Research

Exploring the fundamental principles of chemistry through computational and experimental approaches

Research Interests

Inorganic Chemistry

Investigating transition metal complexes, coordination chemistry, and metal-organic frameworks for catalytic applications.

Physical Chemistry

Studying thermodynamics, kinetics, and spectroscopy to understand molecular behavior and reaction mechanisms.

Computational Chemistry

Using quantum mechanical calculations and molecular dynamics simulations to predict molecular properties and reactions.

Materials Chemistry

Designing and synthesizing novel materials with tailored properties for energy storage and conversion applications.

Magnetochemistry

Exploring magnetic properties of lanthanide coordination compounds using ORCA DFT calculations. Focus on dysprosium triangle complexes and anisotropic magnetism for applications in single-molecule magnets.

Quantum Chemistry

Applying quantum mechanical principles to understand electronic structure and chemical bonding at the molecular level.

Research Projects

Three consecutive UROPS under A/P Liviu Ungur (Ungur Group), Department of Chemistry, NUS.

01

Anisotropic Magnetism of a Dy(III) Trinuclear Cluster

UROPS CM2288 · AY2023/24 Sem 2 · Grade: B+

Evaluated the low-lying energy spectrum, magnetic exchange, and anisotropic magnetic response of a Dy3 triangular cluster (C3 point-group symmetry) — a candidate for toroidal magnetic moment behaviour and single-molecule magnetism. Made the full-cluster ab initio treatment tractable through two cost-reduction strategies: (i) replacing two of the three Dy sites with closed-shell Lu(III) to exploit the C3 symmetry, and (ii) a fragmentation scheme that computes the metal centre and the ligand shell separately and merges their molecular orbitals.

Electronic structure (ORCA): CASSCF(9,7) over the Dy 4f shell (21 spin–orbit configurations), scalar-relativistic DKH Hamiltonian with def2-SVP basis, NEVPT2 dynamic correlation, second-order spin–orbit coupling via QDPT, and single-ion anisotropy tensors extracted with SINGLE_ANISO.

Polynuclear response (POLY_ANISO): Heisenberg-type exchange Hamiltonian on pseudospins S* = 1/2 per Dy site; isotropic exchange parameter J scanned over −0.35 to −0.75 cm−1 and fitted to experiment. Best fit J ≈ −0.50 to −0.55 cm−1 (antiferromagnetic) with Ising-limit single-site anisotropy (gx, gy ≈ 0); computed χT(T) (0–300 K) and low-T M(H) at 1.9 K reproduced unpublished experimental data with excellent agreement, supporting a non-colinear (toroidal) arrangement of local moments in the Dy3 plane.

ORCA CASSCF / NEVPT2 QDPT-SOC SINGLE_ANISO POLY_ANISO Lanthanide SMM
02

Molecular Dynamics of an Organic Solvent System at a Cu Surface

UROPS CM3288 / CM3289 · AY2024/25 Sem 2 – AY2025/26 Sem 1 · Grade: A / A

Investigated the behaviour of a benzene/ethanol mixture in contact with Cu(100) and Cu(111) surfaces using classical molecular dynamics in LAMMPS. Compared four framework combinations — OPLS-AA vs TraPPE-UA force fields × PPPM vs MSM electrostatics — under a unified five-stage protocol (soft relaxation → energy minimisation → staged heating → equilibration → production), so that the effect of force-field choice and long-range Coulomb treatment on interfacial adsorption can be analysed independently.

Developed Python pipelines (ASE, NumPy, SciPy, pandas, matplotlib) for trajectory pre- and post-processing, and for automated analysis of interfacial observables (radial distribution functions, density profiles, surface excess, interfacial tension) across replicate simulations. Intermediate results presented as a poster at the NUS Science Summer Institute 2025 Symposium. A self-contained Korean LAMMPS guide derived from this work is published under /guides/lammps/.

LAMMPS Cu(100) · Cu(111) OPLS-AA TraPPE-UA PPPM · MSM Interfacial MD
03

Reactive Machine-Learning Force Field for Iron Oxidation

Final Year Project CM4288 · AY2025/26 Sem 2 – Present

Developing a reactive machine-learning interatomic potential to describe the oxidation of iron, bridging ab initio accuracy with the time- and length-scales accessible to molecular dynamics. Reference data are generated from periodic DFT in Quantum ESPRESSO and used to train an ML force field capable of capturing bond-breaking and oxide-layer growth — the broader goal is to connect electronic-structure predictions to macroscopic iron-oxidation kinetics.

Quantum ESPRESSO Machine-Learning Potential Reactive MD Fe Oxidation