What failure mechanisms are considered to be bake-recoverable?

SemiconFarEast.com FAQ

- questions most frequently asked by visitors of www.semiconfareast.com

Bake-recoverable Failures

What failure mechanism am I dealing with if the rejects recover (become good again) after bake?

If you encounter a failing unit that becomes good again after it has been subjected to bake (usually 24 hours at 125 deg C), then you can suspect several possible scenarios that can explain the initial failure.

First, and this is the most common conclusion written in reports, is that the device had an internal or external package moisture that enabled ions on the package to become electrically conductive, causing pin-to-pin leakage. This scenario is often applicable only to slight parametric shifts, such as higher-than-normal leakage, and isn't generally used to explain a catastrophic failure (unless there is more in the samples than just moisture). Baking drives away the moisture and makes the units 'recover'.

The presence of a conductive foreign material on the die or the package that bridges electrically active parts is the second possibility. This scenario is different from the first one in the sense that a localized contaminant not normally found anywhere on the package is the culprit more than the presence of moisture in the device. Baking can dry up such a contaminant, either taking away its conductive properties or reducing its size to the point that it can no longer act as a bridge.

The third scenario is the presence of mobile ionic contaminants in the die. Mobile ionic contaminants tend to 'localize' to certain areas with the application of bias (such as during burn-in). Once they've accumulated at a high enough concentration, they can exhibit a charge that's large enough to shift the properties (such as threshold voltage) of a nearby transistor, making the device fail. Baking disperses the mobile ions in random directions, causing the failing unit to become good again. Unfortunately, the failure can return when bias is applied to the unit again.

Hot carrier effects, or the injection and entrapment of electrons or holes in oxide layers and other areas of the device where they shouldn't be, is the fourth possible cause of bake-recoverable failures. Hot carrier injection, which is often caused by excessive bias voltages being applied to the device, results in the formation of a volume charge that can shift the characteristics of a device. Baking can 'dislodge' these trapped carriers and disperse them, eliminating the volume charge formed by their accumulation and making the affected device 'recover' from its failure.

The fifth possible cause of failures that recover after bake is die surface charging, which is often caused by bias over time. Surface charging often manifests as excessive leakage or as a shift in transistor characteristics. Since the charge accumulates on the surface of the device, this mechanism is also die surface wash-recoverable. Baking disperses the surface charge accumulated by bias, making the device recover.

Package stress on the die is the last possible cause of bake-recoverable failures. Studies have shown that a direct application of mechanical stress on an active component of the die circuit can shift its electrical characteristics by a factor that's large enough to make the device fail. Sources of such package-induced die stresses include the TCE mismatch between the die and the molding compound and the molding compound fillers exerting force on the die. Baking can shift the 'stressed' points and make the device recover. Package stress-related failures should be permanently recoverable by decapsulation.