Graphene Image Analysis

Hello!

My name is Jessica and I am a rising sophomore at Mudd. I am currently studying 2D materials with Professor Breznay.

The study of 2D materials such as graphene, stanene, and hexagonal boron nitride is coming into prominence because of the potential it holds for applications in many different fields. Their unique and mostly unresearched mechanical and chemical properties are thought to stem from the weak bonds holding together tightly bonded sheets of repeated subunits consisting of single atoms or molecules. We are focusing on studying graphene because it has been the most widely researched and is the simplest of the 2D materials both to synthesize (using the scotch tape method) and in its basic structure. During the school year, we worked on refining the process of exfoliating graphene onto a silicon substrate and studying the samples under an optical microscope.

The optical microscope works by shining a beam of white light into a beamsplitter through an objective, which then reflects off of the object and into the eyepiece. As the light hits the object, some of it gets reflected back and some of it gets absorbed. Different materials or differently shaped and sized objects (such as single or multiple graphene layers) have wavelength dependent absorption, thus appearing to the eye as a different color.

Microscope diagrams

A picture of the Motic Microscope BA310 (left); A schematic of our graphene sample (right)

 

Over the course of the summer, we have worked on developing a MATLAB code to process the images taken on the microscope. Part of that process involved identifying the visible differences between the substrate, monolayer, and bilayer sections of the sample. 

Lab Optical Microscope Setup

(On the monitor is a sample picture with the microscope to the right)

The picture below shows a cropped 1x101x3 section of a sample image, or a 101 pixel-long line, running through a substrate, then monolayer, then bilayer, then substrate. The RGB values are then plotted on separate graphs, where the x-axis is the location of the pixel along the cropped line (0-101) and the y-axis is the pixel intensity value along the R, G, and B channels respectively.

Labeled sample image with cropped 1x101x3 line (left); Plotted RGB data (right)

Comparatively, between the three channels, the R values are where the layer differences are most apparent.

R Value graph with average substrate, monolayer, and bilayer intensities

Through repeated tests with other sample pictures, we conclude that pictures with similar lighting to this sample have the clearest distinction between R pixel intensity values. The difference between each layer is around 10-20 pixel values, averaging at roughly 15. 

Using this information, we can develop a code that can identify significant monolayer or bilayer sections of the sample by identifying pixels that are ~15 or ~25 R pixel intensity values above a calculated average substrate value. By also utilizing different techniques such as median filtering and circularity tests, we will hopefully be able to create a more accurate automated code to process future sample images.

Jessica Santosa (jsantosa@hmc.edu)

Edited by Prof. Breznay and Inq Soncharoen