By anodizing InP samples under galvanostatic conditions at current densities higher than j=10 mA/cm2 an interesting behavior of the voltage measured between the sample and the Pt electrode in the electrolyte was observed. For example, for low doped samples, i.e. n=1.5×1016 cm-3, at the beginning of anodization the voltage usually increases monotonically from 0 to about 60 V. If the current density is high enough, stable voltage oscillations can be observed as illustrated in Figure 3.26a. The cross sectional SEM image of a porous sample obtained under these etching conditions is shown in Figure 3.26b.
The first conclusion which can be made looking at SEM pictures is that the pores are current line oriented, i.e. they grow mainly perpendicular to the surface and not along <111>B directions as the crystallographically oriented pores do (see section 3.3.1). An additional feature which was not observed at Current-line Oriented Pores anodized at constant voltages (see section 3.3.2 and Chapter 3.5.1) is the strong modulation of the pore diameters. Let us call the local increase of the pore diameters a 'pore node'.
By comparing Figures 3.26a and b it is easy to see that each horizontal set of nodes is directly coupled to one voltage peak. The voltage maxima and the corresponding pore nodes are marked in Figures 3.26a and b as P1 to P22 (note that Figure 3.26b shows only the nodes from P4 to P22). Thus, the second conclusion is that the macroscopic oscillations occurred as a result of a synergetic behavior of the pores, i.e. the pores interact somehow with each other and together "force" the voltage to oscillate.
Figure 3.26: Data taken from an (100) n- InP sample with anodized at a current density of . a) The observed macroscopic voltage oscillations in time; b) Cross-sectional SEM of the sample. The inset shows the magnification of the nodes at the bottom of the porous layer.
Figure 3.26 shows that during 10 min of etching a porous layer with a depth of nearly 60 mm was obtained. Thus, the growth velocity for the Current-line Oriented Pores is much higher than the usual growth rate of the crystallographically oriented pores or the macropores in Si (in the range of 1 mm/min). The main reason for the high growth rate of the Current-line Oriented Pores is the high current density and the efficient dissolution of InP oxides in acidic solutions. The direct correlation between the voltage maxima and pore nodes offers the possibility to calculate the rate of the pore growth. Figure 3.26a allows one to calculate the time interval between two voltage maxima, whereas Figure 3.26b allows one to estimate the growth depth during the corresponding period.
Figure 3.27 shows that the rate of pore growth decreases from nearly 10 mm/min at the beginning to approximately 5 mm/min at the end of the experiment (after 10 min). The retardation of pore growth, in spite of the fact that the electrolyte was continuously pumped through the cell, can be caused by the continuously decreasing diffusion flux (transport) of chemical species to and from the pore tips. At the beginning of the experiment, i.e. etching in the vicinity of the initial surface, more oxide dissolving species are available for the dissolution, while into depth the number of oxide dissolving species (provided by diffusion) decreases and consequently the growth rate decreases significantly.
Figure 3.27: The rate of pore growth during the anodization of InP at constant current densities. The rate decreases due to the lack of oxide dissolving species at the tips of the pores (diffusion problems) and increase of pore diameters. (100) n-InP, n=1.5×1016 cm-3 anodized at j=100 mA/cm2.
However, in a galvanostatic experiment the amount of consumed charges per unit of time at the beginning as well as at the end of the experiment should be constant. Therefore, the pore growth rates should be constant in time as well. This is in contradiction to what was experimentally observed. These contradictory observations can be explained as follows:
It is highly probable that the valence of dissolution can not increase so strongly as to decrease the pore growth rate by a factor of two. This is because the pores with self-induced modulated diameters are Current-line Oriented Pores and according to the current burst model the dissolution at pore tips proceeds mainly via oxide formation. Therefore the valence of dissolution should have its maximum value during the whole experiment, i.e. eight holes per In-P atom pair.
On the other hand, in Figure 3.26b a general tendency to larger diameters is observed as the pores grow into the depth of the substrate. Therefore, it is more probable that the retardation in pore growth rate is a result of diameter enlargement.
Figure 3.28: Data taken from a n- InP sample with n2=3×10e17 cm-3 anodized at a current density j=100 mA/cm2. a) Voltage oscillations; b) SEM image in cross section.
Voltage oscillations have been observed also in n-InP samples with a higher doping level n2=3×1017 cm-3 (see Figure 3.28a). In these experiments the amplitude of the oscillation varies considerably during the time of anodization, see region I and II in Figure 3.28a. In this case the SEM images in cross section show weaker horizontal trajectories (see Figure 3.28b) as compared with lower doped samples (see Figure 3.26b). The main factor for the less stable oscillations at higher doping levels can be the difference in space charge region widths, which is smaller than in the case of low doped samples. As we will see, the space charge region is supposed to induce the synchronization between the pores and thus provokes the oscillations.