SED navigation bar go to NIST home page SED Home Page SED Contacts SED Projects SED Products and Publications Search SED Pages
Invited Session: Process Capability

Invited Session: Process Capability

Organizer: Nien Fan Zhang, NIST
Session Chair: Jim Landwehr, AT&T Bell Laboratories

Process Capability Indices in the Semiconductor Industry

Catherine Lewis
Veronica Czitrom
Quality Assurance Dept., AT&T Microelectronics

Customers of semiconductor manufacturers often require that the process capability (Cpk) of critical processes be reported on a periodic basis. However, practical problems can undermine the usefulness of the Cpk. Manufacturers may have different lists of critical parameters possibly with different specification limits, making comparison between manufacturers difficult. The formulas for Cpk's and the confidence intervals may be used differently (e.g., using the average of lot (sub- group) Cpk values instead of using the individual observations from the whole sample), causing differences which may not be apparent to the customers. The sampling plans and metrology capabilities for data collection are usually intended for statistical process control, and may not be suitable for the calculation of Cpk values. Many of these problems can be addressed by standardization. It is also useful to compute variance components, and to track them as well as the Cpk over time, with goals for improvement and stability.

[Catherine Lewis , AT&T Microelectronics, 9333 John Young Parkway, Orlando, FL 32819-8698 USA; ]


Some Issues in Applications of Process Capability Indices

Nien Fan Zhang
Statistical Engineering Div., NIST

Process capability indices (CPI) are widely used in manufacturing industries to measure a process' performance in meeting preset specification limits. They are also used by supplier companies to demonstrate the quality of their products. Two issues in applications of CPI will be discussed. The first one is about estimating process capability indices for autocorrelated processes. It is well known that in practice process data are often autocorrelated. This is especially true for continuous manufacturing processes such as chemical processes. Interval estimation procedures for Cp and Cpk for autocorrelated processes are proposed and their properties are studied. The second one is about how to summarize the quality of a process via combining sequences of Cpk or Cpm, or Cpmk based on subgroups of the process data. Two different methods of combining the estimators of CPI's will be discussed.

[Nien Fan Zhang, Statistical Engineering Div., NIST, Gaithersburg, MD 20899 USA; ]

Date created: 6/5/2001
Last updated: 6/21/2001
Please email comments on this WWW page to