Neural recommender system for the activity coefficient prediction andUNIFACmodel extension of ionicliquid‐solutesystems

Abstract

For the ionic liquid (IL)‐solute systems of broad interest, a deep neural network based recommender system (RS) for predicting the infinite dilution activity coefficient (γ∞) is proposed and applied for a large extension of the UNIFAC model. In the RS, neural network entity embeddings are employed for mapping each IL and solute, and neural collaborative filtering is utilized to handle the nonlinearities of IL‐solute interactions. A comprehensive experimental γ∞ database covering 215 ILs and 112 solutes (totally 41,553 data points) is established for training the RS, where the cross‐validation and test are performed based on a rigorous dataset split by IL‐solute combinations. The obtained RS shows superior performance than the state‐of‐the‐art γ∞ prediction models and is thus taken to complete the solute‐in‐IL γ∞ matrix. Based on the completed γ∞ database, a large extension of the UNIFAC‐IL model is finally presented, filling all the parameters between involved groups.

Graph neural networks for the prediction of infinite dilution activity coefficients

“…Many models exist that are used for the prediction of γ ∞ i j . We can divide these models into the ones based on mechanistic or phenomenological knowledge [6][7][8][9][10][11][12][13] (refer to as mechanistic models in this work), and those which are mainly constructed using machine learning techniques [14][15][16][17][18][19][20][21] . So far, the use of mechanistic models is more common compared to machine learning methods.…”

Predictive molecular thermodynamic models for ionic liquids

“…23,24 Although this method can fully reflect any structural changes in the cation, anion, and their substituents, the charged functional groups (i.e., cations and anions) necessitate the addition of a Debye-Hückel long-range electrostatic term to Equation (1), similar to solvent-inorganic salt systems. Particularly, Song et al [25][26][27][28] applied this approach to CAILD to significantly extend the designable space and freedom. (iii) The IL consists of several groups, as proposed by Lei et al 2 (see Figure 1C), but the anion and cation skeletons are as one electrically neutral group.…”

An accurate and interpretable deep learning model for environmental properties prediction using hybrid molecular representations

“…Existing methods for predicting log K OW can be classified as: Property‐based, 24–28 substructure‐based, 18,29–34 and machine learning‐based methods 35–39 . The property‐based methods are efficient in computation but rely heavily on correlated properties, for example, the activity coefficient depending highly on temperature, which are not always available.…”