3. Results and discussion
3.1. Etching model
Our experiments were based on the commonly known principles of metal-assisted c-Si etching that were developed for the noble metal catalyst assistant etching. In our study we modified this model by adding specific details about the Ni film properties.
The mechanism was thoroughly described and proposed in [11]. The model explained the c-Si etching in an aqueous solution of HF and H2O2 with the nanoparticles of noble metals, namely Au, Pt and Pd. It was proposed that, during metal assisted etching, the metal–silicon interface can be described as a microscopic electrochemical cell with silicon as the anode and metal as the cathode in it. The electrochemical processes can be described by a set of the following equations:
a) on metal surface:
b) on silicon surface:
According to [12], reactions (3) and (4) occur in the metal–silicon interface due to the hole injection to silicon under metal. The process is controlled by the H2O2 decomposition (Eqs. 1, 2) stimulated by the metal catalyst. The reaction mechanism can be summarized by a single equation, as proposed by Charter et al. [13]:
From reactions (1)–(5) it can be obtained that the etching rate is dependent on the intensity of the hole injection to silicon. The processes are controlled by the catalyst because the hole injection rate depends on the metal catalyst.
It was demonstrated in [14] that H2O2 decomposes on the nickel catalyst and, consequently, reactions (1)–(5) can be used for the description of the nickel-assisted etching. As a result, the H2O2 decomposition on the Ni catalyst surface produces an injection of the holes into silicon. After that, the silicon atoms are removed by the HF etching solution.
The reaction mechanism also includes the mass transport from and to the reaction zone during the process, as described in [15]. More details for the metal-assisted chemical etching mechanism of silicon can be found in the review [12]. The mass transport is very important in the metal-assisted etching and, consequently, in the NACE in our experiments.
It can be concluded from the commonly accepted model of the mechanism that there are a few key factors that control the NACE processes such as the mass transport from the reaction zone, the thickness of the Ni catalyst and the proportions of the components HF and H2O2 in the etching solution. Therefore we studied an influence of the listed factors on the black-Si properties in our study with the aim to optimize the set of the parameters required to produce the minimum optical reflectance over the shortest etching time.
3.2. Investigation of the technological key factors for the black-Si formation
3.2.1. Investigation of the mass transport from the reaction zone
The influence of the mass transport on the characteristics of etched surfaces was studied in the special samples with an individual coverage of the surfaces with thin Ni films. The coverage was intentionally changed by photolithography. For this an individual set of strips was formed in the Ni film on each sample. The width of the Ni strips (Ws) was constant for individual samples and was equal to 0.375, 0.75, 1.5 and 3 mm. Since the Ni strips restricted a direct access to the etching surface for the solution, mass transfer was expected in the interface between the Ni film and Si surface through the channels of an individual length equal to the width of the strips. The dependence of the surface parameters was measured in the samples with an individual length of the mass transport channels. The SEM pictures presented in Fig. 1 show that it is a clear difference between the etched area of silicon where the Ni film was deposited and where it was not (shown by arrows in Fig. 1(a)).
The analysis of the obtained surface (as presented in Fig. 1(b)) showed the structured layer with a thickness in the range 50–100 nm. For the quantitative description we used atomic force microscopy by scanning the samples directly and scanning with a bigger area of the sample. The AFM images of the same as in Fig. 1 etched silicon surface are illustrated in Fig. 2.
From the AFM measurements it is clear that over a larger scale there are big random variations of the surface height up to 300 nm; however, on the same sampling area like in the SEM pictures for lower and higher planes on the sample (measured profiles 1 and 2 in Fig. 2(a)) variations in the structure height are the same and have a value range from 50 to 100 nm (Fig.Fig. 2(b)). As bigger height variations are likely to occur randomly we assume that these deviations are from the instabilities of the etchant by itself, so for the further investigation of the Ni film influence we use areas, which are on the same plane, and we analyse smaller height deviations. The variations in the heights of the sample surface produced by Ni-assistant etching were quantitatively characterized by the root-mean-square roughness (RMS) that represented the average of the measured height deviations taken within the evaluation length and measured from the mean line. The RMS parameters were measured in the centre of the Ni ribbon and on the sides of the bare silicon surfaces after the etching. The results of these tests are presented in Fig. 3.
It must be noted here that the access of etching solution to the surface of substrate can be dependent on the location of surface point and the coverage by catalyst film. Considering this, it was expected that RMS must be dependent on the position under the Ni strip because the mass transport seemed being most effective in the vicinity of the strip edge (positions 2 and 3), whereas the transport effect of the etching solution can hardly be effective over a long distance in the interface channel between the Ni strip and the Si surface. Therefore, the changes produced by Ni-assistant etching were expected being much more visible near the strip edges compared to that far from the edge. Despite this expectation, the results revealed only a very weak influence of the interface channel length on the RMS changes in the Si surface, as can be seen from results in Fig. 2. The mass transport through Ni films explained the access of etching solution to the Si surface under strips during the NACE in our experiments, and is comparable with the result presented in [11] for different metals during metal-assisted etching. Aiming to understand the effect, we have investigated the structure of Ni films.
The AFM imaging of Ni films revealed a fine grain structure of our strips. The dimensions of grains were measured for the as-deposited Ni catalyst films by the analysis of AFM topography images. Typical results obtained for grain diameters are illustrated in Fig. 4.
It can be seen from Fig. 4 that the average grain diameter (D) increased with an increase in the Ni film thickness (hNi). In addition, it was obtained from these measurements that the increase in diameter was followed by a decrease in density of grains in the Ni strips in this study. These results are presented in Fig. 5.
Based on our results, we think that the mass transport channels were formed between the grains in Ni films. These channels were sufficiently large for the flow of etching solution through the film to the reaction area. Consequently, the H2O2 decomposition at Si surface can occur and, hence, the etching mechanism described by equations (2) and (3) can be engaged in Si surface etching. The etching rate for Si surface under metal was dependent on the tunnel-like channels between the grains in Ni film. The dimensions of these channels were dependent on the density and diameter of the grains in Ni film and, therefore, the surface roughness RMS was changed by the film thickness.
3.2.2. Ni layer thickness effect on black-Si surface properties
The influence of Ni layer thickness on the black-Si surface properties was investigated in a series of samples with individual thicknesses of Ni films equal to 10, 20, 40 and 80 nm. These samples were etched in the same etching solution for the same fixed time. The results of RMS analysis are presented for these samples in Fig. 6.
It can be seen in Fig. 6 that the RMS increased with an increase in Ni film thickness and the maximal RMS was obtained for the 80 nm thick films.
3.2.3. Black-Si reflectance dependence on the proportions of etching solutions and the etching time
In our experiments, we described the proportions between the components of etching solution by the parameter ρ that was described by the following equation:
In Eq. (6), M(HF) and M(H2O2) are the molar concentrations of pure acid HF (48%, Sigma-Aldrich) and peroxide H2O2 (35%, Eurochemicals). The relationship between the ρ parameter and the amounts of components in the etching solution are listed in Table 1.
| V(H2O), ml | V(HF), ml | V(H2O2), ml | ρ, % |
|---|---|---|---|
| 10 | 8 | 1 | 93 |
| 10 | 3 | 1 | 80 |
| 10 | 1 | 1 | 72 |
| 10 | 8 | 8 | 59 |
| 10 | 4 | 8 | 47 |
| 10 | 1 | 4 | 21 |
| 13 | 1 | 8 | 14 |
| 17 | 1 | 12 | 9 |
For each ρ parameter in Table 1, the black-Si was produced by etching in the solution for 1, 3, 5, 10 and 20 min. The optical reflectance spectra were measured for these samples. The results of etching tests are illustrated by the spectra in Fig. 7. The spectra were obtained for the series of samples etched in the solution with ρ = 93 % for diverse durations from 1 to 20 min.
The influence of parameter ρ on the optical reflectance was characterized by two calculated parameters, namely the average reflectance Ravg and the minimum reflectance Rmin. The average reflectance was calculated by averaging the optical reflectance over the entire spectra from 400 to 1100 nm. Rmin was obtained by detecting the lowest optical reflectance in the measured spectra. The experimental dependences of these parameters on the parameter ρ and the etching duration are presented in Fig. 8.
From the results in Fig. 8 it was obtained that there was a non-monotonous dependence of RMS on ρ and etching duration. As it followed from our experiments, the lowest obtained values of Ravg and Rmin depended on a certain set of ρ and etching time. The optimal conditions for etching must be identified for an individual pair of the parameters, namely ρ and the etching duration, aiming to obtain the minimal magnitude for Ravg and Rmin. Our experiments suggested that the lowest value of Ravg was ~2% and the lowest Rmin was ~0.4% for the black-Si that was formed by the etching under the conditions with ρ = 9% and duration equal to 10 min.
The SEM pictures of the lowest reflectance value samples are presented in Fig. 9.
From the SEM pictures it is seen that structures with a width of about 100 nm and a depth of around 500 nm have the lowest reflectance value.
3.2.4. Photovoltaic devices using black-Si formed by NACE
To check if the proposed NACE technology can be applied to form photovoltaic devices, we prepared a set of photovoltaic devices the surfaces of which were textured by diverse methods, namely perfectly polished, with the pyramid layer made by KOH/IPA etching and the black-Si layer made by using NACE.
The parameters of the sample PV cells were measured using an AM 1.5 solar simulator. The I–V characteristics of the devices are illustrated in Fig. 10, and the calculated parameters of devices are presented in Table 2.
The results in Table 2 demonstrated that the black-Si made by the NACE technique was improved compared to the c-Si based photovoltaic device. However, the parameters were comparable to that of the samples with the KOH/IPA textured surfaces. It must be noted here that the formation of contacts on the textured surfaces is still an issue, and it typically resulted in worsening of the cell parameters. Therefore, the results in Fig. 10 and Table 2 are highly promising for further development because a simple and accurate control and cheapness of the process can be used to reduce the formation time of black-Si in the process lines with the industrial vacuum technologies when a combination of diverse methods is quite complicated. In addition, it must be noted here that our results do not represent the parameters of the completed PV cells because we are focusing on the development of the features of the NACE method. The cell parameters were significantly reduced by the surface leakage conductance. Therefore, the surface protecting coating and the contact formation methods must be developed before the PV cell parameters can be reliably compared to those of commercially available solar cells.
| Surface preparation | Jsc, mA/cm2 | FF, % | η, % |
|---|---|---|---|
| Polished surface | 17.93 | 68.8±2 | 6.67±0.3 |
| NACE | 22.45 | 70.2±1.9 | 7.25±0.2 |
| KOH etching | 23.75 | 73±1.6 | 9.02±0.3 |