Global Foundries has invested over eight billion dollars in its new Fab 8 campus in Saratoga County, NY, now America’s largest leading-edge semiconductor foundry. There, Global Foundries seeks to develop next-generation chip technology that is smaller, faster, and more energy efficient. But, state-of-the-art infrastructure alone cannot drive innovation at this level. An intellectual power base is required–expertise in physics, chemistry, biology, materials science, and engineering–and expertise in their convergence at the nanoscale.
So, Global Foundries turned to HPCNY–they wanted predictive engineering tools to help drive future innovation. Together with IBM and Rensselaer scientists, they studied ways to apply HPC to modeling molecular electronics.
The problem they sought to address: The limits of existing semiconductor technology. The demand for more computations per second and further miniaturization comes with performance degradation of the underlying interconnects and transistors. For example, copper interconnects, the channels that provide power and carry signals between chip components, have limitations such as low conductivity at nanoscale. Since modern CMOS chips literally use miles of interconnects, this acts as a bottleneck. New technologies and new materials must be developed to overcome this and similar challenges.
HPCNY, with RPI’s Dr. Saroj Nayak and his research group, are at the center of this work. These computational scientists relied on CCNI’s Blue Gene L and Q for system-scale simulation of quantum effects. At nanometer scales, electronic structure-based simulations help solve the problems of molecular placement and interactions. Additional computations provide real-time high-speed visualization and analysis of the physical effects. The software tools they used, some hosted on hpc-ny.org, included a combination of the groups’ own code, open-source codes such as NEMO 3-D and Quantum ESPRESSO, CPMD, Abinit and commercial codes such as Gaussian and VASP.
Global Foundries and Rensselaer gained valuable insight through the use of HPC on these nanoelectronics research issues. By harnessing the best of New York’s high performance computing and intellectual power, research such as this can chart the path of advancement for the semiconductor industry, and its customers.