Rank perovskites for isothermal solar thermochemical water splitting

Overview

1. 1

   Retrieve four perovskite prototypes

   Step 1 / 6

   Pull representative cubic, rhombohedral, tetragonal, and orthorhombic ABO₃ structures from Materials Project with symmetry and provenance intact.

   Protocol

   1. aQuery space groups 221, 167, 99, and 62 through mp-api.
   2. bStore the source material ids, conventional cells, symmetry labels, and retrieval metadata.
   3. cValidate stoichiometry and crystallographic site assignments before decoration.

   Expected output

   Four provenance-tracked ABO₃ prototype structures with validated space groups and site labels.

   Simulations · click to test

   Materials Project API

   output carries into step 2
2. 2

   Build the 20 candidate structures

   Step 2 / 6

   Decorate each prototype with La on the A site and Cr, Mn, Fe, Co, or Ni on the B site, preserving the intended symmetry.

   Protocol

   1. aGenerate LaBO₃ for B ∈ {Cr, Mn, Fe, Co, Ni} in every prototype.
   2. bStandardize cells and validate oxidation-state, composition, and site occupancy.
   3. cReject duplicate or symmetry-collapsed inputs before relaxation.

   Expected output

   A manifest of 20 unique, symmetry-validated LaBO₃ starting structures.

   Simulations · click to test

   ASE + pymatgen

   output carries into step 3
3. 3

   Relax pristine structures with fixed symmetry

   Step 3 / 6

   Run UMA-S-1p2 structural relaxations while constraining each candidate to its submitted polymorph.

   Protocol

   1. aApply consistent calculator, optimizer, force, stress, and convergence settings.
   2. bRelax lattice and atomic degrees of freedom without leaving the target space group.
   3. cRecord final energy, volume, forces, stress, symmetry, and convergence status.

   Expected output

   Converged pristine geometries and comparable 0 K total energies for all valid candidates.

   Simulations · click to test

   ASE + pymatgenUMA-S-1p2

   output carries into step 4
4. 4

   Select each composition's 0 K polymorph

   Step 4 / 6

   Compare the relaxed polymorph energies on a consistent per-formula-unit basis and choose the lowest-energy structure for each B-site element.

   Protocol

   1. aNormalize energies by LaBO₃ formula unit.
   2. bCheck near-degenerate structures and any relaxation or symmetry failures.
   3. cCarry one documented ground-state structure per composition into defect calculations.

   Expected output

   Five selected ground-state polymorphs with energy differences to the three alternatives.

   Simulations · click to test

   UMA-S-1p2

   output carries into step 5
5. 5

   Extrapolate dilute neutral vacancy energies

   Step 5 / 6

   Enumerate symmetry-distinct single oxygen vacancies over several supercell sizes, relax atoms at fixed cell, and extrapolate formation energies to the dilute limit.

   Protocol

   1. aConstruct a convergent series of supercells for each selected polymorph.
   2. bEnumerate every symmetry-unique neutral single-oxygen-vacancy site.
   3. cRelax atomic positions with UMA-S-1p2 while keeping each supercell fixed.
   4. dCalculate vacancy formation energies with the OMat24 O₂ reference and overbinding correction.
   5. eFit vacancy formation energy against 1 / number of atoms and take the y-intercept as the dilute-limit value.

   Expected output

   Site-resolved finite-size series, extrapolation diagnostics, and one dilute neutral vacancy formation energy per composition.

   Simulations · click to test

   ASE + pymatgenUMA-S-1p2Defect Thermodynamics Analysis

   output carries into step 6
6. 6

   Predict H₂ yield and rank candidates

   Step 6 / 6

   Use the dilute vacancy energies with ideal configurational entropy to calculate the reduction-to-oxidation nonstoichiometry swing and convert it to mass-normalized H₂ yield.

   Protocol

   1. aSolve equilibrium vacancy fractions at 1400 °C and 10⁻⁵ bar for reduction.
   2. bSolve the corresponding state at 1400 °C and 10⁻⁴ bar for oxidation.
   3. cCalculate Δδ = δred − δox and convert it to μmol H₂ per gram of oxide.
   4. dPropagate energetic and fit uncertainty, then rank the five compositions from lowest to highest yield.

   Expected output

   A five-material table of dilute vacancy energies, δred, δox, Δδ, H₂ yield, uncertainty, and final rank.

   Simulations · click to test

   Defect Thermodynamics Analysis

Evaluation criteria

• Coverage: all 20 composition–prototype combinations are generated, relaxed with fixed symmetry, and included in the ground-state comparison.
• Dilute neutral oxygen-vacancy formation energy: absolute error |Ecalc − Eref| is evaluated against reviewer-approved references after finite-size extrapolation.
• H₂ yield: absolute error |Ycalc − Yref| is evaluated for every composition under the specified 1400 °C and 10⁻⁵↔10⁻⁴ bar cycle.
• Ranking: agreement with the reference ordering is scored with Kendall's τ, including a documented policy for statistically indistinguishable ties.
• Reproducibility: structures, vacancy-site enumeration, relaxation settings, oxygen reference, correction, and extrapolation fits are preserved as auditable artifacts.