How can one distinguish an EOS damage from an ESD damage?

SemiconFarEast.com FAQ

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

Distinguishing EOS from ESD

In many cases, it is not difficult to distinguish between an electrical overstress (EOS) and an electrostatic discharge (ESD) damage. For instance, we're all quite comfortable assigning gross failure attributes such as extensive metal line burn-out and migration to EOS. When we see that the package is discolored and carbonized, as well as difficult to decapsulate, then we also know that it was exposed to a large amount of heat characteristic of an EOS rather than an ESD event.

On the other hand, subtle defects that usually require high electric fields or voltages to arise such as dielectric or oxide punchthroughs are often attributed to ESD. This is especially true if the affected area is close to an external pin, and historical and technical data show that the area is indeed vulnerable to an ESD event. One example is a capacitor connected directly to an unprotected external pin of an ESD-sensitive device.

Still, there will always be many cases wherein the failure attributes exhibited by a device are definitely due to an EOS or ESD event, yet not absolutely traceable to either one of them. An example of this is mild metal burn-out or migration, which can be caused by either a weak electrical overstress or a strong ESD event. Fused resistors are also known to be caused either by EOS or ESD. 'Gray area' cases such as these are the ones that are difficult to trace to the actual root cause.

In such cases, the engineer analyzing the failures has no choice but to complement the results of the failure analysis techniques with a thorough investigation of the circumstances surrounding the affected units, from their design and qualification to their actual use in the field. This would include analyzing qualification and reliability data, lot histories, equipment set-ups, and even operator assignments, so that any discernible pattern exhibited by the affected lots can be given greater focus. Unfortunately, such an endeavor is complicated by the fact that attributes can change over time and as the units undergo processing. For instance, a gross EOS failure attribute may have originated from a subtle ESD damage that was just aggravated into a full-blown EOS damage during electrical testing.

Simulation is a good method for confirming a cause, but it is only effective if the appropriate simulation conditions can be defined and executed with reasonable control, and if the failure attributes are exactly replicated by the simulation performed. Being able to trace the discharge path of an ESD event or the damage path of an EOS event based on failure attributes exhibited by the samples also adds value to root cause analysis, but it is not applicable to all cases.

Only by gathering as much information as possible and comprehending them holistically can a sound conclusion on whether the failure cause is EOS or ESD be reached. Nonetheless, an engineer must understand that he or she has limited powers, and can not be expected to arrive at the exact failure root cause 100% of the time. Even when the analysis boils down to simply choosing between ESD or EOS as the culprit, attributes can be too vague to reveal the true identity of their source.

Without tell-tale signs from data analysis, or successful simulation or discharge/damage path tracking results, one has to resort to a 'shot gun' approach wherein all possible sources of EOS and ESD on the line are identified systematically and eliminated accordingly. Continuous improvement of the line is a 'must' in such a case, until the problem disappears altogether.

HOME