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.

Introduction

Glasses belong to the earliest materials utilized and produced by humanity and have been rediscovered as modern materials based on diverse novel glass-forming precursors and the concomitant functional properties.[1] Recent examples are flexible semiconducting oxide glasses,[2] multinary chalcogenide glasses for infrared optics and data-storage media,[3] bulk metallic glasses with superior mechanical and magnetic properties,[4] and metal–organic framework glasses.[5,6] Ion-conducting glasses are promising electrolytes for the next-generation of all-solid-state batteries, as they display isotropic behavior, absence of grain boundaries, and often higher ionic conductivity than their ordered crystalline counterparts.[7] In addition to amorphous organic polymers, mixed phosphate, oxide, sulfide, and halide glasses show high ionic conductivities at ambient temperature, combined with enhanced chemical stability and glass-forming ability.[8−10]

Molecular glasses are typically comprised of small organic molecules[11,12] and have been investigated for applications in photolithography, organic electronics,[13−15] and for amorphous pharmaceuticals where their improved dissolution behavior can be exploited.[16,17] The small molecular mass and well-defined composition make molecular glasses attractive for the production of printed organic circuits.[18] However, low glass-transition temperatures Tg and strong tendencies for crystallization remain the major challenges in the design of molecular glasses and hinder long-term applications, in particular, at elevated temperatures.[19]

The general design guidelines for molecular glasses include nonplanar molecular structures, increased molecular size, and bulky substituents,[13] as observed in a series of molecular glasses, based on triphenylamine derivatives, tris(oligoarylenyl)amines or tris(oligoarylenyl)boranes.[20,21] Tuning the glass-transition temperature, long-term stability, and optical and transport properties has been achieved by side-group substitution and variation in molecular size but still relies often on trial-and-error.[11,22] More recent avenues to molecular glass design include the use of atomistic simulations or machine-learning-based algorithms to predict the properties and compositions of possible glass-forming liquids.[12,23,24] For organic electronics, interest has shifted from one-component glasses to mixed molecular glasses, containing two or more molecular species.[25] For applications such as organic LEDs, the combination of light-emitting species with conductive molecules can lead to increased emission efficiency and provides more design flexibility.[26]

Contrary to the glass-forming organic liquids, inorganic oxide, chalcogenide, or halide glasses typically feature a network structure, characterized by the formation of strong directional bonds during the transition from the liquid to the glassy state.[27] The presence of individual molecular units has been reported in a few chalcogenide glasses, notably PS43– and P2S64– in ion-conducting glasses,[10,28,29] P4Se (n = 3–4) in nonstoichiometric P–Se glasses,[30,31] As4S4 molecules in As–S glasses,[32] as well as the analogous multinary P–As–S–Se species.[33] Of special interest is the typically elevated glass-transition temperature (Tg > 100 °C) in inorganic molecular glasses compared to organic glass-forming liquids.[11,25,33]

For the gallium–iodine system, the crystalline binary compounds Ga2I6, Ga2I4, and Ga2I3 have been reported (cf. Figure A).[34−36] The latter two are crystalline salts comprised of Ga+ cations and GaI4– or Ga2I62– anions, respectively, which are isostructural and isoelectronic to the above-mentioned PS43– and P2S64– anions. The initially reported monoiodide GaI appears to correspond to the formula Ga2I3.[35,37] Besides the high-temperature reaction of the elements, the higher iodides can be prepared by sonicating liquid gallium metal in a solution of iodine in benzene.[38] For a 1:1 ratio of Ga/I, this yields a gray-greenish powder, which appears to be a mixture of the above phases and has found use as a gallium source in organic synthesis.[39,40]

Figure 1

(A) Partial phase diagram of the gallium–iodine system adapted from refs (37) and (42). Molecular entities in the crystalline compounds are sketched (Ga: purple spheres, I: green spheres), and the glass composition is indicated as a dashed orange line. (B) DSC on glassy Ga2I3.17. Heating/cooling curves of the air-quenched glass exhibiting a glass transition at Tg = 60 °C and onset of crystallization at Tc ≈ 90 °C. Inset: photographs of a Ga2I3.17 sample in a glassy (g) or crystallized (c) state at room temperature and in a liquid (l) state at 400 °C. (C) Curves (1,2): heating and subsequent cooling of an air-quenched sample; (3,4): subsequent second heating/cooling cycle; (5,6): heating/cooling curves after sub-Tg annealing (50 °C, 1 h). All curves were recorded at a rate of 10 K min–1.

Two early reports noted a possible glass formation in the Ga–I system without providing further details.[37,41] Our study is the first investigation of the gallium halide glass with the composition Ga2I3.17, describing the local structure and chemical properties of the liquid and glassy states in detail. The reported structures of the crystalline phases, c-Ga2I3 and c-Ga2I4, were taken as starting points for the analysis of glassy and liquid Ga2I3.17. The crystalline compounds contain gallium in nominal oxidation states Ga(I), Ga(II), and Ga(III). In c-Ga2I3 and c-Ga2I4, Ga(I) cations coordinate with Ga(II)2I62– heteroethane ions (idealized point symmetry D3) and tetrahedral Ga(III)I4– anions, respectively (cf. Figure A). The reported covalent bond lengths in the molecular anions are 2.39 Å for the Ga(II)–Ga(II) single bond and 2.54–2.61 Å for Ga(I)–I bonds.

Experimental Methods

Gallium iodide samples were prepared in a highly exothermic reaction by carefully heating stoichiometric amounts of gallium metal (Aldrich, 99.99%) and iodine powder (VWR Rectapur GPR, ≥99%) up to 300 °C, forming a dark red melt. Samples of the gallium iodide glass were prepared by air quenching a melt of composition Ga2I3.17 (i.e. Ga38.7I61.3), from 400 °C to room temperature. This composition corresponds to the solubility limit of Ga in the melt at 400 °C. The composition of the glass was determined as Ga2I3.17 by back-weighing the residual Ga metal from samples with Ga metal excess. Differential scanning calorimetry (DSC) data were collected on a PerkinElmer DSC 8000 system at a rate of 10 K min–1. Raman and Fourier transform infrared (FTIR) spectra were recorded under an inert atmosphere on a Renishaw Ramascope (HeNe laser, 633 nm) and a Bruker INVENIO-R spectrometer, respectively. Time-of-flight neutron diffraction data were recorded on the GEM instrument (RAL-ISIS, UK) in the range Q = 0.1–60 Å–1. The DC conductivity and AC admittance data were recorded in a two-probe mode using a UNI-T 61C ohmmeter and an Agilent HP 4294A precision impedance analyzer, respectively. Further experimental details and data evaluation are provided in the Supporting Information.

Results and Discussion

During the preparation of glassy g-Ga2I3.17, a remarkable color change was observed from the yellow crystalline solid to a dark red liquid, which then gives an optically transparent orange glass g-Ga2I3.17 upon quenching (inset to Figure B). The glassy nature of g-Ga2I3.17 was confirmed by DSC of glassy Ga2I3.17. Heating of the air-quenched glass reveals a glass transition at Tg = 60.0 °C with a kinetic overshoot followed by a strong exothermic crystallization of the glass with onset at Tc ≈ 90 °C at 10 K min–1 (Figure B). The glass transition displays a step in specific heat ΔCp = 83.5 J K–1 molGa2I3.17–1 (curve 1 in Figure C). The glass-transition temperature was found as Tg = 57.7 °C in the second heating cycle after cooling the sample at a rate of 10 K min–1 (curve 3 in Figure C). The strong kinetic overshoot observed in the air-quenched material indicates a more relaxed sample which was reduced in the second heating cycle, suggesting that the cooling rate of the air-quenched material was significantly lower due to the large thermal mass. Sub-Tg annealing of the sample for 1 h at 50 °C led to the recovery of the kinetic overshoot and a glass transition with Tg = 60.0 °C and ΔCp = 83.8 J K–1 molGa2I3.17–1 (curve 5 in Figure C).

The temperature-dependent Raman spectra of g-Ga2I3.17 were recorded to determine the molecular make-up of the glass and follow the structural changes upon heating from room temperature to 410 °C, covering the glassy, supercooled liquid, crystalline, and liquid states (Figure ). The reported crystal structures and Raman spectra of Ga2I3, Ga2I4, and a series of A2[Ga2X6] compounds, together with the calculated Raman spectrum of c-Ga2I3, serve as a starting point for assigning the Raman lines to the molecular modes (cf. Table S1 in the Supporting Information).[37,43,44] The composition of the melt and the resulting glass, Ga2I3.17, lies between that of c-Ga2I3 and c-Ga2I4, suggesting a mixture of Ga+ ions with both Ga2I62– and GaI4– molecular ions in the glass.

Figure 2

Raman spectra of glassy Ga2I3.17 recorded upon heating from room temperature to 410 °C. (A) Raman spectra at selected temperatures. Curves shifted vertically for clarity. (B) Contour plot of temperature-dependent measurements. The temperatures of glass transition Tg, crystallization Tc, and complete melting Tm are indicated by dash-dotted lines. Intensity is indicated by the color scale and contour lines. Spectra are normalized to the highest intensity.

The Raman spectrum of g-Ga2I3.17 at 20 °C displays four distinct peaks centered at 106, 118, 142, and 283 cm–1 (Figure A). Modes below 80 cm–1 can be assigned to I–Ga–I and I–Ga–Ga bending modes of GaI4– and Ga2I62– molecules, respectively. Upon close inspection, two very weak broadened features are found at 184 and 244 cm–1 due to the Ga–I asymmetric stretch vibrations of Ga2I62–.[43,44] The peaks at 118 and 283 cm–1 can be identified as the in-phase (A1g) and out-of-phase (A1g) stretch vibrations of the Ga–Ga bond within the Ga2I62– ions, as observed in A2[Ga2I6] (A = H+, Me4N+) compounds and the calculated Raman spectrum for c-Ga2I3 (Table S1). The peak at 142 cm–1 can be assigned to the tetrahedral breathing mode (A1) of GaI4– molecules, the band of the highest intensity in Ga2I4 (cf. Figure S1 in the Supporting Information).[37] The presence of GaI4– molecules is consistent with the determined sample composition of Ga2I3.17, which corresponds to an approximate ratio of 0.4 GaI4– molecules per Ga2I62– molecule. Following this analysis, the overall composition of g-Ga2I3.17 can also be represented as [Ga+]63.2[Ga2I62–]26.3[GaI4–]10.5, reflecting the molecular makeup.

At 106 cm–1, a strong mode is observed in glassy Ga2I3.17 which cannot be attributed to any molecular mode of Ga2I62– or GaI4– units and therefore requires further examination. The peak is of comparable intensity and width to the identified molecular modes, suggesting a localized nature (20 °C curve in Figure A). The mode was not observed in the infrared spectra of g-Ga2I3.17 (Figure S2) and hence is only Raman-active.

Two phenomena are observed for (i) the molecular modes of Ga2I62– and GaI4– and (ii) the unidentified mode at 106 cm–1 upon heating:

(i) The Raman shift and peak width for the Ga–Ga stretch modes at 118 and 283 cm–1 and the GaI4– tetrahedral symmetric breathing mode at 142 cm–1 show a remarkable temperature dependence (cf. Figure S3). All three modes display an overall red shift with increasing temperature, as expected for the thermal expansion behavior of an anharmonic oscillator. Across the glass transition around 60 °C, both the Ga–Ga stretch modes of the Ga2I62– unit experience a rapid but continuous blue shift which stops once the sample crystallizes at around 80 °C.

Contrarily, the GaI4– breathing mode at 142 cm–1 displays a red shift, simultaneous with the above-mentioned blue shift. The opposite behavior in Raman shift agrees well with the separation of Ga2I62– and GaI4– molecules due to the crystallization of g-Ga2I3.17 forming the phases c-Ga2I3 and c-Ga2I4. During melting of the two phases, which begins for c-Ga2I4 at 170 °C and for c-Ga2I3 around 230 °C (cf. Figures and S3), the change in Raman shift is reversed for all three modes.

Extrapolation of the peak positions from the glassy state to higher temperatures coincides well with the observed peak positions in the liquid state, suggesting that the interactions in glass and liquid are of similar nature. The changing line widths of all three modes reflect the narrowing and broadening distributions of the coordination environments for Ga2I62– and GaI4– molecules upon crystallization and melting, respectively.

(ii) Approaching Tg, the peak at 106 cm–1 drastically loses intensity and is completely absent once the sample crystallizes at Tc (100 °C curve in Figure A). Upon melting, the peak reappears as a broad shoulder and gains intensity as the temperature increases (curves for 245 and 410 °C in Figure A). The mode is only present in the disordered glassy and liquid states, which contain both Ga2I62– and GaI4– molecules but not after crystallization when the two types of molecules are separated in two phases. Heating of g-Ga2I3.17 from room temperature across Tg and subsequent cooling, while avoiding crystallization, reveal the reversibility of the intensity loss across Tg. The intensity of the 106 cm–1 peak drops steeply above 60 °C, relative to the highest intensity Ga–Ga stretch mode at 118 cm–1, but regains nearly full initial intensity upon cooling (cf. Figure S4).

The amorphous nature of the material was further corroborated by time-of-flight neutron measurements recorded on glassy (30 °C) and liquid (310 and 400 °C) Ga2I3.17 (GEM diffractometer, RAL-ISIS, UK).[45] The total structure factor S(Q) for glassy Ga2I3.17, presented as a function of the scattering vector Q in Figure A, shows broadened features characteristic of amorphous materials and is dominated by three main peaks in the low-Q region (Table S2). A small but well-defined first sharp diffraction peak (FSDP) is observed at Q1 = 0.96 Å–1. The FSDP can be related to intermediate-range order (IRO) on the length scale of 2π/Q1 ≈ 6.5 Å. In c-Ga2I3, no intramolecular correlations exist beyond this length scale, marking the transition to purely intermolecular pair correlations.[60,61] Comparing S(Q) of g-Ga2I3.17 with the liquid state, an overall broadening of the features is observed at higher temperatures. Interestingly, the positions of the first and second peaks barely change upon melting, and the FSDP is sharpened, suggesting enhanced IRO. The oscillations of S(Q) at high Q are dampened more strongly at high temperatures, as expected, due to thermal broadening.

Figure 3

Neutron diffraction data for Ga2I3.17 in the glassy (30 °C) and liquid (310, 400 °C) states. Data are vertically shifted for visibility. (A) Total structure factors S(Q) after the correction of raw data. (B) Pair distribution functions G(r) obtained by Fourier transformation of the diffraction data. (C) Gaussian least-squares fit to the function r2G(r) for glassy Ga2I3.17. Peaks are labeled with atom pair assignments.

The total pair distribution function G(r) for glassy and liquid Ga2I3.17 (Figure B) was obtained by Fourier transformation of the total structure factor. The close correspondence of G(r) for g-Ga2I3.17, l-Ga2I3.17, and c-Ga2I3 (cf. Figure S6) corroborates the suspected presence of Ga2I62– molecules in glassy and liquid Ga2I3.17. Below r = 2 Å, G(r) of g-Ga2I3.17 shows small irregular oscillations around zero, attributed to Fourier truncation artifacts, and no contributions to G(r) are expected in this range. A comparison with interatomic distances observed in crystalline gallium iodides and the simulated total pair distribution function for crystalline Ga2I3 (Figure S6) allows the assignment of the peaks in G(r) to interatomic distances between atom pairs up to r = 5 Å (Figure B,C).[37,46]

The first peak centered around 2.56 Å contains overlapping contributions of the covalently bound Ga(II)–Ga(II) (∼2.39 Å) pair in Ga2I62– and Ga–I pairs within Ga2I62– and GaI4– molecules (∼2.5–2.6 Å). The contribution of the Ga–Ga pairs to G(r) is quite small due to the high relative abundance of Ga–I bonds and the smaller neutron scattering length of gallium. The third peak in G(r) shows a maximum at r = 4.14 Å in good agreement with the longer nonbonded Ga(II)–iodine distances within Ga2I62– (∼4.1 Å) and intramolecular iodine–iodine (4.2–4.3 Å, geminal) distances within the Ga2I62– and GaI4– units. The asymmetry at high r arises from intramolecular I–I (vicinal) and also intermolecular I–I distances between neighboring molecules, contributing between r = 4.1 and 4.5 Å in the crystalline compounds.[37] The second peak centered at 3.30 Å can then be assigned to the distances between iodine in the Ga2I62–/GaI4– units and the surrounding Ga(I) ions, as observed in c-Ga2I3 and c-Ga2I4 (dGa(I)–I ≈ 3.3–3.7 Å). This distribution is significantly broader than the intramolecular contributions as a result of the relaxed bonding constraints. The Ga(I)–I distribution is also significantly broadened compared to the simulated G(r) of c-Ga2I3 (Figure S6).

For the liquid l-Ga2I3.17, at 310 and 400 °C, G(r) shows a similar overall shape as for g-Ga2I3.17. Upon melting, the first and third peaks in G(r) are slightly broadened, which is enhanced at 400 °C. Overall, the magnitude of the oscillations in G(r) at larger distances decays faster with increasing temperature. The strongest change is observed for the second peak, around 3.3 Å. While this peak is well defined in the glassy solid, it is significantly broadened in the melt at 310 °C and turns into a nearly flat contribution to G(r) at 400 °C. This observation can be well reconciled with the above peak assignment, distinguishing intramolecular Ga(II)–Ga(II) (2.39 Å), Ga(II/III)–I (2.59 Å), and I–I (4.1 Å) distances from the intermolecular Ga(I)–I distances. While the intramolecular distances show only moderate peak broadening upon melting, the distribution of the noncovalently bound Ga(I)–I pairs around r = 3.3 Å is strongly affected, suggesting enhanced mobility and disorder of the Ga(I) ions.

The partial coordination numbers CNGa(II)Ga(II), CNGa(II,III)I, CNGa(I)I, and CNII for the peaks up to r = 5 Å were determined following the formalism derived in the Supporting Information. The fit results from Gaussian contributions to r2G(r) (Figures C and S7) were weighted by the corresponding stoichiometric coefficients and neutron scattering lengths to obtain an estimate for the partial coordination numbers (Tables , S3, and S4).

Table 1

Results of the Least-Squares Fit of Gaussian Contributions to the Functionr2G(r) and Coordination Numbers CN jof Atomsj Around i in g-Ga2I3.17 at 30 °C in the Ranger = 2–5 Å

atom pair i–j	r/Å	fwhm/Å	area/b Å–2	CNij
Ga(II)–Ga(II)	2.388a	0.19(1)	0.47(5)	0.65(6)
Ga(II/III)–I	2.597a	0.284(4)	4.42(7)	3.49(6)
Ga(I)–I	3.336(8)	0.53(1)	5.9(4)	4.7(3)
I–I	4.157(2)	0.56(2)	15.6(17)	10.7(1.1)

Peak position fixed to reported bond lengths in c-Ga2I3. Estimated standard deviations from the least-squares fit are given in brackets.

Coordination numbers for gallium in Ga(II)2I62– or Ga(III)I4– units by Ga(II) and iodine were obtained as CNGa(II)Ga(II) = 0.65(6) and CNGa(II/III)I = 3.49(6), respectively. The individual and summed coordination numbers agree closely with the expected values for tetrahedrally coordinated gallium in a mixture of both Ga2I62– (ideal: CNGa(II)Ga(II) = 1 and CNGa(II)I = 3) and GaI4– (ideal: CNGa(III)I = 4) molecular anions, considering the strong overlap of both peaks. The average coordination number of the individual Ga(I) cations (i.e., Ga+) by iodine CNGa(I)I = 4.7(3) is significantly reduced in glassy Ga2I3.17 compared to crystalline Ga2I3, where eight iodine atoms form the closest coordination shell. Upon melting, the distribution broadens, and CNGa(I)I decreases to 4.2(1) at 400 °C (Tables S3 and S4). At a higher temperature and above r = 4 Å, the coordination numbers are less robust due to the increased overlap of partial contributions and the simplistic approximation as symmetric Gaussian contributions.

The effect of structural changes on the transport properties of Ga2I3.17 was observed in the temperature-dependent measurements of electrical dc conductivity and ac admittance spectra. The DC conductivity σ(T) of g-Ga2I3.17 was measured as a function of temperature while the sample was heated from room temperature through the glass transition and subsequent crystallization. The dc conductivity changes by several orders of magnitude as it passes through three distinct stages, delimited by the glass-transition temperature Tg and the crystallization temperature Tc (Figure A). Below Tg, g-Ga2I3.17 displays a linear trend of ln(σ) versus 1/T, suggesting a thermally activated process for the mobility of Ga(I) ions. The behavior is well described by the Arrhenius law σ(T) = σ0e–, with the fit parameters σ0 = 156.6 S cm–1 and the activation energy Ea = 0.59 eV.

Figure 4

(A) Electrical dc conductivity measured upon heating of a g-Ga2I3.17 sample. Fit functions and parameters for three regions (T < Tg, Tg < T < Tc, Tc < T). (B) Real part YRe(ω) and (C) imaginary part −Yim(ω) of the ac admittance for glassy Ga2I3.17. Inset to (B): Extrapolated low-frequency limit YRe(ω = 0). Arrows in (C) are guide to the eye, highlighting the approximate temperature dependence of the characteristic frequencies for the three suggested relaxation processes.

Above the glass transition, which has a lower onset than in DSC due to the lower heating rate, ln[σ(T)] deviates strongly from the linear behavior, that is, from exponential relaxation, and increases at an accelerated rate. Such behavior is typically observed for glass systems containing mobile ionic species and indicates the onset of a cooperative mechanism for the conduction of Ga(I) ions as Ga2I62–, and GaI4– units gain structural freedom.[47] Conductivity enhancement due to a cooperative mechanism was recently also reported for the molecular glass Li3PS4.[28] The supercooled liquid region Tg < T < Tc can be well fitted using the empirical Vogel–Fulcher–Tammann (VFT) expression σ(T) = σVFTe–.[48] This yields the fit parameters EVFT = 0.014 eV, T0 = 296 K, and σVFT = 1.36 10–4 S cm–1.

The real part YRe(ω) of the spectral ac admittance (i.e., the inverse of the complex electrical impedance) in the glassy state presents a wide low-frequency plateau, extending to higher frequencies as the ionic mobility increases with temperature (Figure B). This corresponds to the bulk ionic conductivity, and the extrapolated values YRe(ω = 0) (inset to Figure B) follow the observed increase in the dc conductivity of g-Ga2I3.17, reproducing a comparable activation energy of 0.8 eV below Tg. A high-frequency limit is not observed within the recorded range (40 Hz – 50 MHz), as data above 3 MHz are influenced by the strong pickup from wiring contributions.[49]

The respective imaginary part −Yim(ω) of ac admittance is characterized by a broad minimum centered around ωmean ≈ 104 Hz at 23 °C (Figure C). With increasing temperature, the relaxation time decreases and the broad minimum splits into two individual contributions around Tg ≈ 37–41 °C, before ending in a single narrow minimum at 0.5–1.4 × 106 Hz above 44 °C, indicating a single relaxation process.[50]

The clear observation of two relaxation processes around Tg with distinct time constants is intriguing and raises questions about their structural origins.[51] In amorphous polymers and glass-forming liquids, the primary or structural α-relaxation determines the glass transition and has been related to cooperative molecular rearrangements.[52,53] For many materials, a secondary β-relaxation, or Johari–Goldstein relaxation, has been observed in their dielectric or impedance spectra, which manifests as a high-frequency contribution on an individual timescale and temperature dependence.[54] Its exact nature is still debated but has been associated with hindered noncooperative reorientations or translations in the local regions of low mobility.[55]

Within this model, the two observed minima may be attributed to the above-described structural primary and secondary β-relaxation.[56] The timescale of the primary relaxation is observable as a low-frequency contribution close to the glass-transition temperature but is indistinguishable from the β-relaxation above Tg. The exact temperature dependence of the low-frequency contribution is hard to determine within the narrow observation window.

The narrowing of the minimum at low temperatures may be attributed to the freeze-out and decrease in magnitude of the primary relaxation mechanism. Thus, the primary relaxation may be understood as cooperative translations and reorientations of Ga(I), Ga2I62–, and GaI4– ions in g-Ga2I3.17.

Below Tg, the movement of molecular units freezes out, and the Ga(I) migration follows pure Arrhenius behavior.[57] A third contribution, corresponding to the slow time constant of the electrode/electrolyte contact, gives rise to a minimum at low frequencies (ω < 102 Hz), which is not resolved within the observed frequency range.

As the sample crystallizes into a physical mixture of c-Ga2I3 and c-Ga2I4 around 80 °C, the dc conductivity drops by 2 orders of magnitude, and the Arrhenius behavior is restored. A linear fit to ln(σ) versus 1/T yields an activation energy of Ea = 0.68 eV, corresponding to a 15% increase compared to that of g-Ga2I3.17 (Ea = 0.59 eV). While the two-phase nature of the crystalline sample complicates interpretation, the increased energy barrier for Ga(I) migration is in line with the change in the average coordination number of Ga(I) by iodine from 4.6 in g-Ga2I3.17 to 8 in c-Ga2I3 and c-Ga2I4, as observed in neutron diffraction. A comparable value for the activation enthalpy of 0.8 eV has been reported for the electrical conductivity of the structurally related crystalline compound Ga2Br3.[58] High ionic conductivity appears as a universal property of the A2[Ga2X6]-related compounds, rendering especially the lithium analogues interesting for application.[46] The electrical ac admittance in the crystallized sample displays a low-frequency plateau in YRe(ω) in line with the observed ionic dc conductivity, and −Yim(ω) indicates a single relaxation mechanism (Figure S8).

Conclusions

A high glass-forming tendency was found for melts with composition Ga2I3.17, located between the two binary crystalline compounds Ga2I3 and Ga2I4. Raman spectroscopy and neutron scattering revealed that the Ga2I3.17 glass can be described as a mixture of the molecular anions Ga2I62– and GaI4– coordinated by Ga+ cations. Remarkably, the glass contains gallium in three formal oxidation states, and its molecular make-up can be summarized by the formula [Ga+]63.2[Ga2I62–]26.3[GaI4–]10.5.

A strong Raman mode at 106 cm–1 was observed for glassy and liquid Ga2I3.17, which is absent in the spectra of the crystalline phases Ga2I3 and Ga2I4 and indicative of strong intermolecular interactions at the glass composition. Temperature-dependent Raman and admittance spectroscopy show that the new mode is strongly connected to the local structure in glassy and liquid Ga2I3.17 and the structural relaxation at the glass transition. The sharp peak shape suggests the resonance of a well-defined local molecular arrangement possible only in the presence of both Ga2I62– and GaI4– units.

Comparison of the extrapolated dc conductivity of the crystallized mixture Ga2I3/Ga2I4 suggests a conductivity increase by several orders of magnitude in glassy Ga2I3.17 compared to the ordered crystalline phases. The existence of the related crystalline compounds LiGaI3, LiGaBr3, and LiGaCl3[46,59] hints that g-Ga2I3.17 may be the first representative of a whole family of mixed molecular glasses with substitutional flexibility, heralding the advent of new ion-conducting glasses for lithium- or sodium-based energy storage concepts.