High-throughput inverse design of a stable topological chalcogenide: The mixed-mass decoupling principle and the priority candidate Ta3PbS6

Computational Materials Data Package: High-throughput inverse design of a stable topological chalcogenide: The mixed-mass decoupling principle and the priority candidate Ta3PbS6 Associated Manuscript "High-throughput inverse design of a stable topological chalcogenide: The mixed-mass decoupling principle and the priority candidate Ta3PbS6" Generation Date: 2026-05-29 Version: v1. 0. 0 Overview This repository contains the complete open-source data package, high-performance computing (HPC) input files, and reproducible workflow for the inverse design of non-ergodic topological quantum materials. By unifying algebraic modular superselection symmetries based on the Z/6Z ring with contemporary materials informatics, this pipeline screens and validates candidates capable of evading the Eigenstate Thermalization Hypothesis (ETH) via a protected Liouvillian gap. The core discovery engine implements a Mixed-Mass Decoupling criterion to solve a fundamental structural paradox in quantum design: heavy elements needed for a strong Spin-Orbit Coupling (SOC) typically collapse the phonon spectrum, leading to severe thermal decoherence. By juxtaposing heavy transition metals (Ta, Pb) with a rigid, light chalcogen sub-lattice (S), our framework forces a massive acoustic-optical phonon desynchronization. This pipeline isolated the ternary chalcogenide Ta₃PbS₆ (mp-20784), a dynamically stable metal resting on the thermodynamic convex hull (E₇ₔ₋₋ = 0\, eV/atom) whose experimental synthesizability is historically validated by its cataloging in the Inorganic Crystal Structure Database (ICSD #83037 and #74693). Repository Structure 1. Main Manuscript & Documentation InversedesingTa3PbS6. pdf: The complete compiled research paper with all high-resolution figures embedded, detailed theoretical models, and comprehensive physical discussions. 2. Reproducible Simulation Workflow notebookcolab. ipynb: An interactive Jupyter Notebook fully optimized for Google Colab. It executes the entire 4-pillar materials data mining workflow, from API data retrieval and Machine Learning Interatomic Potential (MLIP via CHGNet) dynamic screening to electronic structure featurization, proxy ARPES rendering, and automatic HPC control script generation. 3. High-Performance Computing (HPC) Input Bundles To ensure complete transparency and enable the community to compute the exact 3D topological invariants, we provide ready-to-run file bundles for the two main ab initio suites used in the community: z2packₕpcᵢnputs/ (Commercial VASP Workflow): POSCAR: Optimized crystal structure coordinates downloaded from Materials Project for entry mp-20784. INCAR: Input parameters tailored for accurate non-collinear calculations with explicit Spin-Orbit Coupling (LSORBIT =. TRUE. ) and tight electronic convergence criteria (EDIFF = 1E-8). KPOINTS: Dense Monkhorst-Pack mesh grid necessary to accurately resolve multi-band metallic Fermi surfaces. ta3pbs6. win: Configuration input for Wannier90 specifying projection matrices for the target manifold (Ta-d, Pb-p, S-p orbitals). runᵦ2pack. py: Python automation script executing the inner-loop VASP-Wannier interface to track hybrid Wannier charge centers. README. txt: Operational guide for execution on a cluster. z2packqeᵢnputs/ (100% Open-Source Quantum ESPRESSO Workflow): scf. in: Ground-state self-consistent field calculation including fully relativistic pseudopotentials, smearing, and non-collinear spin-orbit parameters. nscf. in: Non-self-consistent grid evaluation enforcing nosym =. true. to guarantee complete compatibility with the Wannier90 gauge translation. pw2wan. in: Interface control file to compute the overlap matrices (. mmn and. amn) linking Bloch states to local functions. ta3pbs6. win: Wannier90 inputs explicitly incorporating the real-space Hamiltonian flag (writeₕr = true) and strict band disentanglement parameters required for a crowded electron manifold. runᵦ2pack. py: An optimized, highly efficient topological script that utilizes the z2pack. hm. System class to extract invariants from the precomputed tight-binding Wannier Hamiltonian (ₕr. dat), avoiding expensive recursive ab initio inner loops. README. txt: Step-by-step documentation for deployment using purely open-source supercomputing stacks. 4. Publication-Grade Graphics PhononDOSₒfTaSe2CHGNetᵥalidation. png: Methodological sanity check demonstrating that the CHGNet MLIP accurately reproduces the acoustic-optical gap (1. 68\, meV) of the control material 2H-TaSe₂ in full agreement with experimental Inelastic X-ray Scattering (IXS) and DFPT literature. ProxyARPESBSTa3PbS6. png: High-resolution computational ARPES band map confirming a poly-electronic metallic ground state crossing the Fermi level (6 active bands, total DOS of 2. 974\, states/eV at EF). Current Status of Topological Invariants In this v1. 0. 0 release, the topological validation is established by a robust electronic parity proxy showing an exact band-occupancy inversion (n = 2) along the -A reciprocal line (80 occupied bands at versus 78 at A). The rigorous calculation of the 3D Z₂ index using the provided HPC inputs is estimated to consume between 500 and 1000 core-hours on an external supercomputing cluster. Fysicists and experimentalists are encouraged to use these open bundles to execute the full calculations and synthesize this highly promising candidate. License & Metadata Data and Inputs: Creative Commons Attribution 4. 0 International (CC-BY-4. 0) Code and Automation Scripts: MIT License Status: Submitted to *Computational Materials Science* (Elsevier).