Physics of Failure Analysis for PCBs and Electronics Assemblies

Before you can put a PCB in a mission-critical system, device, or vehicle, you need to prove its reliability and quality. On the manufacturing side, this is one reason we have copious industry testing, inspection, and quality standards that apply to PCB assemblies and individual components. On the design side, it’s important for designers to perform a risk assessment for mission-critical systems to ensure a design will operate with the expected lifetime, as well as to identify potential points of failure in high-reliability systems.

The area of electronics design and analysis that addresses these reliability concerns is called Physics of Failure. While the field of study was originally applied to integrated circuits and is still used in this area for design purposes, it was later applied to electronics assemblies and, most recently, to failure mechanisms in raw materials. These methods begin with extensive testing and data analysis of prototypes and test coupons, as well as some basic equations that can give some rough estimates of reliability and time to failure.

What is Physics of Failure?

Physics of Failure involves understanding the relationships between a product’s physical characteristics, how they vary due to manufacturing processes, and how they are affected by the product’s operating environment. The field forms the basis of many reliability analysis simulations and calculations, some of which are implemented in field solver applications. The field is sometimes referred to as Reliability Physics Analysis. A related field of reliability analysis and risk assessment is worst-case analysis.

No matter what you call it, some of the basic calculations involved in Physics of Failure provide several benefits that help guide a design.

• Identify critical failure points: The location on a design where a product is expected to fail should be identified early, and it shouldn’t be based on an engineer’s opinion. If a failure point can be identified with hard numbers, it eliminates opinion and subjectivity from a design.
• Predict time to failure: Some of the basic calculations available in the Physics of Failure methodology are used to predict mean time to failure for specific operating conditions. While these are just an average, they help quantify the impact of the environment on reliability.
• Justify design choices: Taking a serious approach to reliability can help justify important design choices, both at the circuit level and the layout level.

In short, the tasks and analysis involved in Physics of Failure help explain what failed and why it failed. Going further, applying some statistical analysis and developing empirical models helps designers identify when a failure might occur for given test conditions based on a totality of possible failure mechanisms. It is this latter area where test engineers and reliability engineers spend their time to ensure PCBAs are as reliable as possible.

Develop Models From Data

Some of the classic models used in early Physics of Failure literature published in the 1970s and 1980s were based on a mix of empirical models and fundamental physics. This is particularly the case in thermally-induced and vibration-induced failures, two areas which have been studied most extensively in the Physics of Failure field. In addition to thermally-induced failures, chemically induced failures have been studied, and there is an empirical model that addresses humidity-related failures in wire bond/bond pad connections.

Empirical and Simulation-Based Methods

Empirical methods focus on determining parameters in physics-based models, or on developing a model to quantify the relationship between two measured variables. Typical techniques involve univariate or multivariate regression with a power law model, something which is simple enough to be performed in Excel. In some cases, simulations can be used on the front-end due to the complexity of some problems.

Thermal and thermomechanical failure could be assessed together in two ways. The principle testing methods include high temperature testing under MIL-STD-810G Method 501.5, or thermal cycling under MIL-STD-810G Method 503.5. The former addresses thermal failures in solid-state devices, while the latter can be used to assess board-level failures from repeated thermal excursions. Areas of focus in thermal cycling tests include mechanical failure in pads, vias (particularly joints and high aspect ratio vias), and solder joints. Due to the complexity of a typical PCBA, simplified simulations may be used, but typically data is taken from tests to determine parameters in an empirical model.

Vibration and mechanical failure mechanisms are more insidious as there is no good way to calculate this by hand in a PCBA. Tests are needed to assess vibration-induced fatigue and failure. There is one text that looks at these problems for electronics generally, and it includes some empirical models that have been used to quantify vibration-induced failures in electronics. You can find this text below:

• D. S. Steinberg, Vibration Analysis For Electronic Equipment, 3rd Edition. John Wiley & Sons Inc., New York (2000).

In examining failures from a random sample of PBCAs or test coupons, accelerated stress/life testing and inspection of failed components can help test engineers pinpoint the specific design element that failed, as well as the failure mechanism. Test and simulation results have been used in the past to develop thermodynamics-based models with empirically determined parameters, which are then used to estimate failure rates once design elements are re-used in a new product. Through successive iteration in complex designs, this guides the design process to continuously identify and eliminate defects.

Statistical Methods

A more general approach is to statistically examine occurrence failures without assuming an underlying physical mechanism, followed by pinpointing the root cause of failure through inspection. After further inspection, it becomes possible to determine the main mechanism driving failure probability. With enough data, one can construct a curve like that shown below; this curve shows a Weibull continuous distribution function (CDF) defining mean time to failure for all times below the value found on the x-axis.

This distribution, and its usage in predicting product failure, are something I’ll save for another article. If you have access to a program like Mathematica or MATLAB, you could take your dataset and perform the above fitting procedure on your own to quantify reliability and mean time to failure.

Once you’ve performed your risk assessment and you’re ready to create a highly reliable physical layout, use the industry’s best CAD tools in Altium Designer®. You can also export your design for reliability simulations using the Ansys EDB Exporter extension. When you’re ready to release these files to your collaborators for more advanced simulations and manufacturing, the Altium 365™ platform makes it easy to collaborate and share your projects. Everything you need to design and produce advanced electronics can be found in one software package.

We have only scratched the surface of what is possible to do with Altium Designer on Altium 365. Start your free trial of Altium Designer + Altium 365 today.