Project Number: C95-22

PHYSICS-OF-FAILURE OF DISCRETE TRANSISTORS

Point of Contact :Dr. Patrick McCluskey e-mail: mcclupa@calce.umd.edu Phone: (301) 405-5323 Fax: (301) 314-9269

Objectives

Background

Approach

Work Accomplished

Objectives

Develop physics-of-failure based models for discrete field effect transistors, which include the dominant failure mechanisms, and the relevant material parameters. Then use these models to examine the effects of novel package materials and geometries on the reliability of power FET packages.

Background

This project builds on previous work in developing physics of failure models for integrated circuits and packages. This previous work provided designers with a way to assess the effects of their designs on product reliability. Because of the differences in the construction of discrete packages and IC packages, as well as the differences in the device design and use environments, discrete transistors can be expected to have similar but not identical failure mechanisms as those identified for ICs.

Approach

The first stage of this study will consist of identifying devices and packages of general interest. This will be accomplished by surveying the types of commercially available discrete packages and input from members. The structure and materials of construction of these packages will then be identified, both by examining actual packages and by reading data sheets on their design.

The second stage will consist of identifying the potential failure mechanisms, both intrinsic and package-related. For this stage, the existing body of knowledge on IC and IC package failures will be utilized, together with a survey of the published literature. Existing models for these failure mechanisms will also be identified where possible. Material property values needed for the models will be collected from published literature and from discussions with package manufacturers and raw materials suppliers.

The third stage will consist of using these models to conduct reliability assessments of commonly used packages to determine dominant failure mechanisms and to provide a reliability baseline for comparison.

The fourth stage will consist of using the models to determine the effect on reliability of using novel materials and geometries.

The final stage will consist of working with the members to develop methods for the experimental verification of models for the dominant failure mechanisms.

Work Accomplished

Commonly used metal, ceramic, and plastic power MOSFET packages have been collected and examined. Their structures and materials of construction have been identified. These include TO-39, TO-3, TO-247, and MRF150
Potential failure mechanisms, both intrinsic and package-related, have been determined for power MOSFETs, appropriate failure models have been identified, and relevant material properties have been collected.
Reliability assessments have been conducted on TO-3, TO-247, and MRF150. Dominant failure mechanisms have been identified, and a baseline of reliability established.
Reliability assessment simulations have been conducted on discrete packages designed using novel materials and package geometries, including the DE-275 and the D3PAK.
Experiments have been designed to verify the models for the dominant failure mechanism, and others performed to verify the predictions of the simulations.