Computational Study of Low Interlayer Friction in Ti_{n+1}C_n (n = 1, 2, and 3) MXene
Summary¶
The study combines PBE + vdw-DF2 DFT for energy paths and LAMMPS ReaxFF simulations using a Ti_{n+1}C_nT_x-compatible reactive potential (T = O, etc.). MD uses Langevin thermostats at 10 K and 298 K with 0.1 fs integration. Friction coefficients from static loading analyses fall near 0.24–0.27 for normal loads under 1.2 GPa for both DFT and ReaxFF. Ti sublayer and terminal O vacancies increase friction but leave μ < 0.31. Hydroxyl and methoxy surface groups on Ti3C2 reduce μ to roughly 0.10–0.14. The work targets solid lubrication and MXene flake contacts where interlayer shear dominates macroscopic friction.
Methods¶
Static QM (DFT). VASP with PAW pseudopotentials; PBE + vdw-DF2; spin-polarized initialization as stated; k-mesh ~1000 points per atom; 520 eV plane-wave cutoff; bilayer cells with 10 Å vacuum separating the stack from its periodic image along c (c fixed while relaxing atomic positions, per §2); Hellmann–Feynman forces < 1 meV/Å and in-plane stresses < 0.1 GPa convergence targets; MEP characterization of interlayer sliding.
MD application (ReaxFF). LAMMPS integrates Ti_{n+1}C_nT_x ReaxFF (T = O, F, OH) for the same bilayer chemistries as DFT—Ti₂CO₂, Ti₃C₂O₂, Ti₄C₃O₂, plus Ti/O vacancy models and Ti₃C₂ with −O, −OH, or −OCH₃ terminations. Setup (§2): conjugate-gradient relaxation of in-plane (x,y) cell parameters for Ti_{n+1}C_nO_2 stacks; Langevin thermostats at 10 K and 298 K; timestep 0.1 fs; PBC in-plane consistent with the DFT cells. Friction readout comes from static loading / constrained sliding pathways rather than a dedicated high-shear-rate production campaign in the indexed Methods summary.
Force-field training is N/A (published MXene parameters).
Equilibration/production durations, NPT usage, barostat settings, electric bias, and metadynamics (if any) beyond the static friction workflow are not captured in the short indexed Methods excerpt—confirm in the full PDF/SI.
Findings¶
Sliding pathways show low barriers for n = 1–3 O-terminated stacks. Friction coefficients ~0.24–0.27 for normal loads < 1.2 GPa agree between DFT and ReaxFF (Ti₂CO₂, Ti₃C₂O₂, Ti₄C₃O₂ table in the paper). Vacancies raise μ through roughness and added adhesion but keep μ < ~0.31. −OH and −OCH₃ on Ti₃C₂ reduce μ to about 0.10–0.14 vs −O. Authors use the DFT/ReaxFF match on μ to argue ReaxFF is suitable for larger/longer tribology sampling where DFT MEP work is costly. Comparisons: side-by-side DFT vs ReaxFF μ tables; vacancy vs pristine; −OH/−OCH₃ vs −O. Sensitivity: normal load (GPa), defect type, and termination chemistry. Limitations / outlook: authors note static friction analysis and MXene chemistry transferability limits; humidity/intercalants may alter adhesion vs vacuum bilayer models. ## Limitations
Static friction estimates and limited MD temperatures may not capture thermally activated slip at all operating conditions; force field transferability to other MXene chemistries is not claimed. Humidity and intercalated solvent could modify adhesion relative to vacuum bilayer models.
Relevance to group¶
Joint DFT + ReaxFF benchmark on MXene friction with van Duin involvement—useful reference for 2D tribology parameter choices.
Reader notes (navigation)¶
- Alternate corpus path (proof PDF): 2017difan-venue-research. Prefer the AMI VOR PDF cited in
pdf_pathwhen reconciling table entries for μ.