Before giving a review of the etching process, it is important to mention some of the main characteristics of the silicon semiconductor in order to gain a better understanding of the etching process.
Silicon is a semiconductor material with an electron conductivity σ between 104 >
σ > 10-8 (Ωcm)-1. This conductivity can be changed by doping. A silicon crystal is built up of many unit crystals arranged periodically. The unit crystal is a cubic structure, see figure 3.1.
Silicon atoms are covalently bonded with each other. Silicon has four valence electrons, and each silicon atom in the bulk is bonded with four neighboring atoms.
On the surface of a silicon wafer, for example, Si bonds are broken (these broken bonds are called “dangling bonds”), i.e., they are not bonded to other Si atoms. Some of these dangling bonds become saturated by forming unstable bonds with atoms in the atmosphere like oxygen, nitrogen, etc.
(0,0,0)
(1/4,1/4,1/4)
Fig. 3.1: The silicon crystal has a face centered cubic (fcc) structure with a base of two identical atoms, which are located at the (0,0,0) and (1/4,1/4,1/4) positions. This is a diamond structure. Every silicon atom is located in the center of a tetrahedron and covalently bonded with another four Si atoms.
Other important characteristics of silicon atoms correspond to the order (arrangement) of the atoms in the crystalline structure. The position of the atoms with respect to an x-y-z coordination system is well described by the Miller indices. The Miller indices are represented by the numbers in parentheses [1: Hummel 2004]. For example, the Miller index (100) describes the positions of atoms located on a plane perpendicular to the x-axes in a unit crystal, see figure 3.2.
Fig. 3.2: Miller indices in a cubic structure [1: Hummel 2004].
Also, from diagrams 3.2 and 3.1, it can be seen that the density of Si atoms on the (111) plane (direction) is higher than on the other planes (directions). This fact has very important consequences for the further explanation of the geometry produced by certain etching processes. Now a review of the etching process will be given.
After the microelectronics revolution and the construction of the first silicon solar cell (1946) and of the first germanium transistor (1949), interest in semiconductors was understandably great. Thus, the first attempts to etch silicon were carried out in the sixties [2: Finne 1967]. Also in 1967, an etch solution was used for the first time
to etch silicon. It consisted of potassium hydroxide (KOH), water (H2O) and isopropyl alcohol (IPA) [3: Price 1967].
In 1969, Lee [4: Lee 1969] used a ternary liquid etch solution for silicon, consisting of hydrazine, iso-propyl alcohol (IPA), and water. Hydrazine was employed as an oxidant, IPA as a complexing agent, and water appeared to function as a catalyst. He explained the etching process of silicon as follows: the oxidant oxidizes silicon to hydrated silica, and the complexing agent reacts with the silica and forms a soluble complex ion. Water provides excess (OH)- ions for the oxidation step.
The above-mentioned etch solutions used to etch silicon show two important characteristics: they act anisotropically and selectively. Anisotropic means that the etch rate depends on the crystal orientation, whereas selective means that the etch rate depends on the doping concentration of silicon.
Knowledge of the anisotropic and selectivity effects of alkaline etch solutions on mono-silicon has been used to produce microstructures in the microelectronic and micromechanical industries [5: Holmes 1974]. And it was also introduced for the first time in the photovoltaic industrial sector by Haynos et al. in the same year [6:
Haynos 1974].
The etch solution employed by Price [3: Price 1967] is now well known in the photovoltaic industry as the standard KOH-IPA solution used to texture mono-Si-wafers. The KOH-IPA solution worked very well for a long time after its introduction in 1974, but the development of new sawing methods and different chemicals to clean slurry from sawed wafers changed the surface characteristics of the as-cut silicon wafers, reducing the effectiveness of this etch solution (see chapter 4).
In 1983, Palik et al. [7: Palik 1983] measured the etching products by recording the Raman spectra in real time as the etching progressed in a 5M KOH solution. He found that the primary etching species were OH- ions, and the etching products were silicate SiO2(OH)=2. Isopropyl alcohol (IPA) does not appear to participate chemically in the etching process. Thus, the role of IPA in the etching process developed by Palik et al. conflicts with the observations of Lee [4: Lee 1969]. The results reported by Palik et al. with respect to the role of IPA are contrary to the findings of other authors, who clearly see the great influence of the complexing agents (IPA for example) on ternary etch solutions [8: Cho 2004]. Seidel also assumed that IPA does not play a major role in the silicon etching process, although he saw that the addition of IPA to the etch solution considerably decreases the etch rate of silicon [9:
Seidel 1986, 10: Seidel 1990]. From these results, it is clear that the role of the complexing agent (IPA for example) in ternary solutions used to etch silicon continues to be a mystery.
Nevertheless, the etching process of silicon (with a minimization of the role of IPA) was well described by Seidel in 1986 and 1990 [9: Seidel 1986, 10: Seidel 1990]. In his work, he also proposed an empirical equation that describes the etching rate of silicon, from which he also concluded that four electrons are needed to remove one silicon atom. Now we take a closer look at Seidel’s explanation of the silicon etching mechanism.
In a first oxidation step, two hydroxyl ions OH- from the alkaline solution are bonded to two Si dangling bonds, and thus Si-OH bonds are established. In this process, two electrons from the OH ions are injected into the conduction band of silicon.
[10: Seidel 1990]
The Si-OH bonds weaken the two silicon back bonds. In order to break the back bonds (which have already been weakened), two electrons from the Si-Si back bonds have to be excited to the silicon conduction band. When this happens, the Si-Si back bonds break, and a positive silicon hydroxide (Si-OH) complex is formed.
[10: Seidel 1990]
The silicon hydroxide complex further reacts with the other two hydroxyl ions from the solution to produce monosilicic acid.
[10: Seidel 1990]
The monosilicic acid can now diffuse into the solution.
The four electrons in the silicon conduction band, which are located near the surface, can be transferred to four water molecules, which are also located near the surface. Thus, a further four water molecules are decomposed into hydroxyl ions and atomic hydrogen. Atomic hydrogen further recombines to produce molecular hydrogen.
[10: Seidel 1990]
Figure 3.3 (see below) offers a schematic representation of the silicon etching process.
The potassium cations K+ and the complexing agent (IPA for example, if it is used in the etch solution) in the solution do not play a major role in the etching process of Si, according to Seidel’s explanation.
Si
Fig. 3.3: Etch process of monocrystalline silicon in an etch solution which contains potassium hydroxide KOH, isopropyl alcohol IPA and water (H2O). In a) an as-cut silicon wafer is shown, dangling bonds (open bonds) cover the silicon surface. In b) monocrystalline silicon is immersed in a KOH etch solution, Si-H bonds are formed on the silicon surface, and furthermore, atomic hydrogen diffuses into the silicon bulk. In c) because Si-H bonds are unstable bonds, they are replaced by Si-OH- bonds. In d) Si-OH- bonds decrease the electron density of Si-Si back bonds, and two thermally excited electrons are given to the conduction band (c.b.). The Si-OH complex further reacts with two OH- ions, and monosilicic acid is formed. In e) the orthosilic acid now diffuses into the solution. The four electrons at the c.b. of silicon are given to four water molecules, producing more OH- ions. In f) broken bonds (dangling bonds) are immediately occupied by atomic hydrogen. The surface remains H-terminated during the etching.
In 1993, Elwenspoek [11: Elwenspoek 1993] proposed a new theory to explain the mechanism of anisotropic silicon etching. His explanation was inspired by theories of crystal growth. He proposed that the physical state (being atomically flat or rough) of the various surfaces is ultimately responsible for the anisotropy of etch rates and activation energies.
However, Elwenspoek’s proposal directly contradicts some results obtained in this study. We found an etch solution (KOH-HBA solution) that acts anisotropically on flat material (polished FZ-Silicon wafers, or flat etched Cz-Si-wafers) and also on rough material (as-cut Cz-Si-wafers), see chapter 5.
Despite Seidel’s explanation of the anisotropic etching of silicon, the role of the complexing agent remains a matter for further investigation.
From the literature and observations of the author, the role of the complexing agent (IPA for example) in the anisotropic etching of silicon can be listed as follows:
1. – It acts as a mask against chemical reactants, as does for example SiO2, but of course to a lesser extent,
2. – It is preferentially absorbed on (111) silicon crystal planes, so that pyramids can be formed, and
3. – It regulates the etching rate for solutions with a constant KOH concentration and constant temperature. It does this by attracting water molecules around itself and thereby controlling the water concentration in the solution.
From points 1 and 2 given above, it is clear that IPA stimulates the formation of a pyramidal texture on the silicon surface.
Figure 3.4 offers a representative illustration of the silicon etching process in a KOH-IPA-water solution.
Fig. 3.4: Anisotropic etching on silicon in a KOH-IPA-Water solution. The masking role of IPA is shown in green. Also, the detachment of hydrogen bubbles and silica can be seen.
The etching process represented in figure 3.4 is similar to that represented in figure 3.3. In figure 3.4, the role of IPA as a mask is enhanced. Also, in figure 3.4, Si-O bonds are considered to be formed instead of Si-H bonds (see figure 3.3). But probably both bonds, Si-O and Si-H, are present on the surface of the silicon.
From point 3 (given above), IPA considerably reduces the etch rate. The latter point suggests that there should be an optimum concentration of IPA in the etch solution. Unfortunately, such an optimum IPA concentration is almost always determined by rule of thumb.
Considering a KOH-IPA solution to etch silicon wafers, as is common in the photovoltaic industry, finding such an optimal IPA concentration also depends on the surface characteristics of as-cut silicon wafers, what makes the work (of finding an optimal etch solution to form a pyramidal texture) very difficult. So, an etch solution which does not suffer from such drawbacks would be desirable in the photovoltaic industry.
The importance of the anisotropic effect on the etching of mono-Si-wafers is related to the formation of random pyramids on the surfaces of silicon wafers. Such a pyramidal texture on Si-wafers reduces total light reflection considerably. Thus the generation of electron-hole pairs in the Si material is enhanced. Such a process is highly desirable in silicon solar cells, for solar cells convert sunlight into electrical energy.
Therefore, the study of chemical solutions which produce a pyramidal structure on mono-Si is a matter of current research interest in the photovoltaic research community. In the next section, the experimental production of a pyramidal texture on monocrystalline silicon wafers using an alkaline KOH-IPA etch solution is described.
3.2 Influence of etching time on pyramidal texture when using a