1. Introduction

In 1956, Uhlir [1] first formed a porous silicon (Si) thin film by electrochemical etching. Then, Canham [2] observed emission spectrum on porous Si, and confirmed the quantum confinement model in 1990. After that, numerous formation skills for porous Si were rapidly developed, including electrochemical anodization [3,4], chemical etching [5–7], vapor etching [8,9], and optical laser breakdown [10,11]. Based on nano-scaled pillars/pores and high surface volume ratio, porous Si is widely used in many applications, such as gas sensors [12], biosensors [13], solar cells [14], photoelectric sensors [15], photoluminescence (PL) devices [16–18] and electroluminescence diodes [19]. Commonly, the porous Si films were fabricated by electrochemical anodization [20] in hydrofluoric acid (HF), also used in this study as well. The overall chemical reaction is represented below [21].

In addition, most researches on porous Si used to focus on the one-dimensional (vertical) etching on sample surfaces previously [3,6,17,26]. In the study, based on the grid ditches, lateral etchings occurring on the sidewalls around the ditched zone-diodes are further studied. The combination of vertical etching and lateral surrounding etching forms a three-dimensional electrochemical etching on the Si films and produces various porous Si structures emitting visible lights with different wavelengths. The different porous Si structures can be formed on a single Si wafers without any mask and lithography process, and those microstructures are examined by scanning electron microscope (SEM) and by PL-spectrum under ultraviolet (UV) light stimulation.

2. Experimental

As shown in Fig. 1(a), in our experiments, a vertical Teflon container with an area of 5.3 cm² and an electrolyte solution of HF: C₂H₅OH=1: 4 are used. All samples were prepared on (100) crystal Si with a resistivity of 27∼33 Ω·cm. Before the electrolyte etching, samples were ditched patterns by laser on their top surfaces. After an etching treatment, followed by the formation of porous Si, the samples were examined their UV-PL response (light intensity versus wavelength) under UV light stimulation, of which photographs on the samples’ top surfaces are called “UV-photos” in the paper for simplicity. Based on the p-layer assistant method with a forward bias on the pn-junction [22], the holes in p-layer can be charge-pumped to the top of n-layer to participate in the etching reaction, as shown in Fig. 1(b), which mechanism is called “hole-pumping” [27]. Of course, while the pn-junction is turned-off, the hole-pumping mechanism is blocked, as shown in Fig. 1(c). In this paper, the turned-on and turned-off diodes are placed together on a single wafer, similar to a colored jigsaw puzzle and a black one tiling together. As a primary experiment (Sample-P0), a pn-structure was open-cut with a depth of 100 um by a laser ditching contour into an inner diode and an outer diode, whose pn-junctions were turned on and off, respectively, during the etching process. The Sample-P0 shows a clear yellow light emission on the inner pattern area of porous Si under UV stimulation with a wavelength of 365 nm, while no light on the outer area, as shown in Fig. 1(d). Such a new patterning method for porous Si is simple, low-cost, mask-free, and conducive to integrated circuit manufacturing, as compared to conventional photolithography with mask patterning, e.g., ion implantation masks, photoresist masks, multilayer electrochemical resistant masks, silicon nitride masks [28–30].

 

Fig. 1. (a) Schematic sketch of experimental setup for samples. (b) Illustrative sketch of hole-pumping effect with a turn-on pn-junction. (c) A turn-off pn-junction blocks hole-pumping mechanism. (d) The UV-photo of Sample-P0 and its layout with contour ditching. (e) The UV-photos of Samples-Q and Q’ with their same layout with grid ditching. (f) Mechanism drawing of different lateral etching conditions on Sidewalls-1∼5.

Download Full Size | PDF

3. Results and discussions

In addition, on Sample-Q and Q’ as shown in Fig. 1(e), not only contour ditching but also grid ditching were applied, and then more inner zone-diodes were formed inside the contour area. Under UV simulation, different light color extends on Sample-Q and Q’. The details on different porous Si zone-diode will reveal a clear insight into lateral etching mechanism.

The sketch of lateral etching mechanism is shown in Fig. 1(f), in which the Sidewalls-1∼12, Ditch_1-2∼Ditch_11-12 and Zones-A∼E were labeled in figures for expression. Obviously, on sample-Q/Q’, visible light emissions only occur in inner zones, while not in outer zone, similar to Sample-P0. Different from Sample-P0, there were grid-ditches on sample-Q/Q’ that should be more concerned. Considering the chemical reaction under a constant etching current, the negative ions in Ditch_3-4 (for example, carrier concentration n = -q) must be shared by Sidewall-3 (n ≅ -0.5q) and Sidewall-4 (n ≅ -0.5q), in which the hole-pumping both in Zones-B and C equally. On the other hand, those negative ions in Ditch_1-2 (n = -q) cannot react with sufficient holes on Sidewall-1 because of lack of hole-pumping. Therefore, most of the negative ions react on Sidewall-2 (therefore n ≅ -q). It is easy to predict that the lateral etching current density on Sidewall-2 is almost two times larger than that on Sidewalls-3 or 4.

The UV-photo of Sample-Q/Q’ in Fig. 1(e) show the real implemented results. Various etching current densities cause different porosities on different zones emitting lights with different wavelengths. To further investigate the effect of lateral etching on the light emission, the relationship between light wavelength (λ) and current density (J) must be clarified first. The second part of preliminary experiments (Samples-P1∼P5) based on a circle pattern area (A=0.79 cm²) with different etching currents (I=25, 30, 35, 40 and 45 mA) for 600 s was studied, and the UV-PL spectrums measured on the inner zones with their UV-photos were shown in Fig. 2. It is clear that a higher current density causes a blue shifting on wavelength.

 

Fig. 2. The UV-photos of Samples-P1∼P5 and their UV-PL spectrums in their inner zones.

Download Full Size | PDF

Figure 3(a) shows the UV-photos and corresponding UV-PL spectrums of five cases (i.e., Samples-Q1, Q2, Q3, Q4 and Q5) based on the same pattern (13-zones) with the same etching time (600 s) with Sample-Q in Fig. 1(e), but with various etching currents of 50, 60, 70, 80 and 90 mA for comparison. Since the pattern has point symmetry, Tiers-1∼4 refer to the four group zones with distance relating to the center point from inner to outer. Obviously, firstly, the zones on the same tier have almost the same wavelength; and secondly, the wavelength is blue shifting from Tier-1 to Tier-4 in order. The cross section views of microstructures near the ditches on Districts-Z1 and Z2 on Sample-Q4 and Q5 are examined by SEM (JEOL-JSM7610F) shown in Fig. 3(a). Both on Districts-Z1 and Z2, it is clearly proved that the lateral etching occurred only on the sidewall with hole-pumping. However, it does not clearly occur on the other sidewall without hole-pumping that is blocked by pn-junction. The results explain why the multiple wavelength lights take place simultaneously even on the same area zone with same vertical etching current (i.e., same vertical current density). Although all square zones are under the same vertical etching condition, but different tiers of zones (Tiers-1∼4) have different lateral etching on their own four sidewalls. Under a constant lump sum etching current, the negative ions in each ditch (e.g., concentration n = -q) may be fully react on one sidewall due to the other sidewall has no hole-pumping (n = -q), or may be half-shared for each adjacent sidewalls (n = -q/2). When all of the ion concentrations on surrounding four sidewalls summed up for each zone, it is found that the total negative ion concentration n of the zones in Tier-1 is 2q^-, Tier-2 is 2q⁺, Tier-3 is 3q, and Tier-4 is 3.5q. (The signs of + and – mean more and less ion concentrations caused by the corner effect, which is minor effect and not discussed here.) It also means J_T1 < J_T2 < J_T3 < J_T4, in which J_T1, J_T2, J_T3, and J_T4 are the lateral current densities on the zones of Tiers-1∼4, respectively. It explains the blue shifting on the wavelength of zones from Tier-1 to Tier-4 in order. The tendency is consistent with the data measured in PL spectrum as shown in Fig. 3(a).

 

Fig. 3. (a) PL-spectrums measured on the zones of Tier-1∼4 for Samples-Q1∼5, respectively, with their corresponding UV-photos. In addition, the SEM photos measured on the districts near Ditch_1-2 and Ditch_11-12 of Samples-Q4 and Q5, respectively, are shown. Lateral etching are clearly observed on Sidewalls-2 and 11 of Samples-4 and 5, respectively. (b) UV-photos and related SEM phots of Samples-P6∼P9.

Download Full Size | PDF

To explore the mechanism associated with porosity-induced color changing, the third part of preliminary experiments (Samples-P6∼P9) are made, as shown in Fig. 3(b). From the SEM photos, the porous Si structure in Sample-P9 cracks more severely than Samples-P8∼P6. In addition, it is clear that a higher etching current density causes a higher porosity structure emitting the light with a shorter wavelength (colors from red to green) as shown in the UV-photos. The results are consistent with the quantum confinement model [2], in which the smaller Si pillars in the higher porosity structures cause blue shifting in the light emission.

In addition, to understand the sequence of forming porous Si on the grid ditched inner zone diodes, six cases (Samples-R1∼R6) based on a same etching current of 70 mA and on a same pattern with Sample-Q are traced on six etching time points of 100, 200, 300, 400, 500 and 600 s, respectively, and shown in Fig. 4(a). It reveals three points. First, for the zones on the same tier, the UV-PL spectrums are almost the same. Second, the peak wavelengths are blue shifting from Tier-1 to Tier-4 (from the inner to the outer) in order. Third, as the etching time is extended from 100 s to 600 s, the peak wavelengths blue shift in order.

 

Fig. 4. (a) The peak values of wavelengths in UV-PL-spectrums measured on the zones of Tiers-1∼4 for Samples-R1∼R6 to trace the six etching time points of 100, 200, 300, 400, 500 and 600 s, respectively. The relevant UV-photos are inset for comparison. (b) The UV-photo and normal-light photo of an interesting attempt on color patterned light emission drawn like a half piece of the Picasso’s painting (Deux Hirondelles) on Si chip.

Download Full Size | PDF

Based on the multi-wavelength light emission approach with three-dimensional etching, one interesting attempt is made on patterned light emission drawn like half piece of Picasso’s painting (Deux Hirondelles) under UV stimulation on a small Si chip, as shown in Fig. 4(b). Under a constant etching current, different zone areas cause different current densities, and then cause light emissions with different wavelengths (colors). In other words, an appropriate pattern design can produce multi-wavelength light emission. Especially, the combination of the lateral etched porous Si pillars on the vertical etched porous Si pillars leads to a light emission with mixed wavelengths (colors) from the mixed porous Si structures. Although the study focuses on the UV-PL, the proposed approach is also suitable to EL. With a proper switching on UV light (for PL) or electrical power (for EL) stimulations, the approach might be applied to the application on the dynamic light emission with multiple wavelengths/colors. As to the pixel resolution of the inner zone, it is also still our fabrication challenge and depends on the limitation of laser beam. In our laboratory, based on the laser machine with a power of 30 watt, a beam diameter of 0.018 mm, wavelength of 1064 nm, and a scan position repetition of ±0.025 mm, then a minimal size (resolution) reached currently is 3 mm in the inner area under HF etching. It is for sure that the resolution minimization must be dependent on the fine-tuning and balance of current density and etching time.

4. Conclusions

In conclusion, a new approach is originally proposed to form a multi-wavelength light emission on a single porous Si chip, based on three-dimensional electrochemistry etching on grid ditched pn-diode structure. This approach is useful for a multi-wavelength light emission chip, wavelength tuning by area size and mask-free patterning. The patterning design with/without hole-pumping, porous Si structures with different porosities on different zones causes a precise light emission pattern consistent with ditching contour. The grid ditching on the inner zone encourages the lateral etchings, which occur on different sidewalls under different conditions, and causes light emissions with different wavelengths. In the study, it is found that a higher current density causes blue shifting on wavelength. Moreover, the ion concentration sharing mechanism on different patterns can be used for the fine-tuning on the lateral etching current density. Therefore, the proposed new approach provides a simple and optimal way for fine-tuning of the multi-wavelength light emission on porous Si.