Statistical simulation of static noise margin variability in static random access memory

Yiming Li*, Hui Wen Cheng, Ming Hung Han

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

19 Scopus citations


In this paper, we examine the impact of random-dopant-fluctuation (RDF), process-variation-effect (PVE), and workfunction-fluctuation (WKF), on 16-nm-gate metal-oxide-semiconductor field-effect-transistor (MOSFET) static random access memory (SRAM) cells. For planar MOSFETs with a threshold voltage (Vth) of 140 mV, the nominal static noise margin (SNM) of six-transistor (6T)-SRAM with unitary cell ratio (CR) is only 20 mV; and the normalized SNM fluctuations (SNM) induced by RDF, PVE, and WKF are 80%, 31%, and 48%, respectively, which may damage SRAM's operation. Two improvement approaches are further implemented; first, eight-transistor (8T)-SRAM and 6T-SRAM with increased CR are examined. Compared with the conventional 6T-SRAM, under the same Vth, the SNM of 8T-SRAM is enlarged to 233 mV and the corresponding RDF, PVE, and WKF-induced SNM are reduced to 9.5%, 6.4%, and 7%, respectively, at a cost of 30% extra chip area. Without increasing chip area, device with raised Vth, doping profile engineering and using silicon-on-insulator fin-type field-effect transistors (SOI FinFETs) are further advanced. The 6T SOI FinFETs SRAM exhibits the smallest σSNM, with merely 5.3%, 1.2%, and 2.3%, resulting from RDF, PVE, and WKF, respectively, where the value of SNM is equal to 125 mV.

Original languageEnglish
Article number5
Pages (from-to)509-516
Number of pages8
JournalIEEE Transactions on Semiconductor Manufacturing
Issue number4
StatePublished - Nov 2010


  • Eight-transistor (8T)
  • electrical characteristic
  • fin-type field-effect transistors (FinFET)
  • fluctuation
  • metal-oxidesemiconductor field-effect- transistor (MOSFET)
  • planar
  • process variation
  • random dopant
  • silicon-on-insulator (SOI)
  • sixtransistor (6T)
  • static noise margin (SNM)
  • static random access memory (SRAM)
  • workfunction fluctuation

Fingerprint Dive into the research topics of 'Statistical simulation of static noise margin variability in static random access memory'. Together they form a unique fingerprint.

Cite this