285°C Resistor Drift and Failure Analysis

Abstract

In order to improve the reliability of down-hole electronics Quartzdyne Electronics has invested over 10 million device test hours in life testing of our circuits in both powered and un-powered modes. In addition to time at temperature, these tests include thermal cycling and high impact drop testing. While resistors tend to be generally reliable, we have observed resistance drift in some units that has pushed circuit performance outside of accepted electrical specifications. In an attempt to comprehend the root cause(s) of resistance drift, failed samples have been studied in a scanning electron microscope. The resulting observations have led to structured designs of experiment to isolate the many possible root causes. This paper will present these observations, experimental outcomes and draw conclusions surrounding thick film vs. thin film performance, variations between value ranges, differences between vendors, and a possible link relating the drift mechanism to the method and extent of value trimming. This information should be useful to resistor vendors interested in improving the quality and performance of their products.

285°C Resistor Drift and Failure Analysis

“…Test results for the thin-film NiCr/Al and thick-film RuO 2 were similar to results from the prior test [5]. As seen in Figure 4, the 63.4k Ω NiCr/Al aged approximately 4% over 1000 hours.…”

“…The resistor aging presented in this paper is a continuation of tests first reported at HiTEC in 2010 [5]. The prior work focused on thin-film Nickel Chromium resistors on Alumina substrates (NiCr/Al), but also included samples of thick-film Ruthinium Oxide (Ru0 2) and thin-film TaN resistors.…”

“…We also tested three high-value NiCr/Si resistors from a different vendor that performed poorly. Clearly, differences between the specific formulations and how they are processed are more critical than the class of resistors The prior testing [5] demonstrated the significance of trimming. Damage to the resistive element near the trim site provides opportunity for crack propagation resulting in increased resistance over time.…”

“…The performance of either resistor is equal to or better than its aging performance and would be considered adequate for our application. Figure 12 shows the performance for each of the tests done both in this work and the prior HiTEC 2010 paper [5]. I-bars are shown at +/-one standard deviation to indicate the relative tracking of units within the same test.…”