Micross

Abstract Preview

Here is the abstract you requested from the IMAPS_2010 technical program page. This is the original abstract submitted by the author. Any changes to the technical content of the final manuscript published by IMAPS or the presentation that is given during the event is done by the author, not IMAPS.

Interconnect Failure Rate Estimation Based on the Extreme Value Distribution
Keywords: Interconnect, Failure , Extreme
Evaluating and predicting the failure rate of interconnect is a time consuming and expensive process. Non parametric techniques such as employing the Chi-square distribution suffer from requiring large sample sizes to achieve a meaningful estimate. Recently there has been resurgence in the use of extreme value theory (EVT). Increases in temperature records, melting in the Polar caps, events such as Hurricane Katrina and the Pacific Tsunami of 2004, and the I-35W bridge collapse in Minnesota have fueled this interest. A novel method that is based on EVT and an accelerated degradation model for estimating the failure rate from a set of stress data is proposed and described. There are many advantages of this technique. It is much more likely to converge relative to the Johnson technique, and no human intervention or interpretation is required. MLE techniques are programmable into a spreadsheet format, and user operation is simple and quick. Failure rate limits can be obtained for the desired confidence level. The main advantage is that it uses all the information in the data set to focus on the tail of the resistance distribution, which is the area of interest in interconnect failures. Also, no assumption is made on the characteristics of the underlying distribution. Recommendations on usage of the technique and on sample size are discussed. Advice on how the total sample should be sectioned before the maximum of each subset is given. Moreover, bias in the estimating parameters and methods to correct for it are addressed. Recommendations on usage of the technique and on sample size are discussed. Interconnect examples are given, generated from Monte Carlo simulations of known distributions, and used for a comparison of the extreme value technique versus Chi-square and Johnson distribution methods.
Mark Plucinski, Engineer
IBM Corporation
Rochsester, MN
USA


CORPORATE PREMIER MEMBERS
  • Amkor
  • ASE
  • Canon
  • EMD Performance Materials
  • Honeywell
  • Indium
  • Kester
  • Kyocera America
  • Master Bond
  • Micro Systems Technologies
  • MRSI
  • NGK NTK
  • Palomar
  • Plexus
  • Promex
  • Qualcomm
  • Quik-Pak
  • Raytheon
  • Specialty Coating Systems