ENBIS-17 in Naples

9 – 14 September 2017; Naples (Italy) Abstract submission: 21 November 2016 – 10 May 2017

Yield Maximizing Test Limits for Longitudinal Measurements using Linear Splines

13 September 2017, 10:50 – 11:10

Abstract

Submitted by
Vera Hofer
Authors
Thomas Nowak (University of Graz), Vera Hofer (University of Graz), Horst Lewitschnig (INFINEON AT, Villach)
Abstract
Design and production of semiconductor devices for the automotive industry are characterized by high reliability requirements, such that the proper functioning of these devices is ensured over the whole specified lifetime.

In our quality control task, we consider longitudinal data from high temperature operating life tests, consisting of repeated measurements for electrical parameter at certain given time points. In case of drift of electrical parameters, manufacturers then need to find appropriate test limits for their final electrical product tests, such that the proper functioning of their devices over the whole specified lifetime is ensured. Based on these datasets, we compute test limits that could then be used by automated test equipment for the ongoing quality control process. First, our test make sure that the devices meet certain predefined reliability requirements and, second, the resulting yield loss should be kept a small as possible.

In calculating these test limits, our approach consists of two steps: First, the observed measurements are transformed in order to capture measurement biases and gauge repeatability and reproducibility. Then, in the second step, we compute test limits based on a spline model that reflects the drift behavior of an electrical parameter. In order to solve the resulting optimization problem, we propose a new derivative-free optimization algorithm.

The capability of the model is demonstrated by computing optimal test limits for several drift patterns, by computing resulting yield losses and by analyzing the performance of the test limits by computing additional devices with a similar drift pattern.

Return to programme