Machine learning algorithms enable progress in 2D material identification and inspection

Recently, Wang Jun, a researcher at the Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, made progress in realizing two-dimensional material layer identification and physical property detection based on machine learning algorithms, and a related article was published in Laser and Photonics Review (Laser) under the title Thickness Determination of Ultrathin 2D Materials Empowered by Machine Learning Algorithms & Photonics Reviews)。

Since the discovery of graphene, a large number of new two-dimensional layered materials have been gradually discovered and prepared, and have now become a huge family covering insulators, topological insulators, semiconductors, semimetals to superconductors. In general, the number of layers of two-dimensional materials is important for adjusting the performance of nanoelectronics and optoelectronic devices, and it is often necessary to determine the optimal thickness of the target sample before further physical research or device fabrication is achieved. At present, after obtaining optical images or spectral information through optical technology, subsequent data processing often relies on the expertise of researchers and is greatly affected by personal experience and subjective factors.

In recent years, artificial intelligence has changed many aspects of modern society, and as its most important subfield, machine learning has brought new development opportunities and solutions to traditional research fields such as physics, chemistry, and materials science by collecting and analyzing data to predict the behavior of complex systems and build models to solve problems. For example, optical images, as the most easily obtained dataset in the laboratory, is a simple method to solve the requirements of high throughput and real-time layer recognition, and machine learning algorithms can extract the basic features in the images and establish decision-making models, and are well applied to different optical systems to meet the requirements of different users for automatic optical recognition and characterization. In addition to optical images, machine learning algorithms can also accurately and efficiently analyze spectral data, which can not only use spectral feature information to quickly obtain the required sample thickness, but also effectively solve the adverse effects of test data errors between different experimental platforms from the intrinsic characteristics of materials. More importantly, these optical solutions enabled by machine learning algorithms significantly promote the establishment of unified, fast, low-cost, non-destructive measurement methods and standards based on data, which in turn strongly promotes the industrial-grade application of two-dimensional materials.

This paper systematically summarizes the development opportunities and problems faced by the deep integration of traditional optical technology and machine learning algorithms, and puts forward the potential risks and challenges brought by the diversity of detection objects, the difference of physical properties, the instability of the test environment, the susceptibility of optical technology and the accuracy of related algorithms to the formulation of cross-laboratory standards. Machine learning algorithms will bring profound changes to the traditional research methods of two-dimensional material thickness determination, gradually liberate manual labor from the existing cumbersome material characterization process, and help promote the rapid development of research and gradually move towards practical application. (Source: Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences)

Related paper information:

Machine learning algorithms enable 2D material identification and detection

Special statement: This article is reproduced only for the need to disseminate information, and does not mean to represent the views of this website or confirm the authenticity of its content; If other media, websites or individuals reprint and use from this website, they must retain the “source” indicated on this website and bear their own legal responsibilities such as copyright; If the author does not wish to be reprinted or contact the reprint fee, please contact us.

Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button