.
Chim., 1998, 43, 799.
12 (a) R. W. Saalfrank, S. Trummer, H. Krautscheid, V. Schünemann,
A. X. Trautwein, S. Hien, C. Stadler and J. Daub, Angew. Chem.,
1996, 108, 2350; (b) R. W. Saalfrank, S. Trummer, H. Krautscheid,
V. Schünemann, A. X. Trautwein, S. Hien, C. Stadler and J. Daub,
Angew. Chem., Int. Ed. Engl., 1996, 35, 2206.
Conclusion
All indications are that the novel ligand system provides a rigid
framework that substantially affects the electronic properties of
the mixed-valence Fe3O core. It also allows reversible formation
of a series of one-electron redox states but still permits the
fine-tuning of the electronic structure by environmental effects.
In the solid state the mixing of valence states persists down to
temperatures as low as 4.2 K, which has not been observed in
any previous systems with this core structure. In both solid and
solution states however, the valences of the three iron atoms
remain inequivalent even up to room temperature. This means
that at any instant in time, valence electron density is unequally
distributed over the three atoms. The extent of this inequality,
and whether it tends towards Fe(3ϩ, 3ϩ, 2ϩ) or Fe(3ϩ, 2.5ϩ,
2.5ϩ), are not specified, but in either case it is clear that the
distribution rotates by an internal electron transfer process at a
rate which is intermediate between the characteristic timescales
13 V. Coropceanu, V. Schünemann, C. Ober, M. Gerdan, A. X.
Trautwein, J. Köhler and R. W. Saalfrank, Inorg. Chim. Acta, 2000,
300–302, 875.
14 R. A. Henry and W. G. Finnegan, J. Am. Chem. Soc., 1954, 76, 923.
3382
J. Chem. Soc., Dalton Trans., 2001, 3373–3383

View Article Online
35 N. Tanaka, T. Tomita and A. Yamada, Bull. Chem. Soc. Jpn., 1970,
15 A. G. Oertli, W. R. Meyer, U. W. Suter, F. B. Joho, V. Gramlich and
W. Petter, Helv. Chim. Acta, 1992, 75, 184.
43, 2042.
16 J. Salbeck, Ph.D. Thesis, Universität Regensburg, 1988, p. 196.
17 J. Salbeck, Anal. Chem., 1993, 65, 2169.
18 M. Büschel, C. Stadler, C. Lambert, M. Beck and J. Daub,
J. Electroanal. Chem., 2000, 484, 24.
19 (a) A. D. Becke, Phys. Rev. A, 1988, 38, 3098; (b) A. D. Becke,
J. Chem. Phys., 1993, 98, 1372; (c) A. D. Becke, J. Chem. Phys., 1993,
98, 5648.
20 Gaussian-98, Revision A.7, M. J. Frisch, G. W. Trucks, H. B.
Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G.
Zakrzewski, J. A. Montgomery, Jr., R. E. Stratmann, J. C. Burant,
S. Dapprich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C.
Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi,
B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski, G. A.
Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, D. K. Malick, A. D.
Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz,
A. G. Baboul, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz,
I. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A.
Al-Laham, C. Y. Peng, A. Nanayakkara, C. Gonzalez,
M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong,
J. L. Andres, C. Gonzalez, M. Head-Gordon, E. S. Replogle and
J. A. Pople, Gaussian, Inc., Pittsburgh PA, 1998.
36 G. Schwarzenbach and J. Heller, Helv. Chim. Acta, 1951, 34, 576.
37 K. J. Vetter and N. Jäger, Electrochim. Acta, 1966, 11, 401.
38 (a) U. Koelle and F. Khouzami, Angew. Chem., 1980, 92, 658; (b)
U. Koelle and F. Khouzami, Angew. Chem., Int. Ed. Engl., 1980, 19,
640.
39 R. P. White, L. M. Wilson, D. J. Williamson, G. R. Moore, U. A.
Jayasooriya and R. D. Cannon, Spectrochim. Acta, Part A, 1990, 46,
917.
40 P. Gütlich, R. Link and A. X. Trautwein, Mössbauer Spectroscopy
and Transition Metal Chemistry, Springer-Verlag, Berlin, 1978.
41 C.-C. Wu, H. G. Jang, A. L. Rheingold, P. Gütlich and D. N.
Hendrickson, Inorg. Chem., 1996, 35, 4137.
42 A. B. Edwards, J. M. Charnock, C. D. Garner and A. B. Blake,
J. Chem. Soc., Dalton Trans., 1995, 2515.
43 R. D. Cannon, U. A. Jayasooriya, R. P. White and S. K. Arap Koske,
Spectrochim. Acta, Part A, 1993, 49, 1787.
44 (a) L. Meesuk, U. A. Jayasooriya and R. D. Cannon, J. Am. Chem.
Soc., 1987, 109, 2009; (b) R. W. Wu, U. A. Jayasooriya and R. D.
Cannon, Spectrochim. Acta, Part A, 2000, 56, 575.
45 R. P. White, Ph.D. Thesis, University of East Anglia, Norwich,
1991, p. 126.
21 P. J. Hay and W. R. Wadt, J. Chem. Phys., 1988, 82, 299.
22 D. T. Cromer and J. T. Waber, in International Tables for X-Ray
Crystallography, ed. J. A. Ibers and W. C. Hamilton, Kynoch Press,
Birmingham, England, 1974.
23 G. M. Sheldrick, Acta Crystallogr., Sect. A, 1990, 46, 467.
24 (a) G. M. Sheldrick, SHELXL-93, Program for refinement of crystal
structures, University of Göttingen, 1993; (b) SHELXL-93: G. M.
Sheldrick, in Crystallographic Computing 3, ed. G. M. Sheldrick,
C. Krüger and R. Goddard, Oxford University, 1993, pp. 175–189;
(c) SHELXL-97: G. M. Sheldrick, University of Göttingen, 1997.
25 Crystal structure determinations were carried out by Frank Hampel
(Erlangen) and by Regine Herbst-Irmer and Cornelia Goebel,
(Institut für Anorganische Chemie, Universität Göttingen,
Tammanstrasse 4, D-37077, Göttingen, Germany).
26 H. G. Jang, K. Kaji, M. Sorai, R. J. Wittebort, S. J. Geib, A. L.
Rheingold and D. N. Hendrickson, Inorg. Chem., 1990, 29, 3547.
27 (a) K. Nakamoto, Infrared and Raman Spectra of Inorganic and
Coordination Compounds, 4th edn., John Wiley, New York, 1986; (b)
V. Galasso, G. de Alti and G. Costa, Spectrochim. Acta, 1965, 21,
660.
28 D. A. Long and E. L. Thomas, Trans. Faraday Soc., 1963, 59, 783.
29 T. Miyazawa, T. Shimanouchi and S.-I. Mizushima, J. Chem. Phys.,
1958, 29, 611.
46 For example a µ-oxo-bridged diiron(,) complex has been
characterised as valence-localised, even though Kcom is similar to the
values reported here: A. S. Borovik, V. Papaefthymiou, L. F. Taylor,
O. P. Anderson and L. Que, Jr., J. Am. Chem. Soc., 1989, 111, 6183.
47 (a) N. S. Hush, Prog. Inorg. Chem., 1967, 8, 391; (b) N. S. Hush,
Coord. Chem. Rev., 1985, 64, 135.
48 M. B. Robin and P. Day, Adv. Inorg. Chem. Radiochem., 1967, 30,
247.
49 For δ[Fe3ϩ] we used the average of the values observed for the mixed-
metal complexes [Fe2MOL3] and the fully oxidised [Fe3OL3]ϩ, i.e.
0.46 mm sϪ1, and for δ[Fe2ϩ] the typical50 value 1.1 mm sϪ1
.
50 D. P. E. Dickson and F. J. Berry, Mössbauer Spectroscopy,
Cambridge University Press, Cambridge, 1968, p. 88.
51 S. I. Boldyrev, V. Y. Gamurar, B. S. Tsukerblat and A. V. Palii, Mol.
Phys., 1994, 81, 621.
52 R. L. Fulton and M. Gouterman, J. Chem. Phys., 1961, 35,
1059.
53 (a) S. M. Oh, D. N. Hendrickson, K. L. Hassett and R. E. David,
J. Am. Chem. Soc., 1984, 106, 7984; (b) S. M. Oh, T. Kambara, D. N.
Hendrickson, M. Sorai, K. Kaji, S. E. Woehler and R. J. Wittebort,
J. Am. Chem. Soc., 1985, 107, 5540; (c) T. Kambara, D. N.
Hendrickson, M. Sorai and S. M. Oh, J. Chem. Phys., 1986, 85,
2895; (d ) M. Sorai, Y. Shiomi and D. N. Hendrickson, Inorg. Chem.,
1987, 26, 223; (e) J. K. McCusker, H. G. Jang, M. Zvagulis, W. Ley,
H. Drickamer and D. N. Hendrickson, Inorg. Chem., 1991, 30, 1985.
54 (a) T. Nakamoto, M. Katada and H. Sano, Chem. Lett., 1990, 225;
(b) T. Nakamoto, M. Katada and H. Sano, Chem. Lett., 1991, 1323;
(c) T. Nakamoto, M. Katada and H. Sano, Hyperfine Interact., 1991,
68, 233; (d ) M. Yoshida, T. Nakamoto, S. Kawata, M. Katada and
H. Sano, Hyperfine Interact., 1994, 84, 583; (e) M. Yoshida,
T. Nakamoto, S. Kawata, M. Katada and H. Sano, Hyperfine
Interact., 1994, 84, 589; ( f ) T. Nakamoto, M. Katada, S. Kawata,
S. Kitagawa, H. Sano and M. Konno, Hyperfine Interact., 1994, 93,
1567; (g) T. Sato, A. Fumitoshi, K. Endo, M. Katada, T. Nakamoto
and H. Sano, J. Am. Chem. Soc., 1996, 118, 3450; (h) T. Nakamoto,
M. Katada and H. Sano, Inorg. Chim. Acta, 1999, 291, 127.
55 G. Wu, Y. Zhang, L. Ribaud, P. Coppens, C. Wilson, B. B. Iversen
and F. K. Larsen, Inorg. Chem., 1998, 37, 6078.
30 A. Bigotto and R. Klingendrath, Spectrochim. Acta, Part A, 1990,
46, 1683.
31 P. Carmona, Spectrochim. Acta, Part A, 1980, 36, 705.
32 The following is a list of differences in standard reduction potential
of the two isostructural couples, ∆E = E(CoIII/II) Ϫ E(FeIII/II),
measured in the same medium at the same temperature—usually
293–298 K. Each consists of the formula of the complex, MLn (M =
Co, Fe), the value of ∆E in V, and the literature reference.
[M(OH2)6]3ϩ/2, 1.13,33 [M(C2O4)3]3Ϫ/4, 0.56;33 [M(acac)3]0/Ϫ, 0.29;34
[M(edta)]Ϫ/2Ϫ, 0.28,35 (Fe),36 (Co); [FeIII2MOL3]0/Ϫ, Ϫ0.05, this work;
[CH2(CH2NCHC6H4O-o)2M]ϩ/0b, Ϫ0.20;37 [M(phen)3]3ϩ/2ϩ, Ϫ0.73;33
[M(C5H5)2]ϩ/0, Ϫ1.30;37 [M(C5Me5)2]ϩ/0, Ϫ1.43.38
33 Standard Potentials in Aqueous Solution, ed. A. J. Bard, R. Parsons
and J. Jordan, IUPAC, Marcel Dekker, New York, 1985.
34 O. Sock, P. Lemoine and M. Gross, Electrochim. Acta, 1981, 26, 99.
J. Chem. Soc., Dalton Trans., 2001, 3373–3383
3383