Computational Approach for Molecular Design of Small Organic Molecules with High Hole Mobilities in Amorphous Phase Using Random Forest Technique and Computer Simulation Method

Abstract

A molecular design system of small organic molecules was developed to realize high hole mobility in the amorphous phase by using the random forest (RF) technique and a computer simulation method. Since there was no accessible datasets of the hole mobilities in the amorphous phase, we have constructed a datasets of experimental hole mobilities for 321 organic molecules with calculated MO energies to utilize machine learning techniques for molecular design procedures. RF was applied to the machine learning technique using the constructed datasets. The optimized RF exhibited correlation coefficients of 0.885 and 0.764 for the training and testing datasets, respectively. The modified ChemTS to use optimized RF was adapted to design molecules with high hole mobilities. The successive conduction (SC) model that uses ab initio molecular orbital (MO) calculations and the Marcus theory was adapted to ensure the predicted mobilities. The generated molecules were verified to have high calculated hole mobility (the order of 10−2 cm2/Vs) owing to the highest transfer integral and lowest reorganization energy by the theoretical successive conduction (SC) model. The datasets and programs used in this work were publicly released on GitHub.

Machine Learning Approach for Predicting the Hole Mobility of the Perovskite Solar Cells

“…Recently, Nakaguro et al. [ 39 ] also similarly developed a model applying the Random Forest method, with 69 descriptors using 321 experimental hole mobility data. Contrasting with the work by Nakaguro et al., our approach to hyperparameter tuning in the random forest model demonstrates the potential for improvements in predictive accuracy, even with limited datasets.…”

“…[ 4,28–37 ] Only a few models have been used to predict the hole mobility of the solar cells. [ 38,39 ] All of these ML models focused on perovskite, and the reduced number of HTMs limited the scope of the models to link the role of the HTM with the performance of the cell. However, in recent years, the number and types of HTMs have shown a significant increase, and we are currently able to gather hundreds of experimental photovoltaic data from the literature.…”

Bio-gel nanoarchitectonics in tissue engineering

“…However, a more rational approach requires the introduction of data processing methodologies. The concepts of machine learning 85 and materials informatics 86 are being used to select, design, and develop optimal materials. There are also proposals to combine nanoarchitectonics and materials informatics.…”

Molecular machines working at interfaces: physics, chemistry, evolution and nanoarchitectonics

“…The organization of molecular machines of multiple species is expected to make the system more complex. Although it has been difficult to build such complex systems using conventional methods, the introduction of artificial intelligence technologies 140 may provide a solution to this problem. We can expect molecular machines to evolve further with the addition of new essences such as the setting of an interface, organization by nanoarchitectonics, and analysis by artificial intelligence.…”