Electronic Quality Enhancement of Multicrystalline Silicon via SiN_x and H₂ Plasma Passivation Using Plasma-Enhanced Chemical Vapor Deposition for Photovoltaic Applications

Abstract

This study explored advancements in photovoltaic technologies by enhancing the electronic quality of multicrystalline silicon (mc-Si) through silicon nitride (SiN_x) and hydrogen (H₂) plasma deposition via plasma-enhanced chemical vapor deposition (PECVD). This innovative approach replaced toxic chemical wet processes with H₂ plasma and SiN_x. The key parameters of silicon solar cells, including the effective lifetime (τ_eff), diffusion length (L_diff), and iron concentration ([Fe]), were analyzed before and after this sustainable solution. The results show significant improvements, particularly in the edge region, which initially exhibited a low τ_eff and a high iron concentration. After the treatment, the τ_eff and L_diff increased to 7 μs and 210 μm, respectively, compared to 2 μs and 70 μm for the untreated mc-Si. Additionally, the [Fe] decreased significantly after the process, dropping from 60 ppt to 10 ppt in most regions. Furthermore, the treatment led to a significant decrease in reflectivity, from 25% to 8% at a wavelength of 500 nm. These findings highlight the effectiveness of the PECVD-SiN_x and H₂ plasma treatments for improving the optoelectronic performance of mc-Si, making them promising options for high-efficiency photovoltaic devices.

1. Introduction

The solar photovoltaic (PV) market is undergoing a rapid and significant transformation [1], solidifying its position as a leading force within the global energy sector [2,3]. This dynamic growth is being driven by the urgent need to reduce greenhouse gas emissions [4], along with decreasing manufacturing costs and favorable government energy policies [5]. As a result, the solar PV capacity is increasing globally at an unprecedented rate. While crystalline silicon (c-Si) technologies continue to dominate the market [6], monocrystalline (c-Si) modules now lead due to their superior efficiency and performance [7,8]. However, multicrystalline silicon (mc-Si), once the dominant technology, still holds a residual share in cost-sensitive applications [9].

High-efficiency solar cell concepts utilize a range of techniques—including passivation layers, rear contacts, and advanced surface texturing—to reduce recombination losses and optimize power output [10,11,12,13]. Additionally, innovative cell architectures, such as heterojunctions and back-contact designs, have shown significant efficiency improvements and enhanced performance under real-world operating conditions [14,15]. However, challenges, such as defects and impurities in silicon and, more specifically, in mc-Si, remain significant barriers to improving its electronic quality [16,17,18]. These defects reduce solar cell efficiency considerably by increasing the charge carrier recombination [19,20], thereby limiting the overall energy output. Additionally, traditional manufacturing processes often rely on hazardous and toxic chemicals, posing risks to both human health and the environment [21]. To overcome these challenges, advanced photovoltaic technologies are being developed to enhance mc-Si electronic quality and eliminate the use of toxic chemicals. Innovations, such as PECVD, gettering techniques, and defect passivation [22,23,24], help to reduce impurities, leading to high efficiency. Furthermore, sustainable and non-toxic manufacturing methods are being explored to minimize the environmental impact [25]. These advancements are crucial for the future of sustainable energy, ensuring that silicon-based solar PV technology remains both efficient and environmentally friendly, leading to clean energy solutions. Advanced photovoltaic technologies aim to enhance the optical, electrical, and optoelectronic properties of multicrystalline silicon wafers through a combination of surface treatments and bulk purification [26,27]. It is important to note that reducing the surface recombination velocity of minority carriers and optical losses significantly impact solar cells’ efficiency. Moreover, the grain boundaries (GBs) can function as a potential barrier for charge carriers and thereby affect the majority carrier mobility, as reported by Maruska et al. [28]. Metallic impurities, particularly transition metals (Fe, Cu, etc.), are typically present in mc-Si with high concentrations. These impurities act as charge carrier trapping centers, leading to a dramatic degradation of the photovoltaic performance of mc-Si-based solar cells. Surface treatments and the gettering of undesirable impurities are widely used techniques to enhance the optical, electrical, and optoelectronic properties of multicrystalline silicon. A surface treatment reduces surface recombination by passivating defects and minimizing optical losses, improving the charge carrier lifetime and overall efficiency. Meanwhile, silicon nitride films effectively getter transition metals impurities from silicon [29,30]. Iron is the most significant metal extracted from the bulk to interface in this process [31], contributing to more than half of the total recombination events in as-grown multicrystalline material. Acting as a recombination center, iron reduces the photovoltaic performance of mc-Si-based solar cells [32,33]. The combination of surface passivation and impurity removal significantly enhances the charge carrier mobility, extends the diffusion length, and improves the overall optoelectronic quality of mc-Si, making it highly suitable for high-efficiency solar cell applications. For this reason, this study first investigated the gettering effect of the iron present in the bulk, along with its impact on the carrier lifetime and diffusion length of the multicrystalline silicon (mc-Si) substrates that were grooved and treated with porous silicon (PS), followed by rapid thermal processing (RTP). In the second stage, after removing the PS layer, plasma-enhanced chemical vapor deposition (PECVD) was employed to deposit silicon nitride and hydrogen plasma onto the mc-Si substrate, aiming to passivate the surface and enhance the optoelectronic properties of the mc-Si, thereby achieving high electronic quality.

2. Materials and Methods

2.1. Sample Preparation for the PECVD Process

The experimental process is described in Figure 1. The substrate used in this experimental process was p-type mc-Si with a thickness of 450 µm, dimensions of 10 × 10 cm², and a resistivity ranging from 0.5 to 2.0 Ω·cm. The sample preparation involved a combination of grain boundary (GB) etchings and the formation of porous silicon. The samples underwent the following sequential treatments: First, the GBs were grooved by immersing the samples in a mixture solution of HF (48%): HNO₃ (65%) with a volume ratio of 4:1 for 10s, followed by rinsing in deionized water and drying under N₂ flow (see Figure 2a). PS layers were formed on both sides of the wafers using the stain etching method [34], which consists of dipping samples in an aqueous HF (40%): HNO₃ (65%): H₂O solution with a 1:3:5 volume ratio for 120s (see Figure 2b). PS formation was followed by a rapid thermal treatment in an infrared furnace under an argon atmosphere for 1 h at 850 °C. The purpose of this thermal treatment was to concentrate unwanted impurities in the inactive region of PS layers on both sides of the substrates. To eliminate the trapped impurities, the porous silicon layer was removed following a two-step process. The first step was to eliminate the silicon oxide (SiO₂) layer formed after annealing using a diluted HF solution (10%), then the PS layer was removed using aqueous sodium hydroxide (NaOH) solution (1 M) to obtain the treated mc-Si, referred to as T-mc-Si.

2.2. PECVD Application

The prepared samples underwent a plasma-enhanced chemical vapor deposition treatment to deposit silicon nitride and hydrogen plasma. To examine the passivation characteristics of PECVD-deposited SiNx, hydrogen plasma pre-treatment was employed, substituting for the conventional toxic wet chemical cleaning. This in situ process, conducted within the PECVD reactor before SiNx deposition, eliminated the need for hazardous chemicals and maintained a continuous vacuum environment [35], in addition to the enhancement of silicon’s antireflective properties. It is important to note that the damage caused by hydrogen plasma treatment [36,37,38] can be reversed through annealing at temperatures above 600 °C, which exceed the typical limits of low-temperature PECVD processes, and at which temperature hydrogen could diffuse into the silicon substrate (grain and grain boundaries). For PECVD deposition of SiN_x, the reactive gases used were SiH₄ and NH₃, with N₂ as the carrier or diluting gas. The SiH₄/NH₃ gas ratio controlled SiNx film stoichiometry (stoichiometric, Si-rich, or N-rich). RF power adjusted film stress via surface bombardment, with higher power yielding compressive stress. Conversely, increased pressure increased tensile stress and reduced film density.

Table 1. Optimized parameters for the SiN_x layer deposition and hydrogen plasma treatment via PECVD.

	Pu (mbar)	P (mTorr)	T (°C)	Time (s)	RF (W)	SiH4 (sccm)	NH3 (sccm)	H2 (sccm)
H2 Plasma	2 × 10−1	700	300	50	60	-	-	100
SiNx Layer	2 × 10−1	700	300	900	5	3	70	-

summarizes the optimized parameters for the SiN layer deposition and hydrogen plasma treatment.

2.3. Characterization Tools

The cross-section and surface structure of the samples were measured with a scanning electron microscope (JEOL JSM-5400, Tokyo, Japan). An FT-IR spectrometer (Nicolet MAGNA-IR 560 ESP FT-IR, Madison, WI, USA) was used to conduct the elemental analysis of the treated mc-Si. Reflectivity spectra were obtained using an UV/Vis/NIR Spectrophotometer (Perkin-Elmer Lambda 950, Springfield, IL, USA) with an integrating sphere. Finally, using a WT-2000PVN (Semilab, Hungary) Multifunction Wafer Mapping Tool high-resolution maps of carrier lifetime, diffusion length, and iron concentration were generated. The tool’s 904 nm laser enabled approximately 30 µm deep excitation of the silicon.

3. Results and Discussion

After subjecting these samples to chemical etching in a HF/HNO3 solution, we observe in Figure 2b that only the GBs are significantly etched, which means that the GBs are the preferred sites for the acidic attack. In fact, the attack speed is mainly influenced by the impurities in and defect density of the mc-Si wafer. As is well known, GBs are the preferential sites of these impurities, which leads to a higher Si dissolution rate at the GBs compared to other locations [39]. Furthermore, we observe that for two sister GBs, the etched thicknesses differ from one another, leading to the need to think about the non-uniformity of the impurities’ distribution. This method produces a polished and rigorous surface with minimum defects at the GBs. In fact, by means of two-dimensional Light-Beam-Induced current imaging, Dimassi et al. have proven that chemical grooving in a HF/HNO3 solution enables the deep penetration of phosphorus and metallic contact and a decrease in the recombination activity in the region close to the GBs [40]. In addition, the V-shape of the GBs obtained after chemical grooving allows for at least a double reflection of the light beam that arrives at the GBs, as shown in Figure 2b. This reflection increases the number of photons absorbed by the mc-Si substrate, leading to the photo-generated carrier’s density increase.

The SiNx: H spectrum shown in Figure 3 consists of typical absorption bands for different N-H bonds and bands related to Si-N, Si-O, and Si-H bonds [41]. The spectrum shows two absorption peaks at 460 cm⁻¹ and around 850–879 cm⁻¹, which are associated with the vibration mode of Si-N. Additionally, there are two absorption peaks related to the vibration mode of the N-H bond at 1150 cm⁻¹ and 3342 cm⁻¹. Furthermore, the absorption peaks associated with the vibration modes of Si-O and Si-H are located at 1073 cm⁻¹ and around 2200–2250 cm⁻¹, respectively. The final absorption peak is located at 3320 cm⁻², corresponding to the vibration of N-Si-H. After the hydrogen plasma treatment, the broad band observed in spectrum (b) is attributed to the convolution of two vibrational modes: Si–H and N–H. Additionally, the disappearance of the N–H bonds and the emergence of Si–O bonds at 1080 cm⁻¹ indicate the onset of oxidation. The oscillations around 500 cm⁻¹ in the IR spectrum of the SiN_x layer are likely due to optical interference within the thin film, with their spacing related to the layer thickness. These are not characteristic vibrational modes of Si–N bonds. The formation of silicon oxynitride (SiO_xN_γ) is unintentional. As a tetrahedral amorphous alloy, it exhibits peaks corresponding to Si–O and Si–N bonds, indicating the presence of a SiO_xN_γ layer [42,43,44,45]. The shift in the maxima of the absorption peaks corresponding to the Si-O and Si-N bonds depends on the substitution of nitrogen atoms for oxygen atoms during the plasma treatment. It is well known that the Si-H content is crucial for the passivation quality.

Scanning electron microscopy (SEM) was used to characterize the surface morphology of the mc-Si samples following these treatments. As shown in Figure 4, the H₂ plasma treatment applied to a SiN_x layer approximately 350 nm thick, as measured by profilometry, significantly enhanced the effective surface area that can interact with incident light compared to the reference sample. This treatment plays a crucial role in improving light absorption in solar cells by inducing multiple internal reflections within mc-Si. The rough surface scatters incoming light instead of reflecting it away, promoting diffused reflection and reducing reflection losses.

Figure 5 shows the total reflectance of the mc-Si sample after the different treatments: Figure 5a for the bare mc-Si, Figure 5b for the PS-treated groove mc-Si, and Figure 5c for the PS-treated groove mc-Si with SiN_x/hydrogen plasma in the wavelength range of 300–1200 nm. It can be observed that a significant reduction in the reflectivity occurred in the sample that underwent a combination of PS-treated groove mc-Si for the entire wavelength range compared to the untreated one, where the reflectivity decreased from 25% to 14% at a wavelength of 500 nm. This reduction in the reflectivity is linked to the etching of the grain boundaries (GBs) and the changes in the grain surface. Indeed, the V-shape of the GBs obtained after etching allowed for at least double the reflection of light beams that reach the GBs. The results further demonstrate that the SiN_x/hydrogen plasma layer effectively reduced the reflectivity to 8% at a wavelength of 500 nm. In contrast, the mc-Si surfaces processed via a multi-step dry etching technique involving NF₃ and Ar exhibited a higher reflectivity of 16.5% [46], primarily due to the formation of nanometric surface pores. Alternatively, the mechanical texturization of the mc-Si wafers achieved a lower reflectivity of approximately 12% by creating grooves 100 μm deep, even in the absence of an antireflection coating. This technique produced microstructured features on the silicon surface; however, it lacked inherent surface passivation and caused damage within the grooved regions [47]. Overall, the reflectivity obtained through the SiNx/hydrogen plasma treatment is superior to the values reported in [46,47]. Moreover, the reduction in reflectivity after the SiN_x/hydrogen plasma layer’s formation is attributed to the intermediate step of the H₂ plasma, which led to the formation of a double-layer antireflective coating with SiN_x and the formation of nanostructures on the mc-Si surface (see Figure 4c).

To evaluate the electronic quality of the mc-Si substrates treated with H₂ plasma and SiN_x by the PECVD technique, 2D mappings of the τ_eff, L_diff, and [Fe] were performed before and after each treatment. The resulting maps are shown in Figure 6.

It can be observed that after the SiN_x and plasma H₂ treatment, the distribution of the τ_eff, L_diff, and [Fe] shows significant changes compared to the bare mc-Si. The main observation is the complete transformation of the edge region, which initially exhibited low τ_eff values and very high [Fe]. Consequently, after the treatment, the most significant improvements in the τ_eff, L_diff, and [Fe] were obtained in the edge region. The mappings of the τ_eff and L_diff show an increase, with the τ_eff and L_diff reaching 7 µs and 210 µm, respectively (Figure 6c), compared to the bare mc-Si, where the average values of the τ_eff and L_diff are approximately 3 µs and 105 µm (Figure 6a). Furthermore, the change in the [Fe] distribution after the treatment was also more significant, with the concentration decreasing in most areas of the wafer, from 52 ppt to a minimum value of 10 ppt. A correlation between the reduction in the [Fe] and the augmentation of the τ_eff and L_diff was observed for the totality of the mc-Si surface, signifying substantial impurity passivation mediated by the SiN_x and H₂ plasma. The implemented gettering protocol exhibited high efficacy, yielding satisfactory values for the τ_eff, L_diff, and [Fe].

Another critical phenomenon examined in this study is the impact of the T-mc-Si on the edge regions of the mc-Si wafers. As shown in Figure 6b, the carrier lifetime mapping of the mc-Si wafers (10 cm × 10 cm), measured using the µ-PCD technique, indicates that the wafers were extracted from the edge of the mc-Si ingot, where the effective minority carrier lifetime (τ_eff) exhibits a pronounced reduction, particularly along the left and lower edges. Figure 6a shows that the region with a very low τ_eff (~1 µs) extends approximately 0–15 mm from the wafer edge and gradually increases between 15 and 100 mm.

Macdonald et al. [48] demonstrated that the low τ_eff of minority carriers in the edge region of mc-Si ingots negatively impacts the electronic quality of wafers. They found no direct correlation between this low τ_eff and the concentrations of substitutional carbon, interstitial oxygen, or the dislocation density. However, iron contamination was identified as the primary cause of τ_eff degradation in the edge region of mc-Si. In this section, we analyze the correlation between the iron concentration and the τ_eff in the mc-Si wafers, considering the effect on the T-mc-Si. The iron concentration in the reference wafer reached 1 × 10¹³ cm⁻³ at the edges and gradually decreased to 3.2 × 10¹² cm⁻³ at the center. This confirms the τ_eff distribution and highlights the detrimental effect of iron (see Figure 6). The progressive increase in the interstitial iron contamination toward the edges may be attributed to iron incorporation during directional solidification. In the T-mc-Si, the [Fe] significantly decreased to 2.6 × 10¹² cm⁻³, demonstrating the effectiveness of the treatment for reducing contamination. The same phenomenon was observed for the L_diff distribution mapping of the Si-mc. For the reference sample, the L_diff values ranged from 59 µm to 105 µm, with the lowest values at the edges. In the T-mc-Si, the L_diff improved significantly, ranging from 69 µm to 118 µm, with a notable increase at the edges (see

Table 2. Measured values of effective lifetime, diffusion length, and iron concentration of mc-Si for etch treatment.

	τeff (µs)	Ldiff (µm)	[Fe] (ppt)
Bare mc-Si	1.0–3.1 ± 0.1	059–105 ± 5	060–220 ± 10
T-mc-Si	1.3–3.9 ± 0.1	069–118 ± 5	052–151 ± 10
SiNx/H2	3.8–7.0 ± 0.1	117–210 ± 5	010–050 ± 10

).

Multicrystalline silicon (mc-Si) contains numerous defects, such as dangling bonds, grain boundaries, and dislocations, which serve as recombination centers for charge carriers. These defects significantly degrade the electronic quality of mc-Si and reduce the efficiency of devices, such as solar cells. During hydrogen plasma and silicon nitride (SiNx) treatments, atomic hydrogen generated from the plasma diffuses into the silicon lattice and bonds with dangling bonds at defect sites and grain boundaries. This process, known as hydrogen passivation [49], neutralizes the electrical activity of these defects. As a result, the carrier lifetime and diffusion length are significantly enhanced, and the overall electronic quality of mc-Si is improved. On the other hand, hydrogen-rich SiN_x films can getter iron even when its concentration in the silicon bulk is below the solid solubility limit at high temperature, suggesting that segregation is the dominant mechanism [31,32,48]. Unlike solubility-based gettering, which requires impurity concentrations above the solubility limit to cause precipitation, segregation gettering creates energetically favorable sites for impurities below this threshold. The hydrogen in SiNₓ films likely aids this process by passivating dangling bonds and altering the film structure or Si/SiNₓ interface, enhancing iron trapping [33,48]. Additionally, high temperatures increase iron diffusivity, promoting its migration to the interface and reducing the iron concentration in the bulk.

4. Conclusions

In summary, an innovative approach utilizing hydrogen plasma and silicon nitride was used in this investigation via the PECVD technique, which revealed a significant improvement in the optical and optoelectronic properties of mc-Si. As a result, the optoelectronic parameters, such as the τ_eff and L_diff, showed substantial enhancement, with the τ_eff and L_diff reaching 7 µs and 210 µm, respectively, compared to the bare mc-Si, where the values of the τ_eff and L_diff were approximately 2 µs and 70 µm. Furthermore, the change in the [Fe] after gettering was also more significant, with the concentration decreasing in most areas of the wafer, reaching a minimum value of 10 ppt. The reflectivity showed a significant reduction from 25% to about 8% for the wafer that underwent a combination of treatment with H₂ plasma/SiN_x at a wavelength of 500 nm, compared to the reference sample. The experimental results suggest that the obtained sample has spectacular passivation and high electronic quality of the mc-Si surface.