• DocumentCode
    761589
  • Title

    Application of statistical analysis to determine the priority for improving LSI technology

  • Author

    Saito, Kazuyuki ; Sakaue, Masahiro ; Okubo, Tsuneo ; Minegishi, Kazushige

  • Author_Institution
    NTT LSI Labs., Kanagawa, Japan
  • Volume
    5
  • Issue
    1
  • fYear
    1992
  • fDate
    2/1/1992 12:00:00 AM
  • Firstpage
    47
  • Lastpage
    54
  • Abstract
    LSI failures are classified into three categories: nonoperation failure, out-of-specification failure, and out-of-reliability failure. The first two are emphasized. The binomial-distribution model is applied for nonoperation failure, and the analysis-of-variance using the F -distribution is applied for the out-of-specification analysis where the distribution of each characteristic parameter is assumed to be the Gaussian distribution. The analysis is performed experimentally using a test chip. An independent evaluation of an elemental structure, which is defined as a critical structure, is the most important principle in test chip design. A new two-level database system is adopted for the statistical analysis based upon the analysis-of-variance. Methods to determine the priority for improving LSI technology are presented for both nonoperation and out-of-specification failures. This procedure effectively assists in improving LSI yield
  • Keywords
    CMOS integrated circuits; failure analysis; integrated circuit testing; large scale integration; statistical analysis; CMOS LSI; Gaussian distribution; LSI failures; LSI technology; LSI yield; analysis-of-variance; binomial-distribution model; nonoperation failure; out-of-reliability failure; out-of-specification failure; statistical analysis; test chip; two-level database system; Condition monitoring; Databases; Electric variables measurement; Failure analysis; Gaussian distribution; Large scale integration; Performance analysis; Quality control; Statistical analysis; Testing;
  • fLanguage
    English
  • Journal_Title
    Semiconductor Manufacturing, IEEE Transactions on
  • Publisher
    ieee
  • ISSN
    0894-6507
  • Type

    jour

  • DOI
    10.1109/66.121976
  • Filename
    121976