Probabilistic Physics of Failure of Electronic Hardware

Project Number: C96-01

Point of Contact:
Dr. Donald Barker
CALCE EPRC
email: dbarker@eng.umd.edu
Phone: (301) 405-5264
Fax: (301) 314-9269

Objective

Background

Approach

Work Accomplished

Objective

Develop and demonstrate a viable Probabilistic Physics of Failure (PPoF) approach for building and supporting highly reliable electronics systems in a cost-effective manner. In addition, develop a roadmap so that this approach can be implemented into design and acquisition practices.

Background

The physics-of-failure (PoF) approach overcomes the shortcomings of empirical reliability prediction methods by directly considering the stresses acting at a potential failure sites; accounting for the physical responses of materials to these stresses; quantitative modeling and characterization of fundamental failure mechanisms; and damage accumulation modeling and failure prediction. Most researchers usually apply the PoF approach in a deterministic manner, dealing with mean values of the critical variables, without accounting for the uncertainties in their behavior. However, without a probabilistic approach to PoF, there are many tasks that cannot be performed, such as assessing the quantitative reliability of a product. Many other logistic and supportability tasks are inextricably linked to reliability estimates. Hence, a probabilistic physics of failure (PPoF) approach is both essential and timely.

Approach

The origins of the PPoF approach starts with a PoF model, such as those developed by CALCE EPRC during the past 5 years. Currently the PoF models are typically suited for deterministic implementation. The purpose of this project is to extend the PoF framework to enable a probabilistic implementation. The key steps in the PPoF approach are:

establish methods to address the uncertainties of the key variables in the PoF model,
collect empirical data from process engineering, test programs (qualification and quality control tests) , and field returns, to assess the distributions of important stochastic variables;
establish methodologies for probabilistic ranking and aggregating of the failure mechanisms at various assembly levels;
develop metrics for probabilistic estimates of product stress-margin;
establish a management systems to implement, support and improve the PPoF approach.

Work accomplished

The work in this project only deals with only the first two steps of the PPoF approach as outlined above.

Existing PPOF methodologies were reviewed in areas outside of electronic packaging to see what methodologies lend themselves to the assessment of electronic packaging reliability.

Review literature for statistical data on key parameters in selected critical PoF failure models in electronic packaging.

Develop a preliminary generic PPoF approach, by considering the uncertainties of all the key variables. Propose methods that can potentially use operational Bayesian methods for parameter identification and updating.

Demonstrate the PPoF approach with practical examples of different failure mechanisms in electronic packaging. For more details, see

Formulate a road-map to pointing out the additional areas of work that need to be done before it is feasible to implement the PPoF approach into design and acquisition practices.

Funding

1 unit

Probabilistic Physics of Failure of Electronic Hardware

Project Number: C96-01

Point of Contact:

Objective

Background

Approach

Work accomplished

Funding