COSMO-derived descriptors applied in ionic liquids physical property modelling using machine learning algorithms

Abstract An application of machine learning algorithms for the prediction of physical properties of ionic liquids is presented herein. Molecular descriptors obtained from quantum-chemistry calculations (COSMO theory (Klamt, 2004)) containing both structural and energetic information are used as input parameters. In this sense, a set of COSMO-based descriptors is proposed by reduction of the original σ-profile (51 descriptors reduced to 9 bins). A critically evaluated set of viscosity data is used for a large number of ionic liquids (159). Artificial neural networks are then trained for the correlation of liquid viscosity and compared with available tools (QSPR).

[1]  Stanley I. Sandler,et al.  A Priori Phase Equilibrium Prediction from a Segment Contribution Solvation Model , 2002 .

[2]  Kamil Paduszynski,et al.  Viscosity of Ionic Liquids: An Extensive Database and a New Group Contribution Model Based on a Feed-Forward Artificial Neural Network , 2014, J. Chem. Inf. Model..

[3]  A. Karunanithi,et al.  New quantum chemistry-based descriptors for better prediction of melting point and viscosity of ionic liquids , 2016, 1612.00879.

[4]  J. Valderrama,et al.  Viscosity of ionic liquids using the concept of mass connectivity and artificial neural networks , 2011 .

[5]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[6]  Ismael Díaz,et al.  Enterprise Ionic Liquids Database (ILUAM) for Use in Aspen ONE Programs Suite with COSMO-Based Property Methods , 2018 .

[7]  Rafiqul Gani,et al.  Product design - Molecules, devices, functional products, and formulated products , 2015, Comput. Chem. Eng..

[8]  G. Járvás,et al.  Estimation of Hansen solubility parameters using multivariate nonlinear QSPR modeling with COSMO scr , 2011 .

[9]  D. Mowla,et al.  Estimation of viscosities of pure ionic liquids using an artificial neural network based on only structural characteristics , 2017 .

[10]  N. Muhammad,et al.  Quantitative structure–activity relationships (QSARs) for estimation of activity coefficient at infinite dilution of water in ionic liquids for natural gas dehydration , 2016 .

[11]  M. El-Harbawi,et al.  Development of QSAR model to predict the ecotoxicity of Vibrio fischeri using COSMO-RS descriptors. , 2017, Chemosphere.

[12]  Suojiang Zhang,et al.  Prediction of Heat Capacity of Ionic Liquids Based on COSMO-RS Sσ-profile , 2015 .

[13]  Lei Zhang,et al.  A grand model for chemical product design , 2016, Comput. Chem. Eng..

[14]  J. Torrecilla,et al.  Development of an a Priori Ionic Liquid Design Tool. 1. Integration of a Novel COSMO-RS Molecular Descriptor on Neural Networks , 2008 .

[15]  J. Coutinho,et al.  A group contribution method for viscosity estimation of ionic liquids , 2008 .

[16]  Andreas Klamt,et al.  COSMO-RS: From Quantum Chemistry to Fluid Phase Thermodynamics , 2005 .