Predicting the chemical reactivity of organic materials using a machine-learning approach

Thiophene Stability in Photodynamic Therapy: A Mathematical Model Approach

Thiophene-containing photosensitizers are gaining recognition for their role in photodynamic therapy (PDT). However, the inherent reactivity of the thiophene moiety toward singlet oxygen threatens the stability and efficiency of these photosensitizers. This study presents a novel mathematical model capable of predicting the reactivity of thiophene toward singlet oxygen in PDT, using Conceptual Density Functional Theory (CDFT) and genetic programming. The research combines advanced computational methods, including various DFT techniques and symbolic regression, and is validated with experimental data. The findings underscore the capacity of the model to classify photosensitizers based on their photodynamic efficiency and safety, particularly noting that photosensitizers with a constant rate 1000 times lower than that of unmodified thiophene retain their photodynamic performance without substantial singlet oxygen quenching. Additionally, the research offers insights into the impact of electronic effects on thiophene reactivity. Finally, this study significantly advances thiophene-based photosensitizer design, paving the way for therapeutic agents that achieve a desirable balance between efficiency and safety in PDT.

Quantitative Structure-Reactivity Relationships for Synthesis Planning: The Benzhydrylium Case

Selective and feasible reactions are among the top targets in synthesis planning. Mayr's approach to quantifying chemical reactivity has greatly facilitated the planning process, but reactivity parameters for new compounds require time-consuming experiments. In the past decade, data-driven modeling has been gaining momentum in the field, as it shows promise in terms of efficient reactivity prediction. However, state-of-the-art models use quantum chemical data as input, which prevent access to real-time planning in organic synthesis. Here, we present a novel data-driven workflow for predicting reactivity parameters of molecules that takes only structural information as input, enabling de facto real-time reactivity predictions. We use the well-understood chemical space of benzhydrylium ions as an example to demonstrate the functionality of our approach and the performance of the resulting quantitative structure-reactivity relationships (QSRRs). Our results suggest that it is straightforward to build low-cost QSRR models that are accurate, interpretable, and transferable to unexplored systems within a given scope of application. Moreover, our QSRR approach suggests that Hammett σ parameters are only approximately additive.

Extracting Polaron Recombination from Electroluminescence in Organic Light-Emitting Diodes by Artificial Intelligence

Direct exploring the electroluminescence (EL) of organic light-emitting diodes (OLEDs) is a challenge due to the complicated processes of polarons, excitons, and their interactions. This study demonstrated the extraction of the polaron dynamics from transient EL by predicting the recombination coefficient via artificial intelligence, overcoming multivariable kinetics problems. The performance of a machine learning (ML) model trained by various EL decay curves is significantly improved using a novel featurization method and input node optimization, achieving an R² value of 0.947. The optimized ML model successfully predicts the recombination coefficients of actual OLEDs based on an exciplex-forming cohost, enabling the quantitative understanding of the overall polaron behavior under various electrical excitation conditions.

Predicting the Solubility of Organic Energy Storage Materials Based on Functional Group Identity and Substitution Pattern

Organic electrode materials (OEMs) provide sustainable alternatives to conventional electrode materials based on transition metals. However, the application of OEMs in lithium-ion and redox flow batteries requires either low or high solubility. Currently, the identification of new OEM candidates relies on chemical intuition and trial-and-error experimental testing, which is costly and time intensive. Herein, we develop a simple empirical model that predicts the solubility of anthraquinones based on functional group identity and substitution pattern. Within this statistical scaffold, a training set of 18 anthraquinone derivatives allows us to predict the solubility of 808 quinones. Internal and external validations show that our model can predict the solubility of anthraquinones in battery electrolytes within log S ± 0.7, which is a much higher accuracy than existing solubility models. As a demonstration of the utility of our approach, we identified several new anthraquinones with low solubilities and successfully demonstrated their utility experimentally in Li-organic cells.

The computational road to reactivity scales

Reactivity scales are useful research tools for chemists, both experimental and computational. However, to determine the reactivity of a single molecule, multiple measurements need to be carried out, which is a time-consuming and resource-intensive task. In this Tutorial Review, we present alternative approaches for the efficient generation of quantitative structure-reactivity relationships that are based on quantum chemistry, supervised learning, and uncertainty quantification. First published in 2002, we observe a tendency for these relationships to become not only more predictive but also more interpretable over time.

OSCAR: an extensive repository of chemically and functionally diverse organocatalysts

The automated construction of datasets has become increasingly relevant in computational chemistry. While transition-metal catalysis has greatly benefitted from bottom-up or top-down strategies for the curation of organometallic complexes libraries, the field of organocatalysis is mostly dominated by case-by-case studies, with a lack of transferable data-driven tools that facilitate both the exploration of a wider range of catalyst space and the optimization of reaction properties. For these reasons, we introduce OSCAR, a repository of 4000 experimentally derived organocatalysts along with their corresponding building blocks and combinatorially enriched structures. We outline the fragment-based approach used for database generation and showcase the chemical diversity, in terms of functions and molecular properties, covered in OSCAR. The structures and corresponding stereoelectronic properties are publicly available (https://archive.materialscloud.org/record/2022.106) and constitute the starting point to build generative and predictive models for organocatalyst performance.

Understanding the charge dynamics in organic light-emitting diodes using convolutional neural network

Knowledge about the charge dynamics in organic light-emitting diodes (OLEDs) is a critical clue to optimize device architecture for enhancing the power efficiency and driving voltage characteristics in addition to the external quantum efficiency. In this work, we demonstrated that the charge behavior according to the operation voltage of OLEDs could be understood by introducing the convolutional neural network (CNN) of the machine learning framework without additional analysis of the unipolar charge devices. The CNN model trained using a two-dimensional (2D) modulus fingerprint simultaneously predicted the mobilities of the charge transport and emitting layers, realizing a deep understanding of the complicated data that humans cannot interpret. The machine learning model successfully describes the electrical properties of the organic layers in the actual devices configurated by different electron-transporting materials and the composition of cohosts in the emitting layer. For the first time, it was revealed that 2D fingerprints extracted using frequency- and voltage-dependent modulus spectra were effective data to represent comprehensive charge dynamics of OLEDs. The interpretation and perspective of the machine learning approach in this work were also discussed.

Nucleophilicity Prediction Using Graph Neural Networks

The quantitative description between chemical reaction rates and nucleophilicity parameters plays a crucial role in organic chemistry. In this regard, the formula proposed by Mayr et al. and the constructed reactivity database are important representatives. However, the determination of Mayr's nucleophilicity parameter N often requires time-consuming experiments with reference electrophiles in the solvent. Several machine learning (ML)-based models have been proposed to realize the data-driven prediction of N in recent years. However, in addition to DFT-calculated electronic descriptors, most of them also use a set of artificially predefined structural descriptors as input, which may result in a biased representation of the nucleophile's structural information depending on descriptors' definition preference. Compared with traditional ML algorithms, graph neural networks (GNNs) can naturally take the molecule's structural information into account by applying the message passing technique. We herein proposed a SchNet-based GNN model that only takes the molecular conformation and solvent type as input. The model achieves a comparable performance to the previous benchmark study on 10-fold cross-validation of 894 data points (R² = 0.91, RMSE = 2.25). To enhance the model's ability to capture the molecule's electronic information, some DFT-calculated parameters are then incorporated into the model via graph global features, and substantial improvement is achieved in the prediction precision (R² = 0.95, RMSE = 1.63). These results demonstrate that both structural and electronic information are important for the prediction of N, and GNN can integrate these two kinds of information more effectively.

A machine learning approach for predicting the nucleophilicity of organic molecules

Nucleophilicity provides important information about the chemical reactivity of organic molecules. Experimental determination of the nucleophilicity parameter is a tedious and resource-intensive approach. Herein, we present a novel machine learning protocol that uses key structural descriptors to predict the nucleophilicities of organic molecules, which agree well with the experimental values. A data driven approach was used where quantum mechanical molecular and thermodynamic descriptors from a wide range of structurally diverse nucleophiles and relevant solvents were extracted and modelled using advanced algorithms against the experimentally available nucleophilicity values. Despite the structural diversity of nucleophiles, we are able to achieve statistically robust models with a high predictive power using tree-based and neural network algorithms trained on an in-house developed unique dataset consisting of 752 nucleophilicity values and 27 molecular descriptors.

Versatile Redox-Active Organic Materials for Rechargeable Energy Storage

With the ever-increasing demand on energy storage systems and subsequent mass production, there is an urgent need for the development of batteries with not only improved electrochemical performance but also better sustainability-related features such as environmental friendliness and low production cost. To date, transition metals that are sparse have been centrally employed in energy storage devices ranging from portable lithium ion batteries (e.g., cobalt and nickel) to large-scale redox flow batteries (e.g., vanadium). Toward the sustainable battery chemistry, there are ongoing efforts to replace the transition metal-based electrode materials in these systems to redox-active organic materials (ROMs). Most ROMs are composed of the earth abundant elements (e.g., carbon, nitrogen, oxygen, sulfur), thus are less restrained by the resource, and their production does not require high-energy consuming processes. Furthermore, the structural diversity and chemical tunability of organic compounds make them more attractive for the versatile design of future energy storage systems. Accordingly, the timely development of high-performance ROM-based electrodes would expedite the shift from the current resource-limited battery chemistry to more sustainable energy solutions.In this Account, we provide an overview of the endeavors to employ and develop ROMs as high-performance active materials for various battery systems. Diverse approaches will be introduced starting from the new ROM design mimicking the energy carrying molecules in biological metabolism to the chemical modifications to tailor the properties for specific battery systems. The molecular redesign of ROM, for example, can be carried out by substituting heteroatoms in the redox center, which leads to the enhancement of the redox potential by the inductive effect. Or, tailoring the ROM molecule by removing redox-inactive functionals results in a reduced molecular weight, thereby an increased specific capacity. The intrinsic limitations of ROMs, such as the low electrical conductivity and the dissolving nature, have been under extensive scrutiny; however, they can be partly addressed through efforts including intermolecular fusion and/or nanoscale hybridization with a conducting scaffold. On the other hand, this problematic dissolving nature of ROMs makes them appealing for some new battery configurations such as redox flow batteries that employ the liquid-state active materials. The high solubility and the stability of the ROM were found to be beneficial in attaining the enhanced energy density and the cycle stability of flow batteries, which could be further optimized by the chemical modifications of ROMs. Besides the role of active materials, the redox activity of ROMs has also enabled their use as catalysts to promote the electrode reaction in metal-air batteries. The redox capability of the ROM was often proven to be effective in the solution-based redox mediation that facilitates both the charging and discharging reaction in metal-air batteries. Finally, we conclude this account by proposing the future research directions regarding the fundamental electrochemistry and the further practical development of ROMs for the sustainable rechargeable energy storage.

Predicting Solvent-Dependent Nucleophilicity Parameter with a Causal Structure Property Relationship

Solvent-dependent reactivity is a key aspect of synthetic science, which controls reaction selectivity. The contemporary focus on new, sustainable solvents highlights a need for reactivity predictions in different solvents. Herein, we report the excellent machine learning prediction of the nucleophilicity parameter N in the four most-common solvents for nucleophiles in the Mayr's reactivity parameter database (R² = 0.93 and 81.6% of predictions within ±2.0 of the experimental values with Extra Trees algorithm). A Causal Structure Property Relationship (CSPR) approach was utilized, with focus on the physicochemical relationships between the descriptors and the predicted parameters, and on rational improvements of the prediction models. The nucleophiles were represented with a series of electronic and steric descriptors and the solvents were represented with principal component analysis (PCA) descriptors based on the ACS Solvent Tool. The models indicated that steric factors do not contribute significantly, because of bias in the experimental database. The most important descriptors are solvent-dependent HOMO energy and Hirshfeld charge of the nucleophilic atom. Replacing DFT descriptors with Parameterization Method 6 (PM6) descriptors for the nucleophiles led to an 8.7-fold decrease in computational time, and an ∼10% decrease in the percentage of predictions within ±2.0 and ±1.0 of the experimental values.

Machine Learning Boosting the Development of Advanced Lithium Batteries

Lithium batteries (LBs) have many high demands regarding their application in portable electronic devices, electric vehicles, and smart grids. Machine learning (ML) can effectively accelerate the discovery of materials and predict their performances for LBs, which is thus able to markedly enhance the development of advanced LBs. In recent years, there have been many successful examples of using ML for advanced LBs. In this review, the basic procedure and representative methods of ML are briefly introduced to promote understanding of ML by experts in LBs. Then, the application of ML in developing LBs is highlighted for the purpose of attracting more attention to this field. Finally, the challenges and perspectives of ML are noted for the further development of LBs. It is hoped that this review can shed light on the application of ML in developing LBs and boost the development of advanced LBs.

BonDNet: a graph neural network for the prediction of bond dissociation energies for charged molecules

A broad collection of technologies, including e.g. drug metabolism, biofuel combustion, photochemical decontamination of water, and interfacial passivation in energy production/storage systems rely on chemical processes that involve bond-breaking molecular reactions. In this context, a fundamental thermodynamic property of interest is the bond dissociation energy (BDE) which measures the strength of a chemical bond. Fast and accurate prediction of BDEs for arbitrary molecules would lay the groundwork for data-driven projections of complex reaction cascades and hence a deeper understanding of these critical chemical processes and, ultimately, how to reverse design them. In this paper, we propose a chemically inspired graph neural network machine learning model, BonDNet, for the rapid and accurate prediction of BDEs. BonDNet maps the difference between the molecular representations of the reactants and products to the reaction BDE. Because of the use of this difference representation and the introduction of global features, including molecular charge, it is the first machine learning model capable of predicting both homolytic and heterolytic BDEs for molecules of any charge. To test the model, we have constructed a dataset of both homolytic and heterolytic BDEs for neutral and charged (-1 and +1) molecules. BonDNet achieves a mean absolute error (MAE) of 0.022 eV for unseen test data, significantly below chemical accuracy (0.043 eV). Besides the ability to handle complex bond dissociation reactions that no previous model could consider, BonDNet distinguishes itself even in only predicting homolytic BDEs for neutral molecules; it achieves an MAE of 0.020 eV on the PubChem BDE dataset, a 20% improvement over the previous best performing model. We gain additional insight into the model's predictions by analyzing the patterns in the features representing the molecules and the bond dissociation reactions, which are qualitatively consistent with chemical rules and intuition. BonDNet is just one application of our general approach to representing and learning chemical reactivity, and it could be easily extended to the prediction of other reaction properties in the future.