Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2012-12-31.

Permanent Link: http://ufdc.ufl.edu/UFE0042374/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2012-12-31.
Physical Description: Book
Language: english
Creator: Masello, Antonio
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010


Subjects / Keywords: Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Statement of Responsibility: by Antonio Masello.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Christou, George.
Electronic Access: INACCESSIBLE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042374:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042374/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2012-12-31.
Physical Description: Book
Language: english
Creator: Masello, Antonio
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010


Subjects / Keywords: Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Statement of Responsibility: by Antonio Masello.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Christou, George.
Electronic Access: INACCESSIBLE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042374:00001

This item has the following downloads:

Full Text




2 2010 A ntonio M asello


3 To my family, for their wholehearted love.


4 ACKNO WLEDGMENTS The first acknowledgement is for my advisor, P rof. George Christou. He guided me through a complex research field with expertise, patience and professionalism He has promoted my critical thinking, led me in developing communication skills and expanded my view of what chemical research is through personal advice or by sponsoring my participation in several conferences I also would like to equally thank my committee members Dr. Adam Veige, Dr. Daniel Talham, Dr. Mark Meisel, Dr. Nicol Omenetto a nd Dr. Stephen Hill for their availability to advise me and follow my research progress I would like to thank my research collaborators Dr. St ephen Hill and his students for the HFEPR studies, Dr Wolfgang W ernsdorfer for the micro S quid measurements, Dr. Yiannis Sanakis and Dr. Athanassios K. Boudalis for the M ssbauer analysis, Dr. Nicol Omenetto and Jonathan Merten for photochemistry experiments, and Dr. Alexander Angerhofer for preliminary EPR investigations. Special thanks go to Dr. Khalil A. Abboud f or the X ray structure determinations, introduction to basics of X ray crystallography and his kind and professional guidance while working as an assistant at the X ray lab. I also warmly thank Dr. Spyros Perlepes, Dr. Anastasios Tasiopoulos and Eleni Mous hi for interesting scientific exchanges and occasions of collaboration. I would like to acknowledge the two teaching advisors Dr. James Horvath and Dr. Adam Veige for their guidance and for developing my teaching skills. T he Department of Chemistry at the University of Florida has also been fundamental in its various components with particular regar d to the graduate coordinator Dr. Benjamin Smith and the secretary Lori Clark. I greatly thank all the Christou group members beginning with the secretaries Sondra, Melinda and Alice, all the postdoctoral students who positively contributed to my research, especially Muralee, Dinos and Ninetta as well as all the


5 graduate and undergraduate students. I will particularly remember my former colleagues Nicole, Abhu, Al ina and Dolos as well as all the current ones I thank my friends here in Florida and in Italy for the constant support and the evenings spent at the "Salty Dog" restaurant followed by a last minute "spaghetti alla carbonara" at 4 a.m. Throughout my studies at UF I have had the constant support of my family for whom I keep m y deepest feelings of gratitude. They have always been ready to listen and shar e my joy on shiny days, and prov id e comfort on the rainy ones.


6 TABLE OF CONTENTS ACKNOWLEDGMENTS .................................................................................................. 4 page LIST OF TABLES ............................................................................................................ 9 LIST OF FIGURES ........................................................................................................ 11 ABSTRACT ................................................................................................................... 15 CHAPTER 1 GENERAL INTRODUCTION .................................................................................. 17 Transition Metal Clusters ........................................................................................ 17 Bond Valence Sum ................................................................................................. 39 2 Mn13AND M n8CLUSTERS FROM THE USE OF 1,1 FERROCENEDICARBOXYLIC ACID ..................................................................... 45 Introduction ............................................................................................................. 45 Properties of FdcH2, and Previous Applications ..................................................... 47 Synthesis of FdcH2 ................................................................................................. 55 Synthesis of [Mn13O8(OR)6(fdc)6], (R = Me, Et), [Mn8O4(fdc)6(DMF)2(H2O)2] and [Mn8O4(fdc)6(DMF)4] ............................................................................................ 60 Description of the Molecular Structures .................................................................. 6 6 [Mn13O8(OMe)6(fdc)6] 8CH2Cl2; (2) 8CH2Cl2 .................................................... 66 [Mn8O4(fdc)6(DMF)2(H2O)2] 4DMF 4H2O; (4)4DMF 4H2O ................................ 70 [Mn8O4(fdc)6(DMF)4] 4DMF; (5)DMF ............................................................. 73 Magnetochemistry .................................................................................................. 77 [Mn13O8(OMe)6(fdc)6] 4H2O 3MeOH; (2) 4H2O 3MeOH and [Mn13O8(OEt)6(fdc)6] 7EtOH; (3)7EtOH ........................................................ 77 [Mn8O4(fdc)6(DMF)2(H2O)2] 4DMF4H2O; (4) 4DMF4H2O ............................... 82 [Mn8O4(fdc)6(DMF)4] 1.5DMF3H2O; (5)1.5DMF3H2O .................................... 88 Electrochemistry ..................................................................................................... 91 Summary and Conclusions ..................................................................................... 94 Experimental ........................................................................................................... 95 [Mn13O8(OMe)6(fdc)6] 4H2O 3MeOH; (2) 4H2O 3MeOH ................................... 95 [Mn13O8(OEt)6(fdc)6] 7EtOH; (3)7EtOH ............................................................ 96 [Mn8O4(fdc)6(DMF)2(H2O)2] 4DMF4H2O; (4) 4DMF4H2O, Method (a) ............ 96 [Mn8O4(fdc)6(DMF)2(H2O)2] 3DMF2H2O; (4) 3DMF2H2O, Method (b) ............ 97 [Mn8O4(fdc)6(DMF)4] 1.5DMF3H2O; (5)1.5DMF3H2O .................................... 97 X Ray Crystallography ...................................................................................... 98 Magnetic Measurement s .................................................................................. 99 Other Measurements ...................................................................................... 100


7 3 NEW F e7CLUSTER FROM NONINNOCENT REACTIONS O F 1,1 FERROCENEDICARBOXYLI C ACID IN POLAR MEDI A ..................................... 101 Introduction ........................................................................................................... 101 Synthetic Strategies for Coordination Compounds with Fdc2 .............................. 103 Synthesis .............................................................................................................. 107 Initial Reactions of FdcH2 with Fe Sources ..................................................... 107 Pho tosensitivity of FdcH2 ................................................................................ 110 Further Study with Other Metal Sources ......................................................... 113 Description of the Molecular Structures ................................................................ 115 Magnetochemistry of the Heptanuclear Clusters [M7O3(OMe)(fdc)6(MeOH)3]n+.... 122 M ssbauer Spectroscopy of the [M7O3(OMe)(fdc)6(MeOH)3]n+ Clusters ............... 127 Electrochemistry of the [M7O3(OMe)(fdc)6(MeOH)3]n+ Clusters ............................. 131 Conclusions .......................................................................................................... 134 Experimental ......................................................................................................... 135 [Fe7O3(OMe)(fdc)6(MeOH)3]3+[FeCl4]2Cl8H2OMeOH; (6)H2OMeOH ........................................................................................ 135 [Fe6.2Mn0.8O3(OMe )(fdc)6(H2O)3]+Cl5H2OMeOH; (7)H2OMeOH ........ 136 [FexCo7 xO3(OMe)(fdc)6(MeOH)3]2+2Clsolv.; (8)solv. ................................. 136 [FexNi7xO3(OMe)(fdc)6(MeOH)3]+Clsolv.; (9)solv. ...................................... 136 [FexZn7 xO3(OMe)(fdc)6(MeOH)3]+ [ZnCl4]2solv.; (10)solv. ....................... 136 X Ray Crystallography .................................................................................... 137 Magnetic Measurements ................................................................................ 139 Other Measurements ...................................................................................... 139 4 NEW APPROACH FOR THE IMPROVEMENT OF THE BOND VALENC E SUM PARAMETERS USING STATISTICAL METHODS .............................................. 141 Introduction ........................................................................................................... 141 Known Examples of BVS Op timization ................................................................. 146 Approach to a New Optimization Method .............................................................. 150 Definitions ....................................................................................................... 150 Exact Methods ................................................................................................ 152 Meta Heuristics .............................................................................................. 153 Description of the Problem and Choice of the MetaHeuristics Method ......... 153 The Variable Neighborhood Search ............................................................... 155 Initial Experimental Setup ..................................................................................... 155 Selec tion of the Structural Data ...................................................................... 155 Proposal for a New BVS Formula ................................................................... 161 Application to the CCDC Database ....................................................................... 163 Data Retrieval from the CCDC Database ....................................................... 163 Data and Software Preparation for the Optimization ...................................... 164 Analysis of the Results ................................................................................... 167 Titanium ................................................................................................... 167 Vanadium ................................................................................................. 168 C hromium ................................................................................................ 168 Manganese .............................................................................................. 168


8 Iron ........................................................................................................... 169 Cobalt ...................................................................................................... 169 Nickel ....................................................................................................... 169 Copper ..................................................................................................... 170 Conclusions .......................................................................................................... 170 APPENDIX A CODES FOR THE COMPOUNDS LISTED .......................................................... 172 B FULL CRYSTALLOGRAPHI C TABLES ................................................................ 173 C BOND LENGTHS AND ANG LES ......................................................................... 181 D BVS OPTIMIZATION DAT A .................................................................................. 198 LIST OF REFERENCES ............................................................................................. 202 BIOGRAPHI CAL SKETCH .......................................................................................... 215


9 LIST OF TABLES Table page 2 1 Selected crystal data for complex 2 8CH2Cl2. ..................................................... 66 2 2 Bond Valence Sum values for complex 2 .......................................................... 68 2 3 Selected crystal data for complex 4 4DMF4H2O. .............................................. 71 2 4 Bond Valence Sum calculations for complex 4 .................................................. 72 2 5 Selected crystal data for complex 5 4DMF. ........................................................ 74 2 6 Bond Valence Sum calculations for complex 5 .................................................. 75 3 1 Selected crystal data for complex 6 2MeOH2H2O .......................................... 116 3 2 Bond Valence Sum calculations for complex 6 ................................................ 118 3 3 Fecentroid and centroidcentroid distances () in complex 6 ......................... 119 3 4 Bond Valence Sum calculations for complex 7 according to the Mn and Fe parameters. ...................................................................................................... 121 3 5 Counterion composition for 7 10 obtained by Cl microanalysis. ....................... 122 4 1 Expected and calculated Co valences in [Co8O4(O2CPh)12(H2O)(MeCN)3] based on published data. ................................................................................. 144 4 2 Comparison of BVS values obtained for a series of model compounds by the newly optimized and published r0 parameters ................................................. 147 4 3 BVS parameters for Fe and Mn obtained from the literature and by a preliminary VNS optimization (with b variable) over a selected database (). .. 157 4 4 BVS values for [Mn12O12(O2CMe)16(H2O)4] calculated with published literature data and by preliminary VNS optimization of r0 and b ....................... 158 4 5 BVS values for [Fe3O(O2CMe)6(H2O)3]+ calculated with published literature data and by preliminary VNS optimization of r0 and b ..................................... 159 4 6 R0 values obtained by computations on the subset data of known valence for iron ( b = 0 .37 ). ............................................................................................... 165 B 1 Crystal data and structure refinement for 2 CH2Cl2. ....................................... 173 B 2 Crystal data and structure refinement for 4 4DMF4H2O. ................................. 174


10 B 3 Crystal data and structure refinement for 5 3DMFH2O. ................................... 175 B 4 Crystal data and structure refinement for 6 2MeOH2H2O .............................. 176 B 5 Crystal data and structure refinement for 7 3DMF3H2O. ................................. 177 B 6 Crystal data and structure refinement for 8 4MeOH2H2O. .............................. 178 B 7 C rystal data and structure refinement for 9 4MeOH. ........................................ 179 B 8 Crystal data and structure refinement for 10MeOHH2O. ............................ 180 C 1 Selected bond lengths [] and angles [] for [Mn13O8(OMe)6(fdc)6], 2 ............. 181 C 2 Selected bond lengths [] and angles [] for [Mn8O4(fdc)6(DMF)2(H2O)2], 4 .... 187 C 3 Selected bond lengths [] and angles [] for [Mn8O4(fdc)6(DMF)4], 5 .............. 189 C 4 Selected bond lengths [] and angles [] for [Fe7O3(OMe)(fdc)6(MeOH)3]3+[FeCl4]2Cl, 6 ................................................. 194 D 1 R0 opt values () obtained for titanium with bopt = 0.447. ................................... 200 D 2 R0 opt values () obtained for vanadium with bopt = 0.414. ................................ 200 D 3 R0 opt values () obtained for chromium with bopt = 0.426. ................................ 200 D 4 R0 opt values () obtained for manganese with bopt = 0.617. ............................. 200 D 5 R0 opt values () obtained for iron with bopt = 0.630. .......................................... 201 D 6 R0 opt values () obtained for cobalt with bopt = 0.449. ...................................... 201 D 7 R0 opt values () obtained for cobalt with bopt = 0.376. ...................................... 201 D 8 R0 opt values () obtained for copper with bopt = 0.433. ..................................... 201


11 LIST OF FIGURES Figure page 1 1 Relationship between the energy level diagram of an M6 system and band diagram of a bulk metal. ..................................................................................... 18 1 2 Common bridging modes for the hydroxide and oxide anions ............................ 19 1 3 Most common structural types adopted by trinuclear and tetranuclear carboxylate clusters. .......................................................................................... 21 1 4 Diagram of the Mn4Ca cluster supporting water oxidation in the OEC of green plants and cyanobacteria.. .................................................................................. 23 1 5 Hypothetical mechanism for the incipient O O bond formation during the final intermediate state "[S4]" of the OEC. .................................................................. 25 1 6 Frontier electronic diagram for a pentacoordinated CuII ion in a sp geometry and structural sketch of the CuII acetate. ............................................................ 27 1 7 Two possible mechanisms for the CuCu magnetic interaction in Cu(II) acetate. ............................................................................................................... 29 1 8 Sketch of trinuclear triangular species displaying some of the possible exchange interactions leading to frustration/satisfaction pathways. ................... 30 1 9 Double well energy diagram for a S = 10 molecule with negative magnetic anisotropy parameter D .................................................................................... 32 1 10 Pov Ray projection of the Mn12acetate cluster .................................................. 34 1 11 MO diagram scheme for the ferrocene molecule in the eclipsed configuration. 36 1 12 Potenti al coordination modes for a generic carboxylate. .................................... 38 1 13 Schematic diagram of the 1,1 ferrocenedicarboxylic acid (fdcH2).. .................... 38 1 14 Scheme explaining the extension of the delocalization of a conjugated substituent to a ferrocene system.. ..................................................................... 39 1 15 Plot of bond strength expressed as valence units (v.u.) vs. the correspond ing average bond length in for a series of metal oxides ....................................... 41 2 1 Known coordination modes of fdc2. ................................................................... 48 2 2 Pov Ray projection of [{Os3H(CO)10}2fdc] ........................................................... 49


12 2 3 Square array of Mo2 4+ units bridged by fdc2 and analogous arrangement in a square of four Ga2 0 units. ................................................................................... 49 2 4 Pov Ray projection of the [Sn8O4(fdc)6] cluster and its cyclic voltammogram in DMF vs. sat. Ag/AgCl. .................................................................................... 52 2 5 Modified synthesis of the ligand precursor 1,1'ferrocenedicarboxylic acid. ....... 56 2 6 1H NMR spectrum of raw 1,1'ferrocenedicarboxylic acid saturated in d6DMSO. ................................................................................................................ 57 2 7 400 MHz 1H NMR spectrum of 1,1'fer rocenedicarboxylic acid 0.040 g. in 0.5 mL d6DMSO. ..................................................................................................... 58 2 8 HREIMS spectrum of the synthesized 1,1'ferrocenedicarboxylic acid. .............. 58 2 9 Comparison of IR spectra for commercial 1,1' ferrocenedicarboxylic acid and the synthesized product. ..................................................................................... 59 2 10 Pov Ray projection of 2 ..................................................................................... 67 2 11 Pov Ray projection of the [MnIVMnIII 6MnII 6( 5O)6( 3O)2]18+ core in [Mn13O8(OMe)6(fdc)6] .......................................................................................... 67 2 12 Packing diagram for complex 2 in Mercury. ........................................................ 70 2 13 Pov Ray projection of [Mn8O4(fdc)6(DMF)2(H2O)2] .............................................. 71 2 14 Pov Ray projection of the core [MnIII 4Mn4 II( 4O)4]12+ in 4 ................................... 72 2 15 Packing diagram for complex 4 in Mercury ......................................................... 73 2 16 Pov Ray projection of complex 5 core. ............................................................... 75 2 17 Packing diagram of complex 5 in Mercury .......................................................... 76 2 18 Plot of MT vs. T for complex 2 H2OMeOH and complex 3 EtOH................ 77 2 19 Plot of reduced magnetization (M/N B) vs. H/T for a sample of 3 EtOH. .......... 78 2 20 g vs. D error surface for the m/N B vs. H/T fit for 3 EtOH.. ............................... 80 2 21 Plot of M T vs. T for complex 3 EtOH. ............................................................. 81 2 22 Plot of M vs. T for complex 3 7EtOH in the 1.815 K. ...................................... 81 2 23 H ysteresis measures for complex 2 8CH2Cl2 in the 0.041 K range. .................. 82 2 24 Plot of MT vs.T for complex 4 DMFH2O. ...................................................... 83


13 2 25 Plot o f reduced magnetization (M/N B) vs. H/T for 4 4DMF4H2O. .................... 84 2 26 Error plot for the estimation of g and D in 4 4DMF4H2O. .................................. 85 2 2 7 Plot of M T vs. T for complex 4 4DMF4H2O in the 1.815 K. ............................ 85 2 28 Plot of M vs. T for complex 4 DMFH2O in the 1.815 K. .............................. 87 2 29 Hysteresis measures for complex 4 4DMF4H2O in the 0.041 K. ...................... 87 2 30 Plot of MT vs. T for complex 5 1.5DMF3H2O ................................................... 88 2 3 1 Plot of M T vs. T for complex 5 1.5DMF3H2O in the 1.815 K. ......................... 89 2 32 Plot of M vs. T for complex 5 1.5DMF3H2O. The inset shows tails of peaks in the 1.8 to 4 K region. ...................................................................................... 90 2 33 Superposition of the MT vs. T plots for 4 DMFH2O and 5 3H2O 1.5DMF. .... 91 2 34 Solid state voltamm ogram of fdcH2, and complex 2. ......................................... 92 3 1 Structure of the pebbm ligand, 1,1 (1,5 pentanediyl)bis 1H benzimidazole, and its coordination in a chain with cadmium ions ............................................ 105 3 2 Overlay of IR spectra for fmcH in pyridine solution with increasing irradiation times. ................................................................................................................ 111 3 3 UVVisible spectra of ferrocene (Fc), benzoylferrocene (BFc) and 1,1 dibenzoylferrocene (DFc) ................................................................................. 112 3 4 Representation of the i nitial irradiation effect on 1,1 benzoylferrocene ............ 113 3 5 Pov Ray projection of [Fe7O3(OMe)(fdc)6(MeOH)3]3+ ....................................... 115 3 6 Pov Ray projection of the core [FeIII 7( 4O)3( 3OMe)]14+ in 6 .......................... 116 3 7 Different chelation modes and respective torsion angles adopted by fdcn in complex 6 ........................................................................................................ 117 3 8 Pov Ray projection of two neighboring molecules of complex 6 in the lattice. .. 118 3 9 Packing diagram of complex 6 in Mercury ........................................................ 120 3 10 Pov Ray projection of the cluster cation 7 and its labeled core ........................ 121 3 11 Direct current measurement of MT vs. T for 6 8H2OMeOH .......................... 123 3 12 Magnetic susceptibility plot for 6 with AC fields ................................................ 125


14 3 13 Direct current measurement of MT vs. T for 7 5H2OMeOH, 8 solv., 9 solv. and 10solv. ...................................................................................................... 126 3 14 AC (in phase) M T vs. T and (out of phase) M plots for 7 10 at 250Hz. 127 3 15 Overlay of the M ssbauer spectra collected at 250 and 140 K for sample 6 ... 129 3 16 Superposition of M ssbauer spectra for 7 10 at 78 K. ...................................... 131 3 17 Cyclic voltammo gram (CV ) and differential pulse voltammogram (DPV) of 6 in acetonitrile .................................................................................................... 132 4 1 Pov Ray projection of [Co8O4(O2CPh)12(H2O)(MeCN)3] and its core ................ 143 4 2 Plot of the variation in B vs. Mn oxidation state in MnO compounds ............... 148 4 3 Labeled core of [Mn12O12(O2CMe)16(H2O)4] ...................................................... 157 4 4 Pov Ray projection of the partially labeled structure of the cation [Fe3O(O2CMe)6(H2O)3]+ .................................................................................... 159 4 5 Histogram of the errors (absolute values) on the valence estimation by t he currently available BVS parameters. ................................................................ 160 4 6 Comparison of error distribution for the optimized BVS parameter r0 and b with the currently published formula ................................................................. 162 4 7 Correlation between published or derived r0 values ......................................... 166 D 1 Exemplified flow chart of chopt algorithm ......................................................... 198 D 2 Histograms for the valence errors ( ) before and after VNS optimization. ....... 199


15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy USE OF THE 1,1 FERROCENEDICARBOXYLATE AS LIGAND IN CLUSTER CHEMISTRY AND NEW IMPROVEMENT APPROACH OF THE BOND VALENCE SUM PARAMETERS THROUGH STATISTICAL METHODS By ANTONIO MASELLO December 2010 Chair: George Christou Major: Chemistry This work focused on the synthesis and study of new transition metal clusters capable of multielectron exchange. Reactions of the ligand precursor 1,1'ferrocenedicarboxylic acid ( fdcH2) in the pres ence of preformed clusters, such as [Mn12O12( O2CMe)16( H2O )4] led to the isolation of two [Mn13O6(OR)6(fdc)6] clusters, where R = Me or Et, and two [Mn8O4(fdc)6(DMF)4n(H2O)n] clusters where n = 2 or 0 (DMF = N,N dimethylformamide) The Mn13 cluster s have been magnetically studied and both were confirmed to posses a S = 9/2 ground state with negative magnetic anisotropy D N e ither of the two species showed singlemolecule magnet behavior or could be further investigated electrochemically due to their low solubility in common solvents. The Mn8 clusters have been obtained by different synthetic procedures reflecting minor differences in the peripheral composition. Such differences induce distortions that lower the symmetry of the structure containing four DMF molecules, which confer different magnetic properties from the homologous Mn8 cluster with two DMF and two water molecules. [Mn8O4(fdc)6(DMF)2(H2O)2] has a spin groundstate S = 5 with negative D and displays out of phase magnetic susceptibility signals and


16 hysteresis below 2 K, whereas [Mn8O4(fdc)6(DMF)4] has S = 2 with almost negligible out of phase signals. Electrochemical investigat ions were impossible due to insolubility in common solvents. Reactions of fdcH2 in polar media and several metal chlorides allowed for the isolation of the novel family of cationic clusters [ FexM7 xO3(OMe)(fdc)6(MeOH)3]n+ (M = Mn, Fe, Co, Ni, Zn) These products result from the photolysis of fdcH2 and only the iron based complex could be fully characterized. The Fe7cluster posses ses S = 2 and is the first true example of a cluster containing an ox idized fdc1 unit. All Fe7M7x clusters display remarkable electrochemical behavior with simultaneous oxidation of the ferrocenyl units and progressive reductions on the clusters core. Along with the synthesis and study of new clusters a novel work on th e optimization of t he parameters for the Bond Valence Sum (BVS) method was performed. The work applied metaheuristic Variable Neighborhood Search ( VNS) methods to a large database of inorganic compounds belonging to the first transition row implying a var iable value of the parameter b. T he study significantly improved the BVS accuracy on each metal and derived parameters for metal donor combinations not yet reported in the literature.


17 CHAPTER 1 GENERAL INTRODUCTION Transition M etal C lusters In inorganic and organometallic chemistry the term "cluster is used to indicate m olecular compounds containing two or more associated and mutually interacting metal ions .1 F. A. Cotton was the first to us e this term in the early 60s for indicating polynuclear species with metal metal bonding. Later other authors defined cluster s slightly differently by referring to any pol ynuclear compound where a net or substantial degree of metal metal bonding was present.2 Nowadays, or ganometallic chemists adopt the term cluster when referring to a polynuclear species where the metals are he ld together by metal metal bonds. I n bioinorganic and general coordination chemistry however, a cluster is a polynuclear species where the structure is mainly held together by bridging ligands with negligible metal metal bonds but perhaps with weak exchange interactions .3 Along the transition series metals in groups 3 8 are recognized to produce stable and relatively inert bonds with oxygen or oxygenated ligands, whereas over the oxo wall (groups 9 12) this tendency fades, giving no stable M=O compounds.4 Analogously, early transition metals have stable interactions with donor ligands and the opposite is seen in groups 810, where the acceptor ligands ( especially C oxidation states and metal metal bonding.5 The effect of the acceptor ligands optim iz es the M M overlap of orbitals, which improves moving down the groups, in other words with the increase in the radial size of metal orbitals .5 One of the major differences between the cluster s that have metal metal bonds and those that do not is revealed upon increase of the cluster size. I n the former type, the properties tend to resemble


18 those of the bulk metal. For large clusters with M M bonding there exists a clear correlation between the dd bandwidth of a metal particle and the highest occupied and lowest unoccupied skeletal molecular orbital (HOMO LUMO gap) (See F ig. 1 1) .5 Although an actual conduction band cannot be achieved in a cluster, a large number of energetically close electronic levels ari se as the size increases allowing t he large systems to achieve numerous redox states.5 Figure 1 1 Relationship between (a) the energy level diagram of an M6 system and (b) band diagram of a bulk metal (adapted from ref. 5). In a cluster where M M bonds are not present each metal carries important properties that it would also carry a s a mononuclear compound, particularly the spin. In clusters with M M bonds the spin of two directly bonded atoms is quenched and the overall number of unpaired el ectrons is lowered, hence modifying the intrinsic properties of the metal ions .6 In the other type of clusters, the initial spin of each metal ion is still available and this characteristic gives rise to all the consequent magnetic properties that distinguish clusters with ligandbridged metals only. The oxide ion is often found to bridge and complete the skeleton structure of most clusters.


19 When a metal cation binds water it mak es the latter more acidic, hence favoring an acid base dissociation that leaves an hydroxide anion as ligand.7 High concentrations of hydrated metal cations can promote metal aggregation through "olation reactions" In this process, a coordinated hydroxide interact s with the proton of a coordinated water molecule on a neighboring hydrated metal. Consequent expulsion of a water molecule leaves the two metals bridged by an hydroxide, as illustrated in Eq. 1 1 ( en is ethylenediamine), (adapted from ref. 5) .7 (1 1) Such a process may be accompanied by further hydroxide dissociation to an oxo ligand, which in turn may promote additional metal aggregation. S ome of the most common ligation modes for the hydroxo and oxo ligand are displayed in Fig. 1 2 .7 Figure 1 2 Common bridging modes for the hydroxide (top row) and oxide (bottom row) anions (adapted from ref. 5). Doubly bridged dinuclear structures are the most common.7 A survey of the structurally characterized compounds on the Cambridge Crystallographic Data Centre (CCDC)8,


20 restricted to the first t ransition row (with the exclusion of Sc and Zn), shows many dinuclear complexes belonging to one of the following three types: {M(OH)2M}, {M(O,OH)M} or {M(O)2M}. In most cases, the peripheral ligation of the metal center is occupied either by bulky monodentate ligands like imidazole9 or cyclohexylamine10, or more often by chelating or even m acrocyclic donors like TACN (1,4,7 triazacyclononane)11. These systems may be stabili zed by large peripheral ligands which hinder other species from attacking the dinuclear center. In the case of {M(OH)2M}, the absence of Mn examples is noteworthy whereas Cu and Cr are the two most common metals, covering together over 70% of the structures. The mixed bridged {M(O,OH)M} form is present in only f our examples three with Fe1214 and one with Cr15. Finally, in the {M(O)2M} case Mn is the dominant metal with about 40% of the total structures, followed by Ti and V where each represent s 25% of the s tructures Ano ther very common group of metal assemblies are the oxo centered trinuclear clusters (se e Fig. 13 a), often referred to as "triangles" or "basic carboxylate triangles" .7,16 Cotton and Wang describe in their work the triangles as trinuclear assemblies with virtual D3h symmetry with a 3oxo ligand coplanar with the three metal centers .17 These authors remark how these units cannot be called clusters because of the long metal metal separation (MM distance 3.3) and thus absence of M M bonding. However, the term is nowadays accepted to i nclude ligand bridged metals clusters with weak M M exchange interactions. Such compounds are called trinuclear or triangular metal clusters In addition to bridging carbox ylates, they also contain monodentate terminal ligands like water or pyridine, while the metal centers generally are at the oxidation state M3 III or the mixed M2 IIIMII.17 A structural survey for trinuclear systems on the CCDC ( 1st


21 transition row with the exclusion of Sc and Zn) shows that early metal s up to Fe have a great tendency to form these oxo centered structures where the central oxide is coplanar or nearly coplanar with the M3 plane. While most examples are based on Fe, the later metals (Co, Ni and Cu) rarely assume such a configuration. However, i t was found that these elements display oxoce ntered trinuclear system with the oxide ion displaced from the trinuclear plane. Beattie et al. discus s in their work how the inclusion of monodentate ligands on positions formerly occupied by a bridging carboxylate may lead to distortions causing the central oxide to move away from the plane.18 The authors also find a correlation between the strength of the ligands and the consequent effect on the distortion, resulting in a change of the oxide's hybridization with increased sp3 character (from a pure sp2 in the coplanar form).18 From the CCDC resul ts it is clear that the distortion can reach an M O M angle of nearly 90, like in an "incomplete cubane" type of structure.8,18 Figure 1 3 Most common structural types adopted by trinuclear (a) and tetranuclear (b) and (c) carboxylate clusters. (a) the full complex [M3O(O2CMe)6L3]+; (b) the core {Mn4O2}n+ of the butterfly arrangement and (c) the core {M4O4}n+ for the cubane architecture ( adapted from ref. 16). As the number of metal ions increases, so does the complexity and the number of possible structures. The nuclearity four is generally characterized by the "butterfly" (Fig.


22 1 3 b) and the "cubane" structures (Fig. 13 c) The butterfly name tries to describe the t opology as an approximation of a butterfly 's body, where the two middle metals constitute the body and the remaining two symbolize the wingtips Both types of tetranuclear assembl y (i.e., butterfly and cubane) are largely found all along the first transiti on row without exceptions While one can find other unique topologies for ligand bridged clusters, most of the higher nuclearit y structures can be described as built from combinations of di nuclear tri nuclear (triangular) and tetranuclear (butterfly and cubane) units. For this reason, a detailed analysis of higher nuclearity structures would be redundant The complexity of the structures and the abi lity of transition metals to reach determined oxidation states are the main characteristics for the interesti ng properties of clusters For example, N ature has adopted transition metal clusters to carry out particular reactions within cavities protected by large proteins that often have the double function of support ing /distort ing the metal structure, and control ling access of substrates to highly reactiv e intermediate species during enzyme turnover .19 Such properties depend on two important characteristics : f irstly the re is a large variety of possible donor groups to bind metals by virtue of the large choice of amino acid residues Secondly, the relative rigidity of some protein structural moieties ( helix, sheet) can cause geometrical distortions (hypothesis of the ent ati c state) to modify the frontier energy levels, h ence tweaking the redox properties.19 Among the many examples that show N ature's use of metal clusters there is the co re of the oxygenevolving center (OEC) This metalloprotein is responsible for the oxidation of water to molecular oxygen and four protons in biological systems capable of


23 photosynthesis.20 It has been studied for a long time due to its importance, but the large multi protein assembly engulfing the OEC has hindered accurate direct structural characterization thr ough X ray crystallography .21 Hence, its structural design and def inite composition could only be achieved with some degree of approximation.2022 According to the last X ray data, a pentanuclear heterometallic Mn4Ca cluster is the heart of the OEC, with a structure similar to the one in Fig. 14 .22 Figure 1 4 D iagram of the Mn4Ca cluster supporting water oxidation in the OEC of green plants and cyanobacteria. The cubane structure is distorted by the presence of a Ca ion instead of Mn at one vertex position (adapt ed from ref. 22) The new structural information on the Mn4Ca cluster allowed for new hypotheses on the catalysis mechanism One of the major changes to the earlier models is the preservation of the cluster structure throughout the complete cycle attributing great stability to the cubane structure while highly reactive species are generated. Theoretical studies have demonstrated that the Mn3Ca unit is stable enough to be unaltered during the catalysis.23 Important cofactors such as one chloride anion bound to the Ca ion and a probable carbonate/bicarbonate anion on t he Ca ion have contributed to the description of a new mechanism. Recent 18O exchange experiment s have demonstrated that one of the substrate water molecules is bound to the Ca ion, suggesting that this


24 water may be coupled with another one located in clos e proximity and possibly bound to the external Mn ion.24 The current assumption on the mechanism proposes four deprotonation and four singleoxidation steps at the end of which the three Mn ions in the cubane unit are MnIV while the Mn outside the cubane is a MnV, containing a doubly bonded oxo ion.23 The formation of the Mn=O moiety is justified by the deprotonation of a water molecule bound to the outside Mn ion at the beginning of the catalytic cycle An other substrate water is bound to the adjacent Ca ion and it i s thought to perform a nucleophilic attack on the oxo (or oxyl radical) located on the MnV cation, establishing the formation of an O O bond (see Fig. 15 ).23 The chloride co factor is believed to support the orientation of the water molecule that performs the attack, while the carbonate/bicarbonate anion is considered helpful to the stability of the cluster assembly, though its function on the catalysis other than for deprotonation purposes has not yet been explained.23 The considerable employment of Mn in biological syst em s is probably a direct consequence of its high chemical flexibility in conjunction with the availability of many oxidation states in aqueous media. On one side, t he states +2, +3 and +4 are quite common and the higher oxidation states can exist under unexceptional conditions .25 On the other side, general considerations of cationic stability within a metal cluster suggest that very high oxidation states can be supported only if a sufficient number of basi c ligands are present. Consequently while MnII compounds do not require oxide ligands to be stable, the higher oxidation states of MnIII and MnIV need one or more oxides to stabilize and balance the high cationic charge. This trend can be extended to the


25 permanganate anion, where the highest oxidation state MnVII is ordinarily stable due to four coordinated oxide ligands. Figure 1 5 Hypothetical mechanism for the incipient O O bond formation during the final intermediate state "[S4]" of the OEC. The chloride anion hydrogenbonds the water substrate to direct its lone pairs toward the oxo (or oxyl radical) on the MnV. Upon O O formation, the new species is deprotonated until molecular oxygen is released (adapted from ref. 23). According to this logic the reactive MnV in the highest oxidation state of OEC ( known as [S4]) may not receive enough stabilization by a single oxide to be feasible as a participant of the catalytic cycle. On the contrary, it is plausible that one of the Mn ions within the cubane can directly participate in the catalytic process achieving the high MnV oxidation state through the support of three bound oxides. Support for this hypothesis comes from the XANES analysis of the PSII, which reveals the absence of the typical signal for a manganyl species in the OEC.26 The cubic feature, Mn3Ca in the Mn4Ca cluster demonstrates a high degree of stability with an outstanding r edox capability and structural adaptation to catalysis. The coexistence of these properties in a metal cluster is the basis for the fundamental characteristics of clusters. While the field of


26 bio inorganic chemistry has several examples of this type, other fields have been developed around the uniqueness offered by metal clusters. Physical chemistry is one of the disciplines within which transition metal clusters have been largely studied for their magnetic properties. T ransition metal complexes have a dem onstrated general propensity to have unpaired electrons ( u.e. 's ) as result of degenerate d orbitals. Each u.e. contributes to building up an overall magnetic moment vector, referred to as spin and is represented commonly by a full arrow ( ) to reflect its vector properties T he Pauli exclusion principle tends to maximize the number of u.e. by single occupancy of degenerate orbitals. Because each resulting singly occupied orbital contributes to the spin it is often referred to as a "magneti c orbital ".6 If in a metal ion the orbital angular momentum is neglected or absent, t he algebraic sum of each u.e. 's magnetic moment (s = ) constitutes approximately the ions total magnetic moment (i.e., the total spin ) In tran sition metal clusters where bridging units do not exist the metal metal bonds have a significant contribution from magnetic orbitals whose overlap produces lower energy molecular orbital s, hence pairing two u.e. 's and thereby quenching their resultant magnetic moment. Clusters of ligandbridged 3d metals on the contrary allow each metal cation to overlap its magnetic orbitals with those of the bridging ligands which in most cases, are already doubly oc cupied. This type of metal ligand interaction preserves the individual number of u.e. 's allowing for the spin of one metal to interact magnetically with the spin of the neighboring metals in the cluster .6 Such an interaction via bridging ligands is known as superexchange (see later). Structure, metal type, metal oxidation state, and ligand type are the main factors in governing the magnetic interaction, or exchange interaction, b etween metal ions in a


27 cluster This interaction is measured by the exchange parameter ( J) also known as the exchange coupling constant The realization of magnetic coupling between metal centers in a compound was noted for the first time in copper( II) acetate, a dinuclear molecule of formula [Cu2( O2CMe )4(H2O)2] where the two Cu ions interact magnetically (Fig. 16 ).27 Figure 1 6 Frontier electronic diagram for a pentacoordinated CuII ion in a sp geometry (the x2y2 is the magnetic orbital), and structural sketch of the CuII acetate (adapted from ref. 6) Each CuII has s = 1/2 and copper acetate is paramagnetic at room temperature (i.e., contains u.e. 's ), whereas it becomes diamagnetic (i.e., does not contain any u.e.) when the temperature is lowered.27 Bleaney and Bowers established the basis for magnetochemistry upon discovering that there is an interaction between the two Cu ions.28 The molecule populated a total spin S = 1 state (corresponding to the sum of two individual CuII spins 1/2 ) at room temperature, but the total spin S was found to quickly decreas e to ~ 0 below ~200 K. Blean ey and Bowers explained t hese observations through the parallel alignment of the vectors relative to each CuII spin magnetic moment ( i.e.: ) at room temperature and the antiparallel alignment at low er temperatures (i.e :


28 ) respectively Initial postulates tried to explain the canceling of the S = 1/2 spins by a so called direct interaction mechanism where the two Cu d(x2y2) magnetic or bitals overlap sideways forming a bond. The resulting molecular bonding orbital would have accommodated both electrons with the antiparallel alignment and quenched the total S Further studies on other dinuclear CuII systems analogous to the Cu acetate d emonstrated how the CuCu distance has negligible influence on the strength of the exc hange constant ( J) and that the coupling is mostly (95%) governed by the four bridging ligands. This other type of interaction is known as the "superexchange mechanism and in t he case of Cu acetate, promotes the antiparallel alignment through a negative J value. In a dinuclear compound, if J is negative, the antiparallel alignment is energetically preferred and we refer to an "antiferromagnetic coupling"; on the contrary the parallel alignment is promoted by a positive J value and the behavior is called "ferromagnetic Although the anti parallel alignment is the ground state i n the case in which J is negative, if its magnitude is lower than ~ 500 cm1, the higher energy p arallel alignment is accessible through thermal populat ion resulting in a behavior similar to the one typical of the Cu acetate (see Fig. 17 ) .6,29 In addition to the two mechanisms of direct interaction and superexchange, there is a third one known as "spin polarization" which acts especially in ionic systems and for this reason is not often very relevant to clusters. The three mechanisms are always present in every molecular structure or polynucl ear assembly, although one of them will generally prevail and direct the magnetic interactions, depending on the nature of the system.6 Trinuclear systems represent the next leve l of complexity for the study of the magnetic properties of clusters. Generally, these units display a more complex magnetic


29 behavior, a result of the increased number of interactions that complicate the number and the distribution of the magnetic levels. Further complications are also due to the effects of possible spin frustration/satisfaction pathways. Figure 1 7 Two possible mechanisms for the CuCu magnetic interaction in Cu( II) acetate: (a) scheme of the sideways dd direct overlap forming the a nd MOs. The top inset shows the possible quenching of the spin at low T, while the bottom one displays how through thermal excitation the S = 1 is reached. (b) representation of the incipient overlap of one acetate ligand responsible for superexchange interaction through the O C O bonds. (c) Relative energy diagram for the two possible states S = 0 and S = 1. S = 0 is the ground state while S = 1 is the thermally accessible excited state (adapted from ref. 6 ). For instance, in a trinuclear triangular system having M1, M2 and M3 metals there will be save for a few exceptions30, a minimu m of two different J. This is due to the tendency of trinuclear clusters to distort from general ized equilateral symmetry down to isosceles (2J system) or scalene (3J system). The spin of a metal ion in a triangle experience s two J simultaneously, each associated with an exchange pathway. The two J can have opposite sign, hence promoting each a different orientation of the spin.


30 When one of the J is much larger in magnitude then the other it governs the spin orientation, otherwise J of comparable strengths will compete leading to cases of doubtful prediction of the spin orientation (spin frustration) (see F ig. 1 8 ) Figure 1 8 Sketch of trinuclear triangular species displaying some of the possible exchange interactions leading to frustration/satisfaction pathways due to comparable J values The magnitude and sign of the exchange interaction in these and more complex systems cannot be known "a priori ", but they are instead calculated by applying the HeisembergDirac Van Vleck spincoupling Hamiltonian.31 For a simple ion pair like in copper(II) acetate, the Hamiltonian operator would be: ij = 2 J ij i j (1 2 ) The Hamiltonian solutions in Eq. 1 2 express the ener gy levels of a dinuclear system as a function of the coupling J and the spin angular momentum operator The Van Vleck equation (see Eq. 1 3 ) establishes the relationship between the molar magnetic susceptibility ( M) and the temperature, considering the various possible energy levels of the system. The expression of such energy levels in terms of couplings and spin operators according to the Hamiltonian will ultimately allow for the estimation of the J values through fitting of the experimental M vs. T plot s. The Van Vleck equation is


31 reported in Eq. 1 3 where the M is expressed as the sum of the magnetic contribution from each energetic level weighted according to its population based on the Boltzmann distribution.6 n kT E n kT E n n Mn ne e E kT E N) / ( ) / ( ) 2 ( ) 1 () 0 ( ) 0 ( 2) 2 / ( (1 3 ) The Hamiltonian description for systems of three or more metal ions soon becomes complicated and its analytical solution often requires the handling of large secular matrices, which hinder the rigorous approach, beginning with pentanuclear systems. A shortcut to the rigorous analysis of small systems takes advantage of the molecular symmetry and the mathematical formulation of the Hamiltonian eigenvalues. This way of proceeding is known as the "equivalent operator method". It involves the use of the Kambes vector coupling and allows for finding the J values without solving the secular matrix .6,32 For larger systems, where the J values can only be obtained through alternative computational methods, some important parameters such as the groundstate spin, S can be determined through low temperature magnetometry or highfield EPR analysis. The description of the total spin in terms of the individual metal ion components is known as "giant spin approximation" and, except for some cases this represents a good approximation for the description of the global spin in a p olynuclear system.33 Another very important parameter to consider in polynuclear system is the behavior of the magnetic anisotropy, through the axial zerofield splitting parameter D The individual anisotropies of each metal ion combin e giving the system an overall expression of D which applies to the total spin of the molecule. According to


32 the meaning of the magnetic anisotropy, the molecular system will have a preferential orientation for the total spin S along one axi s (axial anis otropy). The sign of D is responsible for the disposition of the ms levels on a relative scale of energy, where ms is the projection of S along the preferential magnetization direction. When D is positive the perpendicular projection of S is the most stabl e (i.e.; ms = 0 for S integer, or ms = for S half integer) W hen D is negative the "almost parallel projections are stabilized the most (i.e.: ms = S) .34 A useful way of depicting this system is through the so called "doublewe ll" potential diagram (Fig. 19 ). The presence of a negative D allows the system to be in a bi stable state in the absence of external perturbations, hence the molecular spin vector S can either assume (by convention) the orientation up or down. If such a system is subjected to an external magnetic field, the spin orientation aligned parallel with the external field will be energetically more favorable than the orientation antiparallel, which changes the population of the two states through removal of their degeneration. Figure 1 9 Doublew ell energy diagram for a S = 10 molecule with negative magnetic anisotropy parameter D In the absence of an external field the two most stable levels are the ms = 10 (adapted from ref. 34)


33 Once the external field is quenched, the two states again become degenerat e and the spin orientation will be randomized with equal population. The combination of both a large S and large and negative D value creates the conditions for an energy barrier to the reversal of the magnetization. When the temperature is below a critical valu e, known as "blocking temperature", the thermal energy is significatively lower than the energy barrier and these molecules can retain their magnetic vector alignment for a long time.34 The commonly known materials capable of retaining their magnetization below a critical temperature are known as magnets and their property arises from longrange interactions along an almost infinite lattice of metal ions. On the contrary, in molecular clusters that behave like a magnet the ability of retaining t he magnetization is a property intrinsic to each molecule. Such a cluster can show hysteres is of the magnetization, like conventional magnets, and for this reason they have been called singlemolecule magnets (or SMM) .35 A very well known example of SMM is referred to as "Mn12", a molecular cluster that contains 4MnIV and 8MnIII ions bridged by oxo and acetate groups The core is described as {Mn4 IVMn8 III( 3O)12}16+ in the overall structure of [Mn12O12( O2CMe)16( H2O )4]. The compound initial ly synthesized in 1980 36 did not receive much attention until 1991 37, when its magnetic properties w ere investigated. The Mn12 structure and composition are at the basis of its magnetic characteristics : the four MnIV ions are arranged at the center in a cubanelike fashion and surrounded by an almost planar ring of eight MnIII (see Fig. 110) In the ground state of the molecule the four MnIV ions (s = 3/2 each) are ferromagnetically coupled with each other and antiferromagnetically coupled with the surrounding MnIII ions ( s = 2 each) accounting for the total spin groundstate S = 10.


34 Figure 1 10 Left: Pov Ray projection of the Mn12acetate cluster. Right: core of the Mn12acetate displaying the central cubane surrounded by an almost planar arrangement of eight Mn ions. Color code: blue (MnIV), yellow (MnIII), red (O), grey (C). The presence of the JahnT eller distorted MnIII ions is the main source of the axial magnetic anisotropy34,38, which results in an overall D = 0.50 cm1 (or 0.7 K).39 The barrier to the reorientation of the ms vector from 10 to 10 or vice versa passing though the perpendicular disposition ms = 0 is theoretically expected to be 100D or 70 K. Relaxation experiments have revealed that there is an effective barrier of about 65K to the reversal of the magnetization in Mn12, hence indicating a limited ability to retain the magnetization direction for Mn12.40 However, it was demonstrated that at 1.5 K the relaxation time is long enough to be hardly measurable, creating the basis for the use of information storage devices.34 There are many other examples in the literature of SMMs such as [Fe8O2(OH)12(ta cn)12]Br8 4142, [Mn4O3Cl(O2CMe)3(dbm)3] 35,4344, [Ni(hmp)(MeOH)Cl]4 45, and they all contr ibuted to the development of studies on quantum tunneling of the magnetization46 and SM M application for quantum computing and information storage4748.


35 The two main research fields of interest in polynuclear clusters are based on the electronic properties that an assembly of 3d metals can display. The relatively low interaction among metal ions allows for the stable retention of a particular oxidation state. I n the case of a catalyst this is useful in perform ing and stabilizing reactions with highly reactive intermediates, whereas in an SMM it preser ve s the total spin of the molecule. According to the Robin Day classification49 most of the oxobridged 3d molecular clusters fall into either class I or II, which represent a localized or partially delocalized charge on the metal ions respectively The presenc e of peripheral ligands with a very stable electronic asset represents another factor in conjunction with the poor delocalization that limits the electronic exchange over clusters. Although this combination is advantageous for the applications so far descr ibed, the area of clusters carrying electroactive ligands is still relatively unexplored. One of the main point s of in terest would be the study of the effects of an increased capability for the exchange of electrons on the properties of a cluster. In order to probe this field a very well known electroactive platform was used: ferrocene. Since its first synthesis by Kealy and Pauson in 1951 50, ferrocene showed, on contrary to other ironorganometallic compounds a large degree of stability. The molecule was stable in air and could be heated to a relatively high temperature without significant decom position.50 The first proposal for the correct structure of ferrocene was made in 1952 by Wil kinson et al. 51: a "sandwich" arrangement of one FeII in between two cyclopentadienyl units. In 1952 Woodward et al. proposed the current common name for the molecule (ferrocene or fc ) and showed the similarity with benzene toward reactivity O n this basis they also just ified the stability of the compound.52 The electronic structure of the ferrocene


36 conforms with the 18 electron rule and foresees 16 electrons in bonding type MOs and two frontier electrons paired up in a nondegenerate nonbonding MO ( F ig. 1 11).5354 Figure 1 11. MO diagram scheme for the ferrocene molecule in the eclipsed configuration. The frontier orbital a1 is the HOMO completing the 18 electrons count. Upon oxidation ferrocene becomes ferricenium a paramagnetic species (adapted from ref. 54). Electrochemical investigations soon showed the fast and reversible characteristic oxidation of ferrocene (fc) to the ferricenium (fc+) cation by loss of one electron.5556 The oxidation results in one electron bei ng removed from a nonbonding orbital, leaving the newly formed species (fc+) with an almost unchanged stability with respect to fc.54 The properties of fc make of it an extraordinarily important molecule adopted in many studies and widely used in research without solution of continuity. Ferrocene is also consider ed to be the ideal platform in the present study for another important property


37 resulting from its oxidation. In fact, while fc is diamagnetic (all electrons are paired together), fc+ has one unpaired electron that makes the species paramagnetic with a spin of The fast and reversible electrochemistry of ferrocene is also observed in its derivat ives as well as the switching between the two magnetic states (i.e.: diamagnetic and paramagnetic) This behavior is of great interest in magnetochemistry because it offer s the possibility to analyz e directly the effect of a "spin active ligand on the magnetic core of a cluster Theoretically the magnetic interaction can be turned "on" by oxidation and turned "off by reduction. Despite the small spin value associated with the fc+ unit, its influence on the coupling may modulate the spin coupling pattern of larger spins in the cluster and contribute to the magnetic anisotropy of the system. The typical high rate of the process fc = fc+ + e could enhance a fast er electron exchange in clusters containing fc. Such clusters may be oxidized initially at the fc unit s and later replenish those elect rons with others belonging to the metal core. More generally, t he presence of fc unit s over a cluster could sustain a faster movement of electron s toward an acceptor. In order to anchor the ferrocene on a cluster, the ligand should contain a ferrocenyl moi ety as well as groups capable of promoting aggregation through bridging. A very well known group capable of assisting cluster formation and stabilization is the carboxylate, which by virtue of t he two oxygen atoms can bridge with remarkable flexibility.57 The potential binding modes for a carboxylate are shown in Fig. 1 12.58 The ligand precursor selected to fulfill the abovementioned requirements is the 1,1 ferrocenedicarboxylic acid for short fdcH2 (see Fig. 113). The contemporary presence of two chelating groups on the ferrocenyl unit can offer better coordination and promote


38 metal aggregation. Such an assembly also offers the opportunity to isolate new cluster topologies resulting from the use of dicarboxylate ligands with restricted flexibility. Figure 1 12 P otential coordination modes for a generic carboxylate (adapted from ref. 58) Figure 1 13 Schematic diagram of the 1,1 ferrocenedicarboxylic acid (fdcH2) Ligand precursor of the corresponding dianionic ligand 1,1 ferrocenedicarboxylate. The two cyclopentadienyl moieties stand on parallel or nearly parallel planes and each substituent must lay on one of these planes. The rotation about the ferrocene molecular axis is the only degree of freedom and its energy barrier is very low compared to thermal energy at room temperature.5960 For a conjugated system such as the carboxyl ate, direct attachment to a cyclopentadienyl ring extends the conjugation to the aromatic ring. In the case of ferrocenecarboxamides t his is demonstrated through


39 a decreased rotational barrier for the C N bond as a result of a partial delocalization aft er an intramolecular charge transfer according to the scheme shown in Fig. 1 14.61 The extended delocalization of the ferrocenyl units entails significant magnetic exchange interactions between the metal ions in the cluster core and ligand.62 The final purpose of this study is to obtain new oxobridged 3d transition metal clusters which display a large affinity for multielectron exchange while retaining other important properties such as SMM behavior. Chapters 2 and 3 report the synthesis and full characterization of new clusters of this type. Figur e 1 14 Scheme explaining the extension of the delocalization of a conjugated substituent to a ferrocene system. The increased freedom of rotation along the C N bond is proof of the extension of the delocalization of the system on the ferrocenyl unit ( a dapted from ref. 61). Bond Valence Sum The Bond Valence Sum method (or BVS) represents a useful approach to est imate the oxidation state of an ion (generally a metal) in a coordination compound.63 In an attempt to develop and expand the solid state theory, Linus Pauling approached in 1929 va rious chemical structures in terms of purely electrostatic effects and relative size of the ions .64 Paulings observations resulted in a series of rules and principles describing the ionic arrangement and energy of a crystal lattice. Conforming to the first rule, each cation must be surrounded by a number of anions located at the vertex of a polyhedron, whose shape is determined by the mutual cation/anion radius ratio and the overall coordination number (C.N.) The second rule defines a stable structure as one


40 where the charge of a particular anion is counter balanced by the sum of the partial positive charges each belonging to a surrounding cation. The third and fourth rules empirically correlate lattice stability with the number of anion s shared at the vertices or faces of polyhedra, especially in the case of high valence cations and low coordination numbers. The final rule states that the number of different types of polyhedra found in a lattice naturally tends to be very small as the same typology of atoms will prefer the same and most stable arrangement.64 The important intuition that Pauling expressed in the second rule was confirmed by Bragg in 1930 when he experimentally demonstrated the validity of the principle applied to a silicate structure.65 Later, Pauling realized that a catio n and its neighboring anion shorten their relative ionic radii as a consequence of a n increasingly higher electronic density in between the two species.66 This latter work represent s the first attempt to identify a relation betw een the lengt h of a bond and it s strength. In 1947 Donnay and Allmann deduced the direct correlation standing between valence and bond length through an exponential function.6768 Brown and Shannon confirmed the existence of suc h a correlation in 1973, with the publication of the common exponential function found across a large pool of inorganic compounds based mainly on metal oxides (see Fig. 115).69 The plot in Fig. 115 clearly shows how the sum of the bond strength in terms of valences correlates with the average bond length in several metal oxides. The data is clustered near integer values of valence displaying an exponential behavior. Such profile is a clear proof of the existence of a correlation between the length of a bond and it's strength. Although this general behavior is here evident for metal oxide is also reflected in most metal to ligand bonds.


41 Figure 1 15 Plot of bond strength expressed as valence units (v.u.) vs. the corresponding average bond length in for a series of metal oxides selected and publ ished by Brown and Shannon (reported from ref. 69, Reproduced with permission of the International Union of Crystallography ) Attempts to establish a semi empirical function for the prediction of the valence through the sum of all the bond lengths were made by different author s and resulted in two main expressions reported in Eqs. 14 and 15. s = s 0 (R/R 0 ) N (1 4 ) R = R 1 2k log ( s ). (1 5 ) In these expressions s0, R0, N, R1 and k are all empirical parameters.70 Both expressions were leading to very c omparable results and while Eq. 1 4 is still used for a chieving the valence sum of bridging or terminal ligand anions (especially oxygen) a more general expression was coined by Brown and Altermatt 71, which replaced Eq. 1 5 :


42 B r re s0 (1 6 ) Equation 1 6 introduces with respect to E q s. 1 4 and 15 the new empirical parameter B (note that in the literature we often see "R0" instead of r0 and "b instead of B) Both B (in Eq. 1 6 ) and N (in Eq. 1 4 ) are determined through the fit of experimental data, and their value reflects the concavity of the exponential function. Between the two parameters, B shows a lower degree of oscillation with the nature of the metal ion and for this reason Eq. 1 6 is generally preferred.71 Since Eq s. 1 4 to 16 estimat e the valence along a bond, it is necessary to sum all t he contributions from each bond in order to calculate the valence of a cation, as is inherently suggest ed by the name of the method: Bond Valence Sum. In Eq. 1 6 r0 represents the length in of a bond having unitary valence for a given value of B and its magnitude depends on the metal type and the oxidation state under investigation. On the contrary, B is known as the "universal parameter 72 and it s value is set to 0.37 following Brown and Altermatt findings that most o f the inorganic compounds could be analyzed satisfactorily adopting that number .73 One of the major strengths of this model is its flexibi lity of application, which makes it adapt able to any coordination environment regardless of the coordination number. According to the BVS model, a metal ion with the same oxidation state will generally displays longer bonds w ith an increase in C.N. The BVS model is largely adopted in inorganic chemistry to determine the oxidation state of metal ions in their coordination environment. In fact, transition metals typically form a large variety of coordination compounds while adopting different


43 oxidation states The BVS model represents a powerful method for achieving details on the oxidation states based solely on structural information. The outcome of a BVS analysis is a valence number that approaches an integer, which is suggestive of the actual oxidation state of the metal. Technological advances in X ray diffraction have led to a quicker and more accurate determination of crystal structures hence promoting the wide use of this analytical technique. The consequence is an increas e of the number of structural determinations available. While initial published collection s of r0 and B values w ere sufficient to suggest the correct oxidation state of a metal ion in most cases the increased volume of structural data requires a more representative set of parameters. For this reason, researchers have conducted several studies aimed at the minimization of the error in the estimating the oxidation states through BVS. This task has been most commonly approached by improvements in the fitting procedure of the exponential distributions through mathematical analysis, which also appears to be the only type of analysis reported by the literature. In cluster chemistry the application of the BVS model and relative parameters to metal ions in a cluster can be subject ed to signif icant discrepancies from the actual oxidation states. Among the various factors to consider, physical constraint s in rigid structures may be responsible for bond elongation or compression, which can mislead the BVS results. Another basic cause of BVS discr epancy could reside in the adoption of nonrepresentative r0 and B parameter s for metal ions in a cluster environment. Many BVS optimizations utilize very selective criteria that reject a large portion of structures from crystallographic databases and


44 fo cus on mononuclear species or inorganic polymers These last ones are at the basis of structures describing mineral s and do not well represent metal clusters The purpose of the present study is to approach the BVS parameter optimization under a new perspective that moves away from the precision of rigorous mathematical analysis over a relatively limited n umber of structures and toward modern statistical methods applied to the largest upto date collection of crystallographic data. The study includes all th e metal structures of the first transition row (with the exclusion of organometallic ones ) belonging to the Cambridge Crystallographic Data Centre (CCDC)8 which recently reached over 500000 structures. The work and relative results are discussed in detail in C hapter 4


45 CHAPTER 2 M n13AND M n8CLUSTERS FROM THE US E OF 1,1 FERROCENEDICARBOXYLI C ACID Introduction The synthesis of transition metal clusters has been largely fueled in the last two decades by the discovery of the SingleMolecule Magnet (SMM) phenomenon35,37 and its potential application to ward s molecular storage of information.74 The great advantage offered by SMMs to preserve the direction for the magnetization vector below a particular blocking temperature (Tb) makes these molecu les ideal candidates as ultra small nanoscale magnets.34 Prerequisite s for a metal cluster to function a s a SMM are the presence of a large and well isolated spi n groundstate S and a large negative axial magnetic anisotropy D .75 A simple approach to achieve a cluster possibly possessing a large S is to assemble a large number of 3d transition metal ions. This task is performed with bridging ligands where metal aggregation through oxide formation is favored. The downside of this approach is the poor control on the final molecular assembly, which governs the various intramolecular exchange interact ions among spins. The nature of the metal ions and their disposition in the clust er are responsible for the magnitude of the magnetic anisotropy and not necessarily the number of metal ions This concept is well explained in the literature by the sentence : "Unfortunately, bigger is not necessarily bett er as far as SMMs are concerned" .34 For example, the largest Mn cluster known to date is a Mn84 t orus76 possessing a total spin of only S = 6. In contrast, m uch smaller molecular systems are capable of SMM behavior primarily due to large anisotropies. For exampl e the Mn4 cubane cluster77 with S = 9/2 is a SMM with a wellisolated ground state and has been extensively used for numerous studies on SMM properties. Several strategies for the formation of M M bonded clusters78 have been suggested,


46 however the same does not apply in the case of ligandbridged 3d met al clusters. The complexity of the reaction system leading to cluster formation makes its full description challenging and f or this reason the targeted synthesis of a coordination cluster is usually very difficult .79 New cluster s of chosen nuclearity and ligation may be synthesized via substitution reactions on preformed materials This approach is successful especially in clusters containing carboxylate s. It has been demonstrated that the conjugate acid of a ligand precursor can replac e a bound carboxylate through its transformation to the corresponding acid. T he main factors in the substitution are the relative acidity of the attacking and leaving groups, the presence of preferential substitution sites and the lig and size. The neutral complex [Mn12O12(O2CPh)12( H2O )4] can be prepared either by a direct comproportionation reaction of MnII and MnIII ions in the presence of benzoic acid80 or in a higher yield by two cycles of li gand substitution with benzoic acid on the preformed [Mn12O12(O2CMe)12( H2O )4] 81. The stronger acidity of the benzoic acid (pKa = 4.19)82 with respect to the acetic acid (pKa = 4.75)82 pr omotes the protonation and remov al of the acetate ligands as shown in Eq 2 1 (2 1) Weaker acids can also be employed for such substitutions, but the equilibrium has to be driven in the forward direction through constant removal of acetic acid as a to luene azeotrope.83 In [Mn12O12(O2C R )12( H2O )4], eight carboxylate groups are attached only at the JahnTeller elongated MnIII bonds. This subgroup of ligands can thus undergo easier substit ution allowing for the isolation of partial ly substituted products


47 such as: [Mn12O12(O2CR)8(O2CR )8( H2O )4] 84 or [Mn12O12(NO3)4(O2CCH2But)12(H2O)4] 85. Finally, the large size of an incoming ligand may prevent a complete substitution or make it rat h er difficult because of steric hindrance.86 In the present work the ligand precursor 1 ,1 ferrocenedicarboxylic acid has been used as a potential route to new Mn clusters. T he acidity of fdcH2 (pKa < 5.7 in 50% EtOH)87 is comparable to the one for benzoic acid (pKa = 5.68 in 50 % EtOH)87 hence making the ligand precursor capable of ligand substitutions reactions on carboxylate (especially acetate) clusters such as [Mn12O12(O2CMe)12( H2O )4] Properties of FdcH2, and P revio us A pplications As already summarized in C hapter 1 the fdcH2 ligand precursor is a good candidate because it contains th e electroactive ferrocenyl unit along with two carboxylic acid units. I t has been shown that fdc2 can bind transition metal ions in a variety of ways ( see Fig. 2 1 ) .88 Having a large number of possible chelation modes, the fdc2 opens up the possibility for the isolation of different structural t o pologies The restricted flexibility of fdc2 limit s the degrees of freedom of the two coordinating groups, in contrast to a monocarboxylate ligand where the only limitation may come from steric constraint s. Furthermore, fdc2 is structurally more adaptable then a rigid dicarboxylate, and may allow for the formation of clusters with unprecedented conformations. The presence of the ferrocene group provides a redox capability and allows the possibility of exploring the magnetic influence of a ligand S = 1/2 spin on the magnetic properties of the cluster. Due to its interesting redox properties, ferrocene has been applied extensively in the past to produce multifunctional materials. Its combination with other molecular systems has expanded their properties and consequently, their potential applications.8992 The stability of this organometallic assembly allows for ease


48 of substitution at the cyclopentadienyl positions resulting in a large variety of derivatives that promoted its use.93 Figure 2 1 Known coordination modes of fdc2 ( adapted from ref 88). The first f dcH2 synthesi s was published in 1959 used lithiation o f ferrocene to give the product in 24% yield 93 whereas the f irst metallic adduct reporting f dc2 as ligand was with U and was characterized in 1973.94 The use of f dcH2 as ligand precursor in coordination chemistry aiming towards the formation of polynuclear species was published much later by Lee et al. in 1996.95 They were able to bridge two trinuclear Os units of the type [Os3(CO)10(MeCN)2] with one f dc2 unit giving the new heptanuclear Os6Fe species: [{Os3H(CO)10}2fdc], see F ig. 2 2 Another example of a polynuclear system obtained from the assembly of smaller units was reported by Cotton et al. in 1999 involving a square array of Mo2 4+ and Rh2 4+ units (see Fig. 23a).9697 Almost


49 concurrently, Uhl et al. produced a very similar macrocyclic square assembly of Ga2 0 units supported by four bridging fdc2 ligands (see Fig. 23b).98 Figure 2 2 Pov Ray projection of [{Os3H(CO)10}2fdc], the f irst example of the use of fdcH2 to give a polynuclear system.95 Figure 2 3 (a) Square array of Mo2 4+ units (olive green) bridged by fdc2 and (b) analogous arrangement in a square of four Ga2 0 (green) units.9698 The main interest in using the fdc2 ligand in these coordination compounds resides in the large flexibility offered by the ligand in forming unusual structures while


50 achieving multifunctional systems. In this re gard, many research groups are pursuing properties such as electro chemical activity88,95106, magnetism88,102,107110 and luminescence,88,108,111 within broader programs studying stabilization of M M bonds95 98,112, catalysis95,113 114 and exploration of new topologies in discrete units95,100 103,105 107,109110,112,114117 o r polymeric systems.99,104,108,111,115,118 In cases where the electrochemical activity in polynuclear systems was pursued, the ligand fdc2 showed a variety of behaviors. A s a general guideline, the electrochemical profile of a multiferrocenyl species should display one redox process associated with each ferrocenyl unit. O n the basis of statistical considerations the separation between the first and second peak is expected to be ~ 36.6 mV, the separation between the second and the third ~ 56.5 mV and so on.96 The position of the half wave potentials (E1/2) and the separation between two consecutive redox processes ( E1/2) are complex to predict, however the three main factors that govern the processes are as follows Fi rstly, substantial structural changes such as bond formation or dissociation, variation in the coordination number and geometry upon charge transfer may shift the comproportionation equilibrium. Secondly, a direct Coulombtype interaction may affect the si tuation significantly if two or more redox centers are in close proximity. A (n 1)+ center will be more readily oxidized than the corresponding n+ center just on the basis of ele ctrostatic considerations hence the interaction with a positively charged center will increas e the E1/2 separation for oxidation. Finally, delocalization promotes the stabilization of intermediate oxidation states which again increases the E1/2.99


51 In the specific case of fdc2 units as redox centers crystallographic data have shown that ferrocene119 and its ferricenium120 salts remain almost unchanged with respect to the interatomic distances upon redox processes, implying a negligible energy change in the conversion of fdc2 to fdc and vice versa.54 T he interatomic distance between the two irons of fdc2 units should be comparable to or smaller than ~ 7.6 in order for the iron ions to have a significant electrostatic interaction.99 The only other possib le i nteraction is along the bonds which may be governed by the number and type of bonds s eparat ing the two ferrocenyl units. At one extreme, there are cases where the fdc2 units do not interact and the E1/2 separation follows the thermodynamical statistical distribution previously mentioned. Examples of such systems are the macrocyclic assemblies made by Cotton et al.97 and Uhl et al.98 depicted in Fi g. 23 Other system s of different nuclearity and shape also exe mpl ify non communicating ferrocenyl systems, such as [Pb4Na4( fdc )6(H2O)6] 100, [Mn13O8(fdc)6(OMe )6] 101, bis [Pt2(fdc)(Pet3)2(C14H8)] 103 and [ Sn8O4(fdc)6] 105. In dimeric structures of the type [M2(fdc)2(2,2 bpy)2(H2O)2] (M = Cd, Zn, Ni, Co) 88, [ Cu2(fdc)2(Py)2(DMF)2(H2O)2] 99, [Ni2(fdc)2(py)4(H2O)2] 99, [Sn2(fdc)2R4] (R = n Bu or Bz) 106 or [Te2(fdc)2(C6H4OMe)4] 106 the degree of interaction amongst the two ferrocenyl units varies as suggested by the different E1/2 values Another important factor is the E1/2 value for the oxidation of a coordinated fdc2 unit. It is generally found that the electron withdrawing effect exerted by the coordinated metal ions moves the oxidation of the ferrocenyl units towards more anodic potentials (more positive potentials), which translates as expenditure of more energy for removal of an electron. Although this may be intuitive, there are two examples where such a


52 trend is not observed. The polymeric [Ba (fdc)(H2O)] by Guo et al.104 showed ( in the solid state) that E1/2 for the oxidation of the fdc2 is almost the same as that for uncoordinated fdcH2. I n order to understand this behavior, the author s synthesized another polymeric compound containing Sm: [Sm6(fdc)4(H2O)n]. The conclusion of the study was that both alkaline earth and lanthanide metal ions leave the redox potentials of ferrocenyl based bridging ligands relatively unaffected, contrary to what was generally observed for transition metals. Unfortunately, there is n o other published data of this type to confirm or contest the conclusion. A n important study on the stability of a coordinat ed fdc species comes from the Sn cluster [ Sn8O4(fdc)6] isolated through solvothermal synthesis by Zheng et al. 105 (see Fig. 24 ) In this compound, the six fdc2 units display some degree of lability in DMF solution (solvent of choice for electrochemical investigation). Figure 2 4 Pov Ray projection of the [Sn8O4(fdc)6] cluster and its cyclic voltammogram in DMF vs. sat. Ag/AgCl (the latter reported from ref. 105, Copyright Wiley VCH Verlag GmbH & Co. KGaA. Reproduced with permission)


53 When the potential is scanned in the anodic region a reversible feature is observed which is assigned to the free ligand. At higher potentials appears the irreversible oxidation of the coordinated fdc2 units i n the cluster. The authors explain the irreversibility of this process with the decrease in basicity of the ligand upon oxidation. Such an effect would probably cause the oxidized fdc units to dissociate partially or totally from the Sn core and promote an irreversible rearrangement soon after Quite opposite behavior is found in the large Ga8(fdc)4macroc ycle where the system show s an initial irreversible oxidation of the GaGa bond, then a combined four electron reversible wave followed by another irreversible oxidation.98 The multi electron wave is associated with the simultaneous oxidation of the ferrocenyl units. With respect to magnetic properties, the literature reports only two polynuclear clusters possessing fdc units which have been investigated; [Zn6O2(fdc)5(H2O)(DMF)] and [Zn8O4(fdc)6(H2O)3] by Kim et al. in 2007107. The authors invoke the presence of a mixed valence state of the iron ions due to charge balance considerations and to explain the net paramagnetic behavior of the clusters. T he first Zn cluster is expected to have two FeIII and three FeII ions in the ferrocenyl units, wher e as the second cluster four FeIII and two FeII ions Based on magnetic measurements, the authors suggest the presence of antiferromagnetic interactions amongst the fdcunits in both molecules and no further description of the ground state or coupling pathways was reported. T h e magnetic data, however reveal some mismatch with the assumed composition of the clusters and the orange color of these compounds does not match with the expected intense blue color of ferricenyl units The synthesis of these Zncluste rs was performed by diffusion of the reagents for [Zn6O2(fdc)5(H2O)(DMF)] and by solvothermal methods


54 for [Zn8O4(fdc)6(H2O)3] E xperimental details do not specif y the ambient light conditions and in case of light exposure there may be photodecomposition of the ligand121 and scrambling of Fe into the cluster core. The photoinstability of ferrocenyl units is well kn own in chemistry and it will be discussed in detail in C hapter 3 Another published study reportin g fdc2 units in a Co dimer introduced the possibility of contamination by Fe into the final product to explain some deviations in the expected magnetic behavior.88 N o description was found in the literature about t he magnetic coupling promoted by the fdc2 unit T his could be the consequence of the relatively low number of report s on fdc2 as a ligand i n metal clusters The magnetism of other compounds containing fdc2 is con sidered to be solely due to the cluster core. Furthermore, M ssbauer investigations displayed no charge exchange between fdc2 and the core ions in two Fecore clusters.109 The other often investigated property of ferrocenyl rich system s is fluorescence. This phenomenon is typical of ferrocene and most of its derivatives, like the sodium salt of the ligand (Na2fdc) or its aci dic precursor (fdcH2). The absorption peak at ~ 245 nm results in fluorescence emission at ~ 393 nm in solid Na2fdc.88,111 This property is exhibited in clusters containing fdc groups as a consequence of the intrali gand nature of the fluorescence process. Such observation derives from the relatively unchanged position of the fluorophore emission as free ligand or as coordinating group. The ligandto ligand charge transfer (LLCT) is not remarkably affected by the nature of the polynuclear as sembly of which its is part of. However, presence of relaxing groups nearby such as hydroxide, may be responsible for enhancing the dispersion of energy which accounts for the decreased intensity of the emission band.108


55 Synthesis of FdcH2 The ligand precursor fdcH2 is a commercially available compound used in various fields. However, in order to perform an exhaustive study as a ligand in cluster chemistry relatively large quantities are necessary. Hence, its synthesis was initially attempted in the laboratory 1,1 ferrocenedicarboxylic acid was isolated as the main reaction product and later purified by fractional precipitation and crystallization techni ques. Chromatographic methods w ere not employed due to the bulk amounts of the material to be treated and due to the photoinstability of the components. E xtensive analysis proved fdcH2 purification ineffective for the complete remov al of the by products. The effect of such impurities in the system forming clusters is unknown although the possibility for undesirable side reactions exists. One of the most recent publications to achieve the ferrocene based diacid reports the use of a lithiation reaction followed by electrophilic attack of carbon dioxide on the cyclopentadienyl rings.122 The synthesis has been slightly modified and performed as follow. n B utyllithium in hexane was added dropwise to a solution of TMEDA ( tetramethylethylenediamine) and ferrocene in anhydrous ethyl ether with continuous stirring under argon atmosphere at 0 C After the reagents were completely mixed the system was slowly warm ed up to room temperature and allow ed to stir overnight. The reaction form ed an orange precipitate of the dil ithium salt of ferrocene which was then reacted with an excess of CO2 by bubbling the gas through the solution, yielding the dilithium 1,1'ferrocenedicarboxylate. Deionized water was added until complete dissolution of the precipitate occurred. Aft er filtration the product was extracted with dichloromethane/water and the aqueous phase acidified with diluted hydrochloric acid. The orange precipitate of 1,1'ferrocenedicarboxylic acid resulting from acidification was


56 filtered and washed with deionize d water. After complete drying the isolated product accounted for a 85 % yield, against the 68 % reported in the literature. The increased yield might be due to a longer reaction time between the complex n BuLi TMEDA and ferrocene with respect to the publi shed procedure. The adopted r eaction scheme is reported in Fig. 25 Figure 2 5 Modified synthesis of the ligand precursor 1,1'ferrocenedicarboxylic acid. The yield of isolated product is ~ 85%. The isolated fdcH2 was analyzed using the following techniques: 1H N M R, HR/ EI MS IR and elemental analysis The 1H N M R data were collected at room temperature in a DMSO d6 solution (Fig. 26) and compared with the spectrum obtained fr om published databases .123 DMSO is dime thylsulfoxide and is the solvent of choice for the NMR measurement for both enhancing the solubility of fdcH2 and to reproduce the same experimental conditions of the spectral database.


57 Figure 2 6 1H NMR spectrum of raw 1,1'ferrocenedicarboxylic acid saturated in d6DMSO. The 1H NMR spectrum shows the expected peaks for the hydrogen nuclei belonging to the cyclopentadiene rings at 4.450 and 4.688 ppm with integrated relative areas of 4.0 and 4.1, respectively (reference for the integration is the peak at 4.450 ppm). A low field signal at 12.314 is indicative of the two acidic protons. The intensity of this last signal is not generally reliable due to a fast kinetic exchange. Additional peaks found at 1.232, 4.211 ppm are probably due to solvent impuriti es and unreacted ferrocene respectively hence their presence will not affect the reactions forming clusters Peaks at 4.610 and 4.950 ppm may belong to products of monosubstitution of ferrocene, such as ferrocenecarboxylic acid. For comparison, the 400 M Hz 1H NMR spectrum from a public database is reported (Fig. 2 7 ).123 The spectrum of high purity fdcH2 confirms its presence as the major component in the reaction product.


58 Figure 2 7 400 MHz 1H NMR spectrum of 1,1'ferrocenedicarboxylic acid 0.040 g. in 0.5 mL d6DMSO (reported from ref. 123 with authors permission from SDBSWeb : http://riodb01.ibase.aist.go.jp/sdbs/ (National Institute of Advanced Industrial Science and Technology, June 5th 2010) Further investigations were performed with High Resolution E lec tron Ionization Mass Spectroscopy (HREIMS) showing only a low intensity molecular peak at m/z = 272.9808 due to the low volatility (Fig. 28 ). The theoretical m/z is 272.9844 and these results are supported by published investigations.124 Figure 2 8 HREIMS spectrum of the synthesized 1,1'ferrocenedicarboxylic acid. The low volatility allowed only for the detection of the molecular peak.


59 The characteriz ation by IR spectroscopy of the synthesized fdcH2 against a commercially available source revealed almost superimposable spectra (Fig. 29 ). Some of the characteristic vi brational modes could also be identified by comparison with literature data.125 Figure 2 9 Comparison of IR spectra for commercial 1,1'ferrocenedicarboxylic acid and the synthesized product Typical bands for the carboxylic acid were observed at 2636, 2552 and 917 cm1 in the spectrum of the synthesiz ed fdcH2, which could be assigned to the OH stretching and out of plane OH deformation vibrations. The very strong band found at 1683 cm1 was due to the C=O stretching vibrations. The bands at 1491 and 1297 cm1 were assigned to onplane bending modes of the C O H group.125 Finally, t he results from micro analysis were in close agreement with the expected theoretical composition.


60 S ynthesis of [Mn13O8(OR )6(fdc)6], (R = Me, Et), [Mn8O4(fdc)6(DMF)2(H2O)2] and [Mn8O4(fdc)6(DMF)4] The use of fdcH2 as a ligand precursor with preformed Mn clusters has led to changes in the nuclearity and eventual isolation of two new homonuclear Mn13 and Mn8 cluster complexes. The approach used was the substitution reaction on the [Mn12O12(O2CMe)16(H2O)4]4H2O4MeCO2H ( 1 ) 4H2O4MeCO2H, complex according to published procedures .126128 To a solution of 1 in MeCN, a solution of fdcH2 in CH2Cl2 was added in stoichiometric excess. C omplete removal of the acetate ligands is ensured by removal of acetic acid as a toluene azeotrope (28:72 %; b.p. 101 C / 1 atm)83. The solution was treated repeatedly with small portions of toluene, evaporated at reduced pressure and the product was successively redissolved in MeCN and used for crystallization according to different techniques. The procedure did not lead to crystalline product for structural characterization, although t he IR spectrum of the isolated solid showed the Mn12 starting material to be absent. A variety of other reaction conditions were therefore explored. Complex 1 was suspended in a mixture of MeCN and MeOH in various ratios and heated with fdcH2 (also in variable stoichiometries ). Experimental results showed that a 1:1 ratio MeCN/MeOH is optimal with a stoichiometric rati o 1:4 ( 1 : fdcH2) for the isolation of reddishbrown needles after filtration and crystallization with diethyl ether ( ~ 5% yield) The product was characterized by X ray crystallography as [Mn13O8( OMe )6(fdc)6] 8CH2Cl2 [(2 ) 8CH2Cl2] Further synthetic effor ts based on different solvent mixtures demonstrated the possibility to produce complex 2 using CH2Cl2/MeOH (1:1) as the solvent but in with a low 5% yield.


61 C omplex 2 was reported in the literature in 2003 when M. Kondo et al. synthesized it in 33% yield by diffusion of an acetone solution of fdcH2 into a methanolic solution of Mn(O2CMe)2 under aerobic conditions.101 Although the present work led to this known compound, it could not have been predicted, and it also provided the opportunity for a further study of its properties. The Kondo paper was limited to the electrochemical characterization of the Mn13 cluster in the solid state. This study ha s now carried out the characterization of its magnetic properties. A d etailed description of the mechanism governing such reactions represents a challenge, but it is reasonable to assume the initial binding of one of the carboxylate groups of fdcH2 on 1 with displacement of an acetate group in the form of acetic acid.127128 At this point in analogy with a chelating ligand, the proximal vicinity of the two carboxylates in fdc2 increase the probability of the second carboxylate moiety to bind, once the first carboxylate group enters t he coordination of the Mn core. T he progressive substitution of two acetate groups by an fdc2 likely causes the rearrangement of the cluster core. In this particular case, two or more molecular fragments contribute to increase the nuclearity as all the acetate groups are displaced by fdc2 groups Besides complex 2, another cluster with the same core but slightly different ligation was obtained. The use of a mixture of EtOH and MeCN (1:1) led to an almost identical core where the external methoxide ligati on was replaced by ethoxides. The identification of the new cluster [Mn13O8( O Et )6(fdc)6] (3 ) was based on IR spectra, microanalysis and magnetic behavior The reaction process for complex 3 is actually superior to that for 2 in that it gives a cleaner reaction and slightly higher yield of product


62 (7%) The reactions to both 2 and 3 g ive large amounts of insoluble brown precipitates that justify the low yield for the isolation of these clusters. One unfortunate property of 2 and 3 is their very low solubility. W ith the purpose of increasing the solubility o ther reactions were attempt ed to further change the peripheral ligation. A mixture of 1 /fdcH2 (1:4) was stirred in the presence of benzyl alcohol. This reaction allowed for the isolation of a microcrys talline product not suitable for X ray analysis. Its IR comparison with the previously characterized complex 2 showed that it had a different structure. However, this new product was not pursued, as it did not show sufficient solubility in any common solve nt Detailed synthetic studies of the initial reaction system for 2 was completed with the isolation of another Mn cluster incorporating the ligand fdc2. This new compound has the molecular formula [Mn8O4(fdc)6(DMF)2(H2O)2] 4DMF 4H2O [( 4 ) 4DMF 4H2O]. The brown precipitate resulting from the reaction of 1 with fdcH2 was resuspended in a mixture (1:1) of DMF/CH2Cl2 were it would dissolve only slightly and the filtered D ark needle shaped crystals were obtained by slow diffusion of ether into the filtrate. T he new compound was characterized by singlecrystal X ray diffraction and supported by IR and elemental analysis. Eq. 2 2 summarizes the synthesis of 3 and 4 Unfortunately, the yield of 4 is very low ( 2%), as otherwise expected for a minor reaction product and for the large amount of undissolved precipitate. Careful inspection of the crystalline product under an optical microscope showed the presence of an amorphous substance in addition to the crystals of 4 Manual separation and IR analysis suggested the amorphous product to be structurally very close to the cluster 4 although a complete identification was never achieved.


63 (2 2) Due to the low yield and cocrystallization of unidentified material complex 4 could not be obtained in high purity via this synthetic route. Hence, efforts were made to find another route to the Mn8 cluster. An alternative method was found when different Mn sources were explored and the ligand precursor fdcH2 was used in a comproportionation reaction between MnII and MnVII to generate MnIII. A mixture of Mn(O2CMe)2 and fdcH2 in EtOH/py w as oxidized with tetrabutylammonium permanganate (TBAMnO4) to give a brown suspension. Recrystallization of the collected solid from DMF/CH2Cl2 (1:1) led to complex 4 but with an unidentified byproduct The s cheme for the alternative synthesis of 4 is shown in Eq 2 3 (2 3) Although the new procedure to 4 is still contaminated with some impurity, it is overall superior to the earlier one, and suggested that further modification might allow pure 4 to be obtained. T hrough other experiments, it had been found that addi tion of small amounts of triethylamine (TEA) to the reaction sy stem s can completely prevent the co crystallization of byproducts This modification in the present reaction successfully allowed for the sole isolation of the Mn8 complex in higher yields ( ~ 40%)


64 (s ee Eq 2 4 ) T he product was obtained as a powder, but IR spectra, elemental analysis and magneti c measurement s all ind icated the product to be pure. (2 4) The role of the TEA in preventing the accumulation of impurities in unclear The reaction mechanism may be influenced by TEA in several ways The TEA can change the pH, and also may act as a mild reducing agent. With comproportionation proving a superior approach, further improvements were sought to the yield. A new reaction scheme was subsequently developed by considering factors such as t he average oxidation state of the core, the ratio of the Mn ions to the oxide i ons the number and protonation state of the ligands as well as the number and type of the peripheral ligation. The formula [Mn4 IIIMn4 IIO4(fdc)6(DMF)2(H2O)2] indicated an average oxidation state of +2.5 for Mn There are on average two Mn2.5+ ions per oxid e six fully deprotonated fdc2 units, two DMF molecules and two water molecules To reach the intermediate state of Mn2.5+ a comproportionation reaction between MnII and MnVII could be employed. An improved procedure was devised using [Mn12O12(O2CMe)16(H2O)4] was used as the source of the higher oxidation state ( Mn3.33+) and Mn2+(O2CMe)2 for the lower oxidation state. It was thought that t here may be a possible advantage in a combination where one of the metal sources is a preformed cluster and the other is mononuclear rather than both being mononuclear. I t is conceivable that during the reaction some fragmentation and rearrangement of the Mn12 core occur to form the final


65 product of different nuclearity. In this process, the immediate availabili ty of a mo nonuclear species might assist in completing fragments leading to the Mn8 product. The Mn12 cluster is also a good source of oxide ions with a Mn/O ratio of 1:1. it was suspected that this might also assist formation of Mn8 upon addi tion of the mononuclear MnII, by approaching the expected ratio of 2:1 for the Mn8 cluster. Finally, the presence of DMF as solvent can provide terminal ligands and keep intermediates soluble allowing for clean formation of Mn8. This approach worked spectacularly well. The react ion scheme is shown in Eq. 2 5 The product is the new cluster [Mn4 IIIMn4 IIO4(fdc)6(DMF)4], ( 5 ), which except for two DMF molecules replacing water, is identical in composition and structural topology to 4 The compound forms X ray quality crystals with an overall yield of isolated product of ~ 85% with respect to Mn! Cluster 5 was characterized by X ray crystallography, IR spectroscopy and elemental analysis. (2 5) Attempted application of this synthetic strategy to Mn13 clusters raised immediate challenges due to the solvent. While the Mn13 clusters have alkoxides as peripheral ligands, complex 1 is not soluble in alcohols. The solubility of the reactants is crucial for reaching at the initial stages of the reaction; the targeted Mn oxidation state thro ugh comproportionation. While a solvent like acetonitrile could be suitable for complex 1 the same is not true for fdcH2. To avoid these problems, [Mn12O12(O2CMe)16(H2O)4] was treated in a acetonitrile /toluene mixture with eight equivalent s of trimeth ylac etic acid and evaporated once to drive an acetate substitution reaction. The partial ly substituted


66 product, [Mn12O12(O2CMe)16n(O2CtBu)n(H2O)4] is readily soluble in ethanol and it was employed in a comproportionation reaction with Mn(O2CMe)2 in the presence of fdcH2. The system gives an immediate formation of a glossy brown precipitate that the IR analysis suggested was composed mainly of the Mn8 cluster. This suggests preferential formation of a Mn8 structure over Mn13 and it was not further pursued. Des cription of the M olecular S tructures [Mn13O8( OMe )6(fdc)6] 8CH2Cl2; (2 ) 8CH2Cl2 A Pov Ray projection of complex 2 and its labeled core are shown in Fig. 210 and Fig. 211. The crystallographic data and structure refinement details are listed in Table 21 The complex crystallized in the triclinic space group P 1 and consists of a [MnIVMnIII 6MnII 6( 5O)6( 3O)2( 3OMe)6]12+ core with the peripheral ligation provided by six bridging 4fdc2 ligands. The core can be described as consisting of three layers; a central one where six coplanar MnII surround the central and unique MnIV. Other two layers are present one above and one below and contain each three MnIII ions Six of the eight oxo ligands are unusual in the sense that they are pentacoordinated, whereas the remaining two posses the more common coordination three. Table 2 1. Selected c rystal data for complex 2 8CH2Cl2 (full data in Tab. B 1) Parameter Value Empirical formula C 92 H 94 Cl 16 Fe 6 Mn 13 O 38 Formula weight g mol 1 3424.26 Temperature K 173(2) Wavelength a 0.71073 Crystal system Triclinic Space group P 1 Unit cell dimensions a = 14.1308(12) b = 15.1738(12) c = 15.3656(13) Unit cell angles, deg = 119.4700(10) = 91.4320(10) = 100.339(2) Volume 3 2797.0(4) Z 1 calc g cm 3 1.969


67 Figure 2 10 Pov Ray projection of 2 : MnIV cyan, MnIII light green, MnII yellow, O red, Fe orange. Hydrogen atoms have been omitted for clarity Figure 2 11 Pov Ray projection of the [MnIVMnIII 6MnII 6( 5O)6( 3O)2]18+ core in [Mn13O8( O Me )6(fdc)6] : MnIV cyan, MnIII light green, MnII yellow, O red


68 The assignment of the oxidation states was based on charge balance considerations and quantitative evaluation of the bond lengths through Bond Valence Sum (BVS) calculations. The class ical BVS value for an atom is an estimated value of its valence based on a collection of bond lengths measured through X ray crystallography. The ion's oxidation state (zj) is estimated through t he sum of individual valences assigned to each of its bonds ( sij) Every ( sij) value derives from the existing correlation between bond length and bond strength through two experimentally determined parameter s, R0 and b (this last one usually taken as 0.37 ) .64,7071,129 131 The BVS method has been explained in detail in C hapter 1 (see Eq. 1 6 ) and used for these determinations without application of the newly optimized parameters R0 and b discussed in C hapter 4 Table 22 summarizes the BVS calculations for complex 2 showing in bold the number closer to the actual valence value. Table 22 Bond Valence Sum values for complex 2 Atom Mn II Mn III Mn IV Mn1 3.407 3.142 3.083 Mn2 2.082 1.920 1.884 Mn3 2.025 1.867 1.832 Mn4 1.988 1.833 1.799 Mn5 3.366 3.104 3.046 M n6 3.377 3.114 3.056 Mn7 4.457 4.110 4.033 The number with an asterisk is the one closest to the actual charge for which it was calculated. The oxidation state of a particular atom can be taken as the nearest whole number to the value in bold. The B VS method fails if applied to the esti mation of the oxidation state of a n iron in a ferrocenyl unit. As already mentioned in the I ntroduction, the two common oxidation states for the iron within cyclopentadienyl ligands display only a small difference in t he FeC bond lengths. Furthermore, ferrocenyl units display an opposite trend to the BVS, where the FeC bonds are elongated with the higher metal oxidation state (FeIII) and


69 slightly shorter with the lower oxidation state (FeII).132 133 In this case, t he difficult assessment of the iron oxidation state in the fdc units through statistical methods can be avoided. From charge balance considerations, it is clear that all the iron atoms must be in the FeII state i.e. a s ferrocenyl units The previous description of the core of 2 consisting of three layers (Fig. 2.11) : one comprising a MnII 6 hexagon with a central MnIV atom and MnIII 3 triangular layers above and below implies for this molecule the coexistence of three d ifferent Mn oxidation states as well besides oxygen ions with unusual 5 coordination: the core is held together by six 5O2, two 3O2 and six 3O Me 4fdc2 groups with syn syn syn syn 1: 1: 1: 1: 4 ligation mode. The torsion angle between the two carboxylate groups of the fdc2 is less than 30, accounting for a synperiplanar configuration134. The six MnIII ions have nearly octahedral coordination geometry with JahnTeller (JT) distortion s, as expected for highspin d4 ions. The distortion is typical in the fact that two tran sbonds are elongated and are the weaker among the six. The elongation is in fact observed along the axis containing the carboxylate ( 5O2) Mn ( 4fdc2) rather than two oxides ( 3O2) Mn ( 3O2) with an averaged difference of 0. 262 Such an arrangem ent is often seen in cluster chemistry, where the peripheral ligation can easily rearrange, while the core has a tendency of being more rigid, hence being less prone to distort ions The same arrangement of Mn ions in a cluster was first reported in 1996.135 The packing diagram of the complex revealed a large presence of H interactions involving the H atom s of the cyclopentadienyl rings with the electron cloud of


70 cyclopentadienyl rings on adjacent molecules (Fig 2 12 ). These interactions very likely contribute to the complete insolubility of the crystals in common solvents. Figure 2 12. Packing diagram for complex 2 in Mercury136. The H interactions b etween the molecules in the lattice are shown with dotted lines [Mn8O4(fdc)6(DMF)2(H2O)2] 4DMF 4H2O; (4 ) DMF 4H2O Fig. 213 and 214 show Pov Ray projections of complex 4 and its core respectively. Table 23 presents the main crystallographic data and structure refinement details. The complex crystallizes in the monoclinic space group C2/c. The cluster contains a core comprising a central [MnIII 4( 4O)4]4+ cuba ne where each O2 ion is attached to a MnII ion The overall topology results from two concentric Mn4 tetrahedra. The peripheral ligation is completed by six fdc units and four terminal positions are occupied by two DMF and two water molecules. BVS calcula tions were applied as in the case of the Mn13 and allowed for the clear identification of the oxidation states of the various Mn ions. The inner Mn ions in the core were confirmed as MnIII, whereas all the external pentacoordinated ions were identified as MnII. Overall the core is made of an inner MnIII tetrahedron and outer MnII tetrahedron conferring to the core pleasant aesthetic properties for a virtual Td symmetry


71 Table 2 3. Selected crystal data for complex 4 4DMF4H2O (full data in Tab. B 2) Pa rameter Value Empirical formula C 90 H 102 N 6 Fe 6 Mn 8 O 40 Formula weight g mol 1 2682.40 Temperature K 173(2) Wavelength a 0.71073 Crystal system Monoclinic Space group C2/c Unit cell dimensions a = 23.282(3) b = 19.331(3), c = 22.198(3) Unit cell angles, deg = 90 = 97.885(3) = 90 Volume 3 9896(3) Z 4 calc g cm 3 1.660 Figure 2 13 Pov Ray projection of [Mn8O4(fdc)6(DMF)2(H2O)2] : MnIII light green, MnII yellow, O red, Fe orange, N blue. Hydrogen atoms have been omitted for clarity The four MnII ions are very close to a trigonal bipyramidal geometry with values of 0.78 and 0.81 for Mn1 and Mn4 respectively. is a parameter that quantitates the extent of distortion from a square pyramidal to trigonal bipyrami dal geometry for a general ML5 complex. If the two set s of trans ligands at the base of a square pyramidal


72 complex are called Lx, Lx and Ly, Ly while the apical ligand is Lz, the largest basal angle is called (for instance LxM Lx ) After rearrangem ent 137, will become the axis of the trigonal bipyramid. The other basal angle in the square pyramidal configuration (LyM Ly ) is called The degree of structural distortion is obtained through the formula: = ( ) / 60, where is 0 for an ideal square pyramidal complex and increases to a maximum of = 1 for an ideal trigonal bipyramid geometry .138139 Figure 2 14 Pov Ray projection of the core [MnIII 4Mn4 II( 4O)4]12+ in 4 : MnIII light green, MnII yellow, O red The MnII ions (Mn1, Mn4) are fixed at the corners of an tetrahedron that includes a smaller tetrahedron of MnIII io ns (Mn2, Mn3) R esults of the BVS calculations are shown in Table 24 Table 24 Bond Valence Sum calculations for complex 4 Atom Mn II Mn III Mn IV Mn1 2.113 1.949 1.912 Mn2 3.284 3.028 2.972 Mn3 3.24 0 2.99 0 2.937 Mn4 2.162 1.994 1.956 The num ber with an asterisk is the one closest to the actual charge for which it was calculated, and the nearest whole number to it is t he oxidation state of that atom The four MnIII ions show JahnTeller distortion, as expected, with axial elongations of 0.265 and 0.451 for Mn3 and Mn2, respectively compared with the compressed


73 axe s. The four JahnTeller axes on the cluster are arranged nearly parallel with respect to each other. The coordination around the core consists of six fdc2 groups A s in the case of complexes 2 and 3 it is possible to infer the charge of the ferrocenyl units by charge balance. E ach carboxylate group of an fdc2 bridges a MnII/MnIII pair adopting the same syn, syn, 1: 1 : 4coordination mode seen in complex 2 The peripheral liga tion is completed by O bound DMF molecules for two of the four MnII ions and water for the remaining two MnII. The torsion angle s between the two carboxylate groups of the fdc2 units vary: four fdc2 groups have torsion angles of <30 (synperiplanar confi guration) and the other two have angles between 30 and 90 (synclinal configuration).134 The packing diagram of 4 also shows the presence of H interactions among the ferrocenyl units belonging to adjacent clusters (Fig. 2 15). As for [Mn13O8(O Me )6(fdc)6] complex 4 is very in soluble, and t he H interactions likely contribute significantly to the lattice stabilization. Figure 2 15 Packing diagram for complex 4 in Mercury136. Three molecules are s hown a long with dotted lines indicating some of the H intermolecular interactions. [Mn8O4(fdc)6(DMF)4] 4 DMF; (5 ) 4 DMF The crystallographic structure of complex 5 is analogous to that of 4 with respect to the core description and main ligation. The only m ajor difference is in the peripheral


74 presence of four O bound DMF molecules, instead of two DMF and two water molecules. The terminal DMF molecule directly bound to the Mn6 ion is disordered with one water molecule, a nd the latter appears 30% of time C rystallogr aphic data is listed in Tab. 25 The structure underwent full refinement, but without the creation of complete crystallographic tables. The cluster also shows a lowered symmetry in that it crystallizes in the P21/c space group and all the Mn ions are thus crystallographically inequivalent. Fig. 2 16 shows the core of 5 and its superposition with the core of 4 The central cubanes are essentially superimposable, but there are some small differences in the angles to the external MnII. Table 2 5 Sele cted crystal data for complex 5 4 DMF (full data in Tab. B 3) Parameter Value Empirical formula C 9 6 H 104 N 8 Fe 6 Mn 8 O 36 Formula weight g mol 1 2 720 49 Temperature K 173(2) Wavelength a 0.71073 Crystal system Monoclinic Space group P 2 1 /c Unit cell dimensions a = 24 015(2 ) b = 14.865(14), c = 29.558 (3) Unit cell angles, deg = 90 = 101.088(5 ) = 90 Volume 3 10354.5 (3) Z 4 calc g cm 3 1.745 BVS calculations confirmed the oxidation state of the various Mn ions and the valu es are reported in Table 26 The MnIII ions in clusters 4 and 5 display the common JahnTeller distortion with axial elongation, where the axes are nearly parallel to each other. These distortions cause the two faces of the inner cubanes perpendicular to the elongation axis to be nearly rhombic (sides of equal length) whereas the remaining four faces on each of the two cubanes are deformed close to a rhomboidal shape (opposite sides of equal length)


75 Fig ure. 2 16. Left: Pov Ray projection of com plex 5 core: MnIII light green, MnII yellow, O red Right: structural overlap of the two cluster cores 4 and 5 The green arrow indicate the direction for the JahnTeller elongation. Table 26 Bond Valence Sum calculations for complex 5 Atom Mn II Mn III Mn IV Mn1 3.344 3.083 3.026 Mn2 3.341 3.081 3.023 Mn3 3.434 3.167 3.107 Mn4 3.407 3.142 3.083 Mn 5 2.086 1.923 1.887 Mn 6 2.097 1.934 1.897 Mn 7 2.077 1.915 1.879 Mn 8 2.087 1.924 1.888 The number with an asterisk is the one closest to the actual charge for which it was calculated and the nearest whole number to it is t he oxidation state of that atom The JT elongations for the MnIII ions are 0.589, 0.620, 0.602 and 0.540 for Mn1, Mn2, Mn3, and Mn4, respectively, compared to the compr ession axes. These values are significantly higher if compared than those observed for the Mn3 values in 4 Another peculiarity of cluster 5 is with respect to the geometry of its MnII ions In fact only the Mn6 and Mn7 compare to the ones found in comple x 4 with values of 0.84 and 0.74 (hence close to a trigonal bipyramid). Mn5 and Mn8 on the contrary show much smaller values of 0.22 and 0.54 respectively ( hence closer to a square pyramid). The coordination and disposition of the fdc2 units is almo st unchanged as compared to


76 4 although differences in the ferrocenyl torsion angles reflect the distortion of the cluster core. There is one fdc2 ligand above each face of the inner MnIII cubane and the fdc2 keeps both carboxyl ate unit s nearly parallel while binding two metal ions on that cubane face. The torsion angle of the ligand units is a consequence of the coordination M1 O (O C fc CO) O M2 while satisfy ing the M1 M2 distance on the same cubane face. According to this analysis the smaller torsion angle ( ~ 29) is found for both 4 and 5 on the fdc2 unit s above the nearly rhombic faces of the cubane. T he remaining faces are mor e distorted from an idealized square or rhombic geometry requiring slightly larger fdc2 torsion angles ( ~ 30 and ~ 36) A comparison of the torsion angles on the two molecular systems reveal s differences of only a few degrees, nonetheless these differenc es are significant The molecular lattice shows H interactions analogous to 4 but to a lesser extent In this case, the terminal DMF molecules have a larger contribution in establishing contacts with neighboring molecules. S ome of the intermolecular int eractions in 5 are shown in Fig. 2 17 where as for the Mn13 cluster hydrogen atoms from a ferrocenyl unit interact with the electronic cloud of another ferrocenyl group. Figure 2 17. Packing diagram of complex 5 in Mercury136. D otted lines indicat e some of the H interactions in the lattice.


77 Magnetochemistry [Mn13O8( O Me )6(fdc)6] 4H2O 3MeOH ; (2) 4H2O 3MeOH and [Mn13O8( O Et )6(fdc)6] 7EtOH; (3) 7EtOH Variable temperature DC susceptibility measurements were performed in the 5.0 to 300 K range on pow d ered microcrystalline samples of complexes 2 and 3 The microcrystals were embedded in eicosane to prevent torquing after the application of magnetic field. This experiment requires the use of a constant field which was set at 1.0 kG (0.1 T); (Fig. 2 18). The data for complexes 2 and 3 are nearly superimposable with an initial value of ~ 47 cm3Kmol1 at 300 K that steadily decreases to ~ 11.5 cm3Kmol1 at 5 K. This profile indicates predominant antiferromagnetic interactions among the Mn ions within the cluster. It should be noted that the diamagnetic ferrocenyl units do not have any contribution to the magnetic properti es these clusters, as indicated in the crystal structures. The high nuclearity of these clusters prevent fitting of the data by matrix diagonalization to obtain the individual exchange parameters. T / K 0 50 100 150 200 250 300 MT / cm3Kmol-1 10 20 30 40 50 Cluster 2 Cluster 3 Figure 2 18. Plot o f MT vs T for complex 2 4H2OMeOH ( ) and complex 3 EtOH ( )


78 Also t he Kambe equivalent operator method32 can not be appli ed for this case. However, the groundstate spin ( S ) of the complexes was determined by two independent methods: fitting of the variable te mperature (T) and variablefield (H) DC magnetization data, and extrapolation to low temperature of the AC susceptibility measurements (Fig. 219) DC magnetization data were collected in the 1.810.0 K range in fields varying between 0.1 and 7 T and fit using the program MAGNET140 to a model that assumes exclusive population of the ground state. H/T / kG/K 0 1 2 3 4 5 M/N B 0 1 2 3 4 5 6 0.1 T 0.2 T 0.3 T 0.4 T 0.5 T 0.6 T 0.7 T 0.8 T Fit Figure 2 19. Plot of reduced magnetization (M/N B) vs. H/T for a sample of 3 7EtOH. The analysis includes axial zerofield splitting ( z 2D ) and the Zeeman effect ( B0 H ), and incorporates a full powder average. A good fit could not be obtained using all data up to 7 T. This is typically the observed when there are low lying excited states One way to avoid this problem is to use only data collected in small fields. In this case, a satisfactory fit was obtained using only data up to 0.8 T (solid lines in Fig. 219).


79 The ground state spin S for complex 3 was found to be 9/2, with a g and D value of 1.91 ( 0.10) and 0.23 ( 0. 0 5 ) cm1, respectively. H/T / kG/K 0 1 2 3 4 5 M/N B 0 1 2 3 4 5 6 0.1 T 0.2 T 0.3 T 0.4 T 0.5 T 0.6 T 0.7 T 0.8 T Fit Figure 2 19. Plot of reduced magnetization (M/N B) vs. H/T for a sample of 3 7EtOH To assess the precision of the fit parameters, the root mean square error surface for the fit was generated as a funct ion of g and D using the program GRID.141 The obtained error surface is depicted in Fig. 220 as a convenient 2 D contour plot, and the absolute minimum is the best fit estimation of g and D The nat ure of the minimum in the error surface reflects the precision of the fit parameters Hence, a poorly def ined and "soft" minimum will be correlated with large errors in the estimated g and D fit values. Estimation of 0.10 and 0.05 for the absolute error in g and D respectively is a safe assumption. Another method to estimate the ground state S value for 3 is to measure the AC response of a sample in a particular temperature range, in this case 1.8 to 15.0 K using a 3.5 G AC field oscillating at 50, 250 and 997 Hz. AC


80 measurements are performed in the absence of a DC field, and can thus avoid complication from low lying excited states as expected for 3 Figure 2 20. g vs. D e rror surface for the m/N B vs. H/T fit for 3 7EtOH. The numbers indicate the magnitude of the error and slowly decrease on moving towards the absolute minimum, marked with a star The real, or inphase, AC susceptibility (M' ) of the complex is plotted as M T vs. T in Fig. 221. Extrapolation of the M T to 0 K, from temperatures above ~4 K ( to avoid the effects of w eak intermolecular interactions) gives a value of ~10.5 cm3Kmol1. This value indicates a ground state of 9/2 for this cluster and g 2, as expected for Mn, in agreement with the DC magnetization fits. Data for com plex 2 are similar to those fro complex 3 The alternating current out of phase data down to 1.8 K for complexes 2 and 3 show no out of phase M'' AC signals that would indicate the slow relaxation of the magnetization vector (Fig. 222). To explore whether slow relaxation might be seen at


81 even lower temperatures, single crystals were measured using a microSQUID apparatus down to 0.04 K with the precaution of leaving the crystals in the mother liquor up to the measure to prevent solvent loss .142 Magnetization vs. DC field measurements were performed at different temperatures and at various field scan rates (Fig. 223). T / K 0 2 4 6 8 10 12 14 16 M'T / cm3Kmol-1 10 11 12 13 14 15 16 997 Hz 250 Hz 50 Hz Figure 2 21 Plot of M T vs. T for complex 3 EtOH. T / K 0 2 4 6 8 10 12 14 16 M" / cm3mol-1 -2 -1 0 1 2 997 Hz 250 Hz 50 Hz Figure 2 22 Plot of M vs. T for complex 3 7EtOH in the 1.815 K.


82 The absence of hysteresis indicates the absence of a significant barrier to the relaxation even at 0.04 K. The similar behavior for the two species 2 and 3 isomeric with respect to the core, but with a different set of peripheral alkoxides suggests very poor influence of the alkoxides in the coupling patterns. This argument may be extended to the hysteresis measurement in expecting the same type of behavior for both 2 and 3 Figure 2 23 Hysteresis measures for complex 2 8CH2Cl2 in the 0.041 K range. [Mn8O4(fdc)6(DMF)2(H2O)2] 4DMF4H2O; (4) 4DMF4H2O A m icrocrystalline sample of 4 4DMF4H2O was embedded in eicosane and its magnetic behavior was investigated as previously discussed for complexes 2 and 3 ( vide supra). Variabletemperature DC susceptibility measurements were performed leading to the plot shown in Fig. 224. The MT profile for complex 4 4DMF4H2O remains almost unchanged with decreasing temperature after starting at about 15 cm3Kmol1. The data fluctuate slightly before reaching a final value of 13.64 cm3Kmol1


83 at 5.00 K. The overall decrease of susceptibility is indicative of predominant antiferromagnet ic interactions among the Mn ions within the cluster consistent with depopulation of excited states of higher spin (i.e.: with larger ferromagnetic presence) The contribution of the FeII ions to the magnetism has been disregarded, since the electron exchange rate between the FeII ions and the core is expected to be negligible.110 T / K 0 50 100 150 200 250 300 MT / cm3Kmol-1 5 10 15 20 25 30 Figure 2 2 4 Plot of MT vs. T for complex 4 4DMF4H2O The complexity of the structure despite the relatively high symmetry also compared to the previous Mn13 clusters, again prevents application of the Kambe vector type of approach to determine the various coupling patterns. Fitting of the variabletemperature (T) and variablefield (H) DC magnetizati on data, and AC susceptibility measurements were therefore employed to determine the groundstate spin. The reduced magnetization data w as acquired using the methods previously described for 2 and 3 by fitting of the data with Magnet136 (Fig 225)


84 H/T / kG/K 0 5 10 15 20 25 M/N B 0 2 4 6 8 10 0.1 T 0.5 T 1 T 2 T 3 T 4 T Fit Figure 2 25 Plot of reduced magnetization (M/N B) vs. H/T for 4 4DMF4H2O In the fit of reduced magnetization ( RM ) data vs. H/T, data collected in fields up to 4 T were used for the fit, suggesting a ground state better is olated from excited states than for 2 and 3 The fit suggests a spin groundstate S = 5 with g and D parameters of 1.97 ( 0.10) and 0.32 ( 0. 0 5 ) cm1, respectively. The precision of the fit has been estimated by the error on the magnetization as a surface function of g and D Fig. 2.26 shows the 2D contour plot of this surface where a well isolated absolute minimum appears in the region about the g and D values obtained with the pr ogram M AGNET As for complexes 2 and 3 it is safe to estimate the error on g and D as 0.10 and 0.05 cm1, respectively. Alternate current measurements on a sample of 4 were taken following the same procedure described previously for 2 and 3 The inphase component of the susceptibility, M T shows a monotonic decrease with lowering of the temperature suggesting depopulation of excited states of higher spin with respect to the ground state ( Fig. 2 27). A nonlinear decrease with lowering the temperature may also be due to weak intermolecular interactions.


85 Figure 2 26. Error plot for the estimation of g and D in 4 4DMF4H2O. The numbers indicate the magnitude of the error, which slowly decreases moving towards the absolute minimum, marked with an asterisk. T / K 0 2 4 6 8 10 12 14 16 M'T / cm3Kmol-1 13.0 13.5 14.0 14.5 15.0 15.5 16.0 997 Hz 250 Hz 50 Hz Figure 2 27 Plot of M T vs. T for complex 4 4DMF4H2O in the 1.815 K.


86 The initial M T value of ~ 15.6 cm3Kmol1 at 15 K reaches ~ 13.7 cm3Kmol1 at 2.2 K with a nearly linear progression. At lower temperatures, the M T value has a sudden drop reaching ~ 13.2 cm3Kmol1 at 1.8 K The sharp decrease is due to slow relaxation of the magnetization vector by partial transfer of the inphase susceptibility component to the out of phase. In extrapolating the M T value to 0 K for the estimation of the ground state, the region where slo w relaxation occurs should be avoided as well as data generally at higher temperature up to ~ 6 K because potentially it is affected by intermolecular interactions The plot suggests a M T of ~ 13.8 cm3K mol1 at 0 K for which it corresponds a spin groundst ate S = 5 with g = 1.92. This is a reliable estimation considering that the immediately neighboring integer values of S (i.e., 4 and 6) would lead to unacceptable values of g for a Mn oxide compound. The AC in phase data are hence confirming the S = 5 ground state for 4 as suggested by the reduced magnetization data. As expected from the profile of the in phase magnetic susceptibility vs. T plot the out of phase data display an increase in magnitude from zero at the lower end of the selected temperature range (Fig. 228). The M vs. T plot shows tails of out of phase peaks consistent with the sudden drop in the M T between 2.2 and 1.8 K A process of slow relaxation of the magnetization is also consistent with the appearance at higher temperatures for the out of phase component s M detected at a higher frequency of the oscillating field. Further investigation of this phenomenon requires lower temperatures and a microSQUID. Magnetization versus DC fields were measured on a single crystal to verify if the slow relaxation of the magnetization detected below ~2.5 K may lead to hysteresis that displays temperature and scan rate dependence (Fig. 229), i.e. if 4 is a single molecule magnet. The scanning of the


87 magnetization vs. applied field at different temper atures with a fixed scan rate produced the plot in Fig. 2.29 where it is evident some hysteresis occurs below 2 K. Most importantly there is an increase of the coercivity with the decreasing temperature; one of the diagnostic features of a singlemolecule magnet. T / K 0 2 4 6 8 10 12 14 16 M" / cm3mol-1 -0.02 0.00 0.02 0.04 0.06 0.08 0.10 0.12 997 Hz 250 Hz 50 Hz Figure 2 28 Plot of M vs. T for complex 4 4DMF4H2O in the 1.815 K. Figure 2 29. Hysteresis measures for complex 4 DMFH2O in the 0.041 K.


88 [Mn8O4(fdc)6(DMF)4] 1. 5DMF 3 H2O; (5) 1.5 DMF 3 H2O Complex 5 was studied under the same conditions as compound 4 DC measurements for a microcrystalline sample of 5 show a monotonic ally decreasing value of MT with decreasing temperature (Fig. 2 30) T / K 0 50 100 150 200 250 300 MT / cm3Kmol-1 0 5 10 15 20 25 Figure 2 30. Plot of MT vs. T for complex 5 1.5DMF 3 H2O From the initial 18.64 cm3K mol1 at room temperature, MT has a slow and steady decrease up to ~ 80 K from where it displays a more pronounced decline to the final point of 4.99 cm3K mol1 at 5 K. A reliable estimation of t he ground state from the MT vs. T plot was not possible due to the steep decrease of the data in the low temperature region. It is likely that the weak magnetic exchange interactions associated with the MnII ions produce excited states close to the ground state. With the aim of assess ing the ground state S of 5 the program M AGNET140 was used to attempt the fitting of reduced magnetization vs. H/T data. Unfortunately, the application failed at producing meaningful data for g and D with any combination of the measured fields in the 0.1 to 7


89 T. At this point, the estimation of S was based only on the AC inphase data, measured in the same conditions as for complex 4 (Fig. 2 31) T / K 0 2 4 6 8 10 12 14 16 MT / cm3Kmol-1 0 2 4 6 8 10 12 997 Hz 250 Hz 50 Hz Figure 2 31. Plot of M T vs. T for complex 5 1.5DMF3H2O in the 1.815 K. The plot of M T vs. T shows a decrease in susceptibility with almost constant progression up to the final point of 2.98 cm3K mol1 at 1.8 K from the initial 9.86 cm3K mol1 at 15 K. Extrapolation to 0 K fr om the data at higher temperature led to ~3 cm3K mol1, suggesting of the spin groundstate S = 2 ( g = 2) for complex 5 The adjacent integer spins S = 3 or 1 with respect to the ground state, respectively, are excluded due to the unacceptable g values impl ied being either too low or too high, respectively The AC out of phase measurements do not indicate slow relaxation of the magnetization with decreasing temperature. However, the data points near 1.8 K suggest the beginning of a peak. It is possible that the molecule has a substantial magnetic anisotropy, but its spin groundstate is much lower than 4 and not large enough to create a sufficiently high barrier to the reorientation of the magnetic vector (See Fig. 232 and inset).


90 T / K 0 2 4 6 8 10 12 14 16 M" / cm3mol-1 0.0 0.5 1.0 1.5 2.0 2.5 3.0 997 Hz 250 Hz 50 Hz 2.0 2.5 3.0 3.5 4.0 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 Figure 2 32. Plot of M vs. T for complex 5 1.5DMF3H2O. The inset shows tails of peaks in the 1.8 to 4 K region. Compounds 4 and 5 clearly exemplify how small differences in molecular structure can result in a large difference in magnetic behavior. As described in detail in the structure elucidation, 4 is slightly more symmetric than 5 which contains two nearly square pyramidal MnII ions, and a more asymmetrically distorted central cubane due to a stronger JahnTeller elongation. As previously mentioned, t he unambiguous assessment of the magnetic exchange parameters for cluster 4 is prohibitive due to its high nuclearity and relatively low symmetry. The same applies to an even greater extent for complex 5 due to its lower symmetry with respect to 4 For these reasons, only general conclusions about the different magnetic behavior of 4 and 5 can be made. Compounds 4 and 5 manifest magnetic differences beginning with the room temperature data of the MT vs. T (Fig. 233). According to the spinonly formula, both compo unds should display a MT of ~ 29.5 cm3Kmol1 at 300 K accounting for the independent magnetic behavior of each metal ion.


91 T / K 0 50 100 150 200 250 300 MT / cm3Kmol-1 0 5 10 15 20 25 Cluster 5 Cluster 4 Figure 2 33. Superposition of the MT vs T p lot s for 4 DMFH2O and 5 3H2O 1.5DMF. However, both possess significantly lower susceptibility values that imply strong antiferromagnetic couplings even at room temperature. As the temperature is lowered, both 4 and 5 behave similarly with a steady decrease in MT up to ~ 150 K. Below this point the two molecules div erge showing different MT vs. T profiles and ground state values. Magnetic studies on clusters have largely demonstrated how s mall variations of angles along magnetic exchange pathway or bond lengths can affect the magnitude of the magnetic interaction.143145 Hence, different J values are ultimately responsible for the different magnetic behavior of 4 and 5 Electrochemistry Compounds 2 3 4 and 5 each have six ferrocenyl units bound to the core. This could resul t in the observation of multielectron transfer through cyclic voltammetry. The literature presents such a process for complex 2 .101 The investigation was conducted in the solid state and the results were compared wi th the free ligand precursor fdcH2 (Fig.


92 2 34). The CV shows the oxidation of all the ferrocenyl units on the cluster occurring at the same potential. The authors explained this behavior by describing the cluster as an assembly of ferrocenyl units, which l ack mutual electronic influence, so that the redox processes on each of them can be considered as an independent event. The authors also explain how a shift to a higher potential for the ferrocenyl units in the cluster is due to the electronwithdrawing ef fect exerted by the cluster's core with respect to the free ligand precursor. Solutionphase CV analysis for complexes 2 3 4 and 5 failed due to their low solubility in common nonaqueous media. No significant signals could be observed. This represents t he main obstacle for CV investigation of these clusters. Figure 2 34 Solid state voltammogram of (a) fdcH2, and (b) complex 2. The measurements are referenced against the standard calomel electrode (SCE) (reported from ref. 101, Reproduced by permission of The Royal Society of Chemistry, http://dx.doi.org/10.1039/b210475j )


93 Chemical oxidation of 2 3 and 4 was attempted in order to try to isolate the correspondi ng cationic species that may be more soluble. The presence of fdc2 in the compounds discussed so far allows the possibility for the oxidation of one or more of these ferrocenyl units leading to a net positive charge on the molecule. This is often an advantage for the dissolution of compounds in polar solvents. C ommon oxidizing agents like CeIV salts (NH4)2Ce(NO3)6 or Ce(OH)4, were added to a slurry of the compounds 2 3 or 4 in acetonitri le, benzonitrile, dimethylformamide or dimethylsulfoxide. In no cas e, did clean oxidation to a soluble species occur Despite the discoloration of the solution with time (for CeIV to CeIII), there was no build up of a blue color from charged clusters in solution (ferricenyl units have intense blue color) To e nsure a suff i cient oxidizing strength, attempts were made with the use of a stronger oxidizing agent than for CeIV in MeCN (i.e.: lower reduction potential). The reagent of choice was as a triarylaminium radical cation, [N(aryl)3] +. This type of oxidizing agent is widely used in organic chemistry and to a less extent for inorganic molecules Substitution of the aryl positions with proper electronwithdrawing groups allows these cations to span a wide range of potentials.146 The [N(C6H4Br 2,4)3]+[SbCl6] has a redox potential in MeCN of 1.14 V vs. Fc 146148 and can be easily synthesiz ed.149 Unfortunately, even this oxidizing agent did not allow the isolation of the desired product s. On the other hand, attempts to reduce one of the Mn ions on the core were also made using ferrocene as a mild reducing agent in acetonitrile. It has been demonstrated that fc can reduce MnIII ions in a metal cluster and act as a counter ion in the isolated product.150 Several attempts with different solvents and counterions did not lead to the expected results. In all cases the substrate remained unreacted.


94 Summary and Conclusions It was hoped that 1,1 ferrocenedicarboxylic acid would lead to many new Mnx products when employed in cluster chemistry, but with few exceptions the results have been disappointing. The reaction with preformed [Mn12O12(O2CMe)16(H2O)4]4H2O4MeCO2H has led to the isolation of new polynuclear Mn13 clusters. Although complexes 2 and 3 are based on a core already known in the literature, [MnIVMnIII 6MnII 6( 5O)6( 3O)2( 3OMe)6]12+, t he two clusters have been synthesized differently from the published work101, and a more detailed study of its syn thesis and physico chemical properties has been reported. F urthermore the isolation of 3 demonstrates the possibility of minor changes in the peripheral ligation. Such a characteristic might be useful for studying the electrochemical behavior of the cluster and as a means to improve solub ility. The other Mn polynuclear complex es 4 and 5 exhibit a new core topology with the central [MnIII 4( 4O)4]4+ inside a tetrahedron of four MnII ions A similar core structure i s present in the literature only for a Co8 complex with benzoate peripheral ligation published by the Christou group in 1995 151, and an Fe8 complex by Raptis et al. in 1999.152 The use of dicarboxylate ligands with restricted flexibility has led to the isolation of cluster s possessing different structure and nuclearity The ferrocenyl units are also located in a highly symmetric fashion and can potentially display interesting electrochemical behavior. The high symmetry of complexes 2 and 3 affects t he position of the elongated Jahn Teller axes and this might be the cause of a low magnetic anisotropy A low D value, although negative, in conjunction with the predominant antiferro magnetic coupling may hinder SMM properties i n these molecules On the ot her


95 hand, complex es 4 and 5 posses s nearly parallel JahnT eller axes, which may be the cause of the slow relaxation of the magnetization for 4 Complex 5 however, has a much smaller ground state S Unfortunately, the lack of solubility of these clusters p revents their study in solution particular ly with re spect to the electrochemical behavior one of the original objectives of this work Important questions on the effect of modifications of the redox states of the ferrocenyl units and the resulting magneti c behavior remain unanswered. Experimental All manipulations were performed under aerobic conditions, using materials as received, except where otherwise stated. The preformed cluster [Mn12O12(O2CMe)16(H2O)4]4H2O4MeCO2H ( 1 ) was obtained by a published me thod36 as well as for [N(C6H4Br 2,4)3]+[SbCl6] which was also prepared according to literatu re .149 The ligand precursor fdcH2 was initially synthesized as previously indicated in this chapter and then obtained from commercial sources The experiments were initially all carried in the presence of light and then repeated later in the dark during both the reaction and crystallization of products This wa s done to rule out the possibility of interfering photochemical reactions of the ferrocenyl ligand. [Mn13O8( O Me )6(fdc)6] 4 H2O 3Me O H ; (2) 4H2O 3MeO H To a stirred suspension of fdcH2 (0.5 mmol, 0.137 g) in a mixture CH2Cl2/MeOH (1:1; 25 mL), was added com plex 1 as a solid (0.125 mmol, 0.257g). The mixture was stirred for one hour and then filtered on a P2 filter paper. A dark brown solution was obtained and carefully layered with diethyl ether. Dark brown needle shaped crystals begin to appear after one week. L onger crystallization times leads to the appearance of a precipitate, which has not been characterized. The crystals collected after one week


96 were dried under vacuum yielding ~5% of product. The crystals analyze for ( 2 )4H2OMeOH Anal. Calcd. (found) for C81H86Mn13Fe6O45: C 34.39 (34.42) %; H 3.06 (2.95) %. Selected IR data (KBr pellet, cm1): 1660 (vs), 1581 (vs), 1479 (vs), 1391 (vs), 1357 (vs), 1254 (w), 1193 (m), 1099 (w), 1026 (m), 924 (w), 821 (w), 798 (m), 780 (s), 605 (m), 582 (m), 506 (s), 456 (m). The same compound was obtained using a mixture of MeCN/MeOH (1:1, 25 mL). [Mn13O8(OEt )6(fdc)6] 7EtOH; (3 ) 7EtOH To a stirred suspension of fdcH2 (0.5 mmol, 0.137 g) in a mixture CH2Cl2/EtOH (1:1; 25 mL), was added complex 1 as a solid (0.125 mmol, 0.257g). The mixture was stirred for one hour and then filtered on a P2 filter paper. A dark brown solution was obtained and carefully layered with diethyl ether. Dark brown needle shaped crystals begin to appear after one week. In this case, a longer c rystallization time did not lead to the appearance of a powder. The crystals were isolated by filtraton and vacuum dried. The yield is ~ 7 %. The crystals analyze for ( 3 )7EtOH. Anal. Calcd. (found) for C98H102Mn13Fe6O45: C 38.60 (38.60) %; H 3.37 (3.51) % Selected IR data (KBr pellet, cm1): 1710 (w), 1562 (vs), 1487 (vs), 1394 (vs), 1361 (vs), 1193 (m), 1096 (vw), 1025 (s), 926 (w), 871 (w), 826 (w), 788 (m), 641 (s), 583 (s), 527 (vs), 477 (m). [Mn8O4(fdc)6(DMF)2(H2O)2] 4DMF4H2O; (4) 4DMF4H2O M etho d (a) Resuspending the solid residue from the preparation of ( 2 ) or ( 3 ) in DMF (10 mL) and filtering again on P2 filter paper gave a dark brown solution is obtained. The liquid formed by slow evaporation dark brown needles after 2 weeks. The yield of the v acuum dried product after the first crystal collection is ~2%. Longer crystallization time promotes the formation of a nonidentified precipitate. The crystals analyze for ( 4 )4DMF4H2O. Anal. Calcd. (found) for C90H102N6Mn8Fe6O40: C, 40.30 (41.60) %, H


97 3. 83 (3.59) %, N, 3.13 (2.44) %. Selected IR data (KBr pellet, cm1): 1685 (s), 1577 (vs), 1475 (vs), 1392 (vs), 1537 (vs), 1197 (m), 1029 (m), 923 (w), 829 (w), 798 (w), 780 (m), 635 (m), 517 (s). [Mn8O4(fdc)6(DMF)2(H2O)2] 3DMF2H2O; (4) 3DMF2 H2O M ethod (b) To a slurry of Mn(O2CMe)2 (0.066 mmol, 0.0163 g.) and fcdH2 (0.55 mmol, 0.150 g.) in a 25 mL of EtOH/py/TEA (20:1:0.7) was added solid [t Bu4N]+[MnO4] (0.06 mmol, 0.009 g.). A fine dark powder forms within an hour. The powder was collected and washed with EtOH, then partially redissolved in a mixture DMF/CH2Cl2 to yield block crystals of ( 4 ) 3DMF2H2O by slow diffusion of ether. The yield for the powder product is 40%. ( 4 ) 3 DMF 2 H2O. Anal. Calcd. (found) for C87H91N5Mn8Fe6O37: C, 40. 61 (40. 7 0) %, H 3 .56 (3.84) %, N, 2.72 (2.53) %. Selected IR data (KBr pellet, cm1): 1683 (s), 1578 (vs), 1475 (vs), 1389 (vs), 1534 (vs), 1198 (m), 1025 (m), 920 (w), 827 (w), 800 (w), 782 (m), 634 (m), 518 (s). [Mn8O4(fdc)6(DMF)4] 1.5DMF3H2O; (5) 1.5DMF3H2O To an or ange solution of fdcH2 (0.5 mmol 0.137 g.) in 10 mL of DMF was added solid Mn(O2CMe)2 (0.4 mmol, 0.102 g.) The solution was stirred until all the solid was dissolved. Meanwhile, freshly made crystals of [Mn12O12(O2CMe)16(H2O)4] were crushed in a mortar and one equivalent (0.021 mmol, 0.043 g.) were slowly added as a solid to the stirring solution. The reaction vessel was completely screened from light with aluminium foil, and the solution left stirring for two hours at room temperature. The resulting thic k brown suspension was centrifuged and the supernatant was isolated and put to crystallize in the dark by slow diffusion of diethyl ether The product forms overnight X ray quality crystals of 5 which after few days result in a yield of ~ 85% for the


98 isolat ed product. The crystals analyze for ( 5 ).5DMFH2O. Anal. Calcd. (found) for C90H102N6Mn8Fe6O40: C, 41.01 (40.97) %, H 3.60 (3.45) %, N, 2.97 (3.06) %. Selected IR data (KBr pellet, cm1): 1682 (s), 15 76 (vs), 1480 (vs), 1389 (vs), 1536 (vs), 1201 (m), 1 035 (m), 920 (w), 834 (w), 796 (w), 762 (m), 628 (m), 520 (s). X R ay Crystallography Data were collected at 173 K on a Siemens SMART PLATFORM equipped with A CCD area detector and a graphite monochromator utilizing MoK radiation ( = 0.71073 ). Suitable single crystals of 2 8CH2Cl2, 4 4DMF4H2O and 5 4 DMF were attached to a glass fiber using silicone grease and transferred to the goniostat where they were cooled for characterization and data collection. C ell parameter s were refined using up to 8192 reflections. A full sphere of data (1850 frames) was collected using the scan method (0.3 frame width). The first 50 frames were remeasured at the end of data collection to monitor instrument and crystal stability (max im um correction on I was < 1 %). Absorption corrections by integration were applied based on measured indexed crystal faces. The structure was solved by the Direct Methods in SHELXTL6153, and refined using full matrix least squares. The nonH atoms were treated anisotropically, whereas the hydrogen atoms were calcul ated in ideal positions and riding on their respective carbon atoms. The asymmetric unit of 2 8CH2Cl2 consists of a half Fe6Mn13 cluster and four dichlorom ethane molecules. The latter were disordered and could not be modeled properly, thus program SQUEEZE154, a part of the PLATON155 package of crystallogr aphic software, was used to calculate the solvent disorder area and remove its contribution to the overall intensity data. A total of 615 parameters were refined in the final cycle of refinement using 12232 reflections with I > 2 (I) to yield R1 and wR2 o f


99 4.79% and 11.9%, respectively. Refinement was done using F2. The asymmetric unit of 4 4DMF4H2O consists of a half cluster and two DMF molecules one of which is disordered and refined in two positions and two water molecules. The protons on the coordi nated water, O14, were found and refined freely whereas the protons on the uncoordinated, O17 and O18, were found but were constrained to their respective parent atoms. A total of 670 parameters were refined in the final cycle of refinement using 32126 ref lections with I > 2 (I) to yield R1 and wR2 of 4.56% and 9.27%, respectively. Refinement was done using F2. The asymmetric unit of 5 4DMF consists of the Mn8Fe6 cluster and four DMF solvent molecules. The latter molecules were disordered and could not be modeled properly, thus program SQUEEZE154, a part of the PLATON155 package of crystallographic software, was used to calculate the solvent disorder area and remove its contribution to the overall intensity data. The cluster has one of its coordinated DMF ligand disordered where the O atoms if common to both parts while the rest were refined in two parts. Their site occupation factors were dependently refined. In the final cycle of refinement, 23801 reflections (of which 15611 are observed with I > 2(I)) were used to refine 1210 parameters and the resulting R1, wR2 and S (goodness of fit) were 4.29%, 9.47 % and 0.928, respectively. The refinement was carried out by minimizing the wR2 function using F2 rather than F values. R1 is calculated to provide a reference to the conventional R value but its function is not minimized. Magnetic M easurements The magnetic measurements w ere performed with a Quantum Design MPMS XL SQUID. The samples were colle cted in the crystalline form, briefly dried under vacuum

PAGE 100

100 a nd grinded to microcrystals The microcrystalline pr oducts w ere transferred in a gelatin capsule and embedded in eicosane to prevent torquing. The assessment of the solvation composition was done through microanalysis of the freshly prepared microcrystalli ne product. Unless otherwise stated all the measurements were carried as described later. DC susceptibility was obtained in the 5 to 300 K range with an applied field of 0.1 T. AC inphase and out of phase were performed in the 1.8 to 15 K range of temperature with an oscillating field at 50, 250 and 997 Hz and 3.5 G of magnitude. Reduced magnetization data were acquired using two modalities: low field or high field. In the low field configuration, the magnetization was measured in a set of DC field ranging from 0.1 to 1.0 T, whereas in the high field the magnitude of the applied magnetic vector ranged from 1.0 to 7.0 T. The reduced magnetization procedure and the relative data were appropriately selected to obtain the best fitting profile through the softwa re GRID141 and MAGNET140. T he magnetic data for all measurements were corrected for the diamagnetic response of the sample and the eicosane. Other M easurements IR spectra were recorded on a Nicolet Nexus 670 FTIR spectrometer. The samples were prepared as KBr pellet and analyzed in the range 4004000 cm1.

PAGE 101

101 CHAPTER 3 NEW F e7CLUSTER FROM NONINNOCENT REACTIONS O F 1,1 FERROCENEDICARBOXYLI C ACID IN POLAR MEDIA Introduction The recurring presence of 3d metals in many biological entities as well as new molecular systems capable of carrying out different functions based on their properties has promoted efforts by many research groups worldwide tow ards the synthesis and study of 3d metal assemblies. Quite prominent are the polynuclear 3d metal complexes that are enzyme active sites, and many efforts have been made to achieve metal clusters that can model the Mx sites of such biomolecules. For exampl e, t he synthesis and study of clusters can help in understanding how molecules like ferritin start with simple mononuclear iron sources and build large Fe/ O cores of up to 4500 metal ions.156 The overall process that begins with metal uptake from the environment (homeostas is) to the final core constitution is generally known as "biomineralization" and is important in the perspective of large metal storage sy stems as well as smaller cubanelike cores capable of remarkable electrochemical properties.19 Iron is one of the most abundant and dominant metal s in biological systems, but many other metal ions have been also incorporated within biomolecules and play a catalytic or electron transport role.19 Another well explored area of interest for transition metal clusters is the synthesis of materials capable of retaining their magnetization in the absence of an externally applied magnetic field. Such behavior is commonly found in the many metal, metal oxide and other magnets, but reducing their size to the nanoscale for small scale application result s in the loss of their properties unless the temperature is reduced.6 Some metal clusters can, also preserve the magnetization if kept below a blocking

PAGE 102

102 temperature, and they constitute an example of a molecular approach to magnetic nanoscale material s. The existence of such a magnetic property at the molecular level is termed as single m olecule m agnetism.35 Due to the necessity for a single m olecule m agnet (SMM) to possess a large and well isol ated spin groundstate S and a large and negative magnetic anisotropy D the metal ion of choice to fulfill both these requirements has been Mn at higher (MnIII, MnIV) oxidation states In particular the MnIII ion, with an individual spin of s = 2 and sign ificant D from the presence of JahnTeller distortion, represents an ideal starting point for achieving clusters with SMM properties. With the use of Mn ions, clusters with very large spin states up to (S = 83/2) have been isolated.157 In addition to Mn other metals have also been used for obtaining clusters with a significant net spin. For example, polynuclear iron clusters containing FeIII ions can display interesting magnetic properties. FeIII io ns, as mononuclear species in octahedral geometry, display the large spin S = 5/2 but at the same time their relative electronic isotropy (t2g 3eg 2) is the cause of a small magnetic anisotropy (D) The importance of the topological arrangement in polynuclear iron clusters i s such that despite their tendency for predominant antiferromagnetic couplings it is still possible to find examples of SMM behavior: examples include [(Fe8O2(OH)12(tacn)6)Br7(H2O)]Br (S = 10)158 and [Fe4(OMe)6(dpm)6] (S = 5)159 (where tacn = 1,4,7 triazacyclononane and Hdpm = dipivaloylmethane).160 Numerous iron clusters contain alkoxides and they are synthesized following one of two general strategies: by the use of a chelating agent carrying a short alcoholic chain or by direct attack of carboxylates in alcoholic media.160 In the first case, the chelating group may establish an initial bond to the metal center and then the close

PAGE 103

103 vicinity of one or more alcohol groups cause their coordination with often deprotonation. In the second method, the carboxylate groups exhibit a great affinity for iron ions promoting chelation assisted by an alcoholic solvent for completing the peripheral ligation or acting as bridges.42,161 163 Iron clusters are also often synthesized by reacting salts of the metal with carboxylates or by treatment of preformed cl usters (for example iron triangles) with chelates.164166 In the attempt to isolate iron clusters of a new structural type, several conditions and stoichiometries have been explored with 1,1 ferrocenedicarboxylic acid. As explained in the previous chapter, the purpose of this research was to achieve clusters with expanded redox properties. The following chapter will discuss synthetic strategies that involve different starting materials as w ell as the conditions found to lead to an unexpected behavior of the ligand precursor in the presence of light. As a result of these efforts, a new cluster of iron possessing six ferrocenyl units has been isolated and fully characterized. This iron compoun d was the first of a series of new isomorphous clusters isolated in this work Synthetic Strategies for Coordination Compounds with F dc2 In cluster chemistry, targeted reactions to achieve specific final products may be delusive due to the complexity of t he reaction system involved.79 Reactions with fdcH2 are not an exception to the above and there are only few cases in the literature where this ligand precursor has formed targeted products. A classic example of control over the final product is given by the reaction of the osmium complex [Os3(CO)10(MeCN)2] with fdcH2 in chloroform. The stronger fdc2 replaces the weaker MeCN ligands on two trinuclear units hence acting as a bridge in the final product : [(Os3H(CO)10)2(fdc)].95 Analogously, dinuclear M M bonded complexes can be subject to bridging by fdc by

PAGE 104

104 displacement of some ligands Examples come from the formation of {Mo2fdc}4 and {Ga2fdc}4 macrocyclic units starting with the {Mo2} and {Ga2} dinuclear blocks respectively.96,98 Another example of successful planning of a reaction product comes from the pairing of two molecular units by means of a fdc2. When [( O2NO)(Et3P)2Pt L Pt(Et3P)2(ONO2)], (where L=2,9phenanthrenediyl or 1,8anthracenediyl ) is reacted with fdcNa2 in a 1:1 ratio, it forms a macrocycle by displacing the nitrate with a carboxylate group and releasing sodium nitrate as byproduct.103 Despite the elegance and the control of the systems listed, other authors report similar targeted syntheses resulting in unexpected products. An example is found for the M M dimer {Re2}, where the reaction with fdcH2 did not lead to the expected dimer of dimers {Re2}fdc{Re2}, but rather to chelation by the fdc2 on a single unit to form {Re2}fdc.112 The use of fdcH2 or fdc2 as a bridging unit establishes the feasibility of a controlled oligo merization react ion where the final product contains only a few repeating units instead of an infinite chain. However this type of reaction has proven successful only in the few cases just mentioned, its application towards the preparation of smaller nuclearity metal com plexes has been relatively unsuccessful. In general, polymerization seems to result from simple reaction systems with mononuclear species, leading to the formation of insoluble precipitates and only minor quantities of molecular compounds. The problem of polymerization has been mainly dealt with in three ways in the literature of fdcH2: the use of chelating agents to limit and direct the reactivity to specific metal sites ; reactions performed by slow diffusion of the reagents and the use of solvothermal co nditions. In reactions with 3rd and 4th row transition metals, chelating agents like bpy ( 2,2' b ipyridine ) phen ( 1,10phenanthroline) and tmeda ( N N N' N' tetramethyl ethane-

PAGE 105

105 1,2diamine ) produced molecular species of the type {M2(fdc)2} where M = Co, Ni, Cu, Zn or Cd.88,111,115 The use of bidentate nonchelating ligands with particular geometries like pebbm ((1,1 (1,5 pentanediyl)bis 1H benzimidazole) or prbbm ( 1,1 (1,3 propanediyl)bis 1H benzimidazole) (Fig. 3 1) led to the formation of infinite chains (1D polymerization) or sheets (2D polymerization) with simple cadmium salts. In these cases, the coordination mode of the fdc2 and the nature of the product could be controlled by ligands capable of blocking some coordination positions while guiding the build up of the polymer.111 Figure 3 1. Structure of the pebbm ligand, 1,1 (1,5 pentanediyl)bis 1H benzimidazole, (top left) and its coordination in a chain with cadmium ions (bottom left). Structure of the prbbm ligand, 1,1 (1,3 propanediyl)bis 1H benzimidazole, (top right) and its coordination to form sheets with cadmium ions (bottom right). Adapted from ref. 111, with permission of Elsevier A similar scheme where the ligand exerts lesser control employs pyridine in a solvent with weak coordinating capabilities, such as DMF. As a monodentate ligand,

PAGE 106

106 pyridine blocks some metal positions but cannot bridge to other metal centers. A literature example demonstrates how the added coordinating effect of the solvent is sufficient in reactions with NiII or CuII salts to prevent polymerization, while in other reactions with lanthanide ions, the absence of pyridine or DMF results in the formation of 2D networks.104 The latter literature example represents an intermediate situation where the use of both co ligands and slow diffusion of the reagent led to the is olation of well defined molecular and polymeric products. Reactions with coligands are generally performed with direct mixing of the reagents, while those where there are no coligands, and the solvent is a mild coordinating agent, are generally performed using diffusion. There are two main advantages of diffusion reactions leading to polymers: firstly, the reactants will interact as if in a highly diluted solution and secondly the extended reaction time allows the products to crystallize as they form. The main disadvantage of this reaction i s the long period of completion; it may be avoided in ordinary reaction types, but they result in insoluble precipitates difficult to characterize. The literature r eports different examples of diffusion reactions where the fdc2 ligand is added to the metal ions using different approaches. A mixture of fdcH2 and ZnII ions in DMF forms the {Zn6O2(fdc)5} assembly by slow diffusion of triethylamine (TEA).107 In another case, a solution of the ligand is slowly added of the metal source and TEA solution yi elding a chain of {Eu2(fdc)3(H2O)4}n.118 A third example uses the weakl y basic anion of the metal source, Mn(O2CMe)2, to deprotonate the ligand precursor by slow diffusion and form a Mn13(fdc)6 cluster.101 Finally other methods have employed the already deprotonated fdc2 sodium salt to successfully produce sheet shaped coordination polymers of BaII

PAGE 107

107 and SmIII.104 Analogous reagents (i.e.: fdc2 and the metal ion) in the case of PbII slowly produced a crystalline product instead of a precipitate, probably due to the slow kinetics of the process.100 Reactions under solvothermal conditions represent another pathway to the isolation of well defined compounds of fdc2. The high temperatures and generally long reaction times allow for structural rearrangement of polymeric products into smaller monomeric units or discrete molecular systems. A controlled cooling time is also crucial for favoring crystallization. This method yields molecular clusters like [Zn8O4(fdc)6(H2O)3], [Cu4(fdc)2], [Sn8O4(fdc)6], [R2Sn(fdc)]2 and [R2Te(fdc)]2 (R = aliphatic or aromatic chain).105107 The synthetic methods to M fdc compounds so far discussed are not exclusive nor exhaustive. The synthesis of [Mn13O8(fdc )6(OMe)6], 2 was achieved through slow diffusion by Kondo et al. 101 while as described in C hapter 2, the same compound was isolated by mixing a preformed Mncluster in methanol with fdcH2. Besides 2 other clusters were successfu lly isolated by simple reaction of preformed clusters and fdcH2 leading to assemblies of different topology. Synthesis Initial R eactions of F dcH2 with F e S ources The reaction of fdcH2 with iron sources was found to be an extremely complicated one, and a large number of different reaction conditions were explored in developing the procedure to pure and crystalline products These involved different iron sources, solvent compositions, and reagent ratios. The use of FeII chloride in MeOH was finally found to give the reproducible product [Fe7O3(OMe)(fdc)6(MeOH)3]3+[FeCl4]Cl2, (6 ) in pure form. T he solvent, MeOH played a crucial role because ensur ed sufficient solubility of both the FeCl2 and sparingly soluble fdcH2 reagents and, importantly, to

PAGE 108

108 some extent the product. In addition, the MeOH was of course necessary to provide the MeOligation found in 6 (vide infra). After overnight stirring with the room lights on, the reaction was filtered to give a green/blue filtrate and much light grey blue residue. Layering of the filtrate wit h Et2O/hexanes gave well formed crystals of 6 (solv) not contaminated with any grey blue powder. A number of other crystallization methods were also explored but these gave a mixture of crystals and powder, and the two components were determined not to be the same compound, as the color and also IR spectra indicated. The grey blue powder was essentially ins oluble in common solvents and could not be characterized further. The large amount of solvent molecules in 6 (solv) concluded from the crystal structure is consistent with the presence of Clanions and their hydrogenbonding to solvent molecules. To get an estimate of the content, the crystals were dried very briefly under vacuum, and the elemental analysis (including Cl) was consistent with the formula 6 8H2O This formulation was used for the magnetism studies (vide infra). Many different reaction conditions have been explored in an attempt to increase the very low yield of 6 The formula of 6 has a ratio of fdc2to other Fe of 8:6, but reactions with different fdc2-:FeCl2 reagent ratios showed the optimum reaction ratio to be 2:1; higher or lower fdc2-:FeCl2 ratios were found to give mixtures of products, often containing both microcrystalline solid and fine powders, including unreacted fdcH2 star ting material for the higher ratio reactions. Much of the powder was completely insoluble in all tested solvents, suggesting polymer formation from fdc2adopting a bridging mode between different Fe7 or other Fex units. The addition of base to deprotonat e and solubilize fdcH2 and hopefully increase the yield of 6 also led to

PAGE 109

1 09 mixtures of products as judged from IR spectra and a variety of colored materials. The other products might be completely different compounds from 6 or other oxidation states of 6 o r both; however, the complexity of the reaction and the obtained mix tures of solids has prevented from obtaining other compounds in pure form for identification. Mixed solvent systems were also explored, but with no success. For example, a mixed MeOH/MeCN solvent medium did not have a significant effect on the amount or purity of the product 6 unless the MeCN was present at > 20% (v/v), at which values there was a decreased and even no amount of crystals of 6 obtained FeIII salts were also explored as start ing materials, but this approach proved unsuccessful as a route to 6 ; the main product was a gain an insoluble powder precipitate and no noticeable amount of complex 6 It was then accepted the use of FeCl2 as the starting material in reproducible reactions to pure 6 although in very low yields It was during the above experiments with FeCl2 under different conditions that it was noticed the effect of ambient light on the FeCl2 reactions. It was already realized that long reaction times were necessary for the reaction to give the dark green/ blue solution color indicative of 6 from the initial orange suspension of fdcH2. Significant darkening of the solution is apparent only after a few hours stirring and thus long reaction times (~24 hours) were routinely used. One reaction mixture was divided into two portions, one of which was kept in room light overnight and the other in the dark. The one exposed to light gave the dark colored solution characteristic of 6 whereas the other kept its original orange color, and did so for many weeks. This can be attributed to the known photolability of ferrocenyl units substituted with electronwithdrawing groups discussed in detail in the next section.

PAGE 110

110 Photosensitivity of F dcH2 Although ferrocene is known for its thermal s tability and inertness under several reaction conditions, this organometallic group, once bound to a photoactive compound, is also responsible for remarkable changes in the photoreactivity.167 Bozak et al. have demonstrated how the presence of ferrocene in carbonylated aromatic systems stops their photoreduction.168 Also, it has been shown that the ferrocenyl moiety is susceptible to photodissociation upon exposure to radiation especially in the ultraviolet and visible region. According to L. H. Ali et al.169, Nesmeyanov170 had first described the photodissociation of fdcH2 in aqueous media by prolonged exposure to light. In their work, Ali et al.170, described the effect of photoirradiation on various ferrocenyl ketones. In the case of ferrocenyl aril ketones (fc CO Ar), the presence of water as an impurity proved to be necessary for the photodecomposition. There was no decomposition in rigorously dried solvents until some water was added, and it was also observed that the reaction rate increased as a ratio of 1:1 with the substrate was achieved. On the contrary, carboxylated substrates like ferrocenecarboxylic (fmcH) and 1,1 ferrocenedicarboxylic acids were susceptible to decomposition even in dry solvents. The decomposition was monitored by IR of the dried reaction prod ucts. A common pathway was found for ketones and carboxylate derivatives showing that, besides other changes, there is a decrease in the intensity for the carbonyl vibration while a new band appeared at lower frequencies about 15401600 cm1 ( 100 cm1 in range). This band increases as the substrate is depleted and concurrently a sharp and weak CO2 vibrational band starts appearing (2323 cm1). The CO2 released with prolonged irradiation is at the expense of the newly formed peaks between 1540 and 1600 cm1 (see Fig. 32). Along with the IR studies, Ali et al.169 followe d the formation of the

PAGE 111

111 decomposed product by NMR. In case of both fmcH and fdcH2, the signals of the cyclopentadienyl protons were broadened even after a short exposure. Figure 3 2. Overlay of IR spectra for fmcH in pyridine solution with increasing ir radiation times, top to bottom. As observed for ferrocenyl ketones, the C=O stretching decreases in intensity while a neighboring band grows and a peak for CO2 appears (adapted from ref. 169). The effect became more evident on increasing the irradiation time. As the authors have suggested, the linebroadening is due to a paramagnetically induced relaxation and demonstrates the intramolecular character of the process in the signals for traces of acetone. The final products of the decomposition reactions for fmcH and fdcH2 showed cyclopentadiene, CO2 and some uncharacterized carboxylates of FeIII. More recently, Ding 171 and then Yamaguchi 172, studied in detail the electronic behavior of substitutedferrocene compounds with particular regard t o highly conjugated electronwithdrawing groups. The authors refer to previous work where the effect of benzoyl substitution was studied. Ferrocene shows two doubly degenerate ligand field electronic transitions in the visible region at 442 and 325 nm, whi le higher intensity bands are found in the UV

PAGE 112

112 region corresponding to metal to ligand and ligandto metal transitions (MLCT and LMCT) (Fig. 33). Upon benzoyl substitution, the transitions in the visible are greatly enhanced in intensity and slightly shift ed to lower energies. The authors assign this behavior as a result of the extended conjugation of the substituent with the cyclopentadienyl ring, increased delocalization of d electrons, and consequent relaxation of the Laporteforbidden nature of dd transitions. Figure 3 3. UVVisible spectra of ferrocene (Fc), benzoylferrocene (BFc) and 1,1 dibenzoylferrocene (DFc) (reported from ref. 171, Copyright 2003 American Chemical Society). The inset shows the frontier electronic states for ferrocene. The transition 2a1g 2e1g corresponds to the "band 1" in the spectra, whereas the transition 1e2g 2e1g corresponds to the "band 2" (adapted from ref. 172). The bathochromic effect of the absorption bands of benzoylferrocene and 1,1 dibenzoylferrocene with respect to fer rocene is due to increased stabilization of the LUMO (essentially based on ligands orbitals). Such an interpretation is supported by the negligible fluorescence of the ferrocene as compared to its highly delocalized derivatives. These considerations led Di ng and Yamaguchi to consider the

PAGE 113

113 photodecomposition of the substituted ferrocene due to a MLCTproduced longlived charge separation, which is particularly stable in polar solvents. Ding hypothesized that the intermediate resulting from the irradiation of 1,1 dibenzoylferrocene is a zwitterionic species, where one electron is removed from a metal orbital (mainly d in character) and delocalized over the ligand system (Fig. 34).171 Figure 3 4. Representation of the initial irradiation effect on 1,1 benzoylferrocene as a general example for other electron withdrawing conjugated systems substituted on the ferrocene assembly (adapted from ref. 171 ). The evolution of this species may then lead to attack by polar solvent molecules resulting in the expulsion of one cyclopentadienyl ligand, or there may be an initial dissociation of the cyclopentadienyl unit followed by coordination of the solvent. Although a complete characteri zation of all decomposition products has not been possible, especially due to the high sensitivity of the species initially produced, the effect of UV and visible radiation in the decomposition process of ferrocene and many conjugated derivatives is obvious. Hence, in presence of light partial decomposition with concurrent attack of solvent molecules should be expected. Further S tudy with O ther M etal S ources Something similar to what has been afore described is likely occurring in the reactions to 6 accelerating and directing the reactivity towards the formation of the cluster but the heterogeneous nature and complexity of the present reaction systems

PAGE 114

114 make it difficult to study this in more detail. However, support for the involvement of photoreleased Fe from fdcH2 under the previously described reaction conditions was obtained by alternative means. When the analogous reaction that gives 6 was explored using MCl2 (M = Mn, Co, Ni, Zn), dark colored crystals were obtained again and in low yield. These were f ound to be isomorphous with 6 as determined by IR spectral comparisons and X ray diffraction studies; in the case of M = Mn, a full crystal structure coupled with the results of a complete elemental analysis (C, H, N, Cl, Mn, Fe) revealed the formulation [Fe6.2Mn0.8O3(OMe)(fdc)6(H2O)3]Cl, (7 ), for the isolated product; the product is thus a mixture of species with different Fe:Mn ratios. In addition, the high Fe:Mn ratio is consistent with the structure of 6 being highly favored for Fe. T he iso morphous products from the reactions with M = Co, Ni, or Zn, were not analyzed in the same detail, but it seems clear that a similar situation is occurring as for the Mn product following the reaction scheme in eq. 31 to give the general species [FexM7O3(OMe)(fdc)6(MeOH)3]n+ nCl, where M = Co ( 8 ), Ni ( 9 ) or Zn ( 10) (3 1) T he reaction was also extended to larger CaII, CdII, and CeIV reagents, hoping to crystallographically identify any site preferences for different metals in a mixedmetal product structurally analogous to 6 or otherwise, but these reactions gave crystalline products that were confirmed by crystallography to be isostructural with 6 and to contain only Fe, i.e. they were complex 6

PAGE 115

115 Description of the Molecular S tructures The struct ure and labeled [Fe7O3(OMe)]14+ core of 6 are shown in Fig. 3 5 and 36, respectively ; while Tab. 31 resumes some crystallographic data. The cation consists of an [Fe7( 4O)3( 3OMe)]14+ core comprising a central cubane whose O2ions are each also attac hed to one additional Fe atom The core is very similar to the other previously reported octanuclear Fe cluster possessing a central Fe4( 4O)4 cubane and four other Fe ions bound to the oxides .152,173175 The struc ture of 6 differs from the mentioned octanuclear clusters in that one 4O in the central cubane is a 3OMe, accounting for the missing iron ion in the structure. Figure 3 5. Pov Ray projection of [ Fe7O3(OMe)(fdc)6( MeOH )3]3+: core FeIII yellow, fdc Fe ions orange, O red. Hydrogen atoms have been omitted for clarity

PAGE 116

116 Fig ure. 3 6. Pov Ray projection of the core [FeIII 7( 4O)3( 3OMe) ]14+ in 6 : FeIII yellow, O red. The axis through the atoms C73, O4 and Fe7 describes the ideal C3 rotational axis. Table 3 1. Selected crystal data for complex 6 2MeOH2H2O (full data in Tab. B 4). Parameter Value Empirical formula C 76 H 63 Cl 6 Fe 14 O 31 Formula weight g mol 1 2 466.86 Temperature K 173(2) Wavelength a 0.71073 Crystal system Monoclinic Spac e group P 2 1 /c Unit cell dimensions a = 18.4240 ( 12 ) b = 24.6645(15), c = 24.1746 ( 15 ) Unit cell angles, deg = 90 = 109.349 ( 1 ) = 90 Volume 3 10364.9 ( 11 ) Z 4 calc g cm 3 1.671 These seven Fe atoms all have distorted octahedral geometry. Peripheral ligation is provided by six fdcn groups (n = 1, 2; see below) whose six Fe atoms form a slightly distorted octahedron about the central core, and three MeOH molecules. This gives a complete cation of crystallographic C1 but virtual C3 sym metry, with the virtual rotation axis passing through Fe7 and methoxide atoms O4 and C73. As a result, the fdcngroups separate into two sets of three by virtual symmetry: One set (Fe9, Fe11, Fe13),

PAGE 117

117 those furthest from the methoxide group, bridge four Fe atoms in an 1: 1: 1: 1: 4 mode with each carboxylate bridging a separate Fe2 pair, whereas the other set bridges three Fe atoms in a 1: 1: 1: 1: 3 mode ( Fig. 3 7 ). Figure 3 7. Different chelation modes and respective torsion angles adopted by fdcn i n complex 6 In (a) 1: 1: 1: 1: 4 mode, in (b) 1: 1: 1: 1: 3 mode. The ability of the two Cp rings to twist relative to each other clearly assists in this binding flexibility of fdcn-: the 4 mode has a torsion angle of ~26.7 and thus an almost staggered FeCp2 conformation (ideal value 36), whereas the 3 mode has a torsion angle of ~6.5 and thus an essentially eclipsed conformation ( Fig 3 7 ) .134 Ligation at each external Fe atom is completed by a terminal MeOH group. The orientation of the fdcngroups, and the resulting C3 rather than C3V virtual symmetry, leads to the cation being chiral, and the crystal comprises a racemic mixture of the two enantiomers related by mirror planes (Fig 3 8 ) The seven nonfdcnFe atoms are in the FeIII oxidation state, as suggested by the metric parameters and confirmed by bond valence sum (BVS) calculations,176 which gave values in the 2.94 3.09 range (Tab. 3 2 )

PAGE 118

118 Figure 3 8. Pov Ray projection of two neighboring molecules of complex 6 in the lattice. The object s are related by an ideal mirror plane. Table 32 Bond Valence Sum calculations for complex 6 .176 Atom Fe II Fe III Fe (1) 2.88 3.08 Fe (2) 2.77 2.96 Fe (3) 2.79 2 .99 Fe (4) 2.80 2.99 Fe(6) 2.89 3.09 Fe(7) 2.75 2.94 Fe(10) 2.79 2.99 Fe(14)* 2.87 3.11 The number with an asterisk is the one closest to the actual charge for which it was calculated and the nearest whole number to it represent t he oxidation state. Relative to the counter ion [FeCl4]. Given t he [Fe7O3(OMe)(fdc)6(MeOH)3]3+ formula of the cation, this indicates that one of the ligands is in the oxidized fdcoxidation level, the remainder being fdc2-; in fact, at least one oxidized liga nd was expected from the blue color of the cation, which is not a color expected for Fe/O clusters but which is characteristic of the ferrocenium cation, Cp2Fe+.177 The BVS approach is not reliable for assigning the FeII/FeIII oxidation state within such an organometallic unit, and other distances within 6 were considered to probe this point. Oxidation of a free ferrocene unit to ferrocenium leads to a

PAGE 119

119 lengthening of the centroidcentroid distance (centroid = CT = the center of the Cp ring) by <0.1 on average,132133 which is a small change. Substituents on the rings and binding to additi onal metal centers will affect these changes. Shown in Tab. 3 3 are the FeCT and CTCT distances in 6 and it can be seen that neither parameter shows any major difference that could be assigned to clearly being in the oxidized state. Table 33 Fecent roid and centroidcentroid distances () in complex 6 Atom Fe CT a,b CT CT a,c Fe(5 ) d 1.636, 1.652 3.288 Fe(8 ) d 1.645, 1.652 3.297 Fe(12 ) d 1.656, 1.659 3.314 Fe(9 ) e 1.646, 1.640 3.287 Fe(11) e 1.649, 1.650 3.299 Fe(13) e 1.647, 1.653 3.299 Average 1.64 9 3.297 a CT = centroid. b 0.001 c 0.002 d related by virtual symmetry. e related by virtual symmetry The distances at Fe12 do show slightly longer values, but these differences are statistically borderline. It can be concluded that although ( i) the blue color of the crystal clearly indicates that 6 contains an oxidized (ferricenium containing) fdcligand, and (ii) distances at Fe12 are slightly longer than those at other ferrocenyl Fe atoms, there is nevertheless no unequivocal statistical ev idence from the structural parameters for one of the ligands being in the oxidized (fdc-) level. This is, of course, consistent with the only small changes expected on oxidation and the possibility in 6 of static disorder of the oxidized ligand amongst the three groups related by virtual symmetry, and even amongst all six of them. As in the case of the Mn clusters discussed in C hapter 2, also the lattice of compound 6 displays numerous H interactions and some interactions amongst the fdc units and in some cases between fdc and the [FeCl4] counter ion (Fig. 3 ) The distance of the ferrocenyl units supports from adjacent molecules promoted the formation of these interactions.

PAGE 120

120 Figure 3 9 Packing diagram of complex 6 in Mercury136. Blue dotted lines indicate some of the H and interactions in the lattice. All clusters 7 10 were characterized by X ray c rystallography. Despite minimal differences in the peripheral composition and counterions the isolated cationic cluster s can be described through the general formula [ FexM7 xO3(OMe)(fdc)6(MeOH)3]n+n Cl( solv ) These species are virtually isostructural (with the exclusion of the peripheral ligation) and result in complexes isomorphous to 6 The most well studied of the series is complex 7 and it will be used as the structur ally representative for the family of clusters 7 10 (Fig. 310). The main structural difference between cluster 6 and the series 7 10 is the higher symmetry of the latter for a C3 rotational axis (along C25, O11 and Mn3) that causes many structural element s to be related by symmetry. The crystallographic equivalence of many metal ions represented a problem for the conclusive identification of the oxidation states by BVS. In fact, the BVS analysis for these clusters indicated ambiguous oxidation states. In t he case of complex 7 for

PAGE 121

121 example, application of the BVS method for Mn or Fe ions led to oxidation states intermediate between +2 and +3, Tab. 34 Figure 3 10 Pov Ray projection of the cluster cation 7 (left) and its labeled core (right). In the core, all the metal ions related by symmetry have the same color. Table 34 Bond Valence Sum calculations for complex 7 according to the Mn and Fe parameters Atom Mn II Mn III Mn IV Fe II Fe III Mn 1 3.22 2.97 2.91 2.77 2.96 Mn 2 2.66 2.45 2.40 2.28 2.44 Mn 3 3 .05 2.81 2.76 2.61 2.80 Due to t he uncertainty in the determination of 7 10 metal composition by X ray analysis a Cl micro analysis was obtained with the purpose of achieving information on the total charge. The data allowed for the complete formulation of the counter ions, but leaving unclear the exact metal composition (Tab. 35). As mentioned previously in the chapter, only 7 was fully characterized with the further microanalysis for Mn and Fe. The crystalline product for the reactions with the larger ions CdII, HgII, CeIII and CaII showed the same structural topology as 7 However, in this case all metal ions could be

PAGE 122

122 clearly identified as iron ions due to the relatively small electronic density compared to Cd, Hg, Ce and Ca. Additional crystallogr aphic data for 6 are listed in appendix C Table 35 C ounterion composition for 7 10 obtained by Cl microanalysis Compound Composition 7 [Fe 6 .2 Mn 0.8 O 3 (OMe)(fdc) 6 (H 2 O) 3 ] + Cl 8 [ Fe x Co 7 x O 3 (OMe)(fdc) 6 (MeOH) 3 ] 2 + 2 Cl solv 9 [ Fe x Ni 7 x O 3 (OMe)(fdc) 6 (MeOH) 3 ] + Cl solv 10 [ Fe x Zn 7 x O 3 (OMe)(fdc) 6 (MeOH) 3 ] + [Zn Cl 4 ] 2 solv The complete composition for 8 10 has not been established. Magnetochemistry of the H eptanuclear C lusters [M7O3(OMe)(fdc )6(MeOH)3]n+ Variable temperature magnetic susceptibility ( M) data were acquired in the range 5.0 to 300 K with a steady magnetic field of 1kOe (0.1 T) for the clusters 6 10 The samples were prepared as described in the experimental section and analyzed for their MT vs. T profile. Figure 311 reports the behavior of cluster 7 which shows a steady decrease of the MT value from 16.39 cm3Kmol1 at 300 K to 7.56 cm3Kmol1 at 40 K. Past this temperature the MT is essentially constant. The magnetic data for 6 include the contribution of the paramagnetic high spin anion [FeCl4], which has S = 5/2 and a MT of ~4.38 cm3Kmol1 (for g = 2). In an attempt to remove the contribution of this paramagnetic entity, a sample of [NEt4][FeCl4] was prepared according to a literature procedure178 and measured in the same conditions. With the assumption of an independent magnetic behavior of the cationic cluster and its counterion, the MT for [FeCl4] was subtracted from the MT of 6 The assumption of minimal exchange interactions between the cluster and its ferric cation is supported by observation of only weak ClH C contacts to cyclopentadienyl units in the lattice. The net susceptibility for the cluster decreases steadily from 11.62 cm3Kmol1 at 300 K to 3.17 cm3Kmol1 at 40 K and raises again

PAGE 123

123 after this point to 4.37 cm3Kmol1 at 1.8K. The latter behavior is most probably an artifact originated from the MT subtraction. It is evident from the MT plot that the [FeCl4] units have some intermolecular interactions below 40 K that cause an apparent decrease in the measured MT T / K 0 50 100 150 200 250 300 MT / cm3Kmol-1 2 4 6 8 10 12 14 16 18 Complex 6 Cluster-only [NEt 4 ][FeCl 4 ] Figure 3 11 Direct current measurement of MT vs. T for 6 8H2O6MeOH. The plot also displays the behavior of only the cationic cluster by subtraction of the [FeCl4] MT. It is then appropriate to consider the MT value at 40K as the low temperature plateau for the cation of 6 8H2O6MeOH (which differs by 4.39 cm3Kmol1 from that of the anioncation pair, 4.3 cm3K mol1 being the expected value for the anion). The data just obtained are indicative of an S = 2 ground state with g 2.05. The MT profile indicates strong antiferromagnetic interactions within the FeIII core ions and the ground state can be explained by the antiferromagnetic coupling of a net S = 5/2 spin on core with a S = 1/2 of one oxidized fdc ligand. The value of g slightly higher than 2, is common for high spin FeIII systems and it is also in agreement with the g > 2 generally

PAGE 124

124 found in ferricenyl gro ups.179181 The low symmetry of the cluster cation and the relatively large number of possible pairwise exchange parameters ( Jij) hinder the application of the matrix diagonalization for the fitting of the MT vs. T data as well as the Kambe vector coupling method.32 The ground state was also evaluated by alternating current measurements in the 1.815K range with a 3.5 Oe oscillating field at 50, 250 and 997 Hz. This determination has the advantage of avoiding co mplications from an applied steady field. In the case of an absent out of phase component of the magnetization ( M ) the in phase signal ( M T) corresponds to the DC MT allowing for its determination in the absence of an external DC field. Figure 312 shows the M T vs. T plot of 6 as well as the M confirmed to be negligible throughout the observed temperature range. The M T vs. T plot appears almost T independent with a value of ~ 8 cm3Kmol1 down to ~5K, below which it decreases slightly t o ~7.3 cm3Kmol1. The effect of decreasing M T may result from a combination of weak inter ionic exchange interactions and zerofield splitting. The net value for M T is obtained again by subtraction of the M T for [FeCl4] (4.38 cm3Kmol1) to that for 6 and is ~ 3.6 cm3Kmol1. This value is slightly higher than the one found for the corresponding measurement in the dc field (3.1 cm3Kmol1). This could be due to experimental errors and some degree of magnetic interaction between the cationic cluster and i ts counterion. Such discrepancy, although significant, does not affect the correct identification of the spin groundstate because of the substantial difference from other possible values of S = 1 and S = 3 with expected MT value of 1.0 cm3Kmol1 and 6.0 cm3Kmol1, respectively Hence the assignment of S = 2 with g > 2 for 6 is confirmed.

PAGE 125

125 M'T vs. T T / K 0 2 4 6 8 10 12 14 16 M'T / cm3Kmol-1 6 7 8 9 10 11 997 Hz 250 Hz 50 Hz M" vs. TT / K 0 2 4 6 8 10 12 14 16 M" / cm3Kmol-1 -2 -1 0 1 2 Figure 3 12 Magnetic susceptibility plot for 6 with AC fields The inset shows the out of phase component ( M ) while the main plot represents the inphase component of M T vs. T. T he software M AGNET140 failed in producing a good fit for the reduced magnetization data collected for complex 6 Possible reasons are lowtemperature inter ionic interactions, low spin values, and presence of a ferricenyl unit for its g substantially higher then 2. The magnetic studies on the remaining isomorphous clusters were performed following the same type of approach for 6 Figure 313 show s the behavior of the heptanuclear family of clusters in a variable temperature DC field measurement where a strong component of the antiferromagnetic interactions throughout the series can be observed the steady and steep decrease of the inphase magnetic susceptibility. The data follow approximately the same profile for the decreasing of MT vs. T, from an initial MT value at of ~ 12 cm3Kmol1 300 K down to ~ 2.4 cm3Kmol1 at 5 K. The only exception is for compound 10, for which the MT appears shifted to a lower value

PAGE 126

126 throughout the temperature range by ~ 1.4 cm3Kmol1, hence beginning at ~ 10 cm3Kmol1 at 300 K and ending at ~ 1 cm3Kmol1 at 5 K. T / K 0 50 100 150 200 250 300 MT / cm3Kmol-1 0 2 4 6 8 10 12 14 7, (Mn) 8 (Co) 9 (Ni) 10, (Zn) Figure 3 13 Direct current measurement of MT vs. T for 7 5H2OMeOH, 8 solv., 9 solv. a nd 10solv. The uncertainty in the determinati on of the molecular mass for compounds 8 10 may be cause of a slight vertical shift of the data leaving the overall profile unchanged. However, due to the relatively small variation in mass by passing from an iron for example to a manganese one, the measurement should still allow for qualitative observations. While compounds 7 9 converge to approximately the same low temperature value for the MT of ~ 2.2 cm3Kmol1 at 0 K 10 converge to a MT v alue of ~ 1 at the same temperature, hence suggesting for the latter a spin groundstate S = 1. Alternating current measurements for the detection of the inphase and out of phase signal s confirmed for the clusters 7 10 a similar behavior in the l ow temperature region (Fig. 314). The plot of M T vs. T shows an almost linear decay up to ~ 2.4 cm3Kmol 1 at 1.8 K for 7 9 whereas the lowest temperature value for 10 is ~ 1.1 cm3Kmol 1.

PAGE 127

127 M'T, (in-phase) T / K 0 2 4 6 8 10 12 14 16 M'T / cm3Kmol-1 0 1 2 3 4 5 7 Mn 8 Co 9 Ni 10, Zn M", (out-of-phase) T / K 2 4 6 8 10 12 -1.0 -0.5 0.0 0.5 1.0 Figure 3 14 AC (in phase) M T vs. T and (out of phase) M vs. T (inset) pl ots for 7 10 at 250Hz. The AC in phase confirms compound 10 to posses a spin groundstate S = 1, while showing still overlapping data in the low temperature region for 7 9 The uncertainties in composition and oxidation states of the core do not allow for a unique prediction of the possible spin value. Extrapolation to 0 K of the MT vs. T leads to ~ 2.35 cm3Kmol 1 suggesting two close values for the spin groundstate : S = 2 with g = 1.8 ( MT = 2.43 cm3Kmol 1) and S = 3/2 with g = 2.2 ( MT = 2.27 cm3Kmol 1). In this case, the large presence of iron in the core supports the S = 3/2 because of the g > 2 typical of high spin FeIII. M ssbauer S pectroscopy of the [M7O3(OMe)(fdc)6(MeOH)3]n+ C lusters M ssbauer studies have been employed as a reliable source for oxidation states of iron centers as well as their chemical environment The 57Fe M ssbauer spectra is sensitiv e to the number and different types of iron atoms based on the oxidation state and chemical environment. The spectrum of compound 6 is a complex case to analyze

PAGE 128

128 due to the elaborate structure and variety of iron sites related by virtual symmetry and differ ence in oxidation state. The main features that are expected for [Fe7O3(OMe)(fdc)6(MeOH)3]3+[FeCl4]2Cl are: Five FeII ferrocenyl sites within the fdc2 ligands. These give rise to characteristic Q > 2.0 mm/s .182 One FeIII ferricenyl site within the fdcligand Q is relatively small.182 Q valence complexes in a trappedQ moiety .183 Seven high spin FeIII sites of the octahedral FeO6 type. The isomer shift is expected to be 0.45 Q ( 0.6 mm/s) .182 One FeIII site within the [ FeCl4]anion. The isomer shift of this four coordinated ferri c site is smaller than that Q is negligible .184 According to this description the Fe sites that would be most readily resolved are the five ferrocenyl groups due to their large EQ as compared to the ferricenyl group. On the other hand, the [FeCl4] counter ion as well as the fdc1 will be hard to detect due to their overall small contribution ( ~ 7% each). The M ssbauer spectrum of 6 was collected at 140 and 250 K and consists of tw o quadrupole doublets (Fig. 315). The spectrum at 250 K displays well resolved peaks of an outer doublet with lines at 0.62 and +1.56 mm/s each having small linewidth (FWHM 0.25 mm/s). The high resolution of these peaks allows for their parameters and total contribution to be estimated with high accuracy. The isomer shift ( = 0.48 0.02 mm/s) and quadrupole splitting ( EQ = 2.16 0.02 mm/s) are in agreement with the values expected for a ferrocene species, hence the outer doublet signal is assigned to the fdc2 units. The small line width of these peaks suggests a site with a narrow distribution of energies, which despite the different orientations is predicted by theoret ical calculations.185 The remaining portion of

PAGE 129

129 the spectrum is occupied by a relatively broad and asymmetric doublet whose peaks are located at +0.20 mm/s and +0.65 m/s with an average EQ of ~ 0.40 mm/s and an aver age of ~ 0.45 mm/s. This portion of the spectrum is comprising all the signals from the seven FeO6 sites, the [FeCl4] anion and the ferricenyl moieties. The data in this case cannot be resolved to individual contributing signals that would allow for an unambiguous characterization of the FeIII species. Fig ure. 3 15. Overlay of the M ssbauer spectra collected at 250 and 140 K for sample 6 The outer doublet assigned to the fdc2 units, accounts for only 22% of the total iron signal at 250 K. This value is much lower than the expected ~ 36% for five fdc2

PAGE 130

130 ligands on each molecule. In fact, this value is predicted for only three fdc2 per molecule (theoretically ~ 21% of total iron), and contrasts with other analytical data. It is believed that different DebyeWaller factors are associated with different iron sites. The literature reports examples of molecules featuring a biferrocenium where the latter was reported having a smaller DebyeWaller factor than other iron sites.186 According to this observation it is believed that in 6 due to the smaller DebyeWaller factor, the contribution of the fdc2 units is underestimated as compared to the other sites. To prove this assumption, a M ssbauer spectrum at lower temper ature (i.e., 140 K) was collected and checked for an increased contribution from the ferrocenyl sites (Fig. 315). The obtained spectrum is very similar to the one at 250 K, however as proposed, the contribution of the fdc2 sites has increased to 31 %. The different dependence of the DebyeWaller factor with the temperature for fdc2 and the other FeIII sites, implies that this parameter increased faster for the fdc2 units than it did for other sites in the same temperature range. Further lowering of the temperature up to 4.2 K increased the contribution of the ferrocenyl units to 35 %, consistent with the theoretical value for five fdc2 groups. The M ssbauer spectra of the compounds 7 10 display a singular sim ilitude with each other. For reasons already explained, it is not possible to distinguish the contributions from different sites that result in the central doublet. However, the spectra of 7 10 support a relatively large presence of Fe in the core with a slight diminished contribution from ferric ions to the central doublet when compared to 6 Furthermore, the similarity of the spectra in Fig. 316 suggests a very close fdc/fdc2 ratio for the four cluster cations 7 10.

PAGE 131

131 -6 -4 -2 0 2 4 6 90 92 94 96 98 100 -6 -4 -2 0 2 4 6 92 94 96 98 100 -6 -4 -2 0 2 4 6 92 94 96 98 100 -6 -4 -2 0 2 4 6 92 94 96 98 100 X Axis Title 78 K Zn Ni Mn Co v (mm/s) Figure 3 16 Superposition of M ssbauer spectra for 7 10 at 78 K. Electroche mistry of the [M7O3(OMe)(fdc)6(MeOH)3]n+ C lusters The large variety of metal centers and ferrocenyl units is of great interest for electrochemical investigations. For this purpose, a freshly prepared crystalline sample of 6 was dried under vacuum and used to obtain a solution in dry acetonitrile and used in a standard threeelectrode electrochemical cell. The concentration of the analyte reached saturation before the intended concentration of 1 mM due to its limited solubility in acetonitrile. Figure 317 shows the cyclic voltammogram (CV) at 100 mV/s (top) and the differential puls e voltammogram (DPV) at 20 mV/s ( bottom )

PAGE 132

132 Figure 3 17 Cyclic voltammogram (CV, top) and differential pulse voltammogram (DPV, bottom) of 6 in acetonitrile (n Bu4NPF6 0.1 M as supporting electrolyte) As it is clear from the plots, cluster 6 displays a rich electrochemical behavior corresponding to four reversible reductions and a single envelope for multiple oxidations. The second and third reduction peaks overlap, however bot h seem to involve single electron processes and in particular the one localized at 242 mV takes place on the counterion [FeCl4]. The latter was identified by comparison with the redox

PAGE 133

133 behavior of [NEt4][FeCl4] in the same conditions of solvent and electr olyte. While the three reduction processes on the cluster cation are considered to happen at the core ions, the large oxidation peak is attributed to the almost simultaneous oxidation of the fdc2 groups. Upon starting the measurement at the rest potential (+195 mV vs. fc/fc+) there is a spike in the current which can be assigned to the reduction of one oxidized fdc ligand. Thus the large oxidation should be consistent with the overlapping of six singleelectron oxidations of the fdc2.The fdc2 groups behave near independently and hence are being oxidized essentially at the same potential. Confirmation of this assignment is obtained through the measurement of the peak heights showing a ratio of 10:1 (rather than a 6:1) for the oxidation vs. reduction features. This mismatch can be caused by different electron transfer kinetics from the outer fdc2 redox sites with respect to the inner ones. A difference can also be appreciated by comparing the reduction of the counterion [FeCl4] with a sharper and higher peak than the other reduction peaks. DPV plots provide better resolution of the redox processes. In this case the measurement is performed against a silver wire reference electrode, with the redox couple fc/fc+ at +180 mV. The large oxidation peak exhibits shoulders which confirm its origin from overlapping separate features, consistent with the two sets of fdc groups by symmetry and a minor degree of electronic communication among the groups. Considering all the redox processes localized on the cation of cl uster 6 it is possible to access ten oxidation levels ranging from the fully oxidized form [FeIII 7(fdc)6]15+, through the isolated [FeIII 7(fdc)(fdc2)5]10+, to the [FeIII 4FeII 3(fdc2)6]6+ reduced level. Application of the same technique to clusters 7 10 produced very similar results with minor changes

PAGE 134

134 in the redox potentials and the absence of reduction peak s due to redox active counterions. Co nclusions The reactions of fdcH2 with FeII or FeIII sources have proved to be very challenging due to the formation of amorphous precipitates that are essentially insoluble in all inert solvents. This is probably due to the formation of polymeric products, a situation that is commonly encountered in the chemistry of fdc2-. However, it should also be noted that neutral molecular compounds of fdc2are also often very insoluble, such as the Mn13 and Mn8 clusters described in C hapter 2 and that has unfortunately precluded their study in solution. Nevertheless, the slow reaction of fdcH2 with FeCl2 also gives a more soluble fraction that was crystallized and characterized as 6 8H2O 6H2O The solubility is undoubtedly assisted by its cationic nature. The photore lease of Fe from fdcH2 was shown to occur and give 6 or mixedmetal analogs when other metal reagents were employed (for example 7 10) ; in fact, there are literature studies attempting to characterize the photodecomposition products of fdcH2,169,187188 and under the conditions of this study at least they drive the formation of the cation of 6 The unusual nature of 6 was immediately evident from its blue color hence it can be concluded from the various data that i t contains one oxidized fdcligand. This is the first example of this ligand in inorganic chemistry. There is one other claim of a fdcligand, in the compounds [Zn6O2( fdc )5(H2O)(DMF)] and [Zn8O4( fdc )6(H2O)3] made by hydrothermal reaction of fdcH2 with Zn (NO3)2;107 however, the reported magnetic data and the dark red color are not consistent with fdcand it could be possible that some of the Zn sites are occupied by FeIII released from fdc2under the high energy reaction conditions. The electrochemical studies indicate a new class of compound that has rich

PAGE 135

135 redox behavior, and that offers possibility of being isolated and characterized at different oxidation levels. However, considering the low yield of 6 this requires a route to greater amounts of cluster than are currently available. Nevertheless, the incorporation of multiple fdcnligands into magnetic metal clusters has the potential for interesting new magnetic and redox behavior Experimental All manipulations were performed under aerobic conditions, using materials as received, except otherwise stated. All experiments were initially carried in the presence of light and then repeated later in the dark during both reaction and crystallization periods. This procedure was applied to ensure the role of light in reactions with the ferrocenyl ligand. [Fe7O3(OMe)(fdc)6(MeOH)3]3+ [FeCl4]2Cl8H2OMeOH; (6 )8H2OMeOH A suspension of fdcH2 (0.25 mmol, 0.067 g) and FeCl24H2O (0.5 mmol, 0.099 g) in MeOH (25 mL) was allowed to stir under normal laboratory light conditions for 24 h. The initial orange color changed to dark after a few hours. The mixture was filtered on a P2 filter paper from which a dark brown solution was obtained and set for crystallization by slow diffusion with a ether/hexanes (1:1) mixture. Dark blue hexagonal shaped crystals appeared after one week on the walls of the vials. The crystallization was accompanied by the formation of a blue precipitate, which could not be characterized. The crystals were dried under vacuum. Yield: 2 to 5%. The crystals analyze for ( 6 )H2O 6MeOH. Anal. Calcd. (found) for C82H103Cl6Fe14O45: C 35.13 (35.05) %; H 3.70 (3.49) %; Cl 7.59 (7.65) %. Selected IR data (KBr pellet, cm1): 1575 (vs), 1481 (vs), 1400 (vs), 1362 (s), 1195 (w), 1115 (w), 1082 (w), 1031 (vw), 928 (vw), 832 (vw), 781 (w), 621 (w), 589 (w), 522 (m), 491(m).

PAGE 136

136 [Fe6.2Mn0.8O3(OMe)(fdc)6(H2O )3]+ Cl 5 H2O MeOH; (7 )5 H2O MeOH The procedure is identical to that for 6 with the use of MnCl24H2O as metal source in the same stoichiometric amount. The isolation procedure and yield ar e comparable to that for 6 The crystals analyze for ( 7 )H2O4MeOH. Anal. Calcd. (found) for C77H83ClMn0.8Fe12.2O40: C, 38.39 (38.28) % ; H, 3.47 (3.61) % ; Cl 1.47 (1.65) %; Mn 1.82 (1.58) %; Fe 28.28 (28.70) % Selected IR data (KBr pellet, cm1): 1577 (v s), 1486 (vs), 1395 (vs), 1361 (s), 1197 (w), 1119 (w), 1088 (w), 1029 (vw), 925 (vw), 831 (vw), 777 (w), 627 (w), 585 (w), 524 (m), 493(m). [FexCo7xO3(OMe)(fdc)6(MeOH)3]2+ 2Clsolv.; (8)solv. The procedure is identical to that for 6 with the use of CoCl24H2O as metal source in the same stoichiometric amount. The isolation procedure and estimated yield are comparable to that for 6. Selected IR data (KBr pellet, cm1): 1578 (vs), 1486 (vs), 1395 (vs), 1361 (vs), 1197 (m), 1029 (w), 925 (vw), 831 (w), 778 (m), 630 (w), 586 (w), 523 (s), 492 (w), 451 (w) [FexNi7xO3(OMe)(fdc)6(MeOH)3]+ Clsolv. ; ( 9 )solv. The procedure is identical to that for 6 with the use of NiCl24H2O as metal source in the same stoichiometric amount. The isolation procedure and est imated yield are comparable to that for 6. Selected IR data (KBr pellet, cm1): 1577 (vs), 1485 (vs), 1397 (vs), 1359 (s), 1196 (w), 1118 (w), 1085 (w), 1030 (vw), 924 (vw), 833 (vw), 774 (w), 624 (w), 586 (w), 524 (m), 492(m) [FexZn7xO3(OMe)(fdc)6(MeOH )3]+ [ZnCl4]2solv.; (10)solv. The procedure is identical to that for 6 with the use of ZnCl2 anhydrous as metal source in the same stoichiometric amount. The isolation procedure and estimated yield are comparable to that for 6. Selected IR data (KBr pellet, cm1): 1577 (vs), 1484 (vs),

PAGE 137

137 1393 (vs), 1362 (s), 1196 (w), 1121 (w), 1089 (w), 1028 (vw), 925 (vw), 832 (vw), 775 (w), 626 (w), 583 (w), 521 (m), 491(m). X R ay Crystallography Data were collected at 173 K on a Siemens SMART PLATFORM equipped with a CCD area detector and a graphite monochromator utilizing MoK radiation ( = 0.71073 ). Suitable single crystals of 6 MeOHEt2O, 7 DMFMeOH, 8 MeOH 9 MeOH and 10 2MeOHEt2O were attached to a fiber glass using silicone grease and transferred to the goniostat where they were cooled to the temperature indicated above for characterization and data collection. All cell parameters were refined using up to 8192 reflections. A full sphere of data (1850 frames) was collected using the scan method (0. 3 frame width). The first 50 frames were remeasured at the end of data collection to monitor instrument and crystal stability (maximum correction on I was < 1 %). Absorption corrections by integration were applied based on measured indexed crystal face s. The structure was solved by the Direct Methods in SHELXTL6153, and refined using full matrix least squares. The nonH atoms were treated anisotropically, whereas the hydrogen atoms were calculated in ideal positi ons and were riding on their respective carbon atoms. The asymmetric unit of 6 consists of a Fe13 cluster cation, and FeCl4 anion, two methanol and two ether solvent molecules. All the solvent molecules were disordered and could not be modeled properly, t hus program SQUEEZE154, a part of the PLATON155 package of crystallographic software, was used to calculate the solvent disorder area and remove its contribution to the overall intensity data. Three of the four chloro ligand of the FeCl4 anion in 6 are rotationally disordered

PAGE 138

138 and were refined in two parts with their site occupation factors refining to 0.577(2) and 0.423(2) for the major and minor parts, respectively. A total of 1119 parameters were refined for 6 in the final cycle of refinement using 8375 reflections with I > 2 (I) to yield R1 and wR2 of 4.06% and 8.31%, respectively. Refinement was done using F2. The asymmetric unit of 7 consists of a 1/3 of a Mn7 cluster (located on a 3fold rotation axis), a dimethylformamide and an methanol molecule. The protons on the coor dinated water molecules could not be located in a Difference Fourier map and were not included in the final cycle of refinement. A total of 352 parameters were refined in the final cycle of refinement using 5265 reflections with I > 2 (I) to yield R1 and wR2 of 3.59% and 9.70%, respectively. The asymmetric unit of 8 consists of a third of Fe6Co7 cluster and three methanol molecules of crystallization. A total of 366 parameters were refined in the final cycle of refinement using 14255 reflections with I > 2 (I) to yield R1 and wR2 of 2.53% and 6.27%, respectively. The asymmetric unit of 9 consists of a third of Fe6Ni7 cluster and four methanol molecules of crystallization. A total of 366 parameters were refined in the final cycle of refinement using 13661 re flections with I > 2 (I) to yield R1 and wR2 of 2.74% and 6.50%, respectively. The asymmetric unit of 10 consists of a 1/3 Fe6Zn7 cluster cation (clusters are located on 3fold rotation axes), a 1/6 ZnCl4 dianaion (anions located on 3 rotation axes), thus the ratio is 2 Fe6Zn7 cation to 1 ZnCl4 anion. In the asymmetric unit there is also an ether molecule and two methanol molecules. A total of 381 parameters were refined in the final cycle of refinement using 4282 reflections with I > 2 (I) to yield R1 and wR2 of 7.56% and 19.85%, respectively.

PAGE 139

139 Magnetic Measurements The magnetic measurements were performed with a Quantum Design MPMS XL SQUID. The samples were collected in the crystalline form and briefly dried under vacuum. The microcrystalline products were transferred in a gelatin capsule and embedded in eicosane to prevent torquing. The assessment of the solvation composition was done through microanalysis of the freshly prepared microcrystalline product. Unless otherwise stated all the measurements were carried as described. DC susceptibility was obtained in the 5 to 300 K range with an applied field of 0.1 T. AC inphase and out of phase were performed in the 1.8 to 15 K range of temperature with an oscillating field at 50, 250 and 997 Hz and 3.5 G of m agnitude. Reduced magnetization data were acquired using low field or high field. In the low field configuration, the magnetization was measured in a set of DC field ranging from 0.1 to 1.0 T, whereas in the high field the magnitude of the applied magnetic vector ranged from 1.0 to 7.0 T. The reduced magnetization procedure and the relative data were appropriately selected to obtain the best fitting profile through the software GRID141 and MAGNET140. The magnetic data for all measurements were corrected for the diamagnetic response of the sample and the eicosane. Other Measurements IR spectra were recorded on a Nicolet Nexus 670 FTIR spectrometer. The samples were prepared as KBr pellet and analyzed in the range 4004000 cm1. Elemental analyses (C, H and N) were performed by the inhouse facilities of the University of Florida, Chemistry Department. Chlorine and metal analyses were obtained fr om Desert Analytics ( Tucson, Arizona ). Cyclic voltammetry (CV) and differential pulse voltammetry (DPV) were performed with a BASi CV50W

PAGE 140

140 electrochemistry instrument and a threeelectrode cell, using dry MeCN a s solvent and NBun 4PF6 (0.1 M) as supporting el ectrolyte. A glassy carbon working electrode (BASi model MF2012) and a coiled platinum wire as auxiliary electrode were employed, and the reference electrode was Ag wire; under the same conditions, ferrocene was at 0.18 V. The MeCN was distilled over CaH2 and stored over activated molecular sieves. NBun 4PF6 was recrystallized in the dark from a H2O/EtOH ( 1:1 v/v) and dried under vacuum 57Fe Mssbauer spectra were obtained at NCSR "Demokritos" and were recorded in the constant acceleration mode at temperat ures controlled with a Janis cryostat. Isomer shifts are reported relative to iron metal at room temperature set at 0 mm/s

PAGE 141

141 CHAPTER 4 NEW APPROACH FOR THE IMPROVEMENT OF THE BOND VALENCE SUM PARAMETERS USING STATISTICAL METHODS Introduction Since the wor k done by Brown and Altermatt on an alternative and more general formulation for the Bond Valence Sum (BVS) method, its application to chemistry and closely related disciplines has grown rapidly.71 For example, r esearch on super conductivity has adopted BVS to monitor the valence of Cu ions in metal oxide layers resulting from doping reactions.189 In these cases the doping directly affect s the temperature at which the material act s as a superconductor, thereby making the BVS m ethod of significant importance in this field. In other more exotic studies, BVS was paired with theoretical models to predict acidic strengths (or pKa) ,190 probe the presence and strength of H bonding,191 or localize preferential regions for proton solvation in a crystalline lattice.192 Nevertheless, of the several uses found for the BVS m ethod, two areas have been most dominant in its application: metalloenzymes and solid state materials. In the first case, biological molecules often host metallic ions at the site of enzymatic activity, and the determination of the metals' oxidation state represents a basic requirement for fully understanding the chemical reactivity displayed by this type of system In this area, Thorp began to explore the use of BVS to interpret EXAFS data on metalloproteins.193 His studies confirmed the applicability of the method for biological metal sites al beit with some approximations. The other major applications of BVS span areas ranging from geol ogy to cluster chemistry. Application to polymeric structures, l ike in mineral chemistry seems to have a greater accuracy than the nonpolymeric molecular systems. Despite some discrepancies found in this last application, BVS is widely used for the deter mination of the oxidation states in mononuclear or polynuclear

PAGE 142

142 systems. Examples have been shown in the previous chapters, where the method was used to predict the metal oxidation states in complicated polynuclear systems based on crystallographic data. As mentioned in the introductory chapter, the BVS model uses bond lengths between an i on and its first sphere of coordinated atoms to achieve an estimation of the oxidation state. The method assigns to each bond a contribution to the total valence through the existing correlation between bond length and bond strength as described by the formula (15), repeated here for convenience as (41) where r0 and B are empirical parameters and r is the bond length in B r re s0 (4 1) By sum ming of all th e s values of a coordination sphere, the valence is obtained and it represents an approximation of the formal oxidation state. The empirical values, r0 and B have been calculated and reported for many metal ligand combinations and different oxidation stat es B is taken as constant with a value of 0.37 .71 For instance, by applying the BVS to a metal ion in the +2 oxidation state with r0 values appropriate for that oxidation state, the sum of all its bond valences ( each obtained through Eq. 4 1) will lead to a value close to the integer two The same calculation applied with r0 parameter s for another oxidation state will tend to the integer two but with a larger discrepancy. The initial calculations for r0 and B made by Brown and Altermatt were based on crystallographic data obtained on inorganic ionic crystalline lattices with homoleptic metal ions I n this context, the term homoleptic would refer only t o the first coordination sphere.71 The relatively limited number of high quality crystal structures and the choice

PAGE 143

143 of metal ions with the same environment has led to a set of parameter s that fit accurately for the same type of structures, but they do not work as well when appli ed to molecular systems. S hortly after the introduction of BVS application to metalloenzymes, Thorp and coworkers realized that the r0 parameters could be improved for biomolecules.194 The extensive application of the BVS approach to mononuclear and polynuclear systems has suggested an analogous procedure for improvement by modifying the r0 parameter I t is now possible to find reports that describe efforts aiming to wards more tailored values of r0 (and B in some ins tances).195197 In most cases, application of the BVS to metallic clusters results in negligible dis crepancies from integer values but in some other situations the method fails completely The cluster [ Co8O4(O2CPh)12(H2O)(MeCN)3] for example, has a Cooxo core crystallographically described by eight different cobalt ions. From neutrality considerations the core is described as [CoIII 4CoII 4O4]12+ (Fig. 41) .151 Figure 4 1 Pov Ray projection of [ Co8O4(O2CPh)12(H2O)(MeCN)3] (left) and its core (right). The numbers identify the crystallographic label of the metal ions.151 1 2 3 4 5 6 7 8

PAGE 144

144 The literature reports the following data for Co BVS: r0 (CoIIO) = 1.70 r0 (CoIIIO) = 1.692 and r0 (Co N) = 1.84 (with the assumption of B = 0.37 ).131 Table 41 shows the obtained valence values for these metal ions. The atoms Co1Co4 are CoIII species, wher eas the remaining Co5Co8 are CoII. The BVS indicates a valence close to the expected value only for Co5, Co7 and Co8. All other atoms have significantly higher values than the excepted, making the BVS approach of little use for this cluster. Table 41 E xpected and calculated Co valences in [ Co8O4(O2CPh)12(H2O)(MeCN)3] based on published data.131,151 Atom Actual oxidation state BVS value as Co II BVS value as Co II I 1 +3 3.32 3.39 2 +3 3.39 3.47 3 +3 3.39 3.47 4 +3 3.39 3.46 5 +2 1.97 2.01 6 +2 2.35 2.39 7 +2 2.18 2.21 8 +2 2.15 2.19 The numbers with an asterisk represent the only values close to the actual oxidation state Although it could be hypothesized that the presence of mix edvalen cy is leadin g to intermediate average oxidation states, the authors exclude this possibility and in any case the ligation would not support the total charge for the core ( suggested by BVS. Despite the two different values for the r0 in the case of Co coordination to oxygen and only one for the coordination to nitrogen, the data show better accuracy for Co ions bonded to nitrogen. It is likely that the proposed r0 parameters are better suited for a polymeric type of Co environment rather than a molecular system. T o test this hypothesis the calculation was reversed and the corresponding r0 values were obtained for the expected Co oxidation states of Co8. A lso it was believed that this could provide numbers more suitable for applying the BVS method for other similar molecular Co

PAGE 145

145 system s. The reversal of the BVS formula leading to r0 for a metal ion implies the solution of a system of equations as shown i n Eq. 4 2. z j i z z j j i iv v v V v b r R v b r R v b r R ) ln( ) ln( ) ln(0 0 0 (4 2) In Eq. 4 2, V is the final valence of a metal ion, ri and vi are the bond distance the partial valence for the ith ligand atom respectively, and b = B (0.37 ). The new r0 (or R0) for CoIII has been calculated as the average of the four different r0 values (obtained by use of Eq. 4 2) each belonging to one of the four different CoIII ions in the Co8 cluster and is equal to 1.644 (compared with the current 1.692 ) As expected, application of the new r0 on the same metal ions for which it was obtained leads to values very close to their actual valence (Co1 = 3.132, Co2 = 2.978, Co3 = 2.970 and Co4 = 2.917). Since the same metal ions in a similar coordination environment will produce similar r0 values the averaged r0 found will be a suitable parameter for similar molecular system s. It should be noted that the extraction of an r0 from a homoleptic ion, as in the case of the CoIII ions of Co8, will not account for the influence of one type of ligand over another f or the same oxidation state. It is thus evident that although the current tabulated r0 and B are fairly effective in the estimation of the valence, they may require further modification for molecular system s on which they will be applied. Reasons for signi ficant discrepancies of the BVS values from the expected ones cannot easily be explained; however they might be due to the main differences existing between a polymeric system and a molecular system. In

PAGE 146

146 the first type of arrangement, metal ions show a rel atively limited number of coordination numbers lie in generally high symmetry environments and experience fewer distortions than in a molecular system. The infinite lattice in a mineral generally represents a stable equilibrium where the energies associated with bond distances and coordination geometries achieve a minimum. On the other hand, a molecular compound has a much larger degree of variation. In the case of a metal oxo cluster, for example, the inner metal ions constituting the core will adjust ac cording to the strongest bonds, which are generally to metal s in the higher oxidation state. In large structures there will likely be a number of bonds to lower oxidation state metal ions that w ill be compressed or elongated. Moreover in a polymer each m etal is still surrounded by other metal ions in the same arrangement, in a molecular cluster the core is surrounded by ligands. It may be expected that the outer ligands adjust as per the core; however they may exert significant electronic effects on the metal ions impose geometric restrictions or introduce steric hindrance effects The quantification of these effects is not straightforward due to the large variety of possible cases, nevertheless optimization of the BVS parameters for molecular systems has proven to be an effective way to circumvent these difficulties. Known E xamples of BVS O ptimization One of the first examples of BVS optimization comes from Liu and Thorp194 who worked on the estimation of the oxi dation states in metalloenzymes by BVS. The author s collected a series of published crystallographic data on bond distances of mononuclear and polynuclear systems in biomolecules with the aim of creating a database of structures to obtain a new set of r0. The computation relied on fit ting of the data to the possible combinations of the metals with the most common biological

PAGE 147

147 ligands O, N and S. Hen ce, the probl em consisted of a threeparameter fit for the groups M O, MN and M S The algorithm calculat es the BVS value (independently from the donor combination) for each metal center and by calculating the difference from the expected valence finds a new r0. The adjustment of r0 to a BVS value closer to the actual valence is obtained through a standard Marquar dt algorithm. The authors describe the input files as one entry per each metal center, which consist s of the list of bond lengths and bond types (i.e. the donor). The starting point for the BVS calculation is chosen using the r0 values published by Brown.71 The optimization, performed on 829 entries, displayed a reduction in the error when applied for some examples (Table 42). Table 42 Comparison of BVS values obtained for a series of model compounds by the newly optimized and published71 r0 parameters (reported from ref 194). I on BVS a BVS b I on BVS a BVS b Fe II 2.15 1.99 V IV 4.44 4.00 Fe III 3.28 3.00 V V 5.47 5.00 Mn II 2.07 1.99 Ni II 2.88 1.99 Mn III 3.29 3.06 Ni III 3.44 3.00 Mn IV 4.03 4.00 Mo IV 4.65 3.99 Cu II 1.84 1.98 Mo V 5.49 4.98 V III 3.35 3.00 Mo VI 6.68 5.98 a Calculated taking the r0 values from Brown .71 b Calculated with the optimi zed r0. Another example of BVS optimization is found in the work of Urusov198 who introduced a different type of approach for the optimization of the r0 and B par ameters (in Eq. 4 1 ) The main idea consists in finding a particular B value for a given metal where r0 would represent the length of a bond with unitary contribution to the valence. When this is accomplished, it is possible to demonstrate that the paramet er B becomes a pure function of the softness of the metal ion. Previous work done by Zocchi199200 demonstrated the existence of such a correl ation in the case of Mo. Urusov focused on Mn and analyzed a limited num ber of polymeric structures (mainly from minerals) which display ed a bond with a partial valence of one (i.e.: sij = 1) Such an interaction

PAGE 148

148 possessing unitary valence is not easily observed as it needs homoleptic coordination and well defined structures f or which simple considerations about the electronic distribution can be assumed. This method requires the estimation of which bond has unitary contribution to the valence of the metal center. The need for a homoleptic structure has directed effort s on syst ems carrying O based ligands probably due to the large availability of polymeric structures where O is the bridging ligand. By this method, Zocchi has found that for MoIII, MoIV, MoVand MoVIO compounds the values r0 = 1.8788 and B = 0.3046 can be applied with sufficient accuracy to a MoO complex in that range of valences. Urusov found that for MnII to MnVII while r0 is fixed (i.e. r0 = 1), B shows a progressive variation with the valence (Fig. 42). Mn oxidation state 1 2 3 4 5 6 7 8 B 0.20 0.25 0.30 0.35 0.40 0.45 Figure 4 2. Plot of the variation in B vs. Mn oxidation state in MnO compounds obtained by Urusov.198 The plot in Fig. 42 shows an almost linear dependence of B with re spect to the valence (in homoleptic MnO compounds) with some deviation for MnIII, probably due to the JahnTeller effect and for MnVII due to a nonlinear increase of the rigidity in the

PAGE 149

149 coordination polyhedron. The approaches of Zocchi and Urusov consti tute a new example for treatment of crystallographic data for improving the BVS method, however they are limited to homoleptic systems, and hence to the relatively small number of structures and the category of metal O compounds (polymeric). The strategies described so far are for the improvement o f the method in it s classical formulation. R esearch on this topic has also led to a more radical approach where the classic formulation for the BVS ( Eq. 4 1) has been modified. Datta and coworkers201 while analyzing metal ions with zero or negative formal oxidation states modified Eq. 4 1 and established a new relation between Vij (valence along the ijth bond) and rij (ijth bond length) The authors consider the valence of a bond ( Vij) and its length rij correlated by a linear function rather than an exponential. T his concept was explored in the past by other researchers; however, Datta and coworkers introduced new constant s and terms, which modify this relation to descri be the experimental data more accurately The linear expression for Vij as function of rij is: Vij = rij N (where n is an adjustable constant). The new expression is shown in Eq. 4 where K, K n is the coordination number and Vi is the valence for the ion. j N ij K K ir R n K K V' ,' 1 (4 3) The adaptation of the par ameters for Eq. 4 3 consisted of an initial evaluation of the best constant s K and K (i.e.: 4 and 20 respectively) and then minimization of the valence error by variation of N through a least squares fitting of data ( from homoleptic compounds ) The database contained 415 structures of Ni and Cu wi th common donors like N, O, S, P and for the first time C. Although this method is of great help for metal

PAGE 150

150 ions in low oxid ation states its application as a general method for different type of metal ions may be difficult due to the large number of parameters in consider ation. All of these examples report least squares fitting methods on relatively limited databases either f or the number of entries or the nature of the compounds. The other major limitation of these methods is in not specifying any contribution from a possible dependence of one donor atom on another one of a different nature. For instance, the choice of homoleptic compounds during the optimization procedures has led t o data that fit well for homoleptic systems, but they do not explicitly refer to deviations consequent to adopting these parameter s on heteroleptic and molecular compounds All these problems call for a new approach and different optimization procedures. Approach to a N ew O ptimization M ethod Definitions Optimization represent s part of everyday activities even without realizing it. Fletcher describes the simple act of reaching the work location as an act of optimization.202 The optimal path will minimize the time needed that is then considered the objective ( parameter in need of optimization) of the optimization. The objective can be expressed as a mathematical function, the objective function that associates time with the possible paths. To perform the optimization it is necessary to know the correspondence between each entry in the domain of the objective function (all the paths) with the objective (the time). Essentially, for an objective function y= f(x) are calculated all the solutions for each x belonging to the domain, to find the value x for which y is the minimum possible (the optimal one). It is generally verified that a currently working solution is optimized with attempts that are numerically close in the domain. Translation of these basic concepts to more realistic problems often implies more than

PAGE 151

151 one objective making the optimization more complex in its mathematical formulation ( multiobjective )(for instance, reaching work while stopping for gas). Another important factor may consist in li miting the domain by applying meaningful constraints ; f or example, selecting only the paths accessible with the type of transportation adopted. In biochemistry, the minimization of stable energy states of proteins is found by optimization of the structure within physical constraints.203 The branch of mathematics that deals with extracting useful information from largesize databases is called data mining and it has been used here to improve by optimization the parameters used in the classical BVS formulation ( Eq. 4 1).204 Once t he objective function and the constraints have been established, the actual optimization may be very hard to define especially in the case of a global optimiz ation. A global optimization is performed on functions that can present more than one minimum (or optimum in general) and described by an objective function like: f : x A y B The solution to this minimization problem will be an x A that satisfies all the constraints requirements and that produces a value y which is the smallest in B ( Eq. 4 4). x A: f(x) f(x ) x A (4 4) It is possible to find subgroups of A where there are values of x for which the function is the smallest than for all other solutions and constitute local optima. A global minimum is defined as the best local minimum amongst all the ones that can be found within the whole feasible domain of the objective function. Finally, the most demanding task in an optimization is to create an appropriate function that describes the system behavior. This is called modeling and the quality of the optimization is directly dependent on the coherence of the mathematical expression in describing the real life

PAGE 152

152 application. Although simplifications are often necessary, the model should still represent the problem for the solutions to be acceptable. Exact M ethods An exact method of optimization consists in def ining the domain A the objective function, then apply ing to it all the constraints that define a subset C (where C A ) and solving the function for all the possible values x of C Finally, by comparison of all the possible outcomes, the optimal y value is found and so is the corresponding x (the solution of the optimization). This method is known as exhaustive searc h and is a method that guarantees the solution (or solutions) for the optimization. The application of this method is generally limited by two factors: the complexity of the mathematical function and the number of elements to evaluate. For application to r eal problems the function may result too complex for applying constraints or even to describe its general form. On the other hand, even a relatively limited group of elements may be too demanding computationally as evident by any interval [ a b ] of which containing an infinite number of elements it will require infinite time to analyze. Linear functions (where the variables can be multiplied by constants and the results summed together) are the simplest to approach; however they find only limited application. Nonlinear functions can be approached by differentiation where a descending path in the objective function can guide the optimizatio n toward the rapid localization of minima (derivat iv e = 0) .205 This approach can be limited by those differentiable functions that result in very complex derivatives or that contain multiple minima. A derivati ve of zero can be found in any minima and the algorithm cannot make a distinction between any minima and the global minimum (or absolute minimum).

PAGE 153

153 A complex function is often approximated with one of easier interpretation for the optimization with the prer equisite that the solutions found converge to the one that would be found for the original function.206 All the methods that aim at the analytical solution of an optimization are known as exact methods and they can be applied only to some functions and with some hypothesis. In all those case w here an exact method fails or cannot be applied the optimization is performed by meta heuristic methods. Meta H euristics Meta heuristic methods differently from exact methods do not guarantee a solution and in the cases where a solution is found it only entails some probability of being the actual global minimum. The lower degree of accuracy in finding the solution to a problem is compensated by a large flexibility of application and faster execution. The quality of the solution greatly depends on the algorith m built for the optimization. V arious optimization methods by metaheuristics are inspired by observation of natural phenomena where the scientists build algorithms c apable of finding solutions for the objective functions by adopt ing the same strategi es found by n ature. The most important examples are the physical annealing207, the music improvisation208, the Darwinian theory of evolution209 and the behavior of animals (ants, bees or monkeys).210212 Amongst the metaheuristic methods of opt imization, there are few categories not inspired by natural behaviors T he Variable Neighborhood Search (VNS)213 is an example of these categories and it has been used for the optimization in the present work. Description of the P roblem and C hoice of the M etaH euristics M ethod The BVS method relies on the sum of the terms arising from Eq. 4 1, which only provides the partial valence attributable to a M L bond. The system of Eq. 4 2 shows

PAGE 154

154 how it is possible to adapt the r0 value for a given metallic coordination environment of specified valence so that r0( V M, L) would perfectly fit the known valence ( V ). This approach is appropriate for systems of similar characteristics but it may not be suitable for many other complexes due to the large number of possible environments. There is thus a necessity to formulat e the optimization pr oblem with the purpose of establishing a set of r0 parameters capable of representing the largest possible group of metal environments having same metal type, valence and bound to the same subset of ligands. The error in the estimation of the valence is described by the difference in absolute value ( ) between the estimated valence ( V ) and the actual oxidation state, approximated with the valence ( ), as displayed in Eq. 4 5. = | V | (4 5) In a set of k elements (different coordination environments) for the metal M of possible valences V and having only ligands from the subset L, the total error f ( r0) for the use of the BVS model with r0( V M, L) and b constant is expressed by Eq. 4 6. k i i iV k r f1 01 ) ( (4 6) This formulation provides an aver age error for the BVS model over all possible valences adopted by the metal M and represents the objective function of this optimization. Although a process of optimization based on Eq. 4 6 may appear of easy solution due to the linearity of the terms, it is complicated by the true nature of Vi as an exponential. For this reason, the objective function is not linear, quadratic or differentiable (due to the presence of the absolute values in the sum) and its optimization requires the use of meta heuristics m ethods. Of the various types of metaheuristics methods, the VNS seems to be the most promising for application to the BVS. This method, described in

PAGE 155

155 detail later, focuses on the search for better solution in a close proximity of solutions already known. O nce the interval for each variable has been fully investigated and the method cannot provide any new improved solution, it stops. In this particular case the known solutions are represented by the already known r0 values from the published literature. The Va riable Neighborhood Search The VNS method is based on the search for new improved solutions within a variable's interval ( neighborhood) of adjustable size depending on the new solutions found.213 In the case of a simple objective function dependent on a single variable x, the method consider s as starting point a known solution x0. The x0 is perturbed by an oscillating function within an interval and a new random solution x1 is chosen. The new point x1 becomes the center of a neighborhood of selected extension and a local exhaustive search begins within. When a new solution x0 is found so that f ( x0 f ( x0) (for a minimization problem), x0 is replaced by x0 s from the beginning. If the new solution x0 f ( x0 f ( x0), the size of the interval or neighborhood around x0 is sl ightly increased and rechecked for a new solution. As long as there are not new solutions that improve the minimization, the intervals around the original x0 keep increasing up to reaching the limits imposed by the constraints. Initial E xperimental S etup S election of the S tructural D ata Initial attempts were directed to the optimization of the BVS parameters for Fe and Mn compounds containing only a combination of N or O donor ligands Two easily available databases belonging to the Christou group (IUMSC, Indiana University

PAGE 156

156 Molecular Structure Center and the CXC, Center for X ray Crystallography at the University of Florida) were investigated and the relevant cif (crystallographic information files) were obtained. Initial difficulties were found for many ol der structures that required transformation of the file from a different format to the more common cif extension. The cif files contain various types of information relative to the diffraction experiment and only a portion is relevant to identify ing the di stances between boned atom s. To provide the optimization software with consistent input data, each crystallographic file was manually transformed to a text file containing an entry for each metal ion. Each entry was constituted by five values, which ident ify metal type, unique metal label, ligand type, unique ligand label and finally the bond distance. Other type of problems to overcome came from the presence in the cif files of nonspecific bonding between metal ions and improbable species. For instance, hydrogen atoms in aliphatic chains of a ligand w ere sometimes reported bonded to a metal ion, or a disordered ligand was reported bound twice in the bond list. These problems were solved by manually screening all the crystallographic structures and applyi ng appropriate corrections to the cif files. Many of the crystallographic files contained metal ions equivalent by symmetry which have the same label and bond distances with the corresponding donors It has been chosen to use these coordination environment s only once as they constitute a single measurement reported multiple times. Although this process lowered the overall number of entries in the input files, it would lead to a more reliable set of data representing each unique topology rather than for the number that was reported in a crystal structure. For example, in the core of complex [Mn12O12(O2CMe)16(H2O)4],

PAGE 157

157 shown in Fig. 43 there are only three different types of Mn ions despite the overall dodecanuclear structure. On the other hand, X ray determinations of the same species have been each considered in the optimization as representative of a different measurement. The selection process resulted in a collection of 733 combined entries for Fe and Mn. Figure 4 3. Labeled core of [Mn12O12(O2CMe)16(H2O )4]. There are only three different types of Mn ions by symmetry. Table 43 BVS parameters for Fe and Mn obtained from the literature131 and by a preliminary VNS optimization (with b variable) over a selec ted database (). Parameters M L type r 0 (M II ) r 0 (M III ) r 0 (M IV ) b Current a Mn O 1.790 1.760 1.753 0.37 Current a Mn N 1.849 1.834 1.822 0.37 Optimized Mn O 1.7429 1.7202 1.7351 0.4017 Optimized Mn N 1.7863 1.7828 1.7961 0.4017 Current a Fe O 1.734 1.759 0.37 Current a Fe N 1.769 1.815 0.37 Optimized Fe O 1.7650 1.7397 0.3840 Optimized Fe N 1.8600 1.7424 0.3840 a Data reported from ref. 131 For Fe, only the +2 and +3 oxidation states were analyzed, whereas for Mn the +4 state was included. The computation for the two different metals was executed in

PAGE 158

158 parallel and accounted for the data reported in table 4 3 An important aspect of this optimization is that the parameter b has been considered variable rather than keeping it constant at 0.37 Such an approach does not require the BVS model to fit as many metals as possible. Although t he BVS formula loses in generalization, it can better fit the characteristic behavior of each metal. A plot o f the general exponential form contained in the BVS model displays a change in profile that authors like Urusov ascribed to a different degree of softness of the metal ion.198 According to this interpretation, the r0 values for the metals have been calculated throughout the transition series assuming early or late met als to posses the same softness or hardness. The variability of b in this optimization should then provide a means for a greater accuracy of the BVS model based on a n improved matching of b with the polarizability of the el ectron cloud. As an example of the application of the new parameters to a representative Mn cluster, Fig 4 3 and Table 44 report respectively the core a nd the BVS calculations for [Mn12O12(O2CMe)16(H2O)4]. The BVS data obtained for [Mn12O12(O2CMe)16(H2O)4] using the currently available r0 and b (0.37 ), leads to a clear identif ication of the oxidation states; h owever, by use of the s ame formulation, the VNS optimized r0 and b result in better precision with lower errors. Table 44 BVS values for [Mn12O12(O2CMe)16(H2O)4] calculated with published literature data131 and by preliminary VNS optimization of r0 and b Ion Parameters V i for Mn II V i for Mn III V i for Mn IV Mn1 Current 4.519 4.167 4.089 Mn1 Optimized 4.109 3.884 4.030 Mn2 Current 3.536 3.261 3.200 Mn2 Optimized 3.264 3.085 3.201 Mn 3 Current 3.396 3.131 3.073 Mn3 Optimized 3.150 2.997 3.089 The numbers with an asterisk represent the only ones being acceptably close to the actual oxidation state.

PAGE 159

159 In the case of an Fe cluster the cation [Fe3O(O2CMe)6(H2O)3]+ has been chosen as representative and its labeled structure and BVS data are shown in Fig. 44 and Table 45. As in the case of Mn, the BVS values obtained from the published data are clear enough to distinguish, without uncertainty, the oxidation state of each Fe ion. Nevertheless, the VNS optimized parameter leads to smaller errors allowing for a higher precision. Figure 4 4. Pov Ray projection of the partially labeled structure of the cation [Fe3O(O2CMe)6(H2O)3]+, (CCDC code: cojpum)8. Table 45 BVS values for [Fe3O(O2CMe)6(H2O)3]+ calculated with published literature data131 and by preliminary VNS optimization of r0 and b Ion Parameters V i for Fe II V i for Fe III Fe1 Current 2.962 3.169 Fe1 Optimized 3.293 3.083 Fe2 Current 2.962 3.169 F e2 Optimized 3.293 3.083 Fe3 Current 2.923 3.128 Fe3 Optimized 3.252 3.045 The numbers with an asterisk represent the only ones being acceptably close to the actual oxidation state.

PAGE 160

160 The two clusters may only be representative of a limited number of structural types; hence, chopt has been implemented for the creation of a list reporting the error for each entry based on the currently available BVS parameters and the optimized ones. Figure 45 reports the histogram for the errors ( ) over the entire database. Figure 4 5. Histogram of the error s (absolute values) on the valence estimation by the currently available BVS parameters ( blue bars) and the ones derived from optimization (red bars) over an entire selected structural database. The histogram shows the discrepancies f or the BVS model with the nonoptimized r0 parameters to be distributed over a large range than those resulting from application of the model with the optimized r0 and b The histogram for the latter shows a higher proportion of entries with smaller error and in general is overall more concentrated about zero demonstrating an increased number of structures to statistically provide a smaller error If for example the error = 0.15 I sconsidered, only ~ 50% of the entries fall within the limits when calculated with the currently published r0 values (and b = 0.37 ), whereas ~ 90% are found within = 0.15 application of the newly optimized r0 and b From this point on, any optimized parameter will report the subscript "opt" (for

PAGE 161

161 example, r0 opt or bopt), and any parameter from the literature will be reported without distinctions (for example, r0 or b ). Proposal for a N ew BVS F ormula The availability of a selected database of structures and the flexibility of the software adopted for the initial attempt of optimization has encouraged efforts to improve the formulation of the BVS model. According to the current formulation, Eq. 4 1 is applied to every bond to obtain the individual valences (si) that are then added to gain the bond v alence sum as expressed in Eq. 4 7, where r0 is function of the metal valence (V), metal type (M) and the donor (L), ri is the bond length and B or b is a constant set at 0.37. i n B r L M V r i n iie s V1 ) ( 0 1 (4 7 ) In a metal center containing the same donor atom more than once, each bond of the same type will be considered separately despite being correlated to the same r0 value. A proposed variation would be to approximate the different bond lengths with the same donor type with their average and sum the contribution to the valence for how many times that donor appears in the formula (hence, replace rA rA and rA with Ar and multiply by three its contribution to the valence) see Eq. 4 8 Equation 48 describes the BVS formula for a metal ion (M) possessing n different donors, each appearing m times. It should be noted that the value of B is now relative to the metal. The new formulation has been implemented in chopt and a new optimization iteration was applied to the database using as starting points the r0 opt and bopt previously found

PAGE 162

162 and listed in table 45. The optimization produced only a slight improvement over r0 opt and bopt so that distribution of errors after the optimization is similar to the one before the optimization (with the use of the new BVS formula (Eq. 48). Figure 46 illustrates the comparison of the error distribution before and after optimization of Eq. 48 with the initial r0 opt and bopt. n j M B r m L M r nm i i ne m V1 ) ( 1 ) (1 0 (4 8) Figure 4 6. Comparison of error distribution for the optimized BVS parameter r0 and b with the currently published formula (blue bars) and the optimization of the newly found r0 opt and bopt with the new proposed formulation (red bars). Besides a minor improvement of the errors with the second optimization, the data shows how Eq. 4 8 is not capable of approaching the BVS problem better than Eq. 4 7. The latter could concentrate ~ 90% of the entries within the limit = 0.15 (from the initial ~ 50% before optimization) whereas the new formulation only reaches ~ 73 %

PAGE 163

163 either with or without optimization. This is proof that averaging the same type of metal donor bonds is not a convenient approximation to improve the BVS accuracy, hence there is no reason to adopt Eq. 4 8 in place of Eq. 4 7. The VNS approach was useful for a marginal reduction in error by clustering the data for the objective function closer to zero. Application to the CCDC D atabase Data R etrieval from the CCDC D atabase The successful application of the VNS optimization technique to the initial database of 733 entries has encouraged the extension of such work to an ample database that could provide a larger set of data for further improvement of the published r0 and b There are multiple advantages for retrieving data from the CCDC database. Besides representing the largest collection to date of singlecrystal molecular structures, it also offers dedicated software ( ConQuest v1.12) for interrogating the database and is capable of produc ing outputs that can be further analyzed or compared through logical functions. Another important advantage is the possibility of extracting the d ata as cif files possessing roughly the same architecture, hence facilitating the process of locating and copying the bond lengths. For each metal ion belonging to the first transition row a set of approximately 30 queries has been used. The queries locate d those structures where the metal under scrutiny was bound to the following type of donors: O, N, P, S, F, Cl, Br or I. Cyanide ligands were excluded f rom this subset due to their particular electronic structure, which may not be representative of N donor s. A ll those cases where the metal was bound to one of t he following: H, B, C, As, Se, Te or another metal were excluded because often associated with nonpositive values for the metal oxidation state. The BVS method, does not account for covalency and fai ls whenever

PAGE 164

164 the system diverge from an electrostatic model. Further considerations on the type of structures have le d to the exclusion of those cases that would display obvious bond length distortion, such as macrocyclic chelation. Therefore, all the molec ules containing most common macrocyclic ligands with four to six donors N or Obased have been excluded too. All these queries where applied on the database with a further filter for structures having no disorder, error or powder determinations. This init ial type of search produced large outputs with no discrimination over the oxidation state of the metal. Another set of searches was performed on each metal by adding queries that would isolate those structures including only one type of oxidation state. For each metallic species a general set of structures and a series of subsets each relative to a particular oxidation state was obtained. The software, ConQuest showed some pitfall s in locating all structures according to the set of queries, which required additional workup of the output data. For instance, a search for a metal in a particular oxidation state has produced outcome hits that did not appear in the same set of queries without specification of the oxidation state. Due to these mismatches, all the output for each metal w as crossexamined to e nsure that each structure with known oxidation state would appear also in the more general set for that metal. Data and S oftware P reparation for the O ptimization The large volume o f data to be analyzed required the coding of a new routine capable of automatically access ing all the cif files and producing the corresponding output text files with entries for each metal ion not reported by symmetry. The algorithm associates errors for anomalous cif files containing an unusual number of donors, unknown donors, or otherwise corrupted data that the ConQuest filter failed to exclude.

PAGE 165

165 An entire collection of r0 is necessary for the final stage of optimization, hence t he program was also implemented to calculate r0 valu es from each metal subset of known valence (through Eq. 4 2) The last published data on the BVS model report only some of the possible r0 combinations for the donors O, N, P, S, F, Cl, Br or I for a given metal in common valences The published collection of r0 has been used with no modification as starting parameters f or the optimization stage.214 The aforementioned program was compiled to obtain all the missing r0 data for each M L combination; however it failed in some instances that required manual intervention as explained below. Table 46 reports the results of the preliminary r0 computation in the case of Fe ( b = 0.37 ). Table 46 R0 values obtained by computations on the subset data of known valence for iron ( b = 0.37 ) Donor ( / Valence ( ) 1 2 3 4 5 O 2.29 1.72 1.75 1.74 1.83 N 2.36 1.78 1.84 1.81 1.90 P 2.77 2.15 2.22 2.28 2.31 S 2.66 2.10 2.14 2.10 2.17 F 2.06 1.49 1.52 1.51 1.61 Cl 2.63 2.06 2.09 2.07 2.17 Br 2.78 2.19 2.25 2.22 2. 32 I 2.98 2.40 2.53 2.42 2.45 The numbers with an asterisk represent the r0 ( ) obtained either from the literature214 or by software calculation. All the remaining values have been derived from the ones in bold. The values in Table 46 contain the published data and those calculated by the program (both in bold) as well as those that have been otherwise derived. For N donor s all the values for r0 were found except for the combination FeIN, which was manually calculated as average from homoleptic structures of the type FeIN (i.e.: 2.36 ). An a nalogous approach was used for all the other combinations for which suitable homoleptic structures could be found. It w as also noted that there is an almost parallel profile for the correlation between the r0 values of a donor with the valence. Such

PAGE 166

166 observation has been used for estimating the r0 data of donors with little or no outcome from the software or in any case where they could not be calculated directly from a structure. In the case of F, for exam ple, it was manually calculated the r0 for FeIIF (i.e.: 1.49 ). This datum was then compared with the r0 value for the FeIIN (taken in this case as reference, i.e.: 1.78 ) and the difference (0.29 ) used to create a parallel profile that could represent the other missing data for the Fe F combinations (Fig. 47) Valence number / v.u. 1 2 3 4 5 6 r0 / 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 O N P S F Cl Br I Figure 4 7. Correlation between published214 or derived r0 values with the valence for given donors in iron structures. This approach allowed for the estimation of all possible r0 for every M L combination that had sufficient representation in the subset folders of known valence. The full compilation of r0 provided all the necessary starting points for the VNS optimization performed with an updated version of chopt The optimization was performed on one metal at a time on the entire set of structures including those with unknown valence. Each entry is read by chopt and assigned a tentative valence unless the BVS exceeds a 0.30 limit from the closest integer, in which case it is momentarily

PAGE 167

167 rejected. After an initial cycle of optimization the initial r0 and b are replaced with the newly found r0 opt and bopt, and the previously rejected entries are reanalyzed to include those that now fulfill the 0.30 limit. This procedure is repeated continuously with the aim of extending the optimization over the largest number of entries possible while insuring acceptable BVS estimation through the 0.30 limit. Chopt has also an implemented routine that creates a histogram for the er ror associated with each entry monitoring the overall quality of the optimization. An exemplified flow chart of chopt algorithm is reported in Fig. D 1. The large volume of input data ( ~ 10000 entries per metal on average, corresponding to ~ 60000 bond lengt h s) and the high number of variables to optimize (65 in case of Mn) would required prohibitive computational time on a common calculator. Hence, the data has been moved, along with chopt on the computer cloud Grid5000 ( mainly based in France) with a drast ic reduction for the optimization time. Analysis of the R esults All the newly obtained r0 opt and bopt have been applied with the Eq. 4 7 to the entire set of data. This approach allows for estimation of error over the population of data rather than a sample of it, providing conclusions that are more reliable. Some combinations are not represented in the database; hence, the optimization could not provide the relative r0 opt for them Titanium The most common oxidation state for titanium is +4. The other st ates of l ower valence are air sensitive and its coordination number may range from three to eight .7 Table D 1 and Fig. D 2 report the list of t he new of r0 opt and bopt and the histogram of

PAGE 168

168 the errors obtained by application of Eq. 4 7 respectively The optimization is relative to 1746 files corresponding to 2244 entries. Vanadium Vanadium compounds have positive oxidation states ranging from +1 to +5, displaying coordination numbers from four to seven. The most widely found oxidation state is +4 due in large part to its stability. Other lower oxidation states are readily oxidized in air. Being a hard metal it prefers oxygen coordination, account ing for the majority of its compounds.7 The list of r0 opt and bopt found for this metal are reported in T able D 2 whereas the error distribut ion over the entire V database is in Fig. D 2 The CCDC database did not report structures for the +1 valence. The optimization was obtained from 2850 cif files corresponding to 5464 entries. Chromium Chromium displays in its compounds positive oxidation states ranging from +1 to +6 T he most common s are +2, readily oxidized to the more stable and present +3 and the +6, often found in oxides. The coordination numbers vary from three to eight, with a preference for six.7 Table D 3 reports the new r0 opt and bopt for chromium while the error over the entire database of entries is in Fig. D 2 In the case of chromium, the optimization was conducted over 1566 cif files corresponding to 2071 entries. Manganese Manganese present several oxidation states of which t he +2 is the most stable in the range +1 to +7. Permanganate is the most stable high oxidat ion state followed by manganate. The +5 state is not stable, although it may form as an intermediate. The +3 and +4 states are especially present in hydroxoor oxo based clusters. The predomina nt coordination number is six, although all possibilities between two and eight

PAGE 169

169 have been reported.7 Table D 4 and Fig. D 2 report the list of new r0 opt and bopt in the former and the histogram of the errors in the second. The optimization involved 4675 f iles for 5767 entries. Iron Iron can easily assume the oxidation stated +2 or +3. All other states are instable, and the highest reported is the +6, which is rare. The coordination numbers range from three to eight, although the four and six coordinations (tetrahedral and octahedral respectively) are preferred. The new r0 opt and bopt are reported in Table D 5 while the histogram for the error associated to the entire database F ig. D 2 The optimization on the iron structures was based on 6044 files or 8257 entries. Cobalt Cobalt compounds are stable in the +2 and +3 oxidation state. The +4 state is uncommon, while the +1 is more prevale nt. The coordination number ranges from three to eight and only the four and six coordinate are the most representative.7 The newly obtained r0 opt and bopt are reported in the Table D 6 while Fig. D 2 reports the histogram of the BVS errors applied to the entir e cobalt database. The optimization for cobalt was obtained by analysis of 8897 crystallographic files accounting for 8902 entries. Nickel Nickel ions are prevalent in the oxidation state +2, whereas +3 and +4 are respectively uncommon and rare. In the higher oxidation states, it is thought that the ligands may be responsible for the formal assignment by being oxidized. The +1 state is found although also in this case it may not reflect the actual physical configuration of the metal Nickel coordination is more limited in the number of donors, ranging from three to

PAGE 170

170 six.7 The data for r0 opt and bopt are reported in Table D 7 while the histogram of the relative errors is illustrated in Fig. D 2 The optimization was based on 10150 cif files corresponding to 10204 entries. Copper Copper ions have a restricted number of oxidation states allowed, from +1 to +3. The +4 state is only hypothesized, and t he +3 is uncommon. Coordination numbers range from two to six with conspicuous presence of pentacoordination.7 In Table D 8 are listed the new r0 opt and bopt for the various donors, while Fig. D 2 report s the histogram of the BVS errors counted on the entire copper database. The optimization procedure was carried on 21680 crystallographic files amounting to 22485 entries. Conclusions Current application of the BVS model to molecular system s is known to be feasible despite, in many cases large discrepancies being found. Attempts to improve the BVS method through its parameters r0 and b or even its formulation have led to different set s of data. E ach set is tailored to fit only a particular category of compound, for example inorganic polymers rather than inorganic sites of biomolecules. The proposed o ptimization by metaheuristic methods was applied for the first time to the BVS model and t o larges t database of metallic complexes The unprecedented large set of data allowed for the isolation of parameters that have proven to be a better fit for each transition metal in the first row, as demonstrated by the histograms for the entire population of mol ecular structures. T he influence exerted by a metal dependent parameter b in matching the characteristics of each metal ion rather than using the generalized value of 0.37 throughout the series is not negligible. The consequent lack of generalization is

PAGE 171

171 overcome by a significant increase in accuracy on all metal ions considered, especially on early metals. The improvement decreases moving toward later metals up to copper for which the gain in accuracy is minimal. Another important advantage offered by thi s method consist s in the simultaneous analysis on multiple variables, which granted a vast compilation of r0 opt for almost every non organometallic donor in contrast to previous optimization examples reporting only limited combinations The preparation of datasets as well as the complex development of a computer program capable of extrapolating data and automatically proceed to the optimization are limiting factors ; h owever, it may fi nd ease of applicability to several metal ions with consequent potential application of the BVS method in other areas of chemistry.

PAGE 172

172 APPENDIX A CODES FOR THE COMPOUNDS LISTED [Mn12O12(O2CMe)16(H2O)4] ( 1 ) [Mn13O8(OMe)6(fdc)6] ( 2 ) [Mn13O8(O Et )6(fdc)6] ( 3 ) [Mn8O4(fdc)6(DMF)2(H2O)2] ( 4 ) [Mn8O4(fdc)6(DMF)4] ( 5 ) [Fe7O3(OMe)(fdc)6(MeOH)3]3+[FeCl4] 2 Cl ( 6 ) [Fe6.2Mn0.8O3(OMe)(fdc)6(H2O)3]+Cl ( 7 ) [FexCo7O3(OMe)(fdc)6(MeOH)3]2+2Cl ( 8 ) [FexNi7O3(OMe)(fdc)6(MeOH)3]+Cl ( 9 ) [FexZn7O3(OMe)(fdc)6(MeOH)3]+ [ZnCl4]2 ( 10)

PAGE 173

173 APPENDIX B FULL CRY STALLOGRAPHIC TABLES Table B 1. Crystal data and structure refinement for 2 CH2Cl2. Identification code xm221 Empirical formula C92 H94 Cl16 Fe6 Mn13 O38 Formula weight 3424.26 Temperature 173(2) K Wavelength 0.71073 Crystal system Triclinic Spa ce group P 1 Unit cell dimensions a = 14.1308(12) = 119.4700(10). b = 15.1738(12) = 91.4320(10). c = 15.3656(13) = 100.339(2). Volume 2797.0(4) 3 Z 1 Density (calculated) 1.969 Mg/m3 Absorption coefficient 2.621 mm-1 F(000) 1643 Crystal size 0.12 x 0.12 x 0.04 mm3 Theta range for data collection 1.91 to 27.50. Index ranges 16 19 19 Reflections collected 18432 Independent reflections 12232 [R(int) = 0.0388] Completeness to theta = 27.50 95.2 % Absorption correction Integration Max. and min. transmission 0.8774 and 0.6838 Refinement method Full matrix least squares on F2 Data / restraints / parameters 12232 / 0 / 615 Goodness of fit on F2 1.055 Final R indices [I>2sigma(I)] R1 = 0.0479, wR2 = 0.1190 [9167] R indices (all data) R1 = 0.0668, wR2 = 0.1247 Largest diff. peak a nd hole 0.951 and 0.573 e.3 R1 = (||F o | |F c ||) / |F o |; wR2 = [ w(F o 2 F c 2 ) 2 ] / w F o 2 2 ]] 1/2 ; S = [ w(F o 2 F c 2 ) 2 ] / (n -p)] 1/2 w= 1/[ 2 (F o 2 )+(m*p)2+n*p], p = [max(F o 2 ,0)+ 2* F c 2 ]/3, m & n are constants

PAGE 174

174 Table B 2 Crystal data and structure refinement for 4 4DMF4H2O Identific ation code m89 Empirical formula C90 H102 Fe6 Mn8 N6 O40 Formula weight 2682.40 Temperature 173(2) K Wavelength 0.71073 Crystal system Monoclinic Space group C2/c Unit cell dimensions a = 23.282(3) = 90. b = 19.331(3) = 97.885(3). c = 22 .198(3) = 90. Volume 9896(3) Z 4 Density (calculated) 1.660 Mg/m3 Absorption coefficient 1.950 mm 1 F(000) 4980 Crystal size 0.26 x 0.22 x 0.20 mm3 Theta range for data collection 2.11 to 27.50. Index ranges 30 20 28 Reflections collected 32126 Independent reflections 11193 [R(int) = 0.0499] Completeness to theta = 27.50 98.4 % Absorption correction None Max. and min. transmission 0.6964 and 0.6310 Refinement method Full matr ix least squares on F2 Data / restraints / parameters 11193 / 0 / 670 Goodness of fit on F2 1.010 Final R indices [I>2sigma(I)] R1 = 0.0456, wR2 = 0.0927 [7031] R indices (all data) R1 = 0.0934, wR2 = 0.1117 Largest diff. peak and hole 0.525 and 0.668 e. 3 R1 = (||F o | |F c ||) / |F o | wR2 = [ w(F o 2 F c 2 ) 2 ] / w F o 2 2 ]] 1/2 S = [ w(F o 2 F c 2 ) 2 ] / (n -p)] 1/2 w= 1/[ 2 (F o 2 )+(m*p)2+n*p], p = [max(F o 2 ,0)+ 2* F c 2 ]/3, m & n are constants

PAGE 175

175 Table B 3. Crystal data and structure refinement for 5 3DMFH2O Identificatio n code mas28 Empirical formula C96 H104 Fe6 Mn8 N8 O36 Formula weight 2720.49 Temperature 100(2) K Wavelength 0.71073 Crystal system Monoclinic Space group P21/c Unit cell dimensions a = 24.015(2) = 90. b = 14.8647(14) = 101.088(5). c = 29.558(3) = 90. Volume 10354.7(17) 3 Z 4 Density (calculated) 1.745 Mg/m3 Absorption coefficient 1.839 mm 1 F(000) 5520 Crystal size 0.20 x 0.18 x 0.10 mm3 Theta range for data collection 1.54 to 27.50. Index ranges 31 19 38 Reflections collected 117623 Independent reflections 23801 [R(int) = 0.0826] Completeness to theta = 27.50 100.0 % Absorption correction Numerical Max. and min. transmission 0.8332 and 0.7122 Refinement method Full matrix least squares on F2 Data / restraints / parameters 23801 / 0 / 1210 Goodness of fit on F2 0.928 Final R indices [I>2sigma(I)] R1 = 0.0429, wR2 = 0.0947 [15611] R indices (all data) R1 = 0.0703, wR2 = 0.0999 Largest diff. peak and hole 0.847 and 0.804 e. 3 R1 = (||F o | |F c ||) / |F o | wR2 = [ w(F o 2 F c 2 ) 2 ] / w F o 2 2 ]] 1/2 S = [ w(F o 2 F c 2 ) 2 ] / (n -p)] 1/2 w= 1/[ 2 (F o 2 )+(m*p)2+n*p], p = [max(F o 2 ,0)+ 2* F c 2 ]/3, m & n are constants

PAGE 176

176 Table B 4. Crystal data and structure refinement for 6 2MeOH 2H2O. Identificat ion code mas05 Empirical formula C86 H91 Cl4 Fe14 O35 Formula weight 2608.29 Temperature 173(2) K Wavelength 0.71073 Crystal system Monoclinic Space group P2(1)/c Unit cell dimensions a = 18.4240(12) = 90. b = 24.6645(15) = 109.349(1). c = 24.1746(15) = 90. Volume 10364.9(11) 3 Z 4 Density (calculated) 1.671 Mg/m3 Absorption coefficient 2.075 mm-1 F(000) 5276 Crystal size 0.25 x 0.14 x 0.08 mm3 Theta range for data collection 1.43 to 27.5. Index ranges 23 26 31 Reflections collected 70131 Independent reflections 23599 [R(int) = 0.0675] Completeness to theta = 27.50 99.1 % Absorption correction Integration Max. and min. transmission 0.8552 and 0.6558 Refinement method Full matrix least squares on F2 Data / restr aints / parameters 23599 / 15 / 1119 Goodness of fit on F2 0.702 Final R indices [I>2sigma(I)] R1 = 0.0406, wR2 = 0.0831 [8375] R indices (all data) R1 = 0.0933, wR2 = 0.0859 Largest diff. peak and hole 1.377 and 1.006 e.3 R1 = (||F o | |F c ||) / |F o | wR2 = [ w(F o 2 F c 2 ) 2 ] / w F o 2 2 ]] 1/2 S = [ w(F o 2 F c 2 ) 2 ] / (n -p)] 1/2 w= 1/[ 2 (F o 2 )+(m*p)2+n*p], p = [max(F o 2 ,0)+ 2* F c 2 ]/3, m & n are constants.

PAGE 177

177 Table B 5 Crystal data and structure refinement for 7 3DMF3H2O Identificat ion code xm228 Empirical formula C7 9 H81 Fe6 Mn7 O37 Formula weight 2342.14 Temperature 173(2) K Wavelength 0.71073 Crystal system Rhombohedral Space group R 3 Unit cell dimensions a = 15.8534(6) = 90. b = 15.8534(6) = 90. c = 64.090(5) = 120. Volume 13949.7(13) 3 Z 6 Density (calculated) 1.611 Mg/m3 Absorption coefficient 1.897 mm-1 F(000) 6798 Crystal size 0.09 x 0.09 x 0.04 mm3 Theta range for data collection 1.52 to 25.71. Index ranges 14 18 77 Reflections c ollected 21258 Independent reflections 5265 [R(int) = 0.0537] Completeness to theta = 25.71 88.7 % Absorption correction Integration Max. and min. transmission 0.9306 and 0.8177 Refinement method Full matrix least squares on F2 Data / restraints / parameters 5265 / 0 / 352 Goodness of fit on F2 0.998 Final R indices [I>2sigma(I)] R1 = 0.0359, wR2 = 0.0970 [3756] R indices (all data) R1 = 0.0566, wR2 = 0.1016 Largest diff. peak and hole 0.682 and 0.570 e.3 R1 = (||F o | |F c ||) / |F o | wR2 = [ w(F o 2 F c 2 ) 2 ] / w F o 2 2 ]] 1/2 S = [ w(F o 2 F c 2 ) 2 ] / (n -p)] 1/2 w= 1/[ 2 (F o 2 )+(m*p)2+n*p], p = [max(F o 2 ,0)+ 2* F c 2 ]/3, m & n are constants.

PAGE 178

178 Table B 6 Crystal data and structure refinement for 8 4MeOH 2H2O Identifica tion code mas03 Empirical formula C92 H1 31 Fe6 Co7 O49 Formula weight 2768.62 Temperature 173(2) K Wavelength 0.71073 Crystal system Trigonal Space group P31c Unit cell dimensions a = 16.7433(5) = 90. b = 16.7433 (5) = 90. c = 2 1.0276(12) = 120. Volume 5105.1 (4) 3 Z 2 Density (calculated) 1.80 1 Mg/m3 Absorption coefficient 2.21 mm-1 F(000) 3184 Crystal size 0.20 x 0.18 x 0.18 mm3 The ta range for data collection 1.85 to 27.51 Index ranges 20 15, 26 Reflections co llected 33742 Independent reflections 7491 [R(int) = 0.045 6] Completeness to theta = 27.49 99.9 % Absorption correction Integration Max. and min. transmission 0.7234 and 0.6435 Refinement method Full matrix least squares on F2 Dat a / restraints / paramet ers 7491 / 1 / 352 Goodness of fit on F2 0.935 Final R indices [I>2sig ma(I)] R1 = 0.0357, wR2 = 0.0712 [6155 ] R indices (all data) R1 = 0.0384, wR2 = 0.0672 Ab solute structure parameter 0.477(12 ) Largest diff. peak and hole 0.499 and 0.366 e.3 R1 = (||F o | |F c ||) / |F o | wR2 = [ w(F o 2 F c 2 ) 2 ] / w F o 2 2 ]] 1/2 S = [ w(F o 2 F c 2 ) 2 ] / (n -p)] 1/2 w= 1/[ 2 (F o 2 )+(m*p)2+n*p], p = [max(F o 2 ,0)+ 2* F c 2 ]/3, m & n are constants.

PAGE 179

179 Table B 7 Crystal data and structure refinement for 9 4MeOH Identification c ode mas04 Empirical formula C92 H127 Fe6 Ni7 O47 Formula weight 2731.01 Temperature 173(2) K Wavelength 0.71073 Crystal system Trigonal Space group P31c Unit cell dimensions a = 16.7578(5) = 90. b = 16.7578(5) = 90. c = 20.5954(12) = 120. Volume 5008.8(4) 3 Z 2 Density (calculated) 1.881 Mg/m3 Absorption coefficient 2.22 mm-1 F(000) 2814 Crystal size 0.23 x 0.19 x 0.19 mm3 Theta range for data collection 1.72 to 27.49. Index ranges 21 15 26 Reflections collected 32897 Independent reflections 7427 [R(int) = 0.0416] Completeness to theta = 27.49 99.9 % Absorption correction Integration Max. and min. transmission 0.7283 and 0.6445 Refinement method Full matrix least squares on F2 Data / restraints / parameters 7427 / 1 / 366 Goodness of fit on F2 0.951 Final R indices [I>2sigma(I)] R1 = 0.0274, wR2 = 0.0650 [6190] R indices (all data) R1 = 0.0343, wR2 = 0.0661 Absolute structure parameter 0.464(11) Largest diff. peak and hole 0.516 and 0.371 e.3 R1 = (||F o | |F c ||) / |F o | wR2 = [ w(F o 2 F c 2 ) 2 ] / w F o 2 2 ]] 1/2 S = [ w(F o 2 F c 2 ) 2 ] / (n -p)] 1/2 w= 1/[ 2 (F o 2 )+(m*p)2+n*p], p = [max(F o 2 ,0)+ 2* F c 2 ]/3, m & n are constants.

PAGE 180

180 Table B 8 Crystal data and structure refinement for 10 6MeOH2H2O Identifica tion code mas11 Empirical formula C94 H117 Cl2 Fe6 O49 Zn7.50 Formula weight 2927.15 Temperature 173(2) K Wavelength 0.71073 Crystal system Trigonal Space group P 3 Unit cell dimensions a = 16.7685(8) = 90. b = 16.7685(8) = 90. c = 21.50 7(2) = 120. Volume 5237.1(6) 3 Z 2 Density (calculated) 1.856 Mg/m3 Absorption coefficient 2.635 mm-1 F(000) 2976 Crystal size 0.26 x 0.12 x 0.02 mm3 Theta range for data collection 1.69 to 27.49. Index ranges 12 21 27 Reflection s collected 35569 Independent reflections 8028 [R(int) = 0.1275] Completeness to theta = 27.49 100.0 % Absorption correction Integration Max. and min. transmission 0.9455 and 0.5903 Refinement method Full matrix least squares on F2 Data / restraints / pa rameters 8028 / 8 / 381 Goodness of fit on F2 1.105 Final R indices [I>2sigma(I)] R1 = 0.0756, wR2 = 0.1985 [4282] R indices (all data) R1 = 0.1208, wR2 = 0.2147 Largest diff. peak and hole 2.319 and 1.172 e.3 R1 = (||F o | |F c ||) / |F o | wR2 = [ w(F o 2 F c 2 ) 2 ] / w F o 2 2 ]] 1/2 S = [ w(F o 2 F c 2 ) 2 ] / (n -p)] 1/2 w= 1/[ 2 (F o 2 )+(m*p)2+n*p], p = [max(F o 2 ,0)+ 2* F c 2 ]/3, m & n are constants.

PAGE 181

181 APPENDIX C BOND LENGTHS AND ANG LES Table C 1. Selected bond lengths [] and angles [ ] for [Mn13O8(OMe)6(fdc)6], 2 ______________________________________________________________________ Mn1O19#1 1.876(3) Mn1O5 1.912(3) Mn1O2 1.956(3) Mn1O17#1 1.999(3) Mn1O14#1 2.097(3) Mn1O16#1 2.299(3) Mn1Mn5#1 2.9201(9) Mn1Mn6 #1 2. 9215(9) Mn1Mn7 3.0349(6) Mn1Mn4 3.1757(8) Mn1Mn2 3.2294(9) Mn2O4 2.059(3) Mn2O6 2.074(3) Mn2O18 2.195(3) Mn2O10 2.222(3) Mn2O5 2.227(3) Mn2O16#1 2.421(3) Mn2Mn5 3.1573(9) Mn2Mn3 3.1964(9) Mn2Mn7 3.1993(7) Mn2Mn4 3.2364(9) Mn 3 O11 2.062(3) Mn3O3 2.092(3) Mn3O16#1 2.222(3) Mn3O15#1 2.223(3) Mn3O10 2.227(3) Mn3O17 2.435(3) Mn3Mn6#1 3.1709(9) Mn3Mn7 3.2202(7) Mn3Mn4#1 3.2354(9) Mn3Mn5 3.2369(9) Mn4O7 2.077(3) Mn4O12#1 2.086(3) Mn4O15 2.225(3) Mn4O5 2.22 6(3) Mn4O17#1 2.235(3) Mn4O18 2.465(3) Mn4Mn3#1 3.2354(9) Mn5O19 1.868(3) Mn5O10 1.909(3) Mn5O13 1.959(3) Mn5O18 2.022(3) Mn5O8 2.106(3) Mn5O17 2.321(3) Mn5Mn1#1 2.9201(9) Mn5Mn6 2.9275(9) Mn5Mn7 3.0635(6) Mn6O19 1.876(3) Mn6O15 1.916(3) Mn6O9 1.950(3) Mn6O16 2.010(3) Mn6O1#1 2.108(3) Mn6O18 2.304(3) Mn6Mn1#1 2.9215(9) Mn6Mn7 3.0412(6) Mn6Mn3#1 3.1709(9) Mn7O16 1.894(3) Mn7O16#1 1.894(3) Mn7O18#1 1.901(3) Mn7O18 1.901(3) Mn7O17 1.905(3) Mn7O17#1 1.905(3)

PAGE 182

182 Mn7Mn1#1 3.0349(6) Mn7Mn6#1 3.0412(6) Mn7Mn5#1 3.0635(6) Fe1 C2 2.035(4) Fe1 C10 2.037(5) Fe1 C3 2.041(5) Fe1 C11 2.041(5) Fe1 C12 2.043(5) Fe1 C9 2.049(5) Fe1 C4 2.050(4) Fe1 C6 2.051(4) Fe1 C5 2.052(5) Fe1 C8 2.054(5) Fe2 C15 2.031(4) F e2 C21 2.043(4) Fe2 C22 2.048(4) Fe2 C25 2.049(5) Fe2 C16 2.049(5) Fe2 C19 2.050(5) Fe2 C23 2.053(4) Fe2 C17 2.053(5) Fe2 C18 2.057(5) Fe2 C24 2.057(4) Fe3 C34 2.038(4) Fe3 C29 2.040(4) Fe3 C38 2.041(5) Fe3 C30 2.043(5) Fe3 C35 2.043(5) Fe3 C 28 2.044(4) Fe3 C31 2.050(5) Fe3 C32 2.051(4) Fe3 C37 2.056(4) Fe3 C36 2.060(5) O19#1Mn1 O5 170.75(12) O19#1Mn1 O2 90.19(12) O5 Mn1 O2 92.06(12) O19#1Mn1 O17#1 89.17(12) O5 Mn1 O17#1 86.68(12) O2 Mn1 O17#1 167.40(12) O19#1Mn1 O14#1 88.22(12) O5 M n1 O14#1 100.14(12) O2 Mn1 O14#1 101.08(12) O17#1Mn1 O14#1 91.48(11) O19#1Mn1 O16#1 80.61(11) O5 Mn1 O16#1 90.38(11) O2 Mn1 O16#1 91.05(11) O17#1Mn1 O16#1 76.44(10) O14#1Mn1 O16#1 163.56(11) O19#1Mn1 Mn5#1 38.65(8) O5 Mn1 Mn5#1 138.80(9) O2 Mn1 Mn5#1 128.75(9) O17#1Mn1 Mn5#1 52.32(8) O14#1Mn1 Mn5#1 79.50(8) O16#1Mn1 Mn5#1 84.26(7) O19#1Mn1 Mn6#1 38.86(9) O5 Mn1 Mn6#1 132.74(9) O2 Mn1 Mn6#1 82.17(9) O17#1Mn1 Mn6#1 89.48(8) O14#1Mn1 Mn6#1 127.05(9) O16#1Mn1 Mn6#1 43.27(7) Mn5#1Mn1Mn6#1 60.15(2) O19#1Mn1 Mn7 82.38(8) O5 Mn1 Mn7 89.29(9) O2 Mn1 Mn7 129.63(9) O17#1Mn1 Mn7 37.88(8) O14#1Mn1 Mn7 128.17(8) O16#1Mn1 Mn7 38.59(7) Mn5#1Mn1Mn7 61.883(17) Mn6#1Mn1Mn7 61.370(17) O19#1Mn1 Mn4 133.36(8) O5 Mn1 Mn4 43.64(8)

PAGE 183

183 O2 Mn1 Mn4 135.71(9) O17#1M n1 Mn4 44.30(8) O14#1Mn1 Mn4 89.77(8) O16#1Mn1 Mn4 89.21(7) Mn5#1Mn1Mn4 95.34(2) Mn6#1Mn1Mn4 124.34(3) Mn7Mn1 Mn4 63.022(16) O19#1Mn1 Mn2 128.87(9) O5 Mn1 Mn2 42.36(9) O2 Mn1 Mn2 87.45(9) O17#1Mn1 Mn2 83.18(8) O14#1Mn1 Mn2 142.20(9) O16#1Mn1 Mn2 48.42(7) Mn5#1Mn1Mn2 123.16(3) Mn6#1Mn1Mn2 90.41(2) Mn7Mn1 Mn2 61.329(16) Mn4Mn1 Mn2 60.69(2) O4 Mn2 O6 107.00(12) O4 Mn2 O18 157.39(12) O6 Mn2 O18 92.99(11) O4 Mn2 O10 90.22(11) O6 Mn2 O10 101.54(11) O18 Mn2 O10 75.19(10) O4 Mn2 O5 101.49(11) O6 Mn2 O5 89.81(11) O18 Mn2 O5 88.75(10) O10 Mn2 O5 160.66(11) O4 Mn2 O16#1 90.39(11) O6 Mn2 O16#1 161.47(11) O18 Mn2 O16#1 71.34(10) O10 Mn2 O16#1 84.34(10) O5 Mn2 O16#1 80.28(10) O4 Mn2 Mn5 126.52(9) O6 Mn2 Mn5 93.08(9) O18 Mn2 Mn5 39.48(7) O10 Mn2 Mn5 36.61( 7) O5 Mn2 Mn5 128.22(8) O16#1Mn2 Mn5 81.15(7) O4 Mn2 Mn3 76.31(9) O6 Mn2 Mn3 145.60(9) O18 Mn2 Mn3 81.23(8) O10 Mn2 Mn3 44.14(7) O5 Mn2 Mn3 123.68(8) O16#1Mn2 Mn3 43.95(7) Mn5Mn2 Mn3 61.251(19) O4 Mn2 Mn7 126.14(9) O6 Mn2 Mn7 126.86(8) O18 Mn2 Mn7 35.46(7) O10 Mn2 Mn7 80.66(7) O5 Mn2 Mn7 80.00(7) O16#1Mn2 Mn7 36.15(6) Mn5Mn2 Mn7 57.620(16) Mn3Mn2 Mn7 60.463(16) O4 Mn2 Mn1 94.54(9) O6 Mn2 Mn1 124.66(9) O18 Mn2 Mn1 82.15(7) O10 Mn2 Mn1 129.31(8) O5 Mn2 Mn1 35.34(7) O16#1Mn2 Mn1 45.27(7) Mn5Mn2 Mn1 113 .91(2) Mn3Mn2 Mn1 88.34(2) Mn7Mn2 Mn1 56.338(15) O4 Mn2 Mn4 144.78(9) O6 Mn2 Mn4 76.39(9) O18 Mn2 Mn4 49.57(8) O10 Mn2 Mn4 124.05(7) O5 Mn2 Mn4 43.36(7) O16#1Mn2 Mn4 85.74(7) Mn5Mn2 Mn4 87.45(2) Mn3Mn2 Mn4 121.09(3) Mn7Mn2 Mn4 60.624(16) Mn1Mn2 Mn4 58.833(19)

PAGE 184

184 O11 Mn3 O3 106.76(12) O11 Mn3 O16#1 156.98(12) O3 Mn3 O16#1 92.82(11) O11 Mn3 O15#1 89.48(11) O3 Mn3 O15#1 101.18(11) O16#1Mn3 O15#1 74.63(10) O11 Mn3 O10 102.25(11) O3 Mn3 O10 91.10(11) O16#1Mn3 O10 89.07(10) O15#1Mn3 O10 159.91(11) O11 Mn3 O17 91.13(11) O3 Mn3 O17 161.53(11) O16#1Mn3 O17 70.83(10) O15#1Mn3 O17 83.13(10) O10 Mn3 O17 80.41(10) O11 Mn3 Mn6#1 125.79(9) O3 Mn3 Mn6#1 92.62(9) O16#1Mn3 Mn6#1 38.99(7) O15#1Mn3 Mn6#1 36.56(8) O10 Mn3 Mn6#1 128.04(8) O17 Mn3 Mn6#1 80.13(7) O11 Mn3 Mn2 146.20(9) O3 Mn3 Mn2 77.41(9) O16#1Mn3 Mn2 49.13(7) O15#1Mn3 Mn2 123.21(8) O10 Mn3 Mn2 44.00(7) O17 Mn3 Mn2 85.19(7) Mn6#1Mn3Mn2 86.69(2) O11 Mn3 Mn7 126.77(9) O3 Mn3 Mn7 126.45(9) O16#1Mn3 Mn7 35.03(7) O15#1Mn3 Mn7 79.81(7) O10 Mn3 Mn7 80.10(7) O17 Mn3 Mn7 36.11(6) Mn6#1Mn3Mn7 56.823(15) Mn2Mn3 Mn7 59.813(17) O11 Mn3 Mn4#1 76.21(9) O3 Mn3 Mn4#1 144.46(9) O16#1Mn3 Mn4#1 80.82(7) O15#1Mn3 Mn4#1 43.36(7) O10 Mn3 Mn4#1 123.45(7) O17 Mn3 Mn4#1 43.65(7) Mn6#1Mn3Mn4#1 60.798(19) Mn2Mn3 Mn4#1 12 0.23(3) Mn7Mn3 Mn4#1 60.420(16) O11 Mn3 Mn5 95.46(9) O3 Mn3 Mn5 125.66(9) O16#1Mn3 Mn5 82.25(7) O15#1Mn3 Mn5 128.51(8) O10 Mn3 Mn5 35.11(7) O17 Mn3 Mn5 45.65(7) Mn6#1Mn3Mn5 113.44(2) Mn2Mn3 Mn5 58.779(19) Mn7Mn3 Mn5 56.644(15) Mn4#1Mn3Mn5 88.34(2) O7 Mn4 O12#1 108.77(12) O7 Mn4 O15 103.87(11) O12#1Mn4 O15 90.57(11) O7 Mn4 O5 89.09(11) O12#1Mn4 O5 102.12(11) O15 Mn4 O5 158.04(11) O7 Mn4 O17#1 155.76(11) O12#1Mn4 O17#1 91.98(11) O15 Mn4 O17#1 87.86(10) O5 Mn4 O17#1 74.01(10) O7 Mn4 O18 90.77(11) O 12#1Mn4 O18 159.90(11) O15 Mn4 O18 79.88(10) O5 Mn4 O18 82.32(10) O17#1Mn4 O18 70.19(10) O7 Mn4 Mn1 125.19(9) O12#1Mn4 Mn1 92.43(9)

PAGE 185

185 O15 Mn4 Mn1 126.49(7) O5 Mn4 Mn1 36.36(7) O17#1Mn4 Mn1 38.65(7) O18 Mn4 Mn1 79.50(7) O7 Mn4 Mn3#1 147.15(9) O12#1Mn4 Mn 3#1 76.35(9) O15 Mn4 Mn3#1 43.31(7) O5 Mn4 Mn3#1 122.27(8) O17#1Mn4 Mn3#1 48.75(7) O18 Mn4 Mn3#1 84.69(7) Mn1Mn4 Mn3#1 85.98(2) O7 Mn4 Mn2 76.06(9) O12#1Mn4 Mn2 145.51(9) O15 Mn4 Mn2 122.13(8) O5 Mn4 Mn2 43.40(8) O17#1Mn4 Mn2 79.73(7) O18 Mn4 Mn2 42.67(7) Mn1Mn4 Mn2 60.474(19) Mn3#1Mn4Mn2 118.68(2) O19 Mn5 O10 170.16(12) O19 Mn5 O13 90.67(12) O10 Mn5 O13 92.14(12) O19 Mn5 O18 88.45(12) O10 Mn5 O18 86.51(12) O13 Mn5 O18 165.76(12) O19 Mn5 O8 88.12(12) O10 Mn5 O8 100.42(12) O13 Mn5 O8 102.93(12) O18 Mn 5 O8 91.24(11) O19 Mn5 O17 80.23(11) O10 Mn5 O17 90.33(11) O13 Mn5 O17 90.20(11) O18 Mn5 O17 75.64(10) O8 Mn5 O17 162.58(11) O19 Mn5 Mn1#1 38.85(8) O10 Mn5 Mn1#1 132.46(9) O13 Mn5 Mn1#1 81.81(9) O18 Mn5 Mn1#1 88.65(8) O8 Mn5 Mn1#1 126.96(8) O17 Mn5 Mn1#1 4 2.96(7) O19 Mn5 Mn6 38.66(9) O10 Mn5 Mn6 137.93(9) O13 Mn5 Mn6 129.30(9) O18 Mn5 Mn6 51.61(8) O8 Mn5 Mn6 79.31(8) O17 Mn5 Mn6 83.52(7) Mn1#1Mn5Mn6 59.95(2) O19 Mn5 Mn7 81.69(8) O10 Mn5 Mn7 89.20(8) O13 Mn5 Mn7 128.62(9) O18 Mn5 Mn7 37.26(8) O8 Mn5 Mn7 12 7.27(8) O17 Mn5 Mn7 38.42(7) Mn1#1Mn5Mn7 60.900(17) Mn6Mn5 Mn7 60.963(16) O19 Mn5 Mn2 132.04(9) O10 Mn5 Mn2 43.96(9) O13 Mn5 Mn2 136.01(9) O18 Mn5 Mn2 43.65(8) O8 Mn5 Mn2 90.32(8) O17 Mn5 Mn2 87.98(7) Mn1#1Mn5Mn2 122.72(3) Mn6Mn5 Mn2 94.13(2) Mn7Mn5 Mn2 61.878(16) O19 Mn5 Mn3 128.71(9) O10 Mn5 Mn3 42.15(8) O13 Mn5 Mn3 86.81(9) O18 Mn5 Mn3 82.68(8) O8 Mn5 Mn3 142.19(8) O17 Mn5 Mn3 48.60(7) Mn1#1Mn5Mn3 90.33(2) Mn6Mn5 Mn3 122.33(3)

PAGE 186

186 Mn7Mn5 Mn3 61.402(17) Mn2Mn5 Mn3 59.970(19) O19 Mn6 O15 170.77(12) O19 Mn6 O9 91.18(12) O15 Mn6 O9 91.28(12) O19 Mn6 O16 88.79(12) O15 Mn6 O16 86.63(12) O9 Mn6 O16 165.64(12) O19 Mn6 O1#1 86.97(12) O15 Mn6 O1#1 101.17(12) O9 Mn6 O1#1 102.40(12) O16 Mn6 O1#1 91.94(11) O19 Mn6 O18 80.33(11) O15 Mn6 O18 90.79(11) O9 Mn6 O18 89.72(11) O16 Mn6 O18 76.12(10) O1#1 Mn6 O18 162.65(11) O19 Mn6 Mn1#1 38.87(8) O15 Mn6 Mn1#1 138.06(9) O9 Mn6 Mn1#1 130.04(9) O16 Mn6 Mn1#1 51.64(8) O1#1 Mn6 Mn1#1 79.12(8) O18 Mn6 Mn1#1 83.58(7) O19 Mn6 Mn5 38.47(8) O15 Mn6 Mn5 133.36(9) O9 Mn6 Mn5 81.79(8) O16 Mn6 Mn5 89.32(8) O1#1 Mn6 Mn5 125.40(9) O18 Mn6 Mn5 43.47(7) Mn1#1Mn6Mn5 59.90(2) O19 Mn6 Mn7 82.21(8) O15 Mn6 Mn7 89.29(8) O9 Mn6 Mn7 128.37(8) O16 Mn6 Mn7 37.49(8) O1#1 Mn6 Mn7 128.04(8) O18 Mn6 Mn7 38.65(7) Mn1#1Mn6Mn7 61.153(16) Mn5Mn6 Mn7 61.727(16) O19 Mn6 Mn3#1 132.76(9) O15 Mn6 Mn3#1 43.70(8) O9 Mn6 Mn3#1 134.92(9) O16 Mn6 Mn3#1 44.07(8) O1#1 Mn6 Mn3#1 91.19(8) O18 Mn6 Mn3#1 88.85(7) Mn1#1Mn6Mn3#1 94.51(2) Mn5Mn6 Mn3#1 124.09(2) Mn7Mn6 Mn3#1 62.404(17) O16 Mn7 O16#1 180.0(2) O16 Mn7 O18#1 90.61(12) O16#1Mn7 O18#1 89.39(12) O16 Mn7 O18 89.39(12) O16#1Mn7 O18 90.61(12) O18#1Mn7 O18 180.0(2) O16 Mn7 O17 89.26(12) O16#1Mn7 O17 90.74(12) O18#1Mn7 O17 90.76(12) O18 Mn7 O17 89.24(12) O16 Mn7 O17#1 90.74(12) O16#1Mn7 O17#1 89.26(12) O1 8#1Mn7 O17#1 89.24(12) O18 Mn7 O17#1 90.76(12) O17 Mn7 O17#1 180.00(17) O16 Mn7 Mn1#1 49.20(9) O16#1Mn7 Mn1#1 130.80(9) O18#1Mn7 Mn1#1 92.42(8) O18 Mn7 Mn1#1 87.58(8) O17 Mn7 Mn1#1 40.10(8) O17#1Mn7 Mn1#1 139.90(8) O16 Mn7 Mn1 130.80(9) O16#1Mn7 Mn1 4 9.20(9) O18#1Mn7 Mn1 87.58(8) O18 Mn7 Mn1 92.42(8)

PAGE 187

187 O17 Mn7 Mn1 139.90(8) O17#1Mn7 Mn1 40.10(8) Mn1#1Mn7Mn1 180.00(2) O16 Mn7 Mn6#1 139.78(8) O16#1Mn7 Mn6#1 40.22(8) O18#1Mn7 Mn6#1 49.20(8) O18 Mn7 Mn6#1 130.80(8) O17 Mn7 Mn6#1 92.26(8) O17#1Mn7 Mn6# 1 87.74(8) Mn1#1Mn7Mn6#1 122.523(16) Mn1Mn7 Mn6#1 57.477(16) O16 Mn7 Mn6 40.22(8) O16#1Mn7 Mn6 139.78(8) O18#1Mn7 Mn6 130.80(8) O18 Mn7 Mn6 49.20(8) O17 Mn7 Mn6 87.74(8) O17#1Mn7 Mn6 92.26(8) Mn1#1Mn7Mn6 57.478(16) Mn1Mn7 Mn6 122.522(16) Mn6#1Mn7Mn6 180.00(3) O16 Mn7 Mn5#1 92.48(8) O16#1Mn7 Mn5#1 87.52(8) O18#1Mn7 Mn5#1 40.09(9) O18 Mn7 Mn5#1 139.91(9) O17 Mn7 Mn5#1 130.80(8) O17#1Mn7 Mn5#1 49.20(8) Mn1#1Mn7Mn5#1 122.784(16) Mn1Mn7 Mn5#1 57.216(16) Mn6#1Mn7Mn5#1 57.310(17) Mn6Mn7 Mn5#1 1 22.690(17) O16 Mn7 Mn5 87.52(8) O16#1Mn7 Mn5 92.48(8) O18#1Mn7 Mn5 139.91(9) O18 Mn7 Mn5 40.09(9) O17 Mn7 Mn5 49.20(8) O17#1Mn7 Mn5 130.80(8) Mn1#1Mn7Mn5 57.216(16) Mn1Mn7 Mn5 122.784(16) Mn6#1Mn7Mn5 122.690(17) Mn6Mn7 Mn5 57.310(17) Mn5#1Mn7Mn5 180.00(3) Table C 2 Selected bond lengths [] and angles [ ] for [Mn8O4(fdc)6(DMF)2(H2O)2] 4 ______________________________________________________________________ Mn1O15 2.075(3) Mn1O11 2.090(3) Mn1O5 2.099(3) Mn1O2 2.122(3) Mn1O14 2.168(3) Mn2O16#1 1.921(3) Mn2O15#1 1.921(2) Mn2O3 1.985(3) Mn2O1 1.987(3) Mn2O12#1 2.136(3) Mn2O15 2.219(2) Mn2Mn3#1 2.9096(8) Mn2Mn3 3.0669(8) Mn2Mn2#1 3.1150(11) Mn3O15 1.930(3) Mn3O16 1.979(3) Mn3O6 2.003(3) Mn3O8 2.010(3) M n3 O9 2.084(3) Mn3O16#1 2.138(3) Mn3Mn2#1 2.9095(8) Mn3Mn3#1 3.0926(12) Fe1 C2#1 2.033(4) Fe1 C2 2.033(4)

PAGE 188

188 Fe1 C3#1 2.038(5) Fe1 C3 2.038(5) Fe1 C6 2.046(4) Fe1 C6#1 2.046(4) Fe1 C5 2.054(5) Fe1 C5#1 2.054(5) Fe1 C4#1 2.054(5) Fe1 C4 2.054( 5) Fe2 C14 2.026(4) Fe2 C9 2.034(4) Fe2 C18 2.036(4) Fe2 C8 2.040(4) Fe2 C12 2.043(4) Fe2 C15 2.044(4) Fe2 C17 2.044(4) Fe2 C10 2.048(4) Fe2 C11 2.055(5) Fe2 C16 2.068(5) Fe3 O10#1 2.058(3) Fe3 O16#1 2.065(3) Fe3 O7 2.096(3) Fe3 O4 2.118(3) F e3 O13 2.173(3) Fe4 C24 2.028(5) Fe4 C24#1 2.028(5) Fe4 C20 2.034(6) Fe4 C20#1 2.034(6) Fe4 C21#1 2.043(7) Fe4 C21 2.043(7) Fe4 C23 2.045(6) Fe4 C23#1 2.045(6) Fe4 C22 2.058(7) Fe4 C22#1 2.058(7) Fe5 C26 2.021(4) Fe5 C27 2.034(4) Fe5 C34 2.03 9(5) Fe5 C35 2.040(5) Fe5 C30 2.047(5) Fe5 C33 2.048(5) Fe5 C32 2.049(4) Fe5 C36 2.053(5) Fe5 C29 2.064(5) Fe5 C28 2.064(4) O15 Mn1 O11 95.69(10) O15 Mn1 O5 94.58(10) O11 Mn1 O5 116.63(11) O15 Mn1 O2 91.18(10) O11 Mn1 O2 128.48(11) O5 Mn1 O2 113.57 (11) O15 Mn1 O14 175.52(12) O11 Mn1 O14 82.83(13) O5 Mn1 O14 89.86(13) O2 Mn1 O14 86.47(12) O16#1Mn2 O15#1 83.04(11) O16#1Mn2 O3 95.46(11) O15#1Mn2 O3 175.32(11) O16#1Mn2 O1 173.53(11) O15#1Mn2 O1 92.52(11) O3 Mn2 O1 88.58(11) O16#1Mn2 O12#1 94.87(11 ) O15#1Mn2 O12#1 95.43(11) O3 Mn2 O12#1 89.11(11) O1 Mn2 O12#1 90.23(10) O16#1Mn2 O15 81.92(10) O15#1Mn2 O15 82.37(10) O3 Mn2 O15 93.04(10) O1 Mn2 O15 92.85(10) O12#1Mn2 O15 176.29(10) O16#1Mn2 Mn3#1 42.52(8) O15#1Mn2 Mn3#1 41.04(7) O3 Mn2 Mn3#1 137. 90(8)

PAGE 189

189 O1 Mn2 Mn3#1 133.49(8) O12#1Mn2 Mn3#1 91.78(8) O15 Mn2 Mn3#1 84.60(7) O16#1Mn2 Mn3 43.65(8) O15#1Mn2 Mn3 85.51(8) O3 Mn2 Mn3 90.35(8) O1 Mn2 Mn3 131.52(7) O12#1Mn2 Mn3 138.22(8) O15 Mn2 Mn3 38.82(6) Mn3#1Mn2Mn3 62.26(2) O16#1Mn2 Mn2#1 83.96(8) O15#1Mn2 Mn2#1 44.97(7) O3 Mn2 Mn2#1 130.53(8) O1 Mn2 Mn2#1 89.57(7) O12#1Mn2 Mn2#1 140.34(8) O15 Mn2 Mn2#1 37.72(6) Mn3#1Mn2Mn2#1 61.092(18) Mn3Mn2 Mn2#1 56.147(18) O15 Mn3 O16 81.33(10) O15 Mn3 O6 94.80(11) O16 Mn3 O6 174.97(11) O15 Mn3 O8 173.76(1 1) O16 Mn3 O8 93.44(11) O6 Mn3 O8 90.23(11) O15 Mn3 O9 96.80(10) O16 Mn3 O9 95.86(11) O6 Mn3 O9 87.75(11) O8 Mn3 O9 87.07(11) O15 Mn3 O16#1 83.90(10) O16 Mn3 O16#1 82.29(11) O6 Mn3 O16#1 94.15(10) O8 Mn3 O16#1 92.06(10) O9 Mn3 O16#1 177.92(11) O15 Mn3 Mn2# 1 40.82(7) O16 Mn3 Mn2#1 41.00(8) O6 Mn3 Mn2#1 135.46(8) O8 Mn3 Mn2#1 134.30(8) O9 Mn3 Mn2#1 93.39(7) O16#1Mn3 Mn2#1 85.85(7) O15 Mn3 Mn2 46.13(7) O16 Mn3 Mn2 84.39(8) O6 Mn3 Mn2 90.62(8) O8 Mn3 Mn2 130.30(8) O9 Mn3 Mn2 142.61(8) O16#1Mn3 Mn2 38.34(7) Mn 2#1Mn3Mn2 62.76(2) O15 Mn3 Mn3#1 84.65(7) O16 Mn3 Mn3#1 43.28(8) O6 Mn3 Mn3#1 133.46(8) O8 Mn3 Mn3#1 89.21(8) O9 Mn3 Mn3#1 138.65(8) O16#1Mn3 Mn3#1 39.40(7) Mn2#1Mn3Mn3#1 61.368(19) Mn2Mn3 Mn3#1 56.375(17) Table C 3 Selected bond lengths [] and angles [ ] for [Mn8O4(fdc)6(DMF)4] 5 ______________________________________________________________________ Mn1O3 1.8848(19) Mn1O2 1.895(2) Mn1O5 1.973(2) Mn1O24 1.981(2) Mn1O16 2.167(2) Mn1O1 2.295(2) Mn1Mn3 2.8308(7) Mn1Mn4 3.11 44(7) Mn1Mn2 3.2053(7) Mn2O4 1.884(2)

PAGE 190

190 Mn2O1 1.9037(19) Mn2O9 1.961(2) Mn2O21 1.983(2) Mn2O20 2.145(2) Mn2O3 2.340(2) Mn2Mn4 2.8327(7) Mn2Mn3 3.1279(8) Mn3O3 1.872(2) Mn3O2 1.903(2) Mn3O12 1.938(2) Mn3O25 1.980(2) Mn3O13 2.149(2) Mn3O4 2.310(2) Mn3Mn4 3.1490(7) Mn4O1 1.885(2) Mn4O4 1.8918(19) Mn4O8 1.962(2) Mn4O28 1.987(2) Mn4O17 2.149(2) Mn4O2 2.267(2) Mn5O3 2.0741(19) Mn5O15 2.086(2) Mn5O22 2.111(2) Mn5O11 2.128(2) Mn5O30 2.176(2) Mn6O1 2.075(2) Mn6O 19 2.085(2) Mn6O23 2.109(2) Mn6O7 2.119(2) Mn6O29 2.175(3) Mn7O2 2.065(2) Mn7O14 2.079(2) Mn7O6 2.111(2) Mn7O27 2.123(2) Mn7O31 2.217(2) Mn8O18 2.079(2) Mn8O4 2.0941(19) Mn8O10 2.108(2) Mn8O26 2.117(2) Mn8O32 2.176(2) Fe1 C8 2.0 31(3) Fe1 C2 2.035(3) Fe1 C3 2.043(3) Fe1 C9 2.046(3) Fe1 C12 2.054(3) Fe1 C4 2.058(3) Fe1 C6 2.062(3) Fe1 C10 2.064(3) Fe1 C5 2.074(3) Fe1 C11 2.078(3) Fe2 C21 2.031(4) Fe2 C20 2.031(3) Fe2 C14 2.043(3) Fe2 C18 2.045(3) Fe2 C22 2.046(4) Fe2 C15 2.052(4) Fe2 C24 2.056(4) Fe2 C16 2.059(4) Fe2 C17 2.060(3) Fe2 C23 2.077(4) Fe3 C26 2.028(3) Fe3 C29 2.028(4) Fe3 C33 2.030(3) Fe3 C35 2.034(3) Fe3 C27 2.034(3) Fe3 C30 2.035(4) Fe3 C36 2.039(3) Fe3 C34 2.042(4) Fe3 C32 2.042(3) Fe3 C28 2.045(4) Fe4 C44 2.036(3) Fe4 C41 2.045(3)

PAGE 191

191 Fe4 C48 2.045(3) Fe4 C42 2.046(3) Fe4 C45 2.047(3) Fe4 C40 2.048(4) Fe4 C47 2.051(3) Fe4 C38 2.051(3) Fe4 C39 2.053(3) Fe4 C46 2.058(4) Fe5 C56 2.032(3) Fe5 C51 2.034(3) Fe5 C57 2.037(3) Fe5 C52 2.0 42(3) Fe5 C50 2.044(3) Fe5 C54 2.048(3) Fe5 C60 2.055(3) Fe5 C53 2.055(3) Fe5 C58 2.058(3) Fe5 C59 2.060(3) Fe6 C68 2.032(3) Fe6 C72 2.036(4) Fe6 C62 2.037(3) Fe6 C69 2.041(3) Fe6 C70 2.048(3) Fe6 C66 2.050(3) Fe6 C63 2.051(3) Fe6 C65 2.054(4) Fe6 C64 2.059(4) Fe6 C71 2.062(4) O3 Mn1 O2 82.28(9) O3 Mn1 O5 173.33(9) O2 Mn1 O5 96.01(8) O3 Mn1 O24 89.56(9) O2 Mn1 O24 171.26(8) O5 Mn1 O24 91.74(9) O3 Mn1 O16 96.80(8) O2 Mn1 O16 95.11(8) O5 Mn1 O16 89.77(8) O24 Mn1 O16 88.94(8) O3 Mn1 O1 81.98(8) O2 Mn1 O1 82.74(8) O5 Mn1 O1 91.42(8) O24 Mn1 O1 93.06(8) O16 Mn1 O1 177.64(8) O3 Mn1 Mn3 40.94(6) O2 Mn1 Mn3 41.92(6) O5 Mn1 Mn3 137.90(6) O24 Mn1 Mn3 130.31(6) O16 Mn1 Mn3 92.51(6) O1 Mn1 Mn3 85.24(5) O3 Mn1 Mn4 84.98(6) O2 Mn1 Mn4 46.31(6) O5 Mn1 Mn4 89.18(6) O24 Mn1 Mn4 130.08(6) O16 Mn1 Mn4 140.98(6) O1 Mn1 Mn4 37.03(5) Mn3Mn1 Mn4 63.757(17) O3 Mn1 Mn2 46.29(6) O2 Mn1 Mn2 83.37(6) O5 Mn1 Mn2 127.18(6) O24 Mn1 Mn2 88.73(6) O16 Mn1 Mn2 143.02(6) O1 Mn1 Mn2 35.90(5) Mn3Mn1 Mn2 62.058(17) Mn4Mn1 Mn2 5 3.238(14) O4 Mn2 O1 82.28(8) O4 Mn2 O9 94.04(9) O1 Mn2 O9 174.06(9) O4 Mn2 O21 173.25(8) O1 Mn2 O21 92.85(8) O9 Mn2 O21 90.42(8) O4 Mn2 O20 98.63(8)

PAGE 192

192 O1 Mn2 O20 95.68(8) O9 Mn2 O20 89.47(8) O21 Mn2 O20 86.46(8) O4 Mn2 O3 83.16(8) O1 Mn2 O3 80.37(8) O9 Mn2 O 3 94.60(8) O21 Mn2 O3 91.44(8) O20 Mn2 O3 175.44(8) O4 Mn2 Mn4 41.49(6) O1 Mn2 Mn4 41.37(6) O9 Mn2 Mn4 135.43(7) O21 Mn2 Mn4 134.12(6) O20 Mn2 Mn4 94.08(6) O3 Mn2 Mn4 84.47(5) O4 Mn2 Mn3 47.27(6) O1 Mn2 Mn3 84.19(7) O9 Mn2 Mn3 89.89(7) O21 Mn2 Mn3 127.80(6) O20 Mn2 Mn3 145.73(6) O3 Mn2 Mn3 36.59(5) Mn4Mn2 Mn3 63.553(16) O4 Mn2 Mn1 83.70(6) O1 Mn2 Mn1 44.98(6) O9 Mn2 Mn1 130.17(6) O21 Mn2 Mn1 89.55(6) O20 Mn2 Mn1 140.22(6) O3 Mn2 Mn1 35.60(5) Mn4Mn2 Mn1 61.736(15) Mn3Mn2 Mn1 53.082(15) O3 Mn3 O2 82.38(8) O3 Mn3 O12 93.53(9) O2 Mn3 O12 174.96(10) O3 Mn3 O25 173.61(9) O2 Mn3 O25 92.56(9) O12 Mn3 O25 91.31(9) O3 Mn3 O13 98.86(8) O2 Mn3 O13 96.17(8) O12 Mn3 O13 87.35(8) O25 Mn3 O13 85.53(9) O3 Mn3 O4 84.26(8) O2 Mn3 O4 81.97(8) O12 Mn3 O4 94.73(8) O25 Mn3 O4 9 1.19(8) O13 Mn3 O4 176.16(8) O3 Mn3 Mn1 41.28(6) O2 Mn3 Mn1 41.69(6) O12 Mn3 Mn1 134.59(7) O25 Mn3 Mn1 134.10(6) O13 Mn3 Mn1 94.59(6) O4 Mn3 Mn1 86.27(5) O3 Mn3 Mn2 48.17(6) O2 Mn3 Mn2 85.45(7) O12 Mn3 Mn2 89.63(7) O25 Mn3 Mn2 127.78(6) O13 Mn3 Mn2 146.63( 6) O4 Mn3 Mn2 36.80(5) Mn1Mn3 Mn2 64.860(17) O3 Mn3 Mn4 84.18(6) O2 Mn3 Mn4 45.50(6) O12 Mn3 Mn4 131.36(7) O25 Mn3 Mn4 89.49(6) O13 Mn3 Mn4 141.12(6) O4 Mn3 Mn4 36.64(5) Mn1Mn3 Mn4 62.507(17) Mn2Mn3 Mn4 53.652(15) O1 Mn4 O4 82.56(8) O1 Mn4 O8 96.64(9) O 4 Mn4 O8 176.70(9) O1 Mn4 O28 171.61(9) O4 Mn4 O28 90.86(9) O8 Mn4 O28 89.64(9) O1 Mn4 O17 96.64(8)

PAGE 193

193 O4 Mn4 O17 95.62(8) O8 Mn4 O17 87.65(8) O28 Mn4 O17 89.15(8) O1 Mn4 O2 83.72(8) O4 Mn4 O2 83.39(8) O8 Mn4 O2 93.34(8) O28 Mn4 O2 90.39(8) O17 Mn4 O2 178.90( 8) O1 Mn4 Mn2 41.86(6) O4 Mn4 Mn2 41.28(6) O8 Mn4 Mn2 138.28(7) O28 Mn4 Mn2 132.07(6) O17 Mn4 Mn2 92.69(5) O2 Mn4 Mn2 86.87(5) O1 Mn4 Mn1 47.14(6) O4 Mn4 Mn1 86.20(6) O8 Mn4 Mn1 90.90(6) O28 Mn4 Mn1 127.50(6) O17 Mn4 Mn1 143.32(6) O2 Mn4 Mn1 37.19(5) Mn2M n4 Mn1 65.026(15) O1 Mn4 Mn3 83.86(6) O4 Mn4 Mn3 46.78(6) O8 Mn4 Mn3 130.00(6) O28 Mn4 Mn3 87.84(6) O17 Mn4 Mn3 142.18(6) O2 Mn4 Mn3 36.79(5) Mn2Mn4 Mn3 62.795(17) Mn1Mn4 Mn3 53.736(15) O3 Mn5 O15 95.65(8) O3 Mn5 O22 90.63(8) O15 Mn5 O22 147.99(9) O3 Mn5 O11 93.76(8) O15 Mn5 O11 111.49(9) O22 Mn5 O11 99.32(8) O3 Mn5 O30 161.02(9) O15 Mn5 O30 82.72(8) O22 Mn5 O30 81.36(8) O11 Mn5 O30 104.47(9) O1 Mn6 O19 96.80(8) O1 Mn6 O23 91.59(8) O19 Mn6 O23 125.88(8) O1 Mn6 O7 92.96(8) O19 Mn6 O7 114.64(8) O23 Mn6 O7 1 18.17(8) O1 Mn6 O29 176.49(9) O19 Mn6 O29 85.33(9) O23 Mn6 O29 84.90(9) O7 Mn6 O29 88.66(9) O2 Mn7 O14 100.99(8) O2 Mn7 O6 89.33(8) O14 Mn7 O6 122.85(9) O2 Mn7 O27 92.81(8) O14 Mn7 O27 125.74(9) O6 Mn7 O27 109.37(8) O2 Mn7 O31 169.98(9) O14 Mn7 O31 87.49(9) O6 Mn7 O31 81.53(9) O27 Mn7 O31 86.40(8) O18 Mn8 O4 98.65(8) O18 Mn8 O10 114.77(9) O4 Mn8 O10 91.57(8) O18 Mn8 O26 138.46(9) O4 Mn8 O26 89.82(8) O10 Mn8 O26 105.48(9) O18 Mn8 O32 88.88(9) O4 Mn8 O32 171.09(9) O10 Mn8 O32 89.57(9) O26 Mn8 O32 81.36(9)

PAGE 194

194 T able C 4 Selected b ond lengths [] and angles [ ] for [Fe7O3(OMe)(fdc)6(MeOH)3]3+[FeCl4]2Cl, 6 ______________________________________________________________________ Fe1 O3 1.959(3) Fe1 O18 1.973(3) Fe1 O24 2.004(3) Fe1 O25 2.008(3) Fe1 O22 2.0 28(3) Fe1 O31 2.051(3) Fe2 O13 1.966(3) Fe2 O26 1.990(3) Fe2 O23 2.009(3) Fe2 O3 2.032(3) Fe2 O2 2.073(3) Fe2 O4 2.079(3) Fe3 O12 1.962(3) Fe3 O19 1.985(3) Fe3 O1 1.994(3) Fe3 O6 2.011(3) Fe3 O8 2.020(3) Fe3 O29 2.061(3) Fe4 O21 1.953(3) Fe4 O20 1.964(3) Fe4 O5 1.988(3) Fe4 O1 2.024(3) Fe4 O4 2.077(3) Fe4 O3 2.095(3) Fe5 C26 2.014(5) Fe5 C27 2.023(5) Fe5 C30 2.025(6) Fe5 C33 2.031(5) Fe5 C32 2.037(5) Fe5 C28 2.041(6) Fe5 C29 2.045(5) Fe5 C34 2.047(5) Fe5 C35 2.048(5) Fe5 C36 2.0 52(5) Fe6 O15 1.949(3) Fe6 O7 1.953(3) Fe6 O10 1.973(3) Fe6 O2 2.011(3) Fe6 O1 2.042(3) Fe6 O4 2.101(3) Fe7 O27 1.980(3) Fe7 O11 1.980(3) Fe7 O17 1.982(3) Fe7 O1 2.063(3) Fe7 O3 2.075(3) Fe7 O2 2.079(3) Fe8 C8 2.021(5) Fe8 C12 2.023(5) Fe8 C3 2.035(5) Fe8 C9 2.037(5) Fe8 C4 2.040(5) Fe8 C2 2.041(5) Fe8 C10 2.055(5) Fe8 C6 2.059(5) Fe8 C11 2.062(5) Fe8 C5 2.062(5) Fe9 C14 2.020(5) Fe9 C21 2.023(5) Fe9 C20 2.027(5) Fe9 C24 2.033(5) Fe9 C18 2.034(5) Fe9 C15 2.039(5) Fe9 C23 2.044(5) Fe9 C22 2.045(5) Fe9 C17 2.048(5) Fe9 C16 2.056(5)

PAGE 195

195 Fe10 O28 1.972(4) Fe10 O2 1.985(3) Fe10 O9 1.995(3) Fe10 O14 2.009(3) Fe10 O16 2.042(3) Fe10 O30 2.076(3) Fe11 C44 2.022(5) Fe11 C39 2.024(5) Fe11 C45 2.028(5) Fe11 C48 2.037(5) Fe11 C38 2. 040(5) Fe11 C42 2.043(6) Fe11 C41 2.058(5) Fe11 C40 2.060(5) Fe11 C47 2.068(5) Fe11 C46 2.072(6) Fe12 C50 2.033(5) Fe12 C54 2.039(5) Fe12 C56 2.040(5) Fe12 C57 2.042(5) Fe12 C51 2.047(6) Fe12 C60 2.055(6) Fe12 C59 2.057(6) Fe12 C52 2.058(6) Fe 12 C53 2.065(6) Fe12 C58 2.068(5) Fe13 C62 2.019(5) Fe13 C68 2.028(5) Fe13 C63 2.030(5) Fe13 C72 2.044(5) Fe13 C69 2.045(6) Fe13 C66 2.046(6) Fe13 C71 2.055(6) Fe13 C64 2.057(6) Fe13 C70 2.059(5) Fe13 C65 2.063(6) Fe14 Cl3 2.155(3) Fe14 Cl2 2 .159(3) Fe14 Cl1 2.173(2) Fe14 Cl4' 2.210(4) Fe14 Cl2' 2.228(4) Fe14 Cl3' 2.231(3) Fe14 Cl4 2.255(3) O3 Fe1 O18 92.92(13) O3 Fe1 O24 98.26(13) O18 Fe1 O24 168.80(14) O3 Fe1 O25 94.97(13) O18 Fe1 O25 95.19(13) O24 Fe1 O25 83.30(13) O3 Fe1 O22 95.15(12 ) O18 Fe1 O22 95.55(13) O24 Fe1 O22 84.08(13) O25 Fe1 O22 164.84(13) O3 Fe1 O31 175.52(13) O18 Fe1 O31 82.60(14) O24 Fe1 O31 86.22(14) O25 Fe1 O31 85.68(14) O22 Fe1 O31 85.12(13) O13 Fe2 O26 92.20(13) O13 Fe2 O23 85.99(13) O26 Fe2 O23 96.45(12) O13 Fe2 O3 174.55(12) O26 Fe2 O3 92.36(12) O23 Fe2 O3 96.45(13) O13 Fe2 O2 93.42(13) O26 Fe2 O2 96.51(11) O23 Fe2 O2 167.04(12) O3 Fe2 O2 83.11(12) O13 Fe2 O4 94.27(12) O26 Fe2 O4 172.49(12) O23 Fe2 O4 87.80(11)

PAGE 196

196 O3 Fe2 O4 80.99(11) O2 Fe2 O4 79.32(11) O12 Fe3 O19 94. 56(14) O12 Fe3 O1 92.61(12) O19 Fe3 O1 95.30(13) O12 Fe3 O6 170.22(13) O19 Fe3 O6 85.60(13) O1 Fe3 O6 97.11(12) O12 Fe3 O8 94.64(14) O19 Fe3 O8 164.33(13) O1 Fe3 O8 96.93(13) O6 Fe3 O8 83.19(12) O12 Fe3 O29 84.53(13) O19 Fe3 O29 83.22(14) O1 Fe3 O29 176.66 (13) O6 Fe3 O29 85.78(13) O8 Fe3 O29 85.06(13) O21 Fe4 O20 90.77(14) O21 Fe4 O5 88.63(13) O20 Fe4 O5 98.44(13) O21 Fe4 O1 173.31(12) O20 Fe4 O1 93.84(13) O5 Fe4 O1 95.47(12) O21 Fe4 O4 93.64(13) O20 Fe4 O4 174.86(12) O5 Fe4 O4 84.32(12) O1 Fe4 O4 81.55(12) O21 Fe4 O3 91.51(12) O20 Fe4 O3 97.69(12) O5 Fe4 O3 163.86(13) O1 Fe4 O3 83.08(11) O4 Fe4 O3 79.56(11) C26 Fe5 C27 41.23(19) C26 Fe5 C30 40.86(18) C27 Fe5 C30 69.3(2) C26 Fe5 C33 128.1(2) C27 Fe5 C33 108.0(2) C30 Fe5 C33 165.8(2) C26 Fe5 C32 109.6(2) C27 Fe5 C32 120.8(2) C30 Fe5 C32 127.5(2) C33 Fe5 C32 40.88(17) C26 Fe5 C28 68.5(2) C27 Fe5 C28 40.4(2) C30 Fe5 C28 69.1(2) C33 Fe5 C28 118.6(3) C32 Fe5 C28 154.1(3) C26 Fe5 C29 67.9(2) C27 Fe5 C29 67.8(2) C30 Fe5 C29 40.9(2) C33 Fe5 C29 152.3(3) C32 Fe5 C29 1 65.0(3) C28 Fe5 C29 40.3(2) C26 Fe5 C34 164.5(2) C27 Fe5 C34 125.5(2) C30 Fe5 C34 152.6(2) C33 Fe5 C34 40.32(18) C32 Fe5 C34 68.3(2) C28 Fe5 C34 106.2(2) C29 Fe5 C34 118.2(2) C26 Fe5 C35 154.7(2) C27 Fe5 C35 162.3(2) C30 Fe5 C35 119.2(2) C33 Fe5 C35 67.7(2 ) C32 Fe5 C35 68.0(2) C28 Fe5 C35 125.1(3) C29 Fe5 C35 107.5(2) C34 Fe5 C35 40.18(19) C26 Fe5 C36 121.3(2) C27 Fe5 C36 156.0(2) C30 Fe5 C36 108.2(2) C33 Fe5 C36 68.1(2)

PAGE 197

197 C32 Fe5 C36 40.61(17) C28 Fe5 C36 162.9(3) C29 Fe5 C36 126.9(3) C34 Fe5 C36 67.9(2) C35 Fe5 C36 40.27(18) O15 Fe6 O7 86.30(13) O15 Fe6 O10 97.89(13) O7 Fe6 O10 89.62(12) O15 Fe6 O2 96.59(12) O7 Fe6 O2 176.06(14) O10 Fe6 O2 92.63(12) O15 Fe6 O1 165.44(13) O7 Fe6 O1 92.29(12) O10 Fe6 O1 96.58(13) O2 Fe6 O1 84.25(12) O15 Fe6 O4 85.28(12) O7 Fe6 O4 97.39(12) O10 Fe6 O4 172.50(12) O2 Fe6 O4 80.23(11) O1 Fe6 O4 80.53(12) O27 Fe7 O11 88.87(12) O27 Fe7 O17 90.41(14) O11 Fe7 O17 89.07(13) O27 Fe7 O1 170.76(14) O11 Fe7 O1 88.93(12) O17 Fe7 O1 98.54(12) O27 Fe7 O3 99.89(12) O11 Fe7 O3 171.10(12) O17 Fe7 O3 89.30(13) O1 Fe7 O3 82.65(11) O27 Fe7 O2 89.49(13) O11 Fe7 O2 99.86(13) O17 Fe7 O2 171.06(13) O1 Fe7 O2 82.04(11) O3 Fe7 O2 81.92(11)

PAGE 198

198 APPENDIX D BVS OPTIMIZATION DATA Figure D 1. Exemplified flow chart of chopt algorithm

PAGE 199

199 Ti V Ti Ti Cr Mn Fe Co Ni Cu Figure D 2. Histograms for the valence errors ( ) before and after VNS optimization. Each metal is analyzed independently (blue = before optimization, red = after optimization).

PAGE 200

200 Table D 1 R0 opt values () obtained for titanium with bopt = 0.447. Donor/Valence 2 3 4 O 1.597 1.841 1.771 N 1.908 1.193 P 1.583 2.318 2.348 S 1.713 1.985 2.180 F 1.245 1.673 Cl 2.002 1.963 2.148 Br 2.194 2.172 2.305 I 2.731 2.686 Table D 2 R0 opt values () obtained for vanadium with bopt = 0.414. Donor/Valence 2 3 4 5 O 1.6 15 1.718 1.764 1.800 N 1.630 1.733 1.033 P 1.836 2.161 S 1.434 1.821 1.621 2.076 F 1.612 1.687 1.748 Cl 2.072 1.917 1.762 1.944 Br 2.231 2.301 2.043 I 2.123 Table D 3 R0 opt values () obtained for chromium with bopt = 0.426. Donor/Vale nce 1 2 3 4 5 6 O 1.207 1.589 1.676 1.683 1.733 1.819 N 1.276 1.657 1.431 1.771 1.830 1.837 P 1.849 2.219 2.166 2.204 2.212 S 1.646 2.114 2.013 2.142 2.194 2.264 F 1.148 1.217 1.590 1.702 1.818 Cl 1.589 1.573 2.026 1.934 2.009 2.074 Br 1.564 2.255 I 2.040 2.728 1.193 Table D 4 R0 opt values () obtained for manganese with bopt = 0.617. Donor/Valence 1 2 3 4 5 6 O 1.198 1.493 1.570 1.648 1.708 N 1.268 1.571 1.650 1.713 1.791 1.899 P 1.659 1.956 1.522 S 1.639 1.941 2.066 2.067 F 1.118 1.460 1.508 Cl 1.561 1.876 1.975 2.020 Br 1.728 2.079 2.121 I 2.016 2.317 2.453

PAGE 201

201 Table D 5 R0 opt values () obtained for iron with bopt = 0.630. Donor/Valence 2 3 4 5 O 1.429 1.571 1.638 1.796 N 1.496 1.659 1.657 1.847 P 1.9 41 2.151 2.296 2.296 S 1.962 1.828 1.930 2.004 F 1.223 1.399 Cl 1.870 2.001 2.144 Br 1.990 2.152 2.219 2.408 I 2.288 2.480 2.588 2.738 Table D 6 R0 opt values () obtained for cobalt with bopt = 0.449. Donor/Valence 1 2 3 4 O 1.390 1.599 1.51 2 N 1.590 1.664 1.722 1.670 P 1.589 1.593 1.777 S 1.129 1.485 1.380 F 1.382 1.572 1.680 Cl 1.236 1.375 Br 1.759 2.020 1.712 I 1.798 1.762 1.628 Table D 7 R0 opt values () obtained for cobalt with bopt = 0.376. Donor/Valence 1 2 3 4 O 1.546 1.642 2.059 1.986 N 1.361 1.698 1.532 1.747 P 1.658 1.718 1.712 1.973 S 1.674 2.011 2.097 2.154 F 1.505 1.587 1.637 Cl 1.842 2.008 1.908 2.050 Br 1.959 2.166 2.141 2.306 I 2.183 2.367 2.472 Table D 8 R0 opt values () obtained for copper with bopt = 0.433. Donor/Valence 1 2 3 O 1.421 1.603 1.653 N 1.454 1.657 1.678 P 1.685 2.098 S 1.726 1.921 2.047 F 1.311 1.529 Cl 1.755 1.955 2.050 Br 1.891 2.089 2.173 I 2.071 2.238 2.379

PAGE 202

202 LIST OF REFERENCES (1) Huheey, J. E. Inorganic Chem istry. 3rd Ed ; 3rd ed.; Harper and Row: New York, 1983. (2) Mingos, D. M. P.; Wales, D. J. Introduction to cluster chemistry Prentice Hall: Englewood Cliffs, N.J., 1990. (3) Crabtree, R. H. The Organometallic Chemistry of the Transition Metals ; 5th ed.; John Wiley and Sons, Inc., 2009. (4) Mukerjee, S.; Skogerson, K.; DeGala, S.; Caradonna, J. P. Inorg. Chim. Acta FIELD Full Journal Title:Inorganica Chimica Acta 2000, 297, 313. (5) Dyson, P. J.; Mcindoe, J. S.; Mcindoe, J. S. Transition Metal Carbonyl C luster Chemistry ; Gordon and Breach Science Publishers, 2000. (6) Christou, G. Characterization of Paramagnetic Molecules Gainesville, 2005. (7) Cotton, F. A.; Wilkinson, G. Advanced Inorganic Chemistry ; 5th ed.; John Wiley & Sons, 1988. (8) CCDC, Cambr idge Crystallographic Data Centre. 2010, http://www.ccdc.cam.ac.uk/. (9) Ivarsson, G. J. M. Acta Chemica Scandinavica, Series A: Physical and Inorganic Chemistry 1979, A33 323. (10) Charlot, M. F.; Jeannin, S. ; Jeannin, Y.; Kahn, O.; LucreceAbaul, J.; MartinFrere, J. Inorganic Chemistry 1979, 18, 1675. (11) Chaudhuri, P.; Guttmann, M.; Ventur, D.; Wieghardt, K.; Nuber, B.; Weiss, J. Journal of the Chemical Society, Chemical Communications 1985, 1618. (12) Z ang, Y.; Pan, G.; Que, L., Jr.; Fox, B. G.; Munck, E. Journal of the American Chemical Society 1994 116, 3653. (13) Zheng, H.; Zang, Y.; Dong, Y.; Young, V. G., Jr.; Que, L., Jr. Journal of the American Chemical Society 1999 121, 2226. (14) Honda, Y.; Arii, H.; Okumura, T.; Wada, A.; Funahashi, Y.; Ozawa, T.; Jitsukawa, K.; Masuda, H. Bulletin of the Chemical Society of Japan 2007, 80, 1288. (15) Michelsen, K.; Pedersen, E.; Wilson, S. R.; Hodgson, D. J. Inorganica Chimica Acta 1982, 63, 141.

PAGE 203

203 (16) D im itrou, E., Indiana University Avail. Univ. Microfilms Int., DA9539950., 222 pp. From: Diss. Abstr. Int., B 1995, 56, 3753. (17) Cotton, F. A.; Wang, W. Inorganic Chemistry 1982, 21, 2675. (18) Beattie, J. K.; Hambley, T. W.; Klepetko, J. A.; Masters, A. F.; Turner, P. Polyhedron 1996, 15, 2141. (19) Lippard, S. J.; Berg, J. M. Principles of Bioinorganic Chemistry University Science Books, 1994. (20) Zouni, A.; Witt, H.T.; Kern, J.; Fromme, P.; Krauss, N.; Saenger, W.; Orth, P. Nature (London) 2001 409 739. (21) Kamiya, N.; Shen, J. R. Proceedings of the National Academy of Sciences of the United States of America 2003, 100, 98. (22) Ferreira, K. N.; Iverson, T. M.; Maghlaoui, K.; Barber, J.; Iwata, S. Science (Washington, DC, United States) 2004, 3 03, 1831. (23) Barber, J.; Ferreira, K.; Maghlaoui, K.; Iwata, S. Physical Chemistry Chemical Physics 2004, 6 4737. (24) Hendry, G.; Wydrzynski, T. Biochemistry 2003, 42, 6209. (25) Siegbahn, P. E. M. Current Opinion in Chemical Biology 2002, 6 227. (26) Weng, T. C.; Hsieh, W.Y.; Uffelman, E. S.; Gordon Wylie, S. W.; Collins, T. J.; Pecoraro, V. L.; Penner Hahn, J. E. Journal of the American Chemical Society 2004, 126, 8070. (27) Guha, B. C. Proceedings of the Royal Society of London, Series A: Math ematical, Physical and Engineering Sciences 1951, 206, 353. (28) Bleaney, B.; Bowers, K. D. Proceedings of the Royal Society of London, Series A: Mathematical, Physical and Engineering Sciences 1952, 214, 451. (29) Carlin, R. L. Magnetochemistry ; Springer, 1986. (30) Koegerler, P.; Tsukerblat, B.; Mueller, A. Dalton Transactions 2010, 39, 21. (31) Kolks, G.; Lippard, S. J.; Waszczak, J. V.; Lilienthal, H. R. Journal of the American Chemical Society 1982 104, 717. (32) Kambe, K. Journal of the Physical Society of Japan 1950, 5 48.

PAGE 204

204 (33) Wilson, A.; Yang, E. C.; Hendrickson, D. N.; Hill, S. Polyhedron 2007, 26, 2065. (34) Christou, G.; Gatteschi, D.; Hendrickson, D. N.; Sessoli, R. MRS Bulletin 2000 25, 66. (35) Aubin, S. M. J.; Wemple, M. W.; Adams, D. M.; Tsai, H. L.; Christou, G.; Hendrickson, D. N. Journal of the American Chemical Society 1996 118, 7746. (36) Lis, T. Acta Crystallographica, Section B: Structural Crystallography and Crystal Chemistry 1980, B36 2042. (37) Caneschi, A.; Gatteschi D.; Sessoli, R.; Barra, A. L.; Brunel, L. C.; Guillot, M. Journal of the American Chemical Society 1991, 113, 5873. (38) van Wuellen, C. Journal of Chemical Physics 2009, 130, 194109/1. (39) Jaglicic, Z.; Luznik, J.; Pirnat, J.; Trontelj, Z. Physica B: Condensed Matter (Amsterdam) 2000, 284288, 1219. (40) Evangelisti, M.; Bartolome, J.; Luis, F. Solid State Communications 1999, 112, 687. (41) Barra, A. L.; Debrunner, P.; Gatteschi, D.; Schulz, C. E.; Sessoli, R. Europhysics Letters 1996, 35, 133. (4 2) Wieghardt, K.; Pohl, K.; Jibril, I.; Huttner, G. Angewandte Chemie 1984, 96, 66. (43) Wang, S.; Tsai, H. L.; Libby, E.; Folting, K.; Streib, W. E.; Hendrickson, D. N.; Christou, G. Inorganic Chemistry 1996, 35, 7578. (44) Aubin, S. M. J.; Dilley, N. R.; Pardi, L.; Krzystek, J.; Wemple, M. W.; Brunel, L.C.; Maple, M. B.; Christou, G.; Hendrickson, D. N. Journal of the American Chemical Society 1998 120, 4991. (45) Yang, E. C.; Wernsdorfer, W.; Hill, S.; Edwards, R. S.; Nakano, M.; Maccagnano, S.; Zak harov, L. N.; Rheingold, A. L.; Christou, G.; Hendrickson, D. N. Polyhedron 2003, 22, 1727. (46) Gatteschi, D.; Sessoli, R. Angewandte Chemie (International ed. in English) 2003, 42, 268. (47) Ritter, S. K. Chemical & Engineering News 2004, 82, 29. (48) Bogani, L.; Wernsdorfer, W. Nature Materials 2008, 7 179.

PAGE 205

205 (49) Robin, M. B.; Day, P. Advances in Inorganic Chemistry and Radiochemistry 1967, 10, 247. (50) Kealy, T. J.; Pauson, P. L. Nature (London, United Kingdom) 1951, 168, 1039. (51) Wilkinson, G. ; Rosenblum, M.; Whiting, M. C.; Woodward, R. B. Journal of the American Chemical Society 1952 74, 2125. (52) Woodward, R. B.; Rosenblum, M.; Whiting, M. C. Journal of the American Chemical Society 1952 74, 3458. (53) Jaffe, H. H. Journal of Chemical P hysics 1953, 21, 156. (54) Zanello, P. Inorganic Electrochemistry: Theory, Practice and Application, 2003. (55) Sharp, M.; Petersson, M.; Edstroem, K. Journal of Electroanalytical Chemistry and Interfacial Electrochemistry 1980, 109, 271. (56) Ito, N.; Saji, T.; Aoyagui, S. Journal of Organometallic Chemistry 1983, 247, 301. (57) Gil, M. A.; Maringgele, W.; Dechert, S.; Meyer, F. Zeitschrift fuer Anorganische und Allgemeine Chemie 2007, 633, 2178. (58) Deeth, R. J. Inorganic Chemistry (Washington, DC, United States) 2008, 47, 6711. (59) Rosenblum, M.; Woodward, R. B. Journal of the American Chemical Society 1958, 80, 5443. (60) Laane, J. Journal of Coordination Chemistry 1971, 1 75. (61) Petter, R. C.; Rao, S. J. Journal of Organic Chemistry 1991, 5 6 2932. (62) Pardo, E.; Faus, J.; Julve, M.; Lloret, F.; Munoz, M. C.; Cano, J.; Ottenwaelder, X.; Journaux, Y.; Carrasco, R.; Blay, G.; Fernandez, I.; Ruiz Garcia, R. Journal of the American Chemical Society 2003, 125, 10770. (63) Catlow, C. R. A.; Editor Computer Modeling in Inorganic Crystallography ; Academic: San Diego, 1997. (64) Pauling, L. Journal of the American Chemical Society 1929, 51, 1010. (65) Bragg, W. L. Zeitschrift fuer Kristallographie, Kristallgeometrie, Kristallphysik, Kristallchemie 1930, 74, 237. (66) Pauling, L. Journal of the American Chemical Society 1947, 69, 542.

PAGE 206

206 (67) Donnay, G. Year Book Carnegie Institution of Washington 1970, 292. (68) Donnay, G.; Allmann, R. American Mineralogist 1970, 55, 1003. (69) Brown, I. D.; Sh annon, R. D. Acta Crystallographica, Section A: Crystal Physics, Diffraction, Theoretical and General Crystallography 1973, 29, 266. (70) Brown, I. D.; Wu, K. K. Acta Crystallographica, Section B: Structural Crystallography and Crystal Chemistry 1976, B32 1957. (71) Brown, I. D.; Altermatt, D. Acta Crystallographica, Section B: Structural Science 1985, B41 244. (72) Garcia Rodriguez, L.; RutePerez, A.; Pinero, J. R.; Gonzalez Silgo, C. Acta Crystallographica, Section B: Structural Science 2000, B56 5 65. (73) Brown, I. D. Acta Crystallographica, Section B: Structural Science 1992, B48 553. (74) Ardavan, A.; Blundell, S. J. Journal of Materials Chemistry 2009, 19 1754. (75) Price, D. J.; Batten, S. R.; Moubaraki, B.; Murray, K. S. Polyhedron 2007, 26, 305. (76) Tasiopoulos, A. J.; Vinslava, A.; Werndorfer, W.; Abboud, K. A.; Christou, G. Angewandte Chemie, International Edition 2004 43, 2117. (77) Aliaga Alcalde, N.; Edwards, R. S.; Hill, S. O.; Wernsdorfer, W.; Folting, K.; Christou, G. Journal of the American Chemical Society 2004, 126, 12503. (78) Schmid, G. Structure and Bonding (Berlin, Germany) 1985, 62, 51. (79) Papaefstathiou, G. S.; Perlepes, S. P. Comments on Inorganic Chemistry 2002, 23, 249. (80) Boyd, P. D. W.; Li, Q.; Vincent, J. B.; Folting, K.; Chang, H. R.; Streib, W. E.; Huffman, J. C.; Christou, G.; Hendrickson, D. N. Journal of the American Chemical Society 1988 110, 8537. (81) Sessoli, R.; Tsai, H. L.; Schake, A. R.; Wang, S.; Vincent, J. B.; Folting, K.; Gatteschi, D.; Christou, G.; Hendrickson, D. N. Journal of the American Chemical Society 1993, 115, 1804. (82) Weast, R. C. Handbook of Chemistry and Physics ; 52nd ed. 1972.

PAGE 207

207 (83) Weast, R. C. Handbook of Chemistry and Physics ; 64th ed., 1983. (84) Soler, M.; Artus, P.; Folting, K.; Huffman, J. C.; Hendrickson, D. N.; Christou, G. Inorganic Chemistry 2001, 40, 4902. (85) Artus, P.; Boskovic, C.; Yoo, J.; Streib, W. E.; Brunel, L. C.; Hendrickson, D. N.; Christou, G. Inorganic Che mistry 2001, 40, 4199. (86) Gerbier, P.; Ruiz Molina, D.; Domingo, N.; Amabilino, D. B.; VidalGancedo, J.; Tejada, J.; Hendrickson, D. N.; Veciana, J. Monatshefte fuer Chemie 2003, 134, 265. (87) Little, W. F.; Eisenthal, R. Journal of Organic Chemistry 1961, 26, 3609. (88) Meng, X.; Hou, H.; Li, G.; Ye, B.; Ge, T.; Fan, Y.; Zhu, Y.; Sakiyama, H. Journal of Organometallic Chemistry 2004, 689, 1218. (89) Mata, J. A.; Peris, E.; Llusar, R.; Uriel, S.; Cifuentes, M. P.; Humphrey, M. G.; Samoc, M.; Luther Davies, B. European Journal of Inorganic Chemistry 2001, 2113. (90) Loeffler, U.; Goepel, W.; Speiser, B. Electroanalysis 1991, 3 917. (91) Zhang, X.; Wang, J.; Gao, Y.; Zeng, X. C. ACS Nano 2009, 3 537. (92) Engtrakul, C.; Sita, L. R. Nano Letters 2001, 1 541. (93) Rausch, M. D.; Fischer, E. O.; Grubert, H. Journal of the American Chemical Society 1960, 82, 76. (94) Vigato, P. A.; Casellato, U.; Clemente, D. A.; Bandoli, G.; Vidali, M. Journal of Inorganic and Nuclear Chemistry 1973, 35, 4131. (95) Lee, S. M.; Cheung, K. K.; Wong, W. T. Journal of Organometallic Chemistry 1996, 506, 77. (96) Cotton, F. A.; Daniels, L. M.; Lin, C.; Murillo, C. A. Journal of the American Chemical Society 1999 121, 4538. (97) Cotton, F. A.; Lin, C.; Murillo, C. A. Inorganic Chemistry 2001, 40, 478. (98) Uhl, W.; Spies, T.; Haase, D.; Winter, R.; Kaim, W. Organometallics 2000, 19, 1128. (99) Dong, G.; Li, Y. t.; Duan, C. y.; Hong, M.; Meng, Q. j. Inorganic Chemistry 2003, 42, 2519.

PAGE 208

208 (100) Meng, X.; Cheng, W.; Mi, L.; Tang, M.; Hou, H. Inorganic Chemistry Communications 2006, 9 662. (101) Kondo, M.; Shinagawa, R.; Miyazawa, M.; Kabir, M. K.; Irie, Y.; Horiba, T.; Naito, T.; Maeda, K.; Utsuno, S.; Uchida, F. Dalton Transactions 2003, 515. (102) Masello, A.; Muruges u, M.; Abboud, K. A.; Christou, G. Polyhedron 2007, 26, 2276. (103) Das, N.; Arif, A. M.; Stang, P. J.; Sieger, M.; Sarkar, B.; Kaim, W.; Fiedler, J. Inorganic Chemistry 2005, 44, 5798. (104) Guo, D.; Zhang, B. g.; Duan, C. y.; Cao, X.; Meng, Q. j. Dalto n Transactions 2003, 282. (105) Zheng, G. L.; Ma, J.F.; Su, Z. M.; Yan, L. K.; Yang, J.; Li, Y. Y.; Liu, J. F. Angewandte Chemie, International Edition 2004 43, 2409. (106) Chandrasekhar, V.; Thirumoorthi, R. Organometallics 2007, 26, 5415. (107) Kim, Y. S.; Kim, J.; Kim, D.; Chae, H. K. Chemistry Letters 2007, 36, 150. (108) Meng, X.; Li, G.; Hou, H.; Han, H.; Fan, Y.; Zhu, Y.; Du, C. Journal of Organometallic Chemistry 2003, 679, 153. (109) Mereacre, V.; Prodius, D.; Ako, A. M.; Shova, S.; Turta, C .; Wurst, K.; Jaitner, P.; Powell, A. K. Polyhedron 2009, 28, 3551. (110) Mereacre, V.; Ako, A. M.; Filoti, G.; Bartolome, J.; Anson, C. E.; Powell, A. K. Polyhedron 2010, 29, 244. (111) Wang, L.; Meng, X.; Zhang, E.; Hou, H.; Fan, Y. Journal of Organometallic Chemistry 2007, 692, 4367. (112) Bera, J. K.; Clerac, R.; Fanwick, P. E.; Walton, R. A. Journal of the Chemical Society, Dalton Transactions 2002 2168. (113) Maksakov, V. A.; Slovohotova, I. V.; Golovin, A. V.; Babailov, S. P. Russian Chemical Bu lletin (Translation of Izvestiya Akademii Nauk, Seriya Khimicheskaya) 2001, 50, 2451. (114) Cowley, A. R.; Hector, A. L.; Hill, A. F.; White, A. J. P.; Williams, D. J.; Wilton Ely, J. D. E. T. Organometallics 2007, 26, 6114.

PAGE 209

209 (115) Kuehnert, J.; Rueffer, T.; Ecorchard, P.; Braeuer, B.; Lan, Y.; Powell, A. K.; Lang, H. Dalton Transactions 2009, 4499. (116) Lu, E. N.; Zhang, J.; Zhu, L. M.; Xu, Y.; Fan, Y.T. Wuji Huaxue Xuebao 2008, 24, 201. (117) Wen, L. L.; Zhang, B. G.; Peng, Z. H.; Ren, J. G.; Cai, P.; Duan, C. Y. Wuji Huaxue Xuebao 2004, 20, 1228. (118) Yang, Y. Y.; Wong, W. T. Chemical Communications (Cambridge, United Kingdom) 2002, 2716. (119) Rosenblum, M. Chemistry of the Iron Group Metallocenes: Ferrocene, Ruthenocene, Osmocene, Part 1 (The Chemistry of Organometallic Compounds) 1965. (120) Bernstein, T.; Herbstein, F. H. Acta Crystallographica, Section B: Structural Crystallography and Crystal Chemistry 1968, 24, 1640. (121) Nesmeyanov, A. N.; Sazonova, V. A.; Romanenko, V. I. Doklady Akadem ii Nauk SSSR 1963, 152, 1358. (122) Sorensen, H. S.; Larsen, J.; Rasmussen, B. S.; Laursen, B.; Hansen, S. G.; Skrydstrup, T.; Amatore, C.; Jutand, A. Organometallics 2002, 21, 5243. (123) Yamaji, T.; Saito, T.; Hayamizu, K.; M.Yanagisawa; O.Yamamoto; N.Wasada; K.Someno; S.Kinugasa; K.Tanabe; T.Tamura; K.Tanabe; J.Hiraishi; April, 17 2010 ed. 2010. (124) Kubota, N.; Fukuo, T.; Akawa, R. Journal of the American Society for Mass Spectrometry 1999, 10, 557. (125) Nomura, T. Bulletin of the Chemical Society of Japan 1983, 56, 2937. (126) Eppley, H. J.; Tsai, H. L.; Aubin, S. M.; deVries, N.; Folting, K.; Hendrickson, D. N.; Christou, G. Book of Abstracts, 210th ACS National Meeting, Chicago, IL, August 2024 1995, INOR. (127) Eppley, H. J.; Tsai, H. L.; de Vries, N.; Folting, K.; Christou, G.; Hendrickson, D. N. Journal of the American Chemical Society 1995, 117, 301. (128) Eppley, H. J.; Christou, G.; Law, N. A.; Pecoraro, V. L. Inorganic Syntheses 2002, 33, 61. (129) Palenik, G. J. Inorganic Chemistry 1 997, 36, 4888. (130) Palenik, G. J. Inorganic Chemistry 1997, 36, 122.

PAGE 210

210 (131) Brese, N. E.; O'Keeffe, M. Acta Crystallographica, Section B: Structural Science 1991, B47 192. (132) Cowan, D. O.; Shu, P.; Hedberg, F. L.; Rossi, M.; Kistenmacher, T. J. Journal of the American Chemical Society 1979, 101, 1138. (133) Bats, J. W.; Scheibitz, M.; Wagner, M. Acta Crystallographica, Section C: Crystal Structure Communications 2003, C59 m355. (134) McNaught, A. D.; Wilkinson, A. IUPAC: Compendium of Chemical Terminology ; 2nd. ed. ed., 1997. (135) Sun, Z.; Gantzel, P. K.; Hendrickson, D. N. Inorganic Chemistry 1996, 35, 6640. (136) CCDC 2009, Mercury http://www.ccdc.cam.ac.uk/products/mercury/ (137) Berry, R. S. Journal of Chemical Physics 1960, 32, 933. (138) Muetterties, E. L.; Guggenberger, L. J. J. Amer. Chem. Soc. 1974, 96, 1748. (139) Addison, A. W.; Rao, T. N.; Reedijk, J.; Van Rijn, J.; Verschoor, G. C. Journal of the Chemical Society, Da lton Transactions 1984, 1349. (140) Davidson, E. R. Bloomington, IN 1999, Magnet (141) Davidson, E. R. Bloomington, IN, 1999, Grid (142) Wernsdorfer, W. Advances in Chemical Physics 2001, 118, 99. (143) Crawford, V. H.; Richardson, H. W.; Wasson, J. R.; Hodgson, D. J.; Hatfield, W. E. Inorganic Chemistry 1976, 15, 2107. (144) Atanasov, M. A.; Angelov, S.; Mayer, I. Solid State Communications 1985, 56, 743. (145) Kolks, G.; Lippard, S. J.; Waszczak, J. V.; Lilienthal, H. R. Journal of the America n Chemical Society 1982 104, 717. (146) Connelly, N. G.; Geiger, W. E. Chemical Reviews (Washington, D. C.) 1996, 96 877. (147) Steckhan, E. Angewandte Chemie 1986, 98, 681. (148) Steckhan, E.; Editor Topics in Current Chemistry, Vol. 142: Electrochem istry I 1987.

PAGE 211

211 (149) Yueh, W.; Bauld, N. L. Journal of the American Chemical Society 1995, 117, 5671. (150) KurodaSowa, T.; Nakano, M.; Christou, G.; Hendrickson, D. N. Polyhedron 2001, 20, 1529. (151) Dimitrou, K.; Sun, J.S.; Folting, K.; Christou, G. Inorganic Chemistry 1995, 34, 4160. (152) Raptis, R. G.; Georgakaki, I. P.; Hockless, D. C. R. Angewandte Chemie, International Edition 1999, 38, 1632. (153) Bruker AXS Madison, 2000. Shelxtl6 (154) Van der Sluis, P.; Spek, A. L. Acta Crystallographi ca, Section A: Foundations of Crystallography 1990, A46 194. (155) Spek, A. L. Journal of Applied Crystallography 2003, 36, 7. (156) Andrews, S. C.; Arosio, P.; Bottke, W.; Briat, J. F.; Von Darl, M.; Harrison, P. M.; Laulhere, J. P.; Levi, S.; Lobreaux S.; Yewdall, S. J. Journal of Inorganic Biochemistry 1992, 47, 161. (157) Ako, A. M.; Hewitt, I. J.; Mereacre, V.; Clerac, R.; Wernsdorfer, W.; Anson, C. E.; Powell, A. K. Angewandte Chemie, International Edition 2006, 45, 4926. (158) Delfs, C.; Gattes chi, D.; Pardi, L.; Sessoli, R.; Wieghardt, K.; Hanke, D. Inorganic Chemistry 1993, 32, 3099. (159) Barra, A. L.; Caneschi, A.; Cornia, A.; de Biani, F. F.; Gatteschi, D.; Sangregorio, C.; Sessoli, R.; Sorace, L. Journal of the American Chemical Society 1 999, 121, 5302. (160) Foguet Albiol, M. D. Dissertation, University of Florida, 2006. (161) Gorun, S. M.; Papaefthymiou, G. C.; Frankel, R. B.; Lippard, S. J. Journal of the American Chemical Society 1987 109, 4244. (162) Goodwin, J. C.; Sessoli, R.; Gatteschi, D.; Wernsdorfer, W.; Powell, A. K.; Heath, S. L. Dalton 2000, 1835. (163) Powell, A. K.; Heath, S. L.; Gatteschi, D.; Pardi, L.; Sessoli, R.; Spina, G.; Del Giallo, F.; Pieralli, F. Journal of the American Chemical Society 1995, 117, 2491. (164) Ammala, P.; Cashion, J. D.; Kepert, C. M.; Moubaraki, B.; Murray, K. S.; Spiccia, L.; West, B. O. Angewandte Chemie, International Edition 2000, 39 1688.

PAGE 212

212 (165) Murugesu, M.; Abboud, K. A.; Christou, G. Polyhedron 2004, 23, 2779. (166) Taft, K. L.; Lippard, S. J. Journal of the American Chemical Society 1990, 112, 9629. (167) Ali, L. H.; Cox, A.; Kemp, T. J. Journal of the Chemical Society, Chemical Communications 1972, 265. (168) Bozak, R. E. Advances in Photochemistry 1971, 8 227. (169) Ali, L. H.; Cox, A.; Kemp, T. J. Journal of the Chemical Society, Dalton Transactions 1973, 1468. (170) Nesmeyanov, A. N.; Sazonova, V. A.; Romanenko, V. I.; Zol'nikova, G. P. Izvestiya Akademii Nauk SSSR, Seriya Khimicheskaya 1965, 1694. (171) Ding, W.; Sanderson, C. T.; Conover, R. C.; Johnson, M. K.; Amster, I. J.; Kutal, C. Inorganic Chemistry 2003, 42, 1532. (172) Yamaguchi, Y.; Ding, W.; Sanderson, C. T.; Borden, M. L.; Morgan, M. J.; Kutal, C. Coordination Chemistry Reviews 2007, 251, 515. (173) Chakraborty I.; Baran, P.; Sanakis, Y.; Simopoulos, A.; Fachini, E.; Raptis, R. G. Inorganic Chemistry (Washington, DC, United States) 2008, 47, 11734. (174) Gass, I. A.; Milios, C. J.; Whittaker, A. G.; Fabiani, F. P. A.; Parsons, S.; Murrie, M.; Perlepes, S. P.; Brechin, E. K. Inorganic Chemistry 2006, 45, 5281. (175) Hahn, F. E.; Jocher, C.; Lugger, T. Zeitschrift fuer Naturforschung, B: Chemical Sciences 2004, 59, 855. (176) Kanowitz, S. M.; Palenik, G. J. Inorganic Chemistry 1998, 37, 2086. (177) Gray, H. B. ; Hendrickson, D. N.; Sohn, Y. S. Inorganic Chemistry 1971, 10, 1559. (178) Walters, M. A.; Chaparro, J.; Siddiqui, T.; Williams, F.; Ulku, C.; Rheingold, A. L. Inorganica Chimica Acta 2006, 359, 3996. (179) Duggan, D. M.; Hendrickson, D. N. Inorganic Chemistry 1975, 14, 955. (180) Prins, R. Molecular Physics 1970, 19, 603. (181) Prins, R.; Korswagen, A. R. J. Organometal. Chem. 1970, 25, C74. (182) Greenwood, N. N.; Gibb, T. C. M ssbauer Spectroscopy 1971.

PAGE 213

213 (183) Morrison, W. H., Jr.; Hendrickson, D. N. Inorganic Chemistry 1975, 14, 2331. (184) Edwards, P. R.; Johnson, C. E. Journal of Chemical Physics 1968, 49, 211. (185) Nemykin, V. N.; Hadt, R. G. Inorganic Chemistry 2006, 45, 8297. (186) Jiao, J.; Long, G. J.; Rebbouh, L.; Grandjean, F.; Beatty A. M.; Fehlner, T. P. Journal of the American Chemical Society 2005, 127, 17819. (187) Bergamini, P.; Di, M. S.; Maldotti, A.; Sostero, S.; Traverso, O. Journal of Organometallic Chemistry 1989, 365, 341. (188) Che, D. J.; Li, G.; Yao, X. L.; Zou, D. P Journal of Organometallic Chemistry 1998, 568, 165. (189) Karppinen, M.; Yamauchi, H. International Journal of Inorganic Materials 2000 2 589. (190) Bickmore, B. R.; Rosso, K. M.; Tadanier, C. J.; Bylaska, E. J.; Doud, D. Geochimica et Cosmochimica A cta 2006, 70, 4057. (191) Stoilova, D.; Marinova, D.; Georgiev, M. Vibrational Spectroscopy 2009, 50, 245. (192) Adams, S.; Moretzki, O.; Canadell, E. Solid State Ionics 2004, 168, 281. (193) Thorp, H. H. Inorganic Chemistry 1992, 31, 1585. (194) Liu, W.; Thorp, H. H. Inorganic Chemistry 1993, 32, 4102. (195) Brown, I. D. Acta Crystallographica, Section B: Structural Science 2009, B65 684. (196) Palenik, G. J.; Hu, S. Z. Inorganica Chimica Acta 2009, 362, 4740. (197) Sidey, V. Acta Crystallographica, Section B: Structural Science 2008 B64 515. (198) Urusov, V. S. Doklady Physical Chemistry 2006, 408, 152. (199) Zocchi, F. Solid State Sciences 2000, 2 383. (200) Zocchi, F. Solid State Sciences 2001, 3 383. (201) Naskar, J. P.; Hati, S.; Datta, D. Acta Crystallographica, Section B: Structural Science 1997 B53 885. (202) Fletcher, R. Practical Methods of Optimization; 2nd Ed. ed. New York, 1987.

PAGE 214

214 (203) FarachColton, M.; Roberts, F. S.; Vingron, M.; Waterman, M. DIMACS Series indiscrete mathematics and theoretical computer science; American Mathematical Society: Providence, RI, 1999; Vol. 47. (204) Mucherino, A.; Papajorgji, P.; Pardalos, P. M. Data mining in agricolture; Springer: New York, 2009. (205) Fletcher, R.; Reeves, C. M. Comput. J. 1 964, 7 149. (206) Cafieri, S.; D'Apuzzo, M.; Marino, M.; Mucherino, A.; Toraldo, G. J. Optim. Theory Appl. 2006, 129, 55. (207) Kirkpatrick, S.; Gelatt, C. D.; Vecchi, M. P. Science 1983 220, 671. (208) Lee, K. S.; Geem, Z. W.; Lee, S. h.; Bae, K. w. Engineering Optimization 2005, 37, 663 (209) Goldberg, D. E. Genetic Algorithms in Search, Optimization, and Machine Learning; 1st. Ed. ed.; AddisonWesley: New York, 1989. (210) Corne, D.; Dorigo, M.; Glover, F. New ideas in optimization McGraw Hill: L ondon, 1999. (211) Kennedy, J.; Eberhart, R. In Proc. IEEE Intl. Conf. on Neural Networks Perth, Australia, 1995; Vol. IV, p 1942. (212) Mucherino, A.; Seref, O. In Data Mining, System Analysis and Optimization in Biomedicine Gainesville, Florida, 2007, p 162. (213) Mladenovic, N.; Hansen, P. Computers & Operations Research 1997, 24, 1097. (214) Brown, I. D. Chemical Reviews 2009, 109, 6858.

PAGE 215

215 BIOGRAPHICAL SKETCH Antonio Masello was born in Cagliari, Italy in 1976. He attended the Industrial Technical Institute "M. Giua" earning his diploma in industrial chemistry and biotechnologies. Soon after he enrolled at the "Universit degli Studi" of Cagliari where was awarded a master in chemistry in 2002 with the maximum score. During the same year, he became a member of the Italian National Council of Chemists passing the exam with the highest score. In September 2002 he joined the Italian Army as a volunteer and worked for one year in the Italian Joint School for the NBC (Nuclear Biological Chemical) Defense, in the laboratory of Maj. Dr. Giammaria S. In August 2004, he began his PhD studies at the University of Florida where worked on the synthesis and study of polynuclear clusters containing ferrocene, and performed in collaboration with Dr. Antonio Mucherino a statistical study on the model known as Bond Valence Sum.