There are some common myths related to DCL leakage current that can be heard still today:
Myth 1: IR/DCL leakage current is due to the cracks in dielectric.
This was one of the first imagination theory of why dielectrics have a leakage current without a details understanding of the physical mechanisms inside the insulators. It is true that cracks and “imperfections” in dielectric structure can be a cause of leakage current increase and catastrophic failures on individual “faulty” components. On the other hand, it may not be the prime issue for the basic leakage current level – we have to understand the physical conductivity mechanisms that takes place at the specific capacitor technology dielectric.
The details conductivity mechanisms description is beyond this lecture focus, but lets simplify it that in a capacitor the conductivity through the dielectric can be composed of three major mechanisms (all three are typical for electrolytic capacitors):
Another big influence to the value of DCL is the construction of the capacitor itself – namely the electrical potential between the electrode systems and the insulating dielectric. Metal electrodes may have some sub-oxide layers that are semiconducting and also electrolyte in electrolytic capacitors may exhibit rather semiconducting behavior – so in-fact in many cases on capacitors we are not faced with a simple Metal-Insulator-Metal structures, but more complex Metal-Insulator-Semiconductor systems, where interface barriers may play the leading role to the overall DCL leakage current values.
Myth 2: IR/DCL leakage current is a measure of the component reliability
This common myth is actually related to the Myth 1 as the imagination was that the part with higher leakage current has also larger number of cracks and thus it presents higher reliability risk.
As we learned in the Myth 1 mitigation above, the actual leakage current of a “standard” capacitor is due to its dielectric conductivity mechanisms and construction (electrical potentials matching). DCL of statistically normally produced capacitors is not a measure of reliability and it was confirmed many times that screening of DCL tail distribution is not improving the basic reliability numbers.
HOWEVER, Change of DCL as the structure robustness to external load can be a measure of reliability. There are number of proved screening methods that are part of specifications (MIL,ESA) or applied internally by manufactures as their know-how where a certain (thermo)mechanical and electrical stress is applied with subsequent DCL screening to improve reliability level and sort-out suspicious parts.
Case study: It was quite common to hear that leakage current on tantalum solid electrolytic capacitors with MnO2 electrode is due to the cracks in dielectrics. When conductive polymer electrode was developed, it replaced MnO2 solid electrolyte but DCL increased in 10fold time. This raise natural questions: Why DCL increased when the dielectric is identical and we exchanged electrolyte material only ? Does it mean that polymer type reliability is ten time worse? … of course, it is not true, the answer is that we replace one semiconducting electrolyte to the other and influenced the electrical barriers that are “more open” now and it drains more electrons/holes through it. But this is a natural stage of the capacitor construction without a direct impact to its reliability. How many times I still hear today that cracks are the main cause of leakage current in capacitors … It is cited in literature and still copied and pasted by many authors without any deeper insight.