Anion-exchange of 1We investigated anion-exchange characteristics of 1, prepared using the liquid-liquid interfacial coordination reaction technique established for the synthesis of coordination polymer nanofilms31,32,33,34,35,36. 1 was obtained as an orange film, formed at the liquid-liquid interface. Energy dispersive X-ray spectroscopy under scanning electron microscope observation (SEM/EDS) revealed that 1 includes uniformly distributed constituting elements, C, N, Co, and Cl (Supplementary Fig. 1a). Raman spectrum showed the C=N stretching peak at 1585 cm−1, shifted from that of the free terpyridine ligand by 18 cm−1 (Supplementary Fig. 1b). X-ray photoelectron spectroscopy (XPS) gave the same spectra for 1 as the previous study (Supplementary Fig. 1c)31. The cyclic voltammograms of 1 (Supplementary Fig. 2) showed intense reversible redox waves at −1.13 V vs. (ferrocenium/ferrocene) Fc+/Fc and −1.88 V vs. Fc+/Fc, attributed to the [Co(tpy)2]2+/[Co(tpy)2]+ and [Co(tpy)2]+/[Co(tpy)2]0 redox couples, respectively. The voltammogram also showed weak reversible redox wave at −0.13 V vs. Fc+/Fc derived from the [Co(tpy)2]3+/[Co(tpy)2]2+ redox couple. These characterization data were identical to the previous reports31, and confirmed the preparation of 1.Anion-exchange to [Ni(mnt)2]n− was completed by simply immersing a 1 film into (nBu4N)n[Ni(mnt)2] solutions (n = 1, 2), which afforded anion-exchanged polymers containing [Ni(mnt)2]− (2) and [Ni(mnt)2]2− (3) (Fig. 2a, Supplementary Table 1). The optimal solvents identified for the anion-exchange process were CH3CN and C2H5OH for (nBu4N)2[Ni(mnt)2] and (nBu4N)[Ni(mnt)2], respectively. After the reaction, the polymer films exhibited coloration due to the incorporated metalladithiolate ions (Fig. 2b). SEM/EDS confirmed the absence of Cl and presence of S and Ni in the polymer films for both monovalent and divalent anions (Fig. 2c). In SEM/EDS elemental mapping, uniform distribution of all constitute elements was evident (Fig. 2d,e). The S, Co, and Ni peak area analysis from the SEM/EDS indicated incorporation ratio of [Ni(mnt)2]n− ions in 2 and 3 of 0.55 : 1, aligning with the expected stoichiometric ratio (0.5 : 1) within the bounds of SEM/EDS semi-quantitative analysis accuracy. These findings affirm the efficacy of the anion-exchange reaction in 1. The cross-sectional elemental mapping by scanning transmission electron microscopy confirmed that Cl was not detected in the polymer films (Supplementary Fig. 3), meaning that the anion-exchange reaction efficiently proceeded without the limitation at the surface of 1. XPS further validated the anion-exchange of 1, evidenced by the disappearance of Cl 2p peak as shown in Supplementary Fig. 4. Instead, following the anion-exchange, peaks corresponding to S, and Ni appeared. The N 1s peak from tetrabutylammonium ions around 402 eV (Supplementary Fig. 4) was absent in the XPS, indicating no adsorption of the ion pairs onto the polymers. UV-vis-NIR spectroscopy (Fig. 2f) corroborated the predominant incorporation of the respective metalladithiolates into 2 and 3. Specifically, the distinct peak at 889 nm, linked to the π-π* transition of the [Ni(mnt)2]− ion, was observed in the spectrum of 2 (Supplementary Fig. 5). Conversely, 3 exhibited no NIR absorption band because the [Ni(mnt)2]2− ion has no absorption in the NIR region. Atomic force microscopy (AFM) demonstrated that the polymer sheets thickened after the anion-exchange, suggesting the substitution of chloride ions with the bulkier metalladithiolene anions (Supplementary Fig. 6). The electrochemical rest potentials of 2 and 3 were −0.18 V and −0.30 V vs. Fc+/Fc, respectively. Given that the redox potential of [Ni(mnt)2]−/[Ni(mnt)2]2− was −0.27 V vs. Fc+/Fc37, the measured rest potentials are indicative of the oxidation states of the metalladithiolate ions involved in the anion-exchange reaction. These findings confirm the efficiency of anion-exchange process in 1, whose electrochemical rest potential was −0.46 V vs. Fc+/Fc32.Fig. 2: Anion-exchange of 1.a Photograph of anion-exchange reaction of 1. b Photograph of 1−3. c SEM/EDS spectra of 1−3. d, e SEM images and SEM/EDS mappings of 2 and 3 (Scale bar: 10 μm). f UV-vis-NIR spectra of 1–3.Raman spectroscopy provided the electronic structure information about the polymer films. Raman spectra of 1–3 in the Raman shift below 2000 cm−1 displayed similar peak patterns (Fig. 3a), indicating that the cationic polymer framework remained intact after the anion-exchange. In the spectra after the anion-exchange, the C≡N stretching peaks of [Ni(mnt)2]n− complexes observed at approximately 2200 cm−1 were dependent on their oxidation states. Two peaks appeared at 2221 and 2200 cm−1 for 2, whereas a peak was observed at 2191 cm−1 for 3 (Fig. 3b). According to the Raman spectra of (nBu4N)+ salts of [Ni(mnt)2]n− complexes (Supplementary Fig. 7), C≡N stretching peak at 2215 cm−1 was attributed to the monovalent anion, while the peak at 2191 cm−1 was the divalent anion. Therefore, while the dithiolene complex exists as [Ni(mnt)2]2− in 3, both [Ni(mnt)2]− and [Ni(mnt)2]2− oxidation states coexist in the film in 2. These findings imply a partial charge transfer between the [Ni(mnt)2]− and [Co(tpy)2]2+ moieties.Fig. 3: Evaluation of the electronic states of 1–3.a Raman spectra of 1–3. b Expansion of the Raman spectra in panel a featuring C≡N stretching peaks. c XP spectra of 2 and (nBu4N)[Ni(mnt)2] in S 2p core level (left), and 1 and 2 in Co 2p core level (right).This partial charge transfer is confirmed by XPS. In the XP spectrum of 2 in Fig. 3c, the S 2p peak was observed at 161.5 eV, which was slightly lower than those of (nBu4N)2[Ni(mnt)2] that appeared at 163.0 eV. This result indicates partial reduction of the monovalent nickelladithiolene complexes in the cationic polymer framework. Additionally, the Co 2p3/2 peak shifted to higher binding energy, meaning the partial oxidation of the [Co(tpy)2]2+ moieties. Therefore, electron transfer from [Co(tpy)2]2+ to [Ni(mnt)2]− occurred after the anion-exchange. These peak shifts on XP spectra were not observed in 3, suggesting no charge transfer interaction between [Co(tpy)2]2+ to [Ni(mnt)2]2− (Fig. 3c).To investigate the driving force of the anion-exchange reaction from chloride to [Ni(mnt)2]n−, the inverse anion-exchange reaction was performed for 2 and 3 with 5 mM nBu4NCl solution. The SEM/EDS revealed that the coexistence of metalladithiolene anions and chloride ions after the reaction, with the approximately 60% and 70% anion-exchange, respectively (Supplementary Fig. 8). The excess Cl− did not replace the metalladithiolenes completely. These results indicated that the anion-exchange from chloride to [Ni(mnt)2]n− was a thermodynamically favoured reaction. π-π interactions and charge-transfer interactions between the bis(terpyridine)cobalt(II) polymer backbone and the metalladithiolenes are preferable while chloride-π interactions were less interactive38, which is the plausible origin of the driving force to the anion-exchange reaction.Capacitive response of 1The response of 1 to external electric field was examined using IDA electrodes in a dry condition without additional supporting electrolyte (Supplementary Fig. 9). Figure 4a shows the representative I-V curves of 1 between −1 and +1 V with varying scan rates. The I-V curves are dependent on the scan rate. The width of the I-V curves increases proportionally to the scan rate (Fig. 4b). This relationship is a characteristic of the charge/discharge dynamics typical for an electrochemical supercapacitor and thought to stem from movement or displacement of chloride ions within the polymer framework when external electric field is applied. Based on the width of electrochemical double layers, the areal and volumetric capacitance of the 1 was estimated as 7.8 ± 4.0 μF/cm2 and 0.19 ± 0.09 F/cm3, respectively. Electrochemical impedance spectroscopy (EIS) further elucidated the capacitive nature of 1. The resulting spectrum was modelled using an equivalent circuit that included contact and film resistance (R), charge transfer resistance (R1), Warburg resistance for the diffusion of Cl− (Wdiff), capacitance of the 1 film (C1), leak resistance (R2), and pseudocapacitance for the faradaic process (C2) as shown in Supplementary Fig. 1039,40. The low-frequency phase angle approached approximately 45°, indicative of diffusion-controlled process. The capacitance C1 was consistent to that measured from the cyclic voltammetry. Galvanostatic charge/discharge cycles revealed rapid decrease of capacity within the initial cycles, stabilizing at ca. 25% of the initial capacity (Fig. 4c, d). While these capacity and stability metrics are modest, our findings confirm the potential of chloride-containing bis(terpyridine)metal(II) polymers as solid-state electrolyte for all-solid-state supercapacitor41,42. This paves the way for the future enhancements in microsupercapacitors based on M(tpy)2 complexes through structural refinement.Fig. 4: Capacitive response of 1.a I-V curves for 1 with different scan rate. b Scan rate-dependence of the width of I-V curves in a at 0 V. c Galvanostatic charge-discharge curves of 1 from 0 to +1 V recorded with 1 nA s−1. d Cycle-dependence of volumetric capacity and retention efficiency.Conductive response of 2 and 3The response to the external electric field after the anion-exchange was investigated using IDA electrodes. Anion-exchange reaction was performed for film 1 immobilized on IDA electrodes, and the conductivity measurements were performed by the two probe method at room temperature (ca. 300 K). Figure 5a depicts the I-V curves −1 V and +1 V for 2. While the initial 1 responded as a microsupercapacitor, 2 exhibited the almost linear I-V curve with the conductivity of 1.1 ± 0.2 ×10−8 S cm−1. Temperature-dependent conductivity measurement revealed that the conductivity of 2 decreased with increasing temperature, suggesting the semiconductive nature of 2 (Fig. 5b). From the Arrhenius plot in Fig. 5b, the activation energy (Ea) was calculated to be 0.33 ± 0.01 eV. These results indicated that the anion-exchange from Cl− to [Ni(mnt)2]− endowed the drastic change in responses to external electric field.Fig. 5: Electrical conductivity of 2 and 3.a I-V curves for 2 and 3. b Arrhenius plot detailing the temperature-dependent conductivity of 2. c Schematic illustration of the plausible conductivity mechanism based on electron hopping between partial charge transfer metal complex sites in 2. Electron hopping between redox active [Co(tpy)2]m+ (m = 2 or 3) and [Ni(mnt)2]n− sites is responsible for electron transport. The colours of each metal complex site depict the difference in the oxidation states.Conversely, the conductivity of 3 falls below the detection limit of the measuring apparatus (<10−11 S cm−1), clarifying that 3 is an insulator, whose conductivity is lower than that of 2 by 4 orders of magnitude. Additionally, the I-V curves of 3 showed no dependence on the scan rates. Therefore, [Ni(mnt)2]2− anions neither function as electron conductors nor as electrolytes. The capacitive difference between chloride and the divalent nickeladithiolene complexes can stem from the relatively large volumetric size of [Ni(mnt)2]2−.The modulation of responses to external electric field upon anion-exchange can be explained by electronic interactions between the cationic polymer backbone and anions. The conductive nature of 2 originates from the charge-transfer between [Co(tpy)2]2+ and [Ni(mnt)2]− moieties. Because the redox potentials of [Co(tpy)2]3+/2+ (−0.15 V vs Fc+/Fc) and [Ni(mnt)]−/2− (−0.27 V vs Fc+/Fc) redox couples are close to each other (ΔE = 0.12 V)31,37, partial charge-transfer interactions from [Co(tpy)2]2+ to [Ni(mnt)2]− is expected, causing the mixed-valence states and thus inducing electric conductivity to the 2 film through the redox-conduction mechanism (electron hopping between redox active sites)43,44. Conversely, such charge-transfer does not occur in 3, resulting in its insulating behaviour.The hypothesis on the partial charge-transfer model is also supported by the response to external electric field of a bis(terpyridine)cobalt(II) polymer impregnated with another monoanionic metalladithiolene complex, [Ni(tdt)2]− (tdt: 4-toluene-1,2-dithiolato). The [Ni(tdt)2]−-containing polymer (4) was successfully prepared via anion-exchange reaction (Supplementary Table 1 and Supplementary Figs. 11–13). The electric conductivity of 4 was measured via the same procedure using IDA electrodes, indicating that 4 was insulating with lower electrical conductivity than the detection limit (Supplementary Fig. 14). The redox potential of [Ni(tdt)2]−/[Ni(tdt)2]2− is −0.95 V vs. Fc+/Fc45,46, which does not match the oxidation potential of the polymer backbone (−0.14 V vs. Fc+/Fc). The difference between the redox potentials of the cationic framework and [Ni(tdt)2]− was ca. 0.8 V, indicating that the charge-transfer interaction was not expected. Therefore, the conductive behaviour stems from the host-guest charge-transfer interaction.To investigate the conductivity mechanism of 2, the I-V curve was measured with wider voltage range between −6 V and +6 V (Supplementary Fig. 15a). In the wider range, the current was not linear to the applied voltage (Supplementary Fig. 15b). This non-linear I-V curves were fitted with the simulation based on the redox conduction mechanism47. In addition, the potential-dependent conductivity measurement of 2 shows the conductivity increasing around −0.1 V vs. Fc+/Fc, near to the [Co(tpy)2]3+/[Co(tpy)2]2+ couple (Supplementary Fig. 16). These potential-dependences of conductivity also indicated the electron-hopping-based charge-transfer mechanism32,43. Furthermore, the lower conductivity than 1 at the same potential region32 suggested that the electron hopping between [Ni(mnt)2]n− and [Co(tpy)2]m+ sites is critical in the conductivity. These results indicate that the conductivity of 2 stems from electron hopping between redox active metal complex sites in 2.Further investigation on the conductivity mechanism of 2 was performed through electrochemical analysis (Supplementary Fig. 17). The cyclic voltammograms of 2 revealed a redox wave corresponding to the [Co(tpy)2]2+/[Co(tpy)2]+ redox couple at −1.14 V vs. Fc+/Fc. Notably, the reduction wave of the redox couple decreased in subsequent cycles, stabilizing after the second cycle. This behaviour was not seen in the cyclic voltammograms of 3, which is almost identical to the cyclic voltammogram of 1. The decrease in the redox wave intensity in 2 can be ascribed to the charge-trapping phenomena of the redox couple of [Ni(mnt)2]−/[Ni(mnt)2]2− mediated by the electron transfer based on electron hopping mechanism through the cationic framework32. If the band-like electron transfer between the [Ni(mnt)2]n− sites is critical to the electronic conductivity in 2, the redox wave of the [Ni(mnt)2]−/[Ni(mnt)2]2− couple would be directly observed in the cyclic voltammogram. Therefore, the presence of the charge trapping effect also indicates that the electron hopping between [Co(tpy)2]m+ and [Ni(mnt)2]n− sites based on redox-conduction mechanism is a plausible electron transport pathway in 2 (Fig. 5c).
