Daniil Ivannikov - Final Presentation.pdf

Two-dimensional (2D) materials are a class of nanomaterials that consist of a single- or few-layers of atoms and possess exceptional physical and chemical properties. Such unique properties of 2D materials made them a focal point of research with the goal to use them in production of electronic and nano-electronic devices such as transistors, sensors of various applications, electrodes, etc. There are several ways to make 2D materials, and one of the easiest and widely used research methods is the Mechanical Exfoliation, also known as the Scotch Tape method, where flakes of various size and thickness are randomly split from bulk materials and transferred to a chosen substrate using adhesive tape. Before flakes can be further utilized, they need to be identified on the substrate and that process is usually done manually using optical microscopy as a starting tool and other lower-throughput methods for more reliable final identification. Manual search and identification of flakes is a time consuming and involved process that requires expertise to be done efficiently. Furthermore, it gets less effective the smaller the flakes. There have been efforts to automate flake search and characterization using machine learning (ML) [1,2]. There are examples of ML algorithms that are extremely accurate in identifying flakes. However, ML has some drawbacks that prevent it from being a widely applied tool for 2D materials flakes, especially for new material types or new transfer methods: lack of training data for the model, variable samples/substrates, high requirement for computing power or time. Proposed solution to the efficient recognition and sorting of 2D material supported by the automation is to use a general non-ML algorithm that relies on some user input and general visual properties of 2D material flakes such as transparency and color difference between single and multiple layers of material to identify and possibly characterize flakes with minimum computing power/time required.