Preparation and Structure of the Ion-Conducting Mixed Molecular Glass Ga2I3.17

Modern functional glasses have been prepared from a wide range of precursors, combining the benefits of their isotropic disordered structures with the innate functional behavior of their atomic or molecular building blocks. The enhanced ionic conductivity of glasses compared to their crystalline counterparts has attracted considerable interest for their use in solid-state batteries. In this study, we have prepared the mixed molecular glass Ga2I3.17 and investigated the correlations between the local structure, thermal properties, and ionic conductivity. The novel glass displays a glass transition at 60 °C, and its molecular make-up consists of GaI4– tetrahedra, Ga2I62– heteroethane ions, and Ga+ cations. Neutron diffraction was employed to characterize the local structure and coordination geometries within the glass. Raman spectroscopy revealed a strongly localized nonmolecular mode in glassy Ga2I3.17, coinciding with the observation of two relaxation mechanisms below Tg in the AC admittance spectra.

The obtained glass is, like the crystalline gallium iodides, highly moisture sensitive (color change to redbrown or dark gray ) and was handled inside a glove box with nitrogen atmosphere (p(O 2 /H 2 O) < 0.5 ppm). In the bulk glass sample prepared for neutron diffraction measurements, several sub-millimeter sized particles of crystallized material were observed upon optical inspection. These were detected in the neutron scattering data as small Bragg peaks imposed on the signal of the amorphous glass around Q = 0.98 Å -1 and Q = 1.89 Å -1 , and identified as the crystalline compound Ga 2 I 3 .
Differential scanning calorimetry data were collected on a PerkinElmer DSC 8000 using closed stainless steel DSC pans at a rate of 10 K min -1 . Samples of g-Ga 2 I 3.17 heated only up to 85 °C retained the optically transparent appearance of the glassy state, while samples heated up to 140°C transformed into an opaque yellow crystalline solid. T g and ΔC p were determined from heating curves, by a geometric tangent construction to the inflection point during the transition and the base line below and above the transition. Raw data were baseline corrected by subtraction of a straight line.
Raman spectroscopy was performed on a Renishaw Ramascope equipped with a 633 nm laser and a selfbuilt microscope heating stage (50x magnification). The spectrometer was calibrated using a Ne light source. Spectra were recorded in 5-15 °C steps between 20 °C and 410 °C. The Raman shift was measured from 60 cm -1 up to 800 cm -1 and no modes were observed above 300 cm -1 .
Infrared spectroscopy was performed at room temperature under Argon inert atmosphere on a Bruker Invenio-R FTIR spectrometer in attenuated total reflection mode using a diamond ATR crystal.
Electrical resistivity and admittance data were measured on samples sealed in evacuated glass capillaries (ID = 2 mm, L = 1 cm) which had been fitted with Pt wire electrodes at the ends. DC resistivity and admittance data were measured in two-electrode configuration using a UNI-T 61C ohmmeter and an Agilent HP 4294A precision impedance analyzer. Samples were heated in a micro-tube furnace.
Density functional theory calculations were performed using the CASTEP code version 18.1 using a 4x3x3 k-point grid. 2 The general gradient approximation employing the parametrization of Perdew-Burke-Ernzerhof was chosen for treatment of the exchange correlation potential. 3 After geometry optimization of the reported atomic coordinates for crystalline Ga 2 I 3 , energies of vibrational modes as well as Raman activities were calculated.
Time-of-flight neutron diffraction data in the range Q = 0.1 -60 Å -1 were collected on the general materials diffractometer (GEM) at the ISIS spallation neutron source (Rutherford-Appleton Laboratory, UK). The sample was sealed in a quartz glass ampoule (ID: 8 mm, OD: 10 mm, sample height: 50 mm) and measured at 30 °C, 400°C and 310°C in this order, for a minimum 200 µA of proton current each (beam cross section at the sample: 15 x 40 mm). Additionally, data for a V 0.9486 Nb 0.0514 rod, the empty furnace and an empty silica ampoule were recorded for the purpose of data normalization and correction. The glass ampoules were surrounded by a vanadium foil (0.1 mm thickness) to hold them in place and the resistive heater consisted of a vanadium foil cylinder surrounding the sample. The GudrunN software (version 2) 4 was used to correct the recorded time-of-flight neutron diffraction data for sample container S-3 and background contributions, multiple scattering, absorption and inelasticity effects (Placzek's formula), normalized to and all detector banks merged after elimination of faulty detectors. The low-Q range 〈 〉 2 of the diffraction data showed minor Bragg contributions from trace amounts of the crystalline Ga 2 I 3 phase, which were subtracted from the data. The obtained reduced total scattering function F(Q) was convoluted with a modified Lorch function for Fourier transform to minimize the termination ripples caused by the finite Q-range (Q max = 60 Å -1 ) recorded. Comparison of the obtained G(r)-1 for a convolution with the step function or a modified Lorch function with width L(r) = Δ 0 [1+r β ] ( Figure S5) shows satisfactory suppression of the ripples in G(r)-1, while the concomitant loss of resolution does not appear to conceal additional peaks for Δ 0 = 0.15 and β = 0.4. Absence of any feature in G(r)-1 centered at r ≈ 1.6 Å (corresponding to the Si-O bond distance) indicates adequate correction of the silica ampoule contribution. The pair distribution function G(r) for the crystalline phase c-Ga 2 I 3 was simulated using the RMCProfile software suite. 5

Diffraction notation:
Following the derivation and formalism outlined in literature, [6][7][8] the relationship between the scattered intensity I(Q), the structure factor S(Q), the pair distribution function G(r) and the respective partial quantities g ij (r) and coordination numbers, can be derived as follows: The amplitude Ψ s of a wave scattered by a collection of N atoms as function of the scattering vector Q is given by the sum over all atoms i in the sample where is the average neutron scattering length of the atom located at r i . The intensity of the scattered wave is then given as The spherically averaged intensity for an isotropic sample is given by the Debye scattering equation, where r ij = r i -r j . (3) • Now, one can define the total scattering structure factor S(Q) for an isotropic sample: For a system with M atomic species, the Faber-Ziman average structure factor S(Q) can be defined as the sum over the individual contributions S ij : Here and is the mean scattering length and is the molar fraction of species i.
Fourier transform of gives the reduced pair distribution function G(r): Here, ρ 0 is the numerical density (atoms per Å 3 ) and G(r) is the total pair distribution function which can be decomposed into the individual partial pair distributions g ij (r) The radial distribution function T(r) and the weighted average coordination number CN(r 1 ,r 2 ) within the distance range r 1 to r 2 around an atom at the origin are given by around an atom i can be expressed as (10) ( 1 , 2 ) = 4 0 ∫ 2 1 2 ( ) and the weighted sum over the partial coordination numbers for a system of two atomic species is To completely determine the partial pair distribution functions and partial coordination numbers However, chemical knowledge of the system enables certain assumptions, which allow the estimation of partial pair distribution functions within a limited distance range, despite an incomplete dataset.
Assuming that only one g ij (r) contributes to a certain interval [r 1 ,r 2 ], we can set all other g ij ([r 1 ,r 2 ]) in equation (8) for this interval to zero, and by rearranging (8) for one g ij (r) in the respective distance interval we obtain : The weighting factors for the individual contributions below r = 5 Å were calculated using = 7.288 (2) fm, = 5.28 (2)  By combining (10) and (12), we obtain the general result for the partial coordination number for atoms of type j around atoms of type i, within the distance range [r 1 , r 2 ] which is strictly valid under the assumption that no other partial distribution contributes to G(r) in the interval [r 1 , r 2 ]: