One of the most tedious and daunting tasks for undergraduate assistants in university research labs is staring for hours under a microscope at samples of material, trying to find monolayers.
These two-dimensional materials—less than 1/100,000e the width of a human hair – are highly sought after for use in electronic, photonic and optoelectronic devices due to their unique properties.
“Research labs hire armies of undergraduates to do nothing but research monolayers,” says Jaime Cardenasassistant professor of optics at University of Rochester. “It’s very tedious, and if you get tired you might miss some of the monolayers or start making misidentifications.”
Even after all that work, labs then have to recheck the materials with expensive Raman spectroscopy or atomic force microscopy.
Jesús Sánchez Juárez, doctoral student at Cardenas Laboratoryhas made life much easier for undergraduates, their research labs, and companies that face similar challenges in detecting monolayers.
The revolutionary technology, an automated scanning device described in Express Optical Materialscan detect monolayers with 99.9% accuracy, outperforming any other method to date.
At a fraction of the cost. In much less time. With readily available materials.
“One of the main goals was to develop a system with a very low budget so that students and labs could reproduce these methodologies without having to invest thousands and thousands of dollars just to buy the necessary equipment,” says Sánchez. Juárez, the main author of the paper.
For example, the device he created can be replicated with an inexpensive microscope with a 5X objective and a low-cost OEM (original equipment manufacturer) camera.
“We are very excited,” says Cardenas. “Jesús has done several things here that are new and different, applying artificial intelligence in an original way to solve a major problem in the use of 2D materials.”
Many labs have attempted to eliminate the need for expensive human-scanned backup characterization tests by training an artificial intelligence (AI) neural network to search for monolayers. Most labs that have tried this approach attempt to build a network from scratch, which is time-consuming, Cardenas says.
Instead, Sánchez Juárez started with a publicly available neural network called Alex Net who is already trained to recognize objects.
He then developed a new process that reverses the images of the materials so that whatever was bright in the original image appears black instead, and vice versa. Inverted images are subject to additional processing steps. At this point, the images “don’t look good at all to the human eye,” Cardenas says, “but for a computer it’s easier to separate the monolayers from the substrates they’re deposited on.”
Conclusion: Compared to those long and tedious hours of scanning by undergraduate students, Sánchez Juárez’s system can process 100 images covering 1 centimeter x 1 centimeter samples in nine minutes with nearly 100% accuracy.
“Our demonstration paves the way for the automated production of single-layer materials for research and industry by dramatically reducing processing time,” Sánchez Juárez writes in the paper. Applications include 2D materials suitable for photodetectors, excitonic light-emitting (LED) devices, lasers, optical generation of spin valley currents, single photon emission, and modulators.
Additional co-authors include Marissa Granados Baez, doctoral student at Cardenas Lab, and Alberto A. Aguilar-Lasserre, professor at Instituto Tecnológico de Orizaba.
Express Optical Materials
The title of the article
Automated 2D material detection system using image processing and deep learning
Publication date of articles
Conflict of Interest Statement
The authors declare no conflict of interest.
Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of press releases posted on EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.
#quest #elusive #monolayers #lot #easier