In the coming years, tens of billions of Industrial Internet of Things (IIoT) devices will be connected to each other and generate massive amounts of data collected from sensors and applications. a large portion of IIoT data will eventually be stored, processed and even analyzed at the edge, requiring storage devices there to be able to respond faster with high data integrity.
A major challenge for IIoT edge computing is the harsh environment that these systems will inevitably encounter, especially at high temperatures. Unfortunately, there is a common misconception that by simply using off-the-shelf, industrial-grade NAND components, storage systems serving IIoT devices will be able to operate reliably in often extreme temperatures, which is sufficient to ensure mission-critical system reliability. In practice, taking this approach may result in unacceptable levels of device performance and fault tolerance in NAND flash memory, as explained below.
Translated with www.DeepL.com/Translator (free version)
NAND Characteristics, Die Shrinkage and Extreme Temperature Effects
During manufacturing, lithography node shrinkage or “die shrinkage” tends to increase the number of defective dies, resulting in inconsistent quality of NAND flash modules and ICs. Fewer electrons stored per memory cell can lead to an increase in the number of mis-codes, which reduces data retention and endurance.
Extreme temperatures can further exacerbate the degradation of NAND flash memory and cause changes in electron momentum in the modules and ICs, leading to data retention issues and even data loss. For example, raw bit error rate (RBER) and early life failure rate (ELFR) are two phenomena caused by electron leakage or retention problems in the tunnel oxide layer of the memory cell. During the program/erase (P/E) cycle, high temperatures accelerate electrons into and out of the cell gate and make P/E easier, but at the same time, there is an increase in charge traps (trapped electrons) that form at the tunnel oxide layer. Over time, the release of these charges leads to a threshold voltage shift (Vt), resulting in bit flip and retention failure.
At the other extreme at low temperatures, cell gates can end up with lower charges and increased tunnel oxide degradation can lead to potential dielectric leakage, despite improved data retention.
The only way to prevent such events from occurring in NAND flash devices is through a rigorous reliability testing program.
IC Level Testing and Product Level Reliability Demonstration Testing for Improved Reliability
NAND flash IC testing can be used to verify how error correction codes (ECC) and temperature affect P/E endurance, data retention and the lifetime of NAND flash devices. For example, different ECC levels per 1 KB of memory can be tested across the temperature range in a Reliability Demonstration Test (RDT) to determine the sufficient amount of ECC required for certain environmental factors.
For product-level testing, the same RDT process can be applied by aging tests for read/write quality assurance at temperatures from -40 ºC to +85 ºC and evaluating the entire drive, including firmware, user area, and other memory space, block by block. Validated weak blocks can be filtered out and replaced with spare blocks to enhance the overall endurance of the NAND device throughout its lifecycle, and further validation testing can verify signal integrity across the SATA interface.
ATP’s ITemp MLC NAND flash solutions have been validated in this manner to support high product reliability and long-term product lifecycle requirements in harsh temperatures.
Conclusion
General test methods for NAND IC components are not sufficient to achieve the reliability required for IIoT applications. Advanced RDT for high/low temperatures can improve reliability, extend product life, and reduce total cost of ownership. Is your storage solution up to the task in harsh environments?
Reviewed and edited by Ting Guo
Translated with www.DeepL.com/Translator (free version)
