2,3-Diaminobenzene (1, 77 mg, 0.7 mmol) was added to a solution of diferrocenylethane-1,2-dione (2, 0.3 g, 0.7 mmol) in ethanol (50 mL). The mixture was stirred under reflux overnight during which time an orange solid precipitated, which was isolated by filtration, washed with cold diethyl ether (3 × 10 mL) and finally crystallized in ethanol. Yield 98%. M.p > 300 °C. ¹H-NMR (CD₃CN): δ 4.09 (s, 10H), 4.32 (st, 4H), 4.64 (st, 4H), 7.67 (dd, 2H, J = 3.4 Hz, J = 6.4 Hz), 8.03 (dd, 2H, J = 3.4 Hz, J = 6.4 Hz); ¹³C-NMR (CDCl₃): δ 68.7 (4xCH), 69.7 (10xCH), 71.4 (4xCH), 85.2 (2xq), 128.5 (2xCH), 128.7 (2xCH), 140.4 (2xq), 152.9 (2xq); FAB MS: m/z (relative intensity): 498 (M⁺,100); Anal Calc for C₂₈H₂₂Fe₂N₂: C, 67.57; H, 4.45; N, 5.62. Found: C, 67.80; H, 4.82; N, 5.40.

The quinoxaline-based receptor 3 was prepared following the classical method for synthesizing both quinoxaline itself and its derivatives, which involves the condensation of an aromatic 1,2-diamine with a 1,2-dicarbonyl compound in refluxing ethanol or acetic acid (Scheme 1) [38]. Thus, condensation of the readily available diferrocenylethane-1,2-dione (2) [35] with 1,2-diaminobenzene (1) gave an excellent yield (98%) of the corresponding 2,3-diferrocenylquinoxaline (3) which was fully characterized by using standard techniques: ¹H-NMR and ¹³C-NMR spectroscopies, FAB mass spectrometry and elemental analysis.

The redox properties of receptor 3 was investigated by linear sweep voltammetry (LSV), cyclic voltammetry (CV), and Osteryoung square wave voltammetry (OSWV) in a CH₃CN solution containing 0.15 M [n-Bu₄N]ClO₄ (TBAP) as supporting electrolyte. In spite of the symmetry of the receptor 3 it exhibited, in the range 0−0.9 V, two reversible one-electron redox wave at the half-wave potential value of ¹E_1/2 = 0.47 V and ²E_1/2 = 0.58 V (ΔE_1/2 = 110mV) versus decamethylferrocene (DMFc), demonstrating the existence of a weak interaction between the two iron centres (Figure 1). The criteria applied for reversibility was a separation of ∼60 mV between cathodic and anodic peaks, a ratio of 1.0 ± 0.1 for the intensities of the cathodic and anodic currents Ic/Ia, and no shift of the half-wave potentials with varying scan rates.

The UV−vis spectra for receptor 3 was recorded as 10⁻⁴ M solution in CH₃CN and contains three prominent absorption bands with a maximum at 234 nm (ɛ = 26,000 M⁻¹ cm⁻¹), 277 nm (ɛ = 14750 M⁻¹ cm⁻¹) and 314 nm (ɛ = 9420 M⁻¹ cm⁻¹) which can safely be ascribed to a high energy ligand-centered π−π* electronic transition (L−π*) (HE band). In addition to this band, another two weaker absorptions are visible at 409 nm (ɛ = 1,590 M⁻¹ cm⁻¹) and 490 nm (ɛ = 1,860 M⁻¹ cm⁻¹) which are assigned to another localized excitations with a lower energy produced either by two nearly degenerate transitions, an Fe(II) d−d transition or by a metal−ligand charge transfer (MLCT) process (d_π−π*) (LE band) [39] This assignment is in accordance with the latest theoretical treatment (model III) reported by Barlow et al. [40]. Such spectral characteristics confer an orange color to this species.

One of the most interesting attributes of the new diferrocenylquinoxaline reported here is the presence of metal-ion binding sites on the quinoxaline ring close to a ferrocene redox-active moiety. Due to this structural feature metal recognition properties on the receptor 3 were evaluated by electrochemical, optical and ¹H-NMR techniques.

The electrochemical binding interactions of 3 towards cations of biological and environmental relevance, such as Li⁺, Na⁺, K⁺, Ca²⁺, Mg²⁺, Cu²⁺, Zn²⁺, Cd²⁺, Hg²⁺, Ni²⁺, and Pb²⁺, added as their perchlorate salts, were investigated in CH₃CN (c = 1 × 10⁻³ M). Titration studies with addition of the above-mentioned set of metal cations (2.5 × 10⁻² M in H₂O) to an electrochemical solution of receptor 3 containing [n-Bu₄N]ClO₄ (0.1 M) as supporting electrolyte, demonstrate that while addition of Cu²⁺ and Hg²⁺ ions promotes remarkable responses, addition of Li⁺, Na⁺, K⁺, Ca²⁺, Mg²⁺, Zn²⁺, Cd²⁺, Pb²⁺ and Ni²⁺ metal ions had no effect either on LSV or on the CV or OSWV of this receptor, even when present in a large excess. The results obtained on the stepwise addition of substoichiometric amounts of Hg²⁺ revealed the appearance, in the OSWV, of a new oxidation peak at practically the same potential of the second redox peak in the free receptor (Ep = 0.55 V, ΔEp = 75 mV).This fact suggests that the complex is disrupted after the first monoelectronic oxidation of the complex 3⁺·Hg²⁺ and the second oxidation really takes place on the uncomplexed mono-oxidized 3⁺. The current intensity of this new peak increases until 1 equiv of the Hg²⁺ cation is added [Figure 2(a)]. Moreover, the CV analysis of the complex 3·Hg²⁺ shows that one reduction process takes place at the same reduction potential showed by the uncomplexed ligand 3, indicating that the complex starts to be disrupted after its electronic oxidation [Figure 2(b)]. This behaviour means that this receptor is not only able to monitor binding but it is also able to behave as an electrochemically induced switchable chemosensor for Hg²⁺ through the progressive electrochemical release of these metal cations; as a result of a decrease of the corresponding binding constant upon electrochemical oxidation.

Remarkably, LSV studies carried out upon addition of Cu²⁺ to the CH₃CN solution of this receptor showed a significant shift of the sigmoidal voltammetric wave toward cathodic currents, indicating that Cu²⁺ cations promote the oxidation of the free receptor. On the other hand, the same experiments carried out upon addition of Hg²⁺ revealed a shift of the linear sweep voltammogram toward more positive potentials, indicating the complexation process according to the previously observed by OSWV (Figure 3).

Previous studies on ferrocene-based ligands have shown that their characteristic low energy (LE) bands in the absorption spectra are perturbed upon complexation [41–44]. Therefore, the metal recognition properties of the ligand 3 toward metal ions were also evaluated by UV−vis spectroscopy. Titration experiments for CH₃CN solutions of this ligand (c = 1 × 10⁻⁴ M), and the corresponding cations were performed and analyzed quantitatively. [45] It is worth mentioning that no changes were observed in the UV−vis spectra upon addition of Li⁺, Na⁺, K⁺, Ca²⁺, Mg²⁺, Zn²⁺, Cd²⁺, and Ni²⁺ and Pb²⁺ metal ions, even in a large excess; however, significant modifications were observed upon addition of Hg²⁺.

Thus, the addition of increasing amounts of Hg²⁺ ions in water to a solution of 3 caused a decrease in the intensity of the LE band, at λ = 490 nm, along with the progressive appearance of a new band located at λ = 630 nm (ɛ = 790 M⁻¹ cm⁻¹) as well as a increase of the initial HE band intensity. Two well-defined isosbestic points at 439 and 531 indicate that a neat interconversion between the uncomplexed and complexed species occurs [Figure 4(a)]. The new LE band is red-shifted by 140 nm and is responsible for the change of colour, from orange to deep green, which can be used for a “naked-eye” detection of this metal ion [Figure 4(b)]. Binding assays using the method of continuous variations (Job’s plot) suggests a 1:1 binding model (metal/ligand) with a log Ka = 3.4 ± 0.17 [Figure 4(c)]. Moreover, the calculated detection limit [46] was 1.3 × 10⁻⁵ M. Additionally the peak corresponding to the complex [3·Hg]²⁺ was observed by ES-MS at m/z 700.02. The relative abundance of the isotopic clusters was in good agreement with the simulated spectrum of the 1:1 complex.

In order to get additional information about the coordination between the receptor 3 and Hg²⁺ cations, a ¹H-NMR titration experiment was performed where aliquots of metal cation in D₂O were added to a solution of the receptor in CD₃CN. The free receptor 3 exhibits two sets of signals: one of them corresponding to the ferrocene moiety and another one to the quinoxaline ring. The ferrone moieties show a signal at δ = 4.10 (s), corresponding to the protons present in the unsubstituted ciclopentadienyl (Cp) unit and two psudotriplets at 4.32 and 4.64 ppm assigned to the H_β, and H_α within the monosubustituted Cp ring. On the other hand, the quinoxaline ring displays two double doublets at δ = 7.67 (H-6) and 8.03 (H-5) ppm. An inspection of the ¹H-NMR titration data showed a strong chemical shift for the signals associated with the ferrocene units due to their proximity to the binding sites. The protons within the unsubstituted Cp were shifted Δδ = +0.26 ppm and the H_α and H_β protons Δδ = 0.59 and 0.52 ppm respectively. On the other hand a weaker shift (Δδ = 0.1 ppm) in the H-5 and H-6 protons of the quinoxaline ring were also observed (Figure 5).