{"paper":{"title":"Beyond Diamond: Interpretable Machine Learning Reveals Design Principles for Quantum Defect Host Materials","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"Machine learning on compositions alone extracts consensus design rules to identify 122 high-confidence quantum defect host candidates.","cross_cats":[],"primary_cat":"cond-mat.mtrl-sci","authors_text":"Mohammed Mahshook, Rudra Banerjee","submitted_at":"2025-06-04T11:21:17Z","abstract_excerpt":"Solid-state spin defects in wide-bandgap semiconductors are leading candidates for quantum information processing, but systematic identification of suitable host materials remains limited by the cost of first-principles screening across vast chemical spaces. We address this with a composition-only machine learning framework built on heterogeneous Rashomon set ensembles: by contrasting the feature attributions of seven diverse classifiers, we extract consensus design rules that no single model identifies alone-filled valence s-, d-, and f-shells, low chemical heterogeneity, and enrichment in C,"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"By contrasting feature attributions across seven diverse classifiers in a heterogeneous Rashomon set, the framework extracts consensus design rules (filled valence s-, d-, and f-shells, low chemical heterogeneity, enrichment in C, S, Si, O) that enable identification of 122 high-confidence candidates from ~45,000 compounds, with DFT validation showing dielectric screening as a coherence proxy (R² = 0.89 against experimental T₂) and favorable mid-gap states in TiO₂.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That consensus feature attributions from composition-only classifiers trained on existing data capture the essential physical requirements for quantum defect hosting and that the dielectric-T₂ correlation observed in 12 materials will generalize to the full set of 122 screened candidates without structural or defect-specific details in the initial filter.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A composition-only ML framework with Rashomon ensembles extracts consensus design rules and screens ~45,000 compounds to identify 122 high-confidence quantum defect hosts, recovering known materials and predicting new ones validated by limited DFT.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Machine learning on compositions alone extracts consensus design rules to identify 122 high-confidence quantum defect host candidates.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"dfce17fa0b73531ee8c083adc957df9c6288cae165123e4549906a368e81eed4"},"source":{"id":"2506.03844","kind":"arxiv","version":3},"verdict":{"id":"2e2b3166-fd0e-4cf8-96bc-3ac6d0c5f5a0","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T11:32:16.143277Z","strongest_claim":"By contrasting feature attributions across seven diverse classifiers in a heterogeneous Rashomon set, the framework extracts consensus design rules (filled valence s-, d-, and f-shells, low chemical heterogeneity, enrichment in C, S, Si, O) that enable identification of 122 high-confidence candidates from ~45,000 compounds, with DFT validation showing dielectric screening as a coherence proxy (R² = 0.89 against experimental T₂) and favorable mid-gap states in TiO₂.","one_line_summary":"A composition-only ML framework with Rashomon ensembles extracts consensus design rules and screens ~45,000 compounds to identify 122 high-confidence quantum defect hosts, recovering known materials and predicting new ones validated by limited DFT.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That consensus feature attributions from composition-only classifiers trained on existing data capture the essential physical requirements for quantum defect hosting and that the dielectric-T₂ correlation observed in 12 materials will generalize to the full set of 122 screened candidates without structural or defect-specific details in the initial filter.","pith_extraction_headline":"Machine learning on compositions alone extracts consensus design rules to identify 122 high-confidence quantum defect host candidates."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2506.03844/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":61,"sample":[{"doi":"","year":2024,"title":"Building a quantum- ready ecosystem","work_id":"18fa45f7-a5a6-4fe8-ac25-d7227b903dfb","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2025,"title":"Review of distributed quantum computing: from sin- gle qpu to high performance quantum computing","work_id":"d7101fbb-4ed0-4b30-89a7-385124cd2c3a","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Scalable entanglement certi- fication via quantum communication","work_id":"2a60c6b4-7f00-416d-a3b3-be2df3179cbb","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"Quantum sensing and metrology for fundamental physics with molecules","work_id":"8a179d38-18f8-4e05-ad35-b7ea3a0fab32","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"An elementary review on basic principles and developments of qubits for quantum computing","work_id":"8d6bcba5-f41b-46f9-a79a-b81b6b9650e8","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":61,"snapshot_sha256":"2e1dc9385e11bbd355343136c7471b1bbd2d3be6471dcc59184899cfb9fcb607","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}