1. Introduction

Porous silicon (PS), a silicon (Si)-based material consisting of nano-scaled pores and pillars, is first discovered by Uhlir [1,2]. After four decades passed, the visible light-emission on PS was proposed by Canham [2,3] and led to considerable interests in the development of PS films for yielding photoluminescence (PL) and electroluminescence (EL) [4–7]. In addition, PS has also attracted considerable attention on sensor applications, such as gas sensor, chemical sensor [8–11], etc. And, it has been extended to the bio- and medical-related fields in recent years, e.g. the bio-sensor, drug delivery and medical imaging [12–14]. Also, it is employed as the substrate layer of integrated circuits due to low loss and tunable permittivity for integrating passive devices into a single monolithic chip [15,16].

To produce PS films, several methods have been introduced, including anodization electrochemical etching [17,18], stain etching [19,20], spark erosion [21] and vapor etching [22,23]. Among these approaches, the anodization electrochemical etching is the most common one due to manufacturing convenience and its low cost. During the anodization electrochemical etching, the hydrofluoric (HF) concentration of electrolyte solution, the hole (h⁺) concentration of Si wafer, and the anodization current density are the three essential factors to control the profiles of PS structures, such as pore sizes and pillar depths [2,24].

Some ideas about the h⁺ concentration controlling have been demonstrated in our previous works [25]. In this paper, anodization current distribution and its related edge effect are thoroughly studied by the extra halo baffle design in the electrochemical etching process to observe more details about the curved routes of reactant flux. The differentiation of flux distribution on different areas may cause a patterned PL under ultraviolet (UV) light excitation.

On the other hand, in the past, complicated procedures were required to form a patterned PS-based devices [26–28] since the photoresist in photolithography method would be erosive in HF acid during implementing PS. One simple way has been proposed by our team in previous work [29]. In this paper, based on the edge effect, another selective patterning technique on PS films, originally proposed here, could be alternatively achieved by the baffle design.

2. Experimental setup

Figure 1 shows the sketch of experimental apparatus with four kinds of baffle types (named Cases-A~D) adopted in the experiments. In addition, Teflon material is employed as a container of the HF electrolyte for PS samples preparing. The electrolyte is a 1:3 solution of HF: C₂H₅OH. In the container, the anode and cathode are placed vertically to easily remove the hydrogen bubbles from the surface of PS samples [24]. Meanwhile, a copper (Cu) disk serves as the anode contact to the Si wafer and a silver (Ag) plate serves as the cathode. All PS samples are prepared on (100)-oriented p-type Si wafers with the resistivity of 1~10 Ω⋅ cm. To keep a stable etching condition, a high performance potentiostat/ galvanostat meter is utilized to supply a steady voltage of 15 Volts during the electrochemical etching process for 50 minutes. All experiments and measurements are conducted under room temperature. In the four cases (Cases-A~D), the baffles are placed at the distance d to the surface of Si wafers, and the thickness of the baffles are all in 6 mm. The area inside O-ring is 3.8 cm² and the area of hollow in Cases-A~D are 0.28, 0.64, 0.36 and 0.58 cm², respectively. Since the etching current density is one of crucial parameters during electrochemical etching, two kinds of current densities in this work are shown in Table 1. One is based on the area of circular O-ring, J_o, which is the area of Si wafer contacting to the electrolyte; the other one, J_b, is based on the area of bottle necks on various baffle hollows. Also, five conditions of distances (d = 0, 1, 3, 5, and 10 mm) are investigated in Cases-A~D, and all of the twenty PS samples are labeled as A00, A03, A05, A10, B00, …, and D10, respectively.

 

Fig. 1 Schematic diagram of experimental setup for preparing PS samples with four baffle designs including circular- (Case-A), square- (Case-B), triangle- (Case-C) and triple-circular- (Case-D) hollows inside.

Download Full Size | PDF

Table 1. Current Densities of (J_b, J_o) at baffle hollow and at Si-wafer inside O-ring for all cases.

View Table

3. Experimental results and discussions

Figure 2 shows the naked eye photos of the PL patterns on all PS samples under the excitation of UV light with the wavelength (λ) pitched at 365 nm. It is clear that the PL patterns change significantly as d varies, and the hollow shapes of the baffles are marked on some PS samples. When d = 0 and 1 mm, sharp patterns appear on the PS samples. On the other hand, as d is changed to 3 or 5 mm, the transferred patterns are vague. However, as d increases to 10 mm, no more transferring pattern can be found. As to the effect of dimension on the radius, it is simply demonstrated in Case-D (radius = 2.5 mm) for comparison with Case-A (radius = 3.0 mm). With different dimension sizes, both cases reveal similar results. It show the clear PL patterns and similar phenomena in the smaller-shape case also can be produced especially in the distances of d = 0 mm or 1 mm.

 

Fig. 2 The naked-eye photos of all twenty PS samples under UV-light excitation at room-temperature, corresponding to four baffle cases (Cases-A~D) and five distance conditions (d = 0, 1, 3, 5, and 10 mm).

Download Full Size | PDF

In Fig. 3, furthermore, a PL spectrum measured on the three particular districts (Green-, Red-, and Dark- districts) of the typical patterned PS sample (Sample A01) which sample has a clear edge effect is shown for demonstration, corresponding to the scanning electron microscope (SEM) photos for comparison. Obviously, different porosities can emit light in different colors. The variation trends of porosity distributions on SEM photos and PL spectrums on distinct color districts on Samples-B01, C01 and D01 with different hollow shapes are very similar to Sample-A01 (in Fig. 3), which are not shown here for simplification in the article.

 

Fig. 3 (a) The corresponding top-view SEM pictures for Green-, Red-, and Dark- districts are shown, respectively; (b) PL spectra of Sample-A01 for demonstration.

Download Full Size | PDF

Relatively, in the conventional methods without baffles, the reactant flux distribution is usually supposed to be uniform in the electrolyte and the non-uniform distributions of flux at the edge areas is ignored easily, since the edge effect is very slim and hard to be found. In this study, based on the delicate baffle designs, the edge effect is expanded and large enough to be noticeable and distinguishable.

In the etching on PS samples, the illustrative equation of electrochemical reaction between electrolyte and Si-wafer is expressed as below:

The fluoride ions in the HF-based electrolyte are drifted continually by electric field to reach the Si wafer surface and react with hole-carriers (h⁺) and Si atoms to form the product of H₂SiF₆. Since the H₂SiF₆ molecules are soluble, the nano-sized pores in the crystal Si wafer are generated after the H₂SiF₆ molecules are dissolved in electrolyte and leave the vacancies. Hence the porous structures could be formed. It is noticed that different reaction rates would cause different porosities and structures, as well as different color light emissions.

The mechanism of edge effect is illustrated in Fig. 4 with the reactant flux lines sketched in the electrolyte and Si wafer between anode and cathode. Some flux lines in the center can penetrate the hollow directly; however, the other flux lines near the edge are curved, as shown in Fig. 4. And, the reactants will move to the interface between the electrolyte and Si along the flux lines, and then react in to form a PS structure.

 

Fig. 4 Illustration of the mechanism of edge effect.

Download Full Size | PDF

During the reaction process, the reaction rate (R) is mainly dependent on the supplying speed (v) of reactants. The speed could be approximately described by v = uE = u(V/L) where u is reactant mobility, E the electric field strength, V the potential difference ( = 15 Volts in this study), and L (≅ L₁ + L₂) the total displacements from cathode to baffle hollow (L₁) and then to Si wafer (with anode) (L₂). As depicted in Fig. 4, the shorter L along the inner flux line would results in a higher v and higher R than that along the outer flux line. With a higher R, the porous structure in the center region is more significant than that in the outer region. Such reasoning could be verified in the SEM photos in Fig. 3, in which the porous structure in the central part is reef like; however, outer region is tissue-like structure. Under UV excitation, the reef-like structures emit green light, and tissue-like structures emits orange light. It is worth noting that no color light appears in the center districts of Samples-A00, B00, C00, D00, C01, and D01 in Fig. 2. Following the model presented in Fig. 4, the center districts are under the highest reaction rates (R) which are too high to cause the porous structures to change into the electro-polishing stages and then be polished [30]. The SEM photo of district-a of Sample-A00 in Fig. 5 proves the situation.

 

Fig. 5 Naked-eye photos and related SEM pictures on different districts with different color light emission on Sample-A00, A01, A03, A05, and A10.

Download Full Size | PDF

To further study the relationship between the topology of PS microstructure and its light emission, a detailed view from SEM in Case-A with the parameters of d = 0, 1, 3, 5, 10 mm are shown in Fig. 5 for demonstration, in which the samples are labeled as A00, A01, A03, A05, and A10, respectively. It is easily found that the green lights emit from the reef-like PS structure, such as b-district on Sample-A00 as well as a-district on Samples-A01 and A03. Perversely, some tissue-like PS structures emit orange lights, such as b-districts on Sample-A03, A05, and A10. The results are fully consistent with that in Fig. 4. It indicates that the more concentrated the flux, the higher the reaction rate will be, which causes reef-like structures (precipitous PS islands) with green light emissions, in contrast with the tissue-like structures with orange light emissions. We believed that the quantum confinement effect is more significant in such precipitous PS islands, and a higher frequency light could be emitted. It is also found that the edge effect is declined and the shape becomes vague gradually in PL pattern as the d increases. It indicates that if the distance d is not large enough to let the flux spread out to a uniform distribution, a non-homogeneous profiles would be found obviously.

4. Conclusions

The edge-effect on electrochemical etching process for PS films is originally investigated by a baffle design in this study. The clear evidence of edge-effect on PL patterns and SEM photos shows that the electrochemical flux could be curved by the baffle edge, and a non-uniform flux distribution would be observed. The position arrangement of baffle, d, is one crucial parameter on the edge-effect. When d is shorter enough, the PL pattern of edge effect becomes sharper. However, as the d is larger enough for the curved flux can uniformly spread out, the edge effect becomes indistinct. For reproducibility, four kinds of baffles with circular, triangle, square and triple-circular hollows are designed in our experiments, and the same results and tendency are obtained. Based on the edge effect, the PL patterns with different PS structures and different visible light emission can be obtained on the same sample at the same time. The experiment results proposed in this article can serve as very good evidence of edge effect in electrochemical etching and also provide a possible mask-less patterning technology for PS devices.