Graphene image analysis and artificial intelligence model development

Purpose and goal:

Swedish graphene companies are scaling up their graphene product and it is highly expected to have effective in-line quality control method to support high volume, reliable and repeatable graphene production. This will save manufacturers vital time, money and establish a competitive advantage in the growing market for graphene. The aim of the project is to bring a high-throughput, low-cost, general imaging technique that allows accurate and quantitative evaluation of graphene flakes. This will be achieved by combining automated optical microscope and deep learning algorithm.

Expected results and effects:

In this project, we are going to prepare the samples according to the standard, and apply optical microscope to obtain high quality and large quantity images. The developed deep learning algorithm will address the challenges against changes in optical microscopy conditions. It is expected this method is robust and will provide a generalized 2D material detector that does not require fine-tuning of the parameters.

The TRL will reach 3-4 when the project is finished and it will promote the technological maturation of graphene product in Sweden. Approach and implementation:

This project covers sample preparation and characterization, image processing and analysis, and deep learning method development. Finally, the developed code will be transferred to the industry partners. Monthly meeting will be held to monitor the project progress, and we will educate the industrial partners on basic machine learning knowledge. The microscope suppliers will be contacted to follow the most advanced hardware and software technologies and ready for future application.