Rare Earth Metal Complexes of Bidentate Nitroxide ... - ACS Publications

Dec 21, 2015 - P. Roy and Diana T. Vagelos Laboratories, Department of Chemistry, University of Pennsylvania, 231 South 34th Street, Philadelphia,...
1 downloads 0 Views 3MB Size
Article pubs.acs.org/IC

Rare Earth Metal Complexes of Bidentate Nitroxide Ligands: Synthesis and Electrochemistry Jee Eon Kim, Justin A. Bogart, Patrick J. Carroll, and Eric J. Schelter* P. Roy and Diana T. Vagelos Laboratories, Department of Chemistry, University of Pennsylvania, 231 South 34th Street, Philadelphia, Pennsylvania 19104, United States S Supporting Information *

ABSTRACT: We report rare earth metal complexes with triand bidentate ligands including strongly electron-donating nitroxide groups. The tridentate ligand 1,3,5-tris(2′-tertbutylhydroxylaminoaryl)benzene (H3arene-triNOx) was complexed to cerium(IV) in a 2:1 ligand-to-metal stoichiometry as Ce(Harene-triNOx)2 (1). Cyclic voltammetry of this compound showed stabilization of the tetravalent cerium cation with a Ce(IV/III) couple at E1/2 = −1.82 V versus Fc/Fc+. On the basis of the uninvolvement of the third nitroxide group in the coordination chemistry with the cerium(IV) cation, the ligand system was redesigned toward a simpler bidentate mode, and a series of rare earth metal−arene-diNOx complexes were prepared with La(III), Ce(IV), Pr(III), Tb(III), and Y(III), [RE(arene-diNOx)2]− ([2−RE]−, RE = La, Pr, Y, Tb) and CeIV(arene-diNOx)2, where H2arene-diNOx = 1,3-bis(2′-tert-butylhydroxylaminoaryl)benzene. The core structures were isostructural throughout the series, with three nitroxide groups in η2 binding modes and one κ1 nitroxide group coordinated to the metal center in the solid state. In all cases except CeIV(arene-diNOx)2, electrochemical analysis described two subsequent, ligand-based, quasi-reversible redox waves, indicating that a stable [N−O•] group was generated on the electrochemical time scale. Chemical oxidation of the terbium complex was performed, and isolation of the resulting complex, Tb(arene-diNOx)2· CH2Cl2 (3·CH2Cl2), confirmed the assignment of the cyclic voltammograms. Magnetic data showed no evidence of mixing between the Tb(III) states and the states of the open-shell ligand.



INTRODUCTION

trianionic ligand, has been used to stabilize Cr(IV), Mo(IV), and W(IV) complexes.6−8 In the pursuit of electronic structure studies for 4f-block compounds, we determined that multidentate ligands would be essential to study metal−ligand noninnocence. The high stability of the 3+ oxidation state coupled with the lability of lanthanide−ligand bonds conspires to otherwise disproportionate targeted metal complexes. For example, in lanthanide chemistry, cerocene (Ce(C8H8)2), a complex comprising redoxactive dianionic cyclooctatetraene (COT2−) ligands, exhibits an interesting multiconfigurational ground state described as an admixture of Ce(III, 4f1)[(C8H8)1.5−]2 and Ce(IV, 4f0)[(C8H8)2−]2 configurations. Detailed experiments and computational studies elucidated that the electron hole of cerocene is shared by the cerium cation and the two cyclooctatetraene dianions.9,10 Electrochemical characterization of K[Ce(C8H8)2] showed a reversible Ce(IV/III) couple at −1.4 V versus Fc/Fc+ in tetrahydrofuran (THF) in a one-electron-transfer process. In an attempt to expand the characteristic intermediate valence effects of cerocene to other ions, the analogous compound K[Pr(C8H8)2] was prepared and characterized.11 In the cyclic

The study of redox-active ligands in metal complexes is a fundamental aspect of modern inorganic chemistry. By serving as electron- or hole reservoirs, redox-active ligands can provide access to reaction pathways that mitigate highly energydemanding oxidation state changes of metal cations.1−3 The application of redox-active ligands in f-block chemistry has resulted in unusual electronic situations; the manifestation of metal−ligand noninnocence in the f-block metals is fundamentally different from that in the d-block metals because of the extreme metal−ligand ionicity of the former.4 Thus, because of the lability of metal−ligand bonding in the f block, the targeted synthesis of complexes that show redox noninnocence often leads to decomposition, especially from oxidation reactions.5 Therefore, a critical component of the studies of metal−ligand noninnocence in the 4f block is the stability of oxidized/reduced redox-active ligands, especially against disproportionation. A common strategy used in transition metal chemistry to resist deleterious disproportionation reactions is the use of chelating ligands. The chelate effect is employed to maintain the integrity of the structures and can be used to confer overall complex stability.6 For example, tBuOCO3− (tBuOCO3− = [2,6-(tBuC6H3O)2C6H3]3−), a © XXXX American Chemical Society

Received: September 29, 2015

A

DOI: 10.1021/acs.inorgchem.5b02236 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry voltammograms of K[Pr(C8H8)2], however, the single feature observed was an irreversible oxidation wave at −0.91 V versus Fc/Fc+.11 Consistent with the electrochemical experiment, chemical oxidation of K[Pr(C8H8)2] showed the formation of Pr2(C8H8)3. It was evident that upon the removal of one electron from [Pr(C8H8)2]− the product disproportionated to give Pr2(C8H8)3 and free cyclooctatetraene, which prevented isolation of neutral Pr(C8H8)2. In previous work, our group reported tetravalent cerium and trivalent lanthanum complexes of the 1,3-bis(2′-tertbutylhydroxylaminoaryl)benzene dianion (arene-diNOx2−) that exhibited stabilization of an electron hole in the arenediNOx2− framework on the electrochemical time scale.5 The electron-rich and chelating arene-diNOx2− ligand framework stabilized a Ce(IV) cation and maintained the overall complex integrity upon oxidation. Moreover, the cyclic voltammogram of the metal-redox-inactive lanthanum congener revealed two subsequent reversible redox events at the [N−O−] groups at E1/2 = −1.25 and −0.65 V versus Fc/Fc+, indicating stability upon subsequent oxidation. In this contribution, we report the synthesis of the H3arene-triNOx complex Ce(Harene-triNOx)2 (1) and rare earth−arene-diNOx2− complexes of Pr, Y, and Tb, [RE(arene-diNOx)2 ]− ([2−RE]−). The latter series of complexes [2−RE]− (RE = Pr, Y, Tb) were characterized for comparison with the results of the previously reported complexes [(py)2K(18-crown-6)][La(arene-diNOx)2] (2−La) and Ce(arene-diNOx)2.5 Furthermore, the complex [Tb(arenediNOx)2]− ([2−Tb]−) was oxidized and isolated as Tb(arenediNOx)2·CH2Cl2 (3·CH2Cl2), resulting in a Tb(III) cation and a ligand-based electron hole. These results comprise important steps toward the isolation of non-cerium/non-ytterbium intermediate-valence compounds that avoid disproportionation reactions.

Figure 1. Thermal ellipsoid plot of (H3arene-triNOx)·3THF at the 30% probability level. tert-Butyl groups are depicted using a wireframe model. Hydrogen atoms, except for the [N−OH] protons, and interstitial solvent molecules have been omitted for clarity. Selected bond distance (Å): N(1)−O(1) 1.4544(13).

Each Harene-diNOx2− ligand in 1 possessed two negatively charged nitroxide groups bound to a Ce(IV) cation in an η2 mode and one protonated hydroxylamine ligand arm. The Ce(1)−O bond distances in 1, 2.215(3)−2.223(4) Å, were close to that in the previously reported tetravalent cerium nitroxide complex Ce[η2-ON(tBu)(2-OMe-5-tBu-C6H3)]4,17 which was 2.204(3) Å. The assignment of the η2 binding mode for the two ligand arms was supported by the bond distances, avg. Ce(1)−N 2.641(3) Å, which can be compared to those in the reported complexes [(La(Cp2{ON(C2H4-oPyr)}2)2] (Cp = cyclopentadienide, Pyr = pyridyl) and [(η1ONC5H6Me4)2Sm(μ-η1:η2-ONC5H6Me4)]2, which showed an La−N bond distance of 2.675(3) Å18 and a Sm−N bond distance of 2.537(6) Å,19 respectively, consistent with the ionic radii difference between those ions.20 The N−O and N−OH distances are in the range of fully reduced N−O bond lengths.21−23 The diamagnetic 1H NMR resonances of 1 showed a H−ON resonance at 4.42 ppm, supporting the X-ray structural data (see the Supporting Information). Additionally, the IR spectrum of 1 exhibited a broad H−ON stretch at 3369 cm−1 (see the Supporting Information).24 In order to examine the extent to which the H3arene-triNOx ligand framework stabilized the tetravalent cerium cation, cyclic voltammetry was performed on H3arene-triNOx and complex 1 (Figure 3 and Table 1). In the cyclic voltammogram (CV) of H3arene-triNOx, the first redox waves were observed ranging from −0.67 to 0.42 V versus Fc/Fc+, presumably corresponding to three [N−OH]/[N−O•] processes, which were not welldefined because proton transfer was involved.5,25 The second irreversible oxidation wave at 0.59 V versus Fc/Fc+ was assigned to a [N−O•]/[NO+] couple of H3arene-triNOx. In the CV of Ce(Harene-triNOx)2 (1), the CeIV/III couple was measured at −1.82 V versus Fc/Fc+. The reduction wave observed at −1.1 V versus Fc/Fc+ for 1 was the result of the product of a chemical process arising from the oxidation sweep. The stabilization of the tetravalent cerium cation was attributed to the four electron-rich nitroxide groups in the metal coordination sphere, as previously observed by our group in, for example, the complexes CeIVL4 (L = η2-ON(tBu)(2-OMe-5tBu-C6H3))17 and Ce(arene-diNOx)2.5 Despite the fact that the resulting complex 1 was not the original synthetic target, Ce(arene-triNOx) (a cerium complex with a 1:1 metal−ligand stoichiometry), we were encouraged by the large stabilization of the cerium(IV) cation with the four nitroxide groups. Therefore, the ligand design was simplified to



RESULTS AND DISCUSSION A new tridentate ligand, H3arene-triNOx, was prepared by a synthetic route similar to that for the reported tridentate nitroxide ligand described by our group, tris(2-tertbutylhydroxylaminato)benzylamine (H3TriNOx). The precursor 1,3,5-tris(2′-bromophenyl)benzene was synthesized according to the literature procedure,12 followed by a lithiation reaction at −100 °C with nBuLi and subsequent reaction with 2-methyl-2-nitrosopropane to produce 1,3,5-tris(2′-tertbutylhydroxylaminoaryl)benzene (H3arene-triNOx) in 44% yield. X-ray crystallographic analysis was performed on colorless crystals of the compound (H3arene-triNOx)·3THF grown from a saturated THF solution at −35 °C (Figure 1). H3arene-triNOx was reacted with several Ce(III) and Ce(IV) starting materials, including Ce(OTf)3, (THF)Ce(OAr)3 (Ar = 2,6-tBu-C6H3),13,14 Ce[N(SiMe3)2]3, Ce(OTf)4, CeCl[N(SiMe3)2]3,15 and Ce[N(SiHMe2)2]4,16 to synthesize the corresponding 1:1 Ce(arene-triNOx) complexes, where the cerium metal was expected to be either tri- or tetravalent. However, in these cases intractable brown solids precipitated from the reactions. When Ce(OtBu)4(py)216 was treated with 2 equiv of H3arene-triNOx, the reaction mixture turned dark-brown/ orange. After the volatiles were removed under reduced pressure, the product was crystallized from THF/hexanes, and X-ray crystallographic analysis revealed that the resulting complex comprised one Ce(IV) cation and two HarenetriNOx2− ligands, Ce(Harene-triNOx)2 (1) (Scheme 1 and Figure 2). B

DOI: 10.1021/acs.inorgchem.5b02236 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Scheme 1. Preparation of Ce(Harene-triNOx)2 (1)

Figure 2. Thermal ellipsoid plot of Ce(Harene-triNOx)2 (1) at the 30% probability level. tert-Butyl groups are depicted using a wireframe model. Hydrogen atoms, except for the [N−OH] protons, have been omitted for clarity. Selected bond distances (Å): Ce(1)−O(1) 2.221(3), Ce(1)−O(2) 2.221(3), Ce(1)−O(4) 2.223(3), Ce(1)−O(5) 2.215(3), Ce(1)−N(1) 2.622(3), Ce(1)−N(2) 2.649(3), Ce(1)−N(4) 2.628(3), Ce(1)−N(5) 2.666(3), N(1)−O(1) 1.445(4), N(2)−O(2) 1.429(4), N(3)−O(3) 1.464(4), N(4)−O(4) 1.441(4), N(5)−O(5) 1.427(4), N(6)−O(6) 1.463(4).

Table 1. Electrochemical Analysis Data for H3arene-triNOx and 1 wave 2

H3arenetriNOx 1

CeIV/III

wave 1

Epa (V)a

Epc (V)a,b

Epa (V)a,c

Epc (V)a,c

Epa (V)a

Epc (V)a

0.59















−0.32

−0.41

−1.70

−1.93

a

Potentials are referenced to Fc/Fc+ and were obtained at a scan rate of 100 mV/s. bThe reduction potential was difficult to define because of the formation of unstable [NO+] (see the Supporting Information for the isolated scans). cThe potentials were not welldefined because of the association with proton transfer process.

the goal of generating intermediate Pr(IV/III) or Tb(IV/III) character in the resulting oxidized complexes. Rare earth metal complexes of arene-diNOx2− were prepared following the previously reported synthetic routes,5 either protonolysis from RE[N(SiMe3)2]3 (RE = La, Pr, Y) or metathesis from Tb(OTf)3 in dimethoxyethane (DME) (Scheme 3). For both synthetic routes, [K(DME)2][RE(arene-diNOx)2] (2′−RE) was prepared first. Subsequently, the potassium cation was sequestered and crystallized with 18crown-6 and pyridine. For the La, Pr, and Y complexes, the protonolysis starting material RE[N(SiMe3)2]3 was treated with 2 equiv of H2arene-diNOx in DME followed by 1 equiv of KN(SiMe3)2 (Scheme 3). For the metathesis reaction, Tb(OTf)3 was treated with 2 equiv of H2arene-diNOx in DME, followed by the addition of 4 equiv of KN(SiMe3)2. The

Figure 3. Cyclic voltammograms of H3arene-triNOx (top, black trace) and Ce(Harene-triNOx)2 (1) (bottom, purple trace) recorded with 0.1 M [nPr4N][BArF4] supporting electrolyte in methylene chloride.

a bidentate nitroxide ligand, 1,3-bis(2′-tert-butylhydroxylaminoaryl)benzene (H2arene-diNOx). The synthesis of H2arene-diNOx and its tetravalent cerium metal complex Ce(arene-diNOx)2 (Scheme 2 and Figure 4) was reported previously by us,5 starting from either Ce[N(SiHMe2)2]416 or Ce(OtBu)4(py)2.16 Because of the large Ce(IV/III) redox stabilization of Ce(arene-diNOx)2 of −1.74 V versus Fc/Fc+, we sought to expand the synthesis to nitroxide complexes of other rare earth metals, namely, Pr and Tb, with C

DOI: 10.1021/acs.inorgchem.5b02236 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Scheme 2. Preparation of Ce(arene-diNOx)2

products 2′−RE were obtained in yields ranging from 50 to 77% (Scheme 3). [K(DME)2][Tb(arene-diNOx)2] (2′−Tb) was extracted with toluene, and pale-yellow crystals were grown from DME/hexanes and isolated in 50% yield. A similar procedure was employed for the isolation of the related complexes 2′−RE (RE = La, Pr, Y) (Figure 5). The addition of 1 equiv of 18crown-6 to pyridine solutions of the dissolved 2′−RE complexes followed by layering with hexanes yielded yellow crystals of 2−RE in 64−83% yield (Figure 6). Structural characterization of the complexes 2′−RE (RE = Pr, Tb) (Figure 5) revealed one RE(III) metal cation bound by two arene-diNOx2− ligands in each case, one of which interacts with one potassium cation. The potassium cation also includes interactions with two molecules of dimethoxyethane. Because of the association with the electron-withdrawing potassium cation, the bond distances Pr(1)−O(2) and Pr(1)−O(3) at 2.3216(16) and 2.3413(17) Å, respectively, are slightly longer than the potassium-free Pr(1)−O(1) and Pr(1)−O(4) distances at 2.2880(17) Å and 2.2624(16) Å, respectively. Alkali metal cations are often coordinated by 3 or more equiv of DME in the solid state, such that the cations are sequestered.26,27 However, because of the electron-rich nitroxide groups, the potassium cation in 2′−RE shows interactions with the nitroxide groups and only two DME molecules. Such “−ate” complex interactions of a potassium cation with metalbound electron-rich groups have often been observed with electron-donating ligands, such as halides,28 hard O− or N− anions as in alkoxides or amides, 29,30 or even arene interactions.30,31 The measured [N−O−] bond distances of 2′−Pr range from 1.434(2) to 1.443(2) Å, consistent with fully reduced nitroxide groups. The structures of 2′−RE (RE = La, Pr, Y) in solution examined by 1H NMR spectroscopy in pyridine-d5 solvent exhibited two tert-butyl resonances, consistent with those of Ce(arene-diNOx)2. This demonstrates that the coordination modes of the nitroxide groups in solution were indistinguishable, resulting from the sequestration of the potassium ion by pyridine (see the Supporting Information). For the complexes in 2−RE containing the 18-crown-6sequestered potassium cations, in all cases the coordination environment around the RE cation in the solid-state structure includes three nitroxide groups in η2 binding modes and a fourth nitroxide bound end-on through an oxygen atom (Scheme 2, [2−RE]−, and Figure 6). The average Pr(1)−O and Tb(1)−O bond distances for 2−RE were measured as 2.284(4) and 2.205(3) Å, respectively. These values compare well to the related Sm−O distance of 2.226(2) Å in the reported complex [SmCp2{ON(C2H4-o-Pyr)}2]18 and is consistent with the ionic radii differences.20 The N−O bond distances in 2−Pr and 2−Tb range from 1.426(4) to 1.440(5) Å and 1.429(4) to 1.451(4) Å, respectively, indicating again that all of the nitroxide groups are fully reduced.32 The η2

Figure 4. Thermal ellipsoid plot of Ce(arene-diNOx)2·2THF at the 30% probability level. tert-Butyl groups are depicted using a wireframe model. Hydrogen atoms and interstitial solvent have been omitted for clarity. Selected bond distances (Å): Ce(1)−O(1) 2.2208(15), Ce(1)−O(2) 2.2100(15), Ce(1)−O(3) 2.1813(15), Ce(1)−O(4) 2.0940(15), Ce(1)−N(1) 2.5510(18), Ce(1)−N(2) 2.5976(18), Ce(1)−N(3) 2.5931(18), N(1)−O(1) 1.422(2), N(2)−O(2) 1.431(2), N(3)−O(3) 1.439(2), N(4)−O(4) 1.426(2).

Scheme 3. Preparation of [K(DME)2][RE(arene-diNOx)2] (2′−RE) and [RE(arene-diNOx)2]− ([2−RE]−)

D

DOI: 10.1021/acs.inorgchem.5b02236 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 5. Thermal ellipsoid plots of [K(DME)2][RE(arene-diNOx)2] (2′−RE; RE = Pr (left), Tb (right)) at the 30% probability level. tert-Butyl groups are depicted using a wireframe model. Hydrogen atoms and interstitial solvent have been omitted for clarity. Selected bond distances (Å) in 2′−Pr: Pr(1)−O(1) 2.2880(17), Pr(1)−O(2) 2.3216(16), Pr(1)−O(3) 2.3413(17), Pr(1)−O(4) 2.2624(16), N(1)−O(1) 1.434(2), N(2)−O(2) 1.438(2), N(3)−O(3) 1.443(2), N(4)−O(4) 1.439(2), K(1)−O(2) 2.5905(16), K(1)−O(3) 2.7027(17). The low quality of the structural data for 2′−Tb precluded the precise measurement of its geometrical parameters.

Figure 6. Thermal ellipsoid plots of [RE(arene-diNOx)2]− ([2−RE]−; RE = Pr (left), Tb (right)) at the 30% probability level. tert-Butyl groups are depicted using a wireframe model. Hydrogen atoms, interstitial solvents, and the cation species, [(py)2K+(18-crown-6)] or [(py)K+(18-crown-6)]· (py), have been omitted for clarity. Selected bond distances (Å) in [2−Pr]−: Pr(1)−O(1) 2.288(4), Pr(1)−O(2) 2.302(4), Pr(1)−O(3) 2.266(4), Pr(1)−O(4) 2.283(4), N(1)−O(1) 1.422(6), N(2)−O(2) 1.440(5), N(3)−O(3) 1.430(5), N(4)−O(4) 1.439(6). In [2−Tb]−: Tb(1)−O(1) 2.210(3), Tb(1)−O(2) 2.219(3), Tb(1)−O(3) 2.183(3), Tb(1)−O(4) 2.207(3), N(1)−O(1) 1.429(4), N(2)−O(2) 1.451(4), N(3)−O(3) 1.439(4), N(4)−O(4) 1.444(4).

ionic radii as Tb(III) and Pr(III), respectively, were used for control experiments (see Experimental Procedures and the Supporting Information). The CVs of 2−RE (RE = La, Pr, Y, Tb) and CeIV(arene-diNOx)2 were recorded using a 0.1 M [nPr4N][BArF4] methylene chloride solution (Figure 7 and Table 2). In each case, two reversible redox waves were observed. Per our previous assignments for CeIV(arenediNOx)2, the E1/2 of the CeIV/III couple was assigned to the quasi-reversible wave at −1.74 V versus Fc/Fc+. A reversible ligand oxidation wave in CeIV(arene-diNOx)2, [N−O−]/[N− O•], was observed at E1/2 = −0.37 V versus Fc/Fc+. The appearance of the reversible oxidation wave for CeIV(arenediNOx)2 at E1/2 = −0.37 V versus Fc/Fc+ was important because it implied the stability of the oxidized species, the

binding modes were observed in the reported complexes K[RE(ONiPr2)4] (RE = Y, Sm),33 [RE(TriNOx)(thf)] (RE = La, Nd, Dy, Y; thf = tetrahydrofuran),34 and Ce[η2-ON(tBu)(2OMe-5-tBu-C6H3)]4.17 Similar to the 1H NMR tert-butyl resonances of Ce(arene-diNOx)2 and 2′−RE (RE = La, Pr, Y), the 1H NMR spectra of 2−RE (RE = La, Pr, Y) exhibited resonances for two chemically inequivalent tert-butyl groups at 1.58 and 1.13 ppm (La), −5.07 and −11.75 ppm (Pr), and 1.60 and 1.09 ppm (Y), indicating C2-symmetric solution structures within the metal complexes on the NMR time scale (see the Supporting Information). Cyclic voltammetry was performed on the compounds 2−RE in order to study their stabilities upon oxidation. The redoxinactive yttrium and lanthanum complexes, which have similar E

DOI: 10.1021/acs.inorgchem.5b02236 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

The reversibility of the first [N−O−]/[N−O•] redox wave of 2−Tb indicated that the oxidized ligand was stable and that the integrity of the complex was maintained on the electrochemical time scale. Therefore, we pursued chemical oxidation of 2−Tb to establish its chemical stability against disproportionation and probe its electronic structure. After addition of AgOTf to a solution of 2−Tb in CH2Cl2, black Ag(0) solid precipitated and potassium triflate was detected by 19F NMR spectroscopy. However, because of the similar solubilities of [(py)2K+(18-crown-6)]OTf with the resulting metal complex, purification was problematic. As a result, Tb(arene-diNOx)2· CH2Cl2 (3·CH2Cl2) was prepared using 2′−Tb (Scheme 4). Complex 2′−Tb was treated with AgOTf in methylene chloride, and black Ag(0) was similarly observed to precipitate immediately. After the reaction mixture was stirred for 1 h, the mixture was filtered and the product extracted with hexanes to remove KOTf and residual Ag(0) or AgOTf. Pale-orange crystals of 3·CH2Cl2 were grown from CH2Cl2/hexanes at −21 °C in 34% yield (Scheme 4 and Figure 8). Notably, the solidstate X-ray characterization of 3·CH2Cl2 (Figure 8) revealed that one of the four Tb−O bond lengths is elongated to 2.353(5) Å, compared with the average Tb−O bond distance of 2.205(3) Å in 2−Tb. The remaining three Tb−O bond distances are 2.172(5) Å, which are similar to those in 2−Tb. These results, together with the absence of cations in the structure, suggested that complex 2′−Tb had indeed been oxidized but that the product consisted of three [N−O−] ligand arms and one localized [N−O•] ligand arm. The bond length of the localized, oxidized nitroxide group, N(4)−O(4), decreased to 1.287(8) Å, consistent with the reported [N− O•] distance in [Tb(hfac)3(2pyNO)] (hfac = 1,1,1,5,5,5hexafluoropentane-2,4-dione, 2pyNO = N-tert-butyl-2-pyridyl nitroxide).35 Compound 3·CH2Cl2 showed no signal by 1H NMR spectroscopy. The electronic structural assignments of both 2−Tb and 3· CH2Cl2 were confirmed through magnetic susceptibility measurements (Figure 9). The temperature dependences of the magnetic moments of these complexes were measured from 300 to 2 K using a 1.0 T field. In the case of 2−Tb, the χT product reached a value of 10.23 emu K mol−1 at 300 K, which is somewhat smaller than the value expected for an isolated TbIII ion (J = 6, χTtheoretical = 11.82 emu K mol−1) but largely consistent with previously reported values for TbIII complexes.36,37 In the case of 3·CH2Cl2, the χT product reached a value of 11.09 emu K mol−1, which again is somewhat smaller than the value expected for the sum of an isolated TbIII ion (J = 6, χTtheoretical = 11.82 emu K mol−1) and ligand radical (S = 1/2, χTtheoretical = 0.375 emu K mol−1). On comparison, the magnetic data for 3·CH2Cl2 did show the expected increase in χT at room temperature over the data for 2−Tb based on the addition of a ligand-based radical for 3·

Figure 7. Cyclic voltammograms of 2−RE (RE = La, Pr, Y, Tb) and Ce(arene-diNOx)2 with 0.1 M [nPr4N][BArF4] in methylene chloride.

putative complex [CeIV(arene-diNOx)2]+, on the electrochemical time scale. Further oxidation to a putative doubly charged Ce(arene-diNOx)22+ complex showed smaller currents for the [N−O−]/[N−O•] cathodic and anodic waves, possibly indicating instability or slow heterogeneous electron transfer kinetics for Ce(arene-diNOx)22+. Nevertheless, in this context, neutral complexes comprising RE(arene-diNOx)2 (RE = Pr, Tb) were also targeted for investigation. In the voltammograms of the 2−La and 2−Y complexes, the two observed waves were attributed to two sequential ligandbased [N−O−]/[N−O•] oxidations at −1.20 and −0.57 V versus Fc/Fc+ for 2−La and −1.11 and −0.61 V versus Fc/Fc+ for 2−Y. These data were important baseline results becauses they established that the complexes of the redox-inactive metals La and Y exhibited reversible, ligand-based redox activity. In regard to the data for 2−Pr and 2−Tb, quasi-reversible waves at −1.26 and −0.58 V versus Fc/Fc+ for 2−Pr and −1.13 and −0.65 V versus Fc/Fc+ for 2−Tb compared with those of 2−La and 2−Y, respectively, indicate that Pr(III)/Pr(IV) and Tb(III)/Tb(IV) redox couples were not observed; rather, all of the waves for 2−Pr and 2−Tb were similarly attributed to ligand oxidation processes. Evidently, in all cases, once one ligand-arm [N−O−] group was oxidized, it was more difficult to oxidize a second [N−O−] group, as the redox potentials of wave 1 were shifted to more oxidizing potentials than those of wave 2. Furthermore, the oxidation potentials for wave 2, −1.14 V versus Fc/Fc+ for 2−Y and −1.13 V versus Fc/Fc+ for 2−Tb, are more positive than those of −1.25 V for 2−La and −1.26 V for 2−Pr because of the increased charged density on the smaller metal cations.

Table 2. Electrochemical Analyses of the Ligand Redox Waves in 2−RE (RE = La, Pr, Y, Tb) and Ce(arene-diNOx)2 wave 1 [N−O−]/[N−O•]

a

wave 2 [N−O−]/[N−O•]

complex

Epa (V)

Epc (V)

ΔE (V)

Epa (V)

Epc (V)a

ΔE (V)

2−La Ce(arene-diNOx)2 2−Pr 2−Y 2−Tb

−0.57 0.48b −0.47 −0.61 −0.59

−0.73 0.53b −0.69 −0.69 −0.71

0.16 0.05b 0.22 0.08 0.12

−1.20 −0.32 −1.19 −1.11 −1.08

−1.30 −0.41 −1.33 −1.17 −1.17

0.10 0.09 0.14 0.06 0.09

a

a

a

Referenced against Fc/Fc+. bObtained by differential pulse voltammetry (see the Supporting Information). F

DOI: 10.1021/acs.inorgchem.5b02236 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry Scheme 4. Preparation of Tb(arene-diNOx)2 (3·CH2Cl2)

CH2Cl2 and 2.90 emu K mol−1 for 2−Tb at 2 K. The fielddependent data at 2 K reached values of 5.27μB for 3·CH2Cl2 and 4.17μB for 2−Tb at 7 T (Figure 9). The difference of 1.10μB in the high-field values, which is close to the expected value of 1.0μB for an additional, noninteracting S = 1/2 ligand radical, strongly supports the electronic structure assignments. The magnetic data clearly support the assignment of Tb(III) ions in the electronic structures of both 2−Tb and 3·CH2Cl2. On the basis of the measurements, the alternative formal electronic structure configuration of a TbIV ion (J = 7/2, χTtheoretical = 7.88 emu K mol−1) and a closed-shell nitroxide ligand (S = 0, χTtheoretical = 0 emu K mol−1) was ruled out. Overall, these results are consistent with localized oxidation occurring at the ligand with the oxidation state of the Tb cation as TbIII in both 2−Tb and 3·CH2Cl2.



CONCLUSIONS In summary, a new multidentate nitroxide ligand, H3arenetriNOx, was prepared, and its tetravalent cerium metal complex was isolated. The resulting metal complex was composed of two Harene-triNOx2− ligands, with two [N−O−] groups from each Harene-triNOx2− bound to the cerium cation. Electrochemical analysis revealed that the four nitroxide groups strongly stabilized the tetravalent state of the metal cation. The ligand framework was adjusted to H2arene-diNOx, containing a pair of nitroxide groups, and a series of rare earth metal−arenediNOx complexes, 2−RE (RE = La, Pr, Y, Tb), were synthesized and characterized. Cyclic voltammetry showed quasi-reversible redox waves with potentials at E1/2 = −0.65 and E1/2 = −1.13 V versus Fc/Fc+, which prompted us to perform a

Figure 8. Thermal ellipsoid plot of 3·CH2Cl2 at the 30% probability level. tert-Butyl groups are depicted using a wireframe model. Hydrogen atoms and interstitial solvent molecules have been omitted for clarity. Selected bond distances (Å): Tb(1)−O(1) 2.147(5), Tb(1)−O(2) 2.202(5), Tb(1)−O(3) 2.167(5), Tb(1)−O(4) 2.353(5), N(1)−O(1) 1.450(7), N(2)−O(2) 1.443(7), N(3)−O(3) 1.440(7), N(4)−O(4) 1.287(8).

CH2Cl2 upon oxidation of 2−Tb. In both cases, the measured χT products decreased steadily to 8.82 emu K mol−1 for 3· CH2Cl2 and 7.88 emu K mol−1 for 2−Tb at 20 K as a result of thermal depopulation of the Stark levels created by crystal field perturbations of the J = 6 manifold. Below 20 K, the χT products decreased precipitously to 3.78 emu K mol−1 for 3·

Figure 9. (left) Temperature-dependent and (right) field-dependent magnetic data for [(py)2K(18-crown-6)][Tb(arene-diNOx)2] (2−Tb) and 3· CH2Cl2. G

DOI: 10.1021/acs.inorgchem.5b02236 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

devised; the X-ray data were corrected for the presence of disordered solvent using SQUEEZE.45 Electrochemistry. Cyclic voltammetry experiments were performed using a CH Instruments 620D electrochemical analyzer/ workstation, and the data were processed using CHI software version 9.24. All of the experiments were performed in a drybox under N2 using electrochemical cells consisting of a 4 mL vial, a glassy carbon working electrode, a platinum wire counter electrode, and a silver wire plated with AgCl as a quasi-reference electrode. The working electrode surfaces were polished prior to each set of experiments. Potentials are reported versus ferrocene, which was added as an internal standard for calibration at the end of each run. Solutions employed during these studies contained ∼1 mM analyte and 0.1 M [nPr4N][BArF4] in methylene chloride. All of the data were collected in a positivefeedback IR compensation mode. Magnetism. Magnetic data were collected on a Quantum Design Multi-Property Measurement System (MPMS-7) with a reciprocating sample option. Temperature-dependence measurements were performed under applied 1 T DC fields from 2 to 300 K, and fielddependence measurements were performed at 2 K with varying applied magnetic field strengths ranging from 0 to 7 T. Corrections for the intrinsic diamagnetism of the samples were made using Pascal’s constants.46 Each magnetism sample was prepared in the glovebox and placed in a heat-sealed compartment of a plastic drinking straw. The plastic drinking straws were evacuated overnight prior to use. These straws were then sealed at one end (∼9.5 cm from the top) by heating a pair of forceps and crimping the sides of the straw until the two sides were fused together. Microcrystalline compound (∼10 mg) was loaded into the straw, capped with ∼10 mg of quartz wool (dried at 250 °C prior to use), and packed in tightly using a Q-tip. The other end of the plastic drinking straw was then sealed directly above the quartz wool, forming a small compartment (