High Ambipolar Mobility in a Neutral Radical Gold Dithiolene Complex

A new anionic gold dithiolene complex NBu4·[1] is synthesized from the (1‐((1,1‐biphenyl)‐4‐yl‐)‐ethylene‐1,2‐dithiolene ligand 1, and the cis and trans isomers are separated by recrystallization. The trans isomer is oxidized via electrocrystallisation to the neutral gold dithiolene complex 2. Complex 2 crystalizes in 1D chains, held together by short (3.30–3.37 Å) S–S contacts, which are packed in a herringbone arrangement in the ab‐plane. The complex exhibits semiconductor behavior (σRT = 1.5 × 10−4 S cm−1) at room temperature with a small activation energy (Ea = 0.11 eV), with greater conductivity along the stacking direction. The charge transport behavior of complex 2 is further investigated in single‐crystal field‐effect transistor (FET) measurements, the first such measurements reported for gold dithiolene complexes. Complex 2 shows incredibly balanced ambipolar behavior in the single‐crystal field‐effect transistor (SC‐FET), with high charge‐carrier mobilities of 0.078 cm2 V−1 s−1, the highest ambipolar mobilities reported for metal dithiolene complexes. This well‐balanced behavior, along with the activated conductivity and band structure calculations, suggests that 2 behaves as a Mott insulator. The magnetic properties are also studied by superconducting quantum interference device (SQUID) magnetometry and solid state 1H NMR, with evidence of a nonmagnetic ground state at low temperature.


Synthesis and Characterization
As the dithiolone precursor is asymmetric, the anionic complex NBu 4 ·[1], synthesized via the deprotection of the dithiolone, 4-(4-phenylphenyl)-1,3-dithiol-2-one, with sodium methoxide followed by addition of KAuCl 4 (Scheme 1), is initially obtained as a mixture of the cis and trans isomers. The isomers can be distinguished by a 0.01 ppm difference between the dithiolene protons in the 1 H NMR, and the isomers do not appear to interconvert ( Figure S1, Supporting Information). While rapid cis-trans isomerism in solution has been reported for asymmetric nickel dithiolene complexes, [53,54] rapid interconversion for asymmetric gold dithiolene complexes has not been obse rved. [4,7,8,12,16,17,33,39,40,42,43,[55][56][57] In most cases, the isomers are separated after the oxidation to the neutral species by selective crystallization of one isomer over the other. Here, after cation exchange with tetra-n-butylammonium iodide, we were able to separate the isomers of anion NBu 4 ·[1] via recrystallization to give the pure trans isomer in reasonable yield (45%), confirmed by 1 H NMR. The neutral complex 2 was then synthesized by galvanostatic electrocrystallisation; crystals grew at the anode on application of a 1.0 µA current to a solution of NBu 4 ·[trans-1] in a 0.1 m solution of NBu 4 ·BF 4 in DMF.
The UV-vis spectra ( Figure S2 and Table S1, Supporting Information) of the anionic complex NBu 4 ·[trans-1] show intense (ε > 10 000 dm 3 mol −1 cm −1 ) bands below 400 nm, attributed to π-π* transitions on the ligand and ligand-to-metal charge transfer (LMCT) transitions. At higher concentrations, weak (ε < 1000 dm 3 mol −1 cm −1 ) bands can be observed at 477 and 727 nm, likely corresponding to d-d transitions. On oxidation a band in the NIR (1462 nm) is observed; this has been observed previously for neutral gold radical complexes and has been assigned to either a SOMO-1 to SOMO transition, or an intervalence charge transfer (IVCT) band of the form [Au III (L • )(L)] → [Au III (L)(L • )]. [19,56,57] Cyclic voltammetry measurements (Figures S4 and S5 and  Table S2, Supporting Information) on complex NBu 4 ·[trans-1] show one quasireversible oxidation and one quasireversible reduction at 0.03 and −1.98 V versus Fc/Fc + , corresponding to the −1 to 0 oxidation and −1 to −2 reduction, respectively. The shape of the oxidation peak varies with scan rate and changes on subsequent scans; this is likely due to deposition of the insoluble neutral complex 2, and a similar behavior has been observed in other gold dithiolene complexes. [16,30,36,39,41] Single crystals of NBu 4 ·[trans-1] and 2 suitable for X-ray analysis were grown via slow diffusion of isopropanol into a solution in acetone and electrocrystallisation, respectively. The anion complex NBu 4 ·[trans-1] crystallizes in the triclinic centrosymmetric space P1 with two distinct anion (and cation) molecules in the asymmetric unit ( Figure S6 and Table S3, Supporting Information). The crystal structure confirms the square planar, trans geometry around the gold center, and key bond lengths (Table S4, Supporting Information) of the dithiolene moiety are in good agreement with those of other monoanionic gold complexes. [19,35,39,56] The asymmetric nature of the ligand is reflected in the difference in CS bond lengths. No short S-S contacts are observed between the anionic complexes within the structure.
In contrast, the structure of neutral complex 2 (Figure 1, Table 1) shows significant S-S interactions. The complex crystallizes in the monoclinic P2 1 /c space group, with one unique complex in the asymmetric unit. The geometry about the gold centers is square planar, with only a small distortion between the two S-Au-S planes (4.65°). The complexes form a 1D regular chain along the a axis, and these 1D chains are packed in a herringbone manner in the ab plane. These herringbone structures are then aligned along the c axis. The short S-S contacts (<3.70 Å) seen within the 1D chains are one of the shortest values among those of reported neutral gold dithiolene complexes.
In complexes with dithiolene ligands an important consideration is the "noninnocence" of the ligand. [58,59] In the case of the anion complex NBu 4 ·[trans-1], the CS and CC bond lengths (Table S4, Supporting Information) are consistent with the ene-1,2-dithiolate form of the dithiolene, suggesting that NBu 4 ·[trans-1] can be described as a Au(III) bis(ene-1,2-dithiolate) complex, as has been proposed for other anionic gold dithiolene complexes. [19,39] On oxidation a slight shortening of the AuS bond, a shortening of the CS bond length, along with a slight lengthening in the CC bond length is observed, indicating the adoption of a more dithioketone-like structure ( Table 2). [21,40] These bond changes suggest extensive delocalization of the spin density over the dithiolene ligands, a conclusion that is supported by our computational work, and literature observations of other gold dithiolene complexes. [4,17,34,40,42,55]

Conductive Properties
Electrical conductivity was measured on single crystals of 2 in the ab plane and along the c axis. The room temperature conductivity at ambient pressure for 2 in the ab plane, measured on multiple crystals, was in the range of 10 −4 -10 −5 S cm −1 , with a maximum value of 1.5 × 10 −4 S cm −1 . A lower conductivity was observed along the c-axis (10 −5 -10 −6 S cm −1 ), consistent with the dimensionality of the crystal structure, an observation that is reflected in the band calculations. Upon cooling, semiconductor behavior is observed (Figure 2) in both the ab plane and along the c axis, with an activation energy of 0.11 eV. This is among the lowest activation energies reported for neutral gold dithiolene complexes that do not contain TTF moieties. [4,12,18] Heat capacity measurements ( Figure S7, Supporting Information) supported the presence of a nonmetallic state at low temperatures. The observed semiconductor behavior, relatively low conductivity and uniform stacking in the solid state, along with the band calculations and FET behavior (vide infra), suggest that 2 can be described as a Mott insulator.

Magnetic Properties
The temperature dependence of the magnetic susceptibility of 2 (Figure 3) was measured with a SQUID magnetometer in the range 2-300 K, and corrected for sample diamagnetism. The susceptibility is almost temperature-independent from 300 to 60 K, then decreases down to 45 K, followed by a rapid increase approaching 0 K which is attributed to paramagnetic defects. The susceptibility (χ p ) of 2 is low, only 4 × 10 −4 emu mol −1 at room temperature, far lower than expected for isolated S = 1/2 molecules. This kind of Pauli-like, small paramagnetic susceptibility with weak temperature dependence is reported in other neutral gold dithiolene complexes with a thermally-activated Adv. Funct. Mater. 2019, 29,1904181    conductivity. [4,5,12,16,17,22] The decrease below 60 K could be ascribed to a transition to a nonmagnetic ground state, [60] a suggestion supported by solid state 1 H NMR data, which show no critical divergence of T 1 −1 and almost temperature independent signal shapes ( Figure S8, Supporting Information), unlike that seen in antiferromagnetic transitions. [15,61] There was also no indication of a spin-flop transition as seen in antiferromagnets observed in the magnetization curves below 60 K. Low temperature X-ray diffraction data ( Figure S9 and Table S5, Supporting Information) showed no clear indication of a structural phase transition. The temperature dependence of the electron paramagnetic resonance (EPR) signals displayed Curie-like behavior ( Figure S10, Supporting Information), consistent with the Curie-tail observed in the susceptibility measurements. Below 100 K, the peak-to-peak line widths and g-factors show an almost temperature-independent behavior, with the values of 19 mT and 2.03, respectively. Above 100 K the intensity of the EPR signals was too low to be detected.

Band Structure Calculations
In order to further understand the conductive properties of 2, extended Hückel theory (EHT) calculations were performed on the obtained crystal structure to determine the band structure, as has been performed previously for gold dithiolenes. [2,13,21,29,42] First, the singly occupied molecular orbital (SOMO), SOMO-1 and lowest unoccupied molecular orbital (LUMO) of the isolated molecule were computed at the geometry present in the crystal structure (Figures S11-S13, Supporting Information), and the SOMO is shown in Figure 4, along with the spin density distribution. The SOMO is heavily delocalized across the dithiolene and the phenyl rings of the ligand, with only a 3% contribution from the Au atom, which could explain the strong intermolecular magnetic interactions in the solid state. [17,34,36] The intermolecular interactions considered in the Hückel calculation are shown in Figure 5. Table 3 lists the calculated overlap integrals and transfer integrals. As interactions along the c axis were negligible, we only considered interaction in the ab plane in the calculation; the interactions within the chain (interaction a) is about approximately an order of magnitude smaller than the interchain interactions (b1 and b2). These interactions are comparable to those resulting from the "brick wall" packing of fluorinated oxobenzene-bridged bis-1,2,3-dithiazolyl (FBBO) radicals; [24] in the case of 2 the interchain interactions (b1, b2) are not equal due to the herringbone packing of the chains. The band structure was calculated by the tight-binding approximation, and is shown in    of the four molecules in the unit cell, and the higher four bands originating from the SOMO orbitals. Bands originating from the LUMO orbitals were at significantly higher energy (−9.04 eV) and thus are not shown here. The band structure shows significant dispersion for the SOMO bands along the a* and b* directions, which are approximately and exactly parallel with the a and b directions, respectively. This is consistent with the interactions within and between the chains in the ab plane. The lack of dispersion along Γ-Z is expected, a result of the omission of interactions between planes along the c axis. Due to the lack of dimerization in the crystal structure the SOMO (and SOMO-1) bands are not split; large splitting of the SOMO bands has been being provided as the explanation for the band semiconductor behavior in previous gold radical complexes. [18] The calculations instead predict a metallic character for 2, at odds with the experimentally observed semiconductor behavior, however extended Hückel theory does not take into account strong electron correlation so this is consistent with our Mott insulator interpretation. [24]

FET Measurements
Single-crystal field-effect transistors (SC-FETs) were fabricated (device fabrication shown in Figure 7) to investigate the possibility of an insulator-metal transition and/or a band-like transport in 2. At present, there are no reports of FET devices of gold dithiolene complexes. The temperature dependence of the transfer characteristics were measured from 300 to 250 K under vacuum at a source-drain voltage V D of 1 V (Figure 8) and −1 V ( Figure S14, Supporting Information), which showed identical behavior except for the direction of the current.
Here, the difference of the source-drain current ΔI D is plotted (ΔI D = (I D at V G ) − (I D at V G = 0 V), V G : gate voltage) for each transfer curve due to the relatively large off-currents, likely caused by the bulk conductivity. [68,69] The large off-current also results in relatively small on/off ratios (see Figure S15b, Supporting Information).    The device shows well-balanced, ambipolar characteristics in the measured temperature range, which is characteristic of FETs based on Mott insulator systems. [68][69][70] The value of the field effect mobility µ at 300 K is estimated to be 0.078 cm 2 V −1 s −1 for both electrons and holes, which is among the highest values for electrons in the FETs based on metal-dithiolene complexes, and the highest reported for holes. [44][45][46]48,51,[71][72][73] Furthermore, the ambipolar behavior is incredibly wellbalanced, a behavior that is still a rare example among metal complexes, including other types of molecular skeletons ( Table 4). [44,45,[74][75][76][77][46][47][48][49][50][51]71,72] The activation energy E a for the thermally-activated behavior of the field effect mobility value is estimated to be 0.065 eV for both carriers. While electric-fieldinduced Mott transitions have been reported in organic crystal FET devices previously, [70] in this case higher doping concentrations would be necessary to induce a metal-insulator transition. It should be noted that this device also works under air without significant change in the performance.
The charge transport properties were also investigated theoretically; the intermolecular hopping carrier mobilities were calculated on the basis of Marcus theory. [78][79][80][81] The calculated parameters and intermolecular carrier mobilities along the directions of the intermolecular interactions are listed in Table 5.
In the calculations the dihedral angles between the dithiolene and benzene moieties and between the benzene moieties were restricted (Tables S6-S7, Supporting Information), as when allowed to freely optimize they became larger than 30°, a significant distortion from the geometry obtained in the crystal structure. The results of the calculations for the unrestricted geometries are listed in Tables S8-S10 in the Supporting Information.
While the theoretical values for the hopping mobilities are greater (Table 5) than those obtained experimentally (0.078 cm 2 V −1 s −1 ), it can be seen that the mobilities and reorganization energies for holes and electrons are comparable, in agreement with the observed ambipolar behavior. This can be explained by the fact that electrons and holes would be injected into the same orbital (SOMO), resulting in similar reorganization energies for both charge carriers, which in turn leads to comparable mobilities. Due to the intrachain interactions ( Figure 5, interaction a) being weaker than the interchain interactions (interactions b1 and b2), the calculated mobilities in this direction are significantly lower.
The calculations also support the experimentally obtained activation energy for the mobility of charge carriers (0.065 eV), as it falls between the calculated values of λ and λ/4 for electrons (0.113 and 0.028 eV, respectively) and holes (0.147 and 0.037 eV, respectively).

Conclusions
We have reported the synthesis of a new neutral gold dithiolene complex, 2, and investigated its electronic and conductive   Numerous phthalocyanine complexes have been studied in OFETs. [62] For the sake of brevity, only those with the highest values, or with reported ambipolar behavior, are included.
properties through single crystal conductivity and FET measurements. The crystal structure of complex 2 shows the formation of chains held together by very short S-S contacts, which are packed in a herringbone arrangement, indicating strong intermolecular interactions which provide a mechanism for conductivity. Complex 2 showed incredibly balanced ambipolar behavior in the SC-FET measurements, the first such measurements reported for gold dithiolene complexes, with high charge carrier mobilities of 0.078 cm 2 V −1 s −1 . This, along with the activated conductivity behavior suggests that 2 behaves as a Mott insulator.

Experimental Section
Materials, Synthesis, and Characterization: All commercially available chemicals were used without further purification. Ligand precursor 4-(4-phenylphenyl)-1,3-dithiol-2-one was prepared as described in the literature. [82] Mass spectra (ESI) were recorded with a Xevo QTOF (Waters) high resolution, accurate mass tandem mass spectrometer equipped with Atmospheric Solids Analysis Probe (ASAP) and Bruker MicroToF 2. Elemental analyses were carried out by Stephen Boyer of the Science Centre, London Metropolitan University using a Carlo Erba CE1108 Elemental Analyzer.
Complex NBu 4 ·[1] NaOMe (1.00 g, 18.51 mmol) was added to a stirred solution of 4-(4-phenylphenyl)-1,3-dithiol-2-one (1.00 g, 3.70 mmol) in methanol/THF (15/5 mL) that had been degassed for 15 min with bubbling nitrogen. The mixture was stirred at RT for 1 h. The reaction mixture was heated to 50 °C and KAuCl 4 (0.70 g, 1.85 mmol) was added. The mixture was stirred at 50 °C for 1 h before the solvent was removed in vacuo. The residue was redissolved in acetone (100 mL) and tetrabutylammonium iodide (0.68 g, 1.84 mmol) was added. The mixture was stirred at RT overnight and the solvent was removed in vacuo. Water (50 mL) was added, and the solution was filtered. The solid was dried, washed with acetone and the filtrate collected. The solvent was removed and the solid dissolved in acetone (150 mL) and heated to reflux. Isopropanol (100 mL) was layered over the hot solution and the solution was left for 2 days. The crystals obtained were collected via filtration and washed with isopropanol to give dark brown crystals of the trans isomer of the gold TBA salt, NBu 4 ·[trans-1] (0.77 g, 45%);  Figure S18, Supporting Information), was consistent with the simulated spectrum for the single-crystal X-ray structure of 2.
Electrochemical Characterization: Cyclic voltammetry experiments were recorded using a Hokuto Denko HZ-5000 Automatic Polarization System. The instrument was fitted with a three-electrode system consisting of a Pt disk as the working electrode, a Pt wire as the auxiliary electrode and an Ag/AgCl in 3 m NaCl solution as the reference electrode. Experiments were conducted at room temperature in dry DMF solution with n-BuNBF 4 (0.1 m) as the supporting electrolyte. Cyclic voltammetry experiments were conducted at a scan rate of 200 mV s −1 , unless otherwise specified. Each solution was purged with N 2 prior to the experiment. Ferrocene (Fc) was used as the internal standard in each measurement.
Optical Characterization: UV-vis solution absorption measurements were recorded using a Jasco V-670 UV/vis/NIR spectrophotometer controlled with SpectraManager software. Solution state UV-visible absorption spectra were obtained using freshly prepared solutions of the complexes in CH 2 Cl 2 at low concentrations (<10 −5 m) in a 1 cm pathlength cell. Baseline correction was achieved by reference to pure solvent in the same cuvette.
XRD Characterization: Powder X-ray diffraction studies were performed using a Rigaku SmartLab using Cu-Kα radiation. For the single crystal data of complex NBu 4 ·[trans-1], a suitable crystal [0.17 × 0.15 × 0.04 mm 3 ] was selected and mounted on a MITIGEN holder in Paratone oil on a Rigaku Oxford Diffraction SuperNova diffractometer (Cu-Kα radiation λ = 1.54184 Å). The crystal of NBu 4 ·[trans-1] was kept at T = 120.0 K during data collection. Four distinctly separate twin domains of the measured crystal were identified by using CrysAlisPro. [83] The structure of NBu 4 ·[trans-1] was solved with the SHELXT [84] structure solution program using the Intrinsic Phasing solution method and by using Olex2 [85] as the graphical interface. The model of NBu 4 ·[trans-1] was refined with version 2016/6 of SHELXL [86] using Least Squares minimization. One of the n-butyl chains was modeled as disordered over two sites, consistent with very large displacement ellipsoids. The disordered n-butyl chain was refined with restraints on the anisotropic displacement parameters (RIGU restrain), and constraints on the anisotropic displacement parameters (EADP constraint) and coordinates (EXYZ constraint). For the single crystal data of complex 2, Adv. Funct. Mater. 2019, 29,1904181  Corresponding molecular contacts are indicated in Figure 5; b) Calculated at the B3LYP/LANL2DZ/6-31G* level.  [87] The structure of 2 was solved by direct method (SIR92 [88] ) and standard difference map techniques, and were refined with full-matrix least-square procedures on F 2 . All calculations for 2 were performed using the crystallographic software package, Crystal Structure, [89] except for refinement. The model of 2 was refined with version 2016/4 of SHELXL [86] using Least Squares minimization. Anisotropic refinement was applied to all nonhydrogen atoms, and all hydrogen atoms were placed at calculated positions and refined using a riding model for both crystals. CCDC 1908634 and 1908635 contain the supplementary crystallographic data for this paper. This data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif. Magnetic Characterization: Magnetic susceptibility measurements were performed on polycrystalline samples from 2 to 300 K using a Quantum Design Magnetic Property Measurement System (MPMS)-XL SQUID magnetometer with MPMS MultiVu Application software to process the data. The magnetic field used was 3 T. The paramagnetic susceptibility χ p was calculated by correction with the diamagnetic contributions from the sample (Pascal's constants: −3.13 × 10 −4 emu mol −1 [90] ) and holder (a plastic straw) and the contribution from the ferromagnetic impurity.
Computational Details: DFT calculations for isolated molecular systems were performed using the Gaussian03 package by using the molecular geometry in the crystal structure of 2. The calculations were carried out using the hybrid B3LYP functional, with the 6-31G* basis sets for H, C, and S, and the LanL2DZ basis set and Los Alamos effective core potentials for Au. The intermolecular transfer integrals and tightbinding band structure calculations were of the extended Hückel type, and made using the Caesar 2.0. [91] The modified Wolfsberg-Helmholtz formula was used to calculate the nondiagonal matrix elements of the Hamiltonian. [92] All valence electrons were taken into account in the calculations, and the basis set consisted of Slater-type orbitals of double-ζ quality for Au 5d and S 3s and 2p, and of single-ζ quality for Au 6s and 6p, C 2s and 2p and H 1s. The ionization potentials, contraction coefficients, and exponents were taken from previous work, implemented in Caesar 2.0.
Intermolecular hopping carrier mobilities were calculated on the basis of Marcus theory. [78][79][80][81] First, intermolecular electronic coupling matrix elements (V ab ) were calculated as follows where H ab = intermolecular charge transfer integrals, S ab = overlap integrals and H aa and H bb = the energies of the two molecular orbitals. These were calculated as described for the band structure calculations. Then the intermolecular charge transfer rate constants (k ET ) were calculated as follows where h, k B , and T are Planck's constant, Boltzmann constant, and temperature, respectively. The reorganization energies upon intermolecular hole (or electron) transfer (λ) were obtained from λ ( ) ( ) where E, E + , E*, and E + * are the energies for an optimised neutral molecule, optimised cation (anion, E -) molecule, neutral state on cation (anion) structure, and cation (anion) state on neutral structure, respectively. These were calculated at the B3LYP/LANL2DZ/6-31G* level using the Gaussian16 package. Intermolecular hopping mobilities (µ) were estimated as follows µ = hopping 2 B ET ed k T k (4) where d is the intermolecular center-to-center distance of adjacent molecules.
As described in the manuscript, the optimized structures were calculated by two methods; the first involved restricting the dihedral angles between the dithiolene and benzene rings and between the benzene rings to fit with geometry observed in the crystal structure. These results are shown in Table 5 and Tables S6-S7 (Supporting Information). The second method allowed full optimisation of the molecular geometries, this resulted in large deviations from the geometry present in the crystal structure, even for the neutral molecule. The results of these calculations are shown in Tables S8-S10 in the Supporting Information.
Single Crystal Conductivity Measurements: The temperature dependence of the conductivity of 2 was measured on single crystals with a two-probe method using a Quantum Design Physical Property Measurement System (PPMS) and an Advantest R6245 2 Channel Voltage Current Source/Monitor. The conductivity was measured in the ab plane and along the c axis of the crystal by applying a constant voltage of 1 V. Gold wires were attached to the crystals with a gold paste.
Single-Crystal Field-Effect Transistors: Single-crystal field-effect transistors were prepared by gluing gold wires (source and drain) to a single crystal of 2 in the ab plane, followed by deposition of Parylene C as a dielectric layer (790 nm, determined by UV-vis spectroscopy). Gold paste was then spread over the Parylene-C layer, and a gold wire glued to the surface to form the gate electrode. The field effect mobility µ is estimated by where L, W, and C i are the channel length (130 µm), channel width (210 µm), and dielectric layer capacitance per unit area (3.6 nF cm −2 ), respectively. An activation energy E a for the thermal-activated behavior of the field effect mobility is calculated by

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.