Starting with the known but unexplained fact that high-voltage electrolytic capacitors need operating electrolytes of high resistivity, breakdown (sparking) voltages of aluminum and tantalum anodes have been examined. If interfering side reactions are eliminated, breakdown voltage increases linearly with the logarithm of electrolyte resistivity and is insensitive to variations in electrolyte composition and to changes in temperature between 65~ and 95~ Guntherschulze and Betz, in their classic on electrolytic capacitors (1), considered ionic concentration of electrolytes to be the controlling factor of sparking voltage. They noted the linear dependence of sparking voltage on the logarithm of resistivity of the electrolyte. However, their experiments were limited to aqueous electrolytes and were done at or below room temperature which perhaps was not a good choice because anomalies in the formation process, e.g. etching, porous oxide formation and corrosion, are often experienced under those conditions. Some of their quantitative findings are open to doubt (2). Perhaps the phenomenon of sparking in electrolytic capacitors is still not well understood because side reactions may intervene during the formation process, and thus obscure whatever relationships may exist between sparking voltage and other relevant parameters. The reactions on aluminum in the ethylene glycol/borate system are characterized by electrochemical oxidation of the solvent (3), by metal corrosion and the appearance of a rust-colored solid corrosion product that adheres to the aluminum substrate. However, as will be shown, special measures can be taken to suppress these side reactions, particularly in the case of aluminum, with the result that a clear-cut relationship of wide applicability emerges, linking sparking voltages with the logarithm of electrolyte resistivity.
Experimental TechniqueConstant current formations of foil specimens were carried out in a temperature-controlled, closed, reaction vessel, watching the rate of voltage rise and noting the voltage at which the sound of sparking could be clearly heard for the first time (at which point there occurs also an abrupt slowdown in the rate of voltage rise). All foil specimens, except when otherwise stated, were 3 cm wide and 5 cm long, the lead being formed by an integral central tab, 0.4 cm wide and 5 cm long which emerged from one of the 3 cm edges of the foil. The aluminum specimens were cut from 99.99% pure foil, either plain (0.0090 cm thick) or etched (0.0070 cm thick). Before anodization, the samples were given a pretreatment by immersing them first, at room temperature, in 1N NaOH solution, for 2 min in the case of plain, and for 30 sec in the case of etched foil. This was followed by rinsing in deionized water and then in absolute methanol before drying the specimens at ll0~ for about 5 min. In the case of tantalum, plain capacitor grade, 0.0021 cm thick foil was used. The pretreatment here consisted of degreasing by boiling in trichloroethylene then air drying the specimens, an...
