LONDON – Researchers from France and Switzerland have shown that much higher stresses than previously applied can improve the hole mobility in p-type silicon beyond what had previously been thought possible. The results could change the way chip companies seek to engineer transistors at the leading edge, one of the researchers claimed.
Chip manufacturers already use strain to improve the electron and hole mobility in n- and p-type transistors. The use of stress to produce higher performing devices had previously thought to have reached a limit and that the use of exotic materials such as compound semiconductor layers or carbon sheets would be the next step in improving transistor and IC performance.
Now the researchers – from the Ecole Polytechnique, CNRS, Institut d’Electronique, de Microe´lectronique et de Nanotechnologie and the University of Geneva – have shown p-type silicon can perform under much higher stress than previously thought possible.
While the conductance of n-type silicon saturates at about 45 percent of its zero-stress value, as expected, p-type material conductance increases above the predicted limit without any significant saturation even at 3-gigapascal's uniaxial saturation, the researchers reported.
Theoretical predictions had indicated that the performance gain to be had using this technique levels out at around 1-GPa of applied compressive stress. This is around the level which process engineers can currently produce when fabricating commercial strained-silicon devices. It is also a reason why further investment in the technology had not been considered worthwhile, at least until now.
In p-type wafers, no saturation of the mobility was observed up to 3-GPa, even though these unprecedented mechanical stresses are well above the predicted onset of mobility saturation in p-type silicon. This observation, explained using full band electronic structure calculations, suggests that further investment in strained-silicon technology is justifiable, the researcher said.
The research was reported in Physical Review Letters earlier this month
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.