Machine Learning for Graphene Detection & Perovskite Device to Probe Transport and Optical Properties

Ever since graphene was isolated in 2004 [1], it has attracted a lot of attention from the scientific community due to its fascinating properties. For research purposes, graphene is obtained via mechanical exfoliation of graphite, which frequently requires a subsequent hand-operated search with an optical microscope over the area of interest, usually a Si/SiO2 substrate of about 1 cm2, while graphene flakes typically occupy just a few hundreds of μm2.
An approach to speed up this procedure can rely on a machine learning algorithm to perform the visual recognition task, particularly Neural Networks (NNs). In this work, inspired by previous implementations [2], we develop an automated end-to-end pipeline to fine-tune a NN capable of segmenting graphene in images according to our specific needs and microscope’s characteristics.

Moreover, in this work, we combine graphene and a 2D Ruddlesden-Popper (RP) perovskite to make a device for probing optical and transport properties. We make use of nanofabrication techniques such as e-beam lithography, thin metal film evaporation and dry stacking. Additionally, we investigate how to perform cuts in graphene using an Atomic Force Microscope (AFM) [3]. RP perovskites are emerging as promising materials for optoelectronics since they show compelling properties such as strong quantum confinement, large exciton binding energy, strong electron-phonon coupling and a tunable band gap by either modifying the layer number or chemical compositions [4].

[1] K. S. Novoselov et al., Electric Field Effect in Atomically Thin Carbon Films. Science. 306 (2004), pp. 666–669.
[2] H.-Y. Siao et al., Machine Learning-based Automatic Graphene Detection with Color Correction for Optical Microscope Images (2021)
[3] H. Li et al., Electrode-Free Anodic Oxidation Nanolithography of Low-Dimensional Materials. Nano Lett. 18 (2018)
[4] H. Wang, C. Fang, H. Luo, D. Li, Recent progress of the optoelectronic properties of 2D Ruddlesden-Popper perovskites. J. Semicond. 40 (2019)

If you would like to meet with Renata during her visit, please contact Susan LaMoreaux.