New family of ferric spin clusters incorporating redox-active ortho-dioxolene ligands

Article abstract of DOI:10.1021/ic900729d

Seven new di-, tri-, tetra-, and hexanuclear iron complexes that incorporate a polydentate Schiff base and variously substituted catecholate ligands have been synthesized from the trinuclear precursor [Fe ₃(OAc)₃(L)₃] (1),

Full text of DOI:10.1021/ic900729d

Inorg. Chem. 2009, 48, 7765–7781 7765
DOI: 10.1021/ic900729d
New Family of Ferric Spin Clusters Incorporating Redox-Active ortho-Dioxolene
Ligands
Yanyan Mulyana,† Ayman Nafady,‡ Arindam Mukherjee,† Roland Bircher,§ Boujemaa Moubaraki,‡
Keith S. Murray,‡ Alan M. Bond,‡ Brendan F. Abrahams,† and Colette Boskovic*,†
†School of Chemistry, University of Melbourne, Victoria, 3010, Australia, ‡School of Chemistry, Monash
University, Clayton, Victoria, 3800, Australia, and §Bragg Institute, The Australian Nuclear Science and
Technology Organisation, PMB 1, Menai, New South Wales 2234, Australia
Received April 14, 2009
Seven new di-, tri-, tetra-, and hexanuclear iron complexes that incorporate a polydentate Schiff base and variously
substituted catecholate ligands have been synthesized from the trinuclear precursor [Fe3(OAc)3(L)3] (1), where LH2 =
2-[[(2-hydroxyethyl)imino]phenylmethyl]-phenol. These were isolated as the compounds [Fe3(OAc)(Cat)(L)3] (2),
[Fe6(OAc)2(Cat)4(L)4] (3), [Fe4(3,5-DBCat)2(L)4] (4), [Bu4N][Fe4(OAc)(3,5-DBCat)4(L)2] (5a, 5- is the complex
monoanion [Fe4(OAc)(3,5-DBCat)4(L)2]-), [Fe4(OAc)(3,5-DBCat)3(3,5-DBSQ)(L)2] (6), [Fe2(Cl4Cat)2(L)(LH2)(H2O)]
(7), and [Et3NH]2[Fe2(Cl4Cat)2(L)2] (8a, 82- is the complex dianion [Fe2(Cl4Cat)2(L)2]2-), where CatH2 = catechol; 3,5-
DBCatH2 = 3,5-di-tert-butyl-catechol; 3,5-DBSQH = 3,5-di-tert-butyl-semiquinone, and Cl4CatH2 = tetrachlorocatechol.
While compounds 2-4, 5a, 7, and 8a were obtained by directly treating 1 with the appropriate catechol, compound 6 was
synthesized by chemical oxidation of 5a. These compounds have been characterized by single crystal X-ray diffraction,
infrared and UV-visible spectroscopy, voltammetry, UV-visible spectroelectrochemistry, and magnetic susceptibility and
magnetization measurements. An electrochemical study of the three tetranuclear complexes (4, 5-, and 6) reveals
multiple reversible redox processes due to the o-dioxolene ligands, in addition to reductive processes corresponding to the
reduction of the iron(III) centers to iron(II). A voltammetric study of the progress of the chemical oxidation of compound 5a,
together with a spectroelectrochemical study of the analogous electrochemical oxidation, indicates that there are two
isomeric forms of the one-electron oxidized product. A relatively short-lived neutral species (5) that possesses the same
ligand arrangement as complex 5- is the kinetic product of both chemical and electrochemical oxidation. After several
hours, this species undergoes a significant structural rearrangement to convert to complex 6, which appears to be largely
driven by the preference for the 3,5-DBSQ- ligand to bind in a non-bridging mode. Variable temperature magnetic
susceptibility measurements for compounds 3, 4, 5a, 6, 7, and8a reveal behavior dominated by pairwise antiferromagnetic
exchange interactions, giving rise to a poorly isolated S = 0 ground state spin for compound 3, well-isolated S = 0 ground
state spins for complexes 4, 5-, 7 and 82-, and a well-isolated S = 1/2 ground state spin for complex 6. The ground state
spin values were confirmed by low temperature variable field magnetization measurements. The thermal variation of the
magnetic susceptibility for compounds 3, 4, 5a, 6, 7, and 8awere fitted and/or simulated using the appropriate Hamiltonians
to derive J values that are consistent with magnetostructural correlations that have been reported previously for alkoxo-
bridged ferric complexes.
Introduction
mechanism for bacterial iron uptake, allowing the solubiliza-
tion of iron from minerals or complexed sources. In a
different vein, the catechol dioxygenases are non-heme
iron-containing proteins that catalyze the oxidative ring
cleavage of catechol and its derivatives. There are two distinct
types of these proteins: the intradiol and extradiol catechol
dioxygenases, which, following the insertion of molecular
oxygen into the substrate, afford muconic acids and muconic
semialdehydes, respectively.2 While the active site of the
intradiol dioxygenases contains a high spin iron(III) center,
The ongoing interest in iron complexes of o-dioxolene
ligands arises in part from their relevance to several rather
diverse areas in biology. For example, bacteria produce
catechol-containing siderophores such as enterobactin,
which is a molecule with three catechol groups linked to a
trilactone backbone through amide bonds.1 Enterobactin
and other siderophores act as ligands, forming very stable
chelate complexes with iron(III) centers. This provides a
*To whom correspondence should be addressed. E-mail: c.boskovic@
unimelb.edu.au.
(1) (a) Dertz, E. A.; Raymond, K. N. Compr. Coord. Chem. II 2004, 8, 141.
(2) (a) Broderick, J. B. Essays in Biochemistry 1999, 34, 173. (b) Foster,
T. L.; Caradonna, J. P. Compr Coord. Chem. II 2004, 8, 343.
r
2009 American Chemical Society
Published on Web 07/13/2009
pubs.acs.org/IC

7766 Inorganic Chemistry, Vol. 48, No. 16, 2009
Mulyana et al.
the active site of the extradiol dioxygenases incorporates a
high spin iron(II) center. The variation of iron oxidation state
gives rise to mechanistic differences that lead to the differing
regioselectivity observed for the two types of catechol dioxy-
genases.
Chart 1
An important feature of iron complexes of o-dioxolene
ligands is that both the metal centers and the organic ligands
are redox-active, which leads to multiple possible electronic
states. Isoelectronic complexes can be described at either
extreme as iron(III)-catecholate or iron(II)-semiquinone
species, depending on the charge distribution within the
molecule. A recent report of valence tautomerism in the
complex [FeIII(bispicen)(Cl4Cat)(Cl4SQ)] (bispicen=N,N0-
bis(2-pyridylmethyl)-1,2-ethanediamine, Cl4CatH2 = tetra-
chlorocatechol, Cl4SQH = tetrachlorosemiquinone) is of
some interest in this regard.3 Variable temperature visible,
near-infrared, and electron paramagnetic resonance (EPR)
spectra of this complex in solution and in the solid state
3,5-di-tert-butyl-catechol; 3,5-DBSQH = 3,5-di-tert-butyl-
semiquinone; 3,6-DBSQH = 3,6-di-tert-butyl-semiquinone;
L0 =N,N-bis(2-pyridylmethyl)-N-(2-hydroxyethyl)amine).6-12
As part of an ongoing investigation of the chemistry of poly-
nuclear iron complexes, we have previously reported a new
family of trinuclear and pentanuclear complexes of formula
[Fe3(OAc)3(L00 )] and [Fe5O(OH)(OAc)4(L00)4] (H2L00 =deri-
3
vatives of salicylidene-2-ethanolamine),13 which have proved to
be excellent starting materials for the synthesis of higher
nuclearity complexes. These higher nuclearity complexes were
obtained by reacting the precursor tri- and pentanuclear com-
plexes with ligands that can simultaneously chelate and bridge
metal centers. Replacement of some of the acetate ligands with
the new chelating/bridging ligands, together with structural
rearrangement and agglomeration, has afforded novel Fe8,
Fe11, and Fe12 species.14 Our exploration of the potential of
one such trinuclear complex [Fe3(OAc)3(L)3] (1; H2L = 2-[[(2-
hydroxyethyl)imino]phenylmethyl]-phenol), as a precursor for
mixed-chelate polynuclear complexes of o-dioxolene ligands is
reported herein. We have focused on varying the substituents of
the o-dioxolene ligands (Chart 1) to investigate the steric effects
on the structure of the resulting complexes and the electronic
effects on the electronic properties and redox chemistry of the
new complexes. Complexes with o-dioxolene ligands in the form
of semiquinone radicals were a particular target to study the
magnetic interactions between the metal- and ligand-based
unpaired electrons.
suggest
a
thermally induced transition to [FeII(bis-
picen)(Cl4SQ)2] and a temperature-dependent equilibrium
between the two tautomeric forms. Similar valence tauto-
meric transitions are well established for related bis(o-diox-
olene) cobalt complexes with ancillary ligands with nitrogen
donor atoms.4 The thermodynamic favorability of these
conversions is largely due to the greater vibrational and
electronic entropy of the high spin cobalt(II)-containing
tautomer, compared to low spin cobalt(III)-containing tau-
tomer. Thus the cobalt(II)-containing species is favored at
high temperature. However, in the case of iron it has been
suggested that this entropy driver is not effective.5 This
argument is based partly on the minimal difference between
˚
Fe(II)-O and Fe(III)-O bond distances (ca. 0.03 A) for the
high spin iron centers that exist with these ligand fields,
compared to the significant difference for the observed low
spin Co(III)-O versus high spin Co(II)-O bond distances of
˚
about 0.2 A. In addition, the electronic contribution to the
entropy for the iron complexes is much less than for the
cobalt complexes because antiferromagnetic coupling
between the high spin Fe(II) center and the semiquinone
ligands does not give rise to the increase in spin degeneracy
that accompanies the valence tautomeric transition in the
cobalt case. It is notable that the single iron complex for
which a valence tautomeric transition has been reported
contains tetrachlorocatecholate ligands, which have electron
withdrawing chloro substituents on the catechol. This is in
contrast to the cobalt complexes displaying valence tauto-
merism, which generally have electron-donating tert-butyl or
similar substituents on the catecholate ligands that facilitate
oxidation of the ligand to the semiquinone form.
Although numerous mononuclear iron complexes of
o-dioxolene ligands have been investigated, only a few of
examples of polynuclear complexes have been reported.
These include [Fe2(Cat)4(H2O)2]2-, [Fe2(Cat)2Cl4]2-, [Fe2-
(OAc)(Cat)4], [Fe2(Cat)2(L0)2], [Fe(OMe)(3,6-DBSQ)2]2,
[Fe3{(py)2C(OH)O}2(3,5-DBCat)4]-, [Fe4{(py)2C(OMe)O}2-
{(Hpy)(py)C(OMe)O}2(3,5-DBCat)4]2þ, and [Fe4(3,5-DB-
Cat)4(3,5-DBSQ)4] (CatH2=catechol; 3,5-DBCatH2=
Experimental Section
Syntheses. All manipulations were performed under aerobic
conditions, using materials as received. The precursor complex 1
was prepared as described previously.13
[Fe3(OAc)(Cat)(L)3] (2). A mixture of complex 1 (0.76 g, 0.72
mmol) and CatH2 (0.080 g, 0.72 mmol) in MeCN (40 mL) was
stirred overnight. The resulting precipitate was removed by
filtration. The filtrate was evaporated to dryness on a rotary
(6) Grillo, V. A.; Hanson, G. R.; Wang, D.; Hambley, T. W.; Gahan, L.
R.; Murray, K. S.; Moubaraki, B.; Hawkins, C. J. Inorg. Chem. 1996, 35,
3568.
(7) Zirong, D.; Haltiwanger, R. C.; Bhattacharya, S.; Pierpont, C. G.
Inorg. Chem. 1991, 30, 4288.
(8) Anderson, B. F.; Buckingham, D. A.; Robertson, G. B.; Webb, J.;
Murray, K. S.; Clark, P. E. Nature 1976, 262, 722.
(9) Li, F.; Wang, M.; Li, P.; Zhang, T.; Sun, L. Inorg. Chem. 2007, 46,
9364.
(10) Attia, A. S.; Conklin, B. J.; Lange, C. W.; Pierpont, C. G. Inorg.
Chem. 1996, 35, 1033.
(11) Boudalis, A. K.; Dahan, F.; Bousseksou, A.; Tuchagues, J. P.;
Perlepes, S. P. Dalton Trans. 2003, 3411.
(12) Boone, S. R.; Purser, G. H.; Chang, H. R.; Lowery, M. D.;
Hendrickson, D. N.; Pierpont, C. G. J. Am. Chem. Soc. 1989, 111, 2292.
::
(13) (a) Boskovic, C.; Rusanov, E.; Stoeckli-Evans, H.; Gudel, H. U. Inorg.
::
Chem. Commun. 2002, 5, 881. (b) Boskovic, C.; Labat, G.; Neels, A.; Gudel, H. U.
::
(3) Shaikh, N.; Goswami, S.; Panja, A.; Wang, X. Y.; Gao, S.; Butcher, R.
J.; Banerjee, P. Inorg. Chem. 2004, 43, 5908.
(4) (a) Hendrickson, D. N.; Pierpont, C. G. Top. Curr. Chem. 2004, 234,
63. (b) Evangelio, E.; Ruiz- Molina, D. Eur. J. Inorg. Chem. 2005, 2957.
(5) Attia, A. S.; Bhattacharya, S.; Pierpont, C. G. Inorg. Chem. 1995, 34,
4427.
Dalton Trans. 2003, 3671. (c) Boskovic, C.;Sieber, A.; Chaboussant, G.;Gudel, H.
U.; Ensling, J.; Wernsdorfer, W.; Neels, A.; Labat, G.; Stoeckli-Evans, H.; Janssen,
S. Inorg. Chem. 2004, 43, 5053.
::
(14) Boskovic, C.; Gudel, H. U.; Labat, G.; Neels, A.; Wernsdorfer, W.;
Moubaraki, B.; Murray, K. S. Inorg. Chem. 2005, 44, 3181.

Article
Inorganic Chemistry, Vol. 48, No. 16, 2009 7767
evaporator and redissolved in the minimum volume of MeCN.
The resulting solution was layered with Et2O, giving a mixture
of crystals of 2 and 3 after several weeks. A sample for crystal-
lography was maintained in contact with mother liquor to
prevent the loss of interstitial solvent. Although individual
crystals of 2 could be isolated for single crystal studies, it was
not possible to obtain a pure bulk sample for physical measure-
ments.
with mother liquor to prevent the loss of interstitial solvent. The
crystals were isolated by filtration, washed with hexane and
vacuum-dried to afford a fully desolvated sample in 40% yield.
Anal. Calcd for 6, C88H109N2O14Fe4: C, 64.36; H, 6.69; N, 1.71.
Found: C, 64.32; H, 6.70; N, 1.70. Selected IR data (KBr, cm-1):
3437 (m), 2955 (s), 2905 (m), 2868 (m), 1597 (s), 1522 (m), 1541
(s), 1492 (m), 1479 (m), 1464 (s), 1440 (s), 1413 (m), 1392 (w),
1361 (m), 1326 (m), 1280 (m), 1240 (s), 1205 (w), 1178 (w), 1154
(w), 1112 (w), 1108 (w), 1092 (w) 1027 (m), 987 (s), 953 (w), 901
(w), 859 (w), 832 (w), 754 (m), 725 (w), 702 (m), 686 (m), 655 (w),
637 (w), 618 (w), 576 (w), 556 (w), 491 (m), 454 (w), 428 (w).
[Fe2(Cl4Cat)2(L)(LH2)(H2O)] (7). Solid Cl4CatH2 (0.30 g,
1.21 mmol) was added to a suspension of complex 1 (0.30 g,
0.28 mmol) in toluene (30 mL), and the mixture was stirred at
room temperature for 1 h. The dark brown precipitate obtained
from the reaction was removed by filtration and washed with
toluene. Redissolution of the crude product in Et2O, followed by
slow evaporation, afforded single crystals suitable for X-ray
diffraction. A sample for crystallography was maintained in
contact with mother liquor to prevent the loss of interstitial
solvent. The crystals were isolated by filtration, washed with
cold Et2O, and vacuum-dried to afford a fully desolvated sample
in 30% yield. Anal. Calcd for 7, C42H28N2O9Fe2: C, 45.86; H,
2.57; N, 2.55. Found: C, 45.90; H, 2.48; N, 2.61. Selected IR data
(KBr, cm-1): 3393 (m), 3064 (m), 2947 (m), 1600 (s), 1546 (s),
1520 (m), 1485 (m), 1431 (s), 1380 (s), 1330 (m), 1290 (w), 1247
(s), 1221 (m), 1155 (m), 1122 (w), 1061 (w), 1029 (w), 989 (m), 973
(m), 897 (w), 840 (w), 809 (s), 782 (m), 759 (m), 732 (w), 700 (m),
634 (w), 615 (w), 588 (w), 567 (w), 520 (w), 466 (w), 411 (w).
[Et3NH]2[Fe2(Cl4Cat)2(L)2] (8a). A solution of Cl4CatH2
(0.45 g, 1.8 mmol) and Et3N (0.37 g, 3.6 mmol) in MeCN (5 mL)
was added to a suspension of complex 1 (0.50 g, 0.47 mmol) in
MeCN (30 mL). The mixture was stirred overnight, and the
resulting dark red solution was filtered. Slow evaporation of the
filtrate afforded a dark brown crystalline material. Redissolution in
CH2Cl2 and layering with Et2O afforded single crystals suitable for
X-ray diffraction. A sample for crystallography was maintained in
contact with mother liquor to prevent the loss of interstitial solvent.
The crystals were isolated by filtration, washed with MeCN, and
vacuum-dried to afford a fully desolvated sample in 10% yield.
Anal. Calcd for 8a, C54H58N4O8Fe2: C, 50.42; H, 4.54; N, 4.36.
Found: C, 50.23; H, 4.51; N, 4.16. Selected IR data (KBr, cm-1):
3435 (m), 1597 (m), 1536 (w), 1440 (s), 1375 (w), 1332 (w), 1246 (s),
1146 (w), 1070 (w), 968 (w), 904 (w), 846 (w), 804 (m), 780 (w), 757
(w), 702 (w), 608 (w).
[Fe6(OAc)2(Cat)4(L)4] (3). Solid CatH2 (0.10 g, 0.91 mmol)
was added to a suspension of complex 1 (0.50 g, 0.47 mmol) in
CH2Cl2 (30 mL) and the mixture was stirred overnight. The
resulting solution was filtered and layered with Et2O, and dark
purple crystals grew after 1 day. A sample for crystallography
was maintained in contact with mother liquor to prevent the loss
of interstitial solvent. The crystals were isolated by filtration,
washed with CH2Cl2, then Et2O, and vacuum-dried to afford a
fully desolvated sample in 48% yield. Anal. Calcd for 3,
C88H74N4Fe6O20: C, 57.36; H, 4.05; N, 3.04. Found: C, 57.40;
H, 3.98; N, 2.97. Selected IR data (KBr, cm-1): 3436 (m), 3058
(w), 2926 (w), 2857 (w), 1598 (s), 1580 (s), 1536 (s), 1476 (s), 1438
(s), 1324 (s), 1249 (s), 1028 (m), 1147 (m), 1103 (m), 1075 (m),
1037 (m), 1027 (m), 955 (w), 901 (w), 871 (m), 849 (m), 800 (m),
747 (m), 702 (m), 665 (w), 619 (s), 509 (w), 454 (w), 428 (w).
[Fe4(3,5-DBCat)2(L)4] (4). A solution of 3,5-DBCatH2 (0.195 g,
0.877 mmol) and Et3N (0.10 mL, 1.8 mmol) in MeCN (5 mL) was
added to a suspension of complex 1 (0.30 g, 0.28 mmol) in MeCN
(30 mL). The mixture was then stirred at room temperature over-
night, affording a green precipitate, which was filtered and washed
with MeCN. Single crystals suitable for X-ray diffraction were
grown by layering a CH2Cl2 solution of the crude product with
MeCN. A sample for crystallography was maintained in contact
with mother liquor to prevent the loss of interstitial solvent. The
crystals were isolated by filtration, washed with CH2Cl2, and
vacuum-dried to afford a fully desolvated sample in 30% yield.
Anal. Calcd for 4, C88H92N4Fe4O12: C, 65.20; H, 5.72; N, 3.46.
Found: C, 65.18; H, 5.81; N, 3.38. Selected IR data (KBr, cm-1):
3436 (m), 2950 (m), 2863 (m), 1598 (s), 1587 (s), 1537 (s), 1464 (m),
1440 (s), 1414 (w), 1332 (s), 1245 (s), 1205 (w), 1175 (w), 1147 (m),
1120 (w), 1074 (m), 1040 (m), 1028 (m), 984 (m), 957 (w), 905 (m),
847 (m), 812 (w), 753 (m), 725 (w), 702 (m), 615 (m), 579 (m), 509
(w), 498 (w), 450 (w).
[Bu4N][Fe4(OAc)(3,5-DBCat)4(L)2] (5a). A solution of 3,5-
DBCatH2 (0.30 g, 1.3 mmol) and Et3N (0.30 g, 2.6 mmol) in
MeCN (5 mL) was added to a suspension of complex 1 (0.30 g,
0.28 mmol) in MeCN (30 mL). The mixture was then refluxed
for 1 h. Excess [Bu4N][PF6] (1.00 g, 2.58 mmol) was added upon
cooling, affording blue microcrystals. The product was then
collected by filtration and washed with cold MeCN. Single
crystals suitable for X-ray diffraction were obtained by layering
a CH2Cl2 solution with hexane. A sample for crystallography
was maintained in contact with mother liquor to prevent the loss
of interstitial solvent. The crystals were isolated by filtration,
washed with cold CH2Cl2 followed by hexane, and vacuum-
dried to afford a fully desolvated sample in 40% yield. Anal.
Calcd for 5a, C104H145N3Fe4O14: C, 66.28; H, 7.75; N, 2.15.
Found: C, 66.28; H, 7.68; N, 2.15. Selected IR data (KBr, cm-1):
3436 (m), 2950 (s), 2903 (m), 2869 (m), 1596 (s), 1554 (m), 1539
(s), 1463 (s), 1440 (s), 1415 (s), 1385 (w), 1359 (w), 1327 (m), 1279
(m), 1250 (s), 1205 (w), 1177 (w), 1145 (w), 1115 (w), 1027 (m),
987 (s), 954 (w), 900 (w) 858 (w), 832 (m), 812 (w), 754 (m), 703
(w), 685 (m), 620 (w), 577 (m), 541 (w), 508 (w), 482 (m), 455 (w).
[Fe4(OAc)(3,5-DBCat)3(3,5-DBSQ)(L)2] (6). Solid [Cp2Fe]-
[PF6] (0.050 g, 0.15 mmol) was added to a solution of compound
5a (0.20 g, 0.11 mmol) in MeCN (30 mL), and the mixture was
stirred for 15 h. The resulting dark blue precipitate obtained
from the reaction was filtered and washed with cold MeCN. The
complex was recrystallized by layering a saturated toluene
solution with hexane, affording dark blue plates after several
days. A sample for crystallography was maintained in contact
X-ray Crystallography. The intensity data for compounds
2-4, 5a, and 6 were collected on a Bruker CCD diffractometer
˚
using graphite monochromated Mo KR radiation (λ=0.71073 A),
while the data for compounds 7and 8a were collected on an Oxford
XCalibur diffractometer with graphite monochromated Cu KR
˚
radiation (λ=1.54184 A). Crystals were transferred directly from
the mother liquor to a protective oil, which was used to prevent
solvent loss. A semiempirical correction for absorption was applied
using the SADABS program, and analytical numerical absorption
corrections were carried out using a multifaceted crystal model and
the ABSPACK routine within the CrysAlis software package. The
SHELXTL series of programs were used for the solution and
refinement of the crystal structures. The structures were solved
using direct methods and refined using a full-matrix least-squares
procedure based on F2 using all data.15 The hydrogen atoms were
placed in calculated positions and refined using a riding model.
Crystallographic data for all the compounds are given in Table 1.
Magnetic Measurements. Variable temperature magnetic sus-
ceptibility and magnetization measurements were performed
with a Quantum Design MPMS-5 susceptometer equipped with
(15) Sheldrick, G. M. SHELX97 Programs for Crystal Structure Analysis;
Institut fur Anorganische Chemie der Universitat: Tammanstrasse 4, D-3400
::
Gottingen, Germany, 1998.

7768 Inorganic Chemistry, Vol. 48, No. 16, 2009
Mulyana et al.
Table 1. Crystallographic Data for 2 Et2O, 3 CH2Cl2 Et2O, 4 4CH2Cl2, 5a 5CH2Cl2, 6 4C7H8 C6H14, 7 4Et2O, and 8a 2CH2Cl2
3
3
3
3
3
3
3
3
3
2 Et2O
3 CH2Cl2 Et2O
4 4CH2Cl2
5a 5CH2Cl2
6 4C7H8 C6H14
7 4Et2O
8a 2CH2Cl2
3
3
3
3
3
3
3
3
3
formula
C57H56N3O4-
Fe3
1126.60
P1
12.479(7)
12.853(7)
17.799(1)
77.693(1)
69.545(1)
85.529(1)
2613.3(3)
2
130
1.432
0.886
16587
11452
0.0481, 0.1177
0.0608, 0.1259
C93H86Cl2-
Fe6N4 O21
2001.66
P21/c
14.215(2)
33.054(5)
19.978(3)
90
104.133(4)
90
9103(2)
4
130
1.461
1.063
48163
15995
0.0801, 0.1932
0.1528, 0.2201
C92H100Cl8-
Fe4N4O12
1970.76
P1
13.718(3)
15.525(3)
23.129(5)
70.600(3)
84.280(3)
77.130(3)
4528(2)
2
295
1.438
0.926
23091
15247
0.0781, 0.1736
0.1327, 0.1967
C109H155Cl10
Fe4N3 O14
2309.26
P1
17.728(2)
17.743(2)
19.890(2)
109.829(2)
102.394(2)
90.314(2)
5727(1)
2
130
1.339
0.789
29915
19768
0.0536, 0.1352
0.0720, 0.1420
-
C122H155
-
C58H66N2O13
Fe2Cl8
1394.43
P1
15.510(4)
15.567(4)
16.177(4)
95.875(4)
106.204(5)
113.366(5)
3341(2)
2
130
1.386
6.913
15174
7028
0.0407, 0.0915
0.0605, 0.0971
-
C56H62Cl12
Fe2N4O8
1456.20
-
Fe4N2O14
2094.86
P1
formula weight
space group
P21/n
˚
a, A
14.436(3)
17.963(4)
24.619(5)
92.931(4)
105.270(5)
113.359(4)
5566(2)
2
130
1.251
0.573
29492
18.315(3)
18.744(2)
19.976(3)
90
115.828(2)
90
6173(2)
4
130
1.567
9.013
˚
b, A
˚
c, A
R, deg
β, deg
γ, deg
3
˚
V, A
Z
T, K
F
calc, g cm-3
μ, mm-1
reflns measd
28171
12069
0.0705, 0.1802
0.0837, 0.1924
unique reflns
R1, wR2 [I > 2σ(I)]
R1, wR2 (all data)
19357
0.0708, 0.1520
0.0843, 0.1584
a 5 T magnet. Data were collected on powdered dried crystals,
with two sets of susceptibility data collected with magnetic fields
of 0.01 and 0.1 T. Pascal’s constants were used to estimate the
diamagnetic correction for each complex. Magnetization data
were collected with fields up to 5 T and temperatures between
2.0 and 20 K.
Magnetic susceptibility data for compounds 3, 4, 5a, 7, and 8a
were fitted to the exchange Hamiltonians given in the text. The
computer program MAGMUN was employed for compounds
4, 5a, 7, and8a,16 while thehomemade computerprogramRBFIT
was used for compound 3. This program uses a Levenberg-
Marquardt algorithm for the least-squares fit and employs full
matrix diagonalization methods, making use of total spin sym-
metry to reduce the matrix dimensions.17 The computer program
MAGPACK was used to simulate the magnetic susceptibility
data for complex 6.18
addition of -0.61 V (in CH2Cl2 with 0.05 M [Bu4N][B(C6F5)4])
or -0.55 V (in CH2Cl2 with 0.1 M [Bu4N][PF6]).19 Mechanistic
aspects of the voltammetric processes were investigated by
applying the appropriate diagnostic criteria to cyclic voltam-
metric data as described elsewhere.20
UV-visible spectroelectrochemistry was performed using a
homemade Optically Transparent Thin-Layer Electrolysis
(OTTLE) cell with platinum gauze as the working electrode.
The experiments employed approximately 2 mL of CH2Cl2
solutions containing 0.1 M [Bu4N][PF6] and 0.2-0.75 mM
analyte. The UV-visible spectra were recorded with a Varian
Cary 5000 spectrophotometer.
Other Measurements. Infrared spectra (KBr disk) were recorded
on a BioRad 175 FTIR spectrometer. UV-visible spectra were
measured on a Varian Cary 50 Bio UV-visible Spectrophotometer.
Elemental analyses were performed at Chemical and Microanaly-
tical Services, Belmont, Victoria, Australia.
Electrochemistry and Spectroelectrochemistry. Electrochemi-
cal measurements were performed in CH2Cl2 at 293 ( 2 K using
a standard three-electrode configuration under a flow of nitro-
gen gas or in a Vacuum Atmospheres glovebox connected to a
BASi EC Epsilon computer-controlled electrochemical work-
station. The three-electrode arrangement employed for voltam-
metric measurements consisted of 1.0 or 1.5 mm diameter glassy
carbon (Cypress) disk working electrodes, a platinum wire
counter electrode, and a reference electrode consisting of a
AgCl-coated Ag wire, prepared by oxidative electrolysis of the
wire in 0.1 M KCl solution, separated from the analyte solution
by a glass frit of low porosity containing the same solvent/
supporting electrolyte mixture. For bulk electrolysis experi-
ments, large platinum gauze and platinum mesh baskets were
used as the working electrode and counter electrode, respec-
tively. The Ag/AgCl reference electrode was the same as that
employed in the voltammetric studies. Analyte solutions of 1 to
3 mM were prepared in CH2Cl2 containing either 0.1 M
[Bu4N][PF6] or 0.05 M [Bu4N][B(C6F5)4] as the supporting
electrolyte.
Results and Discussion
Syntheses. Complex 1 proved to be an excellent start-
ing material for the synthesis of new polynuclear ferric
complexes containing substituted catecholate ligands.
The formation of these new complexes from the reaction
of 1 with the deprotonated or acid forms of the catechol
proligands involves the concerted loss of acetate ligands,
coordination of catecholate ligands, and structural
rearrangement. The variety of products isolated sug-
gests that a range of species are present in equilibrium in
solution, with the relative solubilities in the particular
solvent system governing the identity of the species
that crystallizes from solution. Most of the products
obtained were isolated in reasonable yield. The path-
ways for the syntheses of all of the compounds are
summarized in Scheme 1.
All potentials reported in this paper are in volt versus the
ferrocene/ferrocenium redox couple. This was achieved by
addition of decamethylferrocene as an internal standard and
then calibrated to the ferrocene/ferrocenium potential scale by
The reaction of complex 1 with approximately 1 equiv
of catechol in acetonitrile afforded a significant quantity
of precipitate and a dark red solution. Following removal
of the precipitate, evaporation of the mother liquor to
dryness, redissolution in acetonitrile, and layering with
(16) Thompson, L. K.; Waldmann, O.; Xu, Z. Coord. Chem. Rev. 2005,
249, 2677.
(17) (a) Gatteschi, D.; Pardi, L. Gazz. Chim. Ital. 1993, 123, 231.
(b) Waldmann, O. Phys. Rev. B 2000, 61, 6138.
(19) Barrire, F.; Geiger, W. E. J. Am. Chem. Soc. 2006, 128, 3980.
(20) Geiger, W. E. In Laboratory Techniques in Electrochemistry, 2nd ed.;
Kissinger, P. T., Heineman, W. R., Eds.; Marcel Dekker: New York, 1996; Chapter
23.
ꢀ
(18) Borras-Almenar, J. J.; Clemente-Juan, J. M.; Coronado, E.; Tsukerblatt,
B. S. MAGPACK; Inorg. Chem. 1999, 38, 6081.

Article
Inorganic Chemistry, Vol. 48, No. 16, 2009 7769
Scheme 1
The reaction in acetonitrile of complex 1 with excess
3,5-di-tert-butyl-catechol under basic conditions result-
ing from the addition of triethylamine gave different
products depending on the temperature of the reaction.
When undertaken at ambient temperature, the resulting
green precipitate could be recrystallized from dichloro-
methane/acetonitrile to give the neutral tetranuclear
complex 4 in good yield. When the solution was refluxed
and allowed to cool, addition of tetrabutylammonium
hexafluorophosphate to the resulting dark blue solution
afforded a microcrystalline blue product that could be
recrystallized from dichloromethane/hexane to give com-
pound 5a, which is the tetrabutylammonium salt of a
monoanionic tetranuclear complex (5-). The formation
of complexes 4 and 5- are summarized in eqs 3 and 4,
respectively.
diethyl ether, a mixture of crystals of complexes 2 and 3
were obtained. Redissolution of the original precipitate
in dichloromethane and layering with diethyl ether gave
a mixture of crystalline complex 2 and an amorphous
material. Although individual crystals of the trinuclear
complex 2 could be isolated for X-ray crystallography,
it was not possible to obtain the pure bulk sample
required for physical measurements. In contrast,
repeating the original reaction in dichloromethane
and layering the resulting purple solution with diethyl
ether afforded a pure crystalline sample of the hexa-
nuclear complex 3, with an improved yield obtained
when 2 equiv of catechol were employed. The formation
of complexes 2 and 3 are summarized in eqs 1 and 2,
respectively.
The reversible potential (-0.27 V versus ferrocene/
ferrocenium, vide infra) for the first one-electron oxida-
tion process of 5- suggested that this process might be
readily accessible by chemical oxidation. The reaction of
a solution of 5a in acetonitrile with 1.4 equiv of ferroce-
nium hexafluorophosphate afforded a blue precipitate
that could be recrystallized from toluene/hexane to give 6
in 40% yield according to eq 5.
Varying the amount of ferrocenium hexafluoropho-
sphate employed in the reaction did not improve the
yield of 6.
The reaction of a suspension of complex 1 and tetra-
chlorocatechol in toluene or dichloromethane afforded
a brown precipitate that could be recrystallized from
diethyl ether to give complex 7. The reaction of complex
1 with tetrachlorocatechol and triethylamine in acetoni-
trile gave a clear solution. The residue that was obtained
following the evaporation of this solution to dryness was
recrystallized from dichloromethane/ether to give com-
pound 8a, which is the triethylammonium salt of the
dianionic dinuclear complex (82-). The formation of
Although a pure crystalline sample of 3 could be
obtained from the initial crystallization from dichloro-
methane and ether, subsequent attempts to recrystallize
this material from the same solvent mix yielded a
microcrystalline material that appeared to be a mixture
of 2 and 3 according to elemental analysis data. This
suggests that in the absence of excess acetic acid, com-
plex 3 is unstable in solution, partially degrading to
complex 2.

7770 Inorganic Chemistry, Vol. 48, No. 16, 2009
Mulyana et al.
complexes 7 and 82- are summarized in eqs 6 and 7,
respectively.
The direct substitution of tetrachlorocatechol with tetra-
bromocatechol in each of these reactions afforded micro-
crystalline materials that were not suitable for X-ray
diffraction and which proved to be more insoluble in
dichloromethane than the chlorinated analogues.
Dianionic catecholate ligands are evident in all of the
complexes synthesized directly from the trinuclear pre-
cursor 1, regardless of whether the catechol precursor was
deliberately deprotonated prior to reaction with 1. How-
ever, the relatively electron donating tert-butyl substitu-
ents of 3,5-di-tert-butyl-catechol render this species the
weakest acid of the three catechol proligands, and it
was found that, in contrast to the other two catechols
employed, protonated 3,5-di-tert-butyl-catechol did not
react with the precursor complex 1 until base was added
to the solution.
Vacuum drying the isolated crystalline products
afforded fully desolvated samples in all cases. The purity
of bulk samples of compounds 3, 4, 5a, 6, 7, and 8a for
physical measurements was established by elemental
analysis.
Figure 1. Structural representation of (a) complex 2 and (b) complex 3.
Color code: Fe (dark blue), O (red), N (light blue), C (black). Hydrogen
atoms are omitted for clarity.
Structures. Labeled structural diagrams of the com-
plexes of interest are presented in Figures 1 (2 and 3), 2
(4, 5- and 6), and 3 (7 and 82-), while the different binding
modes displayed by the ligands are depicted in Figure 4.
The most relevant interatomic distances and angles are
available in Tables 2-5. All of the iron centers in the
complexes are six coordinate with distorted octahedral
coordination geometry and bond valence sum calcula-
tions indicating that they are all Fe(III).21
Complex 2 crystallizes in the triclinic space group P1,
with one neutral trinuclear complex (Figure 1a) in the
asymmetric unit. This asymmetric complex contains a
{FeIII3(μ3-O)(μ2-O)3}þ core unit, where the μ3-O atom
is from the ethoxo arm of a μ3-L2- (Figure 4), two of the
μ2-O atoms are from the ethoxo arms of two μ2-L2-
ligands, and the remaining μ2-O atom is a phenoxo-type
oxygen atom from the μ2-Cat2- ligand. The structure of
the core can be described as a monovacant distorted
cubane. The remaining coordination sites of the iron
centers are occupied by the L2- and Cat2- ligands, in
addition to a single acetate ligand that binds in the
common syn,syn μ2-mode. The structural differences
between complex 2 and the precursor trinuclear complex
1 arise from the μ3-Oalkoxo atom that is present in 2 but not
in 1 and the single bridging acetate ligand in 2 as opposed
to the three bridging acetate ligands in 1.
Complex 3 crystallizes in the monoclinic P21/c space
group with one neutral hexanuclear complex (Figure 1b)
in the asymmetric unit. The {FeIII6(μ3-O)2(μ2-O)8}2- core
of 3 resembles two of the core units of complex 2 linked by
two μ2-Ophenoxo atoms from two μ3-Cat2- ligands.
Although this complex is strictly asymmetric, it has an
approximate inversion center. The two planes defined by
the three iron centers in each half of the molecule (Fe1,
Fe2 and Fe3; Fe4, Fe5 and Fe6) are nearly parallel, with a
dihedral angle of 6.81(4)°. The iron coordination is
completed by two μ2-Cat2- ligands, two μ3-L2- ligands,
two μ2-L2- ligands and two μ2-acetate ligands.
Complex 4 crystallizes in the triclinic space group P1,
such that the unit cell contains two independent mole-
cules of complex 4, each of which lies on a center of
inversion (Figure 2a). Although the two molecules are
crystallographically distinct they are chemically similar,
with identical connectivities and ligand binding modes.
Each neutral tetranuclear complex has an {FeIII4(μ3-O)2-
(μ2-O)4} core that can be described as a divacant face-
shared distorted double cubane unit, where the individual
(21) Brese, N. E.; O0 Keefe, M. Acta Crystallogr. 1991, B47, 192.

Article
Inorganic Chemistry, Vol. 48, No. 16, 2009 7771
Figure 3. Structuralrepresentation of(a) complex7and (b) complex 82-
Color code as in Figure 1 with Cl (green); the dashed lines indicate
intramolecular hydrogen bonds.
.
complex are coplanar. The two μ3-O atoms in the core are
from the ethoxo arms of μ3-L2- ligands, two of the μ2-O
atoms are from the ethoxo arms of μ2-L2- ligands and the
remaining two μ2-O atoms are phenoxo-type oxygen
atoms from the two μ2-3,5-DBCat2- ligands. The main
conformational differences between the two crystallogra-
phically independent molecules of complex 4 arise from
different orientations of the ligands, particularly the
phenyl groups.
Compound 5a crystallizes in the triclinic space group
P1, with one monoanionic tetranuclear complex (5-) in
the asymmetric unit (Figure 2b). The {FeIII4(μ3-O)2(μ2-
O)4} core of complex 5- is similar to that of complex 4,
although the four iron centers of this asymmetric complex
are only approximately coplanar. As in 4, the two μ3-O
atoms in the core of 5- are from the ethoxo arms of two
μ3-L2- ligands; however, the four μ2-O atoms in the core
of 5- are all phenoxo-type oxygen atoms from the four μ2-
3,5-DBCat2- ligands. Although the core topologies of the
Figure 2. Structural representation of (a) one of the independent molecules
of complex 4, (b) complex 5-, and (c) complex 6. Color code as in Figure 1.
monovacant cubane units are similar to the core of
complex 2. The four iron centers of this centrosymmetric

7772 Inorganic Chemistry, Vol. 48, No. 16, 2009
Mulyana et al.
three μ2-3,5-DBCat2- ligands, while the fourth μ2-O atom
is from one of the L2- ligands that binds overall in an
unusual μ4-mode (Figure 4). Complex 6 is the only
complex in this study to display this binding mode for
L
2-. In the simplest sense the structural rearrangement
that occurs during the formation of 6 from 5- involves
one of the L2- ligands (the one that contains atom N2)
and one of the o-dioxolene ligands (defined by O5 and O6
in 5- and O7 and O8 in 6) exchanging binding sites, while
the coordination of the remaining ligands does not
change. The ligands that change their binding positions
also change their binding modes with the μ3-mode of L2-
in 5- converting to a μ4-mode in 6, while the μ2-bridging
3,5-DBCat2- ligand in 5- converts to a chelating-only
3,5-DBSQ- ligand in 6. This structural rearrangement
involves breaking and reforming five Fe-O and Fe-N
bonds. These rearrangements are likely associated with
the lower affinity of the single 3,5-DBSQ- ligand to
bridge Fe centers, consistent with the relatively electron
poor nature of the 3,5-DBSQ- ligand compared to the
3,5-DBCat2- form. This 3,5-DBSQ- ligand is readily
distinguished (vide infra) from the three 3,5-DBCat2-
ligands crystallographically by the short C-O and long
ring (O)C-C(O) bond lengths (Table 5).
Complex 7 crystallizes in the triclinic space group P1,
with one neutral dinuclear complex (Figure 3a) in the
asymmetric unit. This complex is asymmetric with both
Cl4Cat2- ligands chelating one iron center and the L2-
and LH2 ligands chelating the other iron center. The
molecule has a {FeIII2(μ2-O)2}2þ core, with one of the
core oxygen atoms from the ethoxo arm of the μ2-L2-
ligand and the other a phenoxo-type oxygen atom from a
μ2-Cl4Cat2- ligand. With the other Cl4Cat2- ligand bind-
ing only in a chelating manner, the two Cl4Cat2- ligands
bind in two different modes in this complex. Complex 7
is the only species in this study to incorporate a proto-
nated form of the Schiff base ligand LH2, which acts as a
μ2-bridging ligand (Figure 4), in addition to the μ2-brid-
ging L2-. The C-N bond length for the bound LH2
Figure 4. Ligand binding modes in the complexes.
˚
Table 2. Fe-O Distances (A) for Compounds 2 Et2O, 3 CH2Cl2 Et2O,
3
3
3
4 4CH2Cl2, 5a 5CH2Cl2, 6 4C7H8 C6H14, 7 4Et2O, and 8a 2CH2Cl2.a
3
3
3
3
3
3
compound
Fe-O (μ2-OR)
Fe-O (μ3-OR)
˚
˚
2 Et2O
3 CH2Cl2 Et2O
1.966(2)-2.049(2)
1.935(4)-2.040(5)
1.966(4)-2.069(4)
1.920(2)-2.065(2)
1.932(3)-2.079(3)
1.977(3)-2.093(3)
2.012(3)-2.022(3)
2.112(2)-2.186(2)
2.088(5)-2.159(5)
2.083(4)-2.238(4)
2.079(2)-2.183(3)
2.051(3)-2.173(3)
ligand is 1.304(8) A, which is close to the 1.296(8) A
observed for the L2- ligand and consistent with the
maintenance of imine character, rather than reduction
to an amine. The unusual binding mode of LH2 is coin-
cident with an intramolecular hydrogen bond between
3
3
3
4 4CH2Cl2
5a 5CH2Cl2
3
3
6 4C7H8 C6H14
7 4Et2O
8a 2CH2Cl2
3
3
3
the imine fragment and the phenolic oxygen atom, with a
˚
3
a Distances to bridging O atoms only.
3 3 3
O-H N distance of 2.55 A. Another intramolecular
hydrogen bond is evident between the catecholate oxygen
atom O5 bound to Fe1 and O9 of the terminal water
two complexes are similar, the structural differences arise
from the different ligand combinations in the two com-
plexes, with four L2- and two 3,5-DBcat2- ligands in 4,
versus two L2- and four 3,5-DBCat2- ligands in 5-.
Complex 5- also possesses a single μ2-acetate ligand
bridging Fe2 and Fe4.
Complex 6 crystallizes in the triclinic P1 space group
with one neutral tetranuclear complex (Figure 2c) in the
asymmetric unit. Although the first one-electron oxida-
tion of 5- is chemically reversible on the voltammetric
time scale (vide infra), interesting structural rearrange-
ments occur following the formation of 6 from chemical
oxidation of 5a. Like 4 and 5-, 6 possesses a {FeIII4(μ3-
O)2(μ2-O)4} core with approximately coplanar iron cen-
ters. The two μ3-O atoms are again ethoxo type from the
two L2- ligands, three of the μ2-O atoms are from the
ligand bound to Fe2, with an O-H O separation of
2.66 A.
3 3 3
˚
Compound 8a crystallizes in the monoclinic space
group P21/n with one dianionic dinuclear complex
(82-) in the asymmetric unit (Figure 3b). As occurs in
complex 7, both Cl4Cat2- ligands chelate one iron
center, while the L2- ligands both chelate the other iron
center. Complex 82- also features a {FeIII2(μ2-O)2}2þ
core; however, in this case both bridging oxygen atoms
are from the ethoxo arms of the two μ2-L2- ligands.
Complex 82- is the only species in the present study in
which none of the XCat2- ligands bridge metal centers.
Although the dianion is strictly asymmetric, it has
approximate C2 symmetry, with the C2 axis lying along
the iron-iron vector.

Article
Inorganic Chemistry, Vol. 48, No. 16, 2009 7773
˚
Table 3. Fe Fe Distances (A) for Compounds 2 Et2O, 3 CH2Cl2 Et2O, 4 4CH2Cl2, 5a 5CH2Cl2, 6 4C7H8 C6H14, 7 4Et2O, and 8a 2CH2Cl2
3 3 3
3
3
3
3
3
3
3
3
3
compound
via (μ2-OR)2
via (μ2-OR)(μ3-OR)
via (μ3-OR)2
via (μ2-OR)(μ3-OR)(μ2-OAc)
2 Et2O
3 CH2Cl2 Et2O
3.266(6)-3.307(5)
3.219(1)-3.267(1)
3.253(2)-3.297(2)
3.293(7)-3.310(8)
3.205(8)-3.294(1)
3.058(6)
3.075(2)-3.084(1)
3
3
3.206(1)
3
4 4CH2Cl2
5a 5CH2Cl2
3.172(2)-3.212(2)
3.305(9)
3.206(1)
3
3.086(7)
3.165(1)
3
6 4C7H8 C6H14
7 4Et2O
8a 2CH2Cl2
3
3
3
3.227(1)
3.185(1)
3
complexes obtained from 3,5-DBCat2-. This nuclearity
limitation is also evident in the previously reported tetra-
nuclear complex [Fe4(3,5-DBCat)4(3,5-DBSQ)4], which
contains a {FeIII4(μ3-O)2(μ2-O)4} core similar to those of
4 and 5-.12 Although the binding modes of the 3,5-
DBCat2- and 3,5-DBSQ- ligands in [Fe4(3,5-DBCat)4-
(3,5-DBSQ)4] and in the present complexes differ, the
semiquinone form of the ligand clearly shows a lower
affinity than the catecholate form for bridging Fe centers
in both complex 6 and [Fe4(3,5-DBCat)4(3,5-DBSQ)4].
The structures of the dinuclear complexes 7 and 8a may
also be compared with those of the two previously
reported dinuclear ferric complexes of catechol.6-9
Although a {FeIII2(μ2-O)2}2þ core is common to all, the
complexes [Fe2(Cat)4(H2O)2]2-, [Fe2(Cat)2Cl4]2-, and
[Fe2(OAc)(Cat)4] have both of the core oxygen atoms
provided by two μ2-catechol ligands,6-8 while 7 has one of
the core oxygen atoms provided by a Cl4Cat2-, and 82-
and [Fe2(Cat)2(L0)2] both have the two core oxygen atoms
provided by ancillary ligands.9
Table 4. Fe-O-Fe Bridging Angles (deg) for Compounds 2 Et2O, 3 CH2Cl2
3
3
3
Et2O, 4 4CH2Cl2, 5a 5CH2Cl2, 6 4C7H8 C6H14, 7 4Et2O, and 8a 2CH2Cl2
3
3
3
3
3
3
compound
via μ2-OR
via μ3-OR
2 Et2O
3 CH2Cl2 Et2O
100.29(8)-110.02(9)
101.8(2)-109.3(2)
108.8(2)-110.8(2)
100.27(9)-111.0(1)
102.6(1)-111.0(1)
102.2(1)-107.5(1)
104.0(1)-104.5(1)
91.98(7)-99.74(7)
92.8(2)-100.4(2)
97.6(2)-99.2(2)
92.99(8)-103.53(9)
99.6(1)-101.1(1)
3
3
3
4 4CH2Cl2
5a 5CH2Cl2
3
3
6 4C7H8 C6H14
7 4Et2O
8a 2CH2Cl2
3
3
3
3
The C-O and ring (O)C-C(O) bond distances are
generally considered diagnostic for the oxidation state
of o-dioxolene ligands.11,12,22 Catecholate C-O distances
˚
are typically in the range 1.33-1.39 A, semiquinone C-O
˚
distances are in the range 1.27-1.31 A, and quinone C-O
distances are usually around 1.23 A. These ranges en-
˚
compass the different binding modes of the ligands, with
longer C-O distances observed for an oxygen atom that
bridges two metal centers. The ring (O)C-C(O) distances
are also informative, with distances in the range 1.36-
Infrared Spectroscopy. Infrared spectra for compounds
3, 4, 5a, 6, 7, and 8a were obtained in the range 400-
4000 cm-1 from pressed KBr disks. The infrared spectra
of all the compounds display medium to strong bands in
the two regions 1237-1280 and 1400-1500 cm-1 which
are typical for metal-catecholate complexes.6,23 The infra-
red spectra of 5a and its one-electron oxidized derivative 6
are very similar (Supporting Information, Figure S1).
However, the bands at 1479 and 1492 cm-1 in the
spectrum of 6 that are absent in that of 5a may be assigned
to the C-O semiquinone stretch of the 3,5-DBSQ-
ligand.24 All of the compounds also display a band at
around 1590-1600 cm-1 that is characteristic of the
CdN stretch for the iron-coordinated imine moiety in
˚
˚
1.44 A for catecholates, in the range 1.45-1.48 A for
˚
semiquinones and around 1.5 A for quinones. With the
exception of the single 3,5-DBSQ- ligand in complex 6,
which has C-O and ring (O)C-C(O) bond distances
consistent with a semiquinone, all of the o-dioxolene
ligands in complexes 2-4, 5-, 6, 7, and 82- display C-
O and ring (O)C-C(O) distances that lie within the ranges
typical for catecholate ligands (Table 5).
The coordinative versatility of the ligands used in this
work (Figure 4) gives rise to the multiple structural types
evident in the present complexes. The structural differ-
ences between the products obtained following the reac-
tion of the same trinuclear precursor complex 1 with the
different catechol species can be rationalized in part by a
consideration of both electronic and steric effects of the
catechol substituents. The relatively electron withdraw-
ing chloro substituents render Cl4Cat2- the least likely of
the three XCat2- ligands to bridge iron centers and the
most likely to simply chelate, giving rise to relatively
simple dinuclear complexes. On the other hand Cat2-
and 3,5-DBCat2- are more readily able to bridge the
metal centers, and the monovacant distorted cubane
structural fragment that is evident in complexes, 2, 3, 4,
5-, and 6 is the result. The metric parameters associated
with these units across the four complexes that incorpo-
rate them are very similar (Tables 3-5). These structural
units are linked indirectly in complex 3 and directly
through face-sharing in complexes 4, 5-, and 6. It seems
likely that the steric bulk associated with the tert-butyl
substituents has limited to four the nuclearity of the
ligand L .
2- 25 The spectrum of complex 7 displays a very
broadband with ill-defined features between 3000 and
3700 cm-1 that is not apparent in the spectra of the other
compounds. This is likely due to the O-H and N-H
stretches of the protonated LH2 ligand.
UV-visible Spectroscopy. UV-visible spectra were
measured for solutions of compounds 4, 5a, 6, 7, and 8a
in dichloromethane between 250 and 1100 nm (Table 6).
Spectra were also obtained in acetonitrile for the com-
pounds that were soluble (5a, 6, and 7). Monitoring the
spectra of the solutions over time indicated that com-
pounds 4, 5a, 6, 7, and 8a were all stable in solution for at
(23) Lynch, M. W.; Valentine, M.; Hendrickson, D. N. J. Am. Chem. Soc.
1982, 104, 6982.
(24) Pecoraro, V. L.; Harris, W. R.; Wong, G. B.; Carrano, C. J.;
Raymond, K. N. J. Am. Chem. Soc. 1983, 105, 4623.
ꢀ
(25) (a) Champouret, Y. D. M.; Marechal, J. D.; Chaggar, R. K.;
(22) Ding, Z.; Bhattacharya, S.; McCusker, J. K.; Hagen, P. M.; Hen-
drickson, D. N.; Pierpont, C. G. Inorg. Chem. 1992, 31, 870.
Fawcett, J.; Kuldip Singh, K.; Maseras, F.; Solan, G. A. New. J. Chem.
2007, 31, 75. (b) Sharma, V. K.; Srivastava, S. J. Coord. Chem. 2008, 61, 178.

7774 Inorganic Chemistry, Vol. 48, No. 16, 2009
Mulyana et al.
Table 5. Average C-O and Ring (O)C-C(O) Distances (A) of the Cat2-, Cl4Cat2-, 3,5-DBCat2-, and 3,5-DBSQ- Ligands in Compounds 2 Et2O, 3 CH2Cl2 Et2O,
˚
3
3
3
4 4CH2Cl2, 5a 5CH2Cl2, 6 4C7H8 C6H14, 7 4Et2O, and 8a 2CH2Cl2
3
3
3
3
3
3
C-O
μ2-O
ring (O)C-C(O)
XCat2- 3,5-DBSQ2-
compound
2 Et2O
3 CH2Cl2 Et2O
XCat2- terminal O
3,5-DBSQ2-
1.343(3)
1.368(3)
1.409(4)
3
3
1.322(9)-1.349(8)
1.329(7)-1.345(7)
1.335(4)-1.357(4)
1.341(5)-1.354(5)
1.332(5)-1.341(5)
1.311(6)-1.329(5)
1.333(8)-1.373(8)
1.370(7)-1.376(7)
1.352(4)-1.370(4)
1.350(5)-1.361(4)
1.336(5)
1.36(1)-1.397(9)
1.406(8)-1.415(8)
1.403(4)-1.415(4)
1.396(5)-1.410(6)
1.420(6)-1.424(6)
1.428(6)-1.438(6)
3
4 4CH2Cl2
5a 5CH2Cl2
3
3
6 4C7H8 C6H14
7 4Et2O
8a 2CH2Cl2
1.287(5)-1.291(5)
1.450(6)
3
3
3
3
28
Table 6. UV-visible Spectral Data for Compounds 4, 5a, 6, 7, and 8a in
Scheme 2.
Dichloromethane and Acetonitrile
compound
solvent
λ
max, nm (ε ꢀ 103, M-1 cm-1
)
4
MeCN
CH2Cl2
MeCN
CH2Cl2
MeCN
CH2Cl2
MeCN
CH2Cl2
MeCN
CH2Cl2
insoluble
288a (36.5), 543 (7.97)
5a
6
284 (36.1), 319a (21.0), 607 (12.3)
285 (40.6), 420a (14.0), 600 (15.4)
283a (31.0), 651 (9.34)
286a (38.6), 618 (13.2)
[PF6] as the supporting electrolyte are similar for the first
three oxidations as well as the reduction processes. How-
ever, electrode fouling that requires periodic cleaning of
the electrode surface is encountered when the fourth
oxidation process is involved in the potential scan with
[Bu4N][PF6] as the electrolyte. Thus unless otherwise
specified, the voltammetric data presented here refer to
the use of [Bu4N][B(C6F5)4] as the supporting electrolyte.
Cyclic and square wave voltammograms of complex 5-
(Figure 5) reveal the presence of four well separated
7
294 (17.5), 551 (5.05)
289a (21.6), 549 (5.40)
303 (26.1), 489 (8.36)
303 (24.9), 489 (5.78)
8
a Shoulder.
least a period of several hours. Two intense bands are
evident for each of the compounds, with a less intense
band in the visible region in the range 489-651 nm and
more intense band in the UV region between 280 and
320 nm, which occurs in some cases as a shoulder on an
even more intense absorption. A slight solvent-depen-
dence is observed for both of these bands. The UV band
has been assigned previously as an internal transition
within the o-dioxolene ligands,26 while the band in the
visible region has been attributed to a ligand-to-metal
charge transfer (LMCT) from the catecholate ligands to
the Fe(III) center, that is, Fe(III)-Cat f Fe(II)-SQ.5,27
The energy of this LMCT band varies somewhat between
the complexes, reflecting the varying Lewis basicities of
the differently substituted catecholate ligands.
oxidation processes having formal potentials (E0 (n),
f
n = 1-4) of -0.27, 0.09, 0.52, and 0.89 V versus
[Cp2Fe]0/þ (Table 7), in the potential range 0 to þ1.50
V, consistent with the presence of four oxidizable 3,5-
DBCat2- ligands. The E0 values were calculated from the
f
average of oxidation (Epox) and reduction (Epred) peak
ox
potentials (Ep þEpred)/2, assuming that the oxidized and
reduced components have equal diffusion coefficients.
Scanning the potential in the negative direction (0 to -
2.50 V) led to the detection of a chemically reversible
reduction process at ∼ -1.36 V attributable to one-
electron metal-based Fe(III) to Fe(II) step, together with
a chemically irreversible one at a very negative potential,
Epred = -2.02 V.
Electrochemistry and Spectroelectrochemistry. Both the
iron centers and the o-dioxolene ligands of the complexes
synthesized in this study were expected to exhibit redox
activity. The family of tetranuclear complexes 4, 5-, and 6
was of particular interest in this regard, given the com-
monality in structural core, but variable number (4 and
5-) and differing redox state (5- and 6) of the o-dioxolene
ligands. Although, 6 is derived from the one-electron
chemical oxidation of 5-, X-ray structural data (vide
supra) indicate that a significant ligand rearrangement,
involving the breaking and reforming of up to five Fe-O
and Fe-N bonds, accompanies this oxidation. Thus,
comparative voltammetric studies of 5- and 6 were
expected to provide important insights into this structural
rearrangement.
The diffusion-controlled nature of the oxidation pro-
cesses was confirmed under conditions of cyclic voltam-
metry by the linear plots of the peak currents, Ipox, versus
the square root of scan rate (ν1/2), which passed through
the origin. In addition, Anson plots (not shown) con-
structed from double potential step chronocoulometric
measurements performed at step times of 0.5 to 10 s over
the potential range relevant to each oxidation process
established the absence of reactant or product adsorp-
)
red
tion. The peak-to-peak separations (ΔEp = Epox - Ep
for the first three oxidation processes (76, 82, and 89 mV,
respectively, at a scan rate of 100 mV s-1) are similar
to values obtained for the known electrochemically re-
versible one-electron oxidation of decamethylferrocene
under the same conditions, thereby indicating that these
The voltammetric responses of 5- in dichloromethane
solution containing either [Bu4N][B(C6F5)4] or [Bu4N]-
(26) Benelli, C.; Dei, A.; Gatteschi, D.; Pardi, L. Inorg. Chem. 1989, 28,
(28) No assumption relating to electronic distribution can be inferred
from the given formulations.
1476.
(27) Dei, A. Inorg. Chem. 1993, 32, 5730.

Article
Inorganic Chemistry, Vol. 48, No. 16, 2009 7775
Table 7. E0f Valuesa (volts versus [Cp2Fe]0/þ) in CH2Cl2 Obtained by Cyclic Voltammetry with a 1.5 mm Diameter Glassy Carbon Electrode at a Scan Rate of 100 mV s-1
FeIII/II reduction
o-dioxolene reduction/oxidation
complex
electrolyte
E0 (-1)
E0 (1)
E0 (2)
E0 (3)
E0 (4)
f
f
f
f
f
5-
5-
6c
[Bu4N][B(C6F5)4]
[Bu4N][PF6]
[Bu4N][B(C6F5)4]
[Bu4N][PF6]
[Bu4N][B(C6F5)4]
-1.36b
-1.33b
-1.03
-1.02
-0.80
-0.27
-0.25
-0.59d
-0.57d
-0.10
0.09
0.11
0.12e
0.13e
0.18
0.52
0.51
0.57
0.53
0.89
0.75
0.86
0.69
6c
4f
a Calculated from (Ep þ Epox)/2. b An irreversible reduction process also was detected with Epred = -2.02 V. c In the text the processes are labeled as
Pa, Pb, Pc, Pd and Pe, respectively. d This process is a reduction. e This process is a two-electron oxidation. f Not studied with [Bu4N][PF6] as the electrolyte
because of poor solubility.
red
The differences in the E0 values (Table 7) derived for
f
the four successive one-electron oxidation processes
associated with monoanionic 5-, expressed as ΔE1 =
E 0f (2) - E0f (1), ΔE2 = E0f (3) - E0f (2), and ΔE3 =
E 0f (4)-E0f (3) are 0.36, 0.43, and 0.37 V, respectively.
The total separation between the first and fourth oxida-
tion processes is 1.16 V. The comproportionation con-
stants, Kcomp, calculated from these ΔE values and eq 8
log Kcomp ¼ 16:9 ꢀ ΔE at 298 K
ðΔE in VÞ
ð8Þ
are 1.21ꢀ106, 1.85ꢀ 107, and 1.79ꢀ106, respectively. The
similarity of the individual ΔE values and the agreement
with those reported for structurally related polynuclear
osmium-dioxolene complexes supports the hypothesis
that these oxidation processes are all ligand-based and
involve a similar level of electronic interaction.29
Figure 5. (a) Cyclic voltammogram of 5- (2.0 mM in CH2Cl2 with
0.05 M [Bu4N][B(C6F5)4]) obtained with a 1.5 mm diameter glassy carbon
electrode at a scan rate of 100 mV s-1. (b) Square wave voltammogram
obtained under the same conditions with an amplitude of 25 mV and a
frequency of 25 Hz.
Confirmation that one electron is transferred during
each oxidation process and an assessment of the stability
of the first one-electron oxidation product, neutral 5, over
longer time scales was achieved by controlled potential
bulk electrolysis experiments using a Pt-gauze work-
ing electrode. Exhaustive oxidative bulk electrolysis of
5- at Eappl = -0.1 V in dichloromethane with 0.1 M
[Bu4N][PF6] at 295 K led to a slight change in the intensity
of the deep blue color of the solution. Coulometric
analysis of the data indicated that under these conditions
n = 1.00 ( 0.05, with the deep blue color of the solution of
the generated one-electron oxidized product persisting
for at least 30 min. Voltammetry (Figure 6) performed at
steady state conditions (ν = 2.0 mV s-1) before and
immediately after bulk electrolysis indicate that the start-
ing material had converted quantitatively to the oxidized
product, as the same half wave potential and voltam-
metric profiles were observed, differing only in the sign of
the current. Reductive back electrolysis at Eappl = -0.5 V
regenerated 5- in 90 ( 5% yield. UV-visible spectral
monitoring of the one-electron oxidized material was
performed ex situ by syringe removal of samples
before and after bulk electrolysis or in situ by Optically
Transparent Thin-Layer Electrolysis (OTTLE) experi-
ments, conducted at Eappl = -0.1 V at a platinum gauze
electrode (Figure 6b). In both cases, the weak and broad
bands of 5- at 425 and 610 nm (Table 6) decreased in
intensity during the course of electrolysis, with the oxi-
dized material exhibiting four broad bands at 388, 430,
electron transfer processes are fast (electrochemically
reversible) and also monoelectronic. The small deviation
from the value of 56 mV predicted theoretically for a
reversible process is attributed to uncompensated resis-
tance, Ru, and hence Ohmic IRu drop. The ΔEp value for
the fourth oxidation process (ΔEp=115 mV) is slightly
larger than expected for an ideal reversible one-electron
oxidation process, even after allowance for IRu drop. Non
ideality is also evident in the square-wave voltammogram
(Figure 5b) where the magnitude of the peak current for
the fourth process is smaller than for the preceding ones.
Chemical reversibility was readily established for the first
three oxidation processes, with |Ipox/Ipred| values being
close to unity, indicating that the products of oxidation
are persistent on the cyclic voltammetric time scale, even
when using a scan rate of 100 mV s-1. However, fast scan
rates (ν g 1.0 V s-1) are required to improve the reversi-
bility of the fourth oxidation process. On the basis of
these data, it is probable that each electron-transfer step
associated with oxidation of 5- (Scheme 2) involves fast
and reversible oxidation of one of the coordinated 3,5-
DBCat2- ligands to the corresponding 3,5-DBSQ-, with-
out change in the binding modes and positions of the
ligands. However, some rearrangement may be asso-
ciated with the fourth oxidation process. The net result
of the oxidative voltammetric behavior is that the overall
charge on the tetranuclear complex, 5-, increases from
-1 to þ3 (Scheme 2).
(29) Barthram, A. M.; Reeves, Z. R.; Jeffery, J. C.; Ward, M. D. J. Chem.
Soc., Dalton Trans. 2000, 3162.

7776 Inorganic Chemistry, Vol. 48, No. 16, 2009
Mulyana et al.
Figure 7. Cyclic voltammogram of 6 (1.2 mM in CH2Cl2 with 0.05 M
[Bu4N][B(C6F5)4]) obtained with a 1.5 mm diameter glassy carbon
electrode at a scan rate of 100 mV s-1
.
28
Scheme 3.
process. This postulate was quantitatively confirmed
by sequential controlled potential bulk electrolysis at
Eappl=-0.75, 0.3, and 0.75 V, respectively, which showed
that the oxidation process Pc involves the transfer of two
electrons whereas processes Pa and Pd are associated with
single one-electron transfer steps.
Figure 6. (a) Linear sweep voltammograms of 5- (0.7 mM in CH2Cl2
with 0.1 M [Bu4N][PF6]) obtained with a 1.0 mm diameter glassy carbon
electrode at a scan rate of 2.0 mV s-1 before (black curve) and after (red
curve) one-electron bulk electrolysis at Eappl=-0.1 V. (b) Spectroelec-
trochemical monitoring of the changes in the visible spectrum during the
course of the electrolysis.
Although the E0 values for the three oxidation pro-
f
512, and 660 nm. An isosbestic point was observed at
727 nm during the initial stages of electrolysis, but not at
longer times, indicative of a slow structural rearrange-
ment or another form of instability of neutral 5.
cesses [Pc (n=2), Pd (n=1), and Pe (n=1)] of 6 closely
match their counterparts (n=1) in 5-, the E0 value of the
f
first one-electron reduction process (Pa) is dramatically
shifted by about 0.32 V to more negative potential
compared to the first one-electron oxidation process in
5- (i.e., -0.59 for 6 compared to -0.27 V for 5-). On the
Neutral compound 6, which was obtained from the
one-electron chemical oxidation of complex 5-, exhibits
significantly different voltammetric behavior to that
found with the parent complex 5- and its one-electron
electrochemically oxidized product 5, in terms of formal
reversible potentials and other features. This is attributed
to retention of the molecular structure of 5 following
oxidation of 5- on the cyclic voltammetry time scale,
which is in contrast to the significant structural rearran-
gement that accompanies the longer time scale chemical
oxidation of 5- to 6. Thus, 6 displays five reversible
processes (Figure 7), consisting of three oxidations (Pc,
basis of the E0 values of the parent complex 5-, the single
f
two-electron oxidation process (Pc) can be assigned to the
simultaneous oxidation of the 3,5-DBSQ- ligand and one
of the three 3,5-DBCat2- ligands to a 3,5-DBQ ligand and
another 3,5-DBSQ- ligand, respectively (Scheme 3).
Therefore, the two most positive oxidation processes
(Pd and Pe) are assigned to the oxidation of the remaining
two 3,5-DBCat2- ligands to the 3,5-DBSQ- forms. On
the reduction side, the first process (Pa) is attributed to the
one-electron reduction of neutral 6 to an anionic species
Pd, and Pe) with E0 values of 0.12, 0.57, and 0.86 V,
6-. Cyclic and linear sweep voltammetry(ν= 1-2 mV s-1
)
f
respectively, and two reductions (Pa and Pb) with E0
=
recorded after one-electron reductive bulk electrolysis of
6 provides no evidence for reconversion of the coordinated
3,5-DBSQ- ligand in 6 back to the 3,5-DBCat2- form in
6-. This finding suggests that this ligand-based reduction
process does not involve change in the coordination modes
and/or the structureatleast on this experimental time scale.
Since there is no further ligand-based reduction possible at
the reversible potential of the second reduction process
(Pb), we assign this process to the metal-based Fe(III) to
Fe(II) reduction.
f
-0.59 and -1.03 V (Table 7). The assignment of oxida-
tion or reduction was achieved from the sign of the
current (positive = oxidation, negative = reduction) of
steady-state voltammograms obtained with a rotating
disk electrode (not shown). Detailed analysis of the
voltammetric data reveals that the magnitude of the
current associated with the first oxidation step (Pc) is
close to double that of both the second oxidation (Pd) and
the first reduction process (Pa), which in turn have similar
current per unit concentration values to that of the one-
electron oxidation of 5-, for experiments undertaken
under the same conditions. Thus, process Pc is assigned
as a two-electron rather than one-electron oxidation
The striking differences in the electrochemical proper-
ties of the chemically and electrochemically generated
one-electron oxidation products of 5-, that is, 5 and 6,
suggest that the two methods of oxidation produce

Article
Inorganic Chemistry, Vol. 48, No. 16, 2009 7777
28
Scheme 4.
and 6 are structural isomers, with 5 the kinetic product of
one-electron oxidation of 5-, while 6 is the thermody-
namic product.
Voltammetry of a solution of complex 4 reveals two
well-defined oxidation processes (Supporting Informa-
tion, Figure S2), as expected from the presence of two
oxidizable 3,5-DBCat2- ligands (Scheme 4). On sweeping
the potential in the reverse direction, a new reduction
process is seen at Epred = -0.40 V. This peak is not
observed upon switching the potential immediately after
the first oxidation process (0.1 V), indicating that it is
associated with the second oxidation process. The chemi-
cal reversibility of the second oxidation process dramati-
cally improves as the scan rate is increased (black curve in
Supporting Information, Figure S2), suggesting that the
electron-transfer step for the second process is accompa-
nied by a moderately fast follow-up chemical reaction.
This reaction may involve structural rearrangement of the
coordinated 3,5-DBCat2- ligands from a μ2-bridging
mode to a chelating mode following the one-electron
oxidation and conversion to the 3,5-DBSQ- ligands, as
is observed in the formation of 6 following one-electron
oxidation of 5-. The apparently enhanced rate of struc-
tural rearrangement of 4 versus 5- may be associated with
the presence of only two 3,5-DBCat2- ligands in 4
compared to four in 5-, which, in the latter, would be
more efficient in stabilizing the isostructural oxidized
species.
Figure 8. Square wave voltammograms recorded during the chemical
oxidation of 5- (0.0037 mmol in CH2Cl2 with 0.1 M [Bu4N][PF6]) with
1.0 equiv of [Cp2Fe][PF6] over a period of 7 h, obtained with a 1.5 mm
diameter glassy carbon electrode at a frequency of 15 Hz.
structurally different compounds. To confirm this hypo-
thesis, chemical oxidation of 5- (0.0037 mmol) with 1.0
equiv of [Cp2Fe][PF6] was undertaken in an electroche-
mical cell containing 1.5 mL of dichloromethane with
0.1 M [Bu4N][PF6] as the electrolyte. The progress of the
reaction was monitored voltammetrically, which revealed
the initial generation of the same one-electron oxidation
product, 5, obtained via bulk electrolysis (identical E0
f
values and voltammetric characteristics). However, vol-
tammetric monitoring over a period of 7 h, which mimics
the time scale associated with synthesis and crystalliza-
tion of the chemically oxidized product 6, led to detection
of the voltammetric features associated with 6. Thus as
shown in Figure 8, the processes with E0f values of -1.30
and -0.25 V in 5- are fully replaced by processes with E0
f
values of -1.0 and -0.57 V (similar to processes Pb and
Pa, respectively, for 6) whereas the process at E0f = 0.11 V
shifts to more negative potential and grows with time to a
new process (Pc) that overlays that for ferrocene. The two
most positive oxidation processes seen initially in 5- are
slightly changed with time upon oxidation with
[Cp2Fe][PF6], but after 7 h their potentials match those
of processes Pd and Pe in 6. These results clearly indicate
that chemical and electrochemical oxidation of 5- by one-
electron initially produce the same product, 5, which
upon prolonged periods of time undergoes a major
structural rearrangement to yield 6 as the final product.
This conclusion, reached on the basis of voltammetry, is
supported by the identical nature of the UV-visible
spectra obtained for 6 and for the electrochemically
generated one-electron oxidized species, 5, when left for
an extended period of time. The combined electrochemi-
cal and spectroscopic results from the solution studies are
consistent with the significant structural differences be-
tween 5- and 6 in the solid state, as determined by X-ray
crystallography. The redox-initiated structural rearran-
gement involves breaking and reforming five Fe-O
and Fe-N bonds to alter the binding positions and
modes (Figure 2) of one of the coordinated Schiff-base
L2- ligands (from a μ3-mode in 5- and presumably 5,
to a μ4-mode in 6) and one of the o-dioxolene ligands
(from a μ2-bridging 3,5-DBCat2- in 5- and presumably 5,
to a chelating-only 3,5-DBSQ- ligand in 6). Complexes 5
The structural difference between 4 and 5- is also
evident from the marked difference in their reversible
potentials. Thus, the E0 values of the two oxidation
f
processes of 4, measured at a scan rate of 100 mV s-1
(E0 (1) = -0.10, E0 (2) = 0.18 V) differ by þ0.170 and
f
f
þ0.090 V, respectively, to more positive values relative to
the first two oxidation processes of 5- (Table 1). Similarly,
the metal-based Fe(III) to Fe(II) reduction occurs at E0 =
f
-0.80 V, which is significantly more positive than the
value of -1.36 V obtained for the Fe(III) to Fe(II)
reduction of 5-. The electronic effect that arises from
effectively replacing two 3,5-DBCat2- ligands in 5- with
two extra phenolate Schiff base ligands (L2-) in 4 has a
similar effect on the E0 as is the case when an electron
f
withdrawing substituent is removed from the catechol
ring. This can be understood given that the oxidation will
be thermodynamically more facile when electron density
flows into the catechol ring, thereby leading to stabiliza-
tion of the positively charged cationic species generated
during the oxidation processes. Thus, the overall thermo-
dynamic stabilization of the oxidized products of 4, ex-
pressed as ΔE=0.28 V and Kcomp=5.4 ꢀ 104 (determined
using eq 8), is significantly lower than that for 5-.
Magnetochemistry. Variable temperature magnetic sus-
ceptibility data collected at 0.1 T are plotted in Figures 9-11
and Supporting Information, Figure S8 for compounds 3,
4, 5a, 6, 7, and 8a, respectively, together with the fits of the

7778 Inorganic Chemistry, Vol. 48, No. 16, 2009
Mulyana et al.
Scheme 6
S=2, and S=3 excited states less than 10 cm-1 higher
in energy.
The magnetization isotherms at 2.0 and 3.0 K approach
saturation at 1 T (Supporting Information, Figure S3),
before rising steeply without saturation up to 5 T. This
curve shape is indicative of a poorly isolated ground state
with various close lying states of higher spin, a situation
arising from the multiple pairwise antiferromagnetic
interactions in the given spin topology. The failure to
reproduce both χMT at the lowest measured temperatures
and the isothermal magnetization curves above 0.5 T with
simulations calculated using the parameters obtained
from fitting the susceptibility data results from the overly
simplified model employed to obtain this fit. Data at
lower temperatures and an extended model with addi-
tional exchange parameters would be necessary to deter-
mine the complex energy level scheme for 3.
Figure 9. Plot of χMT and χM vs T for 3. The solid lines are fits as
described in the text.
Scheme 5
For 4 (Figure 10a), χMT decreases steadily from a value
of 10.1 cm3 mol-1 K at 300 K to 0.1 cm3 mol-1 K at 2.0 K.
The room temperature value is substantially less than the
calculated spin-only value of 17.5 cm3 mol-1 K for four
non-interacting Fe(III) centers with g = 2.0 and the
behavior suggests overall antiferromagnetic interactions
and an S = 0 ground state. Over the same temperature
range, χM increases gradually before reaching a broad
maximum at about 120 K and decreasing gradually as the
temperature is decreased to 5 K. Below 5 K, χM increases
rapidly as the temperature is decreased, which is attrib-
uted to the presence of a mononuclear Fe(III) paramag-
netic impurity. The coupling scheme appropriate to a
tetranuclear complex with a planar rhombic core, such as
4, is depicted in Scheme 6.
experimental data for 3, 4, 5a, 7, and 8a and a simulation of
the experimental data for 6. Variable temperature magne-
tization data in the temperature range 2-5.5 K with
magnetic fields up to 5 T are plotted in the Supporting
Information, Figures S3-S7 and S9.
For 3 (Figure 9), χMT decreases gradually from 18.1 cm3
mol-1 K at 300 K to 2.2 cm3 mol-1 K at 2.0 K. The room
temperature value is considerably lower than the spin only
value of 26.3 cm3 mol-1 K calculated for six non-interact-
ing high spin Fe(III) centers with g=2.0. This behavior is
consistent with competing antiferromagnetic interactions
of considerable magnitude. The coupling scheme appro-
priate to the magnetic core of 3 is presented in Scheme 5,
with the low symmetry of the complex requiring seven
different coupling constants.
The symmetry of the structure of 4 affords J2 = J4 and
J3 = J5, while the similarity in the Fe1 Fe2 and
This coupling scheme gives rise to the exchange
Hamiltonian:
3 3 3
Fe1 Fe20 pathways (Table 3) suggests that J2 = J3 =
J4 = J5 is a reasonable approximation. Thus the data
between 2.0 and 300 K were fit to the exchange Hamiltonian:
3 3 3
^
Hex ¼ -2J1ðS1 S2Þ -2J2ðS1 S3Þ -2J3ðS1 S4Þ
3
3
3
-2J4ðS2 S5Þ -2J5ðS2 S6Þ -2J6ðS3 S4Þ -2J7ðS5 S6Þ
3
3
3
3
^
Hex ¼ -2J1ðSB SDÞ -2J2ðSA SB þSB SC
3
3
3
ð9Þ
þSC SD þSA SDÞ
ð10Þ
3
3
To fit the magnetic susceptibility data to the theoretical
expression, simplifications of the model were necessary to
avoid the overparametrization of the fit. No satisfactory
fit of the data could be obtained with a one J model (J1=
J2=J3=J4=J5=J6=J7) or two J models incorporating
the pseudo inversion center of 3 (e.g., J1, J2 = J3 =
J4=J5 =J6=J7). However, a model using three J values
J1=-26.9 cm-1, J3=J4= -6.3 cm-1 and J2=J5=J6=
J7 = -3.0 cm-1 with g fixed to 2.0 reproduced the
experimental data very well (Figure 9). Small devia-
tions from the experimental data were observed only
at the lowest temperatures. The J values obtained give
rise to an S = 0 ground state with multiple S=0, S=1,
with g fixed to 2.0 and the inclusion of a term to account
for a monomeric Fe(III) impurity. Efforts to fit the data
with J1 = J2 did not yield a good fit and the best fit for
J1 ¼ J2 was obtained with J1 = 0 cm-1, J2 = -9.3 cm-1
and 0.6% monomeric Fe(III) S =5/2 paramagnetic im-
purity (solid lines in Figure 10a). This gives rise to a well-
isolated S = 0 ground state with an S = 1 first excited
state 19 cm-1 above the ground state. Low temperature
magnetization measurements (Supporting Information,
Figure S4) are also consistent with a well-isolated S=0
ground state and a small amount of paramagnetic im-
purity.

Article
Inorganic Chemistry, Vol. 48, No. 16, 2009 7779
Scheme 7
fit was obtained with J1 = -3.5 cm-1 and J2 = -7.0 cm-1
(Figure 10b). The adequacy of this fit attests to the validity
of the approximations made and the incorporation of
additional J values was not pursued to avoid overparame-
terizing the solution. The fitting parameters obtainedagain
give rise to an S = 0 ground state, this time with an S = 1
first excited state 14 cm-1 above the ground state. Low
temperature magnetization measurements (Supporting
Information, Figure S5) are also consistent with a well-
isolatedS = 0 ground state and no paramagnetic impurity.
For 6 (Figure 10c), the value of χMT at 300 K is 10.5 cm3
mol-1 K, which is considerably lower than the value of
17.9 cm3 mol-1 K calculated four non-interacting high
spin Fe(III) (S = 5/2) centers with a 3,5-DBSQ- radical
(S= 1/2) ligand, suggesting significant antiferromagnetic
interactions. As the temperature is decreased the value of
χMT decreases steadily to reach 0.37 cm3 mol-1 K at 2.0 K,
which suggests an S=1/2 ground state for the complex.
Over the same temperature range, χM increases gradually to
reach a broad maximum at 90 K before decreasing and then
rapidly increasing below 10 K. The coupling scheme appro-
priate to complex 6 is depicted in Scheme 7.
This scheme is derived from that applied to the structu-
rally related complexes 4 and 5-, although it is extended to
include the radical 3,5-DBSQ- ligand. This ligand chelates
Fe2 (vide supra) and does not bridge to any of the other
Fe centers; thus, the possibility of magnetic coupling
between the radical ligand and the other Fe centers has
been discounted. Again the similarity in the Fe1 Fe2,
Fe1 Fe3, Fe2 Fe4, and Fe3 Fe4 pathways
3 3 3
3 3 3
3 3 3
3 3 3
(Table 3) suggests that J2 =J4 =J5 =J6 is a reasonable
approximation. Thus the exchange Hamiltonian
^
Hex ¼ -2J1ðSB SDÞ -2J2ðSA SB þSB SC þSC SD
3
3
3
3
þSA SDÞ -2J3ðSC SEÞ
ð11Þ
3
3
was employed to simulate the experimental susceptibility
data. For this simulation g was fixed to 2.0, which is
appropriate for both high spin Fe(III) S = 5/2 centers
and the organic 3,5-DBSQ- (S = 1/2) radical. The cou-
pling constant between Fe2 and the radical ligand, J3, was
fixed to -600 cm-1, while J1 and J2 were varied system-
atically. This value of -600 cm-1 for J3 was obtained from
the least-squares fit of the magnetic susceptibility data for
the structurally related complex [Fe(salen)(3,5-DBSQ)], in
which a 3,5-DBSQ- ligand chelates an Fe(III) center in a
similar manner to that observed in 6.30 Extremely strong
antiferromagnetic coupling has been observed in this
and other high spin Fe(III) complexes of 3,5-DBSQ-.31
Figure 10. Plot of χMT and χM vs T for (a) 4, (b) 5a, and (c) 6. The solid
lines are fits or simulations as described in the text.
The thermal variation of the magnetic susceptibility of
compound 5a (Figure 10b) is similar to that observed for 4
(Figure 10a). For 5a, χMT decreases steadily from a value
of 11.7 cm3 mol-1 K at 300 K to 0.01 cm3 mol-1 K at 2.0 K.
The maximum in χM occurs at 85 K, and there is no low
temperature increase, indicating no paramagnetic impur-
ity. The behavior is again consistent with overall antifer-
romagnetic interactions and an S = 0 ground state for
complex 5-. The lack of symmetry in 5- does not allow the
approximations made in the fitting of the data for 4;
however for simplicity, initial attempts to fit the data
between 2.0 and 300 K commenced with the two J model
described by the exchange Hamiltonian (eq 10). With g
again fixed to 2.0, efforts tofit the data with J1 = J2 did not
give a good fit to the experimental data. However, a good
(30) Kessel, S. L.; Emberson, R. M.; Debrunner, P. G.; Hendrickson, D.
N. Inorg. Chem. 1980, 19, 1170.
(31) (a) Buchanan, R. M.; Kessel, S. L.; Downs, H. H.; Pierpont, C. G.;
Hendrickson, D. N. J. Am. Chem. Soc. 1978, 100, 7894. (b) Cohn, M. J.; Xie,
C. L.; Tuchagues, J. P. M.; Pierpont, C. G.; Hendrickson, D. N. Inorg. Chem.
1992, 31, 5028.

7780 Inorganic Chemistry, Vol. 48, No. 16, 2009
Mulyana et al.
The thermal variation of the magnetic susceptibility of
8a (Supporting Information, Figure S7) is very similar
to that observed for 7. For 8a, χMT again decreases
steadily from a value of 6.5 cm3 mol-1 K at 300 K to
0.03 cm3 mol-1 K at 2.0 K. The maximum in χM occurs at
70 K, and the low temperature increase again suggests
a small amount of paramagnetic impurity. The data
between 2.0 and 300 K were fit to the exchange Hamilto-
nian (eq 12) with g fixed to 2.0 and the inclusion of a term
to account for a monomeric Fe(III) impurity. The best fit
was obtained with J1 =-8.5 cm-1 and 0.3% paramag-
netic impurity (solid lines in Supporting Information,
Figure S8). This gives rise to a well-isolated S = 0 ground
state with an S=1 first excited state 17 cm-1 above the
ground state. Low temperature magnetization measure-
ments (Supporting Information, Figure S9) are again
consistent with a well-isolated S=0 ground state and a
small amount of paramagnetic impurity.
The single crystal X-ray structures for solvated com-
pounds 3, 5a, 6, 7, and 8a were determined at 130 K, while
the structural data for solvated compound 4 were col-
lected at 295 K. In every case all Fe-O and Fe-N bond
lengths, and the resulting bond valence sum calculations,
are consistent with Fe(III) centers at these temperature
(vide supra). The magnetic susceptibility data for these
compounds were measured between 2 and 300 K, and the
data are consistent with high spin Fe(III) centers across
this temperature range. There is no evidence for any low
spin Fe(III) contribution, which would be unusual for the
ligand fields associated with the {O6} or {O5N} donor sets
that are present in the present complexes. In addition,
there is no magnetic evidence for the presence of any
Fe(II) centers, nor any indication of any thermally in-
duced valence tautomeric transitions from Fe(III)-cate-
cholate at low temperature to Fe(II)-semiquinone at high
temperature.
Figure 11. Plot of χMT and χM vs T for 7. The solid lines are fits as
described in the text.
An excellent reproduction of the χM and χMT versusT data
was obtained with J1 = -4.0 cm-1, J2 = -7.5 cm-1, J3 =
-600 cm-1, and g = 2.0 (solid line in Figure 10c). It is
noteworthy that the values of J1 and J2 are very close to the
equivalent coupling constants obtained for complex 5-,
which has a very similar magnetic core. These parameters
give rise to an S=1/2 ground state for complex 6 with an
S=3/2 first excited state that is 24 cm-1 above the ground
state. Variable temperature magnetization data between
2 and 5.5 K are plotted in Supporting Information, Figure
S6. The isotherms for 2.0, 3.0, 4.0, and 5.5 K are essentially
superimposable, and the magnetization value of 1.0 Nβ at
2.0 K and 5.0 T is consistent with an S=1/2 ground state
with g = 2.0. However, a slight deviation at large values of
H/T from the Brillouin function calculated for an S=1/2
system with g=2.0 may indicate the presence of S>1/2
excited states lying at lower energy than the S=3/2 first
excited state calculated using the J values derived from the
susceptibility fit.
For 7 (Figure 11), χMT decreases steadily from a value
of 6.5 cm3 mol-1 K at 300 K to 0.03 cm3 mol-1 K at 2.0 K.
The room temperature value is less than the calculated
spin-only value of 8.8 cm3 mol-1 K for two non-interact-
ing Fe(III) centers with g = 2.0, and the behavior is
consistent with antiferromagnetic coupling between the
two Fe centers and an S = 0 ground state. Over the same
temperature range, χM increases gradually before reach-
ing a broad maximum at about 60 K and decreasing
gradually as the temperature is decreased to 4 K. Below
4 K, χM increases rapidly as the temperature is decreased,
suggesting the presence of a paramagnetic impurity. The
data between 2.0 and 300 K were fit to the exchange
Hamiltonian:
Magnetostructural Correlations. A useful magnetos-
tructral correlation has been developed recently by
Christou et al. for polynuclear ferric complexes with
oxo, hydroxo, and alkoxo bridges.32 This was in part
derived from earlier models for dinuclear ferric complexes
proposed by Gorun and Lippard and subsequently Weihe
::
and Gudel.33 The Christou model encompasses both the
radial and angular dependence of the exchange constant,
such that the interaction becomes more strongly antifer-
romagnetic as the Fe-O distance and the Fe-O-Fe
angle increase. The angular dependence is accentuated
as the Fe-O distance decreases and the radial dependence
is more important at larger Fe-O-Fe angles.
Table 8 presents the coupling constants determined
from fitting or simulating the magnetic susceptibility data
for compounds 3, 4, 5a, 6, 7, and 8a, together with the
values calculated using the Christou model. These calcu-
lated values employ the Fe-O distances and Fe-O-Fe
angles derived from the crystallographic data. In general
there is very good agreement between the experimentally
derived and calculated values, bearing in mind that sim-
plified coupling schemes were employed for the present
^
Hex ¼ -2JðS1 S2Þ
ð12Þ
3
withg fixed to2.0 and the inclusion of a term to account for
a monomeric Fe(III) impurity. The best fit was obtained
with J1 = -8.0 cm-1 and 0.4% paramagnetic impurity
(solid lines in Figure 11). This gives rise to a well-isolated
S=0 ground state with an S = 1 first excited state 16 cm-1
above the ground state. Low temperature magnetization
measurements (Supporting Information, Figure S7) are
consistent with a well-isolated S =0 ground state and a
small amount of paramagnetic impurity.
(32) Canada-Vilalta, C.; O0Brien, T. A.; Brechin, E. K.; Pink, M.;
Davidson, E. R.; Christou, G. Inorg. Chem. 2004, 43, 5505.
(33) (a) Gorun, S. M.; Lippard, S. J. Inorg. Chem. 1991, 30, 1625.
::
(b) Weihe, H.; Gudel, H. U. J. Am. Chem. Soc. 1997, 119, 6539.

Article
Inorganic Chemistry, Vol. 48, No. 16, 2009 7781
Table 8. Fe Fe Exchange Interactions for Compounds 3, 4, 5a, 6, 7, and 8a
3 3 3
J (cm-1
)
compound
linkage
from fit or simulation of susceptibility data
calculateda
3
(μ2-OR)2
(μ2-OR)(μ3-OR)
(μ2-OR)(μ3-OR)(μ2-OAc)
(μ3-OR)2
(μ2-OR)(μ3-OR)
(μ3-OR)2
(μ2-OR)(μ3-OR) or (μ2-OR)(μ3-OR)(μ2-OAc)
(μ3-OR)2
-26.9
-3.0, -6.3
-3.0
0
-9.3
-3.5
-7.0
-4.0
-7.5
-8.0
-8.5
-9.2
-7.2, -7.6, -9.3, -10.1
-8.4, -10.5
4
-2.8
-9.1, -11.2
5a
6
-3.9
-6.4, -7.5, -8.6, -11.6
-3.6
(μ2-OR)(μ3-OR) or (μ2-OR)(μ3-OR)(μ2-OAc)
(μ2-OR)2
(μ2-OR)2
-6.4, -7.1, -10.7, -11.5
7
8a
-9.8
-7.8
a Calculated according to reference 32.
complexesand thatthe calculatedvaluesforall compounds
use interatomic parameters for the fully solvated forms of
the compounds, whereas the magnetic susceptibility mea-
surements were performed on desolvated samples.
complexes revealed the initial formation in solution of a
kinetic product, with apparent retention of the structure,
followed by structural rearrangement to a thermodynamic
product, which could be isolated in the solid state and
structurally characterized. This oxidation-induced structural
rearrangement involves breaking and reforming a number of
metal-ligand bonds and appears to be driven by the pre-
ference of the semiquinone ligand, obtained from the oxida-
tion of a catechol ligand, to bind in a non-bridging mode.
Magnetochemical studies of the compounds obtained
indicate behavior dominated by antiferromagnetic pairwise
exchange interactions between the ferric centers. The mag-
netic susceptibility data for the single complex that possesses
a semiquinone ligand suggest a very strong antiferromagnetic
interaction between the radical ligand and the iron center to
which it is bound. The coupling constants determined from
fitting or simulating the thermal variation of the magnetic
susceptibility for the family of complexes are consistent with
magnetostructural correlations that have been reported pre-
viously for alkoxo-bridged ferric complexes.
Conclusions
A trinuclear acetate-containing ferric complex has proved
a useful precursor for a series of polynuclear complexes of
o-dioxolene ligands that display a variety of nuclearities and
structural architectures. Although a distorted monovacant
cubane is a structural fragment common to a number of these
complexes, the substituents on the o-dioxolene ligands clearly
play a significant structure directing role, with steric crowd-
ing apparently preventing the formation of complexes of
nuclearity higher than four when the o-dioxolene ligands are
substituted with bulky tert-butyl groups. The redox state
of the o-dioxolene ligands also affects the structure of
the complexes, with all of the catecholate ligands employed
observed to bridge the metal centers, regardless of the
substituents. However, the smaller electron density on the
semiquinonate form of the ligand appears to lead to a
preference for binding in a non-bridging capacity.
Although the iron centers in these complexes are redox
active, electrochemical studies indicated that the richest
redox behavior is associated with the o-dioxolene ligands.
Of particular interest in this regard are the three tetranuclear
complexes, which despite their possession of a common
structural core, differ in the number and redox state of the
o-dioxolene ligands. A detailed investigation of the chemical
and electrochemical one-electron oxidation of one of these
Acknowledgment. We thank the Australian Research
Council for financial support.
Supporting Information Available: X-ray crystallographic
files in CIF format for 2 Et2O, 3 CH2Cl2 Et2O, 4 4CH2Cl2,
5a 5CH2Cl2, 6 4C7H8 C6H14, 7 4Et2O, and 8a 2CH2Cl2. In-
3
3
3
3
3
3
3
3
3
frared spectra for compounds 5a and 6. Plot of χMT and χM vs T
for 8a. Plots of M/Nβ vs H/T for compounds 3, 4, 5a, 6, 7, and
8a. This material is available free of charge via the Internet at
http://pubs.acs.org.