{
  "meta": {
    "title": "Gallium-oxide family power-electronics screening — case study results",
    "producer": "FluxMateria",
    "version": "1.4",
    "date": "2026-07-14",
    "license": "CC BY 4.0",
    "scope": "Screening-grade predictions to prioritize experiments, not measured device data. This is a HYBRID screening pipeline built on FluxMateria's Universal Materials Engine: electronic properties are derived from its underlying proprietary FLUX Theory physics with zero parameters fit to the target property, while the final ranking also uses literature-measured thermal conductivity, an empirical breakdown-field scaling, and an empirical miscibility gate. Only band gap is validated against a literature reference dataset here; every other property carries an explicit evidence level (see `validation_scope`).",
    "pipeline": {
      "full_physics_evaluations": 10,
      "interpolated_alloy_screens": 1201617,
      "miscible_after_gate": 72599,
      "dropped_by_miscibility_gate": 1129018,
      "full_property_deep_dive_candidates": 6,
      "deep_dive_reference_rows": 3,
      "deep_dive_datasheet_rows_total": 9,
      "note": "The 1.2M figure is an interpolated ranking pass (Vegard interpolation of the 10 Universal Materials Engine-evaluated anchor oxides), not 1.2M independent physics evaluations. The deep-dive datasheet has 9 rows: 6 candidate oxides plus 3 reference rows (beta-Ga2O3 baseline, 4H-SiC, GaN)."
    }
  },

  "blind_protocol": {
    "description": "Order of operations, to make the 'literature convergence' claim auditable. FluxMateria's Universal Materials Engine had no access to the power-electronics literature during enumeration or ranking; the literature comparison was performed only after the ranking was fixed.",
    "steps": [
      "1. FluxMateria's Universal Materials Engine, powered by proprietary FLUX Theory physics, evaluated 10 anchor oxides from chemical formula alone; thermal conductivity was supplied from a measured-literature table.",
      "2. 1,201,617 sesquioxide-alloy compositions (binary–quinary, 5% grid) enumerated and ranked by Vegard interpolation of the anchor properties.",
      "3. Manufacturability gates applied: ionic-radius miscibility gate (dropped 1,129,018 = 94%) and literature-conductivity substitution for supported endpoints; rankings dependent on unsupported Sc2O3/Y2O3/Lu2O3 conductivity inputs are withheld from this public artifact.",
      "4. Shortlist re-evaluated through the Universal Materials Engine (deep-dive datasheet).",
      "5. Ranking compared to the published power-electronics record ONLY at this point (see `prior_art`).",
      "6. The archived campaign reproduced the supported hardened result in approximately 2.4 seconds wall time."
    ],
    "campaign_date": "2026-07-14"
  },

  "validation_scope": {
    "band_gap": {"level": "validated_here", "basis": "17 wide-gap oxides scored against literature reference gaps: 10 from measured sources, 5 from a semiconductor-property compilation, 1 secondary, 1 GW computation (NaGaO2, no measurement located). No DFT-PBE values used as targets. Reference gaps mix optical/fundamental/direct/indirect transitions (per-entry `gap_type`), so individual comparisons are approximate; MAE ~0.6 eV, material-specific (roughly unbiased)."},
    "thermal_conductivity": {"level": "literature_supplied_endpoints", "basis": "Room-temperature values curated per supported anchor oxide; alloys interpolated. The Universal Materials Engine thermal proxy is not used for endpoints. Unsupported Sc2O3/Y2O3/Lu2O3 values and dependent rankings are excluded."},
    "electron_mobility": {"level": "benchmarked_elsewhere", "basis": "Same Universal Materials Engine; benchmarked in the semiconductor mobility atlas study, not re-validated on this oxide set."},
    "breakdown_field": {"level": "estimated", "basis": "Empirical scaling from band gap (E_crit ~ Eg^2.5); standard for FoM ranking, not an independent calculation."},
    "dielectric_constant": {"level": "predicted", "basis": "Universal Materials Engine prediction, not independently validated on this set."},
    "dopability": {"level": "proxy_plus_literature", "basis": "Hole-transport screen is a proxy; ZnGa2O4 bipolar doping is literature-supported."},
    "miscibility": {"level": "empirical_gate", "basis": "Ionic-radius mismatch vs documented oxide-alloy solubility; a literature-grounded heuristic, not a computed mixing free energy."},
    "melting_point": {"level": "not_reliable_excluded", "basis": "The Universal Materials Engine under-predicts this property (beta-Ga2O3 ~1307 K vs literature ~2093 K); excluded from reported outputs."}
  },

  "methods_comparison": {
    "comparison_scope": "Time and resource scale for traversing the same 1,201,617-entry composition grid and reaching shortlist-scale computational triage. This is not a claim that different methods would return an identical shortlist or that screening replaces experimental validation.",
    "fluxmateria": {
      "archived_campaign_wall_time_seconds": 2.4,
      "workload": "10 full-property anchor evaluations, 1,201,617 interpolated alloy screens with manufacturability gates, and 6 candidate deep dives",
      "project_specific_ml_training": false
    },
    "direct_dft_historical_scale_estimate": {
      "source_basis": "The 2013 Materials Project paper reported more than 15 million CPU-hours for over 33,000 compounds and noted that basic properties for one material may require several hundred CPU-hours.",
      "arithmetic_yardstick_cpu_hours": 546189545,
      "arithmetic_yardstick_cpu_years": 62350,
      "illustrative_wall_years_at_10000_continuously_utilized_cores": 6.2,
      "qualification": "Order-of-magnitude historical extrapolation, not a measured head-to-head benchmark or a forecast for modern optimized codes. It is optimistic for disordered alloys because DFT requires explicit structures and often multiple configurations per composition.",
      "refs": [21]
    },
    "realistic_dft_surrogate_workflow": {
      "approach": "Calculate a selected explicit-structure set with DFT, fit a cluster-expansion or ML surrogate, screen the larger space, and return finalists to higher fidelity.",
      "time_scale": "days plus institutional HPC for new-compound iterations; total campaign time depends on structures, convergence, queueing, property stack, and training-set size",
      "qualification": "Cluster expansion reduces rather than removes first-principles reference calculations. A published multi-fidelity workflow reduced optimization cost by about threefold in its test problems.",
      "refs": [21, 22, 23]
    },
    "pretrained_machine_learning": {
      "time_scale": "potentially seconds to minutes for inference after model and representations exist",
      "qualification": "Requires upstream training data, domain coverage, uncertainty control, and high-fidelity verification. GNoME used iterative graph-network filtering followed by DFT verification and targeted stability rather than this power-electronics property stack.",
      "refs": [24]
    },
    "experimental_scale": {
      "combinatorial_example": "A published thin-film workflow reports 8 hours to composition-map 342 measurement areas. Literal arithmetic for 1,201,617 points is about 28,100 hours (3.2 years) for that measurement alone, excluding fabrication and other properties.",
      "autonomous_lab_example": "A-Lab synthesized 36 of 57 inorganic targets over 17 days of continuous operation; this is a synthesis reference, not full power-device validation.",
      "refs": [25, 26]
    },
    "bottom_line": "FluxMateria reduced a million-scale compositional hypothesis space to a validation-sized set in seconds. DFT, defect calculations, crystal growth, thermal measurement, breakdown testing, and device fabrication should now be concentrated on that reduced set."
  },

  "band_gap_validation_eV": [
    {"formula": "Ga2O3", "predicted": 4.3, "measured": 4.8, "gap_type": "fundamental (beta)", "source_type": "measured_review", "source_url": "https://doi.org/10.1002/aelm.202300844"},
    {"formula": "Al2O3", "predicted": 7.4, "measured": 8.7, "gap_type": "optical (alpha, sapphire)", "source_type": "measured_primary", "source_url": "https://doi.org/10.1126/sciadv.abd5891"},
    {"formula": "In2O3", "predicted": 2.7, "measured": 2.9, "gap_type": "fundamental", "source_type": "measured_primary", "source_url": "https://pubs.acs.org/doi/10.1021/acs.jpcc.4c06718"},
    {"formula": "Sc2O3", "predicted": 6.0, "measured": 6.0, "gap_type": "optical", "source_type": "measured_primary", "source_url": "https://www.osti.gov/biblio/4014535"},
    {"formula": "Y2O3", "predicted": 5.9, "measured": 5.5, "gap_type": "optical", "source_type": "measured_primary", "source_url": "https://www.sciencedirect.com/science/article/abs/pii/S0925346711001741"},
    {"formula": "ZnGa2O4", "predicted": 4.4, "measured": 4.67, "gap_type": "optical (spinel)", "source_type": "measured_primary", "source_url": "https://pubs.acs.org/doi/10.1021/acsorginorgau.3c00030"},
    {"formula": "MgGa2O4", "predicted": 4.36, "measured": 4.9, "gap_type": "optical (bulk crystal)", "source_type": "measured_primary", "source_url": "https://pubs.acs.org/doi/10.1021/acsorginorgau.3c00030"},
    {"formula": "LiGaO2", "predicted": 6.3, "measured": 5.35, "gap_type": "optical", "source_type": "measured_primary", "source_url": "https://www.sciencedirect.com/science/article/abs/pii/S0169433217300995"},
    {"formula": "NaGaO2", "predicted": 6.28, "measured": 5.5, "gap_type": "computed (QSGW; no measurement located)", "source_type": "computed_gw", "source_url": "https://arxiv.org/abs/2010.15934"},
    {"formula": "LaGaO3", "predicted": 4.0, "measured": 4.4, "gap_type": "optical (perovskite)", "source_type": "secondary", "source_url": "https://www.researchgate.net/publication/314097300"},
    {"formula": "CuGaO2", "predicted": 1.47, "measured": 3.6, "gap_type": "optical (alpha-delafossite); note beta-wurtzite direct 1.47", "source_type": "measured_primary", "source_url": "https://pubs.acs.org/doi/10.1021/jp2087225"},
    {"formula": "Bi2Ga4O9", "predicted": 4.2, "measured": 2.86, "gap_type": "indirect (mullite; visible-light absorber)", "source_type": "measured_primary", "source_url": "https://pubs.acs.org/doi/10.1021/acs.inorgchem.6b00330"},
    {"formula": "ZnO", "predicted": 3.35, "measured": 3.37, "gap_type": "direct", "source_type": "tertiary_compilation", "source_url": "https://en.wikipedia.org/wiki/List_of_semiconductor_materials"},
    {"formula": "SnO2", "predicted": 3.63, "measured": 3.7, "gap_type": "direct", "source_type": "tertiary_compilation", "source_url": "https://en.wikipedia.org/wiki/List_of_semiconductor_materials"},
    {"formula": "TiO2", "predicted": 2.2, "measured": 3.2, "gap_type": "indirect (anatase)", "source_type": "tertiary_compilation", "source_url": "https://en.wikipedia.org/wiki/List_of_semiconductor_materials"},
    {"formula": "Cu2O", "predicted": 2.5, "measured": 2.17, "gap_type": "direct", "source_type": "tertiary_compilation", "source_url": "https://en.wikipedia.org/wiki/List_of_semiconductor_materials"},
    {"formula": "SrTiO3", "predicted": 3.3, "measured": 3.25, "gap_type": "indirect", "source_type": "tertiary_compilation", "source_url": "https://en.wikipedia.org/wiki/List_of_semiconductor_materials"}
  ],
  "band_gap_source_type_counts": {"measured_primary": 9, "measured_review": 1, "computed_gw": 1, "secondary": 1, "tertiary_compilation": 5},

  "thermal_conductivity_endpoints_W_m_K": [
    {"formula": "Al2O3", "kappa": 35.0, "confidence": "high", "note": "high-purity synthetic sapphire single crystal near room temperature", "source_url": "https://ntrs.nasa.gov/api/citations/19660014513/downloads/19660014513.pdf"},
    {"formula": "Ga2O3", "kappa": 16.0, "confidence": "high", "note": "anisotropic 10.9-27.0; representative average", "source_url": "https://doi.org/10.1002/crat.202200204"},
    {"formula": "In2O3", "kappa": 10.0, "confidence": "medium", "source_url": "https://arxiv.org/abs/2008.13519"},
    {"formula": "La2O3", "kappa": 5.0, "confidence": "medium", "source_url": "https://www.sciencedirect.com/science/article/abs/pii/S0040609008002502"},
    {"formula": "Gd2O3", "kappa": 4.0, "confidence": "low", "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC9655147/"},
    {"formula": "Yb2O3", "kappa": 4.0, "confidence": "low", "source_url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6766251/"},
    {"formula": "B2O3", "kappa": 0.52, "confidence": "high", "note": "vitreous (glass) at room T", "source_url": "https://doi.org/10.1007/BF00522148"}
  ],

  "hardened_ranking_supported_results": {
    "description": "Public-supported output after the miscibility gate and literature-conductivity substitution, ranked by the thermal-power (Keyes-weighted) ensemble. Baliga is relative to silicon; Keyes is the thermal-limited metric; dmax is the maximum pairwise ionic-radius mismatch. Sc-rich rows are withheld because their pure-endpoint conductivity sources were not established.",
    "columns": ["rank", "composition", "band_gap_eV", "mobility_e_cm2_Vs", "kappa_W_m_K", "baliga_rel_si", "keyes", "dmax_radius_mismatch"],
    "rows": [
      [1, "(Al0.95Ga0.05)2O3", 7.27, 15.0, 34.0, 3751, 0.38, 0.15]
    ],
    "headline_result": {
      "composition": "(Al0.95Ga0.05)2O3",
      "baliga_rel_si": 3751,
      "vs_gan": 17.5,
      "band_gap_eV": 7.27,
      "keyes": 0.38,
      "kappa_W_m_K": 34.0,
      "note": "This is the ~3,800x-Si Al-rich (Al,Ga)2O3 entry in the shortlist; the ~3,800 figure is its Baliga FoM (3751)."
    }
  },

  "shortlist": [
    {"candidate": "(Al,Ga)2O3 (Al-rich, e.g. (Al0.95Ga0.05)2O3)", "family": "sesquioxide alloy", "baliga_rel_si": 3751, "vs_gan": 17.5, "keyes": 0.38, "evidence_tier": "literature_supported", "note": "Established gallium-oxide bandgap-engineering alloy; miscible, growable, high thermal conductivity. Top of the hardened ranking.", "refs": [2, 19, 20]},
    {"candidate": "ZnGa2O4", "family": "spinel gallate", "baliga_rel_si": 591, "vs_gan": 2.8, "evidence_tier": "literature_supported", "note": "Actively pursued UWBG oxide with demonstrated bipolar (n- and p-type) self-doping.", "refs": [12, 13]},
    {"candidate": "LiGaO2", "family": "alkali gallate", "baliga_rel_si": 2188, "vs_gan": 10, "evidence_tier": "prospective_lead", "note": "Known UWBG oxide, bulk crystals grown; Li-Ga-O family pursued for power.", "refs": [3, 17]},
    {"candidate": "NaGaO2", "family": "alkali gallate", "baliga_rel_si": 2157, "vs_gan": 10, "evidence_tier": "exploratory", "note": "No power-electronics literature located; the one genuinely under-explored candidate.", "refs": []}
  ],

  "fom_leaderboard_note": "Baliga values are breakdown-only scores and do not use thermal conductivity. Sc-containing entries may therefore appear here even though they are excluded from the supported hardened thermal ranking.",
  "fom_leaderboard_baliga_rel_si": [
    {"material": "(Al0.95Ga0.05)2O3", "family": "sesquioxide alloy", "baliga": 3751},
    {"material": "Al2O3", "family": "sesquioxide", "baliga": 3508},
    {"material": "LiGaO2", "family": "alkali gallate", "baliga": 2188},
    {"material": "NaGaO2", "family": "alkali gallate", "baliga": 2157},
    {"material": "(In0.10Sc0.90)2O3", "family": "sesquioxide alloy", "baliga": 2096},
    {"material": "AlN", "family": "incumbent", "baliga": 2148},
    {"material": "Sc2O3", "family": "sesquioxide", "baliga": 1889},
    {"material": "Y2O3", "family": "sesquioxide", "baliga": 1730},
    {"material": "ZnGa2O4", "family": "spinel gallate", "baliga": 591},
    {"material": "MgGa2O4", "family": "spinel gallate", "baliga": 562},
    {"material": "beta-Ga2O3", "family": "sesquioxide", "baliga": 531},
    {"material": "LaGaO3", "family": "perovskite gallate", "baliga": 399},
    {"material": "GaN", "family": "incumbent", "baliga": 214},
    {"material": "4H-SiC", "family": "incumbent", "baliga": 155},
    {"material": "In2O3", "family": "sesquioxide", "baliga": 77},
    {"material": "Si", "family": "reference", "baliga": 1}
  ],

  "deep_dive_datasheet": {
    "note": "Full-property re-evaluation of the shortlist + baseline + incumbents. Melting point excluded (unreliable). Confidence per validation_scope. mu_e in cm2/V/s; bulk_modulus in GPa; kappa in W/m/K (endpoints literature-supplied, others predicted).",
    "rows": [
      {"material": "(Al0.95Ga0.05)2O3", "band_gap_eV": 7.27, "mobility_e": 15, "dielectric": 7.5, "bulk_modulus_GPa": 250, "kappa_W_m_K": 34, "baliga_rel_si": 3751, "keyes": 0.38, "note": "hardened-ranking #1; the ~3,800x-Si result"},
      {"material": "(Al0.85In0.15)2O3", "band_gap_eV": 7.1, "mobility_e": 21, "dielectric": 7.6, "bulk_modulus_GPa": 356, "kappa_W_m_K": 25, "note": "exploratory; Al-In miscibility concern (dropped by the gate)"},
      {"material": "LiGaO2", "band_gap_eV": 6.3, "mobility_e": 19, "dielectric": 7.6, "bulk_modulus_GPa": 131, "kappa_W_m_K": 11},
      {"material": "NaGaO2", "band_gap_eV": 6.3, "mobility_e": 18, "dielectric": 7.6, "bulk_modulus_GPa": 105, "kappa_W_m_K": 10},
      {"material": "ZnGa2O4", "band_gap_eV": 4.4, "mobility_e": 91, "dielectric": 8.0, "bulk_modulus_GPa": 104, "kappa_W_m_K": 5},
      {"material": "MgGa2O4", "band_gap_eV": 4.4, "mobility_e": 44, "dielectric": 8.2, "bulk_modulus_GPa": 104, "kappa_W_m_K": 6},
      {"material": "beta-Ga2O3", "band_gap_eV": 4.3, "mobility_e": 47, "dielectric": 8.2, "bulk_modulus_GPa": 125, "kappa_W_m_K": 10, "note": "baseline"},
      {"material": "4H-SiC", "band_gap_eV": 3.2, "mobility_e": 30, "dielectric": 6.1, "bulk_modulus_GPa": 409, "kappa_W_m_K": 577, "note": "incumbent"},
      {"material": "GaN", "band_gap_eV": 3.5, "mobility_e": 907, "dielectric": 9.0, "bulk_modulus_GPa": 163, "kappa_W_m_K": 80, "note": "incumbent"}
    ]
  },

  "prior_art": [
    {"candidate": "ZnGa2O4", "status": "actively_pursued", "note": "UWBG energy electronics; demonstrated bipolar self-doping (p-type and n-type epilayers by MOCVD, conductivity controllable over 10 orders of magnitude); crystal growth ongoing.", "refs": [12, 13]},
    {"candidate": "(Al,Ga)2O3", "status": "established_alloy", "note": "alpha-(AlGa)2O3 spans 5.4-8.6 eV by MBE; beta-(AlGa)2O3 grown by MOCVD (up to x~0.52) and bulk Czochralski.", "refs": [2, 19, 20]},
    {"candidate": "MgGa2O4", "status": "emerging", "note": "UWBG + power potential noted; photodetector/phosphor work to date.", "refs": [16]},
    {"candidate": "LiGaO2", "status": "frontier_adjacent", "note": "Known UWBG; Li-Ga-O pursued for power (related LiGa5O8 p-type with beta-Ga2O3).", "refs": [3, 17]},
    {"candidate": "NaGaO2", "status": "under_explored", "note": "No power-electronics literature located.", "refs": []}
  ],

  "references": {
    "1": "Recent Advanced UWBG beta-Ga2O3 Material, Adv. Electron. Mater. (2023), doi:10.1002/aelm.202300844",
    "2": "Crystal-orientation-dictated epitaxy of UWBG 5.4-8.6 eV alpha-(AlGa)2O3 on m-plane sapphire, Sci. Adv. (2021), doi:10.1126/sciadv.abd5891",
    "3": "Optical properties of lithium gallium oxide, Appl. Surf. Sci. (2017), sciencedirect S0169433217300995",
    "4": "Optical band-gap values of spinel AB2O4 (A=Mg,Zn; B=Al,Ga), ACS Org. Inorg. Au (2023), doi:10.1021/acsorginorgau.3c00030",
    "5": "The optical energy gap of scandium oxide, OSTI 4014535",
    "6": "Structure/electronic/optical properties of mullite-type Bi2M4O9 (M=Al,Ga), Inorg. Chem. (2016), doi:10.1021/acs.inorgchem.6b00330",
    "7": "QSGW band structures of LiGaO2 and NaGaO2, arXiv 2010.15934",
    "8": "Klimm et al., Thermal conductivity tensor of beta-Ga2O3 (300-1275 K), Cryst. Res. Technol. (2023), doi:10.1002/crat.202200204",
    "9": "Thermal conductivity of B2O3 glass under pressure (0.52 W/m/K ambient), Int. J. Thermophys., doi:10.1007/BF00522148",
    "10": "Czochralski growth of a mixed cubic Lu2O3-Sc2O3-Y2O3 crystal, including kappa=4.1 W/m/K for one Er-doped mixed composition, IUCr (2021), doi:10.1107/S2052520621005321. Not used as a pure-endpoint conductivity.",
    "11": "Thermal conductivity of bulk In2O3 single crystals, arXiv 2008.13519",
    "12": "Chikoidze et al., p-Type UWBG spinel ZnGa2O4: new perspectives for energy electronics, Cryst. Growth Des. (2020), doi:10.1021/acs.cgd.9b01669",
    "13": "Bipolar self-doping in UWBG spinel ZnGa2O4, Mater. Today Phys. (2021), sciencedirect S2542529321001279",
    "14": "Liu et al., Structural/electronic properties of corundum and monoclinic (Al1-xInx)2O3 alloys, Adv. Theory Simul. (2025), doi:10.1002/adts.202401407  [used for the (Al,In)2O3 first-pass-failure discussion, NOT for the (Al,Ga)2O3 alloy]",
    "15": "Influence of Al content on ternary Al2xIn2-2xO3 alloy films (MOCVD; phase behaviour), sciencedirect S0025540815003219",
    "16": "Beta-to-spinel phase transition of magnesium gallium oxide thin films, ACS Appl. Electron. Mater. (2024), doi:10.1021/acsaelm.4c01079",
    "17": "Ultrawide-bandgap LiGa5O8/beta-Ga2O3 heterojunction p-n diodes, APL Electron. Devices, pubs.aip.org aed/016115",
    "18": "Ga2O3 and related UWBG power semiconductor oxides (review), PMC8838167",
    "19": "MOCVD epitaxy of UWBG beta-(AlxGa1-x)2O3 with high-Al composition on (100) beta-Ga2O3 substrates, Cryst. Growth Des. (2020), doi:10.1021/acs.cgd.0c00864",
    "20": "Alloyed beta-(AlxGa1-x)2O3 bulk Czochralski single crystals and property trends, J. Appl. Phys. 131, 155702 (2022)",
    "21": "Jain et al., The Materials Project: a materials genome approach to accelerating materials innovation, APL Mater. (2013), doi:10.1063/1.4812323",
    "22": "Chang et al., CLEASE: a versatile implementation of cluster expansion for disordered materials, J. Phys.: Condens. Matter (2019), doi:10.1088/1361-648X/ab1bbc",
    "23": "Fare et al., A multi-fidelity machine-learning approach to high-throughput materials screening, npj Comput. Mater. (2022), doi:10.1038/s41524-022-00947-9",
    "24": "Merchant et al., Scaling deep learning for materials discovery, Nature (2023), doi:10.1038/s41586-023-06735-9",
    "25": "Ludwig, Discovery of new materials using combinatorial synthesis and high-throughput characterization of thin-film materials libraries, npj Comput. Mater. (2019), doi:10.1038/s41524-019-0205-0",
    "26": "Szymanski et al., An autonomous laboratory for the accelerated synthesis of inorganic materials, Nature (2023), doi:10.1038/s41586-023-06734-w"
  },

  "provenance_note": "Every retained band-gap and thermal-conductivity reference value, and the growth/doping claims for the shortlisted materials, was located from the cited source during this study. Each band-gap row carries its own source_type (measured_primary / measured_review / computed_gw / secondary / tertiary_compilation). Predicted values are forward evaluations through FluxMateria's Universal Materials Engine, powered by proprietary FLUX Theory physics, with zero parameters fit to the target property. Rankings dependent on unsupported Sc2O3/Y2O3/Lu2O3 thermal inputs are excluded from this public artifact."
}