IBM, Applied Materials, Inc. and the College of Nanoscale Science and Engineering of the University at Albany (UAlbany NanoCollege) recently announced an agreement to jointly develop process modeling technology for manufacturing 22 nanometer (nm) logic and memory chips. The project will combine IBM's semiconductor technology research and development and computer modeling expertise with Applied's semiconductor processing knowledge to develop predictive models that can help minimize process variation, reduce development cost, and improve time to market for 22nm semiconductors.
FinFET transistors -- vertical transistors with fin-shaped silicon channels -- will be used to validate the technology. FinFETs are considered a potential successor to conventional planar transistors for 22nm chips.
Currently the most advanced semiconductors have circuitry at 45 nanometers and larger. Producing circuits at 22nm becomes more challenging since current lithography methods -- the process for creating circuit patterns on silicon wafers -- present physical limitations for critical chip layers. To help overcome these limitations, IBM has led an initiative known as Computational Scaling. Computational-based processes use advanced mathematical techniques, software tools and high-performance computing systems to enable the production of complex, powerful and energy-efficient semiconductors at 22 nanometers and beyond.
"To meet the challenges of 22nm, we need to model the entire physical structure of the transistor, and this is now possible with the advent of ultra-powerful, petascale computer processing. By combining Applied's thin film deposition and etch processing expertise and CNSEs fundamental science know-how with IBM's capabilities, we can extend our modeling beyond lithography to help deliver a more complete and validated manufacturing process," said Gary Patton, vice president, IBM Semiconductor Research and Development Center.
Developing a chip manufacturing process for a new technology node involves complex interactions among multiple process variables, fabrication disciplines, and circuit requirements. With circuit features just a few hundred atoms across, the interactions of each process variable must be taken into account. For advanced technology nodes, this can lead to costly production experiments that consume many thousands of semiconductor wafers. The goal is to develop computational models to perform most of these experiments in a "virtual" laboratory, greatly reducing the need for actual wafer processing.
"The next big scaling challenge is to give these ultra-small transistors consistent geometry and electrical properties which define the speed, reliability and power consumption of a device,"said Hans Stork, Group Vice President and Chief Technology Officer of Applieds Silicon Systems Group. "The key to reducing this variability is to integrate and optimize every fabrication step to precisely construct and repeat the critical transistor geometry. By using Applieds process knowledge to validate and refine IBM's predictive modeling, we plan to bring this new technology to market in a shorter time, with less risk and at a lower development cost than traditional experimental methods."
"This collaboration will extend current research at CNSE's Albany NanoTech to validate fundamental process modeling work for future technology nodes, accelerating the integration of advanced processing and innovative nanoelectronics research and development that is necessary for scaling amid the growing complexity of next-generation transistors, said Richard Brilla, Vice President for Strategy, Alliances and Consortia at CNSE.
The research will be carried out primarily at CNSE's Albany NanoTech Complex in Albany, N.Y., where Applied has a full suite of leading-edge processing tools and a staff of engineers and scientists. Additional modeling and process characterization will be performed at IBM's facilities in East Fishkill and Yorktown, N.Y., Applied's Maydan Technology Center in Sunnyvale, Calif. and the Computational Center for Nanotechnology Innovations (CCNI) at the Rensselaer Polytechnic Institute in Troy, N.Y.
