Inorg. Chem.

2007, 46, 756−766

Odd-Numbered FeIII Complexes: Synthesis, Molecular Structure,

Reactivity, and Magnetic Properties
Ayuk M. Ako,† Oliver Waldmann,*,‡ Valeriu Mereacre,† Frederik Klo1 wer,† Ian J. Hewitt,†
Christopher E. Anson,† Hans U. Gu1 del,‡ and Annie K. Powell*,†

Institut für Anorganische Chemie der UniVersität Karlsruhe, Engesserstrasse Geb. 30.45,
D-76128 Karlsruhe, Germany, and Department of Chemistry and Biochemistry,
UniVersity of Bern, Freiestrasse 3, CH-3012 Bern, Switzerland

Received August 31, 2006

Three isostructural disklike heptanuclear FeIII compounds of the general formula [FeIII7(µ3-O)3(L)3(µ-O2CCMe3)6-
(η1-O2CCMe3)3(H2O)3], where L represents a di- or triethanolamine moiety, display a three-blade propeller topology,
with the central Fe atom representing the axle or axis of the propeller. This motif corresponds to the theoretical
model of a frustrated Heisenberg star, which is one of the very few solvable models in the area of frustrated
quantum-spin systems and can, furthermore, be converted to an octanuclear cage for the case where L is
triethanolamine to give [FeIII8(µ4O)3(µ4-tea)(teaH)3(O2CCMe3)6(N3)3]‚1/2MeCN‚1/2H2O or [FeIII8(µ4O)3(µ4-tea)(teaH)3(O2-
CCMe3)6(SCN)3]‚2MeCN when treated with excess NaN3 or NH4SCN, respectively. The core structure is formally
derived from that of the heptanuclear compounds by the replacement of the three aqua ligands by an {Fe(tea)}
moiety, so that the 3-fold axis of the propeller is now defined by two central FeIII atoms. Magnetic studies on two
of the heptanulcear compounds established unequivocally S ) 5/2 spin ground state for these complexes, consistent
with overall antiferromagnetic interactions between the constituent FeIII ions.

Introduction The thoroughly studied [Mn12O12(OAc)16(H2O)4]3c and [Fe8O2-

(OH)12(tacn)6]Br3d clusters respectively represent prototypes
In the last few decades, the synthesis of polynuclear
of Mn and Fe complexes showing SMM behavior.
compounds of magnetic transition-metal ions1 has become
Another reason for continued interest in the synthesis and
the focus of intense research activities because of their
study of polynuclear paramagnetic 3d metal clusters is the
relevance to biological systems2 and their potential as
aesthetically pleasing nature of some of these species.6 A
molecular species that can function as nanoscale magnetic
particularly intriguing class are the so-called molecular ferric
particles or so-called single-molecule magnets, SMMs.3 For
example, the iron-storage protein ferritin and manganese (4) (a) Taft, K. L.; Delfs, C. D.; Papaefthymiou, G. C.; Foner, S.; Gatteschi,
D.; Lippard, S. J. J. Am. Chem. Soc. 1994, 116, 823. (b) Crichton, R.
clusters have been extensively studied as a biological R. Angew. Chem., Int. Ed. Engl. 1973, 12, 57. (c) Taft, K. L.;
prototype in the inorganic systems4 and as models of the Papaefthymiou, G. C.; Lippard, S. J. Science 1993, 259, 1302.
water-oxidizing complex in photosystem II,5 respectively. (5) (a) Wieghardt, K. Angew. Chem., Int. Ed. Engl. 1994, 33, 725. (b)
Debus, R. J. Biochim. Biophys. Acta 1992, 1102, 269. (c) Bruvig, G.
W.; Thorp, H. H.; Crabtree, R. H. Acc. Chem. Res. 1991, 24, 311.
* To whom correspondence should be addressed. E-mail: (6) (a) Tasiopoulos, A. J.; Vinslava, A.; Wernsdorfer, W.; Abboud, K.
oliver.waldmann@iac.unibe.ch (O.W.), powell@aoc.uni-karlsruhe.de (A.K.P.). A.; Christou, G. Angew. Chem., Int. Ed. 2004, 43, 2117. (b) Low, D.
Tel: +49-721-608-2135 (A.K.P.). Fax: +49-721-608-8142 (A.K.P.). M.; Jones, L. F.; Bell, A.; Brechin, E. K.; Mallah, T.; Rivière, E.;
† Institut für Anorganische Chemie der Universität Karlsruhe. Teat, S. J.; McInnes, E. J. L. Angew. Chem., Int. Ed. 2004, 42, 3781.
‡ University of Bern. (c) Langley, S.; Helliwell, M.; Raftery, J.; Tolis, E. I.; Winpenny, R.
(1) Müller, A.; Peters, F.; Pope, M. T.; Gatteschi, D. Chem. ReV. 1998, E. P. Chem. Commun. 2004, 142. (d) Jones, L. F.; Brechin, E. K.;
239 and references cited therein. Collison, D.; Harrison, A.; Teat, S. J.; Wernsdorfer, W. Chem.
(2) (a) Lippard, S. J. Angew. Chem., Int. Ed. Engl. 1988, 27, 344. (b) Commun. 2002, 2974. (e) Paine, T. K.; Rentschler, E.; Weyhermüller,
Pope, M. T.; Müller, A. Angew. Chem., Int. Ed. Engl. 1991, 30, 34. T.; Chaudhuri, P. Eur. J. Inorg. Chem. 2003, 3167. (f) Overgaard, J.;
(c) Katsoulis, D. E. Chem. ReV. 1998, 98, 359. Iversen, B. B.; Palii, S. P.; Timco, G. A.; Gerbeleu, N. V.; Larsen, F.
(3) (a) Gatteschi, D.; Sessoli, R. Angew. Chem., Int. Ed. 2003, 42, 268. K. Chem.sEur. J. 2002, 8, 2775. (g) Abedin, T. S. M.; Thompson,
(b) Christou, G.; Gatteschi, D.; Hendrickson, D. N. MRS Bull. 2000, L. K.; Miller, D. O.; Krupicka, E. Chem. Commun. 2003, 708. (h)
25, 66. (c) Sessoli, R.; Gatteschi, D.; Caneschi, A.; Novak, M. A. Murugesu, M.; Anson, C. E.; Powell, A. K. Chem. Commun. 2002,
Nature 1993, 365, 141. (d) Barr, A. L.; Debrunner, P.; Gatteschi, D.; 1054. (i) Abati, G. L.; Cornia, A.; Fabretti, A. C.; Caneschi, A.;
Schulz, C. E.; Sessoli, R. Europhys. Lett. 1996, 35, 133. Gatteschi, D. Inorg. Chem. 1998, 37, 1430.

756 Inorganic Chemistry, Vol. 46, No. 3, 2007 10.1021/ic061650s CCC: $37.00 © 2007 American Chemical Society
Published on Web 01/05/2007
Odd-Numbered FeIII Complexes

wheels with nuclearity Fex ranging from 6 to 18 FeIII ions.7,8 Chart 1. teaH3, H3L1 (Left), bdeaH2, H2L2 (Center), and Phdea, H2L3
(Right)
The magnetization of the ferric wheels exhibits steplike field
dependencies at low temperatures due to the importance of
finite-size quantum effects.4a,9 Initially, these systems were
regarded as ideal models for the finite-size version of the
linear 1D Heisenberg chain, but subsequently it was realized
that their magnetic behavior at low temperatures is best
described by a quantized rotation of the Néel vector and,
hence, is more similar to that of 2D and 3D materials.10
Moreover, quantum phenomena such as Néel-vector tunnel- a high spin (S ) 29/2) hepatanuclear mixed-valent FeII,III
ing were demonstrated.11 Because of the rich physics wheel, which showed SMM behavior.15
observed in molecular wheels and a promising potential for In contrast to the large number of even-numbered FeIII
quantum computing,12 this class of molecules has become wheels known, which have been obtained from the reaction
increasingly important. of suitable chelating ligands with preformed oxo-centered
Alkoxo ligands, for example, N-substituted diethanola- FeIII triangles,7e,16 odd-numbered, disklike, all-FeIII complexes
mines, have been employed extensively in the construction are rare.17 Such FeIII complexes are interesting not only from
of hybrid organic-inorganic clusters because of their a chemical and structural perspective but also because
versatile bridging modes, with subtle changes in the reaction elucidation of their magnetic properties and spin topologies
conditions resulting in the formation of unexpected and should help in understanding the magnetic properties of larger
beautiful structures such as cyclic polymetallic clusters.13 ferric systems. The arrangement in such a heptanuclear
Along these lines, Saalfrank and co-workers reported on the structure can be regarded as a ring comprising an even
template-mediated self-assembly of six- and eight-membered number of FeIII ions with an additional central FeIII ion, which
iron coronates {Na[Fe6(tea)6]}Cl (6) and {Cs[Fe8(tea)8]}Cl can be predicted to allow for strong quantum-spin frustration
(7), obtained from the reaction of triethanolamine (teaH3) effects due to competing nearest-neighbor exchange interac-
with FeCl3, NaH, or Cs2CO3.7b In addition, they reported on tion paths. In fact, the exchange-coupling situation realized
a wide range of six-membered iron coronands obtained from in such clusters bears a close analogy to the so-called
the reaction of N-substituted diethanolamines with FeCl3 and frustrated Heisenberg star introduced by Richter and Voigt
CaH2 in tetrahydrofuran.14 Recently, using Schiff base ligands as a rare example of an exactly solvable quantum-spin model
with an alkyl alcohol group, Oshio and co-workers reported with frustration.18
Hence, as an extension of our recent work on the reaction
(7) (a) Sydora, O. L.; Woleczanski, P. T.; Lobkovsky, E. B. Angew. Chem., of diethanolamine ligands with preformed oxo-centered FeIII
Int. Ed. 2003, 42, 2685 and references cited therein. (b) Saalfrank, R. triangles,19 we report here on the reaction of teaH3, N-
W.; Bernt, I.; Uller, E.; Hampel, F. Angew. Chem., Int. Ed. Engl. 1997,
36, 2482. (c) Caneschi, A.; Cornia, A.; Fabretti, A. C.; Gatteschi, D. butyldiethanolamine (bdeaH2), and phenyldiethanolamine
Angew. Chem., Int. Ed. 1999, 38, 1295. (d) Watton, S. P.; Fuhrmann, (Phdea) (Chart 1) with [Fe3O(O2CCMe3)6(H2O)3](O2CCMe3)
P. L.; Pence, E.; Caneschi, A.; Cornia, A.; Abbati, G. L.; Lippard, J.
S. Angew. Chem., Int. Ed. Engl. 1997, 36, 2774. (e) Cañada-Vilalta,
in the presence of metal(II) pivalates (metal ) nickel, copper)
C.; O’Brien, T. A.; Brechin, E. K.; Pink, M.; Davidson, E. R.; Christou, or Ni(OAc)2‚4H2O, which results in the formation of the
G. Inorg. Chem. 2004, 43, 5505. (f) Jones, L. F.; Low, D. M.; heptanuclear FeIII clusters 1-3, respectively, in high yields.
Helliwell, M.; Raftery, J.; Collison, D.; Aromı́, G.; Cano, J.; Mallah,
T.; Wernsdorfer, W.; Brechin, E. K.; McInnes, E. J. L. Polyhedron The crystal structures of compounds 1-3 and the magnetic
2006, 25, 325. properties of complexes 1 and 2 are discussed. We report
(8) Yao, H. C.; Wang, J. J.; Ma, Y. S.; Waldmann, O.; Du, W. X.; Song,
Y.; Li, Y. Z.; Zheng, L. M.; Decurtins, S.; Xin, X. Q. Chem. Commun. here also the structure of the octanuclear clusters [FeIII8(µ4O)3-
2006, 1745. (µ4-tea)(teaH)3(O2CCMe3)6(N3)3] (4) and FeIII8(µ4O)3(µ4-
(9) (a) Cornia, A.; Affronte, M.; Jansen, A. G. M.; Abbati, G. L.; Gatteschi, tea)(teaH)3(O2CCMe3)6(SCN)3] (5) and the mononuclear
D. Angew. Chem., Int. Ed. 1999, 38, 2264. (b) Waldmann, O.;
Schülein, J.; Koch, R.; Müller, P.; Bernt, I.; Saalfrank, R. W.; Andres, complex [Ni(teaH3)2](O2CCMe3)2 (6), obtained from the
H. P.; Güdel, H. U.; Allenspach, P. Inorg. Chem. 1999, 38, 5879. (c) reaction of 1 with excess NaN3, or NH4SCN, and as a
Waldmann, O.; Koch, R.; Schromm, S.; Schülein, J.; Müller, P.; Bernt,
I.; Saalfrank, R. W.; Hampel, F.; Balthes, E. Inorg. Chem. 2001, 40, byproduct of the synthetic procedure used for 1, respectively.
2986.
(10) (a) Waldmann, O. Phys. ReV. B 2002, 65, 024424. (b) Waldmann, (14) Saalfrank, R. W.; Deutscher, C.; Sperner, S.; Nakajima, T.; Ako, A.
O.; Guidi, T.; Carretta, S.; Mondelli, C.; Dearden, A. L. Phys. ReV. M.; Uller, E.; Hampel, F.; Heinemann, F. W. Inorg. Chem. 2004, 43,
Lett. 2003, 91, 237202. (c) Waldmann, O. Coord. Chem. ReV. 2005, 4372.
249, 2550. (15) Oshio, H.; Hoshino, N.; Ito, T.; Nakano, M.; Renz, F.; Gütlich, P.
(11) Waldmann, O.; Dobe, C.; Mutka, H.; Furrer, A.; Güdel, H. U. Phys. Angew. Chem., Int. Ed. 2003, 42, 223 and references cited therein.
ReV. Lett. 2005, 95, 057202. (16) (a) Murugesu, M.; Abboud, K. A.; Christou, G. Dalton Trans. 2003,
(12) (a) Troiani, F.; Affronte, M.; Carretta, S.; Santini, P.; Amoretti, G. 4552. (b) Jones, L. F.; Jensen, P.; Moubaraki, B.; Cashion, J.; Berry,
Phys. ReV. Lett. 2005, 94, 190501. (b) Troiani, F.; Ghirri, A.; Affronte, K. J.; Murray, K. S. Dalton Trans. 2005, 3344.
M.; Carretta, S.; Santini, P.; Amoretti, G.; Piligkos, S.; Timco, G.; (17) While this manuscript was in preparation, Murray et al. reported the
Winpenny, R. E. P. Phys. ReV. Lett. 2005, 94, 207208. structure of compound 1: Jones, L. F.; Jensen, P.; Moubaraki, B.;
(13) (a) Caneschi, A.; Cornia, A.; Lippard, S. J. Angew. Chem., Int. Ed. Berry, K. J.; Boas, J. F.; Pilbrow, J. R.; Murray, K. S. J. Mater. Chem.
Engl. 1995, 34, 467. (b) Watton, S. P.; Fuhrmann, P.; Pence, L. E.; 2006, 16, 2690.
Caneschi, A.; Cornia, A.; Abbati, G. L.; Lippard, S. J. Angew. Chem., (18) (a) Richter, J.; Voigt, A. J. Phys. A: Math. Gen. 1994, 27, 1139. (b)
Int. Ed. Engl. 1997, 36, 2774. (c) Abbati, G. L.; Cornia, A.; Fabretti, Richter, J.; Voigt, A.; Krüger, S. E.; Gross, C. J. Phys. A: Math.
A. C.; Caneschi, A.; Gatteschi, D. Inorg. Chem. 1998, 37, 3759. (d) Gen. 1996, 29, 825.
Dearden, A. L.; Parsons, S.; Winpenny, R. E. P. Angew. Chem., Int. (19) Ako, A. M.; Mereacre, V.; Hewitt, I. J.; Clérac, R.; Lecren, L.; Anson,
Ed. 2001, 40, 151. C. E.; Powell, A. K. J. Mater. Chem. 2006, 16, 2579.

Inorganic Chemistry, Vol. 46, No. 3, 2007 757

 Ako et al.
Experimental Section The resulting dark-brown solution was stirred under ambient
conditions for a further 1 h, filtered, and then allowed to stand
General Procedures. Unless otherwise stated, all reagents were
undisturbed in a sealed vial. Dark-brown hexagonal prisms of 1
obtained from commercial sources and were used as received,
suitable for X-ray crystallography were obtained overnight. The
without further purification. [Fe3O(O2CCMe3)6(H2O)3](O2CCMe3),20
crystals of 1 were maintained in the mother liquor for X-ray
[Cu2(µ-OH2)(O2CCMe3)4(HO2CCMe3)4], and [Ni2(µ-OH2)(O2-
crystallography or collected by filtration, washed with Et2O, and
CCMe3)4(HO2CCMe3)4]21 were synthesized as described elsewhere.
dried in a slow flow of N2. Yield: 30.2 mg, 40% based on Fe.
All reactions were carried out under aerobic conditions.
Anal. Calcd for C63H126Fe7N3O33 (found): C, 41.02 (41.21); H,
The elemental analyses (CHN) were performed using an El-
6.88 (6.69); N, 2.27 (2.22). Selected IR data (KBr disk, cm-1):
ementar Vario EL analyzer. Fourier transform IR spectra were
3447 (b), 2959 (s), 2870 (s), 1567 (s), 1484 (s), 1423 (s), 1359
measured on a Perkin-Elmer Spectrum One spectrometer with
(m), 1228 (s), 1103 (s) 1077 (s), 1039 (s), 921 (s), 789 (s), 685
samples prepared as KBr disks.
(w), 604 (w), 495 (m), 435 (w). The same product is obtained in
X-ray Crystallography. Data were collected at 100 K on a
ca. 30% yield if [Cu2(µ-OH2)(O2CCMe3)4(HO2CCMe3)4] is em-
Bruker SMART Apex CCD diffractometer (1-4 and 6) or at 150 K
ployed in place of [Ni2(µ-OH2)(O2CCMe3)4(HO2CCMe3)4]. Crystals
on a Stoe IPDS II area detector diffractometer (5) using graphite-
of these were obtained after 48 h. The material was identified by
monochromated Mo KR radiation. Crystallographic details and
IR spectral comparison with the title compound.
details of the data collection are summarized in Table 1. Semiem-
A week after the isolation of 1, a few blue crystals were collected
pirical absorption corrections were made using SADABS22a (1-4)
from the mother liquor and identified by an X-ray structure analysis
or XPREP in SHELXTL22b (5 and 6). The structures were solved
(see the Supporting Information) to be the mononuclear complex
using direct methods, followed by full-matrix least-squares refine-
[Ni(teaH3)2](O2CCMe3)2 (6). Anal. Calcd for C22H48N2NiO10
ment against F 2 (all data) using SHELXTL.22b Anisotropic refine-
(found): C, 47.24 (47.32); H, 8.65 (8.54); N, 5.01 (5.22). Selected
ment was used for all ordered non-H atoms; organic H atoms were
IR data (KBr disk, cm-1): 3348 (b), 2974 (s), 2880 (m), 2452 (b),
placed in calculated positions, while coordinates of hydroxo H
2068 (b), 1912 (b), 1527 (b), 1384 (w), 1225 (s), 115 (s), 1062
atoms were refined.
(m), 1002 (s), 903 (s), 790 (s), 601 (s), 429 (w). No corresponding
Magnetic Measurements. The magnetic susceptibility versus
copper byproduct was isolated when copper pivalate was employed.
temperature and the magnetization curves at low temperatures were
measured with a Quantum Design SQUID magnetometer MPMS- Synthesis of [FeIII7(µ3-O)3(bdea)3(µ-O2CCMe3)6(η1-O2CCMe3)3-
XL. The samples used for the magnetic measurements were (H2O)3] (2). The synthesis was similar to that of 1 but replacing
prepared in three different ways. Procedure A: Polycrystalline [Ni2(µ-OH2)(O2CCMe3)4(HO2CCMe3)4] with Ni(OAc)2‚4H2O and
material was prepared by drying crystals under ambient conditions. teaH3 (H3L1) with N-butyldiethanolamine (bdeaH2; H2L2). Orange
About 10 mg of material was encapsulated in Saran foil, put into crystals of 2 were obtained after 72 h, washed with Et2O, and dried
a plastic straw, and held in place with a cotton thread. The weight in a slow flow of N2. Yield: 24 mg, 30% based on Fe. Anal. Calcd
of the sample could be accurately determined to within 0.1%. The for C69.5H138Fe7N3O30 (found): C, 43.42 (43.18); H, 7.24 (7.43);
magnetic data were corrected for the background due to the Saran N, 2.19 (2.21). Selected IR data (KBr disk, cm-1): 3519 (b), 2960
foil. Procedure B: A total of 20-40 crystals were taken out of the (s), 2927 (m), 2866 (s), 1575 (b, s), 1483 (s), 1423 (s), 1407 (m),
mother liquor, put directly into Apiezon grease, and “stirred”. Then 1368 (m), 1355 (m), 1103 (s), 1063 (s), 1049 (s), 1006 (s), 906 (s),
as much grease as possible was removed, and the sample was 785 (s), 680 (b, s), 621 (s), 438 (w). The corresponding mononuclear
weighed (typically 2 mg). Procedure C: A single crystal was taken Ni byproduct was obtained after compound 2 was isolated and the
out of the mother liquor and then treated as it was in procedure B mother liquor was allowed to stand for a further 2 days.
(ca. 1.2 mg). Synthesis of [FeIII7(µ3-O)3(Phdea)3(µ-O2CCMe3)6(η1-O2CCMe3)3-
Because of the unknown amount of grease left over in procedures (H2O)3] (3). The synthesis was similar to that of 1 but replacing
B and C, the measured weight of a sample provides an upper limit [Ni2(µ-OH2)(O2CCMe3)4(HO2CCMe3)4] with Ni(OAc)2‚4H2O and
for the actual weight of the magnetic material, but the accuracy is teaH3 (H3L1) with phenyldiethanolamine (Phdea; H2L3). Orange
estimated to better than 10%. Also, the magnetic data cannot be crystals of 3 were obtained after 1 week, washed with Et2O, and
corrected for the diamagnetic (temperature-independent) contribu- dried in a slow flow of N2. Yield: 27.8 mg, 35% based on Fe.
tion of the grease; its contribution is an estimated 5% at 300 K but Anal. Calcd for C75H126Fe7N3O30 (found): C, 46.42 (46.38); H,
is completely negligible at temperatures below 10 K. The advantage 6.54 (6.45); N, 2.17 (2.21). Selected IR data (KBr disk, cm-1):
of these procedures is that they avoid potential problems with 3509 (b), 2960 (s), 2927 (m), 2858 (m), 1571 (s), 1483 (s), 1425
decomposition and typically minimize contamination by magnetic (s), 1406 (s), 1376 (m), 1355 (s), 1228 (s), 1107 (m), 1059 (w),
impurities. 1049 (s), 1009 (w), 925 (w), 788 (w), 697 (m), 677 (s), 602 (s),
Synthesis of [FeIII7(µ3-O)3(teaH)3(µ-O2CCMe3)6(η1-O2CCMe3)3- 591 (s), 439 (w).
(H2O)3] (1). To a stirred solution of [Ni2(µ-OH2)(O2CCMe3)4(HO2- Synthesis of [FeIII8(µ4O)3(µ4-tea)(teaH)3(O2CCMe3)6(N3)3] (4).
CCMe3)4] (0.09 g, 0.1 mmol) and [Fe3O(O2CCMe3)6](O2CCMe3) The synthesis was similar to that of 1, however, in the presence of
(0.1 g, 0.1 mmol) in MeCN (20 mL) was gradually added a solution NaN3 (0.07 g, 1 mmol). The reaction mixture was stirred at room
of triethanolamine (teaH3; H3L1; 0.3 g, 2 mmol) in MeCN (5 mL). temperature for 2 h to yield an orange solution, then filtered, and
allowed to evaporate slowly at room temperature. Brown crystals
(20) Wilson, C.; Iversen, B. B.; Overgaard, J.; Larsen, F. K.; Wu, G.; Palii, of 4 were obtained after 48 h, washed with Et2O, and dried in a
S. P.; Timco, G. A.; Gerbeleu, N. V. J. Am. Chem. Soc. 2000, 122, slow flow of N2. Yield: 28 mg, 40% based on Fe. Anal. Calcd for
11370.
(21) Chaboussant, G.; Basler, R.; Güdel, H.-U.; Ochsenbein, S.; Parkin,
C55H107.5Fe8N413.5O27.5 (found): C, 35.81 (35.67); H, 5.85 (5.96);
A.; Parsons, S.; Rajaraman, G.; Sieber, A.; Smith, A. A.; Timco, G. N, 10.25 (10.14). Selected IR data (KBr disk, cm-1): 3453 (b),
A.; Winpenny, R. E. P. Dalton Trans. 2004, 2758. 2959 (s), 2903 (m), 2873 (m), 2060 (s), 1575 (b, s), 1484 (s), 1426
(22) (a) Sheldrick, G. M. SADABS (the Siemens Area Detector Absorption (s), 1361 (w), 1227 (s), 1087 (m), 1075 (m), 1039 (w), 1004 (w),
Correction); University of Göttingen: Göttingen, Germany, 1996. (b)
Sheldrick, G. M. SHELXTL, version 5.1; Bruker AXS, Inc.: Madison, 909 (s), 786 (s), 603 (b), 425 (w). The same compound can be
WI, 1997. obtained directly by refluxing compound 1 with excess NaN3 in

758 Inorganic Chemistry, Vol. 46, No. 3, 2007

Odd-Numbered FeIII Complexes

Table 1. Crystallographic Data and Structure Refinement for Complexes 1-6

1 2 3
formula C63H126Fe7N3O33 C69.5H138ClFe7N3O30 C75H126Fe7N3O30
Mr 1844.62 1922.23 1940.74
cryst size [mm] 0.27 × 0.25 × 0.24 0.28 × 0.22 × 0.18 0.26 × 0.17 × 0.17
shape hexagonal prism block block
color dark brown orange orange
cryst syst trigonal monoclinic monoclinic
space group P31c C2/c P21/n
T [K] 100 100 100
a [Å] 15.5804(8) 63.351(3) 10.6609(5)
b [Å] 15.5804(8) 10.4329(5) 31.5950(14)
c [Å] 20.8694(11) 31.4090(16) 27.2343(11)
R [deg] 90 90 90
β [deg] 90 115.078(1) 90.120(1)
γ [deg] 120 90 90
V [Å3] 4387.3(4) 18802.3(16) 9173.3(7)
Z 2 8 4
Fcalcd [g cm-3] 1.523 1.358 1.405
µ(Mo KR) [mm-1] 1.202 1.150 1.151
F(000) 1942 8120 4076
reflns collected 16 786 46 786 46 053
unique reflns 5060 21 285 20 567
Rint 0.0295 0.0329 0.0253
reflns obsd [I > 2σ(Ι)] 4823 14323 14887
parameters/restraints 326/4 1044/24 1087/10
GOF on F 2 1.051 0.986 1.036
R1 [I > 2σ(I)] 0.0338 0.0380 0.0457
wR2 (all data) 0.0869 0.0822 0.1380
largest residuals [e Å-3] +0.90/-0.33 +1.08/-0.56 +1.5/-0.70
CCDC number 618914 618915 618916

4 5 6
formula C55H107.5Fe8N13.5O27.5N3 C61H111Fe8N9O27S3 C22H48N2NiO10
Mr 1844.84 1945.57 559.33
cryst size [mm] 0.34 × 0.28 × 0.25 0.32 × 0.21 × 0.13 0.37 × 0.18 × 0.08
shape octahedron block plate
color brown brown pale blue
cryst syst monoclinic monoclinic monoclinic
space group P21/c C2/c P21/n
T [K] 100 150 100
a [Å] 23.6121(8) 55.495(2) 12.9960(14)
b [Å] 15.0975(5) 15.4971(9) 6.9495(7)
c [Å] 44.6731(14) 21.1361(9) 16.1600(17)
R [deg] 90 90 90
β [deg] 94.523(1) 110.992 102.368(2)
γ [deg] 90 90 90
V [Å3] 15875.6(9) 16969.9(14) 1425.6(3)
Z 8 8 2
Fcalcd [g cm-3] 1.544 1.523 1.303
µ(Mo KR) [mm-1] 1.501 1.477 0.732
F(000) 7680 8096 604
reflns collected 79 377 56 154 6904
unique reflns 35 692 17 443 3196
Rint 0.0391 0.0458 0.0268
reflns obsd [I > 2σ(Ι)] 20 937 11 437 2386
parameters/restraints 1851/11 972/10 172/0
GOF on F 2 0.978 1.012 1.002
R1 [I > 2σ(I)] 0.0504 0.0516 0.0299
wR2 (all data) 0.1276 0.1254 0.0649
largest residuals [e Å-3] +1.92/-1.18 +0.74/-0.81 +0.56/-0.29
CCDC number 618917 618918 618919

CH2Cl2/MeOH (1: 3) for 2 h. Crystals of 4 were obtained in obtained in 30% yield when [Cu2(µ-OH2)(O2CCMe3)4(HO2-
quantitative yield after 24 h. The product was identified using IR CCMe3)4] was used in place of the Ni analogue. Reaction
spectroscopy. Complex 5 was synthesized in a manner similar to of H2L2 with Ni(OAc)2‚4H2O and [Fe3O(O2CCMe3)6-
that of 4, except that NH4SCN was used in place of NaN3.
(H2O)3](O2CCMe3) in MeCN gave orange blocks of the
Results and Discussion heptanuclear cluster 2‚1/2CH2Cl2 in 25% yield, while a similar
reaction using H2L3 as the ligand gave 3 in 35% yield
Synthesis. Reaction of a 10-fold excess of H2L1 with
[Fe3O(O2CCMe3)6(H2O)3](O2CCMe3) and [Ni2(µ-OH2)(O2- (Scheme 1).
CCMe3)4(HO2CCMe3)4] in MeCN afforded dark-brown Compounds 1-3 can be prepared directly from the
crystals of 1 in 40% yield. Alternatively, complex 1 was reaction of the corresponding ligands with the Fe3O pivalate

Inorganic Chemistry, Vol. 46, No. 3, 2007 759

 Ako et al.
Scheme 1. Synthetic Pathways for the Synthesis of Compounds 1-5a nuclearity cluster, consisting of 22 FeIII
ions, the largest
homometallic Fe cluster reported to date.26 Other high-
nuclearity FeIII complexes with non-wheel-like structures and
odd numbers of Fe centers include, for example, the Fe11
cluster with an S ) 11/2 spin ground state, reported by
McInnes et al.,7f and the Fe17 and Fe19 clusters with S ) 25/2
and 33/2 ground states.27 Statistical analysis of recent reports
of FeIII clusters obtained from preformed oxo-centered Fe
triangles as a source of FeIII ions shows that more even-
numbered clusters, generally with S ) 0, have been obtained
a Reaction conditions: (i) [M (µ-OH )(piv) (pivH) ] (M ) Ni, Cu),
2 2 4 4
MeCN; (ii) [Ni2(µ-OH2)(piv)4(pivH)4]/NaN3 (4) or -/NH4SCN (5), MeCN;
than odd-numbered FeIII clusters.
(iii) Ni(OAc)2‚4H2O, MeCN [L ) bdea (2), Phdea (3)]. Compound 1 lends itself to further elaboration, for
example, replacement of terminal pivalates or H2O molecules
triangle. However, the products are obtained after an with pseudohalides. Indeed, introducing excess NaN3 or NH4-
extended reaction time and in lower yields, as was recently SCN into the reaction mixture that produced 1 resulted in
reported for compound 1.17 To our knowledge, complex 3 the isolation of the octanuclear clusters 4 or 5, respectively,
represents the first high-nuclearity Phdea-based compound in high yields. Similarly, when crystalline material obtained
reported so far. Elemental analysis, IR, and single-crystal from the synthesis of 1 was treated directly with excess NaN3
X-ray crystallography (see the Experimental Section) were or NH4SCN, compounds identical with 4 or 5 (as established
used to characterize the compounds. Contrary to what we by IR spectra) were obtained in quantitative yield. The IR
expected, the MII ions are not incorporated in the products. spectra of 4 and 5 exhibit a very strong absorption at 2060
However, a mononuclear Ni complex (6) was isolated as a or 2062 cm-1 corresponding to the azido as well as
byproduct when nickel pivalate or nickel acetate was used thiocyanate asymmetric vibrations,28 respectively. It is
in conjunction with the FeIII source. Although no such noteworthy that heptanuclear FeIII7 clusters with different core
crystalline byproduct was obtained when copper(II) pivalate topologies have been reported, obtained from the reaction
was employed, we conjecture from the nature of the of cyclohexenephosphonic acid (hydrothermally)29 or phen-
crystalline Ni complex formed as the byproduct that these ylphosphonate ligands (at ambient conditions)24 with pre-
MII starting materials provide a means of driving the reaction formed iron(III) carboxylate triangles [Fe3O(O2CR)6(H2O)3]X
equilibria toward the side of the products through the (R ) Ph, Me; X ) Cl, NO3, PhCO2-). To our knowledge,
favorable chelate and ligand-field effects in forming MII the core topology shown by 1-3 is rare for odd-numbered
complexes with the ethanolamine ligands. FeIII clusters obtained via oxo-centered FeIII triangles as a
Previous work on the complex formation of FeIII with source of FeIII ions.17
amino alcohol ligands includes a similar reaction using Description of Crystal Structures of 1-3. Complexes
teaH3,16a diisopropanolamine,16b and phosphonate ligands23 1-3 are analogous, differing only in the nature of the side
or 2-(2-hydroxyethyl)pyridine7e with [Fe3O(O2CR)6(H2O)3]X arm of the ligand, and are among the first examples of this
(R ) Ph, CCMe3; X ) O2CCMe3, O2CPh, NO3), which result disklike heptanuclear motif in all-ferric form. They crystallize
in the formation of even-numbered hexa- or octanuclear FeIII in the space groups P31c for 1, C2/c for 2, and P21/n for 3,
clusters that have S ) 0 spin ground states.7e,12 Equally, the with the Fe7 molecules on general positions. The structure
reaction of the tripodal ligand 1,1,1-tris(hydroxymethyl)- for 2 is shown in Figure 1, while selected interatomic
ethane (H3thme) with [Fe3O(O2CCMe3)6(H2O)3](X) (X ) O2- distances and angles are listed in Table 2. Varying the side
CCMe3, Cl) under solvothermal conditions led to octanuclear arm of the diethanolamine ligands, ethanol (for 1), n-butyl
[FeIII8(O2CCMe3)12(thme)4] or nonanuclear [FeIII9O4(O2- (for 2), and the rigid aromatic ring (for 3) clearly has little
CCMe3)13(thme)2] clusters.7f Meanwhile, the reaction of impact on the way the ligands associate in the FeIII clusters,
benzotriazole (btah) with [Fe3O(O2CR)6(H2O)3]Cl in MeOH and the driving force for the arrangement of the metal ions
under the same solvothermal conditions gave the tetradeca- in the complexes is presumably the bite of the ONO donor
metallic high-spin S ) 23 FeIII cluster [Fe14(bta)6O6- set of the ligands as well as the associated pivalates and H2O
(OMe)18Cl6].6b Other groups have also demonstrated that molecules. Contrary to what might have been anticipated,
diethanolamine ligands form wheel-shaped structures. In this the pendent ethanol arm of teaH3 is not implicated in the
regard, neutral iron coronands [Fe6X6L6] or [Fe6L6] (L ) coordination of the ligand through its supplementary donor
N-substituted diethanolamine; X ) Cl, F) obtained from the O atom of the side arm. Hence, teaH3 behaves essentially as
reaction of diethanolamines with simple metal salts (FeX3) a tridentate ligand and not a tetradentate ligand. The core of
have been reported.24,25 Furthermore, this class of ligands
recently has been used to assemble an unprecedented high- (26) Foguet-Albiol, D.; Abboud, K. A.; Christou, G. Chem. Commun. 2005,
4282.
(23) Tolis, E. I.; Helliwell, M.; Langley, S.; Raftery, J.; Winpenny, R. E. (27) Powell, A. K.; Heath, S. L.; Gatteschi, D.; Pardi, L.; Sessoli, R.; Spina,
P. Angew. Chem., Int. Ed. 2003, 42, 3804. G.; Del Giallo, F.; Pieralli, F. J. Am. Chem. Soc. 1995, 117, 2491.
(24) Rumberger, E. M.; Zakharov, L. N.; Rheingold, A. L.; Hendrickson, (28) (a) Barandika, M. G.; Cortés, R.; Lezama, L.; Urtiaga, M. K.; Arriortua,
D. N. Inorg. Chem. 2004, 43, 6531. M. I.; Rojo, T. J. Chem. Soc., Dalton Trans. 1999, 2971.
(25) Koizumi, S.; Nihei, M.; Nakano, M.; Oshio, H. Inorg. Chem. 2005, (29) Yao, H.-C.; Li, Y.-Z.; Zheng, L.-M.; Xin, X.-Q. Inorg. Chim. Acta
44, 1208. 2005, 358, 2523.

760 Inorganic Chemistry, Vol. 46, No. 3, 2007

Odd-Numbered FeIII Complexes

Figure 1. (Top) Molecular structure of 2 in the crystal. (Bottom) Core of 2. Color code: green, Fe; red, O; blue, N; black, C; white, H.

1-3 is rare for wheel-like diethanolamine complexes.17 It peripheral Fe centers designated as type A [Fe(2), Fe(4), and
is also directly related to that of the Fe8 clusters 4 and 5 Fe(6)] comprise an FeNO5 coordination sphere made up of
(see later). The structural features for the heptanuclear two µ2-O2- ethanolamine ligand O atoms, one ligand N, one
compounds 1-3 are described with reference to the detailed µ3-O2-, and two O atoms from the bridging and terminal
structural description of 2. The structure of 2 consists of an pivalate ligands, respectively; (iii) the peripheral Fe centers
[FeIII7(µ3-O)3]15+ core, with the peripheral ligation provided designated type B [Fe(3), Fe(5), are Fe(7)] comprise an FeO6
by six µ2-bridging pivalate ligands, three bridging nBu- chromophore with one µ2-O2- ligand O atom, one µ3-O2-,
N(CH2CH2O)22- ligands, three terminal pivalate ligands two O atoms from two bridging pivalate ligands, one O atom
bonding in a unidentate fashion, and three terminal H2O from a terminal pivalate ligand, and one O atom from a
molecules coordinated to the Fe(3), Fe(5), and Fe(7) atoms. terminal aqua ligand. Hence, the pivalate ligands exhibit two
All FeIII centers are hexacoordinated, displaying a distorted different coordination modes; bidentate bridging and terminal
octahedral coordination sphere. Three types of FeIII centers monodentate and adjacent peripheral Fe centers are bridged,
can be distinguished based on their coordination environ- on the one hand, by a µ3-O2- connecting pairs of A and B
ment: (i) the central atom Fe(1), the coordination sites of centers to the central Fe and, on the other hand, by two
which are occupied by three µ3-O2- and three µ2-O2- from bidentate pivalate ligands. Hydrogen bonding from the aqua
the ethanolamine ligand to give an FeO6 donor set; (ii) the ligands plays an important role in stabilizing the aggregates,

Inorganic Chemistry, Vol. 46, No. 3, 2007 761

 Ako et al.
Table 2. Selected Bond Distances (Å) and Angles (deg) for Complex 2 similar for all three samples, but the absolute values differ
widely. Apparently, despite careful sample handling, for this
Bond Lengths
Fe(1)-O(4) 1.9944(16) Fe(4)-O(2) 1.8863(16) material we faced problems in properly determining absolute
Fe(1)-O(2) 2.0023(16) Fe(4)-N(2) 2.236(2) magnetic values. Nevertheless, some conclusions concerning
Fe(1)-O(1) 2.0066(16) Fe(5)-O(2) 1.8577(16) the magnetism in 1 can be drawn. At 250 K, the value of χT
Fe(1)-O(3) 2.0085(16) Fe(6)-O(3) 1.8819(15)
Fe(2)-O(1) 1.8878(16) Fe(6)-N(3) 2.238(2) is in the range of 7-11 emu K mol-1 and much smaller
Fe(2)-O(4) 1.9722(16) Fe(7)-O(3) 1.8621(16) than the value of 30.64 emu K mol-1 expected for seven
Fe(2)-O(5) 2.0101(16) Fe(1)‚‚‚Fe(6) 2.9653(5) noninteracting high-spin FeIII ions (spin 5/2, g ) 2). This
Fe(2)-N(1) 2.243(2) Fe(1)‚‚‚Fe(4) 2.9686(5)
Fe(3)-O(1) 1.8597(15) Fe(1)‚‚‚Fe(2) 2.9751(5) indicates very strong overall antiferromagnetic interactions
Bond Angles in the cluster, as suggested also by the decrease of χT with
O(2)-Fe(1)-O(1) 90.34(6) Fe(6)-Fe(1)-Fe(2) 119.037(16) temperature. At low temperatures, χT levels off at values of
O(4)-Fe(1)-O(3) 102.57(7) Fe(4)-Fe(1)-Fe(2) 118.271(15) 3.5-6 emu K mol-1, with a further steep drop at the very
O(2)-Fe(1)-O(3) 90.51(7) Fe(3)-O(1)-Fe(2) 118.93(8)
O(1)-Fe(1)-O(3) 89.68(6) Fe(3)-O(1)-Fe(1) 134.57(9) lowest temperatures. This behavior demonstrates a magnetic
O(1)-Fe(2)-O(4) 81.21(7) Fe(2)-O(1)-Fe(1) 99.58(7) ground state of the cluster with a total spin of S ) 2, 5/2, or
O(1)-Fe(2)-N(1) 59.57(7) Fe(5)-O(2)-Fe(4) 119.13(9) 3 (the expected χT values are 3.001, 4.377, or 6.002 emu K
O(4)-Fe(2)-N(1) 78.93(7) Fe(5)-O(2)-Fe(1) 134.90(8)
N(1)-Fe(2)-Fe(1) 117.89(5) Fe(4)-O(2)-Fe(1) 99.49(7) mol-1, respectively, for g ) 2). The steep drop in χT at the
N(2)-Fe(4)-Fe(1) 117.86(6) Fe(7)-O(3)-Fe(6) 119.37(8) lowest temperatures could be due to zero-field splitting. The
N(3)-Fe(6)-Fe(1) 118.95(5) Fe(7)-O(3)-Fe(1) 134.03(8) magnetic ground state is unambiguously confirmed by the
Fe(6)-Fe(1)-Fe(4) 119.168(15) Fe(6)-O(3)-Fe(1) 99.27(7)
magnetization curves at 1.8 K, which approach values of
with each such ligand forming one hydrogen bond to the O 2.3-3.8 NAµB at high fields, suggesting a spin ground state
atom of the next aqua ligand in a cyclic manner and the in the range of S ) 1-2.
other H forming a hydrogen bond to the noncoordinated O Although the reason for the inconsistent absolute values
atom of a monodentate pivalate, as illustrated in Figure 1. of the magnetic curves is as yet unclear, the similarity of
Description of the Crystal Structures of 4 and 5. the curves for the three samples suggests that the actual
Complex 4 crystallizes in the space group P21/n, while the number of magnetic molecules in the samples is somehow
isostructural complex 5 crystallizes in the space group C2/ overestimated (one possibility could be, for example, that
c. The structure of 4 is shown in Figure 2. The core struc- 20-50% of the clusters lack the central Fe ion, resulting in
ture of 4 and 5 has been previously reported as occurring S ) 0 and, hence, a nonmagnetic ground state of these
in only three compounds and is derived from that of 1-3. clusters). With this assumption, the low-temperature χT value
The octanuclear motif occurs as one of the units of the as well as the saturation magnetization msat can be determined
core of the largest homometallic FeIII cluster [Fe22],26 in the by S and a scaling factor n e 1, which corrects for a smaller
Fe8 cluster [FeIII8O3(O2CPh)9(tea)(teaH)3]‚MeCN,16a and in number of magnetic molecules in the samples (χT ) nS(S
the Fe8 cluster [FeIII8O3(µ4-tea)(teaH)3(O2C(CH2)Me)6F3]‚ + 1)/2; msat ) n(2S); g ) 2). For the three samples, n will
0.5MeOH‚0.5H2O recently reported by Murray et al.17 The be different, but for each sample, the two values of n
core consists of eight FeIII ions bridged by four µ4-O2- ions. determined independently from χT and msat should agree for
It can be described as a central Fe2 pair bridged by three the correct spin ground state S. Indeed, assuming S ) 5/2,
oxide µ4-O2- ions, each of which also bridges an outer Fe2 the n values determined from χT and msat agree to within
pair. These outer Fe pairs are not coplanar, and thus the outer 5% for each of the three samples, while they differ by more
six Fe ions do not form a flat wheel. Instead, the three outer than 17% assuming S ) 2 and by more than 9% assuming
Fe pairs are again arranged like the blades of a propeller, S ) 3 (and even worse for other S values). Thus, one can
with the central pair now representing the axle or axis of conclude that complex 1 experiences very strong overall
the propeller. The Fe8 structure can be regarded as derived antiferromagnetic interactions, resulting in a magnetic cluster
from that of the Fe7 aggregates by the addition of an {FeIII- ground state with a total spin that most likely is S ) 5/2.
(tea)} moiety, in which the chelating O atoms of the (tea)3- Figure 4 presents the temperature dependence of χT and
ligand have coordinated to Fe(1) forming µ2-O bridges, the magnetization curve at 1.8 K for compound 2. For this
replacing the three aqua ligands as shown in Figure 2. material, we also studied several samples prepared by either
Because the three µ3-O ligands in 1-3 now also coordinate procedure A or B using different batches, but here the
to the extra Fe center, they become µ4-O2- ligands. The Fe8 measurements agreed within experimental error for all
structures are completed by the replacement of the terminal samples. The magnetic behavior is similar to that observed
monodentate pivalates by azide (4) or thiocyanate (5) ligands. for 1. At 250 K, χT assumes a value of 6.2 emu K mol-1,
Two types of teaH3 ligands are thus now present, three much smaller than the value of seven uncoupled S ) 5/2 FeIII
doubly deprotonated and one triply deprotonated. ions, indicating very strong antiferromagnetic interactions.
Magnetic Properties of 1 and 2. Figure 3 shows the With decreasing temperature, χT decreases slightly, in
temperature dependence of χT and the magnetization curves agreement with the presence of strong antiferromagnetic
at 1.8 K for three samples of compound 1. Sample #1 was interactions. At low temperatures, χT approaches a value of
prepared by procedure A, and samples #2 and #3 were both 4.4 emu K mol-1, consistent with an S ) 5/2 cluster ground
prepared by procedure B but from different batches. The state (4.375 emu K mol-1 for g ) 2). At the very lowest
functional dependencies of the measured curves are very temperatures, χT shows a slight upturn, which could be due
762 Inorganic Chemistry, Vol. 46, No. 3, 2007
Odd-Numbered FeIII Complexes

Figure 2. (Top) Molecular structure of 4 in the crystal. H atoms have been omitted for clarity. (Bottom) Side views of the cores in 2 (left) and 4 (right)
illustrating their structural relationship.

to a small amount of paramagnetic impurities and/or very temperature dependence of χT and the magnetization curves
weak intermolecular magnetic interactions. The magnetiza- at 1.8 K are presented in Figure 5. The data clearly
tion curve at 1.8 K unambiguously confirms a paramagnetic demonstrate a significant magnetic anisotropy of the easy-
ground state with a total spin of S ) 5/2. axis type due to zero-field splitting in the ground state of 2.
For compound 2, we also measured a single crystal, as Figure 6 shows the magnetization curves of the crushed
prepared by procedure C, with the magnetic field applied sample as measured at temperatures of 1.8, 2.4, 3.5, 5.0, and
parallel to the (approximate) C3 symmetry axis of the clusters 7.5 K, plotted as function of B/T (B is the magnetic field).
and perpendicular to it. After completion of these measure- All curves collapse onto each other. Because of the very
ments, the crystal was crushed into pieces to yield a strong antiferromagnetic interactions in the cluster at tem-
polycrystalline sample and measured. The results for the peratures below 50 K only the ground state is thermally

Inorganic Chemistry, Vol. 46, No. 3, 2007 763

 Ako et al.

Figure 3. (a) Temperature dependence of the magnetic susceptibility,

plotted as χT. (b) Low-temperature magnetization curves of three different
polycrystalline samples of compound 1 (for the preparation details of the
samples, see the text). The solid lines are guides to the eye. Figure 5. (a) Temperature dependence of the magnetic susceptibility,
plotted as χT. (b) Low-temperature magnetization curves of a single-crystal
sample of compound 2 for the magnetic field applied parallel and
perpendicular to the cluster C3-symmetry axis and for the crystal crushed
into pieces. The colored solid lines are guides to the eye. The black lines
represent the best-fit curves obtained from ĤS)5/2.

Figure 6. Magnetization curves at the indicated temperatures, plotted as

a function of B/T, of a crushed single-crystal sample of compound 2. The
black lines (which are mostly covered by the data points) represent the
best-fit curves obtained from ĤS)5/2 (the very small difference at the highest
fields could be improved by assuming slightly smaller g factors in the
calculations; it is recalled that the best-fit procedure used here included
both the magnetization and χT curves; a fit to the magnetization curves
alone would produce perfect-fitting curves).
Figure 4. (a) Temperature dependence of the magnetic susceptibility,
plotted as χT, as determined from different magnetic fields. (b) Low- zero-field splitting, is in seeming disagreement with the
temperature magnetization curve of a polycrystalline sample of compound
2. The solid lines are guides to the eye. significant magnetic anisotropy visible in Figure 5, but, in
fact, this is not the case, as will be shown below.
populated, as confirmed by the near temperature indepen- The low-temperature magnetism of 2 may be described
dence of χT. Hence, the scaling behavior of the magnetization as that of a single S ) 5/2 spin with easy-axis anisotropy,
curves, which is usually interpreted as evidence for negligible corresponding to the spin Hamiltonian
764 Inorganic Chemistry, Vol. 46, No. 3, 2007
Odd-Numbered FeIII Complexes

1
[ ]
ĤS)5/2 ) D Ŝz2 - S(S + 1) + g⊥ µB(ŜxBx + ŜyBy) +
3
g|µBŜzBz (1)

where the z axis is perpendicular to the plane of the molecule.

In view of the (approximate) 3-fold symmetry of the clusters,
any orthorhombic magnetic anisotropy [corresponding to a Figure 7. Exchange-coupling model.
term E(Ŝx2 - Ŝy2) in the spin Hamiltonian] should be small
and was neglected. All of the data for χT and the magnetiza-
tion curves plotted in Figure 5 were least-squares-fitted
simultaneously to the calculated curves obtained from
diagonalizing eq 1. The best-fit results are displayed in Figure
5 as the black solid lines, and the agreement with the data is
excellent. The parameters were obtained as D ) -0.43(2)
K, g⊥ ) 1.99(2), and g| ) 2.02(2). Considering the accuracy
of the determination of the weight of the sample, the values
of the g factors are in perfect accord with the expectation
for FeIII ions. With these best-fit values, also the magnetiza-
tion curves as a function of B/T were calculated for the
appropriate temperatures. The results are plotted as black
solid lines in Figure 6 and are in full agreement with the
Figure 8. Total spin S of the ground state of the exchange model eq 2 as
experiment. Hence, even with a small but significant a function of JC/|JR|, assuming antiferromagnetic JR.
magnetic anisotropy, an approximate scaling behavior is
observed, which, in turn, implies that through a study of the ≡ JC was used, which is based on the rationale that the
deviation from scaling only rather large magnetic anisotropies frustration in 2 originates from the competition between the
may be detected. exchange interactions on the ring of the outer Fe ions (J1
During the preparation of this manuscript, a magnetic study and J2) and those between the ring and the central ion (J3
of compound 1 appeared.17 The reported χT and magnetiza- and J4) and that this key element is correctly incorporated
tion curves are essentially identical with those observed here in the simplified model. The corresponding exchange Hamil-
for compound 2 (Figure 4), but the anisotropy parameters tonian is given by
of the S ) 5/2 ground state were determined from powder 6
electron paramagnetic resonance (EPR) measurements to be
D ) +0.41 K and E/D ) 0.21. The magnitude of D is similar
Ĥ ) -JR( ∑
i)2
Ŝi‚Ŝi+1 + Ŝ7‚Ŝ2) -

to our finding for compound 2 but of opposite sign (we JC(Ŝ2 + Ŝ3 + Ŝ4 + Ŝ5 + Ŝ6 + Ŝ7)‚Ŝ1 (2)
carefully checked that our measurements cannot be explained
with D > 0). Also, the large ratio of E/D reported for 1 is which notably corresponds exactly to the heptanuclear
inconsistent with its crystallographic C3 molecular symmetry. species of the frustrated Heisenberg star introduced in ref
The reason for these discrepancies is unclear. Further studies 18. The energies of Ĥ are easily calculated, as has been noted
such as single-crystal EPR measurements are required to already earlier,18,31 with the spin-coupling scheme ŜR ) Ŝ2
clarify the situation. + Ŝ3 + Ŝ4 + Ŝ5 + Ŝ6 + Ŝ7, Ŝ ) ŜR + Ŝ1, and the energies
Finally, the exchange-coupling situation in compound 2 ER(SR,R) of the Hamiltonian ĤR ) ∑i)2 6
Ŝi‚Ŝi+1 + Ŝ7‚Ŝ2,
along with the origin of the S ) 5/2 ground state can be which is related to the ring of the outer Fe ions (R is an
discussed. This is of interest because of the spin frustration additional quantum number for labeling of the different spin
interactions between the Fe sites. Considering the exchange levels with equal SR). The energies ER(SR,R) can easily be
paths and the approximate 3-fold symmetry, the exchange calculated numerically.32 The energies E(S,SR,R) of the spin
model for 2 should include four different exchange interac- states of Ĥ then simply become
tions, J1, J2, J3, and J4 (see Figure 7). Although the slight
E(S,SR,R) ) -JRER(SR,R) -
deviation from the 3-fold symmetry will have some influence
on the exchange integrals, the effects will be weak, and for JC
[S(S + 1) - SR(SR + 1) - S7(S7 + 1)] (3)
a discussion of the general behavior, the assumption of a 2
3-fold symmetry is certainly a very good starting point; see
In Figure 8, the total spin of the ground state of the
also refs 9c and 30. The available data, however, do not allow
exchange model eq 2 as a function of the exchange constant
for the determination of so many exchange constants (in
JC between the Fe ions on the ring and the central Fe ion
particular, as they are very strong so that the temperature
(an antiferromagnetic exchange constant JR was assumed)
dependence of χT reflects just a few low-lying spin states).
is plotted. For strongly ferromagnetic values of JC, JC/|JR|
Hence, the simplified model with J1 ) J2 ≡ JR and J3 ) J4
(31) Waldmann, O. Phys. ReV. B 2000, 61, 6138.
(30) Waldmann, O.; Güdel, H. U. Phys. ReV. B 2005, 72, 094422. (32) Gatteschi, D.; Pardi, L. Gazz. Chim. Ital. 1993, 123, 231.

Inorganic Chemistry, Vol. 46, No. 3, 2007 765

 Ako et al.
> 4, the ferromagnetic configuration with all spins aligned tion of 1-3 can be used to synthesize a variety of other
parallel, and hence S ) 35/2, becomes the ground state. For compounds, which will be reported elsewhere.
strongly antiferromagnetic values of JC, JC/|JR| < -4, a Concerning the magnetism in compounds 1 and 2, the
ferromagnetic configuration of the spins on the ring is presence of very strong intramolecular exchange interactions
realized, with the central spin oriented oppositely, and hence of overall antiferromagnetic sign can be concluded from the
S ) 6 × 5/2 - 5/2 ) 25/2. A third unfrustrated spin temperature profiles of the magnetic susceptibility. For 2,
configuration is obtained for small values of JC, i.e., for the spin of the ground state was determined to be S ) 5/2
-0.19 < JC/|JR| < +0.27, where a S ) 5/2 spin ground state and single-crystal magnetic measurements revealed a uniaxial
is obtained. Here, the spins on the outer ring assume an magnetic ansiotropy in the ground state of the easy-axis type
antiferromagnetic configuration (SR ) 0) to which the central with D ) -0.43(2) K. With respect to the exchange
spin is coupled so weakly that it is not disturbed by the interactions in 2, it has been shown that the S ) 5/2 ground
interaction JC. state can be either due to a nonfrustrated situation, where
For all other values of JC, frustration is apparently of great the exchange couplings between the FeIII ions on the outer
relevance, and with a proper choice of JC/|JR|, every spin ring and the central FeIII ion are rather weak and below a
state between S ) 1/2 and 35/2 can be stabilized. The strongest certain threshold, or due to frustration with competing
frustration, signaled by the minimal spin ground state of S interactions on the ring and to the central ion. Further studies
) 1/2, is realized in the range -1.10 < JC/|JR| < -0.40. using complementary techniques such as inelastic neutron
Interestingly, a further S ) 5/2 ground state, with a necessarily scattering are needed to resolve this ambiguity. From a
frustrated spin configuration, is obtained in the range -1.65 general perspective, the disklike topology of homonuclear
< JC/|JR| < -1.38. Hence, from the observation of an S ) magnetic centers realized in compounds 1 and 2 exhibits
5
/2 ground state in compound 2, it is not possible to decide close ties to the theoretical model of a frustrated Heisenberg
whether the unfrustrated configuration with JC ≈ 0 or the star, which is one of the very few solvable models in the
frustrated configuration with JC ≈ -1.5|JR| is realized. The area of frustrated quantum-spin systems. Depending on the
temperature-dependent χT curve is of no help here because details of the exchange-coupling strengths, all spin ground
only the very lowest spin levels are involved in the weak states ranging from S ) 1/2 to 35/2 can be realized in such a
upturn of χT at the higher temperatures, which for both topology. Hence, a slight variation of the exchange interac-
configurations are S ) 3/2, 5/2, and 7/2 levels. Using eq 3 and tions by, e.g., a chemical tuning can have pronounced effects
the van Vleck formula, it is easy to calculate the magnetic on the spin ground state. This will be an attractive research
susceptibility for any JC and JR, but the comparison with topic for future work on odd-numbered disklike metal
the experimental data did not yield any further insights clusters.
despite the fact that the exchange constants are numerically
large. From magnetostructural correlations, it was argued that Acknowledgment. This work was supported by the DFG
for compound 1 the exchange interactions J3 and J4 (and, (Grant SPP 1137 and Center for Functional Nanostructures),
hence, also JC) should be ferromagnetic,13 which would imply Alexander von Humboldt Stiftung (postdoctoral fellowship
the nonfrustrated S ) 5/2 ground-state configuration. More for V.M.), MAGMANet (Grant NMP3-CT-2005-515767),
measurements, such as inelastic neutron scattering, would the EC-RTN-QUEMOLNA (Contract No. MRTN-CT-2003-
be needed to clarify the interesting exchange-coupling 504880), and the Swiss National Science Foundation.
situation in 2.
Supporting Information Available: Crystallographic data for
Conclusions compounds 1-6 in CIF format and the structure of 6 with selected
bond lengths and angles. This material is available free of charge
We have synthesized hepta- and octanuclear FeIII clusters via the Internet at http://pubs.acs.org. Crystallographic data have
1-5 presenting a rare topology. These complexes are an also been deposited with the Cambridge Crystallographic Data
addition to a very limited number of this class of compounds Centre under CCDC nos. 618914-618919 for compounds 1-6.
described so far, with compound 3 representing the first Copies of this information may be obtained free of charge from
example of a high-nuclearity Phdea-based complex reported the Director, CCDC, 12 Union Road, Cambridge CB2 1EZ, U.K.
to date. The nature of the side arm of the diethanolamine (fax, +44-1223-336033; e-mail, deposit@ccdc.cam.ac.uk; Web site,
http://www.ccdc.cam.ac.uk).
ligands has no influence on the core topology of the FeIII7
complexes. The synthetic approach employed in the prepara- IC061650S

766 Inorganic Chemistry, Vol. 46, No. 3, 2007