# Bound Rubber as a Transferable Structural Descriptor: Connecting MD-Derived Interfacial Scaling to Continuum Reinforcement Models

## TL;DR

This paper introduces a structural descriptor from molecular dynamics simulations to better connect microscopic interfacial structures with macroscopic reinforcement in filled elastomers.

## Contribution

The novel contribution is a transferable structural descriptor derived from MD simulations that improves cross-scale predictions in elastomer reinforcement.

## Key findings

- MD simulations provide a bound-layer scaling relation for chain length N=50 as a structural probe.
- The regime-partitioned bridge model reduces prediction error from 54.1% to 7.3% by incorporating relaxation physics.
- Linear-viscoelastic constraints significantly improve nonlinear PTT calibration, reducing die-swell error by 87%.

## Abstract

Filled elastomers often exhibit a low-frequency power-law storage modulus (G-prime), yet quantitative links between molecular interfacial structure and macroscopic reinforcement remain unresolved. This gap is addressed using a hierarchical multiscale framework that integrates coarse-grained molecular dynamics (MD) and dynamic mechanical analysis (DMA). Overall, MD contributes transferable structural descriptors rather than direct macro-rheology prediction. MD simulations yield a bound-layer scaling relation for chain length N=50 in coarse-grained simulations serving as a structural probe. For EPDM master curves, the single-phase fractional Maxwell model is statistically preferred (Delta AICc > 147, n = 56), reflecting limited statistical power; larger datasets (e.g., PC/ABS, n = 952) favor the dual-phase formulation. For cross-scale prediction, an MD-derived effective-volume-fraction baseline (MAPE = 54.1%) provides a structural prior; the regime-partitioned bridge model absorbs relaxation physics not resolved at the MD scale, reducing error to 7.3% (blocked-CV MAPE = 9.5%, with a 2.3% fold-to-fold spread). Linear-viscoelastic constraints improve nonlinear PTT calibration, reducing die-swell error by 87%.