Towards Optimal Production of Graphene by Electrolysis in Molten Salts Using Machine Learning
DOI:
https://doi.org/10.54820/entrenova-2023-0007Ključne reči:
graphene, electrolysis, graphite, molten salts, machine learningApstrakt
Graphene is very exciting and one of the best materials in the last decade. The interest in producing non-expensive and high-quality graphene is of great interest to many Departments of materials and researchers worldwide. The main goal of this research considers the design, development, and optimisation of new technology for producing graphene by electrolysis in molten salts using constant cell voltage, reversing cell voltage, and constant overpotential and reversing overpotential methods. Compared to other processes, the electrolysis process is simple, ecological, and economical. Furthermore, it offers possibilities to control various parameters: type of electrolyte, applied voltage, current density, temperature and type of graphite used as a row material for synthesis of graphene. The underlying relationship between the parameters and the quality of the graphene is obtained through an explainable tree-based Machine Learning (ML) model. We trained the model by using the production process parameters as input attributes and the quality of the produced graphene as a target variable. Domain experts provide labels regarding the quality of each experimental instance in our data set. We propose using the extracted rules from the model to serve as a blueprint for optimal graphene production by electrolysis in molten salts. Choosing the correct values for the process parameters results in high-yield graphene production, up to ten times cheaper than the one produced with other existing technologies.
Reference
Alanyalıoğlu, M., Segura, J. J., Oró-Solé, J., & Casañ-Pastor, N. (2012). The synthesis of graphene sheets with controlled thickness and order using surfactant-assisted electrochemical processes. Elsevier (BV).
Andonovic, V., Azemi, M. K., Andonovic, B., & Dimitrov, A. (2022). Optimal Selection of Parameters for Production of Multiwall Carbon Nanotubes (MWCNTs) by Electrolysis in Molten Salts using Machine Learning. Association for Advancing Innovation and Research in Economy.
Andonovic, B., Ademi, A., Grozdanov, A., Paunovic, P., & Dimitrov, A. T. (2015). Enhanced model for determining the number of graphene layers and their distribution from X-ray diffraction data. Beilstein Institut.
Daniya, T., Geetha, M., & Kumar, K. S. (2020). Classification and regression trees with gini index. Union of Researchers of Macedonia.
Geim, A. K., & Novoselov, K. S. (2007). The rise of graphene . Springer Science and Business Media (LLC).
Geim, A. K. (2011). Nobel Lecture: Random walk to graphene. American Physical Society (APS).
Kamali, A. R., & Fray, D. (2013). Molten salt corrosion of graphite as a possible way to make carbon nanostructures. Elsevier (BV).
Novoselov, K. S., Geim, A., Morozov, S., Zhang, Y., & Dubons, S. (2004). Electric field effect in atomically thin carbon films. American Association for the Advancement of Science (AAAS).
Potdar, K., Pardawala, T., & Pai, C. (2017). A Comparative Study of Categorical Variable Encoding Techniques for Neural Network Classifiers. Foundation of Computer Science.
Rao, K. S., Senthilnathan, J., Liu, Y.-F., & Yoshimura, M. (2014). Role of Peroxide Ions in Formation of Graphene Nanosheets by Electrochemical Exfoliation of Graphite. Springer Science and Business Media (LLC).
Raju, V. N., Lakshmi, K. P., Jain, V. M., Kalidindi, A., & Padma, V. (2020). Study the Influence of Normalisation/Transformation process on the Accuracy of Supervised Classification. IEEE.
Schwandt, C., Dimitrov, A. T., & Fray, D. (2010). The preparation of nano-structured carbon materials by electrolysis of molten lithium chloride at graphite electrodes . Elsevier (BV).
Vlassiouk, I., Fulvio, P. F., III, H. M., & Lavrik, N. (2013). Large scale atmospheric pressure chemical vapor deposition of graphene. Elsevier (BV).
##submission.downloads##
Objavljeno
Kako citirati
Broj časopisa
Sekcija
Licenca
Sva prava zadržana (c) 2024 Viktor Andonovikj, Mimoza Kovaci Azemi, Beti Andonovic, Aleksandar Dimitrov
Ovaj rad je pod Creative Commons Autorstvo-Nekomercijalno 4.0 Internacionalna licenca.