A fascinating multifaceted redox-active chelating ligand: introducing the N,N′-dimethyl-3,3′-biquinoxalinium “methylbiquinoxen” platform

Abstract

To intimately combine a chelating ligand function with the numerous properties of a viologen-like redox-active centre would offer a rare possibility to design controllable multi-redox states, whose properties arise from strongly correlated phenomena between the organic ligand as well as with any metalloid coordinated centres. Such a concept previously proved to be feasible, however is not widely applicable owing to challenges in terms of synthesis, isolation, and aerial sensitivity of both the ligand and its metal complexes. Here we report the first stable example of such a redox-active molecule, N,N′-dimethyl-3,3′-biquinoxalinium^2+/˙^+/0 “methylbiquinoxen, MBqn^2+/˙^+/0”, which shows a rich redox chemistry and chelates a metal ion in the case of the metal complex [CdCl₂(MBqn⁰)]. This goes beyond what is possible to achieve using viologens, which are limited by not providing chelation as well as having no accessible biradicaloid state, corresponding to the neutral direduced MBqn⁰ open-shell behaviour we observe here.

---

Introduction

Organic molecules with multiple oxidation levels accessed by electro-, photo-, or chemical means are of extreme importance in many biological, chemical and physical processes.^1–6 Among these, viologens, or N,N′-disubstitued-4,4′-bipyridinium (V²⁺) species, have attracted most attention over the last 80 years. Acting as strong electron-acceptors, viologens undergo two reversible one electron reduction steps leading to a radical cation intermediate (V˙⁺) and neutral form (V⁰).^7,8

The interest in viologen derivatives is motivated, amongst other things, by the ease of synthesis and tunability, the high degree of control concerning their reversible oxidation–reduction process, the range of colours from colourless (V²⁺) to dark green/blue/purple (V˙⁺), the paramagnetic nature of V˙⁺, and their strong structural templating effects (V²⁺).⁹ These possibilities have led to their incorporation in a range of donor–acceptor complexes developed for various applications such as photo- and electrochromic displays,^10,11 charge accumulation,^12,13 fluorescence,^14,15 supramolecular host–guest complexes and molecular machines^4,16 as well as ferroelectrics.¹⁷

Research on viologens is mainly focused on two aspects. The first of these is the diversification of their properties achieved through incorporation of additional functions. The second aims to improve their longevity, in particular through improving the chemical and air stability of the radical intermediates. In this context, Kaim's group have replaced some carbon atoms with nitrogen atoms within the aromatic heterocycles, leading to a new type of redox-active ligand that intimately combines a viologen-like centre with the chelating function of 2,2′-bipyridine.^18–20 Despite an improved π-electron deficiency and an interesting coordination function that provides strong electronic coupling between the coordinated site (typically a metal ion) and the organic redox centre, synthetic challenges and persistent chemical and air instability, render these conceptually interesting molecules currently not useful for applications.

In the context of the potential for radical-based systems and the challenges to be addressed, we report here on the synthesis and characterization of a new stable reversible organic redox-active chelating ligand N,N′-dimethyl-3,3′-biquinoxalinium (MBqn²⁺, 1²⁺, Fig. 1 – grey), “methylbiquinoxen”. In particular the dication 1²⁺ and its monoprotonated direduced form (MBqn⁰-H⁺, 1⁰-H⁺, Fig. 1 – purple) are thermally stable (1²⁺ and 1⁰-H⁺ decomposes at T ≈ 500 K and T ≈ 440 K respectively, Fig. S2 and S4†), and neither moisture nor air-sensitive in the solid state even without complexation to a metal ion. As a consequence, BF₄⁻ salts of these can be prepared and handled as simple crystalline starting pro-ligands. A seemingly facile and reversible chemical reduction at room temperature leads to a doublet radical intermediate which is stable in solution under inert atmosphere (Fig. 1 – blue).

Fig. 1 Path for the synthesis of [1²⁺](BF₄)₂, [1⁰-H⁺]BF₄ and [CdCl₂(1⁰)]. a = from ref. 21; b = [1˙⁺-H⁺]BF₄ not isolated in the solid state. The coloured stickers indicate the relative redox state of the methylbiquinoxen. Grey is for the dication, blue for the doublet radical, and purple for the direduced form.

Using solvothermal conditions under inert atmosphere, the doubly reduced form MBqn⁰ (1⁰) has been isolated in two crystalline forms, firstly as a cation in the monoprotonated species [1⁰-H⁺]BF₄ and secondly as a neutral biradicaloid chelating ligand in the metal complex [CdCl₂(1⁰)]. Noteworthy is that dissolution of [1⁰-H⁺]BF₄ (Fig. 1 – purple) in DMF under inert atmosphere generates the same doublet radical species in solution (Fig. 1 – blue), thus granting the methylbiquinoxen three different accessible redox states which are chemically reversible (vide infra).

Both 1²⁺ and 1⁰-H⁺ have been characterized by X-ray diffraction analysis (single crystal, powder), UV-Vis-NIR, IR and mass spectrometry, thermogravimetric analysis, cyclic voltammetry, spectroelectrochemistry measurements, as well as DFT and TDDFT (Time Dependent Density Functional Theory) calculation. Additional SQUID, NMR and EPR studies of 1⁰-H⁺ in the solid state (SQUID) and dissolved in DMF (SQUID, NMR, EPR) support the presence of the doublet radical intermediate species in solution.

Results and discussion

Synthetic procedures

General procedure for synthesis of compounds [1²⁺](BF₄)₂, [1⁰-H⁺]BF₄, and [CdCl₂(1⁰)] (Fig. 1). Compound [1²⁺](BF₄)₂ was obtained from Meerwein alkylation by mixing biquinoxaline²¹ with an excess of trimethyloxonium tetrafluoroborate in acetonitrile. After 5 d stirring at room temperature, the resulting solid was isolated by filtration, washed with Et₂O, and recrystallized from water to give pale yellow/off-white crystals in 52% yield. Crystals of the direduced monoprotonated species [1⁰-H⁺]BF₄ as well as of the metal complex [CdCl₂(1⁰)] were obtained from solvothermal reaction using the solvent (here methanol) as the reducing agent. In a Schlenk tube, equipped with a hermetic Teflon stopper, [1²⁺](BF₄)₂ were mixed with a mixture of MeOH/DMF. The Schlenk was purged with nitrogen and inserted in a solvothermal oven to follow a heat/flat/cool temperature programme, resulting in a pure phase of dark purple crystals of [1⁰-H⁺]BF₄ in 68% yield. Phase purity of [1²⁺](BF₄)₂ and [1⁰-H⁺]BF₄ was checked by XRPD (Fig. S1 and S3†). For the synthesis of the cadmium complex the same precedent procedure was followed by adding a stoichiometric amount of CdCl₂ in the Schlenk tube. A violet crystal of [CdCl₂(1⁰)] was selected from a mixture of phases which always result from the reaction. In spite of many attempts it was only possible to generate the intermediate doublet radical species in solution, but not to isolate it as a crystalline salt.

Crystallography

Single crystals suitable for X-ray diffraction analysis were isolated for the dication [1²⁺](BF₄)₂ (Fig. 1 – grey) for the direduced monoprotonated form [1⁰-H⁺]BF₄ (Fig. 1 – purple) as well as for the metal complex [CdCl₂(1⁰)]. A summary of the crystallographic data and refinement results are listed Table 1. Datasets for both structures were collected at 180 K. They were refined in the space groups P1̄ for [1²⁺](BF₄)₂ and [CdCl₂(1⁰)], and P2₁/n for [1⁰-H⁺]BF₄. All atoms including hydrogens were identified from Fourier difference maps. All atoms occupy general positions.

Table 1 Summary of crystallographic data for [1²⁺](BF₄)₂, [1⁰-H⁺]BF₄ and [CdCl₂(1⁰)]

	[1^2+](BF4)2	[1^0-H^+]BF4	[CdCl2(1^0)]
Fw (g mol^−1)	461.97	376.16	471.85
Space group	P1̄	P21/n	P1̄
a, Å	7.5470(8)	7.1330(9)	7.4683(5)
b, Å	7.8860(7)	20.0360(16)	10.4739(7)
c, Å	8.9920(9)	11.5260(13)	11.3249(8)
α, °	66.005(7)	90	80.150(6)
β, °	85.305(8)	99.454(9)	82.274(5)
γ, °	83.508(8)	90	81.982(6)
V, Å^3	485.40(8)	1624.9(3)	858.70(10)
Z	1	4	2
T (K)	180	180	180
Obs. reflns (I > 2σ(I)) [Rint]	1527 [0.021]	2273 [0.042]	3386 [0.028]
No. of parameters	177	313	291
R1(I > 2σ(I))/wR2 (all data)	0.044/0.125	0.042/0.126	0.027/0.067
CCDC number	1043174	1043175	1437399

---

The crystallographic asymmetric units of the structures of [1²⁺](BF₄)₂ and [1⁰-H⁺]BF₄ are shown in Fig. 2. The asymmetric unit of [1²⁺](BF₄)₂ contains one independent half [MBqn²⁺] (1²⁺) (N-methylquinoxalinium MQ⁺ type) and one independent ordered BF₄⁻. On the other hand, the asymmetric unit of [1⁰-H⁺]BF₄ contains one independent [MBqn⁰-H⁺] (1⁰-H⁺), as well as one independent ordered BF₄⁻ per compound.

Fig. 2 Crystallographic structures of 1²⁺/1⁰-H⁺: view of the asymmetric unit of [MBqn²⁺](BF₄)₂, [1²⁺](BF₄)₂ (a) and [MBqn⁰-H⁺]BF₄, [1⁰-H⁺]BF₄ (b). Salient C–N and C–C bond lengths are indicated.

In the case of 1⁰-H⁺, after several attempts with different crystals and from different batches, the refinement is such that equal and significant residual electron density peaks (0.4–0.6 e Å⁻³; background 0.18 e Å⁻³) are clearly and systematically located on all aromatic and methyl hydrogen sites, as for 1²⁺, but with an additional peak close to the nitrogen on 2′ position (Fig. 3, full explanation in the ESI Tables S1–S3†).

Fig. 3 Scheme of the C,N skeleton of the molecules overlaid with the residual 3D-Fourier difference map of 1²⁺ (left) and 1⁰-H⁺ (right). For each 1²⁺ and 1⁰-H⁺ an equal and significant residual electron density is clearly and systematically located on all aromatic and methyl hydrogen sites, as well as a on the nitrogen on the N′₂ position for 1⁰-H⁺ (highlighted with the yellow circle).

Thus for 1⁰-H⁺, because 17 H-atoms are clearly present in the molecule, the latter has to be doubly reduced in order to match the singly positive charge balanced by one BF₄⁻. This can be also viewed as the monoprotonated dication (leading to a 16 + 1 = 17H atoms trication) which was further reduced by two electrons to give the so-called direduced monoprotonated cation 1⁰-H⁺.

Complementary FAB and MALDI-TOF mass spectrometry measurement performed on crystals of [1⁰-H⁺]BF₄ are reproducible and in agreement with a single species, differing only by one H-atom either in solution (FAB) or in the solid state (MALDI) (Fig. S5 and S6†). In particular the MALDI spectra in SM matrix shows a clear difference in peak intensity between the 16 H-atoms and 17 H-atoms (m/z = 288.3 and 289.3) which can be assigned as originating from the molecular entities 1²⁺ and 1⁰-H⁺.

As a result of the rotational freedom around the central C₃–C′₃ bond, 1²⁺ and 1⁰-H⁺ could adopt any geometry between planar trans- (ϕ = 180°) and planar cis- (ϕ = 0°), where the dihedral angle is defined as ϕ = C₂–C₃–C′₃–C′₂. What we observe is that for 1²⁺ the geometry is exactly planar trans, whereas for 1⁰-H⁺ it is nearly planar cis (ϕ = 2.5°).

In the dication 1²⁺, constituted of two equivalent MQ⁺ moieties, the structure reveals an aromatic π-conjugated system. The average C–C bond length in the benzene ring [1.392(3) Å] corresponds to the intermediate C–C distance of 1.39–1.40 Å observed in a benzene molecule with electron delocalization.²² In the pyrazinium moiety, the aromaticity is confirmed by the values of the C–C and C–N bond distances, in the range 1.410(2)-1.416(2) Å and 1.311(2)–1.383(2) Å respectively, that are comparable to those found in the literature.²³ The pyrazinium structure is particularly clear on the 1,2,3,4 atomic positions, as shown by the two shortest distances N₁–C₂ [1.311(2) Å] and N₂–C₃ [1.315(2) Å], alternating with an intermediate length for C₂–C₃ [1.410(2) Å] (Fig. 2).

In contrast, in the direduced monoprotonated cation 1⁰-H⁺ the different redox state of the species and the protonation of only one side of the molecule gives rises to an asymmetric environment. On one hand, the non-protonated MQ moiety remains essentially similar to 1²⁺, where the bond lengths of both benzene ring and pyrazinium core are comparable (N₁–C₂ [1.316(2) Å], N₂–C₃ [1.344(2) Å], C₂–C₃ [1.413(2) Å]). On the other hand, in the protonated half molecule, although the benzene ring remains aromatic (C–C ∼ 1.39 Å), significant changes occur mostly within the pyrazinium core. The structure is such that the N′₁–C′₂, N′₁–C′₉ and N′₂–C′₃ bond lengths increase by 0.038 Å, 0.038 Å, and 0.075 Å respectively with a concomitant 0.044 Å decrease in the C′₂–C′₃ bond distance. The central C₃–C′₃ bond also changes dramatically with a net decrease by 0.052 Å, with 1.424(2) Å for 1⁰-H⁺ and 1.476(2) Å for 1²⁺.

These structural features of 1²⁺ and 1⁰-H⁺ have been investigated by quantum chemical calculations using Turbomole²⁴ (for details see Experimental section in ESI†). For both compounds geometrical parameters were fully optimized in the gas phase at the PBE0/def2-TZVPP level of theory (Tables S4, S5 and S8†), the PBE0 hybrid functional being known to provide the best estimates for such π-conjugated systems.^25,26 The calculated bond lengths fit well with the experimental data with mean absolute errors (MAE) of 0.0068 Å (1²⁺) and 0.0071 Å (1⁰-H⁺), whereas the twist angle of the molecule differs only very slightly for 1⁰-H⁺ (ϕ_exp = 2.5°, ϕ_calc = 0°). For 1²⁺ and 1⁰-H⁺ the frequencies calculations, in addition with a scan of the relaxed potential energy surface along the twist angle (ϕ) for 1²⁺, confirmed that the described structures correspond to the ground state of the molecules (Fig. S9 and S13, Tables S6 and S9†). The HOMO and LUMO of 1²⁺, as well as the HOMO of 1⁰-H⁺ are represented in Fig. 4.

Fig. 4 HOMO and LUMO for 1²⁺ (lower and upper left panel). HOMO for 1⁰-H⁺ (upper right panel).

As previously stated by A. di Matteo, “a reduction process increasing or decreasing the electron population, in the bonding or antibonding regions, strengthens or weakens the bonds, which become significantly shorter or longer”.²⁵ Here, by populating the LUMO of 1²⁺ (reduction process), a decrease and increase of the bond length is expected between atoms with bonding and antibonding characteristics respectively. This effect will be even more obvious between atoms where the HOMO and LUMO orbitals exhibit the opposite bonding and antibonding features. Thus focusing on the bipyrazinium core, the experimental parameters clearly show that the chemical reduction of 1²⁺ will lead to a shortening of the C₂–C₃ (C′₂–C′₃) and C₃–C′₃ bond lengths and to an increase of the N₁–C₂ (N′₁–C′₂), N₁–C₉ (N′₁–C′₉) and N₂–C₃ (N′₂–C′₃) bond lengths, both having respectively bonding and antibonding characteristic in the LUMO orbital.

The HOMO of 1⁰-H⁺ and the LUMO of 1²⁺ both exhibit comparable bonds with bonding and antibonding characteristics, which is in line with the above reasoning and the observed bond lengths in 1⁰-H⁺. This also shows that the trans–cis isomerization which exists between 1²⁺ (trans) and 1⁰-H⁺ (cis) has no strong influence on the way the π-orbitals overlap.

However, the N′₂–H situation in 1⁰-H⁺ creates a dissymmetry within the molecule leading to a strengthening of the bonding and antibonding features in the pyrazinium core of the protonated MQ. This results in a partial stiffening of the system, in the sense that the electron density can be rather described as localized and alternate single and double bonds, like in polyene-like structure such as dipyran or dithiopyran.^27,28 The N′₂–H situation also stabilizes the preferential cis-form in the solid. It creates a strong intramolecular hydrogen bond between the protonated and non-protonated amines (N₂⋯H–N′₂ ∼ 2.25 Å), which is also confirmed by the strong bonding characteristic in the N₂⋯H–N′₂ region (HOMO of 1⁰-H⁺Fig. 4, detail of H-bonds Fig. S7†). This apparently stabilizes 1⁰-H⁺ and might also act as a template by constraining the cis-form, i.e. the chelate function of the 2,2′-bipyridine.

Spectroscopy, electrochemistry, magnetic susceptibility, spectroelectrochemistry and TDDFT

Because of the poor solubility of [1⁰-H⁺]BF₄ and in order to get comparative data, both EPR, UV-Vis-NIR, cyclic voltammetry, spectro-electrochemistry, NMR and SQUID experiments have been performed in DMF. Nevertheless, a complementary UV-Vis experiment has been performed in MeCN for [1²⁺](BF₄)₂ in order to fully identify its electronic transitions below 400 nm.

Electron paramagnetic resonance (EPR) (Fig. 5). A sample of [1⁰-H⁺]BF₄ dissolved in DMF yielded a 9-line spectrum at ambient temperature. This implies the presence of a paramagnetic species arising from either an impurity or a new redox state in solution. The rich hyperfine pattern could be modelled by coupling the single electron of the doublet radical to the I = 1 nuclear spin of all four nitrogen atoms (99.7% natural abundance). This is in line with what is expected for a doublet radical in solution. Simulation yielded spin Hamiltonian parameters g_iso = 2.0069 and A_iso = 6.05 × 10⁻⁴ cm⁻¹ with a magnitude of the ¹⁴N hyperfine coupling comparable to that found for N,N′-dialkylated quinoxalinium radical cations.²⁹ The g-value is similar to those reported for phosphine-based cation radicals,^30–32 but is slightly higher compared to the methylviologen radical cation (g ∼ 2.003).^7,8,33 This is consistent with the extra heteroatom in the six-membered ring hosting the unpaired spin. Consequently, the ¹⁴N hyperfine coupling is larger than for the archetypal viologens.^29,34,35

Fig. 5 Isotropic EPR spectrum of the intermediate radical species recorded in DMF solution at 293 K (experimental conditions: frequency, 9.8075 GHz; power, 20 mW; modulation, 0.07 mT). Experimental data represented by the black line and simulation depicted by the red trace.

The absence of additional hyperfine structure from coupling to the methyl and aromatic protons of the organic radical confirms these couplings are less than the isotopic linewidth ∼2 × 10⁻⁴ cm⁻¹, which is invariant irrespective of the measurement temperature. Nevertheless this does not show conclusively whether the paramagnetic species is an impurity or not.

Cyclic voltammetry (CV) (Fig. 6a). In order to gain further insights into the redox process leading to the formation of the paramagnetic species, CV measurements were performed. Furthermore, these were done quantitatively by using the well-known ferrocene/ferrocenium system as internal reference. Regarding its reduction potential (V vs. Fc/Fc⁺), 1²⁺ (E_1/2 = −0.54 V) is an intermediate but strong electron-acceptor between the prototype 3,3′-diazamethylviologen (DAMV˙⁺) (E_1/2 = −0.35 V)^19,20,36 and the methylviologen (E_1/2 = −0.89 V).⁷ Thus, in two experiments, either the dication [1²⁺](BF₄)₂ (Fig. 1 – purple) or the doubly reduced monoprotonated cation [1⁰-H⁺]BF₄ (Fig. 1 – blue) were dissolved in DMF. Both CVs are perfectly reversible and superposable for all investigated scan rates (50–5000 mV s⁻¹).

The deconvolution of the CV allows us to circumvent diffusion problems and to conclude that there is a single-electronic process by comparing with the current peak of the ferrocene/ferrocenium system. This rules out the presence of any impurity and points to the oxidation of the doubly reduced monoprotonated cation [1⁰-H⁺]BF₄ to give the doublet radical. Furthermore, the voltage separation between current peaks of ca. 200 mV is significantly higher than the theoretical value of 59 mV. This is in line with a redox process inducing a significant structural rearrangement such as a trans–cis isomerization.³⁷

Magnetic susceptibility measurement using SQUID and NMR Evans' method. The magnetic behaviour of the redox states in solid and solution states should also show differences due to the diamagnetic nature of the solid state species [1⁰-H⁺]BF₄versus the paramagnetic nature of the doublet radical species in solution.

The magnetic susceptibility of [1⁰-H⁺]BF₄ (Fig. 1 – purple) was measured on a polycrystalline sample over the 4–380 K temperature range under an applied dc field of 1 T. As shown in Fig. S10,† in the solid state the product of magnetic susceptibility with temperature is very close to zero (<0.007 cm³ K mol⁻¹), which confirms the diamagnetic nature of [1⁰-H⁺]BF₄ in the solid state.

Furthermore, crystals of [1⁰-H⁺]BF₄ dissolved in DMF-d⁷ under inert atmosphere exhibit a very strong diamagnetism at −5000 × 10⁻⁶ cm³ mol⁻¹, even after correcting with the expected diamagnetic contribution calculated from Pascal constants. Given that we were expecting to see a paramagnetic signature, which would be expected for isolated paramagnetic species, this behaviour suggests the presence of superdiamagnetism. This has been previously observed in 3D inorganic compounds and was coined superdiamagnetism, in an obvious reference to the Meissner effect and the perfect diamagnetism observed in superconductors.^38,39 More recently superdiamagnetism was induced at low temperatures by the application of a magnetic field in a one-dimensional Pt^II complex.⁴⁰ To the best of our knowledge this has never been observed in a purely organic based compound. In our case the diamagnetic susceptibility reaches 6% of the 1/4π value expected for complete diamagnetism. This unexpected superdiamagnetism was confirmed using the Evans' ¹H-NMR method⁴¹ at room temperature from the strong negative chemical shift of the reference tert-butyl alcohol signal at −27.3 Hz in DMSO-d⁶ and −25.5 Hz in DMF-d⁷ in the presence of our studied molecule (Fig. S11 and S12†). This is typical of an additional diamagnetic environment around the protons of the reference. The calculated magnetic susceptibilities, respectively χ = −962 × 10⁻⁶ cm³ mol⁻¹ and χ = −3510 × 10⁻⁶ cm³ mol⁻¹ are in line with the SQUID measurement, though the effect seems much weaker in DMSO.

This allows us to conclude from the CV, EPR, SQUID and NMR results that a single radical intermediate species is present in solution. Due to the perfect planarity of the molecule (no steric hindrance) and the fact that the spin density is located in π-orbitals, the apparent superdiamagnetic behaviour of such a radical in solution could be tentatively explained from the formation of extended stacks of radicals, in analogy to the unidimensional Pt^II complex.⁴⁰ Strongly spin-paired π-dimers, as already observed in other π-radicals such as viologen and TTF,^34,42–44 would only provide a “normal” diamagnetic behaviour.

However the question remains: what is the nature of the radical species in solution? In their studies on the prototype N,N′-dimethyl-3,3′-bipyrazinium dication,²⁰ and the N-methylpyrazinium,^45,46N-methylquinoxalinium⁴⁷ and N-methylphenazinium⁴⁸ cations, Kaim et al., Rouiller & Laviron, Eaton and Takagi et al. showed that under mild acidic or neutral conditions the chemical reduction of such kinds of cations is accompanied by protonation thus leading to 4-hydro-1-methyl-diazinium 7π electron radical cations in solution. We thus see two possibilities here. Firstly, the dication (1²⁺) is reduced and isomerises through the addition of an electron to give the cis-radical cation (1˙⁺). Secondly, this cis-radical cation is formed and further protonated to give the cis-protonated radical cation (1˙⁺-H⁺).

In order to investigate this, additional spectroelectrochemical (SE), UV-Vis-NIR and TDDFT experiments were performed. The purpose was to identify a possible structure/optical signature relationship for the three different redox states. Optical spectra have been computed using TDDFT at the PBE0/def2-TZVPP level of theory in a closed shell (1²⁺ and 1⁰-H⁺) and unrestricted open-shell (1˙⁺ and 1˙⁺-H⁺) configuration. To reproduce the behaviour of the molecules in solution, the conductor-like screening model COSMO has been applied, with ε = 37 as a mean value for a mixture of DMF and MeCN.

Spectroelectrochemistry (Fig. 6(b and b′)). Known to be a powerful tool for analysing the spectral signature of involved species during an electrochemical process, time resolved spectroelectrochemical experiments (SE) were performed, using the already described home-made cell,^49–51 between 350 and 900 nm on dissolved 1²⁺ (Fig. 1 – grey) and 1⁰-H⁺ (Fig. 1 – purple) (0.5 mM in 0.1 M TBAPF₆/DMF), in order to correlate the changes in their spectra during reduction and oxidation processes.

In Fig. 6(b and b′), we see that both voltabsorptograms (Δ(Abs) vs. initial species) are perfect mirror images of each other, indicating that the two species are part of the same redox couple. Furthermore, the spectra measured during the SE experiments are quite similar to those obtained by dissolving crystals of [1²⁺](BF₄)₂ and [1⁰-H⁺]BF₄ (Fig. 6c2 and 6c1). The results match perfectly those from the CV in terms of the exclusive presence of one reversible and stable 1²⁺/1^radical redox couple in solution.

Fig. 6 (a) Deconvoluted cyclic voltammogram (left) [1⁰-H⁺]BF₄ (0.56 mM) and Fc (0.65 mM) in DMF, with TBAP (0.1 M) as supporting electrolyte on Pt at 200 mV s⁻¹. Evolution of the voltammogram (middle) and the current peak of each redox couple (right) with the variation of the scan rate. (b) Spectroelectrochemistry experiments of 1²⁺ (b) and 1⁰-H⁺ (b′), 0.5 mM in 0.1 M TBAPF₆/DMF on platinum electrode, in thin layer conditions (close to 50 μm) under argon at 10 mV s⁻¹ and 293 K. (b1 and b1′) CVs in current vs. potential representation. (b2 and b2′) 3D representation: (wavelength, time) vs. (Δabsorbance). (b3 and b3′) 3D representation viewed along the absorbance axis: (wavelength, time) vs. (Δabsorbance). (b4 and b4′) CV and d(Δabsorbance)/dt (at 850 nm) vs. time, showing a good agreement between the optical and electrochemical data. (c) Overlay of UV-Vis-NIR and TDDFT spectra of 1²⁺ [crystalline state (c3), dissolved in MeCN (c2)], 1⁰-H⁺ [solid state (c4)] and of the intermediate species 1˙⁺-H⁺ in DMF solution (c1).

Solid and solution state UV-Vis-NIR optical properties of the dication 1²⁺. Crystals of [1²⁺](BF₄)₂ are yellow/off-white in the solid state with no significant absorption in the visible range. The band at λ_max = 380 nm corresponds to the HOMO → LUMO transition (Fig. 6c3, Table S7†). Crystals of [1²⁺](BF₄)₂ dissolved in MeCN give a colourless solution. The SE spectrum of 1²⁺ exhibits no absorption in the visible region (Fig. 6b), which is confirmed from the UV-Vis spectrum with the lowest energy absorption band slightly shifted compared with the solid state spectrum at λ_max = 360 nm (Fig. 6c2).

Solid state UV-Vis-NIR optical properties of the doubly reduced monoprotonated cation 1⁰-H⁺. Crystals of [1⁰-H⁺]BF₄ are deep purple and absorb across the whole UV-Vis range and well into the near-infrared. There are broad absorption bands centred at 270, 350, 570, 820, and 1000 nm (shoulder ≈1300 nm) (Fig. 6c4). The TDDFT spectrum of 1⁰-H⁺ matches perfectly with the electronic transitions observed experimentally in the range 200-950 nm, with λ_max ≈ 820 nm corresponding to the HOMO → LUMO transitions (Table S10†). Analogous to the behaviour of the methylviologen radical cation either in solution or in the solid state,^7,8 the additional transition present at lower energy is most probably explained due to the formation of extended stacks of 1⁰-H⁺ species in the solid. This is actually confirmed by the crystal packing which shows an infinite 1D stack of dimers of 1⁰-H⁺ (Fig. S8†).

Solution state UV-Vis-NIR optical properties of the doublet radical. In the electronic spectrum in solution for a concentration of ≈ 2.4 × 10⁻⁴ M the formation of the doublet radical redox state is accompanied by a colour change from the deep purple of the solid to blue with absorption bands centred at 430, 470, 580, 820 and 920 nm (shoulder ≈1100 nm) in the Vis-NIR region (Fig. 6c1). The SE spectrum of the intermediate radical species looks very similar, although the position of the bands is slightly shifted to higher energy by ca. 40 nm (0.2 eV) (Fig. 6b). To investigate the two previously proposed hypothesis, namely (a) the species in solution is a cis-radical cation 1˙⁺ or (b) the species in solution is the cis-protonated radical dication 1˙⁺-H⁺, two different TDDFT spectra were simulated from fully optimized geometries (Tables S12 and S13†).

The TDDFT spectrum corresponding to hypothesis (a) cis-1˙⁺ is shown with green lines in Fig. 6c1. This is clearly different from the experimental spectrum in that it exhibits a single broad transition at λ_max = 1200 nm and no transitions between 500 and 1100 nm.

The TDDFT spectrum corresponding to hypothesis (b) cis-1˙⁺-H⁺ is shown with violet lines in Fig. 6c1. The main electronic transitions observed experimentally in the range 400 to 850 nm in the SE spectrum are perfectly reproduced theoretically. The lowest energy band λ_max = 820 nm corresponds to the ^αSOMO → ^αLUMO transition (Fig. 6c1, Table S13†).

This leads us to conclude that the radical intermediate species present in solution is intrinsically composed of the cis-protonated radical dication 1˙⁺-H⁺ (Fig. 1 – blue). Furthermore the formation of extended stacks of radicals 1˙⁺-H⁺ in solution is substantiated by the presence of an additional band at lower energy (λ = 920 nm, Fig. 6c1), analogous to the low energy bands observed in aggregated viologens in solution.^7,8

We also note that although at a first glance the spectra of 1˙⁺-H⁺ (Fig. 6c1) and 1⁰-H⁺ (Fig. 6c4) look broadly similar, the fact that the spectrum of 1⁰-H⁺ exhibits no transitions between 400 and 500 nm (Table S10†), provides a rapid fingerprint for the identity of the redox state.

Case of the metal complex [CdCl₂(1⁰)]: chelation and non-Kekulé structure aspect of 1⁰

We provide a first example of a metal complex [CdCl₂(1⁰)] which illustrates the ability of such a ligand to chelate metal ions (Fig. 7). This proves that the neutral direduced species 1⁰ can be obtained without being protonated unlike in the case of 1⁰-H⁺, which reinforces the suggestion that the additional proton on N′₂ is labile in solution. In addition, the crystals obtained are air stable, which is important for the further development of robust compounds.

Fig. 7 Thermal ellipsoid view of the asymmetric unit of [CdCl₂(1⁰)] (left). Corresponding energy level scheme showing the relative energies between singlet RHF, ¹UHF and triplet ³UHF, as well as highlighting the open-shell singlet ground state and the ϕ-angle dependent reduction of the singlet to triplet energy gap (ΔE_ST) (right) with ϕ the dihedral angle defined by the red lines.

The fact that we were unable to isolate a pure phase of the cadmium complex precluded further characterization. Nevertheless, the reliable crystallographic dataset obtained for this compound has been used as a reference structure for further studies, especially regarding the investigation of the neutral direduced methylbiquinoxen 1⁰ and its electronic structure.

The asymmetric unit is made of one entire complex [CdCl₂(1⁰)]. The cadmium adopts a distorted tetrahedral coordination environment of D_2d symmetry.

The 1⁰ chelating ligand is cis-planar (ϕ = 1.3°) and very symmetric in terms of the relative bond distances (Table S15†). As for 1⁰-H⁺, the multiple bond character of the central C₃–C′₃ bond (d = 1.449 Å) indicates a fully π-conjugated molecule, resulting in a homogenous distribution of the electron density.

Unrestricted DFT calculations for the singlet and triplet state of [CdCl₂(1⁰)] have been performed. The energy of the singlet state is slightly lower than the energy of the triplet state. The energy difference (ΔE_ST) amounts to 0.14 eV (Table S14†). However, the ground state shows a significant spin contamination. The analysis of the natural orbitals shows that there is a pair of partly occupied orbitals, one with bonding character between C₃ and C′₃, the other antibonding with occupation numbers of 1.5 and 0.5, respectively, showing partial occupation of the LUMO (Fig. S14†). In addition, the spin density (Fig. S15†) indicates a weakening of the bonding character similar to what is seen in a broken symmetry state. This behaviour is in line with the biradicaloids observed, for example, for under coordinated Sn atoms.⁵²

Regarding 1⁰ in terms of a symmetric and non-protonated structure, not to be confused with the dissymmetric closed-shell monoprotonated cation 1⁰-H⁺, such an open-shell behaviour with a small ΔE_ST gap was predictable due to the non-classical (non-Kekulé)^18,20,53,54 electronic structure of 1⁰, as previously identified by Matheis, et al.^18,20 in the prototype molecule, 1,1′-dimethyl-3,3′-bipyrazinium (Fig. S16†). By definition, non-Kekulé means a molecule which is “fully conjugated, but each of whose Kekulé structures contains at least two atoms that are not π-bonded”.^55,56 Furthermore, the ΔE_ST gap can apparently be significantly reduced by simply twisting the molecule (ΔE_ST = f(ϕ) Fig. 7), thus giving a means to thermally access the triplet state.

Conclusions and Outlook

As is already known for viologens, the reversible and controllable structural, electronic, and electrostatic changes between the dication and radical species formed in solution invest the MBqn^2+/˙⁺ redox couple with electron mediation properties. This should be useful for many radical-assisted reactions (photocatalytic H₂ production, C–C bond cleavage, artificial photosynthesis, electrolyte, electrochromism etc.).^7,57,58 Surprisingly, chelation to a cadmium centre allowed us to access the direduced solid state biradicaloid species MBqn⁰ as ligand. Additionally, as expected for such a biradicaloid species, the calculated small singlet to triplet energy gap opens perspectives for a range of exotic electronic, optical, magnetic, dielectric etc. properties within a single system composed of purely organic moieties or incorporating metalloid centres. Intriguingly, the superdiamagnetic behaviour observed for the doublet radical species in solution is such an unexpected property for an organic molecule that this demands further investigation.

Acknowledgements

The authors thank the Alexander von Humboldt Foundation (fellowship to N. L) and Helmholtz POF “STN” for financial support. We also thank Christopher Anson and Magali Allain for helpful discussions regarding the crystal structure of the direduced monoprotonated species, Ingrid Freuze for the mass spectrometry measurements, Sven Stahl for the TGA measurements and Lutz Greb for the NMR measurement.

Notes and references

1. W. Kaim and B. Schwederski, Coord. Chem. Rev., 2010, 254, 1580–1588.
2. P. J. Chirik and K. Wieghardt, Science, 2010, 327, 794–795.
3. A. Heckmann and C. Lambert, Angew. Chem., Int. Ed., 2012, 51, 326–392.
4. A. C. Fahrenbach, C. J. Bruns, H. Li, A. Trabolsi, A. Coskun and J. F. Stoddart, Acc. Chem. Res., 2014, 47, 482–493.
5. O. R. Luca and R. H. Crabtree, Chem. Soc. Rev., 2013, 42, 1440–1459.
6. N. Martin, Chem. Commun., 2013, 49, 7025–7027.
7. P. M. S. Monk, The Viologens: Physicochemical Properties, Synthesis, and Application of the Salts of 4,4′-Bipyridine, Wiley, 1998.
8. T. M. Bockman and J. K. Kochi, J. Org. Chem., 1990, 55, 4127–4135.
9. N. Mercier, Eur. J. Inorg. Chem., 2013, 19–31.
10. R. J. Mortimer, Electrochim. Acta, 1999, 44, 2971–2981.
11. M.-S. Wang, G. Xu, Z.-J. Zhang and G.-C. Guo, Chem. Commun., 2010, 46, 361–376.
12. N. Leblanc, M. Allain, N. Mercier and L. Sanguinet, Cryst. Growth Des., 2011, 11, 2064–2069.
13. A. S. Abouelwafa, V. Mereacre, T. S. Balaban, C. E. Anson and A. K. Powell, CrystEngComm, 2010, 12, 94–99.
14. A. S. Abouelwafa, C. E. Anson, A. Hauser, H. H. Patterson, F. Baril-Robert, X. Li and A. K. Powell, Inorg. Chem., 2012, 51, 1294–1301.
15. M. Freitag, L. Gundlach, P. Piotrowiak and E. Galoppini, J. Am. Chem. Soc., 2012, 134, 3358–3366.
16. A. Trabolsi, N. Khashab, A. C. Fahrenbach, D. C. Friedman, M. T. Colvin, K. K. Cotí, D. Benítez, E. Tkatchouk, J.-C. Olsen, M. E. Belowich, R. Carmielli, H. A. Khatib, W. A. Goddard, M. R. Wasielewski and J. F. Stoddart, Nat. Chem., 2010, 2, 42–49.
17. N. Leblanc, N. Mercier, L. Zorina, S. Simonov, P. Auban-Senzier and C. Pasquier, J. Am. Chem. Soc., 2011, 133, 14924–14927.
18. W. Kaim and W. Matheis, Chem. Ber., 1990, 123, 1323–1325.
19. W. Matheis and W. Kaim, J. Chem. Soc., Faraday Trans., 1990, 86, 3337–3339.
20. W. Matheis, J. Poppe, W. Kaim and S. Zalis, J. Chem. Soc., Perkin Trans. 2, 1994, 1923–1928.
21. O. N. Chupakhin, E. O. Sidorov, S. M. Shein and I. I. Bil'kis, Zh. Org. Khim., 1976, 12, 2464–2468.
22. E. G. Cox, D. W. J. Cruickshank and J. a. S. Smith, Nature, 1955, 175, 766.
23. F. H. Allen, O. Kennard, D. G. Watson, L. Brammer, A. G. Orpen and R. Taylor, J. Chem. Soc., Perkin Trans. 2, 1987, S1–S19.
24. TURBOMOLE V6.6 2014, a development of University of Karlsruhe and Forschungszentrum Karlsruhe GmbH, 1989-2007, TURBOMOLE GmbH, since 2007; available from http://www.turbomole.com.
25. A. Di Matteo, Chem. Phys. Lett., 2007, 439, 190–198.
26. D. Jacquemin, E. A. Perpète, G. E. Scuseria, I. Ciofini and C. Adamo, J. Chem. Theory Comput., 2008, 4, 123–135.
27. D. Chasseau, J. Gaultier, C. Hauw, R. Fugnitto, V. Gianis and H. Strzelecka, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem., 1982, 38, 1629–1631.
28. H. R. Luss and D. L. Smith, Acta Crystallogr., Sect. B: Struct. Crystallogr. Cryst. Chem., 1980, 36, 1580–1588.
29. A. Schulz, W. Kaim and H.-D. Hausen, J. Chem. Soc., Faraday Trans. 1, 1988, 84, 3207–3214.
30. O. Back, B. Donnadieu, P. Parameswaran, G. Frenking and G. Bertrand, Nat. Chem., 2010, 2, 369–373.
31. X. Pan, X. Chen, T. Li, Y. Li and X. Wang, J. Am. Chem. Soc., 2013, 135, 3414–3417.
32. X. Pan, Y. Su, X. Chen, Y. Zhao, Y. Li, J. Zuo and X. Wang, J. Am. Chem. Soc., 2013, 135, 5561–5564.
33. N. Leblanc, N. Mercier, O. Toma, A. H. Kassiba, L. Zorina, P. Auban-Senzier and C. Pasquier, Chem. Commun., 2013, 49, 10272–10274.
34. A. G. Evans, J. C. Evans and M. W. Baker, J. Am. Chem. Soc., 1977, 99, 5882–5884.
35. D. W. Clack, J. C. Evans, A. Y. Obaid and C. C. Rowlands, Tetrahedron, 1983, 39, 3615–3620.
36. T. Fujihara, T. Wada and K. Tanaka, Inorg. Chim. Acta, 2004, 357, 1205–1212.
37. N. Terkia-Derdra, R. Andreu, M. Sallé, E. Levillain, J. Orduna, J. Garín, E. Ortí, R. Viruela, R. Pou-Amérigo, B. Sahraoui, A. Gorgues, J.-F. Favard and A. Riou, Chem.–Eur. J., 2000, 6, 1199–1213.
38. V. L. Ginzburg, Solid State Commun., 1981, 39, 991–992.
39. V. L. Ginzburg, A. A. Gorbatsevich, Y. V. Kopayev and B. A. Volkov, Solid State Commun., 1984, 50, 339–343.
40. M. Haruki and P. Wachter, J. Magn. Magn. Mater., 1992, 104–107, 475–476.
41. D. F. Evans, J. Chem. Soc., 1959, 2003–2005.
42. W. W. Porter and T. P. Vaid, J. Org. Chem., 2005, 70, 5028–5035.
43. J. M. Spruell, A. Coskun, D. C. Friedman, R. S. Forgan, A. A. Sarjeant, A. Trabolsi, A. C. Fahrenbach, G. Barin, W. F. Paxton, S. K. Dey, M. A. Olson, D. Benítez, E. Tkatchouk, M. T. Colvin, R. Carmielli, S. T. Caldwell, G. M. Rosair, S. G. Hewage, F. Duclairoir, J. L. Seymour, A. M. Z. Slawin, W. A. Goddard, M. R. Wasielewski, G. Cooke and J. F. Stoddart, Nat. Chem., 2010, 2, 870–879.
44. G.-P. Yong, C.-F. Li, Y.-Z. Li and S.-W. Luo, Chem. Commun., 2010, 46, 3194–3196.
45. L. Roullier and E. Laviron, Electrochim. Acta, 1980, 25, 795–804.
46. D. R. Eaton, J. M. Watkins and R. J. Buist, J. Am. Chem. Soc., 1985, 107, 5604–5609.
47. W. Kaim, Rev. Chem. Intermed., 1987, 8, 247–286.
48. T. Yutaka, A. Kazuyuki, I. Toshiaki, K. Hitomi and I. Kazuhiko, Chem. Lett., 1972, 1, 847–848.
49. F. Gaillard and E. Levillain, J. Electroanal. Chem., 1995, 398, 77–87.
50. M. Dias, P. Hudhomme, E. Levillain, L. Perrin, Y. Sahin, F.-X. Sauvage and C. Wartelle, Electrochem. Commun., 2004, 6, 325–330.
51. O. Alévêque, E. Levillain and L. Sanguinet, Electrochem. Commun., 2015, 51, 108–112.
52. C. Schrenk, A. Kubas, K. Fink and A. Schnepf, Angew. Chem., Int. Ed., 2011, 50, 7273–7277.
53. S. Inagaki, Orbitals in Chemistry, Springer, 2009.
54. W. T. Borden, H. Iwamura and J. A. Berson, Acc. Chem. Res., 1994, 27, 109–116.
55. H. C. Longuet-Higgins, J. Chem. Phys., 1950, 18, 265–274.
56. V. I. Minkin, Pure Appl. Chem., 1999, 71, 1919–1981.
57. M. D. Forbes, Carbon-Centered Free Radicals and Radical Cations: Structure, Reactivity, and Dynamics, Wiley, 2010.
58. P. J. Cappillino, H. D. Pratt, N. S. Hudak, N. C. Tomson, T. M. Anderson and M. R. Anstey, Adv. Energy Mater., 2014, 4, 1300566.

---

Footnote

† Electronic supplementary information (ESI) available. CCDC 1043174, 1043175 and 1437399. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c5sc04904k

---