Acceptance sampling: an efficient, accurate method for estimating and optimizing parametric yield [IC manufacture]

Author

Elias, Norman J.

Author_Institution

Philips Lab., Briarcliff Manor, NY, USA

Volume

29

Issue

3

fYear

1994

fDate

3/1/1994 12:00:00 AM

Firstpage

323

Lastpage

327

Abstract

Acceptance sampling is a new yield estimation and optimization method which combines the accuracy of Monte Carlo analysis with the computational efficiency of response surface methods. Response surface approximations are used to guide selection of simulation samples. Formulas based on established statistical methods (viz., confidence intervals, stratified sampling) estimate yield and predict accuracy. Yield optimization procedures employ conventional search algorithms. Examples using 50 to 100 simulations demonstrate accuracy matching 1000 to 10000 Monte Carlo samples

Keywords

Monte Carlo methods; estimation theory; integrated circuit manufacture; optimisation; production; statistical analysis; IC manufacture; Monte Carlo analysis; acceptance sampling; confidence intervals; parametric yield; response surface approximations; search algorithms; statistical methods; stratified sampling; yield estimation; yield optimization; Approximation error; Circuit simulation; Computational modeling; Monte Carlo methods; Optimization methods; Response surface methodology; Sampling methods; Scattering; Surface fitting; Yield estimation;

fLanguage

English

Journal_Title

Solid-State Circuits, IEEE Journal of

Publisher

ieee

ISSN

0018-9200

Type

jour

DOI

10.1109/4.278356

Filename

278356