Editorial S
E
C
T
I
O
N
S
Faster, Cheaper, and Smarter Today’s tech-savvy consumer has insatiable demands for tons of functionality, a mind-boggling quantity of memory, seemingly unending battery life, and low, low cost—and never one without the other. "A tall order?" you might ask. Not as long as the semiconductor industry continues its steady progress along the curve of Moore’s Law. Recently, I bought my son a Nintendo® Game Boy® for his birthday. This ubiquitous hand-held gadget—the I-must-have-it toy for kids (and even some adults)— probably best exemplifies the technologies that have to converge to meet consumer demands. It has to solve three problems simultaneously for my son to be even reasonably happy. First, Mario—the main character in the Super Mario Bros.® game—has to jump in complex scenes (requiring faster transistor switching speed). Second, he has to remember to jump in many different scenes (more memory). Next, he has to continue to jump for hours on end to entertain and alleviate my son’s boredom (longer battery life). On top of all this, Mario has to fit in a small enough form factor that my son can take on the family vacation (or fit in his pocket to hide from his teacher), AND the cost has to be commensurate with his performance at school and the shrinking family budget! While design rule scaling solves the problem of switching speed and memory, it doesn’t help with the issues of power consumption and the skyrocketing capital costs associated with moving to a new technology node. With innovation being the semiconductor industry’s middle name, many leading-edge device manufacturers have turned to new materials and device architectures as a solution to meeting the need for greater speed, lower power consumption, and even lower cost. To minimize leakage current, for example, many device manufacturers are exploring switching from oxide or oxynitride gate dielectrics to new high-k materials like hafnium oxide. However, new high-k and metal gates pose very significant manufacturing challenges because they require incorporating a whole new set of materials in the most critical part of the device technology while, at the same time, avoiding the unwanted effects of these immature
4
Summer 2004
Yield Management Solutions
materials on device yield. Since the 1990s, the adoption of these materials has been escalating at a rapid pace, at approximately ten new materials per decade. By 2010 it is anticipated that over 35 new materials would be in usage. In the face of narrowing market windows and accelerated product lifecycles, our customers are increasingly seeking competitive advantage through faster development and time-to-market. But of all of the possible new materials and processes to consider, what are the criteria by which chip manufacturers determine which materials make it onto their technology roadmaps? It’s not enough that these new materials must meet their raw performance requirements. They also have to be manufacturable and cost-effective. In pursuit of the dual goals of being technology leaders and low-cost producers, many leading-edge customers are no longer waiting until pilot production to make decisions regarding manufacturability and profitability. They are integrating process control into the decisionmaking process in development, and new materials and processes are being qualified for performance, cycle time, inspectability and yield. With an increased burden on process control, IC manufacturers will need to rely on new inspection and metrology solutions that possess all the attributes of a Swiss army knife to help drive down development cycle time and manufacturing cost, while maintaining very high device yields. Nowhere is there a more paradigm shift than at the very foundation of the semiconductor device—the starting substrate. Engineered substrates like silicon-on-insulator (SOI) and strained SOI (sSOI) are gaining traction because they offer higher switching speed at lower power consumption, a critical need for mobile communication infrastructure. However, the unique multi-layer properties of these materials that help enable their improved performance capabilities also create significant process control challenges. New materials require new sets of parameters to be measured and controlled.