The research division of Fujitsu has used advanced supercomputing techniques to successfully simulate the electrical properties of a 3,000-atom nano device, a threefold increase from previous research. The company notes that at the nanoscale level, minute differences in the local atomic configuration can have a major impact on the electrical properties of a device. The challenge is such that it requires the first-principles method of calculation to accurately calculate the behavior of each atom. This method determines physical properties from the basic laws of quantum mechanics that govern atoms and electrons.
Computational simulation is well regarded for supporting a development process that is faster and more cost-effective than physical experimentation. But in this case, there’s a catch. When the first-principles method is applied to electrical property forecasting, the computations involved are so large that forecasts are generally limited to the order of 1,000 atoms. At this scale, only channel regions – the pathways for electricity – are able to be calculated. A more desirable simulation will incorporate interactions with thousands of adjacent electrodes and insulators – which are understood to greatly affect electrical properties. Until now, this has been considered an intractable problem.
Fujitsu Labs came up with a new technique based on massively-parallel supercomputing technology developed by the Japan Advanced Institute of Science and Technology (JAIST) and the Computational Material Science Initiative (CMSI). The new calculation technique reduces memory requirements while maintaining precision. With this approach, it is possible to calculate the electrical properties, not only of individual nano device components, but of the interactions between these components. The advancements have already enabled a successful 3,000 atom scale application, and the researchers anticipate that this development will make way for faster practical implementations of nano devices.
As silicon-based devices come against the limits of miniaturization, there is increased interest in developing new materials and types of structures that can increase speed and energy-efficiency. Nanotechnology offers a promising path to sustaining Moore’s Law-type returns.
“This technology, being capable of modeling the electrical properties of a 3,000-atom nano device, was used to discover the electrical properties of a nano device that included interactions with its environment, making a significant step toward the design of new nano devices,” notes the official release announcement.
As ever-more massive parallel computing technology and more performant supercomputers become available, Fujitsu will continue to experiment with larger-scale and more efficient calculations. Going forward, Fujitsu is pursuing nano device design through total simulations of nano devices at the 10,000 atom scale.