Semiconductor randomness (stochastics) error-based solution provider Fractilia pointed out that in the most advanced process stages, manufacturers lose up to $ billion per wafer factory due to uncontrolled random pattern changes, resulting in lower yields and delays in production processes. These impactful changes are "random" and have now become the maximum resistance to achieving expected yields in the high-volume manufacturing (HVM) stage.
In this regard, Fractilia's analysis brings a complete solution to blue map, helping the industry recover these originally unrealistic values through combined precise measurement, probability-based process control and design strategies with random thinking. Fractilia Technical Director Chris Mack said that random changes have caused advanced process technology to fail to produce smoothly, and the delay caused semiconductor industry losses of up to tens of billions of dollars.
However, traditional process control methods cannot effectively solve these random effects. Stochastics gaps require completely different approaches, and component manufacturers also need to verify and introduce these new approaches to successfully apply advanced process technology to large-scale production. In fact, randomness limits the growth of today's electronic industry.
Fractilia said that there is a gap between the boundary sizes that can be successfully patterned during the R&D stage and the boundary sizes that can be stable in terms of production to meet previous expected yields. This resolution gap mainly comes from random variations, that is, random variations caused by molecules, light sources, and even atoms in materials and equipment in semiconductor microscopes. Unlike other forms of process variations, random variations are inherent characteristics of the materials and techniques used in the process, so they must be solved using a chance analysis different from the current process control method.
Mack emphasized that Fractilia saw that customers produced high-density structures of only 12 nanometers during the R&D stage, but once they entered the production stage, random errors would affect yield, efficiency and reliability, and could not achieve acceptable standards. In the past, random variations had little impact on yield yield, because random effect was relatively small at that time, and the chance of random defects causing yield loss was also low. However, with the application of extreme ultraviolet light (EUV) and high-number pore extreme ultraviolet light (High-NA EUV) technologies, the random variations account for a higher proportion of the tolerance of advanced process errors.
Fortunately, random gaps are not fixed and changeable. Fractilia analyzes the causes of random gaps in detail and proposes solutions, including component design with random thinking, material improvement and process control with random thinking, etc. Mack further pointed out that random gaps are a common problem for the entire industry, but we can resolve and control this problem as long as we start with precise random measurement technology. At present, this solution is not only used for the production of logical chips, but is also further used on DRAM memory chips.