Researchers demonstrated an artificial intelligence system that can find and label 2D materials in microscope images in the blink of an eye. Two-dimensional materials offer a new platform for the creation of electronic devices such as transistors and light-emitting diodes. The family of crystals that can be made just one atom thick includes metals, semiconductors, and insulators.
Many of these are stable under ambient conditions and their properties often differ significantly from those of their 3D counterparts. Even stacking a few layers together can alter the electronic characteristics to make them suitable for next-generation batteries, smartphone screens, detectors, and solar cells. They also can be made using office supplies — atomically thin graphene can be obtained by exfoliating a piece of pencil lead (graphite) with a piece of sticky tape.
Unfortunately, the atomically thin 2D crystals have low fabrication yields and their optical contrasts comprise a very broad range; finding them under a microscope is a tedious job. The researchers have automated this task using machine learning. They used many labeled examples with various lighting to train the computer to detect the outline and thickness of the flakes without having to fine tune the microscope parameters.
The method can be used with many other 2D materials, sometimes without needing any additional data. In fact, the algorithm was able to detect tungsten diselenide and molybdenum diselenide flakes just by being trained with tungsten ditelluride examples. With the ability to determine, in less than 200 milliseconds, the location and thickness of the exfoliated samples, the system can be integrated with a motorized optical microscope.