Predicting Quantum Descriptor Values in B12N12-Based Nanocage Material Using CNN for Toxic Gas Sensor Applications

Air pollution has become an urgent global issue as it has significant scientific, social and economic importance for countries around the world with serious consequences to human health, ecosystems and the global climate. Emissions, air pollution from traffic are major contributors to air pollution in many regions. Long-term exposure to toxic gases can lead to a variety of health problems, including respiratory diseases, heart diseases, and even death (1).
In this context, the development of gas sensors plays a very important role in detecting and monitoring the concentration of toxic gases in the environment. The promising potential for detecting toxic gases based on nanocages, particularly B12N12 has emerged as an interesting research area. Jensen and Toftlund in their study showed that nanocages with six tetragonal rings and eight hexagonal rings are the most stable of the four possible structures for B12N12 (2). Other theoretical studies confirmed that the fullerene-like structure (BN) with the magic number n = 12 (i.e., B12N12) is the most stable structure. The unique structure of this nanocage, which consists of boron and nitrogen atoms forming a nanometer framework, offers great potential in detecting toxic gases. This material has demonstrated sensitivity to changes in chemical environment, as well as high stability, making it a promising candidate for gas sensor applications (3–4).
Although B12N12 materials have been recognized as having potential in gas sensor development, little work has been done on the addition of other atomic elements with atomic numbers 3–85. The incorporation of B12N12 with other materials has the potential to significantly improve the performance of gas sensors, as well as open up new opportunities to gain new insights in the field of gas sensor nanotechnology (5). However, the implementation of this incorporation requires further research as well as the development of appropriate methods to ensure optimal material integration. One approach to overcome this challenge is to use the prediction of quantum descriptor values using the DFT (Density Functional Theory) method. However, optimal material prediction using conventional methods is often time-consuming and costly [6].
Therefore, it is important to adopt more advanced pattern recognition mechanisms, such as Convolutional Neural Network (CNN), in performing predictions. CNNs have been shown to be effective in recognizing complex patterns in multidimensional data, including material structure data [7]. Using CNN, it is expected to identify patterns related to the quantum properties of B12N12 material, as well as predict its response to toxic gases with higher accuracy and in less time, thereby accelerating the development of more sensitive and responsive gas sensors.

Reference :

1) Javad Beheshtian, Zargham Bagheri, Mohammad Kamfiroozi, Ali Ahmadi, Toxic CO detection by B12N12 nanocluster, Microelectronics Journal, Volume 42, Issue 12, 2011, Pages 1400–1403, ISSN 0026–2692, https://doi.org/10.1016/j.mejo.2011.10.010

2) Frank Jensen, F. Jensen, Hans Toftlund, and H. Toftlund, “Structure and stability of C24 and B12N12 isomers,” Chemical Physics Letters, vol. 201, no. 14, pp. 89–96, Jan. 1993, doi: 10.1016/0009–2614(93)85039-q.

3) Silva, A.L.P., de Sousa Sousa, N. & de Jesus Gomes Varela Junior, J. Theoretical studies with B12N12 as a toxic gas sensor: a review. J Nanopart Res 25, 22 (2023). https://doi.org/10.1007/s11051-023-05667-9

4) Alireza Soltani et al., “Influence of the adsorption of toxic agents on the optical and electronic properties of B12N12 fullerene in the presence and absence of an external electric field,” New Journal of Chemistry, vol. 44, no. 34, pp. 14513–14528, Sep. 2020, doi: 10.1039/d0nj01868f.

5) Aidin Bahrami, A. Bahrami, Mohammad Balooch Qarai, M. B. Qarai, Mohammad Balooch Qarai, and N. L. Hadipour, “The electronic and structural responses of B12N12 nanocage toward the adsorption of some nonpolar X2 molecules: X = (Li, Be, B, N, O, F, Cl, Br, I): A DFT approach,” Computational and Theoretical Chemistry, vol. 1108, pp. 63–69, May 2017, doi: 10.1016/j.comptc.2017.03.018.

6) Adilson Luís Pereira Silva, Adilson Luís Pereira Silva, Jaldyr de Jesus Gomes Varela Júnior, and Jaldyr de Jesus Gomes Varela Júnior, “Density Functional Theory Study of Cu-Modified B12N12 Nanocage as a Chemical Sensor for Carbon Monoxide Gas,” Inorganic Chemistry, Sep. 2022, doi: 10.1021/acs.inorgchem.2c01621.

7) Yuchen Wang, Zhengshan Luo, and Jihao Luo, “Research on predicting the diffusion of toxic heavy gas sulfur dioxide by applying a hybrid deep learning model to real case data,” Science of the Total Environment, vol. 901, pp. 166506–166506, Nov. 2023, doi: 10.1016/j.scitotenv.2023.166506.