Porous Germanium

It is challenging, however, to investigate what kind of porous structures will develop
in a semiconductor like Germanium, i.e. in an elementary semiconductor like
Si but with a low quality oxide like III-V compounds.

6.1 Porous Germanium

From the chemical point of view, one of the main differences between Si and III-V compounds is related to the quality of the oxides: very good oxide in Si and low-quality oxides in III-Vs. It is challenging, however, to investigate what kind of porous structures will develop in a semiconductor like Germanium, i.e. in an elementary semiconductor like Si but with a low quality oxide like III-V compounds.

While Ge today plays only a marginal role in semiconductor technology, pore formation in Ge still could be of interest for application, e.g. in the context of photonic crystals [131] or for sensors and special filters. Unfortunately, in spite of the fact that in the older literature there is a wealth of information concerning general electrochemistry of Ge, e.g. [132], less is known about pore formation in this material [133, 134, 135, 136]. More than that, macropores in Ge were reported only recently [137].

The first (unexpected) result observed in Ge is the general presence of electropolishing, even during the stable growth of pronounced macropores. The plain view micrographs thus do not show the original surface, but a cut through the structure at a depth that depends on the total etching time. This seems to be a unique feature that has neither been observed in Si nor in III-V semiconductors.

The second general observation is that homogeneous pore nucleation is quite difficult to achieve. While this effect is also observed in III-V compounds [51, 5, 56] and must be expected to some extent when working in avalanche breakdown conditions, its dependence on the etching parameters (in particular on the current density) is quite different from the other semiconductors.


Porous Germanium I

Figure 6.1: Smooth surface. 5% HCl, t = 120 min, T=20oC; a) j = 0.5 mA/cm2; b) j = 2.5 mA/cm2. b) Single domain.


Figures 6.1 and 6.2 show pore nucleation on polished and rough {100} surfaces, respectively. The homogeneity achieved is distinctly different. For a smooth surface the pore density decreases with increasing current, while for a rough surfaces the pore density increases. The reason for this seems to be the co-existence of pore etching and electropolishing. Since electropolishing does not need nucleation, it will be dominant on smooth surfaces where pore nucleation is difficult. Only by reducing the overall current density, electropolishing can be slowed down to a point were pore nuclei can 'survive' for the time needed to induce stable pore formation. Figure 6.2d shows a cross section of the pore structure obtained at a current density of 7.5 mA/cm2. It demonstrates that distinct macropores with clear features can be obtained. This is a remarkable observation because in n-Si under avalanche breakdown conditions mesopores (diameters <50 nm) are obtained exclusively. Macropores have only been observed under special condition in n+-type Si [67]. However, comparable structures are found in n-type III-V compounds etched under breakdown conditions [62].

The pores show a certain tendency to form side pores in <100> directions. However, since the density is low and the side pores are not visible in the cleavage plane, they are not easily seen in cross section. Figure 6.3a shows an exception (the cleavage plane is not flat in this case).



 Porous Germanium II

Figure 6.2: Rough surface. 5% HCl, t = 120 min, T = 20 oC; a) j = 2.5 mA/cm2; b) j = 5 mA/cm2; c) j = 7.5 mA/cm2; d) cross section view of the pores.


Since electropolishing is continuously exposing areas formerly hidden in the bulk, side pores become visible on the sample surface. This is shown in Figures 6.3 b) and c). These pictures also illustrate several new features of Ge pore etching: The pore walls are very well expressed {110} planes, only sometimes with small {100} facets left in the corners (Figure 6.3d). This has never been observed before. Si macropores have a comparatively weak preference for {100} planes, while III-V compounds strongly favor {111}B planes [138].


Porous Germanium III

Figure 6.3: a) Side pores emerging from a central pore visible at the top, but obscured lower down by uneven cleavage; b) Side pores (marked by arrows) rendered visible by continuous electropolishing of the surface; c) enlarged view; d) top view of a pronounced main pore.


Figure 6.3a also demonstrates that new pores are continuously nucleated. Larger pores nucleated earlier and therefore had more time to grow laterally. The pore shape is generally conical because the electropolishing acts on the pore walls, too. This is demonstrated in Figure 6.4a. Only if neighboring pores come too close, their lateral growth will stop. The thickness of the remaining pore walls correlates with twice the space charge region width. Figure 6.4b shows an example.

The subsequent nucleation processes only stop if the density of pores is so large that there is no more space for new pores. The already existing pores then may still increase their diameters somewhat until the ultimate limit of twice the space charge region width is reached for the wall thickness between two pores (Figure 6.3d).


Porous Germanium IV

Figure 6.4: a) Conical pore shape (the picture has been shortened to enhance the effect); b) Pore walls between close pores.


A particular conspicuous issue preferably found on polished surfaces etched at high current densities is the formation of pronounced pore domains, i.e. clearly expressed systems of secondary pores always centered around a central primary pore, as shown in Figure 6.1c.


Bearing in mind that pore nucleation under those conditions is enhanced for reduced current densities, domain formation can be understood if the potential and thus the current is significantly reduced in an area around a primary macropore. This is precisely what will happen if there are ohmic losses, particularly in the electrolyte.

While it is too early to ponder on details of pore formation in Ge, it appears that two major points can be emphasized:


  • All results are consistent with the assumption that the degree of passivation of Ge surfaces in the electrolytes used so far is smaller than in Si, while the relative difference in degrees of passivation of different crystallographic planes may be even more pronounced. Passivation in this context refers to the density of interface states in the band gap found after prolonged exposure to the electrolyte and with no current flowing [139].
  • The necessary supply of holes is easier achieved compared with other n-type semiconductors.


The first assumption is supported by the independent measurements of Chalzaviel et al. [140] who showed that the surface bond termination of Ge can be changed from H to OH (hydrophobic-hydrophilic) just by changing the applied potential, while in Si always H termination is found. This is a hint that the binding energy of the hydrogen bonds in Si is larger than either the H or OH binding energy on Ge surfaces, since they can easily be broken by applying an external voltage. The pores thus show pronounced crystallographic preferences, but the overall passivation is not strong enough to fully stabilize any surface (including pore walls) against electrochemical dissolution. Assuming that no perfect surface passivation exists, current flows through every point at the surface, even so the current density at pore tips is considerably larger than on flat surfaces. Constant current conditions coupled with pore formation and growth thus lead to a monotonically decreasing average current density since the total surface area increases monotonically.

As soon as the current density is significantly reduced, a new generation of macropores can nucleate and grow. This process can repeat several times and since the diameter of the pores reflects their 'age', a pore size distribution with pronounced peaks should be observed. This is indeed the case as shown in Figure 6.3b.


The second assumption follows from the observation of continuously proceeding electropolishing, implying that there is a hole source in addition to avalanche breakdown at the pore tips. This might simply be the leakage current of the reverse biased Ge - semiconductor junction which is intrinsically larger than in Si owing to the smaller band gap, and contains an additional component due to the postulated unpassivated surface states.



Home | Chapter 1 | Chapter 2 | Chapter 3 | Chapter 4 | Chapter 5 | Chapter 6 | Appendixes | Papers | Top
© Copyright www.porous-35.com , all rights reserved.
Quick, Simple, Free SEO CMS website builder and design software