EUVIMEDEUVIMED entdecken
Crossref

CHARACTERIZATION OF ARTIFICIALLY GENERATED 2D MATERIALS USING CONVOLUTIONAL NEURAL NETWORKS

Cahit Perkgöz, Mehmet Zahit Angi

Eskişehir Technical University Journal of Science and Technology A - Applied Sciences and Engineering · 2022

DOI: 10.18038/estubtda.1149416 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Eskişehir Technical University Journal of Science and Technology A - Applied Sciences and Engineering, 2022. Verfasst von Cahit Perkgöz und 1 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Gefördert von: Eskişehir Teknik Üniversitesi.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Two dimensional (2D) materials have attracted many researchers due to the high-performance of the devices produced by these materials. There are different methods to produce 2D materials such as wet chemical synthesis, chemical vapor deposition (CVD), molecular beam epitaxy, atomic layer deposition, pulsed laser deposition (PLD), all of which require hours during the processes. Once the 2D structures are obtained, their properties including their defects should be revealed by different characterization tools. Characterization process also requires time and expertise. In this respect, deep learning methods such as Convolutional Neural Networks (CNN) can be a solution for the practical and rapid classification of the produced samples. However, there is not enough number of samples in most of the research laboratories because of the above-mentioned long experimental processes. This work presents the performance of a CNN algorithm using artificially created images of MoS2, a commonly studied 2D semiconductor with a high potential in different electronics applications. The synthetic optical microscopic images including normal and defected MoS2 flakes are generated by the intensities of light incident on different materials using Fresnel Equations. A deep CNN algorithm is constructed to detect the normal and defective samples. As a result of the experiments, an average of 88.9% accuracy was obtained. These results can be interpreted that CNN can be used in the future for the characterization of two-dimensional materials with a sufficient number of real images.

Förderung

  • Eskişehir Teknik Üniversitesi

Zitieren

APA

Perkgöz, C., & Angi, M. Z. (2022). CHARACTERIZATION OF ARTIFICIALLY GENERATED 2D MATERIALS USING CONVOLUTIONAL NEURAL NETWORKS. Eskişehir Technical University Journal of Science and Technology A - Applied Sciences and Engineering. https://doi.org/10.18038/estubtda.1149416

Vancouver

Perkgöz C, Angi MZ. CHARACTERIZATION OF ARTIFICIALLY GENERATED 2D MATERIALS USING CONVOLUTIONAL NEURAL NETWORKS. Eskişehir Technical University Journal of Science and Technology A - Applied Sciences and Engineering. 2022. doi:10.18038/estubtda.1149416

BibTeX
@article{Perkgz2022,
  author = {Perkgöz, Cahit and Angi, Mehmet Zahit},
  title = {CHARACTERIZATION OF ARTIFICIALLY GENERATED 2D MATERIALS USING CONVOLUTIONAL NEURAL NETWORKS},
  journal = {Eskişehir Technical University Journal of Science and Technology A - Applied Sciences and Engineering},
  year = {2022},
  doi = {10.18038/estubtda.1149416}
}
Kostenlos starten →