Device Bridges Optical and Electrical Computing, Making Optical Computing Feasible


Author :Justin Brunnette

Category: IT News

In a world’s first, university researchers have created a nanoscopic device that bridges the gap between optical and electrical computing, making optical computing feasible. Developed by researchers at the University of Oxford’s Harish Bhaskaran’s Advanced Nanoscale Engineering group, their device can be used to program with electrons and photons by providing a solution to transition between the two. This has proven to overcome bandwidth limitations, provide energy efficient memory and merge computing and communication technologies.
The potential for computing at the speed of light has been attractive but it has had major obstacles to prevent it from being practical. photons tend to be much larger wavelengths than electrons making their conversion very difficult. Because of this, electrical chips can be made down to the nanoscale while optical chips need to be relatively large.
The researchers need to find a solution to scale down light to the nanoscopic scale. Their solution is something called the “Plasmonic nanogap”. They took a type of phase change material that can provide both electrical and optical modulation, and combined it with something called “plasmonics”.
Plasmonics are particles, usually a metallic material, with a specific size and shape that allow it to interact with light waves in a useful way. This combination gives the device both high electrical conductivity and high plasmonic resonance at certain light wavelengths.
This allowed the researchers to compress light into a nano-size volume in a process called “surface plasmon polariton”. This compression increased the energy density of the system. This increased energy density is what was needed to overcome the gap between optical and electrical signals  and creating a full mixed-mode operation.
This means that their device can program in both light or electrical signals. Their system has also been proven to be both stable and nonvolatile, which allows for practical uses feasible.
This can have massive potential in innovations in computing where high efficient memory is needed such as in-memory computing and multilevel data storage. Artificial intelligence research will also gain a huge boost in this as there are situations that call for higher performance that our current computing infrastructure can provide. There has been a lot of noise in the media about the end of Moore's law but this may be the next generation innovation to keep the law alive.
Original Article: