A robust framework for generating adsorption isotherms to screen materials for carbon capture

In this paper, we present a workflow that is designed to work without manual intervention to efficiently predict, by using molecular simulations, the thermodynamic data that is needed to design a carbon capture process. We developed a procedure that does not rely on fitting of the adsorption isotherms. From molecular simulations, we can obtain accurate data for both, the pure component isotherms as well as the mixture isotherms. This allowed us to make a detailed comparison of the different methods to predict the mixture isotherms. All approaches rely on an accurate description of the pure component isotherms and a model to predict the mixture isotherms. As we are interested in low CO₂ concentrations, it is essential that these models correctly predict the low pressure limit, i.e., give a correct description of the Henry regime. Among the equations that describe this limit correctly, the dual-site Langmuir (DSL) model is often used for the pure components and the extended DSL (EDSL) for the mixtures. An alternative approach, which avoids describing the pure component isotherms with a model, is to numerically integrate the pure component isotherms in the context of IAST. In this work we compare these two methods. In addition, we show that the way these data are fitted for DSL can significantly impact the ranking of materials, in particular for capture processes with low concentration of CO₂ in the feed stream.

本文提出了一种无需人工干预的工作流程，旨在通过分子模拟高效预测设计碳捕集过程所需的 thermodynamic data（热力学数据）。本研究开发了一种无需拟合吸附等温线的程序。通过分子模拟，我们能够获得纯净组分等温线以及混合物等温线的精确数据。这使我们能够对不同预测混合物等温线的方法进行详尽的比较。所有方法均依赖于对纯净组分等温线的准确描述以及预测混合物等温线的模型。鉴于我们对低浓度 CO₂ 感兴趣，这些模型必须能够准确预测低压极限，即正确描述亨利定律。在正确描述该极限的方程中，双位点 Langmuir (DSL) 模型常用于纯净组分，而扩展的 DSL (EDSL) 用于混合物。一种替代方法，即在不使用模型描述纯净组分等温线的情况下，是在 IAST 范围内数值积分纯净组分等温线。在本研究中，我们比较了这两种方法。此外，我们还表明，这些数据在 DSL 中的拟合方式会显著影响材料的排名，尤其是在进料流中 CO₂ 浓度较低的捕集过程中。