Abu Dhabi student wins place in IBM

A PhD student from the capital has become the first person from the Middle East to earn a sought-after international research internship. Sachhidh Kannan, 27, is one of four students out of 80 qualified applicants from the Middle East, Africa and central and eastern Europe chosen to participate in IBM's Great Minds Initiative. The six-month placement in Zurich, beginning in June, will allow him to work alongside world-class scientists in research and innovation. The Indian student has been researching a new memory design for electronic chips, called the memristor, since June at New York University - Abu Dhabi. Memristor technology, also called resistive random access memory (RRAM), will be a more efficient alternative to existing memories, such as static random-access memory semiconductors (SRAM), according to Mr Kannan's faculty adviser, Ozgur Sinanoglu. "Current technology is reaching its limits. We cannot scale any further and things cannot get smaller, so we're looking at these new devices to make things faster and more power-efficient," said Prof Sinanoglu. The memristor, first theorised in 1971, can be used in any device that holds memory - such as mobile phones, cameras and computers - and will hold up to four times more memory than the current chips. "We're basically packing more stuff into the same area," said Mr Kannan. Similar devices are already on the market but Mr Kannan believes there are still "road blocks" to overcome before they are ready to be manufactured on a larger scale. They are expensive to produce and the manufacturing process to make something so small is very complex. The devices on the chip are in the order of nanometres - a billionth of a metre - so there is a "big chance" that something could go wrong, said Prof Sinanoglu. The small size means new types of defect mechanisms are being created during the fabrication of the memristor, which is why their research is focusing on two areas - increasing the capacity of the memory size and creating tools to fix the problems. "We are trying to come up with automated techniques - or built-in circuits - to model, detect, locate and bypass failures," Mr Kannan said. They are doing so by taking existing traditional techniques used for regular memory chips and remodelling them to specifically catch the memristor defects. With this research, they hope to create a testing technique that will cut costs by managing to locate the chips' faults in the shortest possible time. Mr Kannan estimates it will take two or three years before the memristor replaces current technology. Prof Sinanoglu said the IBM internship would give his student "new knowledge and information, which will advance our research even further".