2.1 Synthesis and electrochemical studies

In order to prepare the symmetrical azino-DSeDAF 1, we followed Hünig's strategy which consists of the reaction of quaternary heterocyclic salts with a half equivalent of hydrazine in a basic medium (Scheme 1) [3]. First, we prepared 3-methyl-2-methylthio-benzoselenazolium salt 3 from 2-methylthio-benzoselenazole 2 [6] with a new one-step approach in 91% yield. The synthesis of N,N′-bis(benzo-3H-selenazol-2-ylidene)-hydrazine (azino-DSeDAF) 1 was performed by using salt 3 in the presence of NEt₃ and half an equivalent of hydrazine hydrate, affording azino-DSeDAF 1 in 55% yield. For the synthesis of unsymmetrical azino donors, containing one selenazole and one thiazole cores, we first prepared the appropriate heterocyclic hydrazones 4 and 5 [5] which are consequently coupled with selenazolium salt 3 in basic medium as depicted in Scheme 1. Indeed, the reaction between selenazolium salt 3 and 3-methyl-3H-benzothiazol-2-one hydrazone 4 [7] in MeCN/EtOH in the presence of NEt₃ afforded azinodibenzoselenathiadiazafulvalene (azino-dibenzoSeTDAF) 6 in 70% yield. Treatment of 3,4,5-trimethyl-3H-thiazol-2-one hydrazone 5 with salt 3 under the same conditions resulted in the formation of azino-benzotetramethylselenathiadiazafulvalene (azino-benzoTMSeTAF) 7 which was isolated in 83% yield.

Electrochemical investigations were performed using tetrabutylamonium hexafluorophosphate as supporting electrolyte. In all cases, cyclic voltammetry revealed two monoelectronic reversible oxidation waves with a difference between the two oxidation potentials ΔE larger than 450 mV (ΔE = E₂ − E₁, K ≥ 4.23 × 10⁷ Table 1). This indicates a very high thermodynamic stability of the cation radical species induced by the azino spacer group as for example the parent dibenzo-DSeDAF exhibits a smaller ΔE (K = 5.1 × 10², K = [D⁺]²/[D][D²⁺]). Comparison of the redox potentials between dibenzo-DSeDAF (E_pa¹ = −0.07 V and E_pa² = 0.09 V vs SCE in CH₂Cl₂) [6a] and azinodibenzo-DSeDAF 1 indicates that the presence of the azino spacer group induces an important anodic shift of the two oxidation potentials. Therefore, 1 is stable under atmospheric conditions while dibenzo-DSeDaF oxidizes readily upon air exposure. Furthermore, substitution of one selenium atom in the azino-dibenzo-DSeDAF 1 with sulfur one does not significantly modify the donor ability, since for 5 E₁ is only 20 mV lower than for 1. Similarly, the effect on ΔE is negligible since a difference of 10 mV is observed (Table 1). The most pronounced effect on E₁ was observed when one benzoselenazole core in 1 was replaced with the 3,4,5,-trimethyl-thiazole one. Indeed, azino-benzoTMSeTDAF 6 has the best electron donating ability in the synthesized azino series (E₁ = 0.29 V vs SCE). This unambiguously demonstrates that benzo-fusion decreases the donor ability.

2.2 TCNQ complexes

In order to form charge transfer salts with these azino derivatives, we tried both chemical and electrochemical pathways. For the chemical approach, we simply mixed a hot solution of the donor in CH₂Cl₂ with a hot solution of TCNQ in CH₃CN. Treatment of azine 1, 6 and 7 solutions with TCNQ resulted in the formation of poorly soluble azine–TCNQ complexes which precipitated from the medium as microcrystalline materials in high yields for 1 and 6 (82% and 84% yields, respectively) while in low yield for 7 (19%). Elemental analysis shows that 1:1 complexes were formed in all three cases. FT-IR spectra revealed that all three isolated complexes exhibit a single sharp ν_CN band at 2218 cm⁻¹ for (1)(TCNQ) and (6)(TCNQ) and at 2212 cm⁻¹ for (7)(TCNQ). Based on the nitrile stretching absorption band, it is possible to infer the degree of charge transfer in TCNQ complexes [8]. The value obtained for the three complexes are close to the one observed for neutral TCNQ ν_CN at 2224 cm⁻¹ which indicates that these materials are neutral charge transfer complexes [9]. This was confirmed by EPR experiments as no EPR signal was observed from room temperature to liquid helium temperature. This result is consistent with the presence of neutral complexes.

Crystals were obtained for complex (1)(TCNQ) and the crystal structure determination confirms a stoichiometry of one donor 1 for one TCNQ. It crystallizes in the triclinic system, space group P-1 with both the donor and TCNQ molecules located on an inversion center. Within this complex, donor molecules form alternated stacks with TCNQ. The plane of the donor is not fully parallel to the plane of the acceptor but forms an angle of 2.9(3)° with an interplanar separation of 3.47 Å (Fig. 1). This organization is reminiscent of what we previously observed with (azino-dithiadiazafulvalenes)(TCNQ) complexes [5]. Analysis of the bond lengths in the TCNQ skeleton shows that they are essentially similar to those reported for the neutral TCNQ [10] (see Table 2), confirming that the complex (1)(TCNQ) is neutral. Actually, the potential difference between the reduction potential of TCNQ (0.18 V) and the oxidation potential of 1 (0.56 V) is too large for allowing any redox reaction. Thus, in this case, we are dealing with co-crystallization of the two components rather than with an electron transfer process.

2.3 Electrocrystallization

Then, we focused our attention on the electrochemical pathway which consists in using the electrocrystallization technique, an invaluable tool for the elaboration of ordered electroactive molecular charge transfer salts [11]. Deep blue needles grew on the anode within a couple of days from donor 1 by using n-Bu₄NPF₆ as the supporting electrolyte in CH₂Cl₂. The structure of the salt was determined by X-ray diffraction studies and revealed a stoichiometry of one donor for one PF₆⁻ anion, the first example of isolated cation radical salts in these series.

The salt (1)(PF₆) crystallizes in the monoclinic system, space group C2/c with one donor molecule and one PF₆⁻ anion both in general position in the unit cell, hence a 1:1 stoichiometry. Within this salt, the oxidized donor molecules form segregated stacks (Fig. 2a). Two types of overlap are observed as depicted in Fig. 2b. Mean plane distances between donor molecules within one stack are, respectively, of 3.37 Å and 3.64 Å for overlap A and B. Furthermore, the donor core in its cation radical state is planar, as the structure of the donor 1 under neutral state in (1)(TCNQ). On the other hand, modification of the bond lengths of the donor skeleton is observed in the oxidized donor core essentially on the central spacer group. In the oxidized species the double bond CN is lengthened while the central N–N bond is shortened when compared with the neutral molecule (Table 3). The C–Se and C–N cyclic bonds are also shortened in the oxidized species compared with the neutral one.

The EPR study was carried out on a polycrystalline sample of (1)PF₆. A single Lorentzian shaped line was observed at g = 2.007 with a linewidth of 40 G at room temperature. Since PF₆⁻ is diamagnetic, this signal is due to the radical cation of 1. The temperature dependence of the signal was studied using the well-known approximation that spin susceptibility is proportional to the product: A(ΔB_pp)², where A is the line amplitude and ΔB_pp – peak-to-peak linewidth. The result is displayed in Fig. 3 as χ_sT. From 70 K to ∼200 K spin susceptibility, within the experimental error, is constant, i.e. the Curie law is obeyed. Above 200 K there is a thermally activated slow increase in susceptibility which is a sign of a singlet–triplet behaviour associated with the dimerization of cations observed in the X-ray crystal structure (Fig. 2). Similarly, the linewidth decreases with lowering temperature from 40 G at room temperature to ∼20 G at 200 K. Further decrease in temperature does not change the linewidth significantly (Fig. 4). Both χ_sT and ΔB_pp temperature evolution confirm a thermally activated susceptibility above 200 K. One can estimate singlet–triplet energy gap Δ/k_B of ca. 900 K (630 cm⁻¹) (Fig. 3, right).

2.4 Molecular geometry optimization

DFT calculations [B3LYP/LANL2DZ] were performed on the azinodibenzo-DSeDAF 1 and the oxidized species 1⁺, and, for comparison purpose, on the dibenzo-tetraselenafulvalene (TSeF) and the dibenzo-DSeDAF in order to determine the highest occupied molecular orbitals (HOMO) energies as well as their electron density contours. As shown in Fig. 5, the energy of the HOMO of 1 (E = −4.89 eV) is located in between those of dibenzo-TSeF (E = −5.00 eV) and dibenzo-DSeDAF (E = −4.42 eV), but closer in energy to the TSeF system. This evolution of the HOMOs was anticipated from their redox potentials determined by cyclic voltammetry, as for instance the redox potentials of the dibenzo-TSeF, E₁ = 0.78 V and E₂ = 1.17 V vs SCE in CH₂Cl₂, are closer to the redox potential of 1 compared with the dibenzo-DSeDAF [12]. These energies of the HOMO also explain the higher oxidation potentials observed for the 1 derivatives compared with DSeDAF. The topology of the HOMO for the selenazole rings in 1 and the dibenzo-DSeDAF is similar showing antibonding interactions between the selenium or the nitrogen and the carbon atoms. Concerning the conjugated spacer group in 1, bonding interactions are observed between the nitrogen and the carbon atoms while an antibonding interaction connect the two nitrogen atoms inducing a destabilization of the HOMO and generating a higher oxidation potential than for dibenzo-DSeDAF. It is interesting to note the small coefficient at the C4 carbon atom of the selenazole rings on both the DSeDAF and the azino derivative. This small coefficient on the carbon located alpha to the NMe group has already been noticed on the thiazole ring of sulfur analogues, DTDAF [13]. Optimized bond lengths calculated for azinodibenzo-DSeDAF 1 and the oxidized species 1⁺ are reported in Table 3. If we compare the optimized geometry and bond lengths calculated for 1 with the X-ray data in (1)(TCNQ), we can observe that there is a good agreement for the neutral state of the donor in the complex, (1)(TCNQ). Moreover the optimized bond length calculated for 1⁺ shows that oxidation affects essentially the bonds in which the electron density in the HOMO is concentrated, in agreement with the experimental evidence (1)(PF₆).

4.1 General

¹H NMR spectra were recorded at 300 MHz and ¹³C NMR spectra at 75 MHz with tetramethylsilane as internal reference. Mass spectra were carried out at Centre de Mesures Physiques de l'Ouest, Rennes. Melting points were measured using a Kofler hot stage apparatus and are uncorrected. Elemental analysis results were obtained from the Laboratoire Central de Microanalyse du CNRS (Lyon). Cyclic voltammetry was carried out on a 10⁻³ M solution of metal complex derivative in CH₂Cl₂ containing 0.1 M n-Bu₄NPF₆ as supporting electrolyte. Voltammograms were recorded at 0.1 V s⁻¹ on a platinum disk electrode (1 mm²). Potentials were measured vs Saturated Calomel Electrode (SCE). The EPR measurements were performed on Bruker ESP-300E X-band spectrometer equipped with a helium Oxford kriostat using liquid nitrogen as cooling agent.

4.2 Synthesis and characterization

4.3 Electrocrystallization

Into a 30 mL H-shaped glass cell with a fine frit dividing the anodic and cathodic compartments equipped with platinum wire electrodes was placed a solution of tetrabutylammonium hexafluorophosphate (∼3 g) in CH₂Cl₂ (30 mL). The azine donor 1 (10 mg) was added into the anode compartment. The solution was electrolyzed under a constant current of 2 μA at room temperature. Blue crystals of radical cation salts gradually grew on the anode electrode within several days. The crystals were collected by filtration, washed with cold CH₂Cl₂ and dried. The obtained crystals were utilized for X-ray crystallographic analyses.

4.4 Crystallography

Single-crystal diffraction data were collected on a Nonius KappaCCD diffractometer from the University of Ljubljana for complex (1)(TCNQ) and on a Stoe IPDS diffraction system from the University of Angers for compound (1)(PF₆). Details of the crystallographic are given in Table 4.