Semiconductor device simulator achieves huge increase in calculation speed!

Semiconductor device simulator achieves huge increase in calculation speed!

More than one thousand times faster!

Dec 10, 2013

A JST Strategic Basic Research Program , under the leadership of MORI Nobuya (Associate Professor) and Gennady Mil'nikov (Specially Appointed Researcher, Graduate School of Engineering, Osaka University), a group of researchers developed a semiconductor device simulator enabling calculation of the properties of a device at the atomic level.
In order to overcome technological and economic limitations of current semiconductor ICs, a variety of devices with new structures and materials have been proposed. In addition to these, simulation of device performance has become necessary in order to improve the efficiency of finding materials appropriate for practical use.
In addition to quantum confinement effects and tunneling effects in nanoscale devices, their properties are also affected by atomic structure. For example, atomic disorder in silicon and random discrete dopant distribution affect device properties. That's why it is very difficult to predict performance of devices at the nanometer scale. Furthermore, all current methods proposed require a lot of time for the complex calculations, making it difficult to obtain results quickly when designing circuits.
However, by using a non-equilibrium Green's function (NEGF) method, this group investigated only fluctuations due to atomic disorder and achieved a large increase in speed of calculation. Whereas it normally took more than a year to predict performance of semiconductor ICs at the atomic level, this group's new numerical model enabled calculation in a couple of days -- even with an ordinary personal computer. Thus, the time necessary for calculation was reduced to less than one thousandth of the conventional computation time.

This group's achievement will be used for designing ICs consisting of nano devices with new structures of nanowires and nanopillars. Simulation results will contribute to making design guidelines for increasingly miniaturized semiconductor ICs.
Abstract
Fluctuations of device characteristics due to random discrete dopant (RDD) distribution are numerically investigated in ultra-small Si nanowire transistors. Kinetic Monte Carlo process simulation is performed to obtain realistic RDD distributions, whose effects on the transport characteristics are then analyzed by using a non-equilibrium Green´s function (NEGF) method. Fluctuations due to atomic disorder near the Si/SiO2 interface are also investigated by performing molecular dynamics oxidation simulation for realistic atomic structure models and NEGF device simulation for transport characteristics.

Figure 1

Figure 2

Related link

Technical Glossary