Contact effects between electrodes and dielectrics

Abstract

This report covers a research on the phenomena of so-called “contact resistance”. The research was carried out, on the suggestion of the E.R.A., by the Metropolitan-Vickers Electrical Co., Ltd., and the results show the nature and importance of contact phenomena, also their true bearing on resistivity and dielectric-loss measurements, and that certain conclusions reached by a previous investigator were exaggerated.The authors give a detailed account of a series of experimental investigations and describe the development of methods for the determination of the actual resistance and capacitance of the interface between electrodes and dielectrics.Section (I) of the report deals with two methods which demonstrate the existence of a potential drop across the contact face but which do not lend themselves to accurate quantitative work. A third method is described in which the potential distribution along a cylindrical specimen, subjected to a known d.c. voltage, is ascertained by means of exploring electrodes, and the contact voltage-drops deduced. From these, and the known resistance of the specimen, the contact resistances at each electrode are obtained in ohms per cm2. In the case of slate this was found to be 0.8 to 0.9 × 106, and fibre 5 to 9 × 106.Since the contact voltage-drop with direct current is not affected by direct-contact capacitance effect, a fourth method was developed in which the current and watt-loss were measured for a given a.c. stress of known frequency. From these measurements the equivalent series resistance and capacitance of the specimen can be calculated. The specimen was then machined down to a different thickness and similar calculations were made for a range of thicknesses. These factors plotted against thickness and extrapolated to zero thickness revealed the residual quantities attributable to the interface. In Table 6 it is shown that, over a wide range of power factors, the contact resistance does not vary greatly, it being of the same order as already determined.Ultimately the potential distribution method, previously adopted for direct current was modified for a.c. determinations and a large number of tests were carried out on slate, fibre and bakelite. Table 9 gives a complete summary of the results of 75 tests, including a number of check tests to ascertain the effect of frequency, voltage stress, humidity, temperature, surface leakage, roughness of surface and dirty mercury. Tests 53 to 69 were carried out with electrodes other than mercury. The results are important in that they reveal methods of obtaining better contact than can be obtained by the use of a simple mercury electrode.For comparative purposes the authors give in Table 11 the maximum, minimum and mean values of contact resistance and capacitance for normal conditions at 50 cycles and show that the agreement obtained between the two a.c. methods mentioned above is as good as the agreement obtained in different tests by any one method.The significance of the contact effect in relation to measurements on dielectrics is discussed, certain typical conditions being postulated and the percentage error in power factor and permittivity shown to be negligible except in the case of thin specimens of low-grade material. In the extreme case of a hypothetical material 1 mm thick, having a power factor of 80 per cent, the error attributable to contact impedance is shown to be 18 per cent, while in a more normal case where the power factor is 5 per cent and the thickness 1 cm, the error is appreciably less than 1 per cent.