Methodology for identifying wire bond process quality variation using ultrasonic current frequency spectrum

Abstract

Life time prediction of power electronic modules is becoming increasingly important in order to reduce unscheduled maintenance and unexpected failures. Recent developments in life time estimation of standard power electronic modules determine the nominal life time degradation under operating conditions and/or under harsh environments. However, it is important to obtain life time degradation information considering underling process variation originating from the manufacturing line. In this work, a methodology for obtaining a non-destructive assessment of bonding quality is investigated. This is done with the view of capturing information about bonding quality prior to service exposure, and hence determining the effect of the observed variation in bond quality on reliability. Analysis of the frequency spectra of signals obtained from the ultrasonic generator of a wire bond machine reveals it is a process sensitive parameter. The analyzed results show a good correlation between the frequency and amplitude values of the generator output signals and bond quality. 3D x-ray scans of bonds provide further non-destructive evaluation and validate the observed link between the observed generator output signals and bond quality.

Reliability of thick Al wire: A study of the effects of wire bonding parameters on thermal cycling degradation rate using non-destructive methods

“…Gibson et al and Pufall [10,11] also worked with a similar approach. In our own previous work, [12] we used the ultrasonic current frequency spectrum to identify wire bond quality in different simulated conditions of 'good', 'weak' and 'nonstick' bonds. Most relevantly, Wuwei et al [13] presented a technique for on-line detection of wire bonding quality by a new feature extraction method.…”

Predicting Lifetime of Thick Al Wire Bonds Using Signals Obtained From Ultrasonic Generator

“…Others have looked at the current envelope of the ultrasonic signals [3], [7]- [9]. Although a number of these methodologies are successfully able to discriminate between extremes of bond quality (i.e., good and bad bonds) [9], [10], none have demonstrated the ability to resolve subtle quality differences, as might occur within one production batch as a result of gradual tool wear, for example. More importantly, the influence of such slight changes in quality on the spread in predicted lifetimes has not been characterized.…”