When an electronic product fails under critical circumstances, the emphasis is on quick and accurate analysis. The situation could be a complex electronic system for the military that fails at final demonstration or an electronic game that dies during the New York Toy Convention.
When this happens, failure-analysis services can provide clients with help in determining the problem and instituting corrective action. Typically, the client knows the product has suddenly started failing with a particular diode, transistor or IC. To get the search for the errant device underway, the company provides the failure analysts with several failed units along with new (unused) parts for comparison.
Often, a failure is caused by a flaw in the materials used in a semiconductor. Consequently, the failure analyst must also have in-depth knowledge of metallurgy and material science.
An example of this type of failure was a material-related defect in high-voltage stack diodes that were failing in defibrillator units in hospital emergency situations. Two such failed components sent to the testing lab had caused the deaths of patients before the medics could locate backup defibrillators.
The first step was to perform bench electrical tests to better understand the failure mode. In this case, the diode stacks were short-circuited with such low resistance that cross-sectioning was necessary.
The sections revealed the silicon chips had cracks emanating from the solder used to join them in the stacks. A special X-ray imaging device attached to the electron microscope adjusted the gray scale to the atomic number of the material beneath the E-beam.
The resulting picture is shown in Figure 1. The display brightness is proportional to the atomic number. The dark area is silicon, the white area is solder, and the gray area identified by arrows is nickel.
The failure cracks originated from grayer circles inside the white solder bands. Later these were determined to be nickel balls that had been added to the solder by the diode manufacturer to improve power dissipation. However, nickel has a thermal expansion coefficient quite different from the solder. As a result, the silicon crystal chips cracked during temperature cycling.
These diodes were key to the operation of the defibrillator. The failure analyst’s report suggested substituting a diode stack made by another manufacturer.
Another popular tool for the failure analyst is an infrared microscope that, in contrast to electrons, identifies chemical compounds rather than elements under its tiny beam. Quite often, the failure-analysis task boils down to identifying tiny deposits of metals or chemicals as simply being in the wrong place.
A key technique stimulates the chip to observe the effects on the parameters determined during electrical bench testing. The procedure begins by measuring the electrical anomaly, then chemically stripping away a layer and measuring again to see if the layer removed contained the failure-inducing anomaly.
The strip-away technique was used to analyze failed op-amps in a high-voltage power supply for a missile. Using microprobes on the silicon chip, the analyst found that the small input transistor pair had betas of 2 where normal was 200. Each time a partial layer of the chip protection glass was removed, the beta would drastically improve.
Repeated step-wise strip-aways of the protective glass improved beta until it recovered to about 100. The answer: Sodium ions in the glass poisoned the beta. All the sodium ions naturally scattered around the chip had collected on the edge where the most sensitive input transistor pair was located. The electric field of the power supply was the culprit.
To solve the problem, the analyst convinced the client to relocate circuit-board components so the op-amp was 90° from the original location. That way, the few randomly scattered sodium ions would collect in a less sensitive area of the silicon chip. The failures stopped.
The client was amazed that the analyst could look at only a few IC chips and read enough from them to suggest a successful correction of the problem. As you can see, device physics plays an important part in these failure-analysis investigations.
Independent failure analysis provides very important services to any business that manufactures electronic products. The failure scenarios given in the analysis reports allow managers to focus their corrections quickly, accurately and economically.
About the Author
Porter James is a Staff Engineer at Associated Testing Laboratories. As a failure-analysis expert specializing in semiconductor physics and materials science, he has given testimony in several multimillion-dollar federal liability suits involving the semiconductor industry. Associated Testing Laboratories, Subsidiary of Bell Technologies, 53 Second Ave., Burlington, MA 01803, (617) 272-9050.
Copyright 1996 Nelson Publishing Inc.
September 1996