Component density profiles at vapour–liquid interfaces of mixtures can exhibit a non-monotonic behaviour with a maximum that can be many times larger than the densities in the bulk phases. This is called enrichment and is usually only observed for low-boiling components. The enrichment is a nanoscopic property which can presently not be measured experimentally – in contrast to the classical Gibbs adsorption. The available information on the enrichment stems from molecular simulations, density gradient theory, or density functional theory. The enrichment is highly interesting as it is suspected to influence the mass transfer across interfaces. In the present work, we review the literature data and the existing knowledge on this phenomenon and propose an empirical model to establish a link between the nanoscopic enrichment and macroscopic properties – namely vapour–liquid equilibrium data. The model parameters were determined from a fit to a dataset on the enrichment in about 100 binary Lennard-Jones model mixtures that exhibit different types of phase behaviour, which has recently become available. The model is then tested on the entire set of enrichment data that is available in the literature, which includes also mixtures containing non-spherical, polar, and H-bonding components. The model predicts the enrichment data from the literature (2,000 data points) with an AAD of about 16%, which is below the uncertainty of the enrichment data. This establishes a direct link between measurable macroscopic properties and the nanoscopic enrichment and enables predictions of the enrichment at vapour–liquid interfaces from macroscopic data alone.

中文翻译：

---

混合物汽液界面的富集：建立纳米级和宏观特性之间的联系

混合物汽液界面的组分密度分布可以表现出非单调行为，最大值可能比体相中的密度大许多倍。这称为富集，通常仅在低沸点组分中观察到。富集是一种纳米级特性，目前无法通过实验测量 - 与经典的吉布斯吸附相反。关于富集的可用信息来自分子模拟、密度梯度理论或密度泛函理论。富集非常有趣，因为它被怀疑会影响跨界面的传质。在目前的工作中，我们回顾了有关这一现象的文献数据和现有知识，并提出了一个经验模型来建立纳米富集和宏观特性之间的联系——即汽液平衡数据。模型参数是根据对大约 100 个二元 Lennard-Jones 模型混合物中富集的数据集的拟合确定的，这些模型混合物表现出不同类型的相行为，最近变得可用。然后在文献中可用的整个富集数据集上测试该模型，其中还包括包含非球形、极性和 H 键成分的混合物。该模型预测文献中的富集数据（2,000 个数据点），AAD 约为 16%，低于富集数据的不确定性。