A computational materials investigation combining high-throughput database screening, ab initio molecular dynamics, and machine-learning interatomic potentials has identified the gamma phase of lithium scandium phosphate, chemically designated as Li3Sc2(PO4)3, as an exceptionally promising protective interfacial coating for mid-nickel layered oxide cathodes in all-solid-state lithium batteries. Layered oxide cathode materials containing moderate nickel fractions, specifically LiNi0.5Co0.2Mn0.3O2 or NCM523, have attracted renewed research attention because they offer enhanced structural stability, significantly reduced lattice oxygen evolution, and lower raw material expenses compared to high-nickel alternatives such as NCM811. When operated at elevated cutoff potentials of 4.2 volts or higher, mid-nickel cathodes compensate for their lower theoretical specific capacity, delivering energy densities competitive with high-nickel chemistries. However, continuous operation under high-voltage conditions introduces severe interfacial degradation, primarily driven by deleterious parasitic chemical and electrochemical side reactions at the boundary between the active cathode particles and solid-state electrolytes.
The transition toward all-solid-state battery architectures is widely regarded as a critical milestone for next-generation energy storage, particularly in meeting the demanding safety, durability, and range requirements of electric vehicles targeting operational distances exceeding 300 miles per charge. Replacing liquid formulations with inorganic solid-state electrolytes provides enhanced operational safety, eliminates the need for polymeric separators, and mechanically enables the integration of pure lithium metal anodes, which dramatically boosts cell-level energy density. Nevertheless, sulfide compounds exhibit a comparatively narrow electrochemical stability window and pronounced chemical reactivity toward transition metals and delithiated oxide surfaces, necessitating robust interfacial protection.
In high-nickel cathode systems containing 80 percent or more nickel, surface instability triggers irreversible phase transformations into electrochemically inactive rock-salt structures alongside substantial transition metal dissolution and oxygen release. Identifying coating materials capable of surviving these demanding conditions has historically depended on empirical trial-and-error synthesis, yielding only a limited selection of legacy oxides such as lithium niobate, lithium aluminate, and lithium phosphate. To break this technological bottleneck, researchers implemented an automated, data-driven high-throughput screening framework to systematically evaluate thousands of candidate compounds against multi-property criteria.
The computational workflow commenced with a comprehensive extraction of crystal structures from the Materials Project database, referencing the 2025.02.12 dataset version, which yielded an initial candidate pool of 17,230 lithium- and oxygen-containing oxide materials. Oxides were deliberately chosen as the core chemical library due to their intrinsic oxidative resistance, wide electrochemical stability windows, and established precedent in protective interface engineering. To refine this broad library into an experimentally relevant dataset, the authors implemented rigorous preprocessing filters that removed any compositions containing radioactive elements or structures incorporating more than two constituent elements beyond lithium and oxygen. Furthermore, whenever multiple polymorphs shared an identical chemical stoichiometry, only the phase exhibiting the lowest calculated formation energy was retained, ensuring that the remaining candidates represented thermodynamically stable ground states. This preprocessing pipeline successfully condensed the candidate library down to 4,634 well-defined oxide structures, establishing a consistent basis for property-targeted screening.
The primary screening hurdle evaluated the theoretical electrochemical stability window of each candidate material using the grand potential phase diagram methodology, which calculates compound decomposition energetics as a continuous function of lithium chemical potential. Protective coatings operating in mid-nickel all-solid-state cells must endure both oxidation under high charging voltages and reduction during discharge. By enforcing strict cutoff thresholds that required an oxidation limit above 3.5 volts and a reduction potential below 2.0 volts relative to lithium metal, the candidate library was narrowed to 265 electrochemically stable compounds. Subsequent evaluation focused on interfacial thermodynamic compatibility by calculating the interfacial reaction energy, denoted as Delta Erxn, between each screened candidate and both the NCM523 cathode and the Li6PS5Cl sulfide electrolyte. Applying an established stability threshold of Delta Erxn greater than or equal to negative 100 millielectronvolts per atom reduced the candidate pool to 154 materials, revealing that while most oxides remained stable against NCM523, the majority degraded against the highly reactive sulfur species in Li6PS5Cl.
To guarantee that the protective layer does not create parasitic electronic leakage pathways that accelerate cell self-discharge, the researchers assessed the electronic band gaps of the surviving materials using values calculated via density functional theory with the Perdew-Burke-Ernzerhof generalized gradient approximation. Because standard generalized gradient approximation functionals systematically underestimate band gap values, the investigators applied a conservative threshold requiring an electronic band gap exceeding 2.0 electron volts to confirm insulating character, which passed 150 candidates. However, a viable protective coating must simultaneously allow rapid lithium-ion transport across the cathode-electrolyte boundary. Because performing ab initio molecular dynamics across hundreds of structures is computationally prohibitive, the team devised a structural network descriptor based on lithium-lithium interatomic distances below 3.5 angstroms. Validated against known superionic conductors including Li6PS5Cl, Li7La3Zr2O12, and Li10GeP2S12, this geometric filter identified continuous three-dimensional lithium conduction networks, yielding 88 promising oxide materials while preserving compositional diversity across metallic and non-metallic systems.
To isolate candidates suited for commercial high-voltage operation, the authors subjected the 88 materials to even tighter screening criteria, requiring thermodynamic oxidation stability exceeding 4.0 volts and interfacial reaction energies above negative 50 millielectronvolts per atom against both NCM523 and Li6PS5Cl. This rigorous refinement identified eight outstanding candidate structures, which prominently included boron- and phosphorus-based compositions such as Li3B11O18, LiB3O5, Li3PO4, LiMgPO4, and Li3Sc2(PO4)3, whose stability is enhanced by strong covalent bonding between oxygen and boron or phosphorus atoms. To dynamically verify lithium-ion transport, the researchers performed preliminary ab initio molecular dynamics simulations on the candidates at 1000 Kelvin over a 20-picosecond duration. While most candidates displayed confined mean squared displacements due to the brief simulation timescale, Li3Sc2(PO4)3 exhibited dramatically higher lithium diffusivity than all other tested compounds, confirming that its interconnected framework facilitates exceptional atomic-scale cation mobility under operating conditions.
Delving into the practical viability of Li3Sc2(PO4)3 required reconciling the computational screening hit with established experimental synthesis literature, which documents that the material exists in three distinct crystallographic polymorphs designated as the alpha, beta, and gamma phases. While previous computational screening efforts had occasionally highlighted Li3Sc2(PO4)3, they focused almost exclusively on the high-temperature gamma phase without examining phase transitions or synthesized structural forms. Experimental studies dating back to 1990 demonstrated that the monoclinic alpha phase converts into the fast-ion-conducting gamma phase upon thermal annealing at 573 Kelvin, and that partial chemical substitution with tetravalent titanium or zirconium ions effectively stabilizes the conductive gamma phase at ambient room temperature. Comparative structural examination confirmed that while both the alpha and gamma phases crystallize within the monoclinic space group P2_1/c, the alpha phase features broken, discontinuous lithium-lithium pathways, whereas the gamma polymorph possesses two continuous, highly interconnected percolation networks.
To quantitatively compare the ionic transport characteristics of the two structural polymorphs, the research team performed extended ab initio molecular dynamics simulations at 900 Kelvin over 200 picoseconds, tracking atomic trajectories and three-dimensional probability density isosurfaces. In the alpha phase, lithium probability density remained strictly confined around localized crystallographic positions, yielding a flat mean squared displacement profile indicative of trapped cations. In contrast, the gamma phase displayed continuous, overlapping probability density distributions that mapped directly onto interconnected transport pathways, generating a pronounced upward trajectory in mean squared displacement. Systematic calculations across temperatures spanning 800 to 1200 Kelvin in 100-Kelvin increments, followed by Arrhenius extrapolation to 300 Kelvin, determined that the gamma phase achieves an estimated room-temperature ionic conductivity of 0.2 millisiemens per centimeter, bounded by an uncertainty interval of 0.01 to 2.4 millisiemens per centimeter, whereas the alpha phase reached only 0.006 millisiemens per centimeter.
Because static thermodynamic calculations cannot capture dynamic non-equilibrium interfacial phenomena occurring during battery cycling, the authors constructed large-scale heterointerface supercells containing over 1,000 atoms to directly track atomistic evolution over 500 picoseconds. To overcome the computational scaling limitations of ab initio molecular dynamics, the simulations employed SevenNet, a graph-based universal machine-learning interatomic potential pre-trained on extensive quantum-mechanical datasets. The multi-layer simulation models aligned the most stable crystallographic surfaces: the (104) surface of fully lithiated NCM523, the (100) surface of Li6PS5Cl, and the (101) surface of gamma-phase Li3Sc2(PO4)3, selected to minimize directional lattice mismatch.
Dynamic simulations of the unprotected NCM523(104) and Li6PS5Cl(100) interface revealed severe structural degradation, marked by the spontaneous migration of nickel atoms out of the layered cathode lattice and directly into the sulfide electrolyte. In sharp contrast, simulations of the coated architecture incorporating a Li3Sc2(PO4)3(101) protective interlayer between NCM523 and Li6PS5Cl demonstrated exceptional structural integrity across the entire 500-picosecond trajectory. The internal PO4 tetrahedra and ScO6 octahedra of the coating remained completely stable, preventing nickel ingress into the electrolyte and suppressing PS4 decomposition through the formation of stable scandium-sulfur surface bonds, directly validating the coating material as an effective interfacial barrier for mid-nickel all-solid-state cells.
Subject of Research: High-throughput computational discovery of protective cathode coating materials for all-solid-state lithium batteries
Article Title: High-throughput discovery of Li3Sc2(PO4)3 as a protective coating for stabilizing mid-Ni NCM interfaces in all-solid-state batteries
Article References: Kim, J. H., Lee, S., & Lee, S. U. (2026). High-throughput discovery of Li3Sc2(PO4)3 as a protective coating for stabilizing mid-Ni NCM interfaces in all-solid-state batteries. Nano Convergence, 13(1), Article 27. https://doi.org/10.1186/s40580-026-00555-z
Image Credits: AI Generated
DOI: 10.1186/s40580-026-00555-z
Keywords: all-solid-state batteries, mid-nickel cathode, NCM523, Li3Sc2(PO4)3, solid-state electrolyte, Li6PS5Cl, high-throughput screening, ab initio molecular dynamics, universal machine-learning interatomic potential, SevenNet, electrochemical stability window, interfacial reaction energy, ionic conductivity, density functional theory, Materials Project, protective coating, cathode-electrolyte interphase
Cite Scienmag News
Neil Sanderson. (November 19, 2026). Computational Screening Identifies Protective Lithium Scandium Phosphate Coating for Solid-State Batteries. Scienmag. https://scienmag.com/computational-screening-identifies-protective-lithium-scandium-phosphate-coating-for-solid-state-batteries/
Neil Sanderson. "Computational Screening Identifies Protective Lithium Scandium Phosphate Coating for Solid-State Batteries." Scienmag, 19 November 2026, https://scienmag.com/computational-screening-identifies-protective-lithium-scandium-phosphate-coating-for-solid-state-batteries/. Accessed 19 November 2026.
Neil Sanderson. "Computational Screening Identifies Protective Lithium Scandium Phosphate Coating for Solid-State Batteries." Scienmag. November 19, 2026. https://scienmag.com/computational-screening-identifies-protective-lithium-scandium-phosphate-coating-for-solid-state-batteries/








