Electropolishing silicon

A number of different fluoride salts have been used as electrolytes for electropolishing silicon. Not only must the fluoride salt be sufficiently water soluble but the anode reaction product, the corresponding fiuosilicate, must also be readily soluble in water. Potassium fluoride is highly water soluble but the fiuosilicate is not. At a critical anodic etch rate silicon will passivate in KF solutions due to the precipitation of SiFg on the surface. Ammonium fluoride and ammonium fiuosilicate are both sufficiently water soluble. The acids HF and SiFg are even more water soluble. 

The early work on electropolishing silicon was done in largely nonaqueous solutions because it was believed that silicon could not be polished in aqueous HF solutions. Uhlir (7), for example, anodized silicon in 24 to 48% by weight aqueous solutions of HF at current densities up to 500 ma/cm and obtained only a matte black, brown, or red deposit. Later Turner (29) showed that silicon could be electropolished in aqueous HF solutions if a critical current density was exceeded. 

This will be called the thick anode film throughout this paper to distinguish it from the thin film which is present during electropolishing silicon. The thick film can grow to a thickness of as much as several tenths of a millimeter, whereas the thin film is probably less than 100 A in thickness. 

Fig. 4. Effect of temperature on the critical current density required to start electropolishing silicon in 5% HF.

At a constant silicon temperature, the ic was found to be directly related to the HF concentration. Data obtained at four temperatures, two below and two above 30° C, are shown in Fig. 5. The practical range of HF concentration for electropolishing silicon is 2 to 10% by weight HF. Below 2% HF the... 

Very little has been published on the electrolytic etching of semiconducting materials other than germanium and silicon. There probably have been many unpublished small experiments carried out to determine a suitable electropolishing process for many of the intermetallic semiconductors. Uhlir (36) for example, found that a largely nonaqueous HF solution suitable for electropolishing silicon would also electropolish GaSb. 

Turner TR (1958) Electropolishing silicon in hydrofluoric acid solutions. J Electrochem Soc 105 402-408... 

Electropolishing region does not occur in anhydrous organic solutions due to the lack of water which is required for the formation of oxide film. Figure 5, as an example, shows that in anhydrous HF-MeCN solution the current can increase with potential to a value of about 0.5 A/cm2 without showing a peak current. The relationship between current and potential is linear due to the rate limiting effect of resistance in solution and silicon substrate. 

Such reactions processes are responsible for the transition from PS formation to electropolishing with increasing potential as typically revealed in an i-V curve.18 PS formation can occur when the surface is not or only partially covered by oxide. Once the whole surface is covered with an oxide film further reaction can only proceed through the formation of oxide followed by its dissolution. Further increasing the potential will only result in an increase of oxide film thickness. On the other hand, increasing HF concentration will increase the dissolution rate of oxide. The presence of oxide on the silicon surface in the PS formation region and its increase with potential have been experimentally observed.98... 

In some ways electropolishing and electrochemical pore formation can be understood as the two sides of the same coin. In the first case the rate-limiting species in the chemical reaction is HF, while in the second it is the supply of holes from the electrode. If we assume a rough silicon wafer surface and a reaction that is... 

The need for defect-free, flat silicon surfaces led to the first investigations in this field, which were performed as early as 1958 [Tul]. It was found that electropolishing of silicon is possible in HF if the applied anodic potential is sufficient to produce current densities in excess of the critical value JPS. 

Electropolishing under galvanostatic conditions can be used to remove bulk silicon in a well-defined manner. This can for example be used to profile doping density or diffusion length versus the thickness of the sample, as discussed in Sections 10.2 and 10.3. The thickness D of the removed silicon layer can be calculated from the applied current density J, the anodization time t, the dissolution valence nv, the atomic density of silicon Nsi and the elementary charge e. 

Electropolishing is well established as a simple, in situ method to separate porous silicon layers from the silicon electrode. By switching the anodic current density from values below JPS to a value above JPS, the PS film is separated at its interface to the bulk electrode. The flatness of a PS surface separated by electropolishing is sufficient for optical applications, as shown in Fig. 10.10. 

Anodic oxide formation suggests itself as a passivating mechanism in aqueous electrolytes, as shown in Fig. 6.1a. However, pore formation in silicon electrodes is only observed in electrolytes that contain HF, which is known to readily dissolve Si02. For current densities in excess of JPS a thin anodic oxide layer covers the Si electrode in aqueous HF, however this oxide is not passivating, but an intermediate of the rapid dissolution reaction that leads to electropolishing, as described in Section 5.6. In addition, pore formation is only observed for current densities below JPS. Anodic oxides can therefore be excluded as a possible cause of pore wall passivation in PS layers. Early models of pore formation proposed a... 

Another way to use silicon wafers as DLs was presented by Meyers and Maynard [77]. They developed a micro-PEMFC based on a bilayer design in which both the anode and the cathode current collectors were made out of conductive silicon wafers. Each of fhese componenfs had a series of microchannels formed on one of their surfaces, allowing fhe hydrogen and oxygen to flow through them. Before the charmels were machined, a layer of porous silicon was formed on top of the Si wafers and fhen fhe silicon material beneath the porous layer was electropolished away to form fhe channels. After the wafers were machined, the CEs were added to the surfaces. In this cell, the actual diffusion layers were the porous silicon layers located on top of the channels because they let the gases diffuse fhrough fhem toward the active sites near the membrane. 

---