Phase transitions of ordered ice in graphene nanocapillaries and carbon nanotubes
Summary¶
ReaxFF molecular dynamics maps phase behavior of confined water in graphene nanocapillaries and single-walled carbon nanotubes, emphasizing AA-stacked multilayer square ice, first-order vs continuous melting signatures, elevated freezing temperatures in narrow CNTs, and fast proton/hydroxyl transport relative to bulk water. The study is motivated by ongoing debates about structure and melting of water under extreme nanoconfinement, where dimensionality and template chemistry can stabilize ice polymorphs not seen in bulk.
Methods¶
1 — MD application (confined water + ice). Grand-canonical Monte Carlo (GCMC) at fixed liquid-water chemical potential initializes confined structures (SI details). Production runs use ReaxFF molecular dynamics with 0.10 fs timestep and Nosé–Hoover thermostat (10 fs coupling) under the constant-volume protocols described in Sci. Rep. (the article quotes 1 atm target conditions while keeping cell metrics fixed during heating/melting segments—treat as the paper’s documented NVT-like melting workflow rather than inventing a separate barostat here). PBC: three-dimensional PBC for graphene nanocapillaries and CNT confinement models. System sizes / atom counts: follow figure/table construction in the article (SI for exact counts). Temperature: melting scans span 230–400 K continuous regimes discussed for certain square-ice stacks, include 300 K transport analyses for confined 1D/2D ice, and use broader heating windows tabulated in Sci. Rep. figures. Duration: nanosecond-class cumulative sampling is referenced for selected melting scans in the article/SI (use the PDF for exact segment lengths). Barostat during melting sweeps: N/A — not separated as NPT production in the excerpted description. Pressure / stress control: 1 atm referenced for thermostatized melting studies; no external electric field. Enhanced sampling: N/A — aside from GCMC pre-equilibration.
2 — Force-field training. N/A — uses published ReaxFF water/carbon chemistry.
3 — Static QM. N/A — not an AIMD-first study.
Findings¶
Outcomes / mechanisms: ReaxFF reproduces AA-stacked multilayer square ice in graphene nanocapillaries with interlayer H-bonds, contrasting some older AB-stacked classical pictures. Melting signatures differ: bilayer hexagonal ice shows first-order-like jumps, whereas some square-ice stacks show continuous energy evolution over ~230–400 K windows discussed in the text. CNT confinement raises melting temperatures above 273 K, discussed alongside Raman literature constraints. Proton/hydroxyl mobilities can exceed bulk by up to several-fold at 300 K in ordered confinement.
Comparisons: stacking and melting behavior are compared to prior classical models and selected experiments cited in the paper.
Sensitivity: confinement geometry (2D slit vs 1D tube) steers polymorph selection and melting classification.
Limitations: ReaxFF bulk-water errors propagate to phase boundaries; sampling length affects sharp vs gradual melting.
Corpus honesty: quantitative factors follow papers/Raju_Scientific_Reports_2018.pdf; confirm against any updated SI.
Limitations¶
ReaxFF water properties deviate from experiment and ab initio reference data in some bulk limits; phase boundaries remain sensitive to confinement model and sampling length.
Relevance to group¶
Adri van Duin is a co-author; study extends ReaxFF investigations of nanoconfined water and ice on carbon.