Electronic communication across diamagnetic metal bridges: a homoleptic gallium(III) complex of a redox-active diarylamido-based ligand and its oxidized derivatives

Abstract

Complexes with cations of the type [Ga(L)₂]ⁿ⁺ where L = bis(4-methyl-2-(1H-pyrazol-1-yl)phenyl)amido and n = 1, 2, 3 have been prepared and structurally characterized. The electronic properties of each were probed by electrochemical and spectroscopic means and were interpreted with the aid of DFT calculations. The dication, best described as [Ga(L⁻)(L⁰)]²⁺, and is a Robin-Day class II mixed-valence species. As such, a broad, weak, solvent-dependent intervalence charge transfer (IVCT) band was found in the NIR spectrum in the range 6390 to 6925 cm⁻¹, depending on solvent. Band shape analyses and the use of Hush and Marcus relations revealed a modest electronic coupling, H_ab of about 200 cm⁻¹, and a large rate constant for electron transfer, k_et, on the order of 10¹⁰ s⁻¹ between redox active ligands. The di-oxidized complex [Ga(L⁰)₂]³⁺ shows a half-field ΔM_s = 2 transition in its solid-state X-Band EPR spectrum at 5 K which indicates that the triplet state is thermally populated. DFT calculations (M06/Def2-SV(P)) suggest that the singlet state is 21.7 cm⁻¹ lower in energy than the triplet state.

Introduction

Over the past few decades, the study of mixed-valence (MV) compounds has been pivotal for advancing comprehension of long range electron transfer of importance to both basic biological functions and, potentially, to future molecular electronics applications.¹ A majority² of the MV complexes studied have been of the type Mⁿ⁺-(bridge)-M^{(n 1)+} where the bridge is an organic group such as in the Creutz-Taube ion, [(NH₃)₅Ru^II(μ-pyrazine)Ru^III(NH₃)₅]⁵⁺.³ There has also been a great deal of interest in purely organic systems of the type D-OB-D^+·, where OB is an organic bridge and D/D^+· are the one-electron redox partners of an organic donor.⁴ A popular class of such organic derivatives is those with diarylamine donors that flank and organic bridge, Fig. 1.^5-8 Electronic communication between donor ends of such molecules can vary dramatically by changing: i) the groups, X, along the diarylamine donor;^5c,h-j ii) the type of bridge;^7a,8.9 iii) the bridge length; or iv) the geometric disposition of donors about the bridge,^5i,5j,9 including the dihedral angle between bridging phenylene groups (that also affect the dihedral of orbitals containing the nitrogen lone pair).¹⁰ In cases such as A and B in Fig 1, electronic communication can occur via tunneling, superexchange, or a “hopping” mechanism whereby the bridge becomes an active participant. The latter is important for longer, more highly-conjugated and electron-rich bridges. Both through-bond and through space superexchange interactions become important for short bridges such as found in the tetraanisyl-o-phenylenediamine cation radical.⁹

Figure 1.

Diarylamine-based mixed valent compounds.

An important class of MV complexes is one like Fig 1C (n=1)⁸ that contains organic donors separated by a metal bridge.¹¹ One-electron oxidized or reduced forms of metal dioxolenes,¹² dithiolenes,¹³ diimines,^14-17 o-semiquinones,¹⁸ o-iminosemiquinones,¹⁹ polypyridyls,²⁰ and tridentate catecholates²¹ can all fall into this category. Some important aspects of the chemistry of these and related metal complexes of redox-active ligands were the subjects of a recent special issue of Inorganic Chemistry²² and of several reviews.²³ With relation to the organic derivatives mentioned above, the interjection of the Pt(PEt₃)₂ bridge between (di/tri)arylamine donors, Fig 1C (n=1), permitted weak electronic coupling (H_ab = 350 cm⁻¹) between donor ends, but this coupling was weaker than that found for derivatives where a phenylene (H_ab = 440 cm⁻¹)^5c or a p-dimethoxyphenylene group (H_ab = 520 cm⁻¹)^7a replaces the metal bridge. Thus, despite the former possessing a fewer number of sigma bonds separating donor ends (and a shorter D⋯D^+· separation) than in the pure organic cases, the energetic mismatch between donor and the metal bridge has a small detrimental influence on the electron transfer.

We were interested in further examining how effectively electronic communication could be mediated by using only a single atom bridge between two diarylamido groups. In particular, we recently prepared a series of diarylamines that have a pyrazolyl group situated at an ortho- position of each aryl (Fig. 2).

Figure 2.

Di(2-3R-pyrazolyl)-p-arylamines, H(XY^R).

The coordination chemistry of tricarbonylrhenium(I)²⁴ and rhodium²⁵ complexes showed that these pincer-type ligands are electrochemically active and chemically non-innocent. The electronic properties and reactivity of the complexes could be predictably fine-tuned by substituting at the pyrazolyl, at the para-aryl positions, or even at the metal center. In those studies only one ligand was bound to a metal center. We envisioned constructing molecular wires by assembling strings of M(L = XY^R)₂ complexes together to give species such as LM-[(L-L)M(L-L)]_n-ML (n = 0, 1, 2…). Therefore, it became of interest to examine potential electronic interactions between two ligands across a single metal ion bridge to inform future wire designs. Our investigations began with simple model complexes of redox silent gallium(III) with the added purpose of obtaining structural and spectroscopic markers for ligand-based radicals that should also be of use in future studies that incorporate transition metals. Herein, we report on the preparation and properties of the complete valence series of [Ga(L)₂]ⁿ⁺ complexes (n = 1-3).

Results and Discussion

The reaction between two mol equivalents of “Li(L)” (formed in-situ from Li(n-Bu) and H(L) in THF at −20°C) and one mol equivalent GaI₃ gives blue-luminescent [Ga(L)₂](I), rather surprisingly, as the insoluble product and LiI as the soluble product, a mixture that can be easily separated by filtration. As the signal for iodide oxidation interferes with the ligand oxidation wave in voltammetry experiments (see Supporting Information), an ensuing metathetical reaction between [Ga(L)₂](I), (1)(I), and TlPF₆ afforded [Ga(L)₂](PF₆), (1)(PF₆), in high yield.

Single crystals of (1)(PF₆)·1.75 CH₂Cl₂ suitable for X-ray diffraction were grown by layering hexanes on a CH₂Cl₂ solution and allowing solvents to diffuse. The compound crystallizes with two crystallographically-independent (1)(PF₆) units. Views of the structure of one of the cations are shown in Figure 3. The gallium center in each resides in a compressed octahedral GaN₆ environment as a result of the disparate distances associated with the two types of Ga-N bonds. Those bonds associated with the diarylamido portion of the ligand, Ga-N_Ar, average 1.949(6) Å which is shorter than found in two independent structure determinations of a related hexacoordinate gallium(III) ONO-pincer complex Ga(dbqdi = 3,5-di-tert-butyl-1,2-quinone-1-(2-hydroxy-3,5-di-tert-butyl-phenyl)imine)₂ (avg. 2.020(3) Å^21d and avg. 2.027(3) Å^21e). As expected, the Ga-N_Ar bonds in the current six-coordinate complex are longer than those in three- or four- coordinate diphenylamidogallium(III) complexes which range from 1.85-1.91 Å.²⁶ The gallium-nitrogen bonds in (1)⁺ associated with pyrazolyl groups, Ga-N_pz, range from 2.085(2) Å to 2.141(3) Å and average 2.101 Å. These values are in good agreement with six-coordinate tris(pyrazolyl)borate complexes of gallium(III).²⁷ Notably, in (1)⁺ the amido nitrogen atoms are planar with sum of angles around each of 360°. The six-membered chelate rings (avg. N_pz-Ga-N_pz bite angle, 88(1)°) are non-planar such as to allow the diarylamido NC₂-moieties to be nearly coplanar across the gallium bridge. That is, there is a small dihedral angle of 16.6(8)° between the mean plane containing C1-N1-C31 and that containing C41-N41-C71 (Fig 3, right). Thus, the nitrogen p- orbitals containing the lone-pair electrons are expected to be roughly parallel with each other but are separated by 3.897(3) Å (avg. N⋯N distance). This geometry is in contrast to the case of the ONO-pincer complex, Ga(dbqdi)₂ whose five-member (planar) chelate rings force the two ligands to be orthogonal, with the dihedral angle of 87.05° between mean planes containing the C-N-C atoms. ^21d,e

Figure 3.

Views of one of the two crystallographically-independent cations [Ga(L)₂]⁺, (1)+, in the crystal of (1)(PF₆)·1.5CH₂Cl₂ (left) with a view approximately down the N1-Ga1-N41 vector showing the small dihedral angle between C1-N1-C31 and C41-N41-C71 planes. Selected bond distances (Å): Ga1-N1, 1.947(2); Ga1-N41, 1.953(2); Ga1-N11, 2.099(2); Ga1-N21, 2.094(2); Ga1-N51, 2.101(2); Ga1-N61, 2.085(2). Selected bond angles (°): N1-Ga1-N41, 179.05(11); N11-Ga1-N21, 178.60(9); N51-Ga1-N61, 177.85(9); N1-Ga1-N11, 90.00(10); N1-Ga1-N21, 89.34(9); N41-Ga1-N51, 89.08(10); N41-Ga1-N61, 88.85(10); N11-Ga1-N51, 92.93(9); N11-Ga1-N61, 86.49(10); N21-Ga1-N51, 85.85(9); N21-Ga1-N61, 94.75(9).

Representative cyclic voltammograms of the free ligand, H(L), and of (1)(PF₆) in CH₂Cl₂ are given in Figure 4 while a summary of electrochemical data of (1)(PF₆) in three different solvents is provided in Table 1. The voltammogram of H(L) in CH₂Cl₂ shows a single irreversible oxidation wave with an anodic peak at ca. 1.2 V versus Ag/AgCl (i_a/i_c > 1) whereas that of [Ga^III(L⁻)₂](PF₆) in this solvent shows two overlapping, reversible, one-electron oxidation waves at 0.94 and 1.17 V versus Ag/AgCl. Since gallium(III) cannot be oxidized to gallium(IV), the oxidation waves are unequivocally identified as ligand based affording [Ga^III(L⁻)(L⁰)]²⁺, (2)²⁺, and [Ga^III(L⁰)₂]³⁺, (3)³⁺, respectively. The close proximity of the two ligands connected by a one-atom spacer can give rise to two oxidation waves by simple coulombic means and/or by electronic communication via superexchange or hopping mechanisms. Coulombic interactions do not have a spectroscopic marker whereas electronic communication (via superexchange or hopping) leaves a signature in the form of an intravalence charge transfer (IVCT) band which is indeed observed in the current case, vide infra. The equilibrium constant for comproportionation according to Eq’s. 1 and 2 is on the order of 10² to 10³ (determined from the electrochemical data in various solvents, Table 1) which indicates a small but significant

$${\left[{\mathrm{Ga}}^{\mathrm{III}}{\left({\mathrm{L}}^{-}\right)}_2\right]}^{+}+{\left[{\mathrm{Ga}}^{\mathrm{III}}{\left({\mathrm{L}}^0\right)}_2\right]}^{3+}\leftrightarrow 2{\left[{\mathrm{Ga}}^{\mathrm{III}}\right({\mathrm{L}}^{-}\left)\right({\mathrm{L}}^0\left)\right]}^{2+}$$ (1)

$$K_{\text{com}}={\left[{\left(2\right)}^{2+}\right]}^2∕\left[{\left(1\right)}^{+}\right]\left[{\left(3\right)}^{3+}\right]$$ (2)

degree of electronic communication in mono-oxidized (2)⁺. The relatively small value of K_com ~ 10³ is one indicator that (2)²⁺ is a Robin-Day class II mixed valent species.^{3, 28}

Figure 4.

Overlay of cyclic voltammograms of H(L) and (1)(PF₆) in CH₂Cl₂ obtained at a scan rate of 200 mV/s.

Table 1.

Electrochemical data for (1)(PF6) in various solvents.

Solvent	E1/2(1), Va,b	E1/2(2), Va,b	ΔE, Vc	Kcom d
CH2Cl2	0.939(3)	1.173(4)	0.184	1.39×103
PCe	0.838(2)	0.994(5)	0.156	4.62×102
CH3CN	0.888(1)	1.065(1)	0.177	1.06×103

^aaverage values obtained for scan rates of 50, 100, 200, and 400 mV/s with 0.1 M NBu₄(PF₆) as supporting electrolyte;

^bV versus Ag/AgCl;

^cΔE = E_1/2(1) - E1/2(2);

^dK_com = e(ΔE·F/RT), T = 295 K;

^epropylene carbonate

Theoretical Studies

In order to gain further insight into the nature of the two oxidation waves and to help rationalize the other experimental properties of the oxidized species, the cations (1)⁺, (2)²⁺, and (3)³⁺ were studied by density functional theory. Four models were examined (M06 or B3LYP functionals with either the LANL2DZ or Def2-SV(P) basis sets), each also accounted for solvation in dichloromethane by employing the polarizable continuum model (PCM). While all gave qualitatively similar trends, the combination (U)M06/Def2 SV(P) gave most satisfactory correlation to experimental data (bond distances and spectroscopic parameters) as summarized in the Supporting Information. The major findings of these studies are summarized below. First, despite missing solvated anions in the theoretical study, a 625 mV difference between first and second oxidation potentials was obtained which parallels the experimental finding of two separate oxidation waves. Second, for the doubly oxidized (3)³⁺, the singlet diradical state was found to be 21.7 cm⁻¹ lower energy than the triplet state. Third, the major structural changes along the valence series occur for Ga-N bonds (despite a lack of participating orbitals on the metal center). Thus, upon successive oxidation, the Ga N-bonds associated with the diarylamido, Ga-N_Ar, lengthen while those associated with the pyrazolyls, Ga-N_pz, shorten. The un-oxidized and di-oxidized complexes are more-or-less symmetric about gallium(III). However, the bond distances associated with each ligand of the mono-oxidized species (2)²⁺ are distinct. One ligand has a longer Ga-N_Ar bond and a shorter average Ga-N_pz distance than the other ligand. In (2)²⁺ the longer Ga-N_Ar bond distance 2.081 A resembles the average distance 2.043 Å calculated for the doubly-oxidized complex (3)³⁺ while the shorter Ga-N_Ar distance of 1.937 Å resembles the average distance of 1.966 Å calculated for the unoxidized complex (1)⁺. The intraligand C-C bond distances also show a similar disparity but the differences between each ligand in (2)²⁺ are much less pronounced than those distances involving gallium. Therefore, examination of the Ga-N bond distances allows one to most easily discern which ligand is oxidized. Electronically, the paramagnetic species are ligand–centered radicals with negligible spin density on the gallium center. Finally, Time-Dependent DFT revealed that in the paramagnetic derivatives, a set of pi-radical bands for β-HOMO(-N= 2-7) to SOMO (β-LUMO) transitions should be observed in the 590-830 nm range. For the mono-oxidized complex (2)^2+, an additional weak (oscillator strength, f, ~ 10⁻³), low-energy intervalence charge transfer (IVCT) band for a β-HOMO-SOMO (β-LUMO, see Figure 5) transition was predicted to be found in the NIR region. Moreover, the IVCT band was predicted to show a small solvent dependence, shifting (473 cm⁻¹) from 4237 cm⁻¹ (2657 nm, f = 6.3×10⁻³) in CH₂Cl₂ to 3764 cm⁻¹ (2360 nm, f = 5.3×10⁻³) in CH₃CN, in line with behavior expected for a Class II mixed valence species.

Figure 5.

β-Frontier orbitals for (2)²⁺ from TD-DFT calculations

By careful choice of organic oxidants it was possible to characterize and isolate either the one- or the two electron oxidation products, (2)²⁺ and (3)³⁺, respectively, as mixed SbCl₆-/PF₆- salts. For example, spectrophotometric titration of (CRET⁺)(SbCl₆⁻)²⁹ (E_1/2 = 1.09 V versus Ag/AgCl, top of Figure 6) with substoichiometric amounts of (1)(PF₆) in CH₂Cl₂ showed the disappearance of the signature bands for the organic oxidant at 486 and 518 nm concomitant with the growth of new bands near 590 and 855 nm for pi-radical transitions of (2)²⁺ [β-HOMO to SOMO]. The reaction was complete after an equimolar ratio of starting materials was achieved verifying the one-electron nature of oxidation of (1)⁺. The shape and energies of these pi-radical bands are nearly identical to those found in the rhenium(I) or rhodium(III) complexes of this oxidized ligand.^24,25

Figure 6.

Preparation of (2)(PF₆)(SbCl₆) and spectrophotometric titration using organic oxidant (CRET⁺)(SbCl₆⁻).

As indicated by the theoretical calculations, an IVCT band was predicted to be found in the NIR spectrum. For a weakly coupled Robin-Day Class II mixed valent species, the IVCT band is expected to have a Gaussian shape, be of weak intensity, and have an energy that is solvent dependent.^1c,3c,30 All of these expectations were met for the IVCT band of (2)(PF₆)(SbCl₆). A representative spectrum for (2)(PF₆)(SbCl₆) dissolved in CH₂Cl₂ is shown in Figure 7 while a summary of data obtained from multiple analyses using Gaussian fits of bands in three solvents (CH₂Cl₂, PC = propylene carbonate, CH₃CN) is given in Table 2. That is, the NIR spectra obtained for bulk samples of (2)(PF₆)(SbCl₆) dissolved in various solvents revealed the presence of a very broad (full-width-at-half-maximum, $\Delta {\tilde{v}}_{1∕2}$, ca. 5000 cm⁻¹), weak-intensity (ε ~ 40-80 M⁻¹cm⁻¹) IVCT band in the range of 6390 to 6925 cm⁻¹ (green band in Fig. 7). It is noteworthy that such a band is absent in the NIR spectra of the doubly-oxidized derivative (3)³⁺ and of all (L^·+)MXYZ complexes (M = Re^I, Rh^III) that contain only one singly-oxidized ligand. It is also worthwhile to note that among the numerous reports on gallium(III) complexes of the type [Ga(L^R)(L^R•)]ⁿ⁺ where L^R = a redox active ligand such as N,N-diazabutadiene = DAB¹⁵ variants, di-tert-butyl semiquinone = DBSQ¹⁸ or dbqdi²¹, an IVCT band has not been observed. Perhaps, the broadness and weak intensity of the IVCT band hinders its identification in these other systems. For (2)(PF₆)(SbCl₆), the Gaussian shape of the IVCT band and the indication of a Robin-Day Class II species from the analysis of K_com suggests that the Hush relations³¹ (Eq. 3 -4) can be used to

$${\mathrm{E}}_{\mathrm{OP}}=\lambda$$ (3)

$$H_{\mathit{ab}}({\mathrm{cm}}^{-1}={\left[\right(4.2\mathrm{x}{10}^{-4}\left){\epsilon }_{\max }\Delta {\tilde{v}}_{1∕2}{\mathrm{E}}_{\mathrm{OP}}\right]}^{1∕2_{∕d}}$$ (4)

estimate the strength of the electronic interaction. Here, E_OP is the energy of the absorption maximum, λ is the Marcus reorganization energy, H_ab is the electronic coupling element, ε_max is the molar extinction coefficient, $\Delta {\tilde{v}}_{1∕2}$ is the full-width-at-half maximum, and d is the separation between redox centers in Å. The value d = 3.9735 was used as this represents the distance between amido nitrogen centers obtained by taking into account an average of all crystallographic data for un-oxidized, mono-oxidized and di-oxidized species in an effort to minimize potential errors of a single point structural determination. The following three observations further support that (2)(PF₆)(SbCl₆) is a Robin-Day Class II(A) mixed valent species. First, from the Gaussian fits of the IVCT band, the experimental $\Delta {\tilde{v}}_{1∕2}$ was larger than the theoretical value $\Delta {\tilde{v}}_{1∕2}$(HTL) = [16ln(2)k_BTλ]^1/2·^3c,30 Second, as predicted by dielectric continuum theory, the energy of the IVCT band showed a linear correlation with the solvent parameter,³² γ = 1/ε_s - 1/n² where ε_s is the static dielectric constant and n is the refractive index of the solvent (Figure S-4). Third, the values of H_ab (ca. 200 cm⁻¹) and λ (6390-6925) cm⁻¹ fall within the accepted limits of 0 < H_ab < λ/2 or 0 < 2H_ab/λ < (1 – [$\Delta {\tilde{v}}_{1∕2}$ (HTL)]2λ) for Class II or Class IIA species, respectively.³⁰ The thermal energy barrier to electron transfer, ΔG*, calculated from classical Marcus Theory³³ (Eq. 5) is 1344 to 1515 cm⁻¹. The corresponding rate constant for electron transfer k_et is found to be on the order of 0.76 – 2.9 × 10¹⁰ s⁻¹ from Eq. 6, where Planck’s constant, h = 3.336×10⁻¹¹ cm⁻¹·s,

$$\Delta {\mathrm{G}}^{\ast }={(\lambda -2{\mathrm{H}}_{\mathrm{ab}})}^2∕4\lambda {\mathrm{cm}}^{-1}$$ (5)

$$k_{\mathrm{et}}=(2{H_{\mathrm{ab}}}^2∕h){[{\pi }^3\lambda ∕\mathit{RT}]}^{1∕2}\exp -(\Delta G^{\ast }∕\mathit{RT})$$ (6)

and the gas constant R = 0.695 cm⁻¹K⁻¹. These k_et values are comparable to those organic cation radicals with diarylamido groups linked by unsaturated 12- to 16- atom (phenylethynyl-) spacers but are of approximately one to two orders of magnitude smaller than found for shorter conjugated spacers such as in Fig 1A and their related N,N’-diphenyl-1,4-phenylenediamine cation radical counterparts.³⁴

Figure 7.

NIR spectrum (blue line) of (2)(PF₆)(SbCl₆) in CH₂Cl₂ showing the IVCT band (green), the lowest energy pi-radical band (grey), and unidentified bands (yellow), and the sum of all Gaussian bands used to fit the spectra (red dotted line).

Table 2.

Summary of IVCT band shape fitting and ET parameters of (2)(PF₆)(SbCl₆) in three different solvents.

	CH2Cl2	PC	CH3CN
EOP = λ (cm−1)	6390 (±20)	6725 (±25)	6925 (±25)
εmaxxs (M−1cm−1)	79 (±3)	44 (±3)	55 (±5)
$\Delta {\tilde{v}}_{1∕2}$ (cm−1)	5192 (±17)	4900 (±100)	4900 (±300)
oscillator strengtha, fobs (fcalc)	1.9×10−3 (6.3×10−3xs)	9.9×10−4 (n.d.)	1.2×10−3 (5.3×10−3)
Hab (cm−1), see Eq. 4	264	196	223
$\Delta {\tilde{v}}_{1∕2}$ (HTL)b	3812	3910	3968
θ = $\Delta {\tilde{v}}_{1∕2}$/$\Delta {\tilde{v}}_{1∕2}$ (HTL)	1.36	1.25	1.23
α = Hab/λ	0.0413	0.0291	0.0322
ΔG* (cm−1), see Eq. 5	1344	1491	1515
ket (s−1), see Eq. 6	2.9×1010	7.6×109	8.6×109
γ = 1/εs - 1/n2	0.382	0.480	0.582

^af_obs = (4.6×10⁻⁹)ε_max$\Delta {\tilde{v}}_{1∕2}$, f_calc from DFT calculations;

^bxy2 (HTL) = [16ln(2)k_BTλ]1/2 where k_B = 0.695 cm⁻¹K⁻¹ and T = 295 K;

Figure 8 shows that the titration of (OMN⁺)(SbCl₆⁻)³⁵ (E_1/2 = 1.39 V versus Ag/AgCl) was complete after ½ equivalent of gallium complex was added to the oxidant verifying the two-electron nature of oxidation. In (3)(PF₆)(SbCl₆)₂, the pi-radical bands persisted in the electronic spectrum indicating a diradical species. The effective magnetic moment of the isolated powder, μ_eff = 2.4 μ_B (295 K), was lower than the expected spin-only value of 2.83 μ_B which suggests that the triplet state is probably not wholly thermally populated. Although we do not have access to a magnetometer capable of variable (low) temperature magnetic measurements that would permit elucidation of the ground state properties, the theoretical calculations of (3)³⁺ suggest that the singlet diradical lies 21.7 cm⁻¹ lower than the triplet. This value is on par with the 23 cm⁻¹ singlet-triplet energy difference in a tin(IV) complex of the aforementioned ONO- pincer radical ion, Sn^IV(dbqdi)₂,^21f or the 64.6 cm⁻¹ 2 difference in Zn(tmeda)(3,6-DBSQ)(3,6-DBCat).^18b The presence of a ‘half-field’ signal for a ΔM_s = 2 transition in the EPR spectra of solid (3)(PF₆)(SbCl₆)₂ acquired at 5K in both normal and parallel-modes (Figure 9) verified that the triplet state is thermally populated even at this low temperature. It is noted that the EPR spectrum of an isolated sample of (2)(PF₆)(SbCl₆) only showed an isotropic signal at g = 2.006, a g-value expected for a ligand-based radical, see Supporting Information.

Figure 8.

Spectrophotometric titration of (3)(PF₆)(SbCl₆)₂ and the organic oxidant (OMN⁺)(SbCl₆⁻).

Figure 9.

(a) X-Band (9.63 GHz, 295 K) EPR spectrum of powder sample of (3)(PF₆)(SbCl₆)₂, (b) ‘half-field’ spectrum acquired at 5 K (100 mW) in parallel-mode.

It was possible to obtain X-ray quality, blue single crystals of the di-oxidized complex (3)(PF₆)₂(SbCl₆)·2.33CH₂Cl₂·toluene after mixing (1)(PF₆) with two equivalents of (NO)(SbCl₆) in CH₂Cl₂, layering with toluene, and allowing solvents to diffuse. Obviously, solubility issues dictated the unexpected ratio of P- versus Sb- centered anions. After numerous attempts, X-ray quality violet crystals of “[Ga(L)₂](PF₆)_1.5·1.05 toluene·0.65CH₂Cl₂·0.17H₂O” were obtained from an equimolar mixture of (1)(PF₆) with (CRET)(SbCl₆) in CH₂Cl₂ layered with toluene, as above. After careful scrutiny of the various bond distances (vide infra), this latter structure is best described as the solvate of [Ga(L⁻)₂](PF₆)/[Ga(L⁻)(L⁰)](PF₆)₂. An overlay of cation structures of (1)⁺, (2)²⁺, (3)³⁺, and an intraligand bond labeling diagram are found in Figure 10. Complete structural data are found in the ESI. As suggested by calculations, the most significant structural changes along the valence series involved the Ga-N bond distances, which serve as oxidation number markers for the ligand. The average gallium-amido nitrogen Ga-N_Ar bond distance increased linearly from 1.947(3) Å in [Ga^III(L⁻)₂]⁺ to 2.023(5) Å in [Ga^III(L⁰)₂]³⁺ (0.074 Å change) while the average Ga-N_pz distance (dative bonds from the pyrazolyls) decreased from 2.102(3) Å in [Ga^III(L⁻₂)]⁺ to 2.039(5) Å in [Ga^III(L⁰)₂]³⁺ (0.063 Å change). As described earlier, each of these distances fall within ranges reported for other gallium(III) diphenylamido²⁶ or pyrazolyl²⁷complexes. The bond length changes within the ligand backbone are much less pronounced, and are at the borderline of statistical significance. The most significant change occurs for bond-type G (right of Fig. 10) between ipso- carbons which on average increases from 1.402(7) Å in [Ga^III(L⁻)₂]⁺ 2 to 1.418(8) Å in [Ga^III(L⁰)₂]³⁺ (0.016 Å change). Such a change would imply a bonding interaction between these atoms in (1)⁺, an interaction that is supported by computational studies.

Figure 10.

Left: Overlay of cation structures from X-ray diffraction. Key: Pale blue, (1)⁺; green, (2)²⁺; purple, (3)³⁺; Right: Labeling diagram for bonds within the ligand.

In summary, the homoleptic complex [Ga(L⁻)₂](PF₆) and its mono- and di-oxidized derivatives have been prepared and characterized in solution and in the solid state. The triplet state of the di-oxidized species was found to be thermally populated even at 5 K. For the paramagnetic, mono-oxidized species (2)(PF₆)(SbCl₆), electrochemical and spectroscopic data established that weak electronic communication occurs between electroactive ligands across the gallium(III) bridge. The electronic communication across the diamagnetic metal ion bridge may occur either by direct tunneling,^33,36 by non-resonant charge transfer using the empty, high-energy 4p orbitals on gallium as a coupling medium (McConnell superexchange³⁷), or by a thermally-activated “hopping” mechanism.³⁸ Given the previous magnetic studies of diamagnetic metal complexes of organic diradicals that can promote either ferromagnetic or anti ferromagnetic interactions with J values of different magnitude depending on the metal,^18b the superexchange mechanism seems to be the most probable pathway for electronic communication. Clearly further experimental and theoretical investigations of other [M(L)₂]ⁿ⁺ complexes of redox silent d¹⁰ or d⁰ metal ions and their oxidized counterparts would be needed to elucidate the mechanism. Nevertheless, if oligomeric assemblies of the type LM-[(L-L)M(L-L)]_n-ML (n = 0,1,2…) can be prepared then wire-like behavior is anticipated even for diamagnetic bridging ions. Even stronger electronic communication is expected for transition metal analogues with available d-orbitals that can engage in dπ-pπ interactions with the ligand. Details regarding such monomeric main group and transition metal complexes and their oligomeric assemblies will be reported in due course.

Experimental

General Considerations

The compounds Li(n-Bu) 1.6 M in hexanes, GaI₃, TlPF₆, (NO)(SbCl₆) were purchased commercially and used as received. The compounds H(L),²⁴ (CRET⁺)(SbCl₆⁻),²⁹ (OMN⁺)(SbCl₆⁻)³⁵ 6 were prepared according to literature procedures. Solvents were dried by conventional means and distilled under nitrogen prior to use.

Physical Measurements

Midwest MicroLab, LLC, Indianapolis, Indiana 45250, performed all elemental analyses. Melting point determinations were made on samples contained in glass capillaries using an Electrothermal 9100 apparatus and are uncorrected. ¹H, ¹³C, ¹⁹F and ³¹P NMR spectra were recorded on a Varian 400 MHz spectrometer. Chemical shifts were referenced to solvent resonances at δ_H 5.33, δ_C 53.84 for CD₂Cl₂ or δ_H 1.94, δ_C 118.9 for CD₃CN and δ_H 2.05, δ_C 29.84 for acetone-d₆ while those for ¹⁹F and ³¹P NMR spectra were referenced against external standards of CFCl₃ (δ_F 0.00 ppm) and 85% H₃PO₄ (aq) (δ_P 0.00 ppm), respectively. Abbreviations for NMR and UV-Vis br (broad), sh (shoulder), m (multiplet), ps (pseudo-), s (singlet), d (doublet), t (triplet), q (quartet), p (pentet), sept (septet). Electrochemical measurements were collected under a nitrogen atmosphere for samples as 0.1 mM solutions in CH₃CN and in CH₂Cl₂, each with 0.1 M NBu₄PF₆ as the supporting electrolyte. A three-electrode cell comprised of an Ag/AgCl electrode (separated from the reaction medium with a semipermeable polymer membrane filter), a platinum working electrode, and a glassy carbon counter electrode was used for the voltammetric measurements. Data were collected at scan rates of 50, 100, 200, 300, 400, and 500 mV/s. With this set up, the ferrocene/ferrocenium couple had an E_1/2 value of +0.53 V in CH₃CN and +0.41 V in CH₂Cl₂ at a scan rate of 200 mV/s, consistent with the literature values.³⁹ Solid state magnetic susceptibility measurements were performed using a Johnson-Matthey MSB-MK1 instrument. Electronic absorption (UV-Vis/NIR) measurements were made on a Cary 5000 instrument. Emission spectra were recorded on a JASCO FP-6500 spectrofluorometer. EPR spectra were obtained on both solid powder samples and as solutions ~0.2 mM in 1:1 CH₂Cl₂:toluene mixtures using a Bruker ELEXYS E600 equipped with an ER4116DM cavity resonating at 9.63 GHz, an Oxford instruments ITC503 temperature controller and a ESR-900 helium flow cryostat. The spectra were recorded using 100 kHz field modulation unless otherwise specified.

Syntheses

[Ga(L)₂](I), (1)(I)

A 3.45 mL aliquot of 1.6 M Li(n-Bu) in hexanes (5.52 mmol) was slowly added via syringe to a solution of 1.814 g (5.51 mmol) H(L) in 15 ml THF maintained at −78°C. The resulting bright yellow solution was stirred 15 min, then a solution of 1.241 g (2.76 mmol) GaI₃ in 5 ml of THF was added by cannula transfer under nitrogen. The mixture was maintained at −78 °C for 2 h, then the cold bath was removed and the mixture was allowed to warm naturally with stirring 12 h. The colorless precipitate (which exhibited bright blue luminescence upon irradiation with 354 nm light) was collected by vacuum filtration and was further dried under vacuum 4 h to leave 2.026 g (86%) of (1)(I) as a colorless powder. Anal. Calcd (obs.) for C₄₀H₃₆N₁₀GaI: C, 56.30 (56.22); H, 4.25 (4.27); N, 16.41 (16.19). ¹H NMR (acetone-d₆) δ_H: 8.40 (d, J = 2.6 Hz, 1H, H₅-pz), 7.28 (s, 1H, H₃-Ar), 7.22 (part of AB, 1H, Ar), 7.21 (d, J = 2.5 Hz, 1H, H₃pz), 7.09 (part of AB, 1H, Ar), 6.39 (ps t, J_app = 2.5 Hz, 1H, H₄pz), 2.25 (s, 3H, CH₃). ¹³C NMR (acetone-d₆) δ_C: 143.8, 140.3. 132.9, 130.9, 130.7, 130.2, 127.2, 123.5, 108.3, 20.3. UV-Vis (CH₂Cl₂): nm (ε, M⁻¹cm⁻¹) 249 (57,800), 267sh (32,300), 322 (24,700), 365 (19,300). Very fine needle crystals were grown by layering a CH₂Cl₂ solution with hexanes and then allowing solvents to slowly diffuse. A sample that was exposed to the atmosphere for a few hours analyzed as (1)(I)·H₂O. Anal. Calcd (obs.) for C₄₀H₃₈IGaN₁₀O: C, 55.13 (55.62); H, 4.40 (4.27); N, 16.07 (15.56).

[Ga(L)₂](PF₆), (1)(PF₆)

A 0.618 g (1.77 mmol) sample of TlPF₆ was added as a solid to a solution of 1.510 g (1.77 mmol) (1)(I) in 20 ml dichloromethane. After the mixture had been stirred magnetically 1h, the colorless solution was separated from the pale yellow precipitate of TlI by filtration through a pad of Celite. The CH₂Cl₂ was removed under vacuum to give 1.52 g (99%) (1)(PF₆) as a pale yellow powder. Mp: 280°C dec. Anal. Calcd (obs.) for C₄₀H₃₆N₁₀F₆GaP: C, 54.97 (55.37); H, 4.15 (4.25); N, 16.03 (15.82). ¹H NMR (acetone-d₆) δ: 8.34 (d, J = 2.6 Hz, 1H, H₅-pz), 7.25 (s, 1H, H₃-Ar), 7.22 (part of AB, 1H, Ar), 7.21 (d, J = 2.4 Hz, 1H, H₃pz), 7.09 (part of AB, 1H, Ar), 6.39 (ps t, J_app = 2.5 Hz, 1H, H pz), 2.24 (s, 3H, CH 13 4 3). C NMR (acetone-d₆): 143.8, 140.3, 132.8, 130.9, 130.7, 130.2, 127.3, 123.4, 108.4, 20.3. ¹⁹F NMR (acetone-d₆): δ_F −72.6 (d, J = 707 Hz). ³¹ FP P NMR (acetone-d₆): δ_P −144.3 (sept, J_P-F = 707 Hz) ppm. UV-Vis (CH₂Cl₂): nm (ε, M⁻¹cm⁻¹) 251 (47,500), 269sh (29,500), 323 (25,500), 366 (19,800). Single crystals of (1)(PF₆)·1.75CH₂Cl₂ used for X-ray diffraction were grown by layering a dichloromethane solution with hexanes and allowing solvents to slowly diffuse overnight.

Oxidation Reactions

[Ga(L)₂](PF₆)(SbCl₆), (2)(PF₆)(SbCl₆)

A colorless solution of 0.1055 g (0.121 mmol) (1)(PF₆) in 10 mL CH₂Cl₂ was added to a red solution of 0.0732 g (0.121 mmol) (CRET)(SbCl₆) in 25 mL CH₂Cl₂. The flask originally containing (1)(PF₆) was washed with another 10 mL CH₂Cl₂ to ensure quantitative transfer to the reaction mixture. After the resulting royal blue solution had been stirred 15 min, solvent was removed under vacuum. The resulting blue residue was washed with three 10 mL portions of hexanes to remove the organic by-product and then was dried under vacuum several hours to leave 0.132 g (90 %) of (2)(PF₆)(SbCl₆) as a blue powder. μ_eff (solid, 295 K): 1.8 ±0.1 μ_B. UV-Vis (CH₂Cl₂): nm (ε, M⁻¹cm⁻¹) 250 (59,200), 321 (25,600), 362 (21,600), 596 (1,100), 857 (5,500), 1490 (90).

Violet needle crystals of [Ga(L)₂](PF₆)_1.5·1.05 toluene·0.65CH₂Cl₂·0.17H₂O (see text) were grown by layering and equimolar mixture of (1)(PF₆) and (CRET)(SbCl₆) in CH₂Cl₂ with toluene and allowing solvents to diffuse in a −20°C freezer.

[Ga(L)₂](PF₆)(SbCl₆)₂, (3)(PF₆)(SbCl₆)₂

A colorless solution of 0.1382 g (0.159 mmol) (1)(PF₆) in 20 mL CH₂Cl₂ was added to a colorless solution of 0.1156 g (0.317 mmol) (NO)(SbCl₆) in 30 mL CH₂Cl₂. After the resulting royal blue solution had been stirred 15 min, solvent was removed under vacuum and the blue residue was dried under vacuum to leave 0.213 g (87 %) of (3)(PF₆)(SbCl₆)₂ as a blue powder. μ_eff (solid, 295 K): 2.4 ±0.1 μ_B. UV-Vis (CH₂Cl₂): nm (ε, M⁻¹cm⁻¹) 605 (1,300), 849 (6,200).

Blue crystals of (3)(PF₆)₂(SbCl₆)·2.33CH₂Cl₂·toluene were obtained from by mixing 15 mg (17 Δmol (1)(PF₆) 13 mg (34 μmol) (NO)(SbCl₆) in 5 mL CH₂Cl₂, layering with 15 mL toluene, and allowing solvents to diffuse.

Computational Studies

DFT calculations were performed with the M06 meta-hybrid GGA functional⁴⁰ using the def2-SV(P) double-zeta basis set.⁴¹ Solvent (DCM) effects were accounted for by using the polarizable continuum model IEFPCM,⁴² as implemented in Gaussian 09.⁴³ The chosen model proved superior over other combinations of functionals (M06 or B3LYP⁴⁴) and basis sets (def2-SV(P) or 6311-G*/LANL2DZ⁴⁵) for reproducing bond distances and spectroscopic data, as summarized in the supporting information. Gas phase structures of the metal complexes were optimized using the initial geometry from X-ray structural studies. Analytical vibrational frequency calculations were also carried out to verify that the optimized geometries were stationary points. Time-dependent DFT methodology was used for excitation energy calculations.⁴⁶

Crystallography

X-ray intensity data from a colorless prism of (1)(PF₆)·1.75CH₂Cl₂ and a dark blue plate of (3)(PF₆)₂(SbCl₆)·2.33CH₂Cl₂·toluene were collected at 100(2) K with a Bruker AXS 3-circle diffractometer equipped with a SMART2⁴⁷ CCD detector (Cu Kα radiation, λ = 1.54178 Å). X-ray intensity data from a violet needle of (2)(PF₆)_1.5·1.05 toluene·0.65CH₂Cl₂·0.17H₂O, were collected at 100(2) K with an Oxford Diffraction Ltd. Supernova equipped with a 135 mm Atlas CCD detector, by using Cu Kα radiation, λ = 1.54178 Å Raw data frame integration and Lp corrections were performed with SAINT+⁴⁷ for the data collected from the Bruker instrument but with CrysAlisPro⁴⁸ for that from the Oxford instrument. Final unit cell parameters were determined by least-squares refinement of 9343 reflections from the data set of (1)(PF₆)·1.75CH₂Cl₂, 19744 reflections from the data set of (2)(PF₆)_1.5·1.05·toluene·0.65CH₂Cl₂·0.17H₂O, and 5460 reflections from data set of (3)(PF₆)₂(SbCl₆)·2.33CH₂Cl₂·toluene, with I > 2σ(I) for all cases. Analysis of the data showed negligible crystal decay during collection in each case. Direct methods structure solutions, difference Fourier calculations and full-matrix least-squares refinements against F² were performed with SHELXTL.⁴⁹ Numerical absorption corrections based on the real shapes of the crystals for (1)(PF₆)·1.75CH₂Cl₂, and (3)(PF₆)₂₍SbCl₆)·2.33CH₂Cl₂·toluene were applied using SADABS while an empirical absorption correction using spherical harmonics implemented in the SCALE3 ABSPACK scaling algorithm was used for(2)(PF₆)_1.5·1.05·toluene·0.65CH₂Cl₂·0.17H₂O. The carbon atoms of the highly disordered solvent molecules in each structure were refined with isotropic displacement parameters. The remaining non-hydrogen atoms were refined with anisotropic displacement parameters. Hydrogen atoms were placed in geometrically idealized positions and included as riding atoms. The X-ray crystallographic parameters and further details of data collection and structure refinements are presented in Table 3.

Table 3.

Crystallographic data collection and structure refinement for (1)(PF6)·1.75CH₂Cl₂, (2)(PF₆)_1.5·1.05 toluene·0.65CH₂Cl₂·0.17H₂O , and (3)(PF₆)₂(SbCl₆)··2.33CH₂Cl₂·toluene.

Compound	(1)(PF6)·1.75CH2Cl2	(2)(PF6)1.5·1.05 toluene·0.65CH2Cl2·0.17H2O	(3)(PF6)2(SbCl6)·2.33CH2Cl2·toluene
Formula	C41.75H39.50Cl3.50F6GaN10	C47.99H45.68Cl1.3F9GaN10O0.17P1.5	C49.33H48.66Cl10.66F12GaN10P2Sb
Formula weight	1020.10	1098.43	1640.80
Crystal system	triclinic	monoclinic	orthorhombic
Space group	P -1	P 21/c	Pbca
Temp. [K]	100(2)	100(2)	100(2)
a [Å]	12.9440(3)	17.6021(3)	17.5543(5)
b [Å]	17.4584(4)	24.6732(3)	24.9500(6)
c [Å]	20.9702(5)	23.2482(4)	29.1682(8)
α [°]	73.149(2)	90	90
β [°]	85.8230(10)	108.0987(18)	90
γ [°]	79.4170(10)	90	90
V [Å3]	4457.29(18)	9597.2(3)	12775.1(6)
Z	4	8	8
Dcalcd. [gcm−3]	1.520	1.520	1.706
λ [Å] (Mo or Cu Kα)	1.54178	1.54178	1.54178
μ.[mm−1]	3.716	2.644	9.151
Abs. Correction	numerical	numerical	numerical
F(000)	2078	4492	6526
θ range [°]	3.47 to 67.37	3.30 to 71.02	3.03 to 68.05
Reflections collected	37225	53008	107137
Independent reflections	14703 (Rint 0.0203)	18009 (Rint 0.0384)	11388 (Rint 0.0813)
T_min/max	0.4226 / 0.6243	0.68/0.963	0.2619/0.7110
Data/restraints/ parameters	14703/63/1231	18009 /48/1285	11388/15/790
Goodness-of-fit on F2	0.982	0.914	1.062
R1/wR2[I>2σ(I)]a	0.0484/0.1257	0.0433/0.1067	0.0646/0.1433
R1/wR2 (all data)a	0.0536/0.1294	0.0731/0.1163	0.0853/0.1528
Largest diff. peak/hole / e Å−3	1.763/-0.730	0.84/-0.57	1.79/-1.28

^aR1 = σ||F_o| – |F_c||/σ|F_o| wR² = [σw(|F_o| – |F_c|)²/σw|F_o|²]^1/2.