1. Introduction

Silicon is probably the most important material for applications such as solar cells, sensors, and optoelectronic devices. In most of these applications, light collection is needed but the light collection efficiency of silicon usually suffers from a relatively high reflection at the air-silicon interface because of the refractive index mismatch of air and silicon. In the past, studies showed that one-dimensional (1-D) surface gratings can significantly modify optical properties of semiconductors [1–10] and efficiently suppress the reflection [7,11–13]. Usually, 1-D gratings are fabricated on silicon using optical or electron beam lithography techniques. Previously, studies have shown that 1-D surface gratings can be readily produced on surfaces by directly irradiating silicon using laser pulses, and such structures created using lasers are called laser-induced periodic surface structures [14–18].

Here, we study, for the first time, the antireflection effect of femtosecond laser-induced periodic surface structure (FLIPSS) in the wavelength range of 250 - 2500 nm. Our study shows that the FLIPSS pattern suppresses both the total hemispherical and specular polarized reflectance of silicon surface significantly over the entire studied wavelength range. The total polarized reflectance of the processed surface is reduced by a factor of about 3.5 in the visible and 7 in the UV compared to an untreated sample. The antireflection effect of the FLIPSS surface is broadband and the suppression stays to the longest wavelength (2500 nm) studied here although the antireflection effect in the infrared is weaker than in the visible. The FLIPSS-textured surface also exhibits a polarization effect in the visible region. Compared to other techniques, our technique has advantages of producing FLIPSS structures that are (i) free of chemical contamination, (ii) highly durable, and (iii) easily controllable in size from as small as a tightly focused laser spot, i.e. down to 10 μm, to as large as needed when a raster scanning laser beam is used.

2. Experimental details: setup and measurements

In this work, we produce femtosecond laser-induced periodic structures over a large area on a silicon wafer. To create FLIPSSs on silicon, we use an amplified Ti:sapphire laser system that generates 65-fs pulses with energy around 1.5 mJ/pulse at a maximum repetition rate of 1 kHz and with a central wavelength of 800 nm. The samples used in experiments are Czochralski grown, single-crystal 110-Si wafers (phosphorus-doped, with a resistivity of 1 - 30 Ω⋅cm). The front surface of the wafer is polished, whereas the back surface is chemically etched. The size of the wafer is 25 × 25 × 0.65 mm³. The laser beam is focused normally onto a sample mounted vertically on a X-Y motorized translation stage. An achromatic lens with a focal distance of 20 cm is used for beam focusing. We first scan the sample horizontally across the laser beam followed by a vertical shift and this process is repeated until a large area of FLIPSSs is produced. To find experimental conditions for producing FLIPSSs, the laser fluence, scanning speed in the X-direction, translation step along the Y-direction, and repetition rate of laser pulses were varied. In our experiment, the FLIPSSs are produced in air at the atmospherical pressure. Following the femtosecond laser treatment, the topography of surface structures is studied using a scanning electron microscope (SEM) and atomic force microscope (AFM). The total polarized hemispherical reflectance, R, of the treated samples is measured at an 8° angle of incidence over a wavelength range of λ = 250 – 2500 nm using a Perkin-Elmer Lambda 900 spectrophotometer with a Spectrlon® coated integrating sphere of 60 mm in diameter. We measure two polarized reflectance with the incident light polarization, E, either parallel or perpendicular to the FLIPSS grating vector g. The specular polarized reflectance of the treated sample is measured using specular reflectance accessories for the Perkin-Elmer Lambda 900 spectrophotometer. These specular reflectance accessories allow measuring specular reflection at the fixed incidence angle of 6° and variable incidence angle above 15°.

Figure 1(a) shows an SEM image of the 1-D grating produced at a laser fluence of 0.4 J/cm², scanning speed of 1 mm/s, and with the laser light polarization parallel to the laser beam scanning direction. The grating vector of the produced FLIPSS is parallel to the laser light polarization. We can see in Fig. 1(b) that the period of FLIPSS is about 575 nm and this value is significantly less than the laser wavelength used in our study (800 nm). We note that a significantly reduced FLIPSS period compared to the laser wavelength has also been observed for FLIPSSs generated on metals [19,20]. An average depth of the grooves measured with the AFM is about 90 nm. In contrast to a smooth grating produced by lithography techniques, the FLIPSSs are covered by finer nanostructures on both valleys and ridges of the grating structure. The nanostructures formed are in the shape of nanoprotrusions and nanocavities. Mean value of surface nanoroughness is measured to be about 17 nm. Surface area covered with nanostructures is about 17%. Figures 1(b)–1(d) show the optical images of the treated sample. The structured area appears dark gray to the naked eye at around normal viewing angles [see Fig. 1(b)], but various dark colors at large viewing angles [see Figs. 1(c) and 1(d)]. Figure 2 shows the measured total and specular reflectance of the treated silicon as a function of wavelength for the incident light polarization, E, either parallel (R_||) or perpendicular (R _⊥) to the FLIPSS grating vector g. For comparison, the reflectance of a polished sample before laser processing is also plotted as R_UNTR(λ). The specular reflectance accessories we use allow measuring specular reflection at a fixed incidence angle of 6° and variable incidence angle above 15° and do not allow measuring exactly at the incidence angle of 8° at which the total reflectance is measured. Therefore, we measured the specular reflectance at the incidence angle of 6° and 15°. The wavelength specular reflectance dependencies at these incidence angles were found the same within the measurement uncertainty and we believe that the specular reflectance at 8° is the same as 6°.

 

Fig. 1 (a) SEM image of femtosecond laser-induced periodic surface structures on silicon. (b)-(d) Photographs show that a silicon sample exhibits various shades of dark colors at different viewing angles.

Download Full Size | PDF

 

Fig. 2 Wavelength effect of the polarized-light total and specular reflectance at an incidence angle of 6° from the silicon sample structured with FLIPSSs. For comparison, reflectance of an untreated sample is also shown.

Download Full Size | PDF

3. Discussion

The optical band gap of single-crystal silicon is 1.1 eV (λ = 1.1 μm). Therefore, silicon is essentially transparent at wavelengths below the band gap (λ > 1.1 μm) and opaque at λ < 1.1 μm. We can see from Fig. 2 that FLIPSSs have a significant antireflection effect over the entire studied wavelength range (250 – 2500 nm). The most significant decrease in the total reflection is above the band gap at the wavelength range between 250 and 800 nm, where reflectance is reduced by a factor of about 3.5 for the visible wavelengths and by a factor of about seven in the UV. Below the band gap (λ > 1.1 μm), the total reflectance is reduced by a factor of 1.7 compared to an untreated surface. As seen from a comparison between specular and total reflectance in Fig. 2, the fraction of specular reflection in the total reflection is very small in the UV. However, it rapidly increases in the visible (up to about 60-70% of the total reflection) and reduces to about 50-60% for λ > 1.1 μm. The data in Fig. 2 also show that the FLIPSS surface exhibits clear polarization effect in the visible for both total and specular reflectance. Therefore, FLIPSSs essentially give a polarization-sensitive antireflection surface [13] in the visible.

It is known that optical properties of silicon can be modified through surface structuring and doping [3,4,21,22]. The surface structures produced in this study are surface gratings superimposed by sub-wavelength nanostructures (see Fig. 1(a)). A recent study has shown that randomly distributed sub-wavelength nanostructures on silicon can cause a significant reduction in reflection due to a better refractive index matching over an extremely wide range of wavelengths, from the UV to terahertz region [23]. Moreover, this type of silicon nanostructures suppress the reflection of both s- and p-polarized light over a wide range of angles. Therefore, the nanostructures superimposed on FLIPSSs created here may contribute to the suppression of light reflection in a similar way of a better refractive index matching. The effect of the surface grating on light reflection depends on the relation between the grating period and light wavelength. The FLIPSS period in our study is 575 nm [Fig. 1(a)] and therefore, the FLIPSS period, d, can be larger or smaller than a particular wavelength in the studied wavelength range (λ = 250 –2500 nm) and we will consider the light reflection in the following three cases: d < λ, d > λ, and d ≅ λ. First, it is known that the so-called sub-wavelength gratings (SWGs, d < λ) have antireflection effect [13,24]. In this case, only the zeroth diffraction order exists while higher diffraction orders are evanescent. The antireflection behavior of SWGs can be described by an effective medium theory [7,9,11,23–25]. In this theory, SWGs is modeled as a homogeneous layer with a thickness equal to the grating groove depth and an effective refractive index characterized by a filling factor of the grating. According to Ref. 25, this case of d < λ is called the homogenization region of a grating. For d > λ, the so-called geometrical optics region, grating reflection of light can be reduced by light trapping effect in the grating grooves as in cavities [26]. Finally, for d ≅ λ in the so-called strong diffraction region, surface gratings on undoped silicons also exhibit antireflection behavior [8]. Suppression of reflection by surface gratings has been demonstrated on doped silicon as well [3,4,21,22]. In this case, surface plasmon polaritons can be excited and further modify the optical properties of silicon [4,21].

Durability of silicon surfaces is important in many applications. The durability tests for our FLIPSS surface shows that the surface structures produced here can be washed in solvents such as methanol and acetone and wiped with cleaning tissue without any noticeable loss in performance. This is an advantage over the porous silicon produced by chemical etching methods, which can be easily damaged [27]. Furthermore, our direct laser structuring technique is also free of chemical etching contamination and is suitable for processing complicated surface shapes. The surface area size to produce FLIPSSs can be as small as a tightly focused laser spot, i.e., down to about 10 μm, and as large as needed when a raster scanning laser beam is used.

4. Conclusions

In summary, following direct femtosecond laser pulse irradiation, we produce a unique grating structure superimposed by finer nanostructures on a silicon wafer. We study, for the first time, the antireflection effect of this surface grating in the wavelength range of 250 - 2500 nm. Our study shows that the FLIPSSs suppress the total and specular polarized reflectance of silicon surface significantly over the entire studied wavelength range. The antireflection effect of the FLIPSS surface is broadband and the suppression stays up to the longest wavelength (2500 nm) studied here although the antireflection effect in the infrared is weaker than in the visible. The FLIPSS-textured surface also exhibits a polarization effect in the visible region. Compared to other techniques, our technique has advantages of producing FLIPSS structures that are free of chemical contamination, highly durable, and easily controllable in size from as small as a tightly focused laser spot, i.e. down to 10 μm, to as large as needed when a raster scanning laser beam is used.