[Session 3: Reliability] An Effective Approach to Monitor Potential Reliability Failure in the Advanced Devices
As semiconductor devices scale down in size, the sensitivity of reliability to process variation increases. This is due to the high reliability requirements for advanced system applications, the narrowing process margins induced by scaled pitches and the high sensitivity of devices to material elements. In order to mitigate these challenges for advanced devices, reliability improvements are required at the circuit and process levels.
In search of reliability improvements at the circuit level, circuit designers continue to optimize error correction code (ECC), redundancy and bias temperature instability (BTI). At the process level, however, specific materials or process chambers can degrade reliability of the devices in program disturb, induced by additional charge traps in the adjacent cells and poor data retention due to contamination by killer elements and complex structures, although they may be required from a process integration perspective.
It is difficult to effectively detect killer elements during the process, which can create or accelerate reliability failure. Several killer elements are known to cause reliability degradation: F (fluorine), Cl (Chlorine), local strains in the SiGe layers and plasma-induced traps at the gate oxide interfaces. It is also understood that cosmic rays from the universe can accelerate reliability failures. However, device reliability is complex; only a few failure mechanisms are well understood.
We find that an effective approach to prevent reliability failure is to detect potential killer elements in the process, maintain them under threshold levels and screen out heavily contaminated wafers before wafer fabout. In this paper, we address the relationship between specific killer elements and reliability and review the enablers of in-line metrology that can prevent reliability failure by early detection and monitoring of variation of killer elements in fabs.