Research Article
Development of a Machine Learning Model for Predicting the Structural and Optical Properties of Nanomaterials Based on Quantum-Mechanical Simulations
Issue:
Volume 14, Issue 3, June 2025
Pages:
60-66
Received:
7 May 2025
Accepted:
21 May 2025
Published:
3 June 2025
DOI:
10.11648/j.ijmsa.20251403.11
Downloads:
Views:
Abstract: The rapid advancement of nanotechnology has enabled the development of materials with unique properties that differ significantly from their bulk counterparts. Understanding and predicting the properties of nanomaterials, such as their electronic, optical, and mechanical characteristics, is crucial for their application in fields like electronics, energy storage, and catalysis. However, the computational methods used to predict these properties, particularly through quantum mechanical simulations such as Density Functional Theory (DFT), are computationally expensive and time-consuming, especially when applied to large datasets of nanomaterials. This paper proposes a novel approach that integrates machine learning (ML) techniques with DFT simulations to predict the structural and optical properties of nanomaterials. By utilizing a dataset derived from DFT calculations, we train and evaluate multiple machine learning models, including Random Forest, Support Vector Machine (SVM), and Deep Neural Networks (DNN), to predict key properties such as band gap, conductivity, and optical absorption. The goal is to develop a model that reduces the computational burden of traditional simulation methods while maintaining high accuracy and generalizability. The models were trained on a synthetic dataset that simulates the composition, size, and crystal structure of nanomaterials, with target properties generated based on these features. We evaluated the performance of the models using standard regression metrics such as Mean Absolute Error (MAE), Mean Squared Error (MSE), and R2. Results show that the DNN model provides the best predictive accuracy, closely followed by the Random Forest model, while the SVM model demonstrated lower performance in this context. Additionally, feature importance analysis revealed that material composition, particle size, and crystal structure were the most influential factors in determining the predicted properties of the nanomaterials. This research demonstrates the potential of machine learning to accelerate the discovery of new nanomaterials by providing a fast and scalable way to predict their properties. By combining the predictive power of ML with quantum mechanical simulations, this study offers an efficient framework for material discovery that can be applied to a wide range of nanomaterial systems.
Abstract: The rapid advancement of nanotechnology has enabled the development of materials with unique properties that differ significantly from their bulk counterparts. Understanding and predicting the properties of nanomaterials, such as their electronic, optical, and mechanical characteristics, is crucial for their application in fields like electronics, ...
Show More
