Superionic Conductors: When a Solid's Ions Flow Like a Liquid

What makes a solid 'superionic'

A superionic conductor (also called a fast-ion conductor or solid electrolyte) is a crystalline solid in which one ionic species moves with a conductivity approaching that of a molten salt or aqueous electrolyte — roughly 10⁻³ to 10⁰ S cm⁻¹ — while the remaining framework of counter-ions stays rigidly crystalline. The material is a bona fide solid: it holds its shape, gives sharp Bragg peaks for the immobile sublattice, and has a definite melting point far above the temperature where fast conduction begins. What is unusual is that the mobile sublattice has effectively melted while the crystal has not.

The operational fingerprints are three. First, the ionic transport number tᵢₒₙ ≈ 1: essentially all charge is carried by ions, not electrons, which is exactly what you want in a battery electrolyte (an electronic short would self-discharge the cell). Second, a low activation energy for conduction, typically Eₐ ≈ 0.05–0.25 eV, versus ~0.6–1.5 eV for an ordinary ionic salt. Third, an anomalously large concentration of mobile carriers: instead of the dilute Schottky/Frenkel defects (10⁻⁴–10⁻⁶ per lattice site) that carry current in NaCl, a superionic conductor has a mobile sublattice that is intrinsically disordered — the ions themselves are the 'defects.'

The canonical example is α-AgI. Below 147 °C, β/γ-AgI is an ordinary poorly-conducting, wide-gap ionic solid (a photoconductor, band gap ~2.8 eV). Above 147 °C it transforms to a body-centered-cubic (Im3̄m) arrangement of I⁻ anions, and the two Ag⁺ per formula unit are distributed statistically over 42 crystallographically available sites (tetrahedral 12d, octahedral 6b, trigonal 24h). With so many empty sites and low barriers between them, Ag⁺ diffuses almost freely. Strukturbericht and structural refinements by Strock (1934–1936) established this picture; the conductivity leaps from ~10⁻³ to ~1.3 S cm⁻¹ across the transition.

The mechanism: sublattice melting and a liquid on a lattice

The physically clarifying idea, due largely to Michael O'Keeffe, Robert Huggins, and others in the 1970s, is that the superionic transition is a partial melting. In α-AgI the Ag⁺ sublattice has an entropy of disordering comparable to a real melt: the enthalpy and entropy of the β→α transition (ΔH ≈ 5–6 kJ mol⁻¹, ΔS ≈ 13–15 J mol⁻¹ K⁻¹ near 420 K, with reported values scattering across measurements) are close to those of the actual melting of AgI at ~555 °C (the literature value varies over ~552–558 °C). In effect, the Ag⁺ ions 'melt' roughly 400 °C early, absorbing much of the latent heat, so that the true melt has an anomalously small enthalpy of fusion. The immobile I⁻ framework provides an ordered, low-energy, near-percolating network of interstitial channels through which the silvers roam.

Three structural ingredients recur in every good conductor. (1) A rigid, polarizable anion framework — large soft anions like I⁻ or S²⁻ (a soft-base, soft-acid pairing in HSAB terms) whose polarizability flattens the potential-energy landscape the cation traverses. (2) More available sites than mobile ions, so migration never requires creating a defect — only crossing a low saddle point between adjacent partially-occupied wells. (3) Face-sharing or edge-sharing polyhedra that give continuous, low-barrier diffusion channels rather than isolated cages.

• Type I (order–disorder / first-order): α-AgI, Ag₂S, Cu₂S — a sharp phase transition at which conductivity jumps discontinuously.
• Type II (Faraday / gradual): PbF₂, CaF₂ (F⁻ conductors) — conductivity rises smoothly and continuously toward a broad diffuse specific-heat anomaly (a 'Bredig transition'), with no first-order jump.
• Type III (framework / stoichiometric): β-alumina (Na⁺), NASICON, LISICON, garnets — the mobile-ion channels are built into a permanently rigid host that never melts; these conduct fast even at room temperature.

Modern ab initio molecular dynamics reveals that the fast ions do not hop independently. In the best conductors, motion is concerted and string-like: a cation leaving a site is pushed by, and pulls, its neighbors in correlated chains. This shows up in a Haven ratio H_R = D_tracer/D_σ < 1 (often 0.3–0.7), because collective displacement moves more charge than the sum of independent tracer jumps would predict. It is the crystalline analogue of cooperative rearrangement in a supercooled liquid.

The transport equations: Nernst–Einstein and Arrhenius

Ionic conductivity is the product of carrier density, charge, and mobility, σ = n q μ. Linking mobility to microscopic diffusion via the Nernst–Einstein relation, μ = q D/(k_B T), gives the working equation of solid-state ionics:

σ = n q² D / (k_B T) × (1/H_R),

where n is the number density of mobile ions, q their charge (for a monovalent cation q = e = 1.602×10⁻¹⁹ C), D the self-diffusion coefficient, and H_R the Haven ratio correcting for correlated motion. The 1/H_R factor is what lets concerted diffusion push σ above the naïve tracer estimate; for uncorrelated random-walk hopping H_R = 1 and the correction vanishes.

The diffusion coefficient is thermally activated: D = D₀ exp(−Eₐ/k_B T), and combining the two gives σT = A exp(−Eₐ/k_B T). Plotting ln(σT) versus 1/T yields a straight line whose slope is −Eₐ/k_B. The key superionic signature in such an Arrhenius plot is a low slope (small Eₐ) and — for Type-I materials — a sharp break where σ jumps at the order–disorder temperature. In β-alumina and LGPS the line is continuous with Eₐ as low as ~0.2 eV, meaning σ only doubles over ~30–40 K rather than the order-of-magnitude change per 100 K seen in ordinary salts. Because migration is barrier-limited (not defect-formation-limited), the pre-factor and activation energy are decoupled from the huge Schottky formation enthalpy that dominates in NaCl-type materials.

A worked example: how fast is Ag⁺ in α-AgI?

Let us check that the measured conductivity is consistent with genuinely liquid-like silver mobility. Take α-AgI just above the transition at T = 150 °C = 423 K, with measured σ ≈ 1.3 S cm⁻¹.

The mobile-ion density: α-AgI is bcc I⁻ with lattice parameter a ≈ 5.06 Å = 5.06×10⁻⁸ cm, so the cell volume is a³ ≈ 1.30×10⁻²² cm³. There are 2 I⁻ and therefore 2 mobile Ag⁺ per cell, giving n = 2/(1.30×10⁻²²) ≈ 1.55×10²² cm⁻³. Rearranging Nernst–Einstein (taking H_R ≈ 1 for the estimate):

D = σ k_B T / (n q²) = (1.3)(1.381×10⁻²³ × 423) / [(1.55×10²²)(1.602×10⁻¹⁹)²] C-and-SI-units... converting to CGS-consistent units gives D ≈ 1.9×10⁻⁵ cm² s⁻¹.

That number is the punchline. A self-diffusion coefficient of ~2×10⁻⁵ cm² s⁻¹ is the same order as ions in liquid water (Na⁺ in water is ~1.3×10⁻⁵ cm² s⁻¹ at 25 °C) and roughly what you measure for Ag⁺ in molten AgI. In an ordinary ionic solid at the same temperature, D would be ~10⁻¹² cm² s⁻¹ or smaller. The silver ions in a solid crystal are diffusing at liquid-electrolyte rates — the literal meaning of 'ions flowing like a liquid.' Independent NMR relaxation and radiotracer measurements agree with the value derived from conductivity, which is itself strong evidence that transport is ionic (tᵢₒₙ ≈ 1) and that the Haven ratio is near, though slightly below, unity.

Limits, subtleties, and what can go wrong

Superionic conduction is real but bounded, and several caveats separate laboratory curiosities from useful electrolytes.

• Metastability and phase control. The champion Ag⁺ conductor RbAg₄I₅ (discovered by Bradley & Greene and by Owens & Argue in 1967) reaches ~0.27 S cm⁻¹ at 25 °C — the highest room-temperature ionic conductivity known for a solid — but it is metastable and decomposes below ~27 °C into Rb₂AgI₃ + β-AgI over time (2 RbAg₄I₅ → Rb₂AgI₃ + 7 AgI). Many high-σ phases exist only in a narrow window or must be quenched.
• Electronic leakage. Chalcogenides like Ag₂S and Cu₂S are mixed conductors: they carry both Ag⁺/Cu⁺ ions and electrons, so tᵢₒₙ < 1. A battery separator built from a mixed conductor would self-discharge. The best solid electrolytes deliberately use wide-gap, electronically insulating frameworks (halides, oxides, sulfides with a good band gap).
• Electrochemical window vs. conductivity trade-off. The very softness and polarizability that flatten the migration barrier also narrow the stability window. LGPS (Li₁₀GeP₂S₁₂) conducts superbly (1.2×10⁻² S cm⁻¹) but density-functional calculations (Mo, Ong & Ceder, Chem. Mater. 2012) showed its thermodynamic window is only ~1.7–2.3 V depending on the analysis — it is reduced by Li metal and oxidized above ~2 V, surviving in real cells only via kinetically-passivating decomposition interphases.
• Grain boundaries. Bulk (single-crystal) conductivity often exceeds the practical polycrystalline value by an order of magnitude because grain boundaries block ions; impedance spectroscopy is used to separate bulk from grain-boundary and electrode contributions.

A conceptual subtlety worth stating plainly: the 'melted' sublattice is not a true liquid. Its diffusion is still guided by the crystalline potential of the rigid framework — ions channel along specific crystallographic pathways, not isotropically. β-alumina, for instance, is a two-dimensional conductor: Na⁺ moves fast within the conduction planes and essentially not at all across the spinel blocks. The anisotropy is a direct readout of the host structure and disappears the moment the whole crystal actually melts.

History and applications: from Faraday to solid-state batteries

The phenomenon is old. Michael Faraday, in the 1830s, noticed that hot silver sulfide and lead fluoride conducted electricity by mass transport rather than as metals — arguably the first observation of solid electrolytes. The modern quantitative era begins with Carl Tubandt and Erich Lorenz (1914), who measured the enormous conductivity of α-AgI and demonstrated by electrolysis (silver deposited at the cathode, mass transported) that current was carried by silver ions. Structural understanding followed with Strock's 1930s refinements and blossomed in the 1960s–70s with the discovery of β-alumina (Na₂O·11Al₂O₃, the Na⁺ conductor behind the sodium–sulfur battery, Kummer & Weber at Ford, 1967) and RbAg₄I₅.

The applications are now central to energy technology:

• Solid-state lithium batteries. Sulfide superionic conductors — Li₁₀GeP₂S₁₂ (LGPS) from Kamaya, Kanno and co-workers (Nature Materials, 2011), and the argyrodites Li₆PS₅X (X = Cl, Br) — reach room-temperature σ of 10⁻³–10⁻² S cm⁻¹, matching liquid electrolytes while (in principle) removing flammable organic solvents and enabling lithium-metal anodes. Kanno's Li₉.₅₄Si₁.₇₄P₁.₄₄S₁₁.₇Cl₀.₃ (2016) hit ~2.5×10⁻² S cm⁻¹, the current benchmark.
• Solid-oxide fuel cells (SOFCs). Yttria-stabilized zirconia (YSZ) is an O²⁻ superionic conductor above ~700 °C; doping Zr⁴⁺ with Y³⁺ creates the oxygen vacancies that carry current, powering high-temperature fuel cells and oxygen sensors.
• Sensors and memory. The λ-probe (lambda sensor) in every catalytic-converter car uses YSZ and the Nernst equation to read exhaust O₂ partial pressure. Ag⁺/Cu⁺ conductors underpin electrochemical-metallization (conductive-bridge) resistive memory, where a metal filament is grown and dissolved through a solid electrolyte.

The through-line from Faraday's glowing Ag₂S to a solid-state EV battery is a single idea: if you can persuade one sublattice of a crystal to behave like a liquid, you get the mechanical robustness and thermal safety of a solid together with the ionic throughput of a molten salt — the best of both phases in one material.

Two mechanistic pictures of fast ion transport: dilute point-defect hopping vs. superionic sublattice disorder

Property	Normal ionic solid (defect-hopping)	Superionic conductor
Carrier density	Dilute (Schottky/Frenkel), ~10⁻⁴–10⁻⁶ per site	Near-unity site disorder on mobile sublattice
Conductivity σ (near use T)	10⁻¹⁰–10⁻⁴ S cm⁻¹	10⁻³–10⁰ S cm⁻¹
Activation energy Eₐ	~0.6–1.5 eV (migration + ½ formation)	~0.05–0.25 eV (migration only)
Mobile-ion site occupancy	Ions localized on lattice sites	Ions delocalized over many partial-occupancy sites
Transition character	Smooth Arrhenius, no phase change	Often first-order 'sublattice melting' (α-AgI at 147 °C)
Correlation (Haven ratio H_R)	≈ 1 (uncorrelated hops)	< 1 (concerted, string-like motion)