2.1. General information

All starting materials for synthesis as well as substrates for the catalytic experiments were obtained commercially as reagent grades and used as supplied without further purification. The o-sulfonamide azo-benzene ligands 1 and 2 have been previously prepared and reported [34]. IR spectra were measured with a Bruker Equinox FT-IR spectrometer equipped with a diamond ATR unit in the range of 4000–600 cm⁻¹. UV–Vis measurement was carried out using Varian Cary 5E UV-VIS-NIR Spectrophotometer. Elemental analyses were carried out on Leco CHNS-932 and El Vario III elemental analysers. Mass spectrometry (MS) spectra were measured with a Bruker MAT SSQ 710 spectrometer. ¹H and ¹³C NMR spectra for characterization of the phenoxazinone product were recorded with a Bruker AVANCE 400 MHz spectrometer using deuterated solvents and TMS as the internal standard.

2.2. Synthesis of cobalt(III) complexes

Co1⋅OAc⋅bpy: Ligand 1 (51 mg, 0.14 mmol), 2,2-bipyridine (22 mg, 0.14 mmol) and cobalt(II) acetate tetrahydrate (0.04 g, 0.14 mmol) reacted in 5 mL ethanol after which a 0.1 mL methanol solution of 0.05 M NaOH was added to the reaction mixture. The solution was kept under slow evaporation. After 2 weeks, Co1⋅OAc⋅bpy was obtained as purple crystals suitable for X-ray measurement. Yield (58 mg, 65%). M.p = 223 ^°C. Selected IR data (ATR, cm⁻¹): 3060w (Ar–H), 2965w (methyl), 1653m (C=O, acetate), 1606s (C =C, C =N), 1596vs (C =C, C =N), 1507m, 1445vs, 1243s, 1158m, 1119m, 1035m, 959s, 856s, 731s, 623m. MS (EI, Calc. m/z = 640.58): 502 (M – acetate – SO₂Me, 5%), 424 (M – bpy & OAc, 50%), 368 (ligand 1), 345, 289, 267, 210, 156 (bpy, 100%), 128. Anal. calc. for C₂₄H₂₂CoN₆O₄S₂: C, 49.57; H, 3.81; N, 14.45; S, 11.03%. Found: C, 49.14; H, 3.83; N, 14.20; S, 10.84%.

[Co1₂][Et₃NH]: Ligand 1 (99 mg, 0.27 mmol) and cobalt(II) acetate tetrahydrate (35 mg, 0.14 mmol) reacted in methanol (2 mL) together with 0.1 mL of triethylamine (Et₃N). The solution was allowed to slowly evaporate for 2 weeks, which led to the formation of purple crystals that are suitable for X-ray measurement. Yield (61 mg, 25%). M.p = 301 ^°C. Selected IR data (ATR, cm⁻¹): 3073m (Ar–H), 2933w (methyl), 1595s (C =C, C =N), 1569s, 1474s, 1330s, 1300vs, 1244s, 1041s, 958vs, 867s, 840vs, 751s, 726vs, 625s. MS (ESI, Calc. m/z = 893.95): 916 (M + Na⁺, 15%), 758 (M – 2 methyl – Et₃NH⁺, 45%). Anal. calc. for C₃₄H₄₃CoN₉O₈S₄: C, 45.68; H, 4.96; N, 14.10; S, 14.35%. Found: C, 45.54; H, 4.98; N, 13.87; S, 14.13%.

Co2⋅OAc⋅bpy: Ligand 2 (53 mg, 0.10 mmol), 2,2-bipyridine (15 mg, 0.10 mmol) and cobalt(II)acetate tetrahydrate (26 mg, 0.10 mmol) were added together in 3 mL ethanol. 0.1 mL of 0.05 M methanol solution of NaOH was added to the reaction mixture. The solution was slowly evaporated. After 2 weeks, purple crystals of Co2⋅OAc⋅bpy suitable for X-ray measurement were obtained. Yield (63 mg, 79%). M.p = 236 ^°C. Selected IR data (ATR, cm⁻¹): 3063w (Ar–H), 2978w (methyl), 1606s (C =C, C =N), 1594s (C =C, C =N), 1569m, 1497s, 1451vs, 1357vs, 1295vs, 1247vs, 1159m, 1138vs, 1080vs, 965s, 911s, 836vs, 763vs, 686vs, 624vs. MS (ESI, Calc. m/z + Na⁺  = 815.76): 815 (M + Na, 45%), 733 (M – acetate, 100%), 659 (M – bpy, 25%), 578 (ligand 2 + Co³⁺, 20%). Anal. calc. for C₃₆H₃₀CoN₆O₄S₂⋅2H₂O: C, 55.07; H, 4.50; N, 10.14; S, 7.74%. Found: C, 54.99; H, 4.37; N, 10.11; S, 7.74%.

Co2⋅dmap⋅w: Ligand 2 (53 mg, 0.10 mmol), 4-dimethylaminopyridine (dmap) (12 mg, 0.10 mmol) and cobalt(II)acetate tetrahydrate (26 mg, 0.10 mmol) reacted in ethanol (3 mL). The solution was slowly evaporated. After a week, yellowish crystals of Co2⋅dmap⋅w suitable for X-ray measurement were filtered, washed with ethanol and air-dried. Yield (51 mg, 71%). M.p = 291 ^°C. Selected IR data (ATR, cm⁻¹): 3166w (OH, water), 2918w (methyl), 1612vs (C =C, C =N), 1589vs (C =C, C =N), 1533s, 1465s, 1381s, 1280s, 1228vs, 1045s, 1016vs, 942vs, 834m, 781vs, 707s, 661vs, 633s. MS (EI, Calc. m/z = 716.72): 606 (M – tolyl – H₂O, 100%), 577 (ligand 2 + Co, 20%), 547 (M – 2Me – dmap – H₂O, 80%), 121 (dmap, 20%). Anal. calc. for C₃₃H₃₃CoN₆O₄S₂: C, 55.30; H, 4.64; N, 11.73; S, 8.95%. Found: C, 55.48; H, 4.70; N, 11.47; S, 8.68%.

2.3. Catalytic oxidation of o-aminophenol to 2-aminophenoxazin-3-one

Phenoxazinone-synthase-like activity was studied by deploying 100 equivalents of o-aminophenol (OAPH) relative to concentration of the individual complexes maintained at 5.0 × 10⁻⁵ M in acetonitrile and under aerobic conditions at a thermostat temperature of 25 ^°C. The reaction was followed spectrophotometrically by monitoring the increase in the absorbance of phenoxazinone chromophore as a function of time at 426 nm (ε = 17,278 M⁻¹ cm⁻¹), which is characteristic of 2-aminophenoxazin-3-one in acetonitrile. The 2-aminophenoxazin-3-one molecule was isolated, fully characterized and used for the generation of a calibration curve on the UV–Vis spectrophotometer.

For the catalytic oxidation of 2-amino-4-chlorophenol (Cl-OAPH) to 2-amino-8,10a-dichloro-10,10a-dihydro-3H-phenoxazin-3-one (Cl-APX, Scheme 2), 2.5 × 10⁻⁵ M of the complexes was utilized in the presence of 100 equivalents of Cl-OAPH while the increase in absorbance of the phenoxazinone chromophore was monitored at 433 nm. Absorbance spectra of the solutions were measured at time intervals of 30 minutes for all the complexes. The initial rate method was applied to determine the rate of reaction by linear regression from the slope of absorbance values against time.

2.4. Crystal structure determinations

Single crystals of the complexes were obtained by slow evaporation of the solution of the complexes in ethanol or methanol. The intensity data for the compounds were collected on a Nonius KappaCCD diffractometer using graphite-monochromated Mo-K_α radiation. Data were corrected for Lorentz and polarization effects; absorption was taken into account on a semi-empirical basis using multiple scans [35, 36, 37, 38]. The structures were solved by direct methods (SHELXS) and refined by full-matrix least squares techniques against Fo₂ (SHELXL-97 and SHELXL-2014) [38, 39]. The hydrogen atoms bonded to the water molecule O5 of Co2⋅dmap⋅w were located by difference Fourier synthesis and refined isotropically. All other hydrogen atoms were included at calculated positions with fixed thermal parameters. All non-hydrogen and non-disordered atoms were refined anisotropically [38]. The crystal of Co2⋅OAc⋅bpy contains large voids, filled with disordered solvent molecules. The size of the voids is 281 Å³/unit cell. Their contribution to the structure factors was secured by back-Fourier transformation using the SQUEEZE routine of the program PLATON [40]. The program XP (SIEMENS Analytical X-ray Instruments, Inc. 1994) was used for structure representations [41]. Crystallographic data as well as structure solution and refinement details are summarized in Table 1.

3.1. Syntheses and characterization of cobalt(III) complexes

The octahedral cobalt(III) complexes were generally obtained in good yields by allowing Co(OAc)₂ and any of ligands 1 or 2 to interact at room temperature in a given solvent and in the presence of a selected co-ligand additive or base. Analytical characterization data of the complexes agree with the expected values and are further confirmed by X-ray crystal structures.

Although the complexes Co1⋅OAc⋅bpy, Co2⋅OAc⋅bpy and Co2⋅dmap⋅w possess mixed-ligand coordination environments, our deployment of triethylamine as a potential co-ligand preferentially yielded only the bis-ligand complex [Co1₂][Et₃NH], where the triethylamine only functions as a base and acts as a counter cation after becoming protonated (Scheme 3(a)). Furthermore, the tendency of these dianionic ligands to oxidize the cobalt(II) starting material into cobalt(III) products is also worthy of note. The very rare organometallic cyclocobaltated bond in Co2⋅dmap⋅w (shaded regions in Scheme 3(a)), which results from the C–H bond activation of a tolyl ring of ligand 2 (Scheme 3(b)), is also remarkable.

3.2. X-ray structural analyses

Complexes Co1⋅OAc⋅bpy, [Co1₂][Et₃NH] and Co2⋅OAc⋅bpy crystallized in the monoclinic P2₁/n, C2/c and P2₁/c space groups, respectively, while complex Co2⋅dmap⋅w crystallized in the triclinic Pī space group (Table 1). Crystal structures revealed that the synthesized ligands formed tridentate five–six-membered NˆNˆN chelation for complexes Co1⋅OAc⋅bpy, [Co1₂][Et₃NH] and Co2⋅OAc⋅bpy while tetradentate NˆNˆNˆC chelation is obtained for Co2⋅dmap⋅w due to cyclocobaltation (Figures 1–4). Such chelations suggest higher coordination stability from ligands 1 and 2 relative to the other coordinated co-ligands. Bond lengths and angles around the coordination centre of the complexes are within the expected values (see Table S1 of the Supplementary information) [42]. The diversity of coordinative saturation observed among the obtained complexes fulfils our deliberate aim of studying the influence of diverse coordination environments on catalyst efficiencies.

The interesting C–H activation of an ortho-proton on one of the tolyl rings of ligand 2, which leads to cyclometallation in complex Co2⋅dmap⋅w, can be attributed to the strong electron-withdrawing influence of the bonded sulfonamide function (Figure 3). A further factor that possibly supports the cyclometallation is the absence of sufficient anionic donors or insufficient pyridyl co-ligands to satisfy the coordinative saturation required by the Co(III) centre. Even a water molecule had to be taken up as the co-ligand to achieve an octahedral geometry. The only few reports of cyclometallated cobalt complexes have been obtained under classical organometallic inert conditions [43, 44, 45, 46]. To the best of our knowledge, the formation of a cyclocobaltated complex in the presence of moisture and air as in the preparation of Co2⋅dmap⋅w is rare. It is therefore plausible to conclude that incorporation of a sulfonamide substituent could be important for the purpose of facilitating C–H bond activations for organometallic syntheses [47].

In order to properly describe the coordination polyhedra around each cobalt(III) for Co1⋅OAc⋅bpy, [Co1₂][Et₃NH], Co2⋅OAc⋅bpy and Co2⋅dmap⋅w, Continuous Shape Measurement (CShM) calculations were carried out by using the experimentally obtained X-ray structural coordinates of the central cobalt atoms and their directly coordinated donor atoms. The results, which are summarized in Table 2, revealed that the coordination polyhedron around each cobalt(III) centre can best be described as octahedral (O_h) [48, 49, 50]. The distortion path analysis, which provides the percentage deviation from an ideal polyhedron on a scale of 0%–100%, revealed the lowest distortions from an ideal octahedron (i.e. 0.348–0.748%) relative to hexagonal pyramid (30.155–32.295%), pentagonal pyramid (25.530–27.265%), trigonal prism (11.964–19.84%) and Johnson pentagonal pyramid (29.178–30.560%) [51]. The complex Co2⋅dmap⋅w possesses the closest geometry to an octahedron arguably because of the water donor, which allows for the lowest distortive ligand–ligand repulsion and may imply some sort of coordinative stability in Co2⋅damp⋅w. However, the chelate effect in the bis-ligand complex [Co1₂][Et₃NH] would represent a much stronger coordinative stability.

3.3. Phenoxazinone synthase mimicking activity by cobalt complexes

The phenoxazinone synthase mimicking activity of the complexes was tested via oxidative coupling of OAPH or its derivative 2-amino-4-chlorophenol (Cl-OAPH) in the presence of 1 mol% of cobalt(III) and at 25 ^°C (Scheme 2). The relative catalytic efficiencies for the various complexes were evaluated via absorbance changes (Figure 5(a and b)) as a function of time. Estimates of the initial rate values (V) were determined by linear regression from the slopes of absorbance versus time plots (Figures 6 and 7). These estimates were used for quantitative comparison of catalytic efficiencies among the prepared complexes (Tables 3 and 4).

Insignificant absorbance changes were observed in the absence of the complexes, which proves the importance of the metal ions (Figure 6(a and b) and Table 3). Correlation could be observed between the coordination features and the catalytic outcomes. The bis-ligand complex [Co1₂][Et₃NH], which is the most strongly chelated analogue, has remarkably the least catalytic productivity among the complexes, which could be attributed to the difficulty in releasing the metal centre from the hold of two tridentate chelators. This emphasizes the importance of free coordination sites for the catalysis and that a ligand with strong chelation characteristics as obtainable with porphyrin ligands should be avoided when designing metalloenzyme models.

On the other hand, from a comparison among complexes assembled along with co-ligands [i.e. Co1⋅OAc⋅bpy (23.30 × 10⁻³ hr⁻¹) > Co2⋅OAc⋅bpy (15.48 × 10⁻³ hr⁻¹) > Co1⋅dmap⋅w (13.30 × 10⁻³ hr⁻¹)], the catalyst efficiencies appear to correlate with the percentage extent of deviation from the regular octahedron according to the CShM calculations (i.e. Co1⋅OAc⋅bpy = 0.748% > Co2⋅OAc⋅bpy = 0.598% > Co1⋅dmap⋅w = 0.348%; Table 2). It is therefore plausible to conclude that the general state of steric hindrance in the various coordination polyhedra, which is associated with ligand and co-ligand sizes and the corresponding ligand–ligand steric repulsions, may be accountable for the observed differences in catalytic productivities of the different metal centres. A strained coordination polyhedron hints at the readiness for dissociative events, which is required for creating free substrate-binding sites around the metal ion. Additionally, of the three mixed-ligand complexes, the cyclocobaltated complex Co2⋅dmap⋅w has the tetradentate trianionic complexation feature of ligand 2 as further reason for its poorer catalyst performance.

Complexes Co1⋅OAc⋅bpy and Co2⋅OAc⋅bpy, which possess higher catalytic yields for the coupling of OAPH, are further deployed for the coupling of Cl-OAPH at 25 ^°C. Their resulting initial rates are 32.06 × 10⁻³ hr⁻¹ and 26.06 × 10⁻³ hr⁻¹, respectively, which reveal higher initial rate values relative to the experiments without cobalt(III) complex (Figure 6(b)). A similar coupling reaction for Cl-OAPH in the absence of cobalt(III) also yielded negligible products (see Supplementary information Table S2). Yet it is noteworthy that Co1⋅OAc⋅bpy shows a higher catalytic efficiency than Co2⋅OAc⋅bpy towards the oxidative coupling of Cl-OAPH. It is also worthy of mention that coupling of the chloro-substituted substrate Cl-OAPH is less favoured compared to the non-substituted OAPH, which can be expected on the grounds of steric influence of the chloro-substituent (Figure 6(b) vs. (a)). A proposed scheme of catalytic coupling by complex Co2⋅OAc⋅bpy has been summarized in Scheme 4.

3.4. Temperature variation and phenoxazinone mimicking activities

The effect of temperature variation from 25 ^°C to 65 ^°C on the phenoxazinone activity was examined using complexes Co1⋅OAc⋅bpy, Co2⋅OAc⋅bpy and Co2⋅dmap⋅w as catalytic models. The results show a remarkable increase in catalytic activity as the temperature increases (Table 4 and Figure 7). This agrees with the increasing tendency to dissociate co-ligand donors, which then creates vacant coordination sites for substrate binding. Thus, the tendency of the complexes to generate free coordination sites could on one hand depend on the coordinative strain as well as on thermal dissociations on the other hand.

A very important observation during the increase in catalysis temperature from 25 ^°C (Figure 6(a)) through 35 ^°C, 45 ^°C (Figure 7(a)) and 55 ^°C to 65 ^°C (Figure 7(b)) is the steady improvement in performance for complex Co2⋅OAc⋅bpy. This complex displayed the second best efficiency at 25 ^°C and thereafter overtakes the originally most active complex Co1⋅OAc⋅bpy from 35 ^°C (Table 4). The superior activity for Co2⋅OAc⋅bpy at higher temperatures relative to Co1⋅OAc⋅bpy might be attributed to a higher susceptibility to the thermal dissociation of acetate or bipyridine co-ligands in Co2⋅OAc⋅bpy. The bulkiness of the tolyl substituents in Co2⋅OAc⋅bpy appears to be a contributory factor to its thermal susceptibility [52]. In general, the catalysis outcomes suggest that the availability of a vacant binding site during the catalytic process is important.