Nanoporous Material Recognition via 3D Convolutional Neural Networks: Prediction of Adsorption Properties

Abstract

Nanoporous materials can be effective adsorbents for various energy applications. Because of their abundant number, brute-force-based material discovery can, however, be challenging. Data-driven approaches can be advantageous for such purposes. In this study, we demonstrate for the first time the applicability of a 3D convolutional neural network (CNN) in material recognition for predicting adsorption properties. 2D CNNs have been widely applied to image recognition, where the CNN self-learns important features of images, without the need of handcrafting features that are subject to human bias. This study explores methane adsorption in zeolites as a case study, where ∼6500 hypothetical zeolites are utilized to train/validate our designed CNN model. The CNN model offers highly accurate predictions, and the self-learned features resemble the channel and pore-like geometry of structures. This study demonstrates the extension of computer vision to materials science and paves the way for future studies such as carbon capture.

Recommendation System to Predict Missing Adsorption Properties of Nanoporous Materials

“…Machine learning plays an important role in the discovery and deployment of NPMs [26][27][28][29][30][31][32]. Supervised machine learning models have been widely used to predict the adsorption properties of NPMs [33][34][35][36][37][38][39][40][41] from vectors of hand-crafted structural features [42,43] or from a graph representation [44]. Unsupervised machine learning methods have been used to embed NPMs into a lowdimensional "material space" [45] and cluster together NPMs with similar structures [46][47][48].…”

“…Machine learning plays an important role in the discovery and deployment of NPMs. − Supervised machine learning models have been widely used to predict the adsorption properties of NPMs − from vectors of hand-crafted structural features , or graph representations. , Unsupervised machine learning methods have been used to embed NPMs into a low-dimensional “material space” and cluster together NPMs with similar structures. − Genetic algorithms, − Monte Carlo tree search, and Bayesian optimization , have been used to more efficiently search for the NPM(s) with an optimal adsorption property. Finally, recently an autoencoder enabled inverse design , of NPMs, where one specifies a desired adsorption property, and the machine learning model generates a NPM structure with that property.…”

Learning the Use of Artificial Intelligence in Heterogeneous Catalysis

“…Discovering new materials is of the utmost importance from both technological and scientific points of view. That is why, in recent years, the use of artificial intelligence techniques to speed up their search has proliferated (Goldsmith et al, 2018;Gupta et al, 2018;Gómez-Peralta and Bokhimi 2020;Chen et al, 2021;Cho and Lin, 2021;Konno et al, 2021). Of particular interest is the finding of nanoporous materials because they have a large surface area that is attractive in heterogeneous catalysis (Cho and Lin, 2021).…”

“…However, nowadays, selecting those with an attractive surface area for heterogeneous catalysis occurs through the slow procedure of trial and error. As an alternative to this discovery procedure, Cho et al (Cho and Lin, 2021) developed a learning machine based on convolution neural networks (CNN) that predicts the surface area of such materials.…”

Interpolating Nonadiabatic Molecular Dynamics Hamiltonian with Artificial Neural Networks

“…ANN is a widely used ML model that has been applied to predict thermodynamic stability, potential energy, formation energy, and other properties of perovskites and many other materials. − The ANN architecture contains an input layer receiving a vector of features of a given length, several hidden layers which conduct linear transformation on the input followed by application of a nonlinear activation function, and lastly, an output layer giving the prediction, Figure S1. In the present work, the features are calculated using the modified symmetry function with the following form: …”