DFT · Machine-Learned Interatomic Potentials · Reaction Kinetics · Hands-on UHV STM/MBE
📍 Nancy, France
Profile · Currently · Skills · Projects · Publications · Education
Ph.D. in Materials Science and Solid-State Physics (Université de Lorraine / CNRS – Institut Jean Lamour). I bridge hands-on ultra-high vacuum (UHV) surface physics with first-principles electronic structure and equivariant machine-learning interatomic potentials, from the STM tip to the supercomputer.
- ⚛️ Building autonomous active-learning pipelines (MACE + jobflow) that train MLIPs to DFT accuracy and extract kinetics (CI-NEB, BEP scaling) for single-atom catalysts.
- 💥 Running statistical molecular dynamics of Ar bombardment of graphene (ASE + LAMMPS, 514 impacts, convergence-tested) to quantify vacancy yields and self-healing; N impacts next, with a MACE potential.
- 🧪 Studying H₂ dissociation on N-doped graphene-supported transition-metal single-atom catalysts.
|
⚛️ Electronic Structure & DFT Periodic plane-wave and grid-based DFT (VASP, Quantum ESPRESSO, GPAW), spin polarization, adsorption energetics, CI-NEB transition-state search, and electronic descriptors (Bader charges, d-band center, COHP/ICOHP via LOBSTER). |
🤖 Machine Learning & MLIPs Training and fine-tuning of equivariant MLIPs (MACE, NequIP, SevenNet), on-the-fly active-learning molecular dynamics, compressed-sensing symbolic regression (SISSO), and classical/reactive MD with LAMMPS (AIREBO, ZBL, ReaxFF). |
|
🔭 UHV Surface Science Variable-temperature polar STM (4 K liquid He / ~90 K liquid N₂), thin-film growth by Molecular Beam Epitaxy (MBE), surface preparation (Ar+ sputtering, annealing, defect engineering), and STM/LDOS image simulation from DFT configurations. |
🖥️ HPC & Automation End-to-end Python workflows (ASE, Pymatgen) for high-throughput screening, SLURM job handling, and convergence checks on Tier-1 supercomputers (GENCI, several million CPU hours). |
|
MACE-ActiveLearning-SAC Autonomous active-learning pipeline (MACE + jobflow): MLIP training to DFT accuracy, automated CI-NEB and BEP scaling for H₂ dissociation on N-doped graphene-supported Pd single-atom catalysts.    
|
Autonomous_Pipeline_SAC Multi-phase active-learning pipeline coupling GPAW (DFT) and MACE (MLIP) for high-throughput screening, CI-NEB kinetics and dynamic stability of Pt–NxCy single-atom catalysts. 
|
|
Graphene-Defect-Engineering-MD Statistical MD (ASE + LAMMPS, AIREBO + ZBL) of Ar bombardment of a thermalized 2048-atom graphene sheet: 514 random impacts at 50–200 eV, automatic outcome classification (single/di-vacancies, self-healing, projectile fate), vacancy yields with 95% intervals, an onset near 63 eV and convergence tests on quench time, window, sheet size and thresholds.   
|
Python_poisson High-precision numerical integration of 1D Poisson equations using the Numerov method. 
|
- S. Ziat, F. Brix, A. Tsaturyan, B. Kierren, É. Gaudry, "How N-Doping Promotes Hydrogen Dissociation at Graphene-Based Single-Atom Catalysts", J. Phys. Chem. Lett., 2026.
doi:10.1021/acs.jpclett.5c03805 - Théo Bequet, Florian Brix, Safouan Ziat, Corentin Martinez, Laurent Piccolo, et al., "The Tradeoff behind Optimal 3-Fold (C,N)-Coordinated Single-Atom Catalysts", Nano Letters, 2026.
doi:10.1021/acs.nanolett.6c02777 - U. Khan, J. A. Okolie, S. Ziat, "Activated carbon as catalyst support and electrocatalyst for industrial chemical reactions", in Activated Carbon: Progress and Applications, Ch. 9, Elsevier, 2025.
doi:10.1016/B978-0-443-13840-9.00009-3
| Period | Degree | Institution |
|---|---|---|
| 2022–2026 | Ph.D. in Materials Science & Solid-State Physics Thesis: DFT Study of H₂ Dissociation on Transition-Metal Single-Atom Catalysts Supported on N-Doped Graphene |
Université de Lorraine, CNRS – Institut Jean Lamour |
| 2021–2022 | M.Sc. in Condensed Matter & Nanophysics (With Honours) | University of Strasbourg, IPCMS |
| 2017–2019 | M.Sc. in Advanced Materials & Renewable Energies (With Honours) | Université Moulay Ismail, Meknès |