Clapeyron Neural Networks for Single-Species Vapor-Liquid Equilibria

2026-02-20

AI Summary

This work addresses the challenges of scarce experimental data and thermodynamic inconsistency in predicting pure-component vapor–liquid equilibrium properties by proposing a thermodynamics-informed graph neural network (GNN). For the first time, the Clausius–Clapeyron equation is incorporated as a physical constraint within a multi-task learning framework to jointly predict vapor pressure, liquid- and vapor-phase molar volumes, and enthalpy of vaporization. The equation is embedded into the loss function as a regularization term, ensuring that model predictions inherently satisfy thermodynamic consistency. Compared to both single-task and purely data-driven multi-task baselines, the proposed method achieves significantly improved prediction accuracy across all target properties, with the most pronounced gains observed for the property with the scarcest data.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Computer Vision: Multi-modal Vision
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Machine learning (ML) approaches have shown promising results for predicting molecular properties relevant for chemical process design. However, they are often limited by scarce experimental property data and lack thermodynamic consistency. As such, thermodynamics-informed ML, i.e., incorporating thermodynamic relations into the loss function as regularization term for training, has been proposed. We herein transfer the concept of thermodynamics-informed graph neural networks (GNNs) from the Gibbs-Duhem to the Clapeyron equation, predicting several pure component properties in a multi-task manner, namely: vapor pressure, liquid molar volume, vapor molar volume and enthalpy of vaporization. We find improved prediction accuracy of the Clapeyron-GNN compared to the single-task learning setting, and improved approximation of the Clapeyron equation compared to the purely data-driven multi-task learning setting. In fact, we observe the largest improvement in prediction accuracy for the properties with the lowest availability of data, making our model promising for practical application in data scarce scenarios of chemical engineering practice.

Problem

• vapor-liquid equilibria
• thermodynamic consistency
• data scarcity
• molecular property prediction
• pure component properties

Innovation

• Clapeyron equation
• thermodynamics-informed machine learning
• graph neural networks
• multi-task learning
• vapor-liquid equilibria