By using computer vision and machine learning, computers can design and manufacture miniaturised detectors based on nanowires. This could make detectors more affordable, more reliable, and be used for a larger range of applications from security screening to medical diagnosis.
Nanowires are very thin wires, so thin that you would not be able to see them with the naked eye. To give you some scale, you could fit a million of these nanowires in a single strand of human hair. However, unlike typical wires that you see, nanowires are made of semiconductors, which are a special kind of material that could conduct or block electricity depending on how you control them.
So why do we want to work with these nanowires? Due to their one-of-a-kind shape and composition, nanowires offer unique properties which make them suitable for building Terahertz (THz) detectors. THz detectors are detectors that capture and measure electromagnetic waves based in the THz region, which is the region of the spectrum between microwaves (like those that power the ones in our kitchen) and infrared radiation (radiation that we feel as heat, or those used in TV remotes and night-vision goggles). THz waves can penetrate many non-conductive materials such as textiles, paper and plastic, and are non-ionising, meaning they can be used in anything from airport security screening to wireless communication between our smartphones.
However, as nanowires are so small, being able to detect, pick up, and manipulate them into THz detectors is challenging. Imagine Lego building using human hair instead of Lego blocks! Currently, researchers would take days, and maybe even weeks, just to make a single device using one or two nanowires. Not only that, but given the difficulty, it is not uncommon that mistakes will be made, and the whole process will have to be restarted.
For my PhD, I will create a system where researchers will be able to automatically build THz detectors. By using computer vision, computers will be able to detect these nanowires and a rate and accuracy greater than any human. After detection, the computers would design the ideal circuit based on the different locations and orientations of these nanowires. This method would allow us to make 1000s of devices on the same chip in just a matter of days, saving time, improving the accuracy and number of devices made, and being more efficient with our resources.
With this breakthrough, not only can we make these THz detectors cheaper and more readily accessible for use in applications mentioned before, but we could also unlock new device designs and transform the way we use THz detectors. One such example is spectral imaging, a technique which could be used in different fields, like helping farmers monitor the health of crops, or in satellite imaging to study the Earth’s surface, and even in medical diagnosis, to help with early detection of diseases like skin cancer.
NanoDTC PhD Student, c2022