Group contribution-based property estimation methods: advances and perspectives

Prediction of Thermochemical Properties of Long-Chain Alkanes Using Linear Regression: Application to Hydroisomerization

Linear regression (LR) is used to predict thermochemical properties of alkanes at temperatures (0-1000) K to study chemical reaction equilibria inside zeolites. The thermochemical properties of C1 until C10 isomers reported by Scott are used as training data sets in the LR model which is used to predict these properties for alkanes longer than C10 isomers. Second-order groups are used as independent variables which account for the interactions between the neighboring groups of atoms. This model accurately predicts Gibbs free energies, enthalpies, Gibbs free energies of formation, and enthalpies of formation for alkanes which exceeds the chemical accuracy of 1 kcal/mol and outperforms the group contribution methods developed by Benson et al., Joback and Reid, and Constantinou and Gani. Predictions from our model are used to compute the reaction equilibrium distribution of hydroisomerization of C10 and C14 isomers in MTW-type zeolite. Calculation of reaction equilibrium distribution inside zeolites also requires Henry coefficients of the isomers which can be computed using classical force field-based molecular simulations using the RASPA2 software for which we created an automated workflow. The reaction equilibrium distribution for C10 isomers obtained using the LR model and the training data set for this model are in very good agreement. The tools developed in this study will enable the computational study of hydroisomerization of long-chain alkanes (>C10).

Modeling Dipolar Molecules with PCP-SAFT: A Vector Group-Contribution Method

Predicting thermodynamic equilibrium properties is essential to develop chemical and energy conversion processes in the absence of experimental data. For the modeling of thermodynamic properties, statistical associating fluid theory (SAFT)-based equations of state, such as perturbed-chain polar (PCP)-SAFT, have been proven powerful and found broad application. The PCP-SAFT parameters can be predicted by group-contribution (GC) methods. However, their application to the dipole term is substantially limited: current GC methods neglect the dipole term or only allow for a single dipolar group per substance to avoid handling the molecular dipole moment's symmetry effects. Still, substances with multiple dipolar groups are highly relevant, and their description substantially improves by including the dipole term in SAFT models. To overcome these limitations, this work proposes a vector-addition-based (Vector-)GC method for the dipole term of PCP-SAFT that accounts for molecular symmetry. The Vector-GC employs information on the substance's molecular 3D structure to predict the molecular dipole moment through a vector addition of bond contributions. Combining the proposed sum rule for dipole moments with established sum rules for the remaining parameters yields a consistent GC method for PCP-SAFT for dipolar substances. The prediction capabilities of the Vector-GC method are analyzed against experimental data for two substance classes: nonassociating oxygenated and halogenated substances. We demonstrate that the Vector-GC method improves vapor pressure and liquid density predictions compared to neglecting the dipole term. Moreover, we show that the Vector-GC method enables differentiation between cis- and trans-isomers. The Vector-GC method, hence, substantially increases the predictive capabilities and applicability domain of GC methods. All parameters are provided as JSON and CSV files, and the Vector-GC method is available through an open-source python package. Additionally, the developed regression framework for GC methods for PCP-SAFT is openly available. The regression framework can be employed to regress the Vector-GC method to other substance classes and is easily adaptable to other sum rules for PCP-SAFT.

Crystal Phase Ionic Liquids for Energy Applications: Heat Capacity Prediction via a Hybrid Group Contribution Approach

In the selection and design of ionic liquids (ILs) for various applications, including heat transfer fluids, thermal energy storage materials, fuel cells, and solvents for chemical processes, heat capacity is a key thermodynamic property. While several attempts have been made to develop predictive models for the estimation of the heat capacity of ILs in their liquid phase, none so far have been reported for the ILs' solid crystal phase. This is particularly important for applications where ILs will be used for thermal energy storage in the solid phase. For the first time, a model has been developed and used for the prediction of crystal phase heat capacity based on extending and modifying a previously developed hybrid group contribution model (GCM) for liquid phase heat capacity. A comprehensive database of over 5000 data points with 71 unique crystal phase ILs, comprising 42 different cations and 23 different anions, was used for parameterization and testing. This hybrid model takes into account the effect of the anion core, cation core, and subgroups within cations and anions, in addition to the derived indirect parameters that reflect the effects of branching and distribution around the core of the IL. According to the results, the developed GCM can reliably predict the crystal phase heat capacity with a mean absolute percentage error of 6.78%. This study aims to fill this current gap in the literature and to enable the design of ILs for thermal energy storage and other solid phase applications.

Process Simulation and Optimization on Ionic Liquids

Ionic liquids (ILs) are promising alternative compounds that enable the development of technologies based on their unique properties as solvents or catalysts. These technologies require integrated product and process designs to select ILs with optimal process performances at an industrial scale to promote cost-effective and sustainable technologies. The digital era and multiscale research methodologies have changed the paradigm from experiment-oriented to hybrid experimental-computational developments guided by process engineering. This Review summarizes the relevant contributions (>300 research papers) of process simulations to advance IL-based technology developments by guiding experimental research efforts and enhancing industrial transferability. Robust simulation methodologies, mostly based on predictive COSMO-SAC/RS and UNIFAC models in Aspen Plus software, were applied to analyze key IL applications: physical and chemical CO2 capture, CO2 conversion, gas separation, liquid-liquid extraction, extractive distillation, refrigeration cycles, and biorefinery. The contributions concern the IL selection criteria, operational unit design, equipment sizing, technoeconomic and environmental analyses, and process optimization to promote the competitiveness of the proposed IL-based technologies. Process simulation revealed that multiscale research strategies enable advancement in the technological development of IL applications by focusing research efforts to overcome the limitations and exploit the excellent properties of ILs.

Transformer-Based Representation of Organic Molecules for Potential Modeling of Physicochemical Properties

In this work, we study the use of three configurations of an autoencoder neural network to process organic substances with the aim of generating meaningful molecular descriptors that can be employed to develop property prediction models. A total of 18,322,500 compounds represented as SMILES strings were used to train the model, demonstrating that a latent space of 24 units is able to adequately reconstruct the data. After AE training, an analysis of the latent space properties in terms of compound similarity was carried out, indicating that this space possesses desired properties for the potential development of models for forecasting physical properties of organic compounds. As a final step, a QSPR model was developed to predict the boiling point of chemical substances based on the AE descriptors. 5276 substances were used for the regression task, and the predictive ability was compared with models available in the literature evaluated on the same database. The final AE model has an overall error of 1.40% (1.39% with augmented SMILES) in the prediction of the boiling temperature, while other models have errors between 2.0 and 3.2%. This shows that the SMILES representation is comparable and even outperforms the state-of-the-art representations widely used in the literature.

Solid-liquid equilibria of triacylglycerols and vitamin E mixtures

Oils and fats are important ingredients for food and pharmaceutical industries. Their main compounds, such as triacylglycerols (TAG), are responsible for determining their physical properties during food storage and consumption. Lipid-rich foods are also sources of minority compounds, which is the case of vitamin E, mainly represented by (±)-α-tocopherol. These compounds can interact with the main lipid molecules in food formulation leading to modification on lipids' physicochemical properties during processes, storage, as well as during digestion, possibly altering their nutritional functionalities, which is the case of vitamin E antioxidant abilities, but also their solubility in the systems. In this case, the study of the phase-behavior between (±)-α-tocopherol and lipid compounds can elucidate these physicochemical changings. Therefore, this work was aimed at determining the solid-liquid equilibrium (SLE) of binary mixtures of TAG (tripalmitin, triolein and tristearin) and (±)-α-tocopherol including the complete description of their phase diagrams. Melting data were evaluated by Differential Scanning Calorimetry, Microscopy, X-Ray Diffraction, and thermodynamic modeling by using Margules, UNIFAC, and COSMO-SAC models. Experimental results showed that systems presented a monotectic-like behavior, with a significant decreasing in TAG melting temperature by the addition of (±)-α-tocopherol. This high affinity and attractive strengths between these molecules were also observed by thermodynamic modeling, whose absolute deviations were below 2 %. Micrographs and X-ray diffraction evidenced the possible formation of solid solutions. Both behaviors are interesting by avoiding phase separation on food in solid and liquid phases, possibly improving the antioxidant role the (±)-α-tocopherol in lipid-base systems.

Molecular modelling of the thermophysical properties of fluids: expectations, limitations, gaps and opportunities

This manuscript provides an overview of the current state of the art in terms of the molecular modelling of the thermophysical properties of fluids. It is intended to manage the expectations and serve as guidance to practising physical chemists, chemical physicists and engineers in terms of the scope and accuracy of the more commonly available intermolecular potentials along with the peculiarities of the software and methods employed in molecular simulations while providing insights on the gaps and opportunities available in this field. The discussion is focused around case studies which showcase both the precision and the limitations of frequently used workflows.

Combining Group-Contribution Concept and Graph Neural Networks Toward Interpretable Molecular Property Models

Quantitative structure-property relationships (QSPRs) are important tools to facilitate and accelerate the discovery of compounds with desired properties. While many QSPRs have been developed, they are associated with various shortcomings such as a lack of generalizability and modest accuracy. Albeit various machine-learning and deep-learning techniques have been integrated into such models, another shortcoming has emerged in the form of a lack of transparency and interpretability of such models. In this work, two interpretable graph neural network (GNN) models (attentive group-contribution (AGC) and group-contribution-based graph attention (GroupGAT)) are developed by integrating fundamentals using the concept of group contributions (GC). The interpretability consists of highlighting the substructure with the highest attention weights in the latent representation of the molecules using the attention mechanism. The proposed models showcased better performance compared to classical group-contribution models, as well as against various other GNN models describing the aqueous solubility, melting point, and enthalpies of formation, combustion, and fusion of organic compounds. The insights provided are consistent with insights obtained from the semiempirical GC models confirming that the proposed framework allows highlighting the important substructures of the molecules for a specific property.

Screening for New Efficient and Sustainable-by-Design Solvents to Assist the Extractive Fermentation of Glucose to Bioethanol Fuels

The production of bioethanol fuels using extractive fermentation increases the efficiency of the bioconversion reaction by reducing the toxic product inhibition. The choice of appropriate solvents to remove the bioethanol product without inhibiting the fermentation is important to enable industrial scale application. This work applies computer-aided molecular design technologies to systematically screen a wide variety of candidate solvents to enhance the separation, also considering the microorganisms that perform the fermentation. The performance of the candidates was evaluated using a rigorous process simulator for extractive fermentation, assisted by functional group-contribution (QSPR/QSAR) models for the prediction of various solvent properties, including toxicity and life cycle impacts. The solvent designs generated through this approach can provide powerful insights on the kind of molecular structures and functionalities that satisfy the process objectives and constraints, as well the desired sustainability features.

Paradigm Shift: The Promise of Deep Learning in Molecular Systems Engineering and Design

The application of deep learning to a diverse array of research problems has accelerated progress across many fields, bringing conventional paradigms to a new intelligent era. Just as the roles of instrumentation in the old chemical revolutions, we reinforce the necessity for integrating deep learning in molecular systems engineering and design as a transformative catalyst towards the next chemical revolution. To meet such research needs, we summarize advances and progress across several key elements of molecular systems: molecular representation, property estimation, representation learning, and synthesis planning. We further spotlight recent advances and promising directions for several deep learning architectures, methods, and optimization platforms. Our perspective is of interest to both computational and experimental researchers as it aims to chart a path forward for cross-disciplinary collaborations on synthesizing knowledge from available chemical data and guiding experimental efforts.

Group Contribution Estimation of Ionic Liquid Melting Points: Critical Evaluation and Refinement of Existing Models

While several group contribution method (GCM) models have been developed in recent years for the prediction of ionic liquid (IL) properties, some challenges exist in their effective application. Firstly, the models have been developed and tested based on different datasets; therefore, direct comparison based on reported statistical measures is not reliable. Secondly, many of the existing models are limited in the range of ILs for which they can be used due to the lack of functional group parameters. In this paper, we examine two of the most diverse GCMs for the estimation of IL melting point; a key property in the selection and design of ILs for materials and energy applications. A comprehensive database consisting of over 1300 data points for 933 unique ILs, has been compiled and used to critically evaluate the two GCMs. One of the GCMs has been refined by introducing new functional groups and reparametrized to give improved performance for melting point estimation over a wider range of ILs. This work will aid in the targeted design of ILs for materials and energy applications.

Prediction of ionic conductivity of imidazolium-based ionic liquids at different temperatures using multiple linear regression and support vector machine algorithms

The conductivity of various imidazolium-based ILs has been predicted via QSPR approach using MLR and SVM regression coupled with stepwise model-building. This will aid the screening of suitable ILs with desired conductivity for specific applications.