Synergistic Ion-Bridging and Disjoining Pressure in Smart Brine EOR: A Hybrid Molecular Dynamics and Machine Learning Approach

Abstract The synergistic integration of nanotechnology and “smart brine” chemistry represents a complex Frontier for enhanced oil recovery (EOR), where the coupling between ion-bridging kinetics and macroscopic displacement remains a subject of rigorous investigation. This study presents a robust, computationally validated hybrid computational framework integrating molecular dynamics (MD) simulations with machine learning (ML) architectures to investigate the mechanistic drivers of oil mobilization in siliciclastic (quartz) and carbonate (calcite) lithologies. Atomistic analysis confirms the significance of the multi-ion-bridging (MIB) mechanism, wherein divalent cations (Ca2+, Mg2+) reach up to 3.0 MPa at the three-phase contact line, contributing to a wettability alteration that reduces the contact angle by up to 72% in calcite systems. A distinct “Golden Window” of ionic concentration (3.0–4.0 wt %) was identified, where interfacial partitioning achieves an optimal thermodynamic balance. To rigorously analyze the MD data set, an XGBoost predictive engine was implemented; cross-validation and feature sensitivity analyses demonstrate exceptional predictive capability (R2 ≈ 0.89), identifying disjoining pressure and binding energy as primary mechanistic drivers. Unlike traditional bulk-parameter models, this approach employs unsupervised clustering to provide a broader understanding of the geochemical regimes governing recovery. These findings establish a comprehensive, high-precision roadmap for designing chemically engineered recovery processes, providing a simulation-driven, mechanically guided approach to estimate atomistic oil displacement (reaching ηdisp ≈ 52.3% under controlled simulated conditions).