Space Charge

           "Space charge" is charge in the dielectric, electrons, protons, and ions, that is moved around by the applied voltage.  Charge tends to build up at "discontinuities" in the dielectric, such as the dielectric-electrode interface in film capacitors, at the grain boundaries in crystalline dielectrics, and at various molecular sites simply called "charge traps".  Space charge is a major factor in dielectric behavior, such as leakage, high-voltage reliability, and dissipation factor vs. frequency and vs. temperature.  Its role in dielectric aging is still the subject of research.  Researchers have shown a special interest in cross-linked polyethylene (XLPE) because of its widespread use as a high-voltage insulation in power transmission, but work is also being done with polyethylene napthalate, PMMA, silicon dioxide, and other common dielectrics.  Space charge has been extensively studied for many years, but I can´t remember any books or journal articles specifically relating space charge phenomena to capacitor behavior.  It may be that capacitor people and dielectric theorists don´t talk to each other much. 

          Dielectric absorption is easily understood in terms of space charge.  When a capacitor is charged, electrons (-) in the dielectric are attracted to the (+) electrode.  When the capacitor is shorted, the electrons are slowed down in returning to their holes by the dielectric´s high resistance.  This gives the appearance that the capacitor is still charged, and in a sense, it is. 

          Space charge also explains why some capacitor reliability factors, such as breakdown voltage and corona starting voltage, tend to correlate with the peak-to-peak applied voltage better than with the peak applied voltage.  It´s almost as if the capacitor somehow remembers the voltage applied to it in the past, and adds it to the voltage applied now.   In fact, the capacitor does have a memory in the form of space charge.

           The tendency for high-voltage capacitors in DC operation, to fail when voltage is reversed is one example. When the applied voltage is suddenly reversed, previously displaced charge in the dielectric now adds to the voltage stress. There are others.  Corona may more easily start when an AC waveform is applied than when a DC voltage is applied, even when the DC voltage is higher than either of the AC peaks.  For example, 1000 VDC is less likely to start corona at a void in a capacitor dielectric than is +750 volts quickly followed by -750 volts, all things being equal.  This complicates high-voltage capacitor design in that you must not only know the highest voltage your capacitor will see, but how it will be applied. 

          The breakdown voltage of polymers is highly time dependent.  As the electrons move under the influence of the applied voltage, charge accumulation can cause voltage stress that in some spots, will be several times that of the applied voltage alone.  The isolation voltage of opto-isolators for example, may be rated both steady-state and after one minute.  Capacitors are often rated both steady-state, and "surge". 

          Space charge probably also explains why the breakdown voltage of polymers is nonlinear with thickness.  Above about 5-10 mils, breakdown voltage increases roughly as the square root of the thickness.  Presumably, the thicker the insulation, the more space charge can accumulate.  There have been reports that layered polymer insulations have a higher breakdown voltage than if they were made of just one solid piece.  One reason this is done with capacitors is make reduce the likelihood of failure from defects in the film.  It is unlikely that a defect in one layer will line up with a defect in the next layer.  However, layering may also help keep the space charge more evenly distributed throughout the insulation.  In some high-voltage capacitors, the dielectric is broken up into two layers with a floating foil or metallization layer between them.  This is to increase corona starting voltage.  Breaking up the dielectric into more layers would presumably increase breakdown voltage even more, without increasing total dielectric thickness.  There is apparently some research in this area. 

          Space charge behavior in dielectrics is still being actively studied, and some of the above is still not understood in detail. 

For more information see:  Space Charge in Solid Dielectrics, ed. Fothergill and Dissado. 1998, ISBN 0 9533538 0 X