The Chemical Dissolution of Silicon

The chemical dissolution of silicon can be obtained in both liquid and gaseous media. The latter is known as dry etching or reactive ion etching (RIE) and is used in today s microelectronic manufacturing. However, wet processes related to silicon are also very important, as one third of the total number of process steps for the fabrication of today s integrated circuits involve... 

When the surface is completely covered by an oxide film, dissolution becomes independent of the geometric factors such as surface curvature and orientation, which are responsible for the formation and directional growth of pores. Fundamentally, unlike silicon, which does not have an atomic structure identical in different directions, anodic silicon oxides are amorphous in nature and thus have intrinsically identical structure in all orientations. Also, on the oxide covered surface the rate determining step is no longer electrochemical but the chemical dissolution of the oxide.1... 

Fig. 4.1 Reaction scheme proposed for the chemical dissolution of (100) oriented silicon surfaces in alkaline solutions.

Electron injection has been observed during the chemical dissolution of an oxide film in HF [Mai, Ozl, Bi5]. The injected electrons are easily detected if the anodized electrode is n-type and kept in the dark. Independently of oxide thickness and whether the oxide is thermally grown or formed by anodization, injected electrons are only observed during the dissolution of the last few monolayers adjacent to the silicon interface. The electron injection current transient depends on dissolution rate respectively HF concentration, however, the exchanged charge per area is always in the order of 0.6 mC cm-2. This is shown in Fig. 4.14 for an n-type silicon electrode illuminated with chopped light. The transient injection current is clearly visible in the dark phases. 

The etch rate of anodic oxide can be determined by methods similar to those for thermal oxide or deposited oxides. It may also be estimated from the anodic current of the oxidized electrode. The anodic i-V curve of a silicon electrode typically shows a passivation-like peak above which the dissolution occurs through a two-step process the formation of oxide film followed by the chemical dissolution of the oxide. The steady-state anodic current measured at an anodic potential above the peak potential indicates the dissolution rate of the anodic oxide. Thus, the passivation current, /p, listed in Table 5.5 can be used for estimation of the etch rate of the oxide film formed at the anodic potentials. (A current density of 1 mA/cm corresponds to a silicon etch rate of 3.1A/S or to a silicon oxide etch rate of about 7 A/s.) For example, in 1% HF solution, ip is 5mA/cm, and thus the etch rale of the oxide film formed at a potential anodic of the first current peak is about 35 A/s. hi 2M KOH solution at room temperature, ip == 0.002mA/cm equivalent to an etch rate of about 0.014 A/s. These numbers appear to be in general agreement with the data in Table 4.1. 

Except for reaction path (3), which is purely chemical in nature, all the other reaction paths are of electrochemical nature, at least partially. These electrochemical reactions depend on the carrier transfer between the states at the interface and those in the semiconductor and thus their rates increase with increasing potential or illumination. While the reaction paths ( ), (3), and (4) result in the direct dissolution of silicon, the reaction paths (2) and (5) result in the formation of Si—O—Si bonds, the dissolution of which results in an indirect dissolution path. The rate of reaction paths (2) and (5) also increase oxide formation with potential. As the coverage of the surface by Si—O—Si bonds increases with increasing potential, the surface becomes increasingly less active and becomes passivated when these bonds fully cover the surface. Further reaction has to proceed via the breaking of Si—O—Si bonds, which is fast in HF solutions but very slow in KOH solutions. 

The measures of solid state reactivity to be described include experiments on solid-gas, solid-liquid, and solid-solid chemical reaction, solid-solid structural transitions, and hot pressing-sintering in the solid state. These conditions are achieved in catalytic activity measurements of rutile and zinc oxide, in studies of the dissolution of silicon nitride and rutile, the reaction of lead oxide and zirconia to form lead zirconate, the monoclinic to tetragonal transformation in zirconia, the theta-to-alpha transformation in alumina, and the hot pressing of aluminum nitride and aluminum oxide. 

In contrast to acidic electrolytes, chemical dissolution of a silicon electrode proceeds already at OCP in alkaline electrolytes. For cathodic potentials chemical dissolution competes with cathodic reactions, this commonly leads to a reduced dissolution rate and the formation of a slush layer under certain conditions [Pa2]. For potentials slightly anodic of OCP, electrochemical dissolution accompanies the chemical one and the dissolution rate is thereby enhanced [Pa6]. For anodic potentials above the passivation potential (PP), the formation of an anodic oxide, as in the case of acidic electrolytes, is observed. Such oxides show a much lower dissolution rate in alkaline solutions than the silicon substrate. As a result the electrode surface becomes passivated and the current density decreases to small values that correspond to the oxide etch rate. That the current density peaks at PP in Fig. 3.4 are in fact connected with the growth of a passivating oxide is proved using in situ ellipsometry [Pa2]. Passivation is independent of the type of cation. Organic compounds like hydrazin [Sul], for example, show a behavior similar to inorganic ones, like KOH [Pa8]. Because of the presence of a passivating oxide the current peak at PP is not observed for a reverse potential scan. 

Fig. 4.2 The chemical dissolution rate of (111) orien ted silicon surfaces in alkaline solutions is negligible as a result of insufficient polarization of the Si back-bonds by only one OH group.

If a silicon electrode is anodically oxidized in an acidic electrolyte free of HF, the oxide thickness increases monotonically with anodization time. This is also true for alkaline electrolytes if the oxide formation rate exceeds the slow chemical dissolution of the anodic Si02. This monotonic behavior, however, is not necessarily associated with monotonic current-time or potential-time curves. 

On the supposition that the total number of unit cells keeps invariable and no aluminum atoms are lost during the boronation, the composition of unit cell and the population of vacancies can be estimated as listed in composition of unit cell (I) in Table 2. It can be seen that the vacancies occupy about 30-50% of total T sites after the boronation. However, it should be noted that the population of vacancies thus obtained by chemical analysis is only a bulk average result. The composition on the surface of crystallites is actually different from that in the bulk because the dissolution of silicon starts first from the outer surface, so that the vacancies on the surface are much more than those in the interior of crystallites. Such a large number of vacancies on the surface will result in corrosion and dissolution of the surface parts of crystal particles. Therefore, the number of unit cells in the sample after the boronation is actually less than that before the boronation, whereas boron atoms in each unit cell should be more than those shown in composition of unit cell (1) in Table 2. On the other hand, if all the 64 T sites are occupied by silicon and trivalent atoms, we can give another set of compositions as shown in composition of unit cell (II) in Table 2. The real composition of a unit cell should be between these two sets of compositions, that is, the 64 T sites are neither occupied completely nor vacated so severely that the collapse of the framework occurs. It can also be seen that the introduction of boron atoms is so limited that there are no more than 1.5 atoms per unit cell even though the repeated boronation is performed. 

Chemical etching is a process for removal of silicon dioxide films through dissolution in solutions. Dissolution of silicon oxides, in the context of this book, is related to the anodic behavior of silicon electrodes. However, the dissolution of anodic oxides is not well studied. In contrast, there is a wealth of information on the dissolution of other types of oxides. Much of this information must also be applicable, at least the qualitative and mechanistic nature, to that of anodic oxides. Also, because oxides of different types are commonly used in device fabrication, compiling the etch rate data of these oxides and those of silicon (presented in Chapter 7) in the same volume would be useful in practice. Additionally, because silica-water interaction, which has been extensively investigated in the geological field, is fundamental to the etching of silicon oxides, some of the results from the investigations on the dissolution of rocks and sands are also included. 

---