SURF 2026 Enthalpic vs. Entropic Drivers of Ion Adsorption at Charged Surfaces

Caltech Summer Undergraduate Research Fellowship (SURF) 2026 Host Lab: Fong Lab, Caltech Student: Defne Nihal Ertugrul

---

1. Scientific Objective

This project uses classical molecular dynamics (MD) to disentangle the enthalpic (ΔH) and entropic (ΔS) contributions to the adsorption free energy (ΔG = ΔH − TΔS) of monovalent (Na⁺) and divalent (Ca²⁺) cations at a charged planar electrode in explicit water.

Three primary systems are studied:

System	Electrolyte	Purpose
Monovalent	NaCl only	Baseline for Na⁺ adsorption thermodynamics
Divalent	CaCl₂ only	Baseline for Ca²⁺ adsorption thermodynamics
Mixed	NaCl + CaCl₂	Competitive adsorption between Na⁺ and Ca²⁺

Each system is simulated at a baseline temperature (T₀ ≈ 298 K) and an elevated temperature (T₀ + 40 K) so that the temperature dependence of the adsorption free energy can be used to decompose ΔG into ΔH and ΔS via a finite difference van 't Hoff analysis.

---

2. Repository Organization

SURF2026/
├── README.md
├── .gitignore
├── environment.yml              # Conda environment for Python analysis
│
├── setup/                       # System construction (Weeks 1–2)
│   ├── packmol/                 # Packmol input files (.inp)
│   ├── moltemplate/             # Moltemplate .lt topology files
│   └── forcefields/             # SPC/E water, Joung–Cheatham ions, etc.
│
├── simulations/                 # LAMMPS input decks (Weeks 3–6)
│   ├── monovalent/              # NaCl-only systems
│   ├── divalent/                # CaCl2-only systems
│   └── mixed/                   # NaCl + CaCl2 competitive systems
│
├── hpc_scripts/                 # Purdue Anvil Slurm job scripts
│   └── submit_anvil.slurm
│
├── analysis/                    # Post-processing (Weeks 6–9)
│   ├── scripts/
│   │   └── analyze_hydration.py # RDF + coordination-number skeleton
│   ├── notebooks/               # Jupyter notebooks for exploratory analysis
│   └── results/                 # Plots, CSVs (small files only)
│
├── visualization/               # Weeks 7–9
│   ├── vmd/                     # .tcl scripts: PBC wrap, representations
│   └── ovito/                   # OVITO Python modifiers
│
├── data_management/             # Automated Anvil ↔ macOS backups
│   ├── rclone_backup.sh
│   └── com.defne.surf.backup.plist   # macOS launchd agent
│
└── docs/                        # Weekly reports, figures, final poster

The directory layout mirrors the 10 week SURF workflow:

Weeks	Stage	Directories
1/2	Literature + system building	setup/, docs/
3/4	Equilibration + validation	simulations/, hpc_scripts/
5/6	Production runs (2 T, 3 sys)	simulations/, hpc_scripts/
6/8	Structural analysis	analysis/, visualization/
8/9	Thermodynamic decomposition	analysis/notebooks/
10	Poster + final report	docs/

---

3. Installation & Environments

3.1 Python analysis environment

conda env create -f environment.yml
conda activate surf2026

3.2 LAMMPS on Purdue Anvil

module load gcc/11 openmpi/4 lammps/2Aug2023

See hpc_scripts/submit_anvil.slurm for a ready to edit Slurm template.

3.3 System building tools (local)

• Packmol
• Moltemplate
• VMD
• OVITO

---

4. Simulation Workflow

1. Build the simulation box with Packmol (water + electrodes + ions) and assign force field parameters with Moltemplate.
2. Minimize and equilibrate in LAMMPS: NVT → NPT → NVT with the charged walls held rigid.
3. Production NVT runs at T₀ and T₀ + 40 K for each of the three electrolyte compositions.
4. Sync trajectories from Anvil to local storage nightly via data_management/rclone_backup.sh (scheduled by launchd).
5. Analyze with analysis/scripts/analyze_hydration.py using MDAnalysis: ion density profiles, RDFs, hydration shell coordination numbers, and ion clustering statistics.
6. Decompose ΔG(T) → ΔH, ΔS via two temperature finite difference.
7. Visualize in VMD (custom .tcl PBC wrap scripts) and OVITO.

---

5. Reproducibility Notes

• LAMMPS input decks are version controlled; trajectory outputs are not (see .gitignore). Raw trajectories live on Anvil scratch and are mirrored via rclone.
• Random seeds for velocity create and fix langevin are recorded in each input deck.
• Module and compiler versions are logged at the top of every Slurm job via module list.

---

6. Acknowledgments

Supported by the Caltech SURF program and the Fong Lab. Computations performed on Purdue Anvil via ACCESS.