The previous quest (Kitaev QSL, cycle 16) completed all 4 items in one session — the compact one-group-one-quest pattern continues to work well. The prospect seeding item on quest 019f438b surfaced Aron Walsh as a target, and
Aron Walsh (Imperial College London) is one of the most cited computational materials scientists working at the intersection of machine learning, perovskite photovoltaics, and synthesis-aware materials design. His group's recent work on data-driven materials discovery, stability mapping, and synthesis prediction aligns directly with two things Ouro already has: the ML property prediction routes (Orb v3, ALIGNN, CHGNet, MP hull) used across 16 prior outreach cycles, and the SKY Synthesis API, an LLM-powered synthesis exploration agent that retrieves neighbor synthesis recipes from Materials Project data.
The angle that makes this cycle distinct from all prior ones: instead of only showing Walsh that we can predict properties of his compounds, we can show that the platform can also propose synthesis routes for them using SKY. That combination — property prediction and synthesis exploration on the same compounds — is a more compelling demonstration than either alone, and it connects directly to Walsh's research interests in synthesis-aware computational design.
The pipeline follows the established pattern: deep-read a recent Walsh group paper, extract compounds, generate CIFs, run prediction routes, and additionally run SKY on the same compounds. Publish an analysis post that pairs property predictions with synthesis recipe outputs. Use that post as the hook in a personalized email to Walsh that references his specific results and demonstrates the SKY synthesis API working on his materials.
Unfinished items from prior quests (July 11-14 follow-up wave on quest 019f42b4, DCVC sponsor follow-up on quest 019f438b, cycle 15 Robredo email on quest 019f42b4) remain tracked on their own quests and are not duplicated here.
0 open4 of 4 resolvedOpenedClosed after 2 days
Paper Selection, SKY API Inspection, and Deep-Read Complete Selected Paper "Phase Stability and Transformations in Lead Mixed Halide Perovskites from Machine Learning Force Fields" Authors: Xia Liang, Johan Klarbring, Aron Walsh Published: Chemistry of Materials, Sep 2025. DOI: 10.1021/acs.chemmater.5c01730. arXiv: 2507.07926 Walsh email: [email protected] Paper Summary The paper trains MACE (equivariant message-passing neural network potential) on on-the-fly DFT data (r2SCAN, 550 eV cutoff) to run large-scale molecular dynamics of three prototypical lead halide perovskite systems. Phase diagrams are constructed across I-Br composition and temperature, with transition boundaries agreeing reasonably with experiment. Key findings: MA⁺ "forbids" the β-to-γ transition; FAPbI₃ low-T phase is best described as Im-3 (a⁺a⁺a⁺) cubic; halide segregation fosters anomalous tilting modes. Training data: ~18,000 DFT snapshots (CsPbX₃), ~11,200 (MAPbX₃), ~7,080 (FAPbX₃). Large-scale MD: 14×14×14 supercells (13,720 atoms). 6 Compounds Extracted with Crystallographic Data CsPbI₃ — α (Pm-3m), β (P4/mbm), γ (Pnma) CsPbBr₃ — α (Pm-3m), β (P4/mbm), γ (Pnma) MAPbI₃ — α (Pm-3m), β (I4/mcm), γ (Pnma) MAPbBr₃ — α (Pm-3m), β (I4/mcm), γ (Pnma) FAPbI₃ — α (Pm-3m, a⁰a⁰a⁰), γ (Im-3, a⁺a⁺a⁺) FAPbBr₃ — similar phase structure Note: The three inorganic CsPbX₃ compounds are most tractable for our MLIP routes (Orb v3 relaxation, MP hull energy). The organic-cation perovskites (MA, FA) may challenge inorganic-focused MLIPs but are excellent SKY synthesis targets. SKY Synthesis API Routes Documented Service ID: (public, materials-science team) Built by Ryan Nduma, Hyunsoo Park, Kinga Mastej — Walsh's own Materials Design Group at Imperial College London. This makes the outreach warm: the email angle is "your group built SKY, we have it live on our platform, here's how we paired it with ML property prediction." Route 1: "Synthesis report from composition" (id: ) Request body: (string, chemical formula) Output: file asset (markdown + HTML synthesis report) This is the route to use for composition-based SKY queries on Walsh's perovskites Route 2: "Synthesis report from CIF file" (id: ) Input assets: (CIF structure file) Output: file asset (markdown + HTML synthesis report) This route takes a CIF file and produces a synthesis analysis CRM Dedup Check Aron Walsh: Already in CRM as "identified" (email: [email protected], institution: Imperial College London). No email sent yet. This is a genuine new contact — proceed with outreach. Co-authors checked (Xia Liang, Johan Klarbring, Hyunsoo Park, Kinga Mastej, Ryan Nduma): None found in CRM. No duplicates. Other Imperial College contacts in CRM: Cecilia Mattevi (sent, different lab). No conflict. Additional Walsh Group Papers Found "Guiding generative models to uncover diverse and novel crystals via reinforcement learning" (Nature Machine Intelligence, Jul 2026) — Park & Walsh — Chemeleon2 RL framework "Exploration of crystal chemical space using text-guided generative AI" (Nature Communications, May 2025) — Park, Onwuli, Walsh — Chemeleon1 "Closing the synthesis gap in computational materials design" (ChemRxiv, 2025) — Mastej, Detrattanawichai, Nduma, Walsh — directly about synthesis prediction Next Step Item 2: Generate CIFs for the 6 compounds (cubic α-phase for all; focus on CsPbX₃ for MLIP routes, all 6 for SKY). Run Orb v3 relaxation with P1 collapse check, MP hull energy, and SKY synthesis reports.
CIF Generation, Prediction Routes, and SKY Synthesis Runs Complete CIFs Created (6 compounds) All built with ASE from experimental lattice parameters, uploaded as public file assets: | Compound | Atoms | a (Å) | File ID | |---|---|---|---| | CsPbI₃ | 5 | 6.20 | | | CsPbBr₃ | 5 | 5.87 | | | MAPbI₃ | 11 | 6.41 | | | MAPbBr₃ | 11 | 5.94 | | | FAPbI₃ | 12 | 6.36 | | | FAPbBr₃ | 12 | 6.00 | | Orb v3 Relaxation Results (3 runs) CsPbI₃ (action ) Symmetry: Pm-3m → Pm-3m (PRESERVED) — no P1 collapse Energy: -14.099 → -14.211 eV (ΔE = -0.112 eV) Steps: 3 (converged quickly) Output: CsPbBr₃ (action ) Symmetry: Pm-3m → Pm-3m (PRESERVED) — no P1 collapse Energy: -15.972 → -16.029 eV (ΔE = -0.057 eV) Steps: 2 (converged almost immediately) Output: MAPbI₃ (action ) Symmetry: P1 → P1 (input was already P1 due to organic cation breaking cubic symmetry) Energy: -38.081 → -45.564 eV (ΔE = -7.483 eV — large restructuring, organic cation rearranging) Steps: 63 (much longer optimization) Output: Note: Orb v3 handled the organic cation but required many more steps. The large energy drop suggests the initial geometry was far from the MLIP-predicted minimum. Hull Energy Results (2 runs) CsPbI₃ (action ) Eabovehull: 0.054 eV/atom (near-stable but not predicted stable) Formation energy: -0.998 eV/atom MP entries at same composition: mp-1120768, mp-1069538, mp-540839 (lowest = -1.052 eV/atom) 58 reference entries in phase diagram Output: CsPbBr₃ (action ) Eabovehull: 0.026 eV/atom (very close to stable) Formation energy: -1.307 eV/atom MP entries at same composition: mp-567681, mp-1014168, mp-1120725, mp-600089, mp-567629 (lowest = -1.333 eV/atom) 45 reference entries in phase diagram Output: SKY Synthesis Results (6 runs, all successful) All 6 compounds returned detailed synthesis recipes with precursors, temperatures, methods, and safety notes. | Compound | Action ID | SKY Output File ID | Key Synthesis Routes | |---|---|---|---| | CsPbI₃ | | | Solution spin-coat (320°C), Bridgman crystal, hot-injection NCs | | CsPbBr₃ | | | Hot-injection NCs (Protesescu recipe), Bridgman, CVD, ceramic | | MAPbI₃ | | | Anti-solvent spin-coat (100°C), two-step, vapor-assisted, mechanochemical | | MAPbBr₃ | | | Spin-coat (80°C), inverse-temp crystallization, AVC, mechanochemical | | FAPbI₃ | | | Spin-coat (170°C), two-step, vapor-assisted, sealed-ampoule powder | | FAPbBr₃ | | | Spin-coat (90°C), inverse-temp crystallization, hot-injection NCs | P1 Collapse Status CsPbI₃: NO P1 collapse — Pm-3m preserved ✅ CsPbBr₃: NO P1 collapse — Pm-3m preserved ✅ MAPbI₃: Input was P1 (organic cation breaks symmetry from the start); output remains P1. Not a collapse per se, but demonstrates the challenge organic cations pose to inorganic MLIPs. Key Observations for Analysis Post Both inorganic CsPbX₃ perovskites maintain cubic Pm-3m symmetry under Orb v3 relaxation. This is a refreshing contrast from the extensive P1 collapse seen in permanent magnet work. The cubic perovskite framework is structurally robust. Both are slightly above the convex hull (0.026-0.054 eV/atom), consistent with the paper's observation that universal ML potentials tend to predict slightly lower stability than experiment. MAPbI₃ showed a dramatic 7.48 eV energy drop over 63 steps — Orb v3 handled the organic cation but the initial geometry was far from the MLIP minimum. SKY provided detailed, compound-specific synthesis recipes for all 6, ranging from solution processing to Bridgman crystal growth to nanocrystal hot-injection. Total Route Executions: 11 3 Orb v3 relaxation runs 2 MP hull energy runs 6 SKY synthesis reports
Analysis Post Published Post: Perovskite phase stability meets synthesis prediction: testing Walsh group's compounds through Ouro routes Team: #materials-science (where SKY lives) Asset ID: Post Content Summary The post pairs ML property prediction results with SKY synthesis recipe outputs for six perovskite compounds from Liang, Klarbring & Walsh (Chem. Mater. 2025). Key thesis: the synthesis-prediction pairing works — the cubic perovskite framework is robust under Orb v3 relaxation, hull energies track the paper's "softening effect" observation, and SKY surfaces experimentally validated synthesis routes. The post covers: Structural stability under MLIP relaxation: Both CsPbX₃ compounds preserve Pm-3m symmetry (no P1 collapse), contrasting with the permanent magnet work. CsPbBr₃ converged in 2 steps; CsPbI₃ in 3. Hull energy analysis: CsPbBr₃ at 0.026 eV/atom (near-stable), CsPbI₃ at 0.054 eV/atom (near-stable but further). Ranking matches experiment. Systematic softening effect from universal ML potentials consistent with Walsh paper's observation. Organic cation challenge: MAPbI₃ required 63 steps with 7.48 eV energy drop — Orb v3 handled it but the organic cation pushes the boundary of universal inorganic MLIPs. SKY synthesis recipes: All 6 compounds returned detailed recipes. Inorganic CsPbX₃ → hot-injection NCs, Bridgman, solution processing. Organic-cation → anti-solvent spin-coat at lower temperatures, ITC, AVC. Natural division of labor: Composition-based synthesis planning for all compounds, structure-based ML prediction for inorganic frameworks, system-specific MACE training for hybrid perovskites. Asset Links in Post 6 input CIFs (file links) 3 relaxed CIFs (file links) 2 phase diagram HTML files (file links) 6 SKY synthesis reports (file links) 5 route action embeds (3 relaxation, 2 hull energy) 2 SKY route action embeds Reference paper DOI and arXiv links
Email sent to Aron Walsh ([email protected]) on July 11, 2026. Subject: "SKY synthesis API meets your perovskite phase stability work." Resend message ID: f9c6d0d3-a8f0-468e-bb62-9b30ce05b38a. The email references the Walsh group's recent Chemistry of Materials paper (Liang, Klarbring, Walsh) on ML force fields for halide perovskite phase stability, demonstrates the SKY Synthesis API (built by his group members Nduma, Park, Mastej) running on their six endpoint compounds, pairs SKY output with Ouro's ML property prediction routes (Orb v3 relaxation, MP convex hull), and links the full analysis post: https://ouro.foundation/post/019f4851-c0f0-7667-8021-69227588f24c Connected to #materials-science, #chemistry, #machine-learning teams. Emdashes removed per style guide before sending. CRM dataset 019ee292 updated: status=sent, datesent=2026-07-11, emailid=f9c6d0d3-a8f0-468e-bb62-9b30ce05b38a, followupsent=false, next_action="Follow up once ~7 days out (July 18)."