Gibbs–Helmholtz 图神经网络：捕获无限稀释时活度系数的温度依赖性,Digital Discovery

当前位置： X-MOL 学术 › Digital Discovery › 论文详情

Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)

Gibbs–Helmholtz 图神经网络：捕获无限稀释时活度系数的温度依赖性
Digital Discovery ( IF 6.2 ) Pub Date : 2023-05-02 , DOI: 10.1039/d2dd00142j
Edgar Ivan Sanchez Medina ₁ , Steffen Linke ₁ , Martin Stoll ₂ , Kai Sundmacher _{1,

3}

Affiliation

准确预测混合物中化合物的物理化学性质（例如无限稀释时的活度系数γ _ij^∞）对于开发新颖且更具可持续性的化学过程至关重要。在这项工作中，我们分析了先前提出的基于 GNN 的模型预测γ _ij^∞的性能，并将它们与一系列 9 等温研究中的几个机械模型进行了比较。此外，我们开发了用于预测 ln γ _ij^∞的 Gibbs–Helmholtz 图神经网络 (GH-GNN) 模型不同温度下的分子系统。我们的方法将 Gibbs–Helmholtz 派生表达式的简单性与一系列图神经网络相结合，这些图神经网络结合了明确的分子和分子间描述符以捕获色散和氢键效应。我们使用实验确定的 ln γ _ij^∞训练了这个模型涉及 1032 种溶质和 866 种溶剂的 40 219 个二元系统的数据，与流行的 UNIFAC-Dortmund 模型相比，总体表现出优越的性能。我们分析了 GH-GNN 在连续和离散内插/外推方面的性能，并给出了模型的适用范围和预期精度的指示。一般来说，如果训练集中至少包含 25 个具有相同溶质-溶剂化学类别组合的系统，并且 Tanimoto 相似性指标高于0.35 也存在。该模型及其适用领域建议已在 https://github.com/edgarsmdn/GH-GNN 上开源。

"点击查看英文标题和摘要"

Gibbs–Helmholtz graph neural network: capturing the temperature dependency of activity coefficients at infinite dilution

The accurate prediction of physicochemical properties of chemical compounds in mixtures (such as the activity coefficient at infinite dilution γ_ij^∞) is essential for developing novel and more sustainable chemical processes. In this work, we analyze the performance of previously-proposed GNN-based models for the prediction of γ_ij^∞, and compare them with several mechanistic models in a series of 9 isothermal studies. Moreover, we develop the Gibbs–Helmholtz Graph Neural Network (GH-GNN) model for predicting ln γ_ij^∞ of molecular systems at different temperatures. Our method combines the simplicity of a Gibbs–Helmholtz-derived expression with a series of graph neural networks that incorporate explicit molecular and intermolecular descriptors for capturing dispersion and hydrogen bonding effects. We have trained this model using experimentally determined ln γ_ij^∞ data of 40 219 binary-systems involving 1032 solutes and 866 solvents, overall showing superior performance compared to the popular UNIFAC-Dortmund model. We analyze the performance of GH-GNN for continuous and discrete inter/extrapolation and give indications for the model's applicability domain and expected accuracy. In general, GH-GNN is able to produce predictions with a mean absolute error below 0.3 for extrapolated binary-systems if at least 25 systems with the same combination of solute–solvent chemical classes are contained in the training set and a Tanimoto similarity indicator above 0.35 is also present. This model and its applicability domain recommendations have been made open-source at https://github.com/edgarsmdn/GH-GNN.

更新日期：2023-05-02

点击分享查看原文

点击收藏

公开下载

阅读更多本刊新发论文