1 Introduction

During the last decade, photovoltaic (PV) production has increased rapidly. It is currently dominated by multicrystalline silicon (mc-Si) and its demand is increasing each year [1]. Mc-Si is a competitive material for photovoltaic industry due to many reasons such as non-toxicity, abundance, low production cost, which are the most advantage reasons compared to the monocrystalline substrate, and it is also one of the most friendly material for the environment [2]. To minimize the reflection from surface, multicrystalline silicon wafers were nanostructured using many methods. Several approaches have been proposed to reduce the reflectivity of silicon surface such as laser scribing [3, 4], double anti-reflection coating [5], porosification [6], texturization [7, 8], mechanical grooving [9] and so on. In this content, chemical etching the silicon surface into nanowires appears very advantageous to reach this goal and to improve the efficiencies of industrial multicrystalline silicon solar cells. Recently, due to their unique structural, electrical, optical, and thermal properties, silicon nanowire (SiNW) arrays have received significant attention. It is broadly used in lithium ion rechargeable batteries, energy conversion and storage devices, chemical and biological sensors, thermo-electrics devices, photocatalytic applications [10, 11], supercapacitors, photodetectors and light emitting diodes [12].

There are several methods to produce nanowires including vapor liquid solid growth mechanism, solution-liquid solid growth mechanism, gas solid growth mechanism, laser assisted catalytic growth, template-based synthesis methods, physical vapor deposition and chemical vapor deposition (CVD) methods [13–15].

In addition to these methods, metal assisted chemical etching (MACE) technique, is a top down approach based on the oxidation-dissolution consecutive reactions between the reagents and the silicon substrate [16]. This technique is very suitable for producing large area oriented nanowire  arrays on silicon substrates at room temperature under ambient atmosphere. This method does not need any vacuum equipment and can be performed even on large surfaces in a short time [17]. Today, MACE is considered to be the most popular technique that produce a mass production of SiNWs.

In this study, silicon nanowires have been grooved on multicrystalline silicon wafer via a facile and low cost method starting from inexpensive silicon wafers. On the basis of characterization of the obtained SiNW arrays, we present a correlative study between morphological, optical and structural properties of the nanostructured surface in connection to the etching times. Morphological characterizations of etched silicon have been conducted using atomic force microscopy (AFM) and scanning electron microscopy (SEM). The optical reflectance measurements were performed on an UV–Vis–NIR spectrophotometer to show the reflectance of the nanostructured surface. Furthermore the structural properties was examined by using Raman spectroscopy.

2 Experimental procedure

Figure 1 illustrates the fabrication process of silicon nanowires on multicristalline silicon. Firstly, The mc-Si wafers with a thickness of 380 μm were cleaned in an aqueous acid solution (HNO3: 64%, HF: 16%, CH3COOH: 20%) for 30 s followed by copious rinsing with deionized water then dried. After that, the clipped samples are sequentially cleaned by ultrasonication in acetone for 10 min, ethanol for 10 min, and deionized water for 10 min. In the first step, clean surfaces were placed in a teflon beaker containing HF (4.6 M)/AgNO3 (0.03 M) aqueous solution for 1 min. Silver (Ag) nanoparticles are deposited on mc-Si by electroless displacement reaction. The second step consist of immersing Ag decorated wafers in the mixture of 4.6 M of HF (40 wt%) and 0.1 M of H2O2 (30 wt%) solution and we leave the samples to be etched for five various durations (5, 10, 20, 30 and 60 min). After etching, the synthesized samples were rinsed copiously in deionized water and dried at room temperature, then, they were dipped thirdly in 30 wt% HNO3 aqueous solution for 15 min to remove residual silver nanoparticles (AgNPs) from surfaces. Finally, the samples were washed with de-ionized water and dried in air before characterization. The modified surface have been investigated by using SEM, AFM, optical reflectance and Raman spectroscopy.

Fig. 1
figure 1

Illustration showing the fabrication of nanowire arrays on multicrystalline silicon

Full size image

3 Results and discussions

3.1 Atomic force microscopy

Atomic force microscopy is a non destructive apparatus useful for the investigation of surface morphologies. It was used to observe the surface topography and to calculate the roughness of our samples. As it can be seen in Fig. 2, AFM topography show a remarkable modification of the surface morphology of six various samples (Si Ref, SiNW5, SiNW10, SiNW20, SiNW30 and SiNW60 corresponding to bare mc-Si, mc-Si etched for 5, 10, 20, 30, and 60 min, respectively). For an etching time of 5 min, the Si nanostructures were like pinecones and the surface appears nearly similar to Si ref, the topography of this surface does not present a significant change. As the etching time increased to 10 min, more pinecones were formed and we see clearly a turnover of the surface morphology compared to untreated silicon. For 20 min, Silicon pinecone like patterns disappear and SiNWs appear clearly. As the etching time increased to 30 min, SiNWs become taller but congregate together and they are strongly bonded by Van der Waals electrostatic forces [18]. Due to these forces, the SiNW arrays appear as a package forest. Each package is composed of a limited number of wires ranging from 20 to 40 wires. AFM image shows that SiNW30 sample is composed by two kinds of structures: porous and islands. The latest structures show the high density. The morphology of this etched sample was typically a forest like island of wires distributed randomly on the surface of the substrate. This structure is interesting for its use in light trapping and light diffusing. The root mean square (rms) roughness of the six sample was measured. We remark that if the etching duration increases, the surface roughness increases too ranging from 55 to 186 nm as seen in Fig. 3. Results show that the surface of SiNW30 reveals the highest roughness of about 186 nm. After 30 min of etch, AFM image shows a surface like the one etched for 5 min (SiNW5), it appears clearly that the wires are broken from the surface which is confirmed by the decrease of the surface roughness from 186 to 110 nm.

Fig. 2
figure 2

Atomic force microscopy (AFM) images in three dimensional of mc-Si surface carry out to a tapping mode for samples before and after silicon nanowires formation

Full size image
Fig. 3
figure 3

The root mean square (rms) roughness of etched samples

Full size image

3.2 Scanning electron microscopy analysis

The six samples: Si Ref, SiNW5, SiNW10, SiNW20, SiNW30 and SiNW60 were observed by SEM. Figure 4a shows that untreated multicrystalline silicon wafers appears different to crystalline silicon due to the existence of defects and grain boundaries (GB) in the mc-Si. Mc-Si consists of different grains with different orientations. Figure 4b shows that for an etching time of 5 min, a porous nanostructure was observed. As the etching time increase (Fig. 4c–e), we see clearly the formation of silicon nanowires and we remark that the length of the wires increase linearly with the etching time from 5 to 30 min. As shown by AFM, the wires collapsing together and they appear in package shape due to the Van der Waals interactions between adjacent vertically SiNWs [18]. After 30 min of etching, SEM image shows the nanowires with a short length less than the others, this may be due to the destruction of the wires after 30 min of etching which confirm the decrease of the surface roughness. We therefore attribute this behaviour to the existence of defects in multicrystalline silicon which facilitate the etching of the nanowires.

Fig. 4
figure 4

Top view images of untreated mc-Si (a) and treated one (b–f) corresponding to SiNW5, SiNW10, SiNW20, SiNW30 and SiNW60, respectively

Full size image

Figure 5a shows that with diluted nitric acid there is some Ag dendrites on the silicon substrate and we can see clearly the formation of wires between randomly distributed dendrites structure in directions nearly perpendicular to the mc-Si grains on the substrate. The existence of these Ag dendrites is advantageous to the fabrication of SiNWs [19].

As can it be seen from the side view of SiNW30, the nanowires are relatively vertical to the facets of the crystalline silicon grains on multicrystalline silicon and the wires are closely interrelated, held together in packages and oriented perpendicularly to the facets of grains on mc-Si. SEM image of the cross section substrate shows that the length of the nanowires can be determined to be about 3 µm and the diameter was 70 nm (Fig. 5b).

Fig. 5
figure 5

a Typical SEM image of SiNWs after being treated with diluted HNO3 solution for few minutes, b cross section view of the formed SiNWs on the multicrystalline silicon wafers after etching for 30 min (SiNW30)

Full size image

3.3 UV–visible measurement

When compared to bare mc-Si, all the etched samples present a remarkable reduction in the reflectivity. The physical reason of the strong reduction of reflection is apparent from the AFM micrographs and SEM images. The etched multicrystalline silicon wafers for 30 min and after removing the silver layer appeared dark black and non reflective to the naked eye, reflecting the difference in the visible light absorption which is moreover indicative of high surface absorption. This has motivated us to examine the reflectivity property of nanowires structure fabricated by chemical etching method (Fig. 6). It is observed that the reflectivity decreases with increasing the etching time. We can relate the decrease of the reflectance of the etched samples to the formation of rough ordered structure suitable for light absorption [20]. More remarkably, as the etching duration increases from 5 to 30 min, the length of silicon nanowire increases; making a decrease in the R% within the whole, especially for short wavelength range. The reflectivity is reduced significantly for all samples as compared to bare mc-Si and it attains a minimum of 2.9% in the 250–500 nm range and 5% in the spectral range 500–900 nm for the sample etched for 30 min. The Fig. 7 shows the measured mean reflectivity in the 250–1200 nm of SiNWs fabricated at different durations). However, we notice that the reflectivity increases for sample etched during 60 min and reaches more than 6% in the 300–1000 nm range. We conclude that the SiNW30, sample etched for 30 min, exhibits the lowest reflectance in the range of 300–500 nm. The thickness of this SiNW films is only few micrometers when Si wafer shows hundreds of micrometers in thickness, the absorption of the visible light is greater than 97%, which means a remarkable light trapping effect. Karyaoui et al. [21] assigned this low reflectance to numerous causes such as: the roughness and the geometry of the silicon wires, as well as the collective light scattering interactions within the nanowire structure, which trap the light and make it travel many turns over distances much longer than the array thickness. Peng et al. [22] assigned also this to surface nanostructure of the SiNW and for two other important factors: (1) extremely high surface area for the greater length of the SiNW which enhances light trapping and conducts to much lower reflectance and (2) a gradual change in the refractive index (RI) with the increase on the depth of the SiNW arrays that closely resembles a multi antireflection layer coating. The refractive index pass from air (RI ≈ 1.0) to Si nanowire and to the substrate (RI ≈ 3.5).

Fig. 6
figure 6

The reflectivity (R%) spectra measurements of SiNWs with different length (the inset spectrum is attributed to the m-Si wafer before etching)

Full size image
Fig. 7
figure 7

Mean reflectivity in the 250–1200 nm range of SiNWs fabricated at different etching time

Full size image

3.4 Raman analysis

As shown in Fig. 8, the spectra of mc-Si etched for different durations are similar to that of unetched one. We can obviously observe that each spectrum has three major peaks. This three peaks are located around 300, 520 and 930 cm−1 corresponding to phonon modes: 2TA(X), TO(Γ), and 2TO(X,L), respectively and they appear in the full spectra including the untreated multi-crystalline silicon [23].

Fig. 8
figure 8

Raman spectra of the six various samples synthesized on multicrystalline silicon wafers exihibit similar patterns

Full size image

The spectrum corresponding to untreated silicon has a dominant peak with a frequency of about 520.34 cm−1, is thinner and less intense than the other peaks. It corresponds to the first-order phonon mode, an optical active mode triply degenerated. On either side of it, we can see two less intense peaks corresponding to the second-order Raman spectra involving two phonons [24].

There is no novel peak and the Raman spectra are relatively similar except for the magnitudes of the Raman intensity, shift and broadening. More phonons appear in the Raman spectra of SiNWs compared with those of mc-Si and the intensities of these phonons increase and shift to lower frequencies.

All the Raman spectra in Fig. 9 after etching showed an enhancement in the dominant peak, these spectra reveal well resolved optical phonon peaks at frequency varied between 519 and 520 cm−1. It is well known that the length of silicon nanowires increase with increasing the duration of etching [25]. We note that for all etched samples, except SiNW60, there is an increase in the Raman scattering intensity by an order of magnitude, it attain the maximum for 30 min, after this duration there is a decrease of intensity. The sample fabricated for 30 min exhibited the largest enhancement of the Raman intensity compared to other. The peak intensity of untreated mc-Si is 12,032 arbitrary units (A.U.), while for the SiNW30 it increases to 39,160 A.U. All silicon nanowires samples provide a higher Raman intensity compared to the unetched silicon. This enhancement in the Raman intensity as compared to mc-Si may be due to the silicon nanowires witch can efficiently absorb the inelastic scattered Raman light. It may be attributed to network of SiNWs, which cause multiple reflection of excitation intensity eventually leading to maximum absorption which is confirmed by the optical reflectance. The dark black that appear after every etching of the silicon indicate also the minimization of reflectance and the scattering losses at the surface of samples [26]. As reported by Tian et al. [27] two effects that could lead to the enhancement of Raman intensity of silicon nanowire. The first one may possibly be attributed to the transmitted excitation intensity into the material which should be increased due to the decrease of area fraction of silicon after etching leading to minimization of losses due to diffuse scattering. The second is the Raman back scattering traveling toward the surface may have encountered the nanointerstice surface between the SiNWs causing an enhancement factor for the light excitation. Khorasaninejad et al. [28] report also that the enhanced Raman scattering is due to an increased electric field intensity resulting from the confinement of light within the nanowires. As mentioned earlier, Fig. 9 shows that TO peaks shifts to lower wavenumbers in SiNWs when compared to the bulk silicon (Table 1). The full width at half maximum (FWHM) values of the most shifting and the most intense peak witch correspond to SiNW30 sample, is 3.96 cm−1. The energetic shift of this peak has been attributed to the quantum confinement effect [29]. According to the phonon confinement model proposed by Richter, when the crystalline size decreases Raman active modes will not be limited to be at the center of the Brillouin zone leading to a redshift and asymmetric broadening of the first order Raman band [30]. Dariani et al. [31] report also that Raman spectra are dependant to the phonon coherence length. When compared to nanostructured silicon, ideal bulk silicon present a very long phonon coherence length. Its Raman spectrum contains a very narrow peak from q = 0 phonons. When the crystal size decreases, the phonon coherence length will also decrease and this will shift the peak to lower frequencies.

Fig. 9
figure 9

The first-order Stokes Raman of the bulk Si and that of etched samples

Full size image
Table 1 The Raman peak shifts of samples and there corresponding FWHM
Full size table

4 Conclusion

In this work we report a rapid, cost effective and simple approach to prepare low cost antireflective surface by using silver assisted chemical etching (Ag-ACE) method starting form a cheap multicrystalline silicon wafer. It was found that the surface etched for 30 min minimize reflection loss to as low as 3% in the spectral range of interest for silicon solar cells. We showed that the silicon nanowires were perfectly crystalline, oriented perpendicularly to the facets of grains on mc-Si. It is clear from results that the roughness of samples prepared with silver assistance by electroless deposition method varies from a sample to another depending on the etching times. When comparing the Raman spectra for the SiNWs and the bulk silicon, we conclude that the first-order phonon mode increase when silicon nanowires length increase. The nanowires still with excellent defect-free diamond crystalline structure after etching. These results demonstrate that a thin layer of SiNW arrays formed on mc-Si after a subsequent etching time equal nearly to 30 min is beneficial for solar light trapping and consequently for enhancing the mc-silicon solar cell efficiency.