Low resistivity (∼0.02 ohm cm), (111)‐oriented, p‐type silicon slices were anodized in a solution of 0.04N KNO3 in ethylene glycol containing 2% water, and the dielectric behavior of the system normalSi/anodic SiO2/normalelectrolyte has been studied. It was found that the reciprocal capacitance decreased proportionally to the square root of time during which the sample was immersed in a measurement electrolyte. Denning a “diffusion constant,” 2D=false(εεo·A)2∂false(1/C)2/∂t , it was found that D was not affected by the pH of the measurement electrolyte but increased (i) with the water content of the electrolyte, (ii) with temperature, and (iii) with anodic bias. An activation energy of 0.66 ev has been determined between 0° and 75 °C. It is suggested that water, after entering the anodic oxide, dissociates into hydroxyl groups which migrate into the oxide and cause the effective dielectric thickness to decrease. Partial dehydration was possible by annealing or by evacuation. The frequency dependence of the capacitance, when measured in an electrolyte of low water content, (i) increased with film thickness for oxide layers formed at the same current density, and (ii) decreased with increasing formation current density for constant film thickness.

The temperature dependence of the field coefficient B in the relation for the high field ionic current, i ~-io exp[BF--(W/kT)], has been investigated for the anodic oxidation of tantalum. Results from constant voltage and constant current anodizations demonstrate that B is inversely proportional to the absolute temperature as expected for a barrier theory which predicts B = a.q./kT. For the half jump distance, a, an average value of about 1.5A was found when q was assumed to be equal to the charge on the five valent Ta-ion.The relation between the ionic current and the field during the anodic growth of oxide films on valve metals iswith k the Boltzmann constant, T the absolute temperature, and F the field across the oxide. The factor to, the barrier height W, and the field coefficient B depend in their interpretation on the proposed mechanism (1), e.g., whether the rate-determining barrier for the ionic motion is at the metal-oxide interface or within the oxide. Several theories agree (2) that the temperature dependence of B should bewith q the charge on the moving ion and a half the width of the barrier W. Adams and Kao (3) studied the formation of very thin oxide films on niobium at constant current and found agreement between the measured temperature dependence of B and Eq.[2]. Much experimental research (4) of the steady state and the transient kinetics has been reported for the anodization of tantalum which fails to show the temperature dependence of B as expressed in Eq.[2]. It was observed that plots of log i vs. field were not straight lines but very slightly curved. The curvature was such that B decreased with temperature at a constant field. To account for this, Young (5) introduced a field dependent B by making the activation distance a in Eq.[2] a linear function of the field, a = ~ + ~F. The earlier observation of a temperature independent B or Tafel slope at different fields (the inverse of B is referred to as the Tafel slope) was explained (5) by a field effect compensating the temperature effect. Later (6) the effect of condenser pressure was discussed as a possible explanation of the field dependence of coefficient B. In this paper Eq.[1] is once more applied to the anodic oxidation of tantalum at different formation fields with particular interest focused on the temperature dependence of the field coefficient B. New evaluations are presented which are based on the current decay during constant voltage formations (see Results) and on temperature changes during constant current anodizations (see Results). ExperimentalSpecimens of about 18.8 cm 2 area and with a narrow tab were punched from 10 rail tantalum sheet (Fansteel, impurities less than 0.1%). The samples were degreased, chemically polished for 15 sec in 5:2:2 conc H2SO4:conc HNO3:48% HF, leached in boiling deionized water for 10 min, and then vacuum annealed (<10 -4 Torr) at 2100~The tabs of the foils were anodized to a potential of about 150v to confine the area. A new foil was used for each run. Just prior to the start of the experiment ...

Experimental evidence is presented which indicates that HF attacks preferentially along cracks, pores, or other singularities which exist in anodic tantalum oxide films. These defects when enlarged by the HF etch permit an electrochemical emf to arise. It is discussed how the electrochemical emf affects the photoresponse of tantalum oxide.Many phenomena observed with electrolytic capacitors or anodized valve metals have been connected to pin holes or flaws in the oxide layer. It is known that, for a given field across the oxide film, the anodic leakage current density cannot be reproduced very well and that the leakage current density increases with increasing oxide thickness. This is in agreement with the concept of flaws or microfissures which penetrate the oxide and whose effect becomes more severe as the oxide thickness increases. Young (1) measured tan 6 as a function of frequency and found by varying the electrolyte conductivity that the increase of tan ~ at low frequencies is due to microfissures in the oxide. He suggested that the rectification properties of the anodized valve metals are also connected to the pores which are blocked by oxygen gas but remain more or less free during hydrogen evolution. A model based on the idea that the formation of oxide above surface irregularities will cause stresses in the oxide, and hence cracks, was also proposed by Young (2). The onset of recrystallization (3), which occurs at the metal-oxide interface in the presence of a strong electric field, also will lead to cracks. Further support of "weak" spots in the oxide film is reported by Vermilyea (4) who tested Ta205 films with small area contacts. It was found that the Ta205 layer is a good insulator except where it is formed over singularities at the tantalum surface. It is proposed that the oxide formed over these small imperfections is highly conductive.In this work we are concerned with tantalum, and it is probably safe to assume that its oxide film has pores or "weak" spots. The density of these imperfections will depend on the purity of the tantalum metal, the pretreatment, and the anodization. These imperfections, if present, may be attacked differently than the bulk oxide by HF solutions. Vermilyea (5) found by comparison with an optical step gauge that the dissolution of TaeO5 in HF is uniform. However, the method of optical comparison would not be sensitive enough to reveal minute preferential etching in HF. In a very recent paper Vermilyea (6) showed electron micrographs of flaws in anodic Ta205 films. Several possibilities leading to flaws were outlined and discussed. On comparing the measured capacity with that calculated from the optical thickness, Vermilyea concluded that the etching occurs preferentially along the flaws as well as uniformly along the surface. This was confirmed by electron microscopy studies. A very sensitive indicator should also be the leakage resistance of the oxide film. Preferential etching along "weak" spots will change the leakage resistance and should, therefore, affect leakag...

The phenomenon of the residual discharge current following an inverse time law, inormalres∝b/t , was confirmed for tantalum electrolytic capacitors. It was shown that the coefficient b was independent of charging fields up to 2/3 of the formation field and that the residual current increased with temperature. The activation energy was 2.5 kcal/mole. Furthermore, it was shown that, if the external resistor was removed and the cell was left with open terminals, the internal residual current continued to flow until the build‐up field modified the flow. If the oxide film were illuminated during discharging, two different photoeffects were recorded: the transient effects due to wavelengths of 320 mµ to about 600 µ and the stationary photocurrent mostly due to wavelengths shorter than 300 mµ.Several models are discussed which lead to an inverse time law. Preference was given to the idea that a net positive space charge arises within the oxide during charging; that this space charge causes two internal fields of opposite directions and that the residual discharge current is the difference of two currents flowing in the opposing fields, thus neutralizing the space charge. Irradiation during charging and discharging affected the positive transient effect and the space charge field.

The growth of an anodic oxide film on tantalum metal immersed in diluted sulfuric acid has been investigated using potential and capacitance measurements. The tantalum metal was connected through an external resistance to a platinum electrode in the electrolyte. An analysis of voltage vs. time confirms the exponential field dependence of the ionic current for the later phases of oxide growth. Analysis of current false(Ifalse) vs. time false(tfalse) measurements provides BFo and λFo where B is the field coefficient, Fo is the initial field in the oxide, and λ is the growth rate per unit current, provided that the emf of the reaction, E , is known. E can be determined by matching the B ‐values derived from I vs. t to those derived from capacitance measurements. For sufficiently thin oxide films the growth rate exceeds the value expected from Faraday's law based on the external current which indicates the presence of internal electron currents for which the empirical expression IE=IEo e−d/l with IEo∼3×10−4 normalamp/cm2 normaland 1∼3.0Aå , was obtained. There are indications that either the inverse Tafel slope, B , or the emf, E , varies with film thickness in the range where the internal electron current is appreciable.