Variable coordination of tris ( 2-pyridyl ) phosphine and its oxide toward M ( hfac ) 2 : a metal-speci fi able switching between the formation of mono-and bis-scorpionate complexes †

An unexpected substitution of the anionic chelating ligands at the M centre by a neutral tripodal ligand has been observed in the reaction of Mn, Co, Ni and Cu hexafluoroacetylacetonates (hfac) with tris (2-pyridyl)phosphine (Py3P) or its oxide (Py3P = O). The nature of the metal ion in M(hfac)2 and the M/L ratio determine the degree of substitution of hfac-anions (partial vs. total) and therefore, the structure of the complex formed (scorpionate vs. bis-scorpionate ones, respectively). Hence, the reaction of the ligands with [Cu(hfac)2(H2O)2] in an equimolar ratio affords scorpionate [Cu(N,N’,N’’-Py3P = X)(O,O’-hfac) (O-hfac)], wherein one hfac-ligand chelates metal, while the other hfac acts as an O-monodentate one. Using the two equivalents of Py3P in this reaction leads to [Cu(N,N’,N’’-Py3P)2](hfac)2, which contains a bis-scorpionate cation [Cu(Py3P)2] 2+ and two noncoordinated hfac-anions. [Co(hfac)2(H2O)2] and [Ni(hfac)2(H2O)2], regardless of the M/L molar ratio, react with Py3P = O to give cationic scorpionates [M(N,N’,N’’-Py3P = O)(O,O’-hfac)(H2O)](hfac), in which one hfac-anion is noncoordinated. In contrast, [Mn(hfac)2(H2O)2], on interaction with Py3P, results in the cationic complex [Mn(N,N’,N’’-Py3P)2] [Mn(hfac)3]2 bearing a bis-scorpionate cation [Mn(Py3P)2] 2+ and two [Mn(hfac)3]2 − counterions. The synthesized scorpionates have been characterized by X-ray diffractometry, cyclic voltammetry, SQUID magnetometry, FT-IR and UV-Vis techniques.


Introduction
During the past decades, pyridylphosphines and their chalcogenides have attracted increasing attention in coordination chemistry, catalysis and material science. [1][2][3] The combination of "soft" (P-atom) and "hard" (N-atoms) donor sites in pyridylphosphines makes them very important and versatile ligands for the design of unique catalysts, 4 luminescent materials 5 and prospective drugs. 6 Among pyridylphosphines, tris (2-pyridyl) phosphine (Py 3 P) is gaining a special interest owing to the useful properties of its metal complexes. For instance, Pd, 7,8 Cr, 9 Mo, 9 W, 9,10 Ru 11 and Rh, 11 Fe, 12 Co, 12,13 Ni 13 and Mn 13 complexes with Py 3 P have been explored as catalysts for methoxycarbonylation of alkynes, 7,8 Diels-Alder cycloaddition, 9 Friedel-Crafts/aldehyde cyclotrimerization, 10 hydroformylation of alkenes, 11 ethylene polymerization, 12 and O 2 -oxidation of tetraline. 13 Most significantly, Py 3 P has recently become readily available due to the development of its synthesis directly from elemental phosphorus and 2-bromopyridine. 14 Tris(2-pyridyl)phosphine, owing to its tripodal structure and heminal disposition of the N atoms towards the P atom, exhibits numerous coordination patterns, e.g. N-15 and P-monodentate, 14,16 P,N-bridging, 17 N,N′-chelating, 18,19 N,P,N′pincer, 20 N,N′/P-bridging 21 and N,N′,N″-tripodal 9,22-24 ones. A while ago, N,N′,N″-/P-coordination of substituted tris (2-pyridyl) phosphine was found in the dinuclear complex [(MeCN) 3 Cu {P(6-Me-2-Py) 3 }Cu(MeCN)](PF 6 ) 2 , wherein this ligand acts as a Janus head ligand. 25 In contrast to the plethora of studies related to tris (2-pyridyl)phosphine based complexes, the data on those with tris(2-pyridyl)phosphine chalcogenides, Py 3 P = X (X = O, S or Se), are scarce, although the latter are equally important ligands. In recent years, these complexes have been employed for the synthesis of the scorpionate complexes [Cu(N,N′,N″-Py 3 P = X)]Hal, 26,27 showing yellow to red TADF emission with good quantum yields. 27 Later, a series of Cu I thiocyanate complexes with Py 3 P = O was designed. 28 The Ru II scorpionate supported by the Py 3 P = O, viz. [Ru(N,N′,N″-Py 3 P = O)(bpy)(OH 2 )] (OTf ) 2 , has proven to be an excellent electrocatalyst for water oxidation, thus surpassing any known Ru catalyst. 29 Therefore, the further development of the coordination chemistry of tris(2-pyridyl)phosphine and its chalcogenides appears to be particularly appealing to access novel functional compounds. Herein, as a part of our ongoing interest in this area, we report on the variable coordination of tris(2-pyridyl) phosphine (1) and its oxide (2) towards M(hfac) 2 (M = Cu, Ni, Co, Mn) leading to either scorpionate or bis-scorpionate complexes. In the course of this study, we have observed an unexpected replacement of the anionic chelating ligand at the metal center by the neutral tripodal ligand. The degree of such substitution ( partial vs. complete) and, hence, the structure of the resulting complexes (scorpionate vs. bis-scorpionate) depend on the nature of the metal ion in M(hfac) 2 and, to a lesser extent, on a metal-to-ligand ratio.

Synthetic aspects
As our experiments have shown, the reaction of [Cu (hfac) 2 (H 2 O) 2 ] with Py 3 P (1) at the 1 : 1 molar ratio occurs immediately upon mixing of the reactants (CHCl 3 , r.t.) to give scorpionate 3 in 88% yield (Scheme 1). The reaction is strictly selective: a possible formation of the coordination polymers does not take place. In the structure of 3, Py 3 P acts as N,N′,N″tripodal ligand toward the octahedral Cu II ion. Importantly, during the reaction, one of the hfac-anions retains an O,O′chelating pattern, whereas the second hfac-anion adopts an O-monodentate mode. The reaction of Py 3 P = O (2) with [Cu(hfac) 2 (H 2 O) 2 ] under similar conditions (1 : 1 ratio, r.t., CHCl 3 ) instantly leads to structurally related scorpionate 4 in 80% yield (Scheme 1). Therefore, the charge of the P atom in 1 and 2 (δ− vs. δ+, respectively) has no influence on the result of the reaction with Cu(hfac) 2 . This can be explained by weak or a complete lack of conjugation between the phosphorus-centered orbitals and the π-electronic system of the pyridine rings.
Moreover, the interaction of [Cu(hfac) 2 (H 2 O) 2 ] with Py 3 P in a 1 : 2 ratio at ambient temperature (CHCl 3 or MeCN) produces complex 5 in 78% yield (Scheme 1). Thus, both chelating hfacanions within the coordination sphere of Cu II are replaced by the Py 3 P ligands. In the complex obtained (5), the Cu II ion octahedrally coordinates six N atoms of two Py 3 P molecules to form a bis-scorpionate cation [Cu(Py 3 P) 2 ] 2+ . The two hfacanions are in the second coordination sphere.
Notably, Cu(acac) 2 does not react with Py 3 P or Py 3 P = O; after removing the solvent from the reaction mixture, the starting reagents remain unchanged. The observed discrepancy is likely due to a higher Lewis acidity of Cu II in Cu(hfac) 2 caused by the presence of two strong electron-acceptor hfac-ligands.
To our surprise, the reaction between Py 3 P = O and [Co (hfac) 2 (8), as shown in Scheme 3. Using a stoichiometric ratio of the reactants (3 : 2) affords 8 in 69% yield. In the bis-scorpionate [Mn(Py 3 P) 2 ] 2+ cation, the metal has an octahedral environment arranged from six N atoms of two Py 3 P ligands. In the crystal of 8, there are two stereochemically different [Mn(hfac) 3 ] − anions (Δ-and Λ-isomers), in which the Mn II coordinates three hfac-anions to form a MnO 6 twisted prism.
These results suggest that the outcome of the reaction between M(hfac) 2 and Py 3 P or Py 3 P = O is strongly influenced by the nature of M II ions and, in the case of Cu(hfac) 2 , by the M/L ratio specified. In the reaction with Cu(hfac) 2 , the final product may be the scorpionate (3 or 4) or bis-scorpionate (5) complex, whilst Co(hfac) 2 and Ni(hfac) 2 give only mono-scorpionates (6 and 7). Unlike these, Mn(hfac) 2 interacts with 1, furnishing the unexpected complex [Mn(N,N′,N″-Py 3 P) 2 ] [Mn(hfac) 3 ] 2 (8). Therefore, in the reactions investigated, the replacement of the charged chelating ligands by the neutral tripodal ligands is observed. When M(acac) 2 (exemplified by Cu II acetylacetonate), is used as the starting reagent, no reaction with the Py 3 P or Py 3 P = O is observed. The different behavior of M(acac) 2 and their polyfluorinated analogues is probably due to the better leaving group ability of the hfac-anion compared with that of the acac-anion.
The observed that the difference in reactivity of M(hfac) 2 toward Py 3 P or Py 3 P = O can be rationalized by the interference of two factors. First, it is well known 32 that within a series of octahedral complexes, the lability toward ligand substitution generally increases in the sequence: Cu 2+ [t 2g (6) e g (3) ] > Mn 2+ [t 2g (3) e g (2) ] > Co 2+ [t 2g (5) e g (2) ] > Ni 2+ [t 2g (6) e g (2) ] due to the filling of the t 2g and e g orbitals. Therefore, [Cu(hfac) 2 2 ] as well as in the resulting complexes 6 and 7 becomes more difficult.
The synthesized complexes are air-stable crystals, which are well soluble in MeCN, Me 2 CO and CHCl 3 . Moreover, complex 8 is soluble in Et 2 O and, partially, in hexane. In the solid state, complexes 3-8 have been characterized by X-ray diffraction analysis (the crystallographic data are given in Table S1 †), SQUID magnetometry and ATR-IR spectroscopy, while their redox behaviour has been studied using cyclic voltammetry (CV).

Crystal structure description
In scorpionate 3 ( Fig. 1), the Cu atom has a 4 + 2 pseudooctahedral environment, where the equatorial plane is defined by Scheme 3 Synthesis of bis-scorpionate complex 8. two O atoms of the chelating hfac-ligand (Cu-O 1.9798 ± 0.0069 Å) and two N atoms of Py 3 P. The axial positions are occupied by the third N atom of Py 3 P and the O atom of second hfac-anion, the O3 atom of which remains noncoordinated. The Cu-N eq distance is about 2.01 Å, whereas the Cu-N ax bond length is significantly longer [2.2575(14) Å], indicating a Jahn-Teller axial elongation. 35 On the whole, these values are typical for the Cu II -N Py lengths. 36,37 The geometry of the [Cu(Py 3 P)] cage of 3 deviates from the perfect C 3 -symmetry; the dihedral angles between the pyridine plane are 55. 66 The structure of scorpionate 4 ( Fig. 2) is almost perfectly consistent with that of 3 (overlay of these molecules is depicted in Fig. S1 †). The major difference between them is the presence of an O atom of the PvO bond in 4, the length of which [1.468(2) Å] is comparable with that in the free Py 3 P = O [1.4792(11) Å]. 38 Similarly, the Cu ion of 4 shows a typical Jahn-Teller axial elongation in its distorted octahedral geometry.
The asymmetric unit of 5 contains one noncoordinated hfac-anion and half of the cation [Cu(Py 3 P) 2 ] 2+ , where the Cu atom is located in an inversion center. The two Py 3 P molecules are ligated to the Cu atom through six N atoms to form the bis-scorpionate structure that looks like a paddle-wheel (Fig. 3). Thus, the metal atom adopts a distorted octahedral geometry. It should be noted that in the CuN 6    hydrogen bonds (their parameters are given in the captions of Fig. 4

Electrochemical properties
To elucidate the electrochemical features of the prepared complexes and possible products of their redox transformations, typically two successive cycles were measured. For the initial scan starting at 0.0 V (Ag/Ag I ), a potential was swept up to the positive direction. After two cycles were recorded, the experiment was repeated with the negative direction of the initial potential sweep. In all cases, voltammetry results at both Pt and glassy carbon (GC) electrodes were similar. The data for the first oxidation and first reduction peak potentials of the studied complexes have been summarized in the Table 1.    CV measured in the Py 3 P = O solution of the same concentration demonstrates a weak quasi-reversible couple at the mentioned potentials, whereas the oxidation peaks of the studied complexes are electrochemically irreversible. It is worthwhile to note that neither the nature of the metal nor the type of ligand significantly influences the potential of the first oxidation peak. The peculiarity of Co II complex (6) is the appearance of the second anodic oxidation peak at E = 1.414 V and the cathodic peak at E = 0.552 V on the reverse branch of CV. In our opinion, these peaks might be associated with the Co II /Co III redox transformations.
On the cathodic potential sweep in the solutions of Co II (6) and Ni II (7) complexes at potentials more negative than −0.5 V, the growth of a cathodic current was observed, although there were no cathodic peaks in the studied potential window. Even so, the products of cathodic reduction appeared in the solution and might be responsible for the anodic peak at −0.150/−0.140 V on the anodic branch of the CV.
The Cu II scorpionate 4 demonstrated different cathodic behavior. Thus for its solution, a clear cathodic current peak was measured at −0.349 V. Some parts of the shape of this peak were quite complicated and probably corresponded to several overlapping redox reactions: one part was the same as for the above-mentioned reduction process of the Co II and Ni II complexes and was associated with the anodic peak at −0.140 V on the anodic branch of the CV, and another might have been associated to Cu II /Cu I reduction. This process is electrochemically irreversible, although the appearance of the broad anodic current peak at 0.317 V indicates an oxidation of a reduced complex on the reverse potential scan. Noteworthy, in this case, the cathodic scan was limited by potentials less negative than that needed for the first reduction, with the only anodic feature being an irreversible oxidation peak at the potential close to that of Co II and Ni II complexes.
The data shown in Fig. 8 allow one to estimate the influence of the ligands' nature and their number on the Cu II complexes' voltammetry. The first oxidation current peaks appear at ca. 0.95 V for all of the studied substances, again evidencing that the nature of the ligand has a minor effect on the oxidation peak position. In contrast, the number of ligands coordinated to the metal strongly affects the value of the anodic current measured at the peak. Indeed, comparing the CV measured in the phosphine complexes' (3 and 5) solution of the same concentration, one could notice that for complex 5, bearing two ligands, the oxidation current is two times higher than that for mono-scorpionate 3.
In the series of additional experiments by varying the potential sweep rate, it has been found that the oxidation current peak value linearly depends on the square root of the sweep rate (see ESI † for corresponding data). Such a result suggests a diffusion control on the oxidation process and the absence of the hindrances from adsorption. 44 Taking into account the above-mentioned oxidation current ratio for mono-and bis-phosphine complexes, two assumptions can be deduced. First, the oxidation of complexes occurs presumably   through the ligands. Second, for bis-scorpionate complex 5, both ligands are equally accessible for oxidation in the same potential scan.
On the cathodic branches of the CV (Fig. 8), there are irreversible peaks of cathodic current at −0.35/−0.45 V corresponding to the first reduction of the studied compounds. In contrast to the oxidation, the potentials of the reduction peak depend significantly on the type of ligand. In fact, for both phosphine complexes 3 and 5, their reductions occur at potentials about 100 mV more negative than that for phosphine oxide 4. Considering the oxygen atom in the phosphine oxide ligand as an electron withdrawing substituent, one could speculate on the role of such a center facilitating the reduction of the complex at less negative potentials. Despite the irreversibility of the cathodic peaks, the reduction products of all copper complexes undergo oxidation on the second anodic potential sweep, contributing to the growth of the anodic current in a potential region of 0.0/0.8 V.
As complex [Mn(Py 3 P) 2 ][Mn(hfac) 3 ] 2 (8) contains Mn 2+ ions both in the cationic and in the anionic parts, we attempted to separate their voltammetric responses. Because the Mn II scorpionate composes the cationic part of the complex, CVs were measured at the Nafion coated GC electrode. According to Zook et al., 45 Nafion behaves as a swelling membrane in acetonitrile in a similar way as in water solutions. The application of Nafion as a cation exchange membrane in non-aqueous solutions containing TBA + as a background cation is also known. 46 Voltammograms measured at the Nafion coated GC electrode remarkably differ from those measured at a bare GC (see ESI † for corresponding data). Fig. 9 shows CVs measured at the Nafion modified GC electrode in the 8 solution. At first glance, the anodic behavior of 8 differs significantly from all other studied complexes and demonstrates quasi-reversible characteristics of the first oxidation. However, in fact, the first oxidation current wave should be considered as a result of at least two overlapping processes. By varying the value limiting the anodic potential sweep, it was possible to confirm this sug-gestion. Indeed, until the anodic potential limit does not exceed 0.95 V, the oxidation remains irreversible and resembles other studied complexes. An inflection appears on the anodic branch of the voltammogram at the mentioned potential value and indicates the transition to the next redox process. Expanding the anodic potential limit above 0.95 V, leads to a registration of the cathodic current peak at 0.8 V on the reverse potential scan. Finally, at potentials more positive than 1.0 V, there is one more irreversible anodic current wave. Comparing the anodic behavior of 8 with other studied complexes, one can conclude that, in all cases, oxidation first involves the ligand moiety and demonstrates an irreversible characteristic. At the more positive potentials, the Mn site might be considered as an apparent target for oxidation.
CV measured at the bare Pt electrode in the 8 solution shows the irreversible cathodic current peak at −0.93 V and the anodic peak at 0.09 V on the reverse potential sweep. However, none of them are reproduced at the modified GC electrode. Indeed, in the studied potential region, there were no reduction peaks on the cathodic branch of the CV measured in the solution of 8 on the Nafion coated GC electrode, except for those assigned to the reduction of the products obtained in the anodic potential sweep. Such a difference can be attributed to the presence of Mn II complexes both in cationic and in anionic parts of 8.

Magnetic properties
Magnetic susceptibility data for complexes 3-8 were collected as a function of temperature from 2 to 300 K. Fig. 10 shows a representative plot of Cu II complexes 3-5. For 3, the μ eff value at 300 K is 1.91, which is in good agreement with the theoretical value (1.86µ B ) for Cu II ion, i.e. one paramagnetic centre with spin S = 1/2 and g = 2.15. The μ eff value virtually does not change at lower temperatures down to 2 K. Thus, the characteristic of the μ eff (T ) dependence reveals the absence of significant exchange interactions and isolation of the paramagnetic  centres that is consistent with X-ray analysis data. As seen from Fig. 10, complexes 4 and 5 exhibit nearly the same magnetic behavior; their μ eff values at 300 K are 1.84 and 1.83µ B , respectively.
The temperature dependence μ eff for Co II complex (6) is presented in Fig. 11. At a lower temperature, μ eff smoothly decreases from 4.67µ B (300 K) to 3.46µ B (2 K). The high temperature value of μ eff is higher than the theoretical spin-only value for Co II (S = 3/2, g = 2), which is consistent with the orbital contribution to the magnetic susceptibility, typical for Co II ions in the octahedral environment and g-factors larger than 2. The μ eff value at 300 K agrees well with typical values of 4.3-5.2µ B for the Co II ion. The decrease of μ eff at lower temperatures is due to the spin-orbit interaction. The μ eff value at low temperature is close to the theoretical spin-only value for the Co II ion and indicates the absence of significant exchange interactions between the paramagnetic centres.
For Ni II complex (7), the μ eff value at 300 K is 3.07µ B (Fig. 11), which agrees well with the theoretical value of 3.11µ B for Ni II with spin S = 1 and the g-factor of 2.2. Therefore, the dependence μ eff (T ) for 7 is quite similar to that for Cu II complexes 3-5 (Fig. 10), showing the absence of significant exchange interactions between the paramagnetic centres.
For Mn II complex (8), the μ eff value at 300 K is 10.17µ B and practically does not change when the temperature is lowered to 25 K but decreases to 9.77µ B at 2 K. The μ eff value for the temperature range of 300-25 K is in good agreement with the theoretical spin-only value, 10.25µ B , for three non-interacting paramagnetic centers with S = 5/2 and g = 2 (Mn II ions). The decreasing of μ eff value below 25 K is due to weak antiferromagnetic exchange interactions.
Thus, magnetic measurements show that the paramagnetic nature of complexes 3-8 is defined by the M II ions only; any exchange interactions between the paramagnetic centres are negligibly small. A magnetic saturation plays a role at low temperatures, where μH ∼ kT causes a small decrease of μ eff . For comparison, the theoretical dependence of μ eff (T ), that takes into account the magnetic saturation, is shown in Fig. 12 by the solid line.

Conclusions
In summary, the coordination of tris(2-pyridyl)phosphine or its oxide toward M(hfac) 2 is accompanied by an unexpected replacement of the hfac-ligands by one or two tripodal ligands to afford mono-or bis-scorpionate complexes, respectively. The switching between the formation of one or another structure strongly depends on the nature of M II and, in the case of Cu(hfac) 2 , on the M/L ratio used. In the reaction with Cu(hfac) 2 , the alteration of the Cu/L ratio from 1 : 1 to 1 : 2 alters the selectivity of the product formation from a mono-scorpionate to a bis-scorpionate complex, correspondingly. Moreover, Ni II and Co II hexafluoroacetylacetonates yield exceptional mono-scorpionates, even if the M/L ratio is 1 : 2. In contrast, Mn(hfac) 2 , upon reacting with Py 3 P, gives a complex composed of a bis-scorpionate cation [Mn(Py 3 P) 2 ] 2+ and two [Mn(hfac) 3 ] − counterions.
CV measurements have shown that all synthesized compounds undergo irreversible oxidation at 0.95-1.00 V. The first electrochemical oxidation can be considered as an apparently ligand-centered process. Both the metal nature and the ligand type significantly influence the reduction of the complexes. The irreversibility of the voltammetric peaks suggests that an oxidation as well as a reduction of the studied complexes leads to significant structural changes or can be followed by some chemical transformations.
The magnetic measurements display that the coordination of tris(2-pyridyl)phosphine or its oxide to the M II ion leads to the disappearance of the exchange interactions. As a result, paramagnetic M II ions are well isolated magnetically from each other. Therefore, the tris(2-pyridyl)phosphine and its oxide are promising ligands in the design of single molecular (SMM) or single ion (SIM) magnets. Fig. 11 The temperature dependence of the effective magnetic moment μ eff (T ) for 6 (•)and 7 (▲). Keeping in mind the known data, 12,13 the synthesized complexes can be regarded as potential catalysts for organic transformations. Moreover, they are useful building blocks for the assembly of new heterometallic complexes, e.g. through coordination of the P atoms of 3, 5 or 8 to the "soft" metal ions.
All of the results obtained contribute to the coordination chemistry of tripodal ligands and scorpionate complexes. The disclosed substitution of the anionic chelating ligand in the coordination sphere by the neutral tripodal ligand opens up new opportunities for the structure-oriented design of coordination compounds.

General details
Solvents were purified following standard procedures prior to use. Tris(2-pyridyl)phosphine (1) was synthesized from red phosphorus and 2-bromopyridine according to the published method. 14 Phosphine oxide 2 was prepared by oxidation of 1 with H 2 O 2 in water/acetone media. The M(hfac) 2 were prepared by the method of Bertrand and Kaplan. 47 Elemental analyses were performed using a Flash EA 1112 CHNS analyzer. Melting points were determined with a Kofler micro hot stage. The IR spectra were recorded on a Varian 3100 FT-IR spectrometer with an ATR sample setting. XPRD analysis of samples 3-8 (see Fig. S20-S24 †) was performed on a Shimadzu XRD-7000 diffractometer (Cu Kα radiation, Ni-filter, 5-35°2θ range, 0.03°2θ step, 5 s per point).

Electrochemistry
CVs have been recorded using Autolab PGSTAT 128N (The Netherlands) in MeCN solutions containing 0.1 M TBAP as a supporting electrolyte. Measurements were performed at room temperature in a convenient three electrode cell with either Pt (area 2.01 mm 2 ; BAS) or glassy carbon (GC, area 7.07 mm 2 ) working electrodes, Pt mesh counter electrode and a Ag/Ag I reference electrode prepared with a BAS reference electrode kit filled with a CH 3 CN solution containing 0.1 M TBAP and 0.01 M AgNO 3 . The Ag/Ag I reference electrode has been characterized in 0.1 M TBAP CH 3 CN solution containing 1 mM ferrocene (FeCp 2 ). Both on Pt and on GC working electrodes, the FeCp 2 + /FeCp 2 couple shows E 1/2 = 0.120 V [estimated as (E ap − E cp )/2] vs. a thus-prepared Ag/Ag + reference electrode. All potentials are given against this reference electrode. The concentrations of studied complexes were 1 mM, and a potential sweep rate (ν) was 50 mV s −1 . Prior to CV measurements, argon gas (99.99%) was passed through the solution in the cell for 10 min. For several CV experiments, the GC electrode modified by Nafion was prepared. Nafion solution (Aldrich) was coated onto the GC surface in a manner described earlier. 48 In order to equilibrate the Nafion film against the background electrolyte before CV measurement, the electrode was kept overnight in the MeCN 0.1 M TBAP solution.

Magnetochemistry
The magnetic susceptibility of the polycrystalline samples 3-8 was measured with a Quantum Design MPMSXL SQUID magnetometer in the temperature range of 2-300 K with a magnetic field of up to 5 kOe. None of the complexes exhibited any field dependence of molar magnetization at low temperatures. Diamagnetic corrections were made using the Pascal constants. The effective magnetic moment was calculated as µ eff (T ) = [(3k/N A µ B 2 )χT] 1/2 ≈ (8χT ) 1/2 .

X-Ray crystallography
The single crystals of 3-7 and 8·2Me 2 CO were grown by slow liquid diffusion of hexane (ca. 7 mL) into an acetone or chloroform solution of these complexes (about 50 mg in 5 mL of the solvent). The mixture was allowed to stay at room temperature for a day. The precipitated crystals were collected, washed with a hexane/acetone (or chloroform) mixture and dried in air. Data were collected on a Bruker D8 Venture diffractometer with Mo Kα (λ = 0.71073 Å) radiation using the φ and ω scans. An empirical absorption correction was applied using the SADABS program. 49 The structures were solved and refined by direct methods using SHELX. 50 All non-hydrogen atoms were refined anisotropically using SHELX. 50 The coordinates of the hydrogen atoms were calculated from geometrical positions. The CF 3 groups within 3 (C 4 ) and 8 (C 20 ) were rotationally disordered, and two conformers were thus observed in the solid state. CCDC 1431158 (3), 1431160 (4), 1431159 (5), 1033904 (6), 1036445 (7) and 1035512 (8) contain the supplementary crystallographic data for this article.