P. M. Solomon
IBM, T. J. Watson Research Center, Yorktown Heights, NY, 10598, USA
At this date of writing, we should be approaching the end of CMOS as the premier IT logic technology with new “beyond CMOS” nanotechnologies waiting in the wings. However, CMOS is surging ahead, with noteworthy modifications, while its putative rivals have not materialized. The rivals with self-described revolutionary impact have barely progressed to the single device level and rarely to the level of simple circuits and never to CMOS-competitive switching speeds. Here we examine what went wrong and whether there is any path to a true post-CMOS logic technology.
A previous article,1 part of this series of conferences dating back 20 years, reviewed a number of nanodevices that had the possibility of going beyond CMOS in their ability to achieve high performance at much lower switching energies. At the time, CMOS appeared to be rapidly approaching its scaling limits. Today, 6 years later, scaling of gate lengths has slowed down, yet the technology has continued aggressively to higher densities and improved functionality.2 3 Competing nanotechnologies have advanced in a different direction – not toward commercialization, in the main, but rather toward the application of novel physical principles to logic devices. The commercialization threshold is lower for memory4 devices, where novel materials offer data retention capabilities at densities beyond CMOS, ...