Off-campus UMass Amherst users: To download dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.
Non-UMass Amherst users, please click the view more button below to purchase a copy of this dissertation from Proquest.
(Some titles may also be available free of charge in our Open Access Dissertation Collection, so please check there first.)
Managing lithographic variations in design, reliability, & test using statistical techniques
Much of today’s high performance computing engines and hand-held mobile devices are products of aggressive CMOS scaling. Technology scaling in semiconductor industry is mainly driven by corresponding improvements in optical lithography technology. Photolithography, the art used to create patterns on the wafer is the heart of the semiconductor manufacturing process. Lately, improvements in optical technology have been difficult and slow. The transition to deep ultra-violet (DUV) light source (193nm) required changes in lens materials, mask blanks, light source and photoresist. It took more than ten years to develop a stable chemically amplified resist (CAR) for DUV. Consequently, as the industry moves towards manufacturing end-of-the-roadmap CMOS devices, lithography is still based on 193nm light source to print critical dimensions of conman, 32nm and likely 22nm. Sub-wavelength lithography creates a number of printability issues. The printed patterns are highly sensitive to topographic changes due to metal planarization, overlay errors, focus and dose variations, random particle defects to name a few. Design for Manufacturability (DFM) methodologies came into being to help analyze and mitigate manufacturing impacts on the design. Although techniques such as Resolution Enhancement Techniques (RET) which involve optical proximity correction (OPC), phase shift masking (PSM), off-axis illumination (OAI) have been used to greatly improve the printability and better the manufacturing process window, they have not been able to perfectly compensate for these lithographic deficiencies. DFM methods were primarily devised to predict and correct systematic patterning problems that arise during manufacturing. Apart from systematic errors, random manufacturing variations may occur during photolithography. This is where a statistical approach to modeling of error behavior and its impact on different design parameters may prove to be effective. For example, by incorporating statistical analysis to parameter variation, an effective, non-conservative design can be obtained. IC manufacturing yield is the foremost measure that determines the profitability of a given semiconductor manufacturing process. Early prediction of yield detractors is an important step in the design process. Such predictions are based on models that mimic the behavior of the underlying manufacturing process. Success of yield prediction is based on quality of models. The models must capture all physical phenomena and yet be efficient for computation. In this work, we present a lithography-based yield model that is computationally practical for use in the design process. The work also provides a methodology to perform statistical lithography rules check to identify hot spots in the design that can contribute to yield loss. Yield recovery methods aimed at minimally modifying the design ultimately produce more printable masks. Apart from IC manufacturing yield, ICs today are vulnerable to various reliability failures including electromigration (EM), negative bias temperature instability (NBTI), hot carrier injection (HCI) and electro-static discharge (ESD). Though such reliability issues have been examined since the beginning of CMOS, manufacturability impacts have created a renewed interest in analyzing them. This dissertation work introduces the concept of Design for reliable manufacturability (DFRM) to consider the effect of linewidth changes, gate oxide thickness variations and other manufacturing artifacts. A novel Litho-aware EM calibration and analysis has bee shown in this work. Results indicate that there is a significant difference in EM estimation when litho-predicted layouts are considered during analysis. DFM has always looked at linewidth and material thickness variation as detractors to the design. However, the increase in such variations with technology scaling is inevitable. Part of this dissertation aims at utilizing these fluctuations to improve manufacturing test quality. Test structures sprinkled all over the wafer encounter varying process fluctuations. This can be harnessed to predict the current lithographic process corner which will later be used to choose the test pattern set that results in maximum fault coverage. In summary, the objective of this dissertation is to consider the impact of subwavelength lithography on printability and the overall impact on circuit reliability and manufacturing test development.
Sreedhar, Aswin, "Managing lithographic variations in design, reliability, & test using statistical techniques" (2011). Doctoral Dissertations Available from Proquest. AAI3445186.