A mononuclear metal complex, Ni(Hm-dtc)₂, was prepared using a procedure similar to that previously reported [33,34]. The reagents were purchased from Tokyo Kasei Kogyo Co., Ltd. (Tokyo, Japan) and Aldrich Chemical Co., Inc. (St Louis, MO, USA). All the chemicals were used without further purification.

An acetonitrile solution (10 mL) of CuBr·S(CH₃)₂ (0.041 g, 0.2 mmol) was diluted with 10 mL of acetone, and this solution was added to 20 mL of a CHCl₃ solution of Ni(Hm-dtc)₂ (0.041 g, 0.10 mmol). The reaction mixture was stirred for 5 min and then filtered. Crystals of complex 1a suitable for X-ray analysis were obtained from the filtrate in several days. Anal. Calcd for [Cu^I₂Ni^IIBr₂(Hm-dtc)₂(CH₃CN)₂]_n (C₁₈H₃₀Br₂Cu₂N₄NiS₄): C, 27.85; H, 3.90; N, 7.22. Found: C, 27.64; H, 3.81; N, 7.18.

An acetonitrile solution (10 mL) of CuI (0.038 g, 0.2 mmol) was diluted with 10 mL of acetone, and this solution was added to 20 mL of a CHCl₃ solution of Ni(Hm-dtc)₂ (0.041 g, 0.10 mmol). The reaction mixture was stirred for 5 min and then filtered. Crystals of complex 1b suitable for X-ray analysis were obtained from the filtrate in several days. Anal. Calcd for [Cu^I₂Ni^III₂(Hm-dtc)₂(CH₃CN)₂]_n (C₁₈H₃₀Cu₂I₂N₄NiS₄): C, 24.84; H, 3.47; N, 6.44. Found: C, 24.67; H, 3.43; N, 6.41.

Data collections of the crystals for complex 1a and 1b were performed on a Rigaku RAXIS RAPID imaging plate area detector with graphite-monochromated Mo-Kα (λ = 0.71073 Å) radiation. The crystal structures were solved by direct methods (SHELX97 [35] for 1a and SIR2008 [36] for 1b) and expanded using the Fourier techniques. The non-hydrogen atoms were refined anisotropically. Hydrogen atoms were refined using the riding model. The final cycles of full-matrix least-squares refinement [37] on F² for 1a and 1b were based on 2966 and 3060 observed reflections and 142 and 142 variable parameters, respectively. All calculations were performed using the CrystalStructure [38] crystallographic software package except for refinement, which was performed using SHELXL-97 [35].

The coordination polymers (0.3 mol), [Cu^I₂Ni^IIX₂(Hm-dtc)₂(CH₃CN)₂]_n [X = Br^– (1a), I^– (1b)], [Cu^I₂Cu^IIX₂(Hm-dtc)₂(CH₃CN)₂]_n [X = Br⁻ (2a), I⁻ (2b)] and {[Cu^I₄Cu^II₂Br₄(Pyr-dtc)₄]·CHCl₃}_n (3) were mixed with TiO₂ paste (0.1 mL) for the electrode, where the TiO₂ paste for low-temperature sintering (PECC-K01) was purchased from Peccell Technologies Co. (Kanagawa, Japan) and the mixtures of the TiO₂ paste and the coordination polymers were grinded in a mortar for 10 min, and painted on an indium tin oxide (ITO, sheet resistance 10 Ω/sq) glass substrate with spacers of Kapton tape (thickness 50 µm). It was annealed at 50 °C for 0.5 h. For the counter-electrode, a solution of poly(3,4-ethylenedioxythiophene)-tetramethacrylate (PEDOT-TMA) was mixed with ethanol at a volume ratio of 5:1. The mixed solution was spin-coated (2000 rpm for 10 s) on ITO glass and annealed at 50 °C for 10 min. The two prepared electrodes were assembled to fabricate a sandwich-type cell with a 50-µm spacer, and the electrolyte, which was a solution of lithium iodide (LiI, 0.5 M) and iodine (I₂, 0.05 M) in polyethylene glycol, was injected between the electrodes. The performance of the solar cell was measured under irradiation with an AM 1.5 G (100 mW/cm²) solar simulator (Bunkoukeiki Co., KP-3201). The cell device area was 0.30 cm².

UV-Vis-NIR spectra were monitored on a U-4100 UV/VIS/NIR Spectrophotometer (HITACHI). The photoemission yield spectra were measured by Surface Analyzer AC-2 (RIKEN KEIKI Co.). The DC conductivities were investigated with an Ultra High Resistance Meter R8340 (ADVANTEST) on a pellet of the powder sample sandwiched by 13 mm diameter brass electrodes. The impedance and dielectric measurements were carried out on a pellet of the powder sample sandwiched by 13 mm diameter brass electrodes with a 6440B (Wayne Kerr Electronics) Series Precision Component Analyzer in a frequency range 100 to 3 MHz.

Single-crystal X-ray analysis of complex 1a reveals the formation of a coordination polymer, [Cu₂NiBr₂(Hm-dtc)₂(CH₃CN)₂]_n, with a 1D infinite chain structure [39], as shown in Figure 1a. This complex consists of mononuclear copper units, Ni(Hm-dtc)₂, and dinuclear copper units, Cu₂Br₂(CH₃CN)₂, with bromide ions and acetonitrile molecules to construct the 1D infinite chain. The nickel ion of the mononuclear unit, Ni1, has a square-planar coordination geometry containing four sulfur atoms of two dithiocarbamate ligands, which form four-membered chelate rings. Each copper ion in the dinuclear unit, Cu1, has a distorted tetrahedral coordination geometry with two bridging bromide anions, one nitrogen atom of acetonitrile, and one sulfur atom of the mononuclear unit. In the mononuclear Ni(Hm-dtc)₂ unit of the coordination polymer 1a, the average Ni1–S distance (2.2212(7) Å) is slightly longer than those of the typical average Ni(II)–S distances in Ni(II)–dithiocarbamate complexes such as Ni^II(EtOH₂dtc)₂ (2.201 Å) [40], Ni^II(Pip-dtc)₂ (2.202 Å) [41], Ni^II(echdtc)₂ (2.2056 Å) [42] and Ni^II{S₂CN(CH₂CH₂OMe)₂}₂ (2.2038 Å) [43], and is shorter than those in Cu(II)–dithiocarbamate complexes [33] such as Cu^II(Me₂dtc)₂ (2.3105 Å), Cu^II(Et₂dtc)₂ (2.3117 Å), Cu^II(n-Pr₂dtc)₂ (2.3183 Å) and Cu^II(i-Pr₂dtc)₂ (2.2886 Å). The structural data for the Ni1–S distances therefore indicate that the Ni1 ion is divalent. The Cu1 ion in the dinuclear unit has a distorted-tetrahedral coordination geometry typical of monovalent copper ions. The Cu1–Br1 and Cu1–Br1* distances are 2.4638(4) Å and 2.5519(5) Å, respectively. In order to confirm the oxidation state of the Cu1 ion, we performed bond valence sum (BVS) calculations [44]. The estimated BVS value for the Cu1 ion was 1.19, indicating the monovalent oxidation state. On the basis of its charge neutrality, it is concluded that the chemical formula of 1a is [Cu^I₂Ni^IIBr₂(Hm-dtc)₂(CH₃CN)₂]_n; the square-planar Ni1 of the mononuclear unit Ni(Hm-dtc)₂ is divalent and the tetrahedral Cu1 in the dinuclear unit is monovalent. Figure 1c shows the packing diagram of the 1D chains for complex 1a,viewed parallel to the a-axis. The nearest-neighbor S•••S, S•••Br, and Cu•••Cu separations between the 1D chains were 3.8910(6) Å, 6.9921(3) Å, and 8.1909(6) Å, respectively. These separations were larger than those of the sums of the van der Waals radii of the sulfur, bromide, and copper atoms.

[Cu^I₂Ni^III₂(Hm-dtc)₂(CH₃CN)₂]_n (1b), which contains iodide ions, is isostructural (Figure 1b) [45] with complex 1a. The average Ni1–S distance (2.2113(8) Å) in the mononuclear unit is similar to that of complex 1a, and the estimated BVS value for the Cu1 ion with tetrahedral coordination geometry is 1.19. The packing diagram of the 1D chains for complex 1b is shown in Figure 1d. The nearest-neighbor S•••S, S•••Br, and Cu•••Cu separations between the 1D chains were 4.0004(3) Å, 6.9452(8) Å, and 8.2724(8) Å, respectively.

Figure 2a shows the absorption spectra obtained via diffuse reflectance spectroscopy of the mononuclear complex Ni(Hm-dtc)₂ (black) [46] (Figure S1 in the Supporting Information), and coordination polymers 1a (red) and 1b (blue), doped in MgO; the measurements were carried out with pellet samples made by mixing each metal complex (0.01 mmol) with MgO powder (80 mg); the absorbances were plotted as the Kubelka–Munk function [47]:

(1)

where R is the absolute reflectance of the samples. The 1D coordination polymers exhibit two absorption bands in the visible region. The larger absorption bands at around 380 nm for Ni(Hm-dtc)₂, 1a, and 1b could be attributed to charge transfer. The small absorption bands at around 640 nm for Ni(Hm-dtc)₂, 1a, and 1b may arise from d–d transitions of the nickel(II) ion. The HOMO–LUMO gaps (band gaps) of the Ni(Hm-dtc)₂ coordination polymers 1a and 1b were estimated from the absorption edges (Figure 3b) to be 1.58 eV, 1.34 eV, and 1.57 eV, respectively. The HOMO–LUMO gaps of the coordination polymer 1a are smaller than those of 1b and those of other 1D coordination polymers with dithiocarbamate ligands, such as 1.48 eV for 2a and 1.48 eV for 2b, and larger than that of the 3D coordination polymer 3 (1.01 eV). In order to compare the absorption intensity, the transmission spectra with the KBr powder pressed pellets including the coordination polymers 1a, 1b, 2a and 2b (0.5 µmol in 50 mg KBr) were measured and shown in Figure S3, which indicates the decreasing of the transition probability of the d-d transitions around 600 nm for the Cu(I)–Ni(II) heterometal coordination polymers 1a and 1b as against the Cu(I)–Cu(II) mixed-valence coordination polymers 2a and 2b.

The photoemission yield curves of the coordination polymers 1a and 1b are shown in Figure 3. The HOMO energy levels of complexes 1a, 1b, and 3 (Figure S2) are −5.24, −5.23, and −5.28 eV, respectively. These values are sufficiently low for the complexes to accept electrons from the redox electrolyte (−4.8 eV). From the HOMO values, the LUMO energy levels obtained from the band edges of the absorption spectra are determined to be −3.90 eV for 1a, −3.66 for 1b, and −4.27 eV for 3. These indicate that the energy level of the LUMO for the 3D coordination polymer 3 is lower than that of the conduction band (E_CB) of the TiO₂ semiconductor (−4.0 eV); in contrast, the energy levels of the LUMO for the 1D Cu(I)–Ni(II) heterometal coordination polymers 1a and 1b are well matched with the E_CB of the TiO₂ semiconductor, enabling electron injection from the dyes 1a and 1b.

Most of the Cu(I)–Cu(II) mixed-valence coordination polymers with dithiocarbamate ligands show semiconducting properties because the overlaps of the d-orbitals and HOMOs and/or LUMOs of the ligands in the Cu(II)–dithiocarbamate coordination polymers cause the formation of narrow conduction and/or valence bands, inducing strong magnetic interactions and conducting properties. It is predicted that Cu(I)–Ni(II) heterometal coordination polymers containing Ni(II)–dithiocarbamate units will show lower conductivities than those of Cu(I)–Cu(II) mixed-valence coordination polymers with dithiocarbamate ligands since the overlaps of the d-orbitals and the HOMOs and/or LUMOs of the dithiocarbamate ligands in the Cu(I)–Ni(II) coordination polymers are smaller than those in the Cu(II)–dithiocarbamate coordination polymers because of the higher d-levels of the nickel(II) ions than those of copper(II) ions. The direct-current (DC) electrical resistivities of powder-pressed pellet samples of the coordination polymers 1a and 1b, sandwiched by brass electrodes (diameter 13 mm), were measured in the temperature range 250–400 K. The thicknesses of the pellet samples of the coordination polymers 1a and 1b were 0.239 mm and 0.164 mm, respectively, and the applied DC voltage was 100 V. The variations in the DC conductivities of 1a (red) and 1b (blue) at different temperatures are shown in Figure 4. The conductivities of the Cu(I)–Ni(II) coordination polymers 1a and 1b are smaller than those of the Cu(I)–Cu(II) mixed-valence coordination polymers 2a, 2b, and 3 by several orders of magnitude, and do not show the temperature dependences of thermally activated semiconductors below 350 K, that is, these 1D Cu(I)–Ni(II) heterometal coordination polymers were insulators (σ_300K < 10⁻¹² S cm⁻¹) below 350 K, and the observed DC conductivities below 350 K may be not as the bulk samples for the coordination polymers but as the leak current of the samples. The conducting properties of the Cu(I)–Ni(II) coordination polymers 1a and 1b were also investigated using complex impedance spectroscopy—a powerful technique for studying the carrier transport and dielectric properties of bulk samples and electric devices. The impedance measurements were carried out using the same powder-pressed pellet sample. Most of the semiconducting samples show semicircular arcs in the Cole–Cole plots of the impedance and the electric modulus, but the Cole–Cole plots of the impedance and electric modulus for 1a and 1b do not show semicircular arcs because of their quite low conductivities. Figure 5a,b show the temperature and frequency dependencies of the dielectric properties converted from the impedance data. Both samples showed temperature and frequency independent dielectric properties below 350 K, which indicated the insulation properties of the Cu(I)–Ni(II) heterometal coordination polymers.

Research on solar cells containing coordination polymers is less developed than that on mononuclear metal complexes such as N3 dye ([Ru(4,4'-(COOH)₂-byy)₂(NCS)₂]) [30] and black dye ([Ru(4,4',4"-(COOH)₃-terpy)(NCS)₃]) [48], because the process of forming thin films of coordination polymers is still undeveloped. This is because coordination polymers are generally insoluble in organic solvents and retain their infinite structures, and they do not have functional groups such as carboxyl groups to connect with the TiO₂ surface of the photoelectrode. To the best of our knowledge, there has been no example of the application of coordination polymers to solar cell devices, other than our research. In our research, the photoelectrodes of DSSCs were fabricated using TiO₂ paste for low-temperature sintering (PECC-K01, Peccell Technologies Co., Kanagawa, Japan), directly mixed with 1a, 1b, 2a, 2b or 3. The mixtures were prepared by mixing the TiO₂ paste and the coordination polymer using an agate mortar. Figure 6 shows the photocurrent density–voltage (J–V) curves of the DSSCs with the 1D Cu(I)–Ni(II) heterometal coordination polymers [Cu^I₂Ni^IIX₂(Hm-dtc)₂(CH₃CN)₂]_n [X = Br⁻ (1a), I⁻ (1b)], the 1D Cu(I)–Cu(II) mixed-valence coordination polymers [Cu^I₂Cu^IIX₂(Hm-dtc)₂(CH₃CN)₂]_n [X = Br⁻ (2a), I⁻ (2b)], and the 3D Cu(I)–Cu(II) mixed-valence coordination polymer {[Cu^I₄Cu^II₂Br₄(Pyr-dtc)₄]·CHCl₃}_n (3). The details of the short-circuit current densities (J_SC), open-circuit voltages (V_OC), fill factors (FF), and PCEs (η) of cells with different coordination polymer dye materials are summarized in Table 1.

The coordination polymers showing the highest performances in the DSSCs are the 1D Cu(I)–Cu(II) mixed-valence coordination polymers 2a and 2b, which have appropriate HOMO (−5.20 eV for 2a and −5.10 eV for 2b) and LUMO (−3.72 eV for 2a and −3.62 eV for 2b) levels as dye sensitizers, that is, the LUMO levels are higher than that of the conduction-band (E_CB) of the TiO₂ semiconductor (−4.0 eV) and the HOMO levels are lower than that of the redox electrolyte (−4.8 eV). These coordination polymers also exhibit semiconducting properties (σ_340K = 1.07 × 10⁻⁷ S cm⁻¹ for 2a and σ_340K = 2.46 × 10⁻⁷ S cm⁻¹ for 2b) with small activation energies [E_a = 0.562 eV (2a) and E_a = 0.479 eV (2b)]. In contrast, the J_SC values for the DSSCs with the 1D Cu(I)–Ni(II) heterometal coordination polymers 1a and 1b are less than 1/3 of the J_SC values for 1a and 1b, although they also have appropriate HOMO (−5.24 eV for 2a and −5.23 eV for 2b) and LUMO (−3.90 eV for 1a and −3.66 for 1b) levels, as dye sensitizers. This might be the result of low absorptions in the visible region and the low conductivities of the 1D Cu(I)–Ni(II) heterometal coordination polymers 1a and 1b. The DSSC with the 3D Cu(I)–Cu(II) mixed-valence coordination polymer 3 shows lower efficiency than those with the 1D Cu(I)–Cu(II) mixed-valence coordination polymers 2a and 2b, and an efficiency similar to those of the 1D Cu(I)–Ni(II) heterometal coordination polymers 1a and 1b, despite the high conductivity of 3 as a semiconductor (σ_300K = 5.2 × 10⁻⁷ S cm⁻¹, E_a = 0.29 eV) and large absorption in the visible region [13]; this must be because the LUMO levels of 3 (−4.27 eV) are lower than that of the E_CB of TiO₂. The V_OC values for the DSSCs with the 1D Cu(I)–Ni(II) heterometal coordination polymers 1a and 1b are higher than those of the Cu(I)–Cu(II) mixed-valence coordination polymers 2a, 2b, and 3. This difference may be attributed to the more strongly positive shift of the E_CB of TiO₂ in the case of the Cu(I)–Cu(II) mixed-valence coordination polymers 2a, 2b, and 3 [49].