Abstract
For SRAMs with high logic complexity, hard defects, design debug, and soft defects have to be tackled all at once early on in the technology development while innovative integration schemes in front-end of the line are being validated. This paper presents a case study of a high-complexity static random access memory (SRAM) used during a 32nm technology development phase. The case study addresses several novel and unrelated fail mechanisms on a product-like SRAM. Corrective actions were put in place for several process levels in the back-end of the line, the middle of the line, and the front-end of the line. These process changes were successfully verified by demonstrating a significant reduction of the Vmax and Vmin nest array block fallout, thus allowing the broader development team to continue improving random defectivity.
