Salicylaldoxime (H2salox) in iron(III) carboxylate chemistry ...

Polyhedron 24 (2005) 711–721www.elsevier.com/locate/polySalicylaldoxime (H 2 salox) in iron(III) carboxylate chemistry:Synthesis, X-ray crystal structure, spectroscopic characterizationand magnetic behavior of trinuclear oxo-centered complexesCatherine P. Raptopoulou * , Yiannis Sanakis, Athanassios K. Boudalis, Vassilis PsycharisInstitute of Materials Science, N.C.S.R. ‘‘Demokritos’’, 15310 Aghia Paraskevi, Athens, GreeceReceived 20 January 2005; accepted 11 February 2005AbstractTwo new neutral trinuclear oxo-centered complexes, [Fe 3 (l 3 -O)(O 2 CPh) 5 (salox)L 1 L 2 ] [L 1 =L 2 = MeOH (1), L 1 = EtOH,L 2 =H 2 O(2)] are reported. In both complexes, the central oxygen atom is bridging the three iron(III) ions, which form a trianglewith an average edge of 3.3 Å. The central [Fe 3 (l 3 -O)] 7+ core is planar. The benzoato ligands are coordinated through the usualsyn,syn l 2 :g 1 :g 1 mode and the salicylaldoximato ligand shows the common l 2 :g 1 :g 1 :g 1 coordination mode. Two of the iron(III)ions have an O 6 coordination environment while the third has an O 5 N one. The two crystallographically independent trimers inthe structure of 1 (a and b, respectively) through intermolecular interactions form polymeric chains of the ...a–a–b–b...type, alongthe ac diagonal of the unit cell. In 2, again through intermolecular interactions, the trimers form layers lying parallel to the [1 0 1]plane. Mössbauer spectra at room temperature from polycrystalline samples of 1 and 2 give rise to two well-resolved doublets withan average isomer shift consistent with high spin iron(III) ions. The two doublets, at 2:1 ratio, are characterized by different quadrupolesplitting and are assigned to the nonequivalent iron(III) ions of the clusters. Magnetic measurements of 1 and 2 in the2–z300 K temperature range show antiferromagnetic interactions between the iron(III) ions in the trimers stabilizing a S = 1/2ground state. The derived values for the exchange interaction constants fall in the range usually found in [Fe 3 (l 3 -O)] 7+ clusters. Solidstate X-band EPR spectra of 1 and 2 at liquid helium temperatures give rise to broad lines extending several hundred Gauss, indicatingweak intermolecular interactions. EPR spectra from 1 and 2 in an acetone glass at 4.2 K exhibit a strong symmetric signal atg 2.0 which is attributed to the S = 1/2 ground state of the clusters.Ó 2005 Elsevier Ltd. All rights reserved.Keywords: Trinuclear oxo-centered Fe(III) complexes; Salicylaldehyde oxime; Magnetochemistry; Crystal structures; EPR spectroscopy; Mössbauerspectroscopy1. Introduction* Corresponding author. Tel.: +3210 6503365; fax: +3210 6519430.E-mail address: craptop@ims.demokritos.gr (C.P. Raptopoulou).Our interest in the coordination chemistry of phenolicoximes is related to our continuing effort to synthesizepolynuclear 3d metal clusters in a quest for newmolecule-based magnetic materials. ÔMetal-oximatesÕhave been attracting the interest of inorganic chemists,since they can be used as Ôbuilding blocksÕ for the designedsynthesis of polynuclear clusters [1]. Accordingto this synthetic route, the ligands already bound toone metal have free coordination sites and can bind toa second metal of the same or different kind. The majoradvantages of this method are, the better control of thereaction and of the products formed, factors that cannotbe controlled by following the self-assembly route. Byusing the strategy of Ômetal oximatesÕ as building blocks,various homo- and hetero-metal complexes have beensynthesized [2–8]. The propensity of phenolic oximesto form polynuclear complexes by using both the oxime0277-5387/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.doi:10.1016/j.poly.2005.02.002

712 C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–721and phenolate groups as chelating and/or bridging units,along with their ability to act in their mono- or di-anionicform, make them interesting candidates for theexploration of their metal coordination chemistry [9].In their mono-anionic form, phenolic oximes usuallyact as chelating agents through their oximatenitrogen and phenolate oxygen atoms (mode A inthe following scheme) in a variety of mono- andpoly-nuclear metal complexes [10–14]. A more interestingcoordination mode, includes the use of the phenolateoxygen atom as both chelating and bridging agent(mode B) [10,15].In their di-anionic form, phenolic oximes usually usetheir phenolate oxygen and oximate nitrogen atoms tochelate to one metal ion and their oximate oxygen asbridging unit to bind a second metal (mode C). A varietyof di- [16–18], tri- [14,19] and tetranuclear [20–22] metalcomplexes presenting the l 2 :g 1 :g 1 :g 1 coordinationmode have been reported. The oximate oxygen atomcan also bind two metal ions, presenting thel 3 :g 1 :g 2 :g 1 coordination mode (mode D) giving rise tohigher nuclearity complexes, such as hexanuclear[23,24], although this coordination mode is also observedin trinuclear [25] and tetranuclear [11] complexes.There is also one example [10] where the dianion of phenolicoximes can act as chelating tridentate ligandaccording to mode E.ACH=N-OHOOCMCH=N-OMMCH=N-OHOMBODMCH=N-OMMMiron(III) salicylaldoximato complex is the tetranuclear[11] cluster [{Fe(Hsalox)(salox)} 4 ], in which the fourHsalox act as chelating agents through their phenolateoxygen and oximate nitrogen atoms (mode A), whilstthe four salox 2 further use their oximate oxygen atomto bind to two additional metal centers (mode D). In theremaining examples, coligands have been used to completethe coordination spheres of iron(III) centers inpolynuclear complexes. In the dinuclear compound[(Me 3 tacn)Fe(III)(salox) 3 Fe(III)] [16], as well as in thetetranuclear compound [(Me 3 tacn) 2 Fe 4 (salox) 2 (l 3 -O) 2(l 2 -CH 3 CO 2 ) 3 ](ClO 4 ) [20] (where Me 3 tacn = 1,4,7-trimethyl-1,4,7-triazacyclononane)the dianion ofsalicylaldoxime adopts the l 2 :g 1 :g 1 :g 1 coordinationmode. The only trinuclear iron(III) salicylaldoximatocomplex known so far is [(C 2 H 5 ) 3 NH][Fe 3 O(Hsalox)(salox)2 (salmp)] [19] (salmp = 2-(bis(salicylideneamino)-methyl)phenol, a new ligand formed during thereaction process), where the Hsalox ligands act as chelatingunits and the salox 2 ligands adopt thel 2 :g 1 :g 1 :g 1 coordination mode.The aim of our project is to explore the use of salicylaldoximein iron(III) carboxylate chemistry, encouragedby our previous results on manganese(III) [24]. Herein,we report the first results of our project, in particularthe synthesis, X-ray crystal structure, spectroscopic (solidstate and frozen solutions EPR, Mössbauer) characterizationand magnetic behavior of two neutraltrinuclear oxo-centered iron(III) complexes, [Fe 3 (l 3 -O)-(O 2 CPh) 5 (salox)L 1 L 2 ](L 1 =L 2 = MeOH (1), L 1 = EtOH,L 2 =H 2 O(2)). The reported complexes are the first examplesof trinuclear carboxylato iron(III) complexes withsalicylaldoxime, and can be considered as precursorsfor the synthesis of higher nuclearity complexes, basedon the principle of using Ômetal oximatesÕ as ÔbuildingblocksÕ for polynuclear clusters. In both compounds,the salox 2 ligands adopt the l 2 :g 1 :g 1 :g 1 coordinationmode (mode C) thus, the possibility of further usingthe oximate oxygen atom, under certain reaction conditions,to bind to a second metal is left open.2. ExperimentalCH=NO2.1. Compound preparationsOEMAll manipulations were performed under aerobic conditionsusing materials as received (Aldrich Co). Allchemicals and solvents were of reagent grade.Further restricting our discussion to polynucleariron(III) complexes with the simplest member of thephenolic oximes, 2-hydroxy-benzaldehyde (salicylaldoxime,H 2 salox), only a few examples have beenreported. As far as we know, the only example of2.1.1. [Fe 3 (l 3 -O)(O 2 CPh) 5 (salox)(MeOH) 2 ] Æ1.25MeOH Æ 1.05H 2 O(1)A methanolic solution of H 2 salox (0.069 g,0.50 mmol) was stirred under reflux, until sodium benzoate(0.216 g, 1.50 mmol) was added. The refluxcontinued for 30 min and then Fe(NO 3 ) 3 Æ 9H 2 O

C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–721 713(0.202 g, 0.50 mmol) was added. The color of the solutionimmediately turned to deep brown. After 24 h of reflux,small amounts of a brown-red solid were filtered,washed with MeOH and identified as compound 1 byFT-IR spectroscopy. The brown filtrate was sealed andafter a period of one week X-ray quality brownish-redcrystals of 1 were formed (Yield: 0.34 g, 70%). Thecrystals of 1 were collected by filtration, washed withcold MeOH and dried in vacuo. The resulting powderanalyzed as solvent-free. Anal. Calc. for (1) (C 44 H 38 -NO 15 Fe 3 ): C, 53.47; H, 3.88; N, 1.42. Found: C, 53.45;H, 3.86; N, 1.40%.2.1.2. [Fe 3 (l 3 -O)(O 2 CPh) 5 (salox)(EtOH)(H 2 O)] ÆEtOH (2)The procedure followed was the same as describedabove, but FeCl 3 Æ 6(H 2 O) and ethanol were used instead(Yield: 0.35 g, 70%). The powder of 2 analyzed as solvent-free.Anal. Calc. for (2)(C 44 H 38 NO 15 Fe 3 ): C, 53.47;H, 3.88; N, 1.42. Found: C, 53.45; H, 3.87; N, 1.39%.Table 1Crystallographic data for complex 1 Æ 1.25MeOH Æ 1.05H 2 OFormula C 45.25 H 46.1 Fe 3 NO 17.3Formula weight 1048.30Space group P2 1 /cUnit cell dimensionsa (Å) 19.67(2)b (Å) 26.96(2)c (Å) 19.70(2)b (°) 98.84(3)V (Å 3 ) 10323(2)Z 8T (°C) 298RadiationMo Kaq calcd (g/cm 3 ) 1.348l (mm 1 ) 0.899aR 1 0.0775awR 2 0.1994R 1 = P (jF o j jF c j)/ P (jF o j)andwR 2 ¼ P ½wðF 2 o F 2 o Þ2 Š= P ½wðF 2 o Þ2 Šfor 5082 reflections with I >2r(I).a w ¼ 1=½r 2 ðF 2 o ÞþðaPÞ2 þ bPŠ and P ¼ððmax F 2 o ; 0Þþ2F 2 c Þ=3;a = 0.1046, b = 86.0886.1=22.2. General and physical measurementsElemental analysis for carbon, hydrogen, and nitrogenwas performed on a Perkin Elmer 2400/II automaticanalyzer. Infrared spectra were recorded as KBr pelletsin the range 4000–500 cm 1 on a Bruker Equinox 55/SFT-IR spectrophotometer. EPR spectra were recordedon a Bruker ER 200D-SRC X-band spectrometerequipped with an Oxford ESR 9 cryostat in 4.2–300 Ktemperature range. Variable temperature magnetic susceptibilitymeasurements were carried out on polycrystallinesamples of 1 and 2 in the 2.0–300 K temperaturerange using a Quantum Design MPMS SQUID susceptometerunder a magnetic field of 1000 G. Magnetizationmeasurements were carried out at 2.5 K over the 0–5 Tmagnetic field range. Diamagnetic corrections for thecomplexes were estimated from PascalÕs constants.Mössbauer spectra were taken with a constant accelerationspectrometer using a 57 Co (Rh) source at RT and avariable temperature Oxford cryostat.2.2.1. X-ray crystallography and solution of structuresBrownish-red prismatic crystals of 1 (0.05 · 0.25 ·0.75 mm) and 2 (0.05 · 0.15 · 0.25 mm) were mountedin capillary with drops of mother liquid. Diffraction measurementswere made on a Crystal Logic Dual Goniometerdiffractometer using graphite monochromated Moradiation. Important crystal data and parameters fordata collection for 1 are reported in Table 1. Unit celldimensions were determined and refined by using theangular settings of 25 automatically centered reflectionsin the range 11°

714 C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–721and of very poor diffraction quality. Repeated efforts togrow larger crystals proved unsuccessful. We continuedwith data collection in order to establish the gross structureof the complex. Since the structure of 2 proved quitesimilar to that of 1, we felt that complete structural characterizationwas not crucial. Thus, the refinement wasdone with anisotropic thermal parameters for the iron(III)ions only, whilst all the rest atoms were refined isotropically.No H-atoms were included in the refinement.The phenyl rings of the salox 2 and benzoate ligandswere fitted to the geometry of a regular hexagon. Thecrystallographic analysis revealed the presence of oneethanol of crystallization per trimeric complex.3. Results and discussion3.1. SynthesesAs mentioned earlier, our aim was to investigatethe use of salicylaldehyde oxime (H 2 salox) to theiron(III) carboxylate chemistry, extending our previousexperience on manganese [24]. The 1:1 molar ratioreaction of Mn(O 2 CR) 2 (R = Me, Ph) and H 2 salox inEtOH led to the hexanuclear complexes [Mn 6 O 2(O 2 CR) 2 (salox) 6 (EtOH) 4 ] (R = Me, Ph). These complexescontain the novel ½Mn III 6ðl 3 -OÞ 2ðl 2 -ORÞ 2Š 12þcore, whose topology consists of six Mn ions arrangedas two {Mn 3 (l 3 -O)} subunits bridged by two oximatooxygen atoms. Both compounds proved to behave assingle molecule magnets (SMM) and are the firstmembers of a new class of manganese-based SMMscontaining exclusively Mn III ions with blocking temperaturegreater than 2 K. In order to investigate thepossibility of synthesizing the corresponding polynucleariron(III) complexes and to study their magneticproperties, we have designed the 1:3:1 molarratio reaction of Fe 3+ /PhCO 2/H 2 salox in alcoholicmedia. The 1:3:1 molar ratio has been chosen so thatthe M/carboxylate/H 2 salox ratio remains the same ingoing from manganese(II) to iron(III) ions. But itturned out that in the iron(III) case, the amount ofthe carboxylate ions was very high, and the trinuclearoxo-centered complexes 1 and 2 were formed. The salox2 ions are coordinated through the usuall 2 :g 1 :g 1 :g 1 mode while the possibility of further useof the oximato oxygen atom to bridge a second metaland to increase the nuclearity of the derived complexhas failed. Nevertheless, we still consider compounds 1and 2 as useful staring materials for polynucleariron(III) complexes, under different reactionconditions.The synthesis of 1 and 2 can be summarized inEqs. (1) and (2), respectively, based on the reasonableassumption that H 2 O from the starting materials and/or the solvent is the source of the O 2 ion.3FeðNO 3 Þ 3 9H 2 O þ 5PhCO 2 Na þ H 2 salox þ 2MeOHMeOHƒ ƒ! ½Fe3 OðO 2 CPhÞ 5ðsaloxÞðMeOHÞ 2Šþ5NaNO 3þ 4HNO 3 þ 26H 2 Oð1Þ3FeCl 3 6H 2 O þ 5PhCO 2 Na þ H 2 salox þ EtOHƒ EtOH ƒ! ½Fe 3 OðO 2 CPhÞ 5ðsaloxÞðEtOHÞðH 2 OÞŠ þ 5NaClþ 4HCl þ 16H 2 Oð2Þ3.2. IR spectroscopyIn the IR spectrum of both 1 and 2, medium-intensitybands at 3451, 3063, 2924 cm 1 (for 1) and 3414, 3063,2980 cm 1 (for 2), are assigned to the m(OH) H2 O,m(CH) aromatic , m(CH 3 ), respectively. Strong intensitybands at 1598 cm 1 and medium intensity bands at1286 cm 1 observed in the spectrum of both complexesare assigned to the m(C@N) and m(N–Oox) of the salicylaldoximatoligand [28]. The frequencies of the m s (CO 2 )bands of PhCO 2coincide in the spectra of 1(1406 cm 1 ), and 2 (1405 cm 1 ). The bands at1557 cm 1 in 1 and 2, respectively, are assigned to them as (CO 2 ) of the benzoato ligands. The difference D(D = m as (CO 2 ) m s (CO 2 )) for 1 is 151 cm 1 and for 2 is152 cm 1 , less than that for NaO 2 CPh (184 cm 1 ), as expectedfor the bridging modes of benzoate ligation [29].A strong intensity band at 480 cm 1 in the spectra of 1and at 478 cm 1 in the spectra of 2 is characteristic ofthe presence of the [Fe 3 O] moiety in the structures of 1and 2.3.3. Structural description of [Fe 3 (l 3 -O)(O 2 CPh) 5(salox)(MeOH) 2 ] Æ 1.25MeOH Æ 1.05H 2 O(1)Compound 1 crystallizes in the monoclinic spacegroup P2 1 /c with two crystallographically independenttrinuclear complexes in the asymmetric unit (further reportedas molecules 1a and 1b). The structure of themolecule 1a is given in Fig. 1. Selected bond distancesand angles for both 1a and 1b are listed in Table 2.The structure of 1 consists of a trinuclear iron(III)oxo-centered complex. The three Fe III ions in 1a forman isosceles triangle whilst in 1b the triangle is best describedas scalene (see Table 2). The Fe–O oxo bond distancesare in the range 1.86–1.94 Å in both molecules 1aand 1b, and the central [Fe 3 (l 3 -O)] 7+ moiety is planar.Ions Fe(2) and Fe(3) in molecule 1a and the correspondingions Fe(5) and Fe(6) in molecule 1b have an octahedraloxygen rich coordination environment, while ionsFe(1) and Fe(4) in molecules 1a and 1b, respectively,have a distorted octahedral O 5 N coordination geometry.The angles within the tetragonal plane of the octahedronrange from 82.2° to 97.5° for 1a and from 78.5° to 99.1°for 1b. The angles involving the axial positions of the

C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–721 715Fig. 1. The molecular structure of the molecule 1a with the atomiclabelling (40% thermal probability ellipsoids).octahedron range from 168.5° to 178.5° and from 163.9°to 178.1° in 1a and 1b, respectively.The salicylaldoximato ligand shows the commonl 2 :g 1 :g 1 :g 1 coordination mode with the N–O oximatogroup and the phenolate oxygen atom lying above and belowthe [Fe 3 (l 3 -O)] 7+ plane, respectively. The Fe–O oximatobond distances are 1.982 Å in both molecules 1a and 1b,the Fe–O phenoxy are 1.914 and 1.930 Å in molecules 1aand 1b, respectively. The Fe–N oximato bond lengths are2.133 and 2.145 Å in molecules 1a and 1b, respectively.The whole salicylaldoximato ligand is planar and makesan angle of 51.1° and 65.3° with the [Fe 3 (l 3 -O)] 7+ planein molecules 1a and 1b, respectively.The benzoato ligands are coordinated in the commonsyn,syn l 2 :g 1 :g 1 mode. The Fe–O carboxylato bond distancesare in the range 1.977–2.029 and 1.990–2.052 Åin molecules 1a and 1b, respectively.The polymeric lattice structure of 1, shown in Fig. 2,arises from the strong intermolecular interactions betweenthe isolated trimers. More specifically, the oximatooxygen atom O(1) in molecule 1a is strongly interacting(most likely through hydrogen bonding interactions)to a coordinated methanolic oxygen atom Om(1) belongingto a centrosymmetrically related molecule 1a[O(1)...Om(1 0 ) = 2.574 Å (1 x, y, 1 z)]. Thus, a dimerof trimers of the a–a type, is formed though twointermolecular interactions across a center of symmetry.The closest FeFe interatomic distance in the a–a dimeris Fe(2)Fe(2 0 ) = 5.067 Å. The same feature is observedin the case of molecule 1b, where the oximato oxygenatom O(61) is strongly interacting (most likely throughH-bonding interactions) to a coordinated methanolOm(3) from a centrosymmetrically related molecule 1b[O(61)...Om(3 00 ) = 2.666 Å (2 x, y, 2 z)]. Thus, adimer of trimers of the b–b type, is formed due to thedouble interactions described above. The closest FeFeinteratomic distance in the b–b dimer is Fe(5)Fe(5 00 )=5.207 Å. Apart from the formation of dimers, anotherfeature is also observed in the lattice structure of 1.The phenoxy oxygen atom O(62) of molecule 1b isTable 2Selected bond distances (Å) and angles (°) for 1 Æ 1.25MeOH Æ 1.05H 2 OMolecule 1aBond distancesFe(1)–Ox(1) 1.944(8) Fe(2)–Ox(1) 1.882(9) Fe(3)–Ox(1) 1.855(8)Fe(1)–O(2) 1.914(10) Fe(2)–O(1) 1.982(9) Fe(3)–O(42) 1.977(11)Fe(1)–O(21) 1.999(10) Fe(2)–O(12) 2.004(11) Fe(3)–O(22) 1.983(11)Fe(1)–O(31) 2.013(12) Fe(2)–O(41) 2.012(10) Fe(3)–O(32) 2.006(10)Fe(1)–O(11) 2.020(12) Fe(2)–O(51) 2.017(11) Fe(3)–O(52) 2.029(10)Fe(1)–N(1) 2.133(11) Fe(2)–Om(1) 2.070(10) Fe(3)–Om(2) 2.110(9)Fe(1)...Fe(2) 3.254(2) Fe(1)...Fe(3) 3.327(2) Fe(2)...Fe(3) 3.251(2)Bond anglesFe(1)–Ox(1)–Fe(2) 116.5(4) Fe(2)–Ox(1)–Fe(3) 120.9(2) Fe(3)–Ox(1)–Fe(1) 122.3(4)Molecule 1bBond distancesFe(4)–Ox(2) 1.932(9) Fe(5)–Ox(2) 1.877(9) Fe(6)–Ox(2) 1.868(9)Fe(4)–O(62) 1.930(10) Fe(5)–O(61) 1.982(10) Fe(6)–O(82) 1.998(10)Fe(4)–O(81) 1.990(11) Fe(5)–O(111) 2.008(10) Fe(6)–O(92) 2.006(12)Fe(4)–O(71) 2.012(11) Fe(5)–O(72) 2.009(9) Fe(6)–O(102) 2.011(12)Fe(4)–O(91) 2.052(11) Fe(5)–O(101) 2.045(12) Fe(6)–O(112) 2.028(11)Fe(4)–N(61) 2.145(13) Fe(5)–Om(3) 2.099(10) Fe(6)–Om(4) 2.103(10)Fe(4)...Fe(5) 3.222(2) Fe(4)...Fe(6) 3.317(2) Fe(5)...Fe(6) 3.284(2)Bond anglesFe(4)–Ox(2)–Fe(5) 115.5(4) Fe(5)–Ox(2)–Fe(6) 122.5(4) Fe(6)–Ox(2)–Fe(4) 121.5(5)

716 C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–721Fig. 2. Plot of the polymeric chains of 1 along the ac diagonal. The intermolecular interactions between the trimers are shown as dashed lines.strongly interacting (most likely through H-bond) to thesecond coordinated methanolic oxygen Om(2) of molecule1a [O(62)Om(2 00 ) = 2.693 Å (2 x, y, 2 z)].Thus, polymeric chains of alternative dimers of the...a–a–b–b... are formed along the ac diagonal ofthe unit cell. The closest FeFe interatomic distancein the a–b part of the chain is Fe(3)Fe(4 00 )= 5.518 Å.3.4. Structural description of [Fe 3 (l 3 -O)(O 2 CPh) 5(salox)(EtOH)(H 2 O)] Æ EtOH (2)The structure of 2 (Figure S1) is analogous to that ofcompound 1, the only difference being in the monodendateligands completing the octahedral coordinationaround the iron(III) ions (one ethanol and one watermolecules instead of two methanols in 1). The three Fe IIIions form a scalene triangle with edges 3.207, 3.277 and3.306 Å. The Fe–O oxo bond distances are in the range1.835–1.923 Å and the central [Fe 3 (l 3 -O)] 7+ moiety isplanar. Ions Fe(2) and Fe(3) have an octahedral oxygenrich coordination environment, while Fe(1) has a distortedoctahedral O 5 N coordination geometry.Despite its similarities in the molecular structure withthat of 1, compound 2 presents a different latticestructure originated by the existence of intermolecularinteractions between the isolated trimers. Strong intermolecularinteractions (possibly through H-bonds) betweenthe oximato oxygen atom O(1) and thecoordinated ethanolic oxygen atom Oe(1) of a centrosymmetricallyrelated trimer [O(1)...Oe(1 0 ) = 2.674 Å( x, y, 1 z)], and between the phenolate oxygenatom O(2) and the coordinated water molecule Ow(1)of a neighboring trimer [O(2)...Ow(1 00 ) = 2.967 Å(0.5 x, 0.5 + y, 0.5 z)], are responsible for the 2Dpolymeric network (parallel to the [1 0 1] crystallographicplane) in 2 (Figure S2). The closest Fe...Fe interatomicdistances due to the H-bonding interactions areFe(2)...Fe(2 0 ) = 5.075 Å, and Fe(1)...Fe(3 00 ) = 5.913 Å.Compounds 1 and 2 can be considered as part of thelarge Ôbasic carboxylatesÕ family of trivalent oxo-centeredtrinuclear compounds with the general formula[M 3 (l 3 -O)(O 2 CR) 6 L 3 ] (L = terminal monodentateligand). The only difference between the Ôbasic carboxylatesÕand compounds 1 and 2 is that, one of the carboxylatesand one of the terminal ligands have beenreplaced by the oximato and phenoxy group of the salicylaldoxime,respectively. In that sense, bond distancesand angles in the coordination sphere are in the rangesfound for several iron(III) Ôbasic carboxylatesÕ [30–37].3.5. Mössbauer spectraFig. 3 shows Mössbauer spectra from powder samplesof 1 and 2 recorded at room temperature in theabsence of an external magnetic field. The spectra forboth compounds consist of two well resolved doublets.Solid lines above the spectra are theoretical simulationsassuming two iron sites at 2:1 ratio with theparameters quoted in Table 3. For both compoundsthe isomer shifts for sites I and II are typical for highspin Fe III in an octahedral environment comprisingO/N atoms.From the crystal structure of complex 1, site I whichrepresents the majority species is readily assigned to theiron sites with O 6 environment whereas site II, theminority species, is assigned to the iron site with O 5 Ncoordination. Site II is characterized by a significantlylarger DE Q relatively to site I. The DE Q parameter reflectsthe degree of asymmetry around the Fe ion. Thiscan be asymmetry of Fe–L bond lengths and coordinationgeometry, or asymmetry in the nature of donoratoms (N or O). In terms of coordination geometry, significanttrans effects are experienced by Fe(2), Fe(3) in

C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–721 717the spectrum exhibits fairly narrow lines suggesting thatthe iron sites in molecules 1a and 1b are characterized bysimilar Mössbauer parameters and especially DE Q . Thesituation for complex 2 is similar and the O 5 N coordinatediron atom [Fe(1)] with a larger DE Q is assignedto site II, despite its more symmetric octahedral coordinationgeometry.At 4.2 K and in the absence of external magnetic fieldsa slight broadening of the doublets for both complexes isobserved (Fig. S3). No magnetic ordering is evidenced at4.2 K. The broadening can be attributed to the onset ofmagnetic relaxation effects. Alternatively, structuralchanges may occur leading to three distinct sites.Although the deconvolution of the spectra is not uniqueboth can be simulated assuming two sites at 2:1 ratio asin the room temperature case with the parameters quotedin Table 3 (Fig. S3). The average isomer shift for bothcompounds at 4.2 K is 0.54 mm s 1 and the differencein the isomer shifts with the room temperature spectrais due to the second order Doppler effect [38].Fig. 3. Room temperature Mössbauer spectra from solid powdersamples of 1 and 2. Solid lines are theoretical spectra assuming twodifferent iron sites with the parameters quoted in Table 3.Table 3Mössbauer parameters for 1 and 2Complex Site Temperature(K)d a,b,cDE Qa,cC/2 a,d Area (%) d1 I 300 0.42(1) 0.56(1) 0.15 674.2 0.54(1) 0.61(1) 0.19II 300 0.41(1) 1.24(1) 0.15 334.2 0.53(1) 1.15(1) 0.192 I 300 0.42(1) 0.70(1) 0.14 674.2 0.55(1) 0.79(1) 0.19II 300 0.41(1) 1.29(1) 0.14 334.2 0.54(1) 1.31(1) 0.19a In mm s 1 .b Isomer shifts are reported relative to iron metal at roomtemperature.c Number in parentheses represent errors in the last digit.d Fixed.3.6. Magnetic behaviourFor complex 1, the v M T product at 300 K is3.94 cm 3 mol1 K(Fig. 4), significantly lower than thetheoretically expected value for three non-interactingS = 5/2 spins (13.14 cm 3 mol1 K), suggesting antiferromagneticinteractions. Upon cooling, the v M T productdecreases to 0.31 cm 3 mol1 K at 2 K, without extrapolatingto zero as the temperatures tends to zero, whichagrees with the interplay of antiferromagnetic interactions,and a magnetic ground state.As analyzed in the description of the structures, complex1 actually consists of two independent molecules(1a and 1b), with two rather different geometries; isosceles(for 1a) and scalene (for 1b, see description of structures).Due to the different geometries it is expected thatmolecule 1a, and by Fe(5), Fe(6) in molecule 1b, with theshortest bond being the Fe–(l 3 -O 2 ) in each case. Thecorresponding trans effect experienced by atoms Fe(1)and Fe(4) is insignificant (see Table 2). In terms of donoratoms, there are two iron atoms with asymmetric O 5 Nchromophores [Fe(1) and Fe(4)] and four atoms withmore symmetric O 6 chromophores [Fe(2), Fe(3), Fe(5)and Fe(6)]. The Mössbauer spectra, therefore, indicatethat the O 5 N coordination should be considered responsiblefor the larger DE Q value of site II.The resolution of the spectrum of complex 1 does notallow us to distinguish the two crystallographically differentmolecules 1a and 1b. We observe however thatFig. 4. v M T vs. T experimental data for complex 1, and theoreticalcurves based on the Hamiltonian of Eq. (3), using the parameters ofsolution A.

718 C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–7211a and 1b have different magnetic properties. However,consideration of two different magnetic entities for theanalysis of the bulk magnetic susceptibility data wouldlead to a large number of parameters with a small degreeof reliability. Therefore, we have analyzed the dataassuming one single trinuclear species.Initial attempts to simulate the magnetic properties ofthe complex by considering a single exchange constant,J did not yield satisfactory results. It was, therefore,concluded that a model taking into account a second exchangeconstant, J 0 , should be considered. The Hamiltonianused was^H ¼ 2Jð^S 1^S 3 þ ^S 2^S 3 Þ 2J 0^S 1^S 2 : ð3ÞThe fitting process yielded two solutions of comparablequality (with g fixed to 2.0), as is usually the case forcomplexes containing the [Fe 3 O] 7+ core, with a S = 1/2ground state [39,40]. The best-fit parameters were, A:J = 27.3 cm 1 , J 0 = 42.7 cm 1 , g = 2.0 (R = 1.2 ·10 3 ) and B: J = 34.7 cm 1 , J 0 = 24.6 cm 1 , g = 2.0(R = 1.7 · 10 3 ). This is better depicted by an error surfaceplot of J versus J 0 , which reveals the existence oftwo minima (Figure S4). Both these solutions imply aS = 1/2 ground state.In order to verify the validity of these results, a simulationof the M versus H curve at 2.5 K was carried outwith the MAGPACK program package [41,42], using thebest-fit parameters of solutions A and B. These reproducedthe experimental curves very nicely, producingpractically superimposable curves, without the use of aparamagnetic impurity fraction (Fig. 5).The situation is similar in the case of complex 2, withthe v M T product decreasing smoothly from 3.65cm 3 mol1 K at 300 K down to 0.28 cm 3 mol1 K at2 K, without extrapolating to zero (Fig. 6). For the analysisof the data we assumed the same model as above.The fit shown in Fig. 6 was carried out over the 8–Fig. 6. v M T vs. T experimental data for complex 2, and the theoreticalcurve based on the Hamiltonian of Eq. (3) (solution B).300 K temperature region. Two solutions of comparablequality were determined, their parameters being, A:J = 35.9 cm 1 , J 0 = 29.8 cm 1 , g = 2.0 (R = 3.7 ·10 4 ) and B: J = 31.3 cm 1 , J 0 = 41.2 cm 1 , g = 2.0(R = 4.3 · 10 4 ). Fitting over the 2–300 K temperaturerange yielded small discrepancies at low temperatures.These discrepancies are attributed to the presence ofweak intermolecular interactions.The existence of two minima in the fitting procedureof the magnetic susceptibility data is an inherent propertyof trinuclear complexes [40]. It would be reasonableto correlate the value of the exchange coupling constantswith the metric characteristics of the complexes. This approachhowever may not be realistic in the present case.First, complex 1 consists of two different molecules andthe unambiguous determination of the individual magneticproperties is not feasible from the present bulkmagnetic susceptibility measurements. Second, even inthe simpler case of equilateral complexes the necessityof at least two different exchange coupling constantsfor the reproduction of the magnetic susceptibility measurementsindicate that magnetostructural correlationsare not straightforward and other factors have to be takeninto account. We observe however that on average,the derived values for the exchange constants for solutionsA and B for both complexes fall well within therange of values usually found in trinuclear complexes[43–47].3.7. EPR spectroscopyFig. 5. Variation of the magnetization M vs. the applied field H at2.5 K for complex 1. The curves calculated according to the parametersof solutions A or B are superimposable.EPR spectra from powder samples for both complexesat 4.2 K are shown in Fig. 7. For compound 1the spectra reveal a feature at g 2.0 accompanied bybroad features at lower magnetic fields. For compound2 an additional broad feature is observed at higher magneticfields. The magnetic susceptibility data indicatethat the clusters are characterized by a S = 1/2 groundstate, however the EPR signals cannot be attributed to

C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–721 719Fig. 7. EPR spectrum from a solid powder sample of 1 and 2 at 4.2 K.EPR conditions: microwave frequency, 9.41 GHz; microwave power,50 mW; modulation amplitude 10 G pp .Fig. 8. EPR spectrum from 1 and 2 in acetone glass at 4.2 K. EPRconditions: microwave frequency, 9.41 GHz; microwave power, 2 mW;modulation amplitude 10 G pp .isolated S = 1/2 systems. We attribute this behaviour tointermolecular interactions present in the solid state.This is supported by the crystal structure revealing intermolecularinteractions between neighbouring trimers.We note that the low temperature magnetic susceptibilitydata for complex 2 cannot be reproduced with isolatedtrimers and the EPR spectra support weakintermolecular interactions. For compound 1 the intermolecularinteractions appear to be weak enough, sothey do not affect the magnetic susceptibility data. However,weak dipolar interactions may substantially affectthe EPR spectra. For instance a weak dipolar interactionbetween two S = 1/2 systems of the order of0.05 cm 1 can be hardly discernible in bulk magneticsusceptibility measurements, but it will critically affectthe EPR measurements leading to spectra extending ina large magnetic field region.In order to eliminate the solid-state effects, we studiedby EPR both compounds in an acetone glass, a weaklycoordinating solvent. The 4.2 K spectra are shown inFig. 8. A strong signal is observed at g 2.0 for bothcompounds. The signal is characterized by a symmetricderivative feature with a line-shape which is better describedby a Lorenztian curve. The temperature dependenceof this signal indicates that it arises from aground state. The dependence of the signal on microwavepower and the line-shape argues against a free radicalspecies. Further, it is improbable that this signalarises from monomeric Fe III (S = 5/2) species resultingfrom decomposition of the compound upon solvationin acetone. Usually such species gives rise to signals athigher g-values (for instance at g 4.3) as a result ofzero-field splitting effects. No such signals are observedin the present case. A monomeric Fe III (S = 1/2) low spinspecies can be also excluded because in this case fairlyanisotropic signals are expected.On the other hand, it is reasonable to assume that theEPR signals of Fig. 8 arise from the S = 1/2 ground stateof isolated trimers in both compounds. The symmetricline-shape of the signals suggests an isotropic system.This is indeed expected from a trimer comprising Fe III(S = 5/2) sites antiferromagnetically coupled with theisotropic Heisenberg–Dirac–van Vleck Hamiltonian asthe main interaction. Although symmetric EPR linesare expected this is rarely met in trinuclear complexesof Fe III (S = 5/2) or Cr III (S = 3/2) [39,48–53]. Usually,in such systems, non-Heisenberg interactions such asantisymmetric exchange and/or single-ion zero-fieldsplitting terms result in an anisotropic S = 1/2 groundstate. Due to the anisotropy, broad EPR signals are observedextending to high field values with g 2.0. In thepresent case the signal is symmetric and confined at theg 2.0 region indicating that the contribution of theseterms is negligible. From this point of view compounds1 and 2 are rare cases of triferric complexes where theanticipated S = 1/2 EPR signals are indeed observed.4. ConclusionsAs part of our continuing efforts to synthesize polynucleariron(III) complexes with novel magnetic properties,we have reacted H 2 salox with iron(III) in thepresence of carboxylates. The two new neutral iron(III)complexes, [Fe 3 (l 3 -O)(O 2 CPh) 5 (salox)L 1 L 2 ] (L 1 =L 2 =MeOH (1), L 1 = EtOH, L 2 =H 2 O(2)) derived, containthe [Fe 3 O] 7+ core found in Ôbasic iron carboxylatesÕ. Incomplexes 1 and 2 however two iron atoms are in anO 6 coordination environment whereas the third is foundin an O 5 N one. A detailed characterization of complexes1 and 2 including spectroscopic and magnetic studieshas been carried out. The nonequivalence of the

720 C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–721iron(III) ions is depicted in the Mössbauer spectra. Themagnetic properties are interpreted within the context ofa trimer with exchange constants typical for complexeswith the [Fe 3 (l 3 -O)] 7+ core and a S = 1/2 ground state.The crystal structure analysis of 1 reveals intermolecularinteractions between the trimers forming 1D polymericchains. In 2 the intermolecular interactions lead to theformation of a 2D network. These interactions give riseto broad EPR spectra in the solid state. The salox 2 ligandadopts the usual l 2 :g 1 :g 1 :g 1 coordination mode,while the possibility of further use of the oximate oxygenatom to bridge to a second metal is left open. In thatsense, compounds 1 and 2 can be considered as importantstarting materials for the synthesis of higher nuclearityclusters under certain reaction conditions.AcknowledgementsThe Greek General Secretariat of Research andTechnology is acknowledged for supporting the presentwork within the frame of the Competitiveness EPAN2000–2006, Centers of Excellence # 25. We also thankDr. D. Stamopoulos for the magnetic measurements.Appendix A. Supplementary dataFull crystallographic details have been deposited inCIF format with the Cambridge Crystallographic DataCentre. Copies of the data can be obtained free ofcharge on request from the CCDC, 12 Union Road,Cambridge CB2 1EZ, UK (fax: +44 1223 336408,e-mail: deposit@ccdc.cam.ac.uk or www. at http://www.ccdc.cam.ac.uk) quoting the deposition numbersCCDC 261079 (1), and CCDC 261080 (2). Four figuresshowing the molecular and lattice structure of 2, theMössbauer spectra from powder samples of 1 and 2 at4.2 K, and the surface error plot of J versus J 0 (withg = 2.00), as well as a table of the bond distances andangles around the coordination sphere of the threeiron(III) ions in 1, are also given as supplementary material.Supplementary data associated with this article canbe found, in the online version, at doi:10.1016/j.poly.2005.02.002.References[1] O. Kahn, Molecular Magnetism, VCH-Verlagsgesellschaft,Weinhein, 1991.[2] D. Luneau, H. Oshio, H. Okawa, S. Kida, J. Chem. Soc., DaltonTrans. (1990) 2283.[3] P. Chaudhuri, M. Winter, P. Fleischhauer, W. Haase, U. Florke,H.-J. Haupt, J. Chem. Soc., Chem. Commun. (1990) 1728.[4] P. Chaudhuri, M. Winter, B.P.C. Della Vedova, E. Bill, A.Trautwein, S. Gehring, P. Fleischhuer, B. Nuber, J. Weiss, Inorg.Chem. 30 (1991) 2148.[5] D. Luneau, H. Oshio, H. Okawa, S. Kida, Chem. Lett. (1989) 443.[6] H. Okawa, M. Koikawa, S. Kida, D. Luneau, H. Oshio, J. Chem.Soc., Dalton Trans. (1990) 469.[7] R. Ruiz, F. Lloret, M. Julve, M.C. Munoz, C. Bois, Inorg. Chim.Acta 219 (1994) 179.[8] E. Colacio, J.M. Dominguez-Vera, A. Escuer, M. Klinga, R.Kivekas, A. Romerosa, J. Chem. Soc., Dalton Trans. (1995) 343.[9] P. Chaudhuri, Coord. Chem. Rev. 243 (2003) 143.[10] A.G. Smith, P.A. Tasker, D.J. White, Coord. Chem. Rev. 241(2003) 61.[11] J.M. Thorpe, R.L. Beddoes, D. Collison, C.D. Garner, M.Helliwell, J.M. Holmes, P.A. Tasker, Angew. Chem. Int. Ed. 38(1999) 1119.[12] L.F. Larkworthy, D.C. Povey, J. Crystallogr. Spectrosc. Res. 13(1983) 413.[13] M. Gallagher, M.F.C. Ladd, L.F. Larkworthy, D.C. Povey,K.A.R. Salib, J. Crystallogr. Spectrosc. Res. 16 (1986) 967.[14] P. Chaudhuri, M. Hess, T. Weyhermüller, E. Bill, H.-J. Haupt,U. Flörke, Inorg. Chem. Commun. 1 (1998) 39.[15] R. Willem, A. Bouhdid, A. Meddour, C. Camacho-CamachoF. Mercier, M. Gielen, M. Biesemans, F. Ribot, C. Sanchez,E.R.T. Tieknik, Organometallics 16 (1997) 4377.[16] P. Chaudhuri, M. Winter, P. Fleischhauer, W. Haase, U. Flörke,H.-J. Haupt, Inorg. Chim. Acta 212 (1993) 241.[17] C.N. Verani, E. Bothe, D. Burdinski, T. Weyhermüller, U.Flörke, P. Chaudhuri, Eur. J. Inorg. Chem. (2001) 2161.[18] S. Liu, H. Zhu, J. Zubieta, Polyhedron 8 (1989) 2473.[19] E. Bill, G. Krebs, M. Winter, M. Gerdan, A.X. Trautwein, U.Flörke, H.-J. Haupt, P. Chaudhuri, Chem. Eur. J. 3 (1997) 193.[20] P. Chaudhuri, E. Rentschler, F. Birkelbach, C. Krebs, E. BillT. Weyhermüller, U. Flörke, Eur. J. Inorg. Chem. (2003) 541.[21] P. Chaudhuri, M. Winter, P. Fleischhauer, W. Haase, U. Flörke,H.-J. Haupt, J. Chem. Soc., Chem. Commun. (1993) 566.[22] P. Chaudhuri, F. Birkelbach, M. Winter, V. Staemmlar, P.Fleischhauer, W. Haase, U. Flörke, H.-J. Haupt, J. Chem. Soc.,Dalton Trans. (1994) 2313.[23] P. Chaudhuri, M. Hess, E. Rentschler, T. Weyhermüller, U.Flörke, New J. Chem. (1998) 553.[24] C.J. Milios, C.P. Raptopoulou, A. Terzis, F. Lloret, R. Vicente,S.P. Perlepes, A. Escuer, Angew. Chem. Int. Ed. 43 (2004) 210.[25] V. Zerbib, F. Robert, P. Gouzerh, J. Chem. Soc., Chem.Commun. (1994) 2179.[26] G.M. Sheldrick, SHELXS-86: Structure Solving Program, Universityof Göttingen, Germany, 1986.[27] G.M. Sheldrick, SHELXL-97: Crystal Structure Refinement, Universityof Göttingen, Germany, 1997.[28] M. Alexiou, C. Dendrinou-Samara, C.P. Raptopoulou, A. Terzis,D.P. Kessissoglou, Inorg. Chem. 41 (2002) 4732.[29] G.B. Deacon, R.J. Phillips, Coord. Chem. Rev. 33 (1980) 227.[30] F. Degang, W. Guoxiong, T. Wenxia, Y. Kaibei, Polyhedron 20(1993) 2459.[31] E.M. Holt, S.L. Holt, W.F. Tucker, R.O. Asplund, K.J. Watson,J. Am. Chem. Soc. 96 (1974) 2621.[32] V.M. Lynch, J.W. Sibert, J.L. Sessler, B.E. Davis, Acta Crystallogr.C 47 (1991) 866.[33] A.B. Blake, L.R. Fraser, J. Chem. Soc., Dalton Trans. (1975) 193.[34] R.V. Thundathil, E.M. Holt, S.L. Holt, K.J. Watson, J. Am.Chem. Soc. 99 (1977) 1818.[35] K. Anzenhofer, J. De Boer, J. Rec. Tav. Chim. 88 (1969) 286.[36] A.M. Bond, R.J.H. Clark, D.G. Humphrey, P. Panagiotopoulos,B.W. Skelton, A.H. White, J. Chem. Soc., Dalton Trans. (1998)1845.[37] C.E. Anson, J.P. Bourke, R.D. Cannon, U.A. Jayasooriya, M.Molinier, A.K. Powell, Inorg. Chem. 36 (1997) 1265, and refstherein.[38] N.N. Greenwood, T.C. Gibb, Mössbauer Spectroscopy, Chapmanand Hall, London, 1971.

C.P. Raptopoulou et al. / Polyhedron 24 (2005) 711–721 721[39] A.K. Boudalis, Y. Sanakis, F. Dahan, J.-P. Tuchagues, Inorg.Chem., submitted for publication.[40] D.H. Jones, J.R. Sams, R.C. Thompson, J. Chem. Phys. 81 (1984)440.[41] J.J. Borrás-Almenar, J.M. Clemente-Juan, E. Coronado, B.S.Tsukerblat, Inorg. Chem. 38 (1999) 6081.[42] J.J. Borrás-Almenar, J.M. Clemente-Juan, E. Coronado, B.S.Tsukerblat, J. Comput. Chem. 22 (2001) 985.[43] R.D. Cannon, R.P. White, Prog. Inorg. Chem. 36 (1988) 195.[44] R.D. Cannon, U.A. Jayasooriya, R. Wu, S.K. arapKoske, J.A.Stride, O.F. Nielsen, R.P. White, G.J. Kearley, D. Summerfield,J. Am. Chem. Soc. 116 (1994) 11869.[45] F.E. Sowrey, C. Tilford, S. Wocadlo, C.E. Anson, A.K. Powell,S.M. Bennington, W. Montfrooij, U.A. Jayasooriya, R.D. Cannon,J. Chem. Soc., Dalton Trans. (2001) 862.[46] G.J. Long, W.T. Robinson, W.P. Tappmeyer, D.L.J. Bridges,J. Chem. Soc., Dalton Trans. (1973) 573.[47] J.F. Duncan, C.R. Kanekar, K.F. Mok, J. Chem. Soc. A (1969)480.[48] Y.V. Rakitin, Y.V. Yablokov, V.V. Zelentzov, J. Magn. Reson.43 (1981) 288.[49] A. Caneschi, A. Cornia, A.C. Fabretti, D. Gatteschi, W.Malavasi, Inorg. Chem. 34 (1995) 4660.[50] Y.V. Yablokov, V.A. Gaponenko, A.V. Ablov, T.N. Zhkhareva,Sov. Phys. Solid. State 15 (1973) 251.[51] H. Nishimura, M. Date, J. Phys. Soc. Jpn. 54 (1985) 395.[52] M. Honda, M. Morita, M. Date, J. Phys. Soc. Jpn. 61 (1992) 3773.[53] A. Vlachos, V. Psycharis, C.P. Raptopoulou, N. Lalioti, Y.Sanakis, M. Fardis, M. Karayanni, G. Papavassiliou, A. Terzis,Inorg. Chim. Acta 357 (2004) 3162.