Liquid water is a dynamic polydisperse branched polymer
Evidence and attribution¶
Authority of statements
Prose summarizes the PNAS article identified by doi. RexPoN is a Goddard-group water force field distinct from ReaxFF; use this page as a comparative reference for QM-fitted classical water models.
Summary¶
The RexPoN force field for water is parametrized from quantum mechanics without empirical fitting to experimental liquid properties. Molecular dynamics with RexPoN is reported to reproduce bulk thermodynamics and structure, then supports a structural interpretation of hydrogen-bond connectivity in the liquid as a dynamic, polydisperse, branched network. The authors motivate the work as a route to reconcile anomalies in liquid and supercooled water with microscopic connectivity statistics beyond simple tetrahedral pictures (abstract; introduction).
Methods¶
1 — MD application. Molecular dynamics simulates a water system with on the order of hundreds to thousands of atoms in 3D PBC (see PNAS for the exact supercell and H₂O count). NVT/NVE-style control through melting from ice to liquid; thermostat and any NPT barostat / hydrostatic pressure are as in Methods. Time step in fs, equilibration and production ps/ns spans, and temperature K ramps follow the paper (not retyped here at ~298 K checkpoints). External electric field: N/A in the main SHB analysis description. Umbrella / metadynamics: N/A unless stated. NEMD (thermal gradient MD): N/A for the RexPoN bulk liquid connectivity study summarized here.
2 — Force-field training (RexPoN, classical water). RexPoN is a Goddard-group classical force field for water built from QM (no empirical fit to bulk liquid PVT in the same way as many TIP models). Contributors include vdW terms tied to H\(_2\)/O\(_2\) equations of state, electrostatics, and many-body couplings (see PNAS and prior RexPoN references). This is not a ReaxFF reparameterization.
3 — Static QM. QM data underpin RexPoN parameters; there is no separate DFT Results “application” block as the main finding of the paper is classical MD on RexPoN water.
4 — Review / non-sim. N/A.
Findings¶
- Thermodynamic checkpoints at 298 K (density, \(\Delta H_\mathrm{vap}\), entropy, dielectric constant) and melting point lie close to experiment in the reported comparisons.
- Upon melting, SHBs per molecule drop from ice-like values near four to about two in the liquid regime described; lifetimes are on the order of tens to ~90 fs at 298 K in their analysis.
- SHBs link into dynamic branched chains averaging ~150 water molecules at 298 K with branch points carrying three SHBs and terminators one SHB—a picture offered to rationalize long-range angular correlations and anomalies in supercooled water.
- The “polymer” metaphor is explicitly structural (connectivity statistics), not a claim that water is a covalent polymer in the usual chemical sense (discussion framing in the article).
Corpus honesty. For definitive locators and any update to RexPoN text, use the version-of-record PNAS PDF; this page is a curated summary, not a substitute for figure captions and SI.
Limitations¶
Interpretive claims about macroscopic “polymer” behavior are model-dependent; subsequent literature debates alternative explanations—read alongside critical discussions.
RexPoN parameters prioritize bulk water; transfer to interfaces and solutes requires separate validation against QM or experiment for each new chemistry.
Relevance to group¶
Complementary reference water model (RexPoN) outside the ReaxFF lineage; useful contrast for aqueous interface projects parametrized with ReaxFF.