Syntheses of Assembles Containing POMs and Multinuclear Platinum Complexes

For POMs, Keggin-type [PMo₁₂O₄₀]³⁻(={PMo₁₂}³⁻) and Dawson-type [P₂Mo₁₈O₆₂]⁶⁻(={P₂Mo₁₈}⁶⁻) were selected; these have spherical structures constructed of twelve or eighteen molybdenum atoms containing PO₄³⁻ anions inside the cages. As reported previously^{3, 7} and confirmed in Figures S1–S11, both {PMo₁₂}³⁻ and {P₂Mo₁₈}⁶⁻ can be reduced by multiple electrons at the degenerate Mo dxy orbitals, facilitating easy oxidation of their counterparts. As platinum-based multinuclear complexes, [Pt₂(piam)₂(NH₃)₄]₂ⁿ⁺ (={Pt₄}ⁿ⁺, piam=pivalamidate, n=4–6)⁸ and [Pt₂Pd(piam)₄(NH₃)₄]ⁿ⁺ (={Pt₂Pd}ⁿ⁺, n=2–4),⁹ which have 1D tetranuclear Pt−Pt−Pt−Pt and trinuclear Pt−Pd−Pt alignments, respectively, were selected. Both compounds can be oxidized and reduced by two electrons, {Pt₄}⁴⁺↔{Pt₄}⁵⁺↔{Pt₄}⁶⁺ and {Pt₂Pd}²⁺↔{Pt₂Pd}³⁺↔{Pt₂Pd}⁴⁺, at σ-type orbitals overlapping with each of the other metal dz² filled frontier molecular orbitals.^{6, 9} Compounds {Pt₄}⁵⁺ and {Pt₂Pd}³⁺ have delocalized unpaired electrons over their 1D backbones;^{8a, 9} in particular, {Pt₄}⁵⁺ is well known as “platinum blue”¹⁰ and is also categorized as a Class-II mixed-valent compound.^4a Simply mixing {PMo₁₂}³⁻ or {P₂Mo₁₈}⁶⁻ with {Pt₄}⁴⁺ or {Pt₂Pd}²⁺ in the appropriate solvents and slow evaporation afforded dark blue single crystals of [{PMo₁₂O₄₀}{Pt₂(piam)₂(NH₃)₄}₂]_n ⋅ 6nMe₂CO (1), dark green single crystals of [{PMo₁₂O₄₀}{Pt₂Pd(piam)₄(NH₃)₄}]_n ⋅ 5nMeCN (2), and green single crystals of (NH₄)_n[{P₂Mo₁₈O₆₂}{Pt₂(piam)₂(NH₃)₄}₂]_n ⋅ nH₂O ⋅ 4nMeCN (3).

Crystal Structures of Assembles 1–3

A single-crystal X-ray analysis revealed that compound 1 had a 1D structure (Figure 1a), where {PMo₁₂} and {Pt₄} were alternately aligned with close distances of 3.42 Å between the bridging oxygen and platinum atoms, supported by quadruple hydrogen bonds with donor ligands in {Pt₄}. This close distance was shorter than the oxygen–platinum van der Waals radii, which are approximately 1.5 Ź¹ and 2.1 Ź² for O_e and Pt, respectively. As shown in Figures 1b and 1c, compounds 2 and 3 also formed 1D structures composed of alternate alignments of POM and multinuclear complexes. In 2, the distance between {PMo₁₂} and {Pt₂Pd} was 3.40 Å, similar to that in 1, whereas in 3, shorter 2.84 Å and 3.41 Å distances between bridging oxygen and platinum atoms were observed due to the presence of four crystallographically independent platinum atoms. In the crystals, each 1D chain was packed in a parallel fashion (Figures 1d–f), and solvent molecules were accommodated among the chains. In particular, in 3, channels with diameters of approximately 6 Å contained acetonitrile, water, and ammonium ions. The shortest intercluster O_t−O_t distances were estimated to be 6.1 Å (1), 3.1 Å (2), and 3.2 Å (3). In 1 and 2, the central P atom is surrounded by a cube of eight disordered oxygen atoms, and the Mo atoms are situated at the corners of a regular cuboctahedron due to the crystallographic symmetry.

Metal Oxidation States and Behaviours of Unpaired Electron in Assembles 1–3

Considering the chemical formulae, the overall charges were 0 for {PMo₁₂}–{Pt₄} in 1, 0 for {PMo₁₂}–{Pt₂Pd} in 2, and −1 for {P₂Mo₁₈}–{Pt₄} in 3. The formal metal oxidation states in {Pt₄} and {Pt₂Pd} were assumed by comparing the metal–metal distances because the HOMOs in these complexes are dz² orbitals that sensitively influence the metal–metal bonds.⁶ The intra- and inter-Pt−Pt distances were 1: 2.8153(6) Å and 2.9068(8) Å, and 3: 2.7446(4), 2.8188(4) Å and 2.8443(5) Å, which were similar to those of [Pt₂(piam)₂(NH₃)₄]₂⁵⁺ with a Pt(+2.25) average oxidation state.^8a Additionally, in {Pt₂Pd}, the oxidized form had shorter metal–metal distances, where the Pt−Pd distance (2.6535(3) Å) in 2 was closer to that of [Pt₂Pd(piam)₄(NH₃)₄]³⁺ (2.6601(9) Å) than that of [Pt₂Pd(piam)₄(NH₃)₄]²⁺ (2.8500(5) Å),⁹ resulting in {Pt₂Pd(+2.33)}³⁺. Consequently, the charges of the POMs were {PMo₁₂}⁵⁻ in 1, {PMo₁₂}³⁻ in 2, and {P₂Mo₁₈}⁶⁻ in 3, and the molybdenum average oxidation state in 1 was Mo(+5.83) due to two-electron reduction, although those in 2 and 3 were not changed to Mo(+6) from parent ones. The reduced {PMo₁₂}⁵⁻ in 1 was also confirmed by the alternating short and long bond lengths (ABLs),¹³ where the O_e atoms with off-centre displacements among two Mo atoms led to short and long distances in the Mo−O_e bonds. The differences in ABL distortions tend to decrease with the reduction in POMs. For example, when comparing the short and long Mo−O_e distances in {PMo₁₂}³⁻ and one-electron reduced {PMo₁₂}⁴⁻, the differences of in {PMo₁₂}⁴⁻ is smaller than those in {PMo₁₂}³⁻. As explained in Supporting Information with Tables S3–4 and Figure S15, the average distances for the short and long bonds were 1.868(3) Å and 1.939(3) Å for 1, 1.816(2) Å and 1.988(2) Å for 2, and 1.834(3) Å and 2.041(3) Å for 3; the average distances found in 1 strongly supported {PMo₁₂}⁵⁻.¹³

Infrared (IR) spectroscopy and X-ray photoelectron spectroscopy (XPS) also supported these assignments. It is possible to estimate the metal oxidation states of POM by comparison with IR spectra.¹⁴ The Mo=O_t stretching vibration in the IR spectrum for 1 was observed at 952 cm⁻¹ (Figure 2b), shifting lower from 962 cm⁻¹ in Na₃[PMo₁₂O₄₀] (Figure 2a), indicating the lower oxidation state in Mo atoms.¹⁴ Additionally, the Mo−O_b−Mo (890–850 cm⁻¹) and Mo−O_c−Mo (800–760 cm⁻¹) stretching vibrations, where O_b and O_c were the O_e atoms in the corner-sharing and edge-sharing octahedra, respectively, became broader than the original ones.

The XPS spectra of 1 showed that the Mo 3d_3/2 and 3d_5/2 binding energies were 236.1 and 233.0 eV, respectively, which were characteristic of Mo(+6),¹⁵ accompanied by those for Mo(+5) at 234.9 and 231.7 eV (Figure 2c). Deconvolution of these four peaks resulted in a Mo(+6) to Mo(+5) area ratio of 1.00 : 0.20, leading to an average oxidation state of Mo(+5.83), which was similar to the value estimated by single-crystal X-ray analysis. However, in the Pt 4f_5/2 and 4f_7/2 regions for 1 (Figure 2d), the binding energies for Pt(+3) were 77.9 and 74.6 eV, and those for Pt(+2) were 76.4 and 73.1 eV,¹⁶ with an area ratio of 0.17 : 1.00 for Pt(+3) and Pt(+2), respectively, leading to the average oxidation state of Pt(+2.15). However, in regions 2 and 3, the peaks in the Mo region indicate a single component of Mo(+6) (Figure S18); however, a weak Mo(+5) peak was observed in region 2, and the peaks in the Pt region constitutes the two components for Pt(+3) and Pt(+2) (Figure S19). Based on the area ratio in 2 to 3, Pt(+3) : Pt(+2)=0.55 : 1.00 and 0.28 : 1.00, the average Pt oxidation states were calculated as Pt(+2.35) and Pt(+2.22), respectively. Furthermore, for 2, the Pd 3d_3/2 and 3d_5/2 binding energies were 344.4 and 338.3 eV for Pd(+3) and at 343.0 and 337.5 eV for Pd(+2), respectively, with an area ratio of Pd(+3) : Pt(+2)=0.51 : 1.00 (Figure S20), although the parent [Pt₂Pd(piam)₄(NH₃)₄](PF₆)₃ with {Pt₂Pd(+2.33)} was difficult to be distinguish with XPS.⁹ Consequently, IR and XPS supported the formal oxidation states estimated with the single-crystal X-ray analyses, {PMo(+5.83)₁₂}–{Pt(+2.25)₄} in 1, {PMo(+6)₁₂}–{Pt₂Pd(+2.33)} in 2, and {P₂Mo(+6)₁₈}–{Pt(+2.25)₄} in 3, whose mixed-valence resulted in unpaired electrons in the metals.

Figure 2e shows the magnetic susceptibility data for 1–3 over the temperature range 2–300 K. The χT values at 300 K were 0.38, 0.40, and 0.34 cm³ K mol⁻¹ for 1, 2, and 3, respectively, similar to the spin-only value of 0.375 cm³ K mol⁻¹, indicating that one unpaired electron was present in the {POM}–{Pt complex}. With decreasing temperature, the χT values gradually decreased to 0.25, 0.31, and 0.32 cm³ K mol⁻¹ at 2 K; the data were fitted to the Curie–Weiss law χ=C/(T–θ), yielding 1: C=0.39 cm³ K mol⁻¹ and θ=−9.9 K; 2: C=0.42 cm³ K mol⁻¹ and θ=−12.2 K; and 3: C=0.33 cm³ K mol⁻¹ and θ=−0.4 K (Figures S21–S23). The negative θ values indicated weak antiferromagnetic interactions. Figure 2f–h shows the electron paramagnetic resonance (EPR) spectra of powders 1–3 at room temperature. The spectrum for 1 contained an isotropic signal with g=2.07, whereas those for 2 and 3 contained axial-type signals with hyperfine splitting. Whether the axial-type or isotropic signals are observed in EPR in this case depends on the mobility of the unpaired electrons, as explained below. Five separate signals with g_⊥=2.29, g_//=1.95, A_⊥=345×10⁻⁴ cm⁻¹, and A_//=325×10⁻⁴ cm⁻¹ were found for 2, which is typical for dz² spin delocalized on Pt−Pd−Pt,⁹ where two ¹⁹⁵Pt nuclei (I=1/2 with an abundance of 33.7 %) exhibited splitting without the ¹⁰⁵Pd nuclei (I=5/2 with an abundance of 22 %), supporting the oxidation state {Pt₂Pd(+2.33)}. In contrast, in 3, nine separated signals with g_⊥=2.36, g_//=1.95, A_⊥=220×10⁻⁴ cm⁻¹, and A_//=180×10⁻⁴ cm⁻¹ were typical for dz² spin delocalized on Pt−Pt−Pt−Pt,^8a where four ¹⁹⁵Pt nuclei affect the splitting, supporting the oxidation state {Pt₄(+2.25)}.

Figure 2i shows the variable-temperature EPR spectra (T=4–280 K) for a powdered sample of 1. In addition to a broad isotropic signal without hyperfine structure, axial-type signals with g_⊥=2.48 and g_//=1.91 and sharp isotropic signals with g=1.95 appeared below 100 K (Figure S24). Considering these g values, the axial-type and sharp isotropic signals were attributed to dz² spin delocalized on Pt−Pt−Pt−Pt^8a and Mo dxy spin,^{4b, 17} respectively, and the intensities increased with decreasing temperature. These two signals indicated that 1 contained two unpaired electrons, a Pt dz² spin with {PMo(+5.83)₁₂}–{Pt(+2.25)₄} and a Mo dxy spin with {PMo(+5.92)₁₂}–{Pt(+2)₄}, suggesting that the broad isotropic signal was a coreless signal. Consequently, bistable oxidation states, {PMo(+5.83)₁₂}–{Pt(+2.25)₄}↔{Pmo(+5.92)₁₂}–{Pt(+2)₄}, were observed for 1, and the unpaired electron hopped between {PMo₁₂} and {Pt₄}.

As shown in Figure 2i, the coreless isotropic signals became sharper as the temperature increased from 4 to 40 K, whereas above 40 K, the g values of the coreless signals are similar over the whole temperature range (Figure S25b), and both the Pt dz² and Mo dxy signals completely disappeared above 100 K. The electron hopping frequency monotonically increased with increasing temperature, which was proportional to the peak-to-peak linewidth of the EPR signals.^{4b, 17d} Assuming that the hopping frequency can be expressed by the Arrhenius equation, an activation energy of 0.02 eV was obtained (Figure S25c), which was approximately half that for hopping in {PMo₁₂}⁴⁻ (0.035 eV).^{4b, 17d} In other words, the activation barrier of {PMo(+5.83)₁₂}–{Pt(+2.25)₄}↔{PMo(+5.92)₁₂}–{Pt(+2)₄} was 0.02 eV, which was lower than that for delocalization in {PMo₁₂}⁴⁻, meaning that the unpaired electron in {PMo₁₂} more easily hopped to {Pt₄}. Additionally, for 2, below 100 K, Mo dxy signals appeared in addition to the signal for the dz² spin (Figures S26 and S27), which was caused by partially reduced Mo(+5), as discussed later. However, for 3, only Pt dz² signals were observed over the whole temperature range (Figure S29).

Absorption Spectra and DFT Calculations

Figure 3a shows diffuse reflectance spectra of 1–3 at room temperature. All three compounds exhibited intense and broad absorption peaks in the visible and near-infrared regions, with peak maxima at 705 (1), 1037 (2), and 725 (3) nm. It is well known that mixed-valent compounds exhibit intervalence charge-transfer (IVCT) bands at approximately 600–1200 nm with deep blue or green colours,^4a which were also observed for both the POMs^{1a, 18} and platinum multinuclear complexes.^6b To determine the reason for the absorption band observed in Figure 3a, the molecular orbitals of the {Pt complex}–{POM}–{Pt complex} were determined with density functional theory (DFT) and time-dependent (Td)-DFT calculations. Figure 3b shows the frontier orbitals of 1 calculated with the model [{Pt₂(NHCOCH₃)₂(NH₃)₄}₂{PMo₁₂O₄₀}{Pt₂(NHCOCH₃)₂(NH₃)₄}₂]⁵⁺ with S=1 (Figures S30–31). Both the α- and β-HOMOs were Mo dxy orbitals in {PMo₁₂}, whereas both the α- and β-HOMO–1 and HOMO–2 orbitals were dz² orbitals over Pt−Pt−Pt−Pt. The estimated Mulliken charges were {Pt₄}^4.86+–{PMo₁₂}^4.74−–{Pt₄}^4.88+ (Table S7), which were consistent with the experimental results. Td-DFT calculation revealed the transitions from these dz² orbitals to vacant Mo dxy orbitals serving as α-LUMOs at 666.2 nm (f=0.1389) and to vacant Pt dz² orbitals serving as β-LUMOs at 660.1 nm (f=1.2804), which coincided well with the observed values (Figure S30).

For 2, the Td-DFT calculations based on [{Pt₂Pd(NHCOCH₃)₄(NH₃)₄}{PMo₁₂O₄₀}{Pt₂Pd(NHCOCH₃)₄(NH₃)₄}]³⁺ with S=1 showed an intense absorption at 503.8 nm (f=0.5148) attributed to the IVCT in {Pt₂Pd}, which is quite far from the observed value (Figure S32). Therefore, similar calculations were conducted for one-electron reduced [{Pt₂Pd(NHCOCH₃)₄(NH₃)₄}{PMo₁₂O₄₀}{Pt₂Pd(NHCOCH₃)₄(NH₃)₄}]²⁺ with S=1/2 (Figures S34–35) or two-electron reduced [{Pt₂Pd(NHCOCH₃)₄(NH₃)₄}{PMo₁₂O₄₀}{Pt₂Pd(NHCOCH₃)₄(NH₃)₄}]⁺ with S=1 (Figures S36–37), and the results were relatively consistent with the observed results. As observed for the α- and β-HOMO in the one-electron reduced model and for the α-HOMO–1 in the two-electron reduced model, the added electrons were in the Mo dxy orbitals of {PMo₁₂}, which led to destabilization of the filled orbitals with lower energy absorptions. For example, Td-DFT calculations for a one-electron reduced model show that the absorption at 1093.7 nm (f=0.0051) was attributable to α/β-HOMO→LUMO+15 (Figure S34), and the two-electron reduced model showed that the absorption at 1069.7 nm (f=0.0063) was attributable to α/β-HOMO→LUMO+11 (Figure S36), both of which corresponded to charge transfer from {Pt₂Pd} to {PMo₁₂}. Consequently, partial reduction at {PMo₁₂} could occur, although crystal structure and IR results showed that the formal oxidation state of 2 was {PMo(+6)₁₂}–{Pt₂Pd(+2.33)}. In fact, the XPS data showed small peaks attributed to the reduced Mo(+5) (Figure S18d), and the EPR spectrum obtained at low temperature showed Mo dxy spins (Figure S26), supporting this assignment. However, for 3, DFT calculations were conducted with the [{Pt₂(NHCOCH₃)₂(NH₃)₄}₂{P₂Mo₁₈O₆₂}{Pt₂(NHCOCH₃)₂(NH₃)₄}₂]⁴⁺ model and S=1, and absorptions at 626.7 nm (f=1.4065) and 620.3 nm (f=0.2570) were attributed to IVCT in {Pt₄} (Figure S39). Since there was a slight difference between the calculated and observed values, calculation of two-electron injection model with S=1 were also conducted (Figure S41), revealing that the absorption at 708.3 nm (f=0.1942) was attributed to IVCT in {P₂Mo₁₈}; however, 3 was not involved in the reduction based on the EPR results.

Electrical Conductivity

Figure 4 shows the results of DC electrical conductivity measurements. The single crystal was too small to be terminalized properly, so the possibility of using pellets was investigated (Figures S45). The current–voltage (I–V) characteristics were linear, and the conductivities were evaluated from the slopes of the lines as 1.0×10⁻⁸ (1), 7.0×10⁻⁸ (2), and 3.0×10⁻⁷ (3) Scm⁻¹. The conductivities at room temperature increased in the order 1<2<3. The conductivity of Dawson-type [Pr₄N]₆[S₂Mo^V₂Mo^VI₁₆O₆₀] ⋅ solv with mixed valence was 1.6×10⁻¹² S cm⁻¹,¹¹ meaning that for 3 was approximately 10⁶ times greater than that of the parent compound. As shown in Figure 4, the resistivities for 1–3 increased exponentially with decreasing temperature, indicating thermally activated semiconductor behaviour. The activation energies were estimated from the slopes of the Arrhenius plots as 1: 0.60 eV at 270–297 K, 2: 0.48 eV at 239–298 K, and 3: 1.70 eV at 273–300 K (Figures S46–S48). The temperature dependence of conductivity was also successfully measured using single crystal 3, and the activation energy was found to be 0.35 eV at 238–286 K (Figure S49). Since the maximum absorption energies found in the diffuse reflectance spectra were 1.76 (1), 1.20 (2), and 1.71 (3) eV, indicating approximately 1 eV optical band gaps for 1–3, the estimated activation energies were valid. The greater activation energy of 3 compared with those of 1 and 2 was attributed to the structural geometry of {P₂Mo₁₈}. Unlike the Keggin-type {PMo₁₂}, where each metal occupies an equivalent site, the Dawson-type {P₂Mo₁₈} contained two types of Mo atoms, six Mo atoms at the polar edges and twelve Mo atoms at other edges.

To date, two mechanism for POM conductivity have been reported: proton conductivity¹⁹ and framework conductivity.²⁰ Both are realized in channel-surrounded POMs, where the latter conductivity is related to organic conductors such as tetrathiafulvalene (TTF) accommodated in the pores.²⁰ Table 1 summarizes the conductivity and activation energies of related previous compounds, indicating that compound 1–3 conducts electricity relatively well compared to these compounds. To our knowledge, there is few examples of an intrinsically conducting framework, where (TTFPyH)₂[PMo^VMo^VI₁₁O₄₀] with strong electron-donor TTF derivatives shows a conductivity of 6.3×10⁻⁷ Scm⁻¹ at room temperature.²¹ The conductivities of mixed-valent assemblies 1–3 were similar to this pioneered compound, and the intrinsic conductivity of the POMs was enhanced by the electronic interaction between the O_e p orbital and Pt dz² orbital.