Etching of silicon in fluoride solutions
Introduction
Gerhard Ertl’s first two publications were written with Heinz Gerischer [1], [2]. From a reading of these papers it would appear that it was under Gerischer’s aegis that Ertl began to recognize that true dynamical understanding of elementary processes in surface reactions [3] required the establishment of an approach that had all of the rigor of the gas-phase school of chemical reaction dynamics. Ertl began by building up from the structure of the surface [4], [5]. This was a necessary prerequisite that would be followed by development of the electron and vibrational spectroscopic techniques, which are required to obtain a true molecular understanding of chemical transformations on surfaces. This new school has come to be called the surface science approach. To reduce complexity and obtain the requisite rigor and reproducibility, Ertl turned away – at least at first – from the liquid/solid interface and moved to the gas/solid interface.
Coincidentally, the timeframe of their move to the University of Munich is also when Gerischer first turned to the study of the etching of semiconductors in aqueous solutions. The first studies were on germanium electrodes [6] with silicon following some years later [7]. Ertl’s first surface chemical studies were also on Ge surfaces [8], [9], and the two shared an interest in the elucidation of the fundamental properties of semiconductor surfaces [10]. Gerischer recognized that semiconductor electrodes provided a unique platform for the study of the dynamics of electrons and holes in electrochemical reactions [11] as well as for surface photochemical studies [12].
Studies of the electrolyte/semiconductor interface have had a profound impact on the field of electrochemistry as well as solid state physics [13]. One needs only mention the multifaceted TiO2 [14] or its photocatalytic properties and use in Grätzel type dye-sensitized solar cells [15], [16]. Semiconductors, of course, are at the heart of photovoltaic cells used for hydrogen production [17], [18] or as solar cells [19], [20]. Gerischer’s early work on the electrolyte/semiconductor interface soon turned to studies of photoelectrochemical energy conversion, and his work lead to fundamental advances in this area [21]. It is on semiconductor surfaces that the link between electrochemistry and photochemistry was made.
The etching of silicon in fluoride is a wonderful example of how identifying reactants and products, as well as initial and final thermodynamic states, gives us no insight into surface chemical processes. What is required is a surface science approach to elucidate the dynamics. The overall corrosion reaction for Si dissolution in fluoride media under the most commonly studied conditions is given by the equationin which h+ represents a hole injected into the valence band and an electron is injected into the conduction band. At low illumination intensities on n-type Si, there appears to be a second competing reaction.Reaction (1.1) is responsible for a process known as current doubling and Rxn (1.2) for current quadrupling. In current doubling the photocurrent quantum yield is two, that is, for each absorbed photon two charge carriers contribute to the measured photocurrent. In photocurrent quadrupling, four charge carriers are counted for every one photon absorbed. These simple electrochemical reactions belie a wealth of complexity in the reaction dynamics, which can only be completely understood by taking a surface science approach to the electrolyte/semiconductor interface.
First we need to recognize that fluoride solutions are themselves complex and not entirely well understood [22]. To state that HF exists either in an undissociated or dissociated form in water is an oversimplification. Instead it appears to oscillate between an undissociated form and a contact ion pair in which both H+ and F− are bound strongly on either side of a molecule of water [23], [24]. HF(aq) contains not only solvated H+ and F−, but also complex ions such as and , any and all of which may be able to participate in the reaction mechanism.
Second, we need to ask why Si does not spontaneously dissolve in acidic fluoride. When kept in the dark, the etch rate of H/Si in concentrated HF, Rdark, is roughly 2.5 × 1012 cm−2 s−1 (∼0.3 Å min−1) [25], [26], [27]. In other words, unbiased and unilluminated Si is virtually inert in acidic fluoride solutions even though the formation of the products in Rxn (1) is thermodynamically favored. Keep in mind, of course, that fluoride solutions do spontaneously etch silica surfaces,These reactions, in which is the more reactive of the two species [28], [29], are important to recognize whenever investigating the interaction of Si surfaces with aqueous solutions because Si surfaces generally start with a native oxide film or may become covered with an oxide layer under certain conditions. While the reaction rate depends on the composition of the fluoride solution, generally the dissolution of silica is very rapid and isotropic. Hence, dissolution of Si appears to be kinetically hindered, whereas dissolution of SiO2 is not.
Third, we can ask why etching in alkaline solutions leads to such different final states compared to etching in fluoride. Alkaline etching of silicon follows the reaction [30]While this reaction appears to be completely different than reactions (1.1) and (1.2), we shall see below that essentially all of the steps in F− and OH− induced etching of Si are the same except the crucial first step. Because of this alkaline etching is a chemical reaction that occurs catalytically under the influence of OH− [31]. Alkaline and fluoride etching of silicon are both anisotropic but in very different respects. We shall see that because of this, alkaline etching, which is much the same as the so-called chemical reaction that occurs in moderate to high pH fluoride solutions, can be used to create flat, nearly perfect surfaces, while fluoride etching can be used to form nanocrystalline porous silicon films. The versatility of fluoride etching also allows it to be used in the selective area formation of macropore arrays [32], silicon nanowires [33], [34], and in micromachining [35], [36].
Section snippets
Surface termination
A bare silicon surface is highly reactive and would never survive for long when exposed to either air or an aqueous solution. Silicon exposed to the atmosphere rapidly oxidizes to form a native oxide layer with a thickness of several angstroms [37]. After degreasing in solvents, when a native oxide surface is placed in a fluoride solution, Rxns. (2.1) and (2.2) rapidly remove the oxide. This is a fundamental difference between fluoride and alkaline (non-fluoride containing) solutions. If the
The initiation step: role of holes
The role of valence band holes in controlling anodic oxidation of semiconductors was recognized by Brattain and Garrett for Ge [83] and Uhlir for Si [84]. Beck and Gerischer [85] proved that the reaction rate on p-type Ge is proportional to the surface concentration of holes. Because of the different reactivities of surface electrons and holes, the doping type of the crystal leads to very different reactivities for n-type and p-type doping (see Fig. 1).
The underlying premise of Gerischer’s
The Gerischer mechanism step by step
The revised and improved Gerischer mechanism of silicon etching in fluoride solutions is presented in Fig. 3. This figure contains not only Steps (1) through (5a), which correspond to an improved version of the revised Gerischer mechanism, but also Steps (5b) through (7), which comprise the current quadrupling branch not included in the original model. Whether or not both branches are required and which is the dominant branch is still a question in need of resolution.
The first chemical change
Conclusion
The development and current status of an extended and revised Gerischer mechanism of silicon dissolution in fluoride solutions has been presented. The model is an example of how the surface science approach is essential for molecular level mechanistic understanding of etching reactions. Several concepts are key to understanding etching by both fluoride and hydroxide: (i) surface passivation provided by a H-terminated surface, (ii) the role of the initiation step in determining the rate,
Acknowledgement
A reliable source once told me that Heinz Gerischer considered Gerhard Ertl his brightest ever Doktorant. The Swedish Academy’s awarding of the 2007 Nobel Prize in Chemistry confirms this assessment spectacularly. It was a pleasure and honor to work in Ertl’s Abteilung at the Fritz-Haber-Institut in Berlin; though oddly based on the topic of this manuscript, it was only some years later that I took an interest in silicon etching in fluoride solutions. Gerhard Ertl’s towering professional and
References (116)
Surf. Sci.
(1967)Surf. Sci.
(1967)- et al.
Electrochim. Acta
(1968) Surf. Sci.
(1969)Electrochim. Acta
(1990)Curr. Opin. Solid State Mater. Sci.
(2006)- et al.
Comptes Rendus Chimie
(2006) - et al.
Sol. Energy Mater.
(2006) Surf. Sci.
(1994)- et al.
Curr. Opin. Solid State Mater. Sci.
(2005)