Current status and challenges for hole-selective poly-silicon based passivating contacts

Silicon photovoltaics (Si-PV) has become a key technology in the global shift toward clean and sustainable energy. The rapid adoption of Si-PV has been driven by continuous innovations across the entire value chain. Currently, the passivated emitter and rear cell (PERC) technology [Fig. 1(b)] is the workhorse of the Si-PV market because of its ability to achieve higher cost-to-performance ratio than its predecessor, the aluminum back surface field (Al-BSF) cell technology [Fig. 1(a)].¹ Continuous improvements in wafer quality, stability, and production tools have led to consistent improvements in PERC solar cell efficiency in mass production. These developments have now brought PERC solar cell efficiency of 24.5% in the laboratory and average 23.5% in mass production bringing it closer to its practical limit of achievable efficiency, around 24.5% for this architecture.^2–5 To overcome the inherent efficiency limitation mainly due to recombination at the metal contacts of the PERC technology, passivating contact solar cell technologies have emerged as the next generation of solar cells for mass production. The doped poly-Si films with a thin oxide interlayer the commonly known as tunnel oxide passivated contact (TOPCon) cell technology^6,7 [Fig. 1(c)], have emerged as industrially viable option due to their compatibility with the existing PERC technology.¹ Additionally, doped hydrogenated amorphous (a-Si) silicon heterojunction (SHJ) cell technology^8,9 [Fig. 1(d)], has also remained in the race for next generation of industrial solar cells. However, SHJ cell technology which currently holds the single junction crystalline silicon record at 26.8%¹⁰ is not covered in this review. This review is focused only on the passivating contacts featuring doped poly-Si films which is expected to gain market share from about 10% in 2022 up to 60% within the next 10 years.¹ 

Doped poly-Si contacts are formed by inserting an ultrathin silicon oxide (SiO_x) as an interlayer between the crystalline Si wafer and a highly doped poly-Si film. The interfacial oxide layer provides excellent chemical passivation of the dangling bonds and prevents recombination that would otherwise occur at the metal–silicon interface.¹² Additionally, these interlayers enable electrical transport through tunnelling^10–20 and/or pinholes.^24,25 Doped poly-Si films can serve as either electron-selective or hole-selective passivating contacts by doping them n⁺ and p⁺, respectively. Over the years, extensive research has demonstrated that doped poly-Si contacts can achieve both excellent passivation quality and low contact resistivity (ρ_c) using various interfacial oxide layer and poly-Si deposition methods.^15,26–37 Consequently, many high-efficiency (>25%) silicon solar cells based on doped poly-Si contacts have been demonstrated by research groups and PV manufacturers.^6,38–43 Notably, n⁺ poly-Si contacts have consistently shown remarkable performance, with typical single-sided surface recombination current densities (J₀) of less than 5 fA/cm² achieved on planar surfaces, regardless of the formation method of the interfacial oxides and the poly-Si layers.^{26–29,31,34,44} This underscores the adaptability and robustness of n⁺ poly-Si contact fabrication.

The current solar cell architecture based on a rear-side n⁺ poly-Si contact (TOPCon) is also approaching its practical limit (∼26%) in mass production. A detailed loss analysis of state-of-the-art doped poly-Si solar cells suggested that the recombination loss on the front contacts is one of the main source of efficiency loss.^42,45 Therefore, full potential of poly-Si based passivating contact solar cells could only be realized by incorporating p⁺ poly-Si contacts to replace the traditional boron diffused contacts on the front surface. Other report has also concluded that high efficiency solar cells with doped poly-Si passivating contacts will require both n⁺ and p⁺ poly-Si contacts.⁴⁶ That work projected that fully passivating contact solar cells [in this case, the polycrystalline silicon on oxide (POLO) interdigitated back contact architecture] could achieve efficiencies greater than 27.5%, surpassing the limitation of current n-type TOPCon solar cells (26%).⁴⁶ 

While studies of both p⁺ and n⁺ poly-Si contacts started around the same time,⁴⁷ there has been a notable discrepancy in the progress of their performance. A comparison between the present status of p⁺ poly-Si and n⁺ poly-Si contacts is shown in Fig. 2(a), based on reported efficiencies of both-sides-contacted solar cells. Numerous research groups and PV manufacturers have consistently demonstrated high-efficiency (>25%) solar cells employing n⁺ poly-Si contacts.^6,38–43 The efficiency of solar cells utilizing n⁺ poly-Si contacts has shown rapid improvement, with PV manufactures, such as Jinko, Longi, Trina, Jolywood, and DAS solar unveiling impressive efficiency improvements within a short timeframe.^38–41,48 However, progress in the performance of solar cells with p⁺ poly-Si contacts has been slower. The record efficiency (23%) of solar cells using only p⁺ poly-Si contacts was achieved in 2018.¹⁵ Therefore, there is a significant gap between the performance of p⁺ poly-Si contacts and n⁺ poly-Si contacts. The primary challenge faced by p⁺ poly-Si contacts lies in their comparatively lower surface passivation, typically characterized by J₀ values ranging from $5 to 15$ fA/cm² on planar surfaces,^{21,22,49–52} and their limited compatibility with current screen-printed fire-through metallization processes.

Furthermore, the discrepancy between the performance of n⁺ and p⁺ poly-Si contacts is also attributed to the inadequate optimization of p⁺ poly-Si contacts, as shown in Fig. 2(b). Assuming that each publication related to p⁺ poly-Si contacts contribute equally to the understanding and performance advancement, gradual progress hinges on factors such as the number of researchers, their effectiveness, time allocation, and available resources. These factors are often correlated with publication volumes.⁵³ However, breakthrough single papers may have a more significant impact on the development of the field. Regardless, a comparison of publication counts (peer-reviewed journal articles) related to n⁺ poly-Si and p⁺ poly-Si contacts since 2014 when the interest in the doped poly-Si contacts was revived, proves instructive. The publication statistics are obtained from publicly available databases, including Scopus, ScienceDirect, and Google Scholar. As illustrated in Fig. 2(b), there are notable disparities on publication volumes between research articles on n⁺ poly-Si and p⁺ poly-Si contacts, underscoring the more pronounced emphasis and endeavors directed toward the optimization of n⁺ poly-Si contacts.

In order to improve the performance of p⁺ poly-Si contacts, a comprehensive understanding of their current status and challenges is essential. This review offers an update on the progress of p⁺ poly-Si contacts, building upon previous works.^23,72 Subsequently, it will examine the current challenges faced by p⁺ poly-Si contacts, explore potential solutions, and provide an outlook of options for the realization of p⁺ poly-Si contacts in various high-efficiency solar cell architectures.

The methods commonly employed for the formation of p⁺ poly-Si contacts are similar to those used for n⁺ poly-Si contacts and have been extensively reviewed elsewhere.⁷³⁻⁷⁶ Here, we will only provide a brief overview of the process flow, as outlined in Fig. 3. Similar to n⁺ poly-Si contacts, the formation of p⁺ poly-Si contacts also can be divided into five primary steps: (i) growth of the ultrathin interfacial oxide layers, (ii) deposition of intrinsic polycrystalline or amorphous silicon, (iii) crystallization and dopant introduction (ex situ) or dopant activation (in situ) at an elevated temperature, (iv) hydrogenation, and (v) metallization.

The interfacial oxides can be formed through various methods, including immersion in hot nitric acid,^20,77 UV-assisted ozone oxidation,^36,78 and thermal oxide growth in a furnace at temperatures ranging from 550 °C $to$ 700 °C.^13,17,27,36 In situ oxide growth can also be achieved using thermal oxidation followed by chemical vapor deposition (CVD), such as low-pressure CVD (LPCVD) or plasma-enhanced CVD (PECVD).^49,79–81 The type of SiO_x layers has a significant impact on the performance of doped poly-Si contacts, particularly in p⁺ poly-Si contacts, as shown in Fig. 4. This is primarily due to boron induced damaged of oxide interlayers due to high levels of B segregation into the SiO_x layers.^82,83 Generally, the typical J₀ values for p⁺ poly-Si contacts are in the range of $5−15$ fA/cm² on planar wafers, whereas n⁺ poly-Si contacts can achieve J₀ values ranging from $1 to 5$ fA/cm² (Refs. 12, 26, 27, 64, and 84–90) for the same type of oxide layers.

The deposition of intrinsic poly-Si or a-Si layers is primarily carried out using various versions of CVD. LPCVD is the preferred method for n⁺ poly-Si in mass production [70% market share in 2023 (Ref. 1)] due to its better uniformity, consistent performance, no blistering, and ease of integration with PERC technology.^13,20,27,36 However, LPCVD is a non-directional double-sided deposition process, which can increase processing costs due to the need to remove the poly-Si layer from one side of the cell, for example, the current TOPCon cells. LPCVD deposited p⁺ poly-Si contacts have demonstrated some of the best J₀ values (⁠ $1−6.5$ fA/cm²)^36,44,95,96 as shown in Fig. 4. The one of the lowest J₀ values (1 fA/cm²) and iV_oc of 732 mV for p⁺ poly-Si contacts were achieved by thermal oxide as an interlayer with in situ boron-doped poly-Si layer deposited by LPCVD and hydrogenated by SiN_x layers.⁹⁵ On the other hand, PECVD deposited p⁺ poly-Si contacts have also shown promising passivation performance (J₀ =  $3.5–12$ fA/cm²) comparable to LPCVD.^17,22,49 The single-sided deposition capability of PECVD makes it an attractive option for industry. Nevertheless, PECVD-deposited passivating contacts are susceptible to blistering issues.^71,78,97,98 Beyond PECVD and LPCVD, some studies have explored the use of other methods for forming p⁺ poly-Si contacts, such as atmospheric pressure CVD (J₀ = 6 fA/cm²)⁹⁹ and hot wire CVD (HWCVD).¹⁰⁰ However, the adoption of these methods is limited to few research laboratories. Another group of methods used for p⁺ poly-Si contact formation is physical vapor deposition (PVD), particularly sputtering which has achieved a slightly higher J₀ =  $20–30$ fA/cm², which is likely due to sputter damage of poly-Si layers.^15,101 However, this single-sided method has increasingly gained attention from the research institutes and PV industries due to the additional benefits of avoiding the use of hazardous precursor gases (such as silane and diborane), eliminating blistering due to a low hydrogen content, and low deposition temperatures.

The doping of poly-Si contacts can be achieved through two broad methods: ex situ^13,17,20,102 and in situ.^20,22,49 Most research on p⁺ poly-Si contacts have focused on ex situ doping methods. Ex situ doping methods include vapor diffusion (using BCl₃ or BBr₃),^{15,101,103,104} ion implantation,^14,105 and liquid-based doping.^14,105–107 Ion implantation and liquid-based doping have also been used to explore alternative doping elements, such as Ga, for the formation of p⁺ poly-Si contacts. In in situ methods, both the dopant elements and silicon are deposited simultaneously, and the dopant activation is performed in an inert environment at high temperatures. PECVD is considered more suitable for in situ deposition of doped poly-Si contacts but is susceptible to hydrogen induced blistering, resulting in a severe degradation of the poly-Si contacts after dopant activation.^47,69,78 Nonetheless, Morisset et al.⁷⁸ demonstrated that optimizing deposition temperature and gas ratios can yield blister-free p⁺ poly-Si contacts by PECVD, achieving one of the best iV_oc values of 734 mV for p⁺ poly-Si contacts on planar n-type samples (⁠ $t=$ 275 μm and resistivity = 2  $−$ 3 Ω·cm).

Following the formation of doped poly-Si contacts, an additional crucial step involves hydrogenation. This step aims to improve the passivation performance of both n⁺ and p⁺ poly-Si contacts by passivating electronically active dangling bonds at the SiO_x/c-Si interface. Various hydrogenation methods have been explored for doped poly-Si contacts, including the deposition of hydrogen-rich dielectrics such as AlO_x and SiN_x films, as well as annealing in a hydrogen-rich atmosphere (forming gas anneal or remote hydrogen plasma).^{21,44,49,50,106,108,109} Atomic hydrogenation (H-rich dielectric layers or H₂ plasma) has proven to be more effective compared to molecular hydrogenation (FGA).^71,110,111 While hydrogenation is expected to improve the passivation performance of doped poly-Si contacts, recent investigations have demonstrated that excessive hydrogen diffusion could also detrimentally impact their passivation effectiveness.^112,113 Currently, the industrial TOPCon processes do not require an additional hydrogenation step and is achieved during contact firing. The dielectric capping (SiN_x) releases hydrogen into poly-Si contacts. A study by Kang et al.¹¹⁴ revealed that both n⁺ and p⁺ poly-Si passivating contacts exhibit increased recombination with firing temperature, though the impact on p⁺ poly-Si contacts was less severe. The excessive accumulation of hydrogen near the thin interfacial oxide layer was found to significantly degrade the firing stability of n⁺ poly-Si contacts, whereas p⁺ poly-Si contacts exhibited relatively higher thermal stability over a broader range of hydrogen concentrations. The improved firing stability of p⁺ poly-Si contacts can be attributed to their comparatively lower hydrogen diffusivity, leading to slower hydrogen migration^115–118 or to the accumulation of boron at the surface between interfacial SiO_x and Si substrate,^98,119 which could hinder hydrogen diffusion and effusion. However, Mack et al.⁹⁵ have reported an opposite trend as J₀ values decreased [J_{0(before firing)} = 13 fA/cm², J_0(780C) = 5 fA/cm², and J_0(900C) = 1 fA/cm²] with increased firing temperatures up to 900 °C for in situ boron doped p⁺ poly-Si contacts with SiN_x as a capping layer.

The formation of metal contacts is achieved through methods such as screen printing,^{87,89,90,95,120–126} electroplating,¹⁵ or evaporated metallization.^15,101,104 The Si-PV industry primarily uses screen-printed fire-through metallization.¹ However, contact formation on doped poly-Si contacts presents greater challenges compared to conventional back surface fields. To prevent metal spiking through the oxide interlayer during firing, n⁺ poly-Si contacts with a thickness greater than 100 nm have been necessary thus far. This has enabled low contact resistivities ρ_c ≤ 1 mΩ·cm² (Refs. 89, 122, and 125) and low J_0,metal of $30–50$ fA/cm² (Refs. 87, 89, and 122). However, when applying the same screen-printed and fire-through metallization method to p⁺ poly-Si contacts, inferior results have been obtained. Typically, p⁺ poly-Si contacts have exhibited higher ρ_c values of $2–10$ mΩ·cm² and higher J_0,metal values of $160–800$ fA/cm² (Refs. 95, 120, and 125), even with thicker poly-Si contacts of up to 200 nm. The limited compatibility of p⁺ poly-Si contacts with the fire-through metallization process could be partially addressed through further optimization of screen-printed metal paste. Nonetheless, promising outcomes have been observed in the metallization of p⁺ poly-Si contacts with evaporated metallization and with low-temperature, non-fire-through screen-print pastes.^90,123,124 However, these methods are unlikely to be compatible with the mass production metallization of doped poly-Si solar cells.

Another benefit of poly-Si contact is its strong impurity gettering effect.^{81,103,127–129} The overall gettering effectiveness of the doped poly-Si contact is determined by both the heavily doped poly-Si layers and interfacial oxide layers.¹²⁷ Recently, Yang et al.¹⁰³ provided a comprehensive overview of the gettering effects observed in poly-Si contacts formed using diverse methods, including vapor diffusion, ion-implantation, and in situ doping. The study shed light on the distinct mechanisms underlying impurity gettering in both n⁺ and p⁺ poly-Si contacts. By using iron (Fe) as a tracer impurity, the research demonstrated that the segregation coefficients remained within one order of magnitude for in n⁺ poly-Si contacts. However, a much stronger variability in gettering effects was observed in p⁺ poly-Si contacts. The study highlighted the strong gettering effects exhibited in vapor-diffused and in situ (LPCVD) p⁺ poly-Si contacts, surpassing the effects of n⁺ poly-Si contacts. In contrast, p⁺ poly-Si contacts formed by boron ion-implantation showed relatively weak gettering effects.

As reviewed above, the performance of poly-Si contacts demonstrates a significant dependence on their fabrication processes. However, p⁺ poly-Si contacts are consistently inferior to their n⁺ counterparts. In the subsequent section, we delve into the fundamental limitations associated with dopant incorporation in p⁺ poly-Si contacts, followed by a brief overview of strategies aimed at mitigating these limitations. Additionally, we outline other limitations that primarily stem from inadequate optimization efforts.

The effective operation of passivating contacts relies on the controlled diffusion of dopants from poly-Si layers into the c-Si bulk. Achieving the desired performance of doped poly-Si contacts (J₀ and ρ_c values) hinges on the careful management of peak doping within the poly-Si layers and the subsequent in-diffusion into the c-Si substrate. The observed performance gap between p⁺ and n⁺ poly-Si contacts can be partially attributed to the distinctive properties of boron (B) and phosphorus (P) in silicon, as shown in Table I. The segregation coefficient of B (k =  $0.1−0.3$⁠) is significantly smaller than of P (k = 10).¹³⁰ As a result, boron accumulates in the oxide interlayer,^82,131,132 which can lead to the degradation of oxide layer and an increase in the density of interface states. Furthermore, the implied solid solubility limit of each dopant in the oxide can be determined from the solid solubility limit in silicon and the segregation coefficient. This is approximately two orders of magnitude higher for B (⁠ $0.3−1× 10 21 cm − 3$⁠) than for P (⁠ $2.5× 10 19 cm − 3$⁠). This implies that the B concentration in the oxide layer is much higher than that of the P concentration. Note the reported segregation coefficients of dopant elements shown in Table I are representative of stoichiometric oxide (SiO₂) layers. However, in realistic conditions, the oxide interlayer might exist in a sub-stoichiometric state, which could further increase the solubility limits. Furthermore, the smaller atomic radius of B (0.087 nm) compared to P (0.098 nm) leads to a greater strain induced by B atoms, leading to the formation of more defects in the oxide interlayer.

To address the damage caused by boron accumulation within the oxide interlayer, Ga has been explored as an alternative doping element.^{14,107,119,133} Ga offers advantages such as a much higher segregation coefficient (⁠ $k$⁠) compared to B,^134,135 as shown in Table I. This helps to reduce the accumulation of Ga within the oxide interlayer. Furthermore, the implied solubility of Ga in the oxide is several orders of magnitude lower than that of B, as shown in Table I. Additionally, Ga atoms are significantly larger than B atoms, which minimizes dopant-induced damage to interfacial oxides. Therefore, Ga doping has the potential to improve the passivation performance of p⁺ poly-Si contacts.

Young et al.¹⁴ conducted a study on the depth profiles of hydrogenated p⁺ poly-Si contacts [PECVD in situ p⁺ (B) poly-Si and ion-implanted p⁺ (Ga) poly-Si] through secondary ion mass spectrometry (SIMS), as shown in Fig. 5. The profile illustrates that in p⁺ (B) poly-Si contacts, the boron and hydrogen concentrations peaked within the SiO₂ layer, as shown in Fig. 5(a). However, in the case of p⁺ (Ga) poly-Si contacts, the Ga distribution does not exhibit a pronounced peak within the SiO₂ layers, as shown in Fig. 5(b). Notably, the p⁺ (Ga) poly-Si contacts achieved doping of approximately $2× 10 19$ cm⁻³, consistent with the solid solubility limit of Ga in Si, as shown in Table I. The hydrogen peak is observed only in the p⁺ (B) poly-Si contacts, which could be attributed to the hydrogen passivation of oxide/interface defects. Note that the H levels in the c-Si wafer are representative of the background level of the SIMS instrument. In this study, the PECVD in situ p⁺ (B) poly-Si and ion implanted p⁺ (Ga) poly-Si contacts achieved optimal J₀ values of 18 and 1 fA/cm², respectively, and reported an excellent iV_oc with a record of 737 mV for p⁺ (Ga) poly-Si samples.

However, utilizing Ga also presents inherent challenges for the fabrication of p⁺ poly-Si contacts. The diffusivity of Ga in c-Si is slightly smaller than B in c-Si, which can pose difficulties in achieving uniformly doped poly-Si contacts,¹³⁴ as shown in Fig. 5(b). Another significant issue arises from the solid solubility of Ga in c-Si, which is nearly one of magnitude lower than of boron at the same diffusion temperature.^134,136 This limitation reduces the achievable dopant concentration in poly-Si contacts, posing a particular challenge in achieving low ρ_c values in doped poly-Si contacts. Thus, the use of non-equilibrium doping methods becomes necessary to activate excess Ga above its solubility limit and achieve a low contact resistivity.^138–140 

Recently, Chen et al. employed a nanosecond-scale pulsed laser melting technique to overcome this limitation of conventional thermal processing, enabling doping above the equilibrium solid solubility of Ga in poly-Si layers to form p⁺ poly-Si contacts. The study achieved Ga and B concentrations of $3× 10 20$ and $1× 10 21$ cm⁻³, respectively, significantly surpassing the solid solubility in c-Si, as shown in Table I. Impressively, the approach led to p⁺ (Ga) poly-Si contacts with a low ρ_c value of 35.5 mΩ·cm², in contrast to the 2000 mΩ·cm² obtained through conventional furnace annealing. Remarkably, the hyper-doped p⁺ poly-Si (Ga) contacts maintained a low J₀ value of 4.1 fA/cm² (iV_oc = 735 mV) indicating the absence of laser-induced damage and establishing this method as promising for p⁺ poly-Si contacts. It is worth noting that the study utilized relatively a thick poly-Si layer (⁠ $∼$ 250 nm), further exploration would be necessary to determine whether the process can maintain performance with thinner poly-Si layers.

Nevertheless, using both B and Ga as dopants presents an opportunity to improve p⁺ poly-Si contacts performance, particularly on the rear side of solar cells. One strategy could involve employing p⁺ poly-Si (B) contacts in the metal-contacted areas, while employing p⁺ (Ga) poly-Si contacts to passivate the non-contacted areas. This dual-doping strategy could help overcome the limitation of each dopant and optimizes the performance of p⁺ poly-Si contacts. Truong et al.¹⁰⁷ demonstrated co-doping of Ga with B using a spin-on doping method to form p⁺ poly-Si passivating contacts. However, they found even a small fraction of B co-doping caused a large reduction in surface passivation. Therefore, an optimization of B and Ga concentration is crucial.

Two different models have been established to describe the transport of charge carriers through the oxide interlayers in poly-Si contacts: direct tunneling across the oxide interlayer and drift-diffusion through pinholes in the oxide. It is well-understood that charge carrier predominantly tunnel through ultrathin oxide layers when they are less than 2 nm thick.¹⁴¹ On the other hand, charge carrier transport through thicker oxide layers (>2 nm) occurs primarily through pinholes formed after a high-temperature annealing (T > 1000 °C).¹⁴² The formation of pinholes upon thermal annealing has been confirmed by several works using various methods such as transmission electron microscopy^142–144 and electron-beam-induced current (EBIC).¹⁴⁴ It is crucial to control pinhole density and size to maintain the high passivation of passivating contacts based on pinhole oxides, commonly referred to as polysilicon on POLO structures.^24,142 In principle, thicker pinhole oxides are expected to be more resilient to boron-induced defects, achieving lower J₀ for p⁺ (B) poly-Si contacts. However, published data do not seem to support this expectation, as shown in Fig. 6. The typical J₀ values for pinhole oxides 3–8 fA/cm² (Refs. 44 and 90) is within the range of best J₀ values for p⁺ (B) poly-Si contacts based on tunneling oxides. It is worth noting that there is limited optimization work conducted on pinhole-based p⁺ poly-Si contacts compared to their tunneling-based counterparts, which should be explored further to improve performance.

The use of SiO_x films as tunneling layers in hole-selective contacts brings inherent drawbacks. The valence band offset of SiO_x layer, which is approximately 4.5 eV, along with the large tunneling effective mass of holes (0.58 m_h) within SiO_x collectively contribute to limited hole tunneling. Consequently, this may result in suboptimal hole-selective contacts compared to electron-selective contacts with conduction band offset of 3.2 eV and smaller tunneling effective mass of electron (0.42 m_e), as shown in Table II. Note that m_e and m_h are the free electron mass and free hole mass, respectively. In response to these challenges, alternative ultrathin tunnel interlayers such as AlO_x and SiN_x films have been explored. These interlayers are interesting due to their exceptional uniformity and thickness control when deposited by atomic layer deposition (ALD).^102,149,150 Furthermore, these interlayers exhibit more favorable valence band offsets with silicon compared to SiO_x, as shown in Fig. 7. It is worth noting the uncertainty regarding the persistence and homogeneity of fixed charges in the ultrathin tunneling dielectric layers after the heavy doping and high temperature processes. Consequently, the influence of the fixed charges is not taken into account when illustrating the band energy diagrams.¹⁵¹ 

However, both SiN_x and AlO_x interlayers can have a high concentration of intrinsic charge, which can impact their performance as hole-selective contacts, as shown in Table II. In the case of AlO_x, the presence of negative charge can create a beneficial accumulation regime in a hole-selective contact. This can improve hole accumulation/transport and facilitate efficient carrier extraction, which is desirable for achieving low ρ_c values. On the other hand, the positive charge in SiN_x interlayers may lead to the generation of a depletion region at the interface. This depletion region can adversely affect the contact properties for hole-selective layers, as it can hinder hole transport and increase ρ_c values. Recent studies have shown promising indications of improvement when incorporating SiN_x and AlO_x interlayers into p⁺ poly-Si contacts,^{102,149,150,157–159} as shown in Fig. 8. However, further research is needed to explore the advantages and drawbacks of alternative interlayers such as SiN_x and AlO_x, particularly when employing ultrathin layers with expectedly lower intrinsic charge compared to bulk films.

Moreover, some of studies have explored the possibility of incorporating nitrogen into SiO_x interlayers, either during N₂O plasma oxidation or through post-nitridation to form silicon oxynitride (SiO_xN_y) layers^74,82 to improve the performance of p⁺ poly-Si contacts. These investigations observed less in-diffused boron concentration with SiO_xN_y interlayers. The reduction of boron in-diffused concentration was expected due to a low B segregation coefficient in SiO_xN_y (∼20) compared with pure SiO₂ (stoichiometric) interlayers.¹⁶⁰ Some of the studies also suggest that the reduction of in-diffusion is attributed to the formation of B⁺-N bonds within the interlayers.^160–162 

In recent years, there has been a growing emphasis on controlling the in-diffusion of dopants as a means to improve the performance of doped poly-Si contacts.^96,163,164 One avenue explored involves employing a multilayer system that incorporates additional oxide layers [Fig. 9(b)],^96,109,163 or intrinsic a-Si or poly-Si layers [Fig. 9(c)],^109,140 within the structure of doped poly-Si contacts. The purpose of these approaches is to mitigate the detrimental effects of boron segregation-induced damage on the interfacial oxides.

Stodolny et al.⁹⁶ conducted a study comparing the performance of p⁺ poly-Si contacts using a conventional poly-Si/SiO_x/c-Si [Fig. 9(a)] configuration with a multilayer system that introduced an extra oxide layer positioned at 20 nm from the interface (poly-Si/SiO_x/poly-Si/SiO_x/c-Si) [Fig. 9(b)]. With the introduction of this additional oxide layer, they observed a decrease in the concentration of in-diffused boron from 1  $×$ 10²⁰ cm⁻³ to 6  $×$ 10¹⁹ cm⁻³ near the interface, and a reduction in the in-diffused tail depth from 80 to 40 nm. These modifications yielded improvements in J₀ values from 21 to 9 fA/cm² on textured surfaces and from 5 to 3 fA/cm² on planar surfaces (after hydrogenation). This improvement was primarily attributed to the reduced boron segregation into the interfacial oxide and reduced Auger recombination below the interfacial oxide. Similarly, Polzin et al.¹⁰⁹ demonstrated that the insertion of $10–15$ nm intrinsic a-Si layers [Fig. 9(c)] resulted in a more uniform boron distribution within p⁺ poly-Si contacts, and reduced boron in-diffusion. As a result, the passivation performance was significantly improved from J₀ = 23 fA/cm² to J₀ = 11 fA/cm² on planar surfaces.

However, it is important to acknowledge the implementation of a multilayer strategy for forming doped poly-Si contacts, which may be difficult in the industry due to the added complexity and the processing steps. Their impact on the contact resistivity would also need to be carefully monitored.

The performance of ex situ p⁺ poly-Si (B) contacts has also seen improvement incorporating a high-temperature pre-annealing step (≥1000 °C) prior to thermal boron diffusion, as shown in Fig. 3. This pre-annealing step serves a dual purpose in enhancing the passivation of doped poly-Si. First, it is believed to impact passivation by promoting stoichiometry of the interfacial oxides.¹⁰⁴ It is widely recognized that the passivation quality of doped poly-Si contacts strongly depends on the stoichiometry of the interfacial oxides.^{15,20,104,136,137} Sub-stoichiometric SiO_x layers are more susceptible to disruptions, during high-temperature dopant diffusion processes, particularly during the diffusion of boron.^30,165 Stucklberger et al.¹⁰⁴ conducted a comprehensive study investigating the impact of different pre-annealing temperatures (ranging from 850 to 1050 °C) on the performance of LPCVD deposited p⁺ poly-Si contacts. Their study demonstrated that a pre-annealing step at 1050 °C during the formation of p⁺ poly-Si contacts led to improvements in iV_oc of $5–$ 10 and 24 mV on planar and textured (both hydrogenated) samples, respectively.

Second, it is also believed that pre-annealing improves the passivation of doped poly-Si contacts by intentionally generating pinholes in the oxide layer. Yang et al.¹⁶⁶ showed that ion-implanted p⁺ poly-Si passivated samples with pinholes (pre-annealed) showed a gain of 10 mV in iV_oc compared to p⁺ poly-Si passivated samples prepared without intentionally pre-forming pinholes (without pre-annealing) on textured samples. However, there is also an optimal pinhole density that could have a minimal impact on the in-diffused doping concentration but a positive effect on the passivation quality.^82,167 A higher pinhole density could increase in-diffusion of dopants and Auger recombination, leading to lower surface passivation. Nevertheless, the sensitivity of the passivation quality to pinhole density for p⁺ poly-Si contacts is much lower than in the case of n⁺ poly-Si contacts.¹⁶⁶ This is attributed to the reduction of oxide damage by boron atoms with alternative transport path provided by pinholes.

Furthermore, a good performing doped poly-Si layers could possibly require a crystal size as large as possible. It is known that the during the ex situ doping process poly-Si is expected to undergo significant crystal size growth enhancement when doped with phosphorus.^168,169 Consequently, pre-annealing is not crucial for n⁺ poly-Si layers to pre-crystallize and increase crystal size. However, crystal size growth enhancement effect is negligible during B doping, particularly when concentration is within the range of 1 × 10¹⁹ $−$ 1 × 10²¹ cm⁻³ (Refs. 169–171). Therefore, pre-annealing is also believed to be beneficial for p⁺ poly-Si contacts due to enlargement of crystal size prior to B doping.

The current state-of-the-art TOPCon cells are only textured on the front side. Texturing on the rear-side is desirable for bifacility and improved light trapping as well as wafers become thinner. However, the passivation performance of doped poly-Si contacts is clearly influenced by the surface morphology, as shown in Fig. 10.

The quality of interfacial oxide layers emerges as a pivotal factor in achieving excellent chemical passivation through doped poly-Si contacts. Textured surfaces introduce complexities as the ultrathin SiO_x layers experience disruptions along the peaks and edges of the pyramid features. Additionally, non-uniformity in SiO_x thickness arises from different oxide growth rates on different crystallographic planes, such as <100> and <111>,¹⁷² and this non-uniformity is further accentuated across the peaks and valleys of the textured surface.^173,174 It is worth highlighting that a SiO_x/c-Si interface on <111> exhibits a higher density of interface defect states compared to the SiO_x/c-Si interface on <100>, with values approximately at 10¹⁰ cm⁻² eV⁻¹ for <100> and 3 × 10¹¹ cm⁻² eV⁻¹ for <111> surfaces.^175,176 Additionally, the sharp edges of the pyramid structures exhibit elevated stress concentration levels,¹⁷⁷ which are known to increase recombination activity.^178,179 These characteristics inherent to textured surfaces collectively contribute to the reduction in their passivation performance with doped poly-Si contacts. These challenges become even more pronounced with the diffusion of boron during the formation of p⁺ poly-Si contacts, as shown in Fig. 10.

Larionova et al.³⁶ investigated the recombination behavior of p⁺ poly-Si contacts on silicon with different crystal orientations, elucidating the poor performance on the textured surfaces. The study reported J₀ values to be 90, 70, and 10 fA/cm² on <111> planar, <100> textured, and <100> planar surfaces, respectively.

Several researchers are actively exploring strategies to address the challenges of doped poly-Si contact passivation performance on both planar and textured surfaces. These strategies encompass optimization of deposition parameters, surface treatments, passivation layer design, and interface engineering technique.^{44,96,162–164,180–182} Hydrogenation stands out as a prevalent approach to improve the performance of doped poly-Si contacts on both types of surfaces. Stodolny et al.⁹⁶ demonstrated advanced hydrogenation schemes, involving novel dielectric stacked layers to improve the passivation of p⁺ poly-Si contacts on textured surfaces, yielding J₀ of 44, 21, 15, and 11 fA/cm² values on textured surfaces before hydrogenation and after hydrogenation with SiN_x, AlO_x/SiN_x, and AlO_x/SiN_x/AlO_x layers, respectively.

Furthermore, Mack et al.¹⁸⁰ demonstrated that the utilization of slightly thicker SiO_x layers (while still maintaining tunneling transport characteristics) compared to planar surfaces could yield improved J₀ values on textured surfaces for p⁺ poly-Si contacts. Lozac'h et al.^181,182 proposed the application of atomic layer deposition (ALD) to create ultrathin SiO_x layers instead of the conventional thermal or chemical SiO_x, affording enhanced control over SiO_x thicknesses across different crystal orientations and ensuring excellent conformal coverage on textured surfaces.

Moreover, other studies have shown that rounding of sharp edges, peaks, and valleys on standard pyramid structures can improve passivation quality without significant loss in optical benefits.^{51,65,183,184} Several works have demonstrated that alkaline or acidic polishing selectively polishes the vertices of random pyramid textured surfaces by forming a pyramidical base, resulting to a decreased proportion of Si <111> crystal orientation surfaces.^65,184,185

At present, the predominant technique employed in the Si-PV industry to create doped poly-Si contacts uses the LPCVD method coupled with ex situ doping. However, the international technology roadmap for PV (ITRPV) predicts a growing preference (50% market share from 2025 onwards) for in situ doped poly-Si contacts in the upcoming years, due to not requiring a single-side etch to remove the poly-Si layers from the other side.¹ However, one challenge related to this approach is the amorphization or reduced crystallinity of the poly-Si layers, which can occur during in situ B doping.¹⁸⁶ The introduction of B atoms can induce strain and disrupt the crystal structure, ultimately leading to amorphization. This does not occur for n⁺ poly-Si contacts as the larger size of P atoms does not introduce sufficient strain to disrupt the crystal structure.

Typically, reducing the concentration of boron doping helps to mitigate amorphization and enhance carrier mobility in the poly-Si layer during in situ B doping. Nevertheless, achieving rapid deposition, uniform coverage over a large area, and the desired dopant concentration of in situ B-doped poly-Si contacts simultaneously is a challenge.

To date, the metallization of p⁺ poly-Si contacts using the conventional screen-printed and fire-through metallization method has not been extensively studied. The majority of research has focused on n⁺ poly-Si contacts using Ag pastes, which have been meticulously optimized for mass production. However, screen-printed metallization of p⁺ poly-Si contacts still significantly lag n+ poly-Si contacts after metallization, especially in terms of ρ_c values, as shown in Fig. 11.

Typically, the formation of metal contacts with p⁺ Si (boron-diffused surfaces) using traditional Ag pastes has yielded a relatively high ρ_c values (>50 mΩ·cm).¹⁸⁷ Studies have shown that the incorporation of Al into the Ag pastes can reduce ρ_c values.^188,189 Nevertheless, the utilization of Ag–Al pastes leads to the growth of substantial metal spikes that penetrate the Si surface to a depth of up to 1 μm. In the co-firing process, the metal pastes must locally permeate the SiN_x capping layers, facilitated by glass frits, to establish a low-Ohmic contact. However, the penetration through the interfacial oxide layers and beyond into the c-Si bulk by the aggressive glass frit and metal spikes must be avoided in order to preserve the outstanding passivation of the doped poly-Si/SiO_x contacts. Recent advances have demonstrated the viability of p⁺ poly-Si contact properties through the application of fire-through Ag pastes⁹⁵ and Ag–Al pastes.^120,125,190

Importantly, for similar doped poly-Si thicknesses a wide range of ρ_c and J_0,metal values are achieved across different studies, as shown in Fig. 11. One of the potential reasons for this discrepancy could be attributed to the usage of different metal pastes (different suppliers and chemistry), which appears to heavily influence the performance of doped-poly-Si contacts.

A comprehensive study conducted by Padhamnath et al.¹²¹ extensively examined the impact of surface morphology (planar vs textured surface) and poly-Si thickness on the contact formation of p⁺ poly-Si, using fire-though Ag-Al pastes. They showed that ρ_c values were poorer on planar surfaces compared to textured surfaces. Furthermore, this difference could not be solely attributed to the increased surface area of textured surfaces. However, the presence of pyramid features was discovered to create disruptions within the glass layer (fired contacts). These disruptions played a crucial role in facilitating the formation of metal crystallites, which are vital for establishing effective contacts. Additionally, the study demonstrated that the ρ_c values increased to 10.5 from 3 mΩ cm² when the p⁺ poly-Si layer thickness increased from 22 to 235 nm. This phenomenon was primarily ascribed to the penetration of oxide interlayers by the metal crystallites. These observations indicate that the challenge in achieving low ρ_c values of screen-printed metal pastes with p⁺ poly-Si contacts lie in the inability to form small, dense, and shallow metal crystallites.

While several paste suppliers have already undertaken the optimization of Ag pastes for n⁺ poly-Si contacts, further dedicated effort is required to tailor pastes that are specifically suitable for p⁺ poly-Si contacts. These pastes should be designed to restrict the formation of metal crystallites within a few tens of nm in the p⁺ poly-Si contacts.

Furthermore, in the context of the Si-PV industry's trajectory toward the multi-terawatt scale, the utilization of Ag screen-printed contacts on both sides of solar cells will depend on the available supply of silver.^192,193 In this context, the electroplating of Ni/Cu/Ag contacts provides a viable pathway to address the challenges associated with screen-printed and fire-through metallization of doped poly-Si solar cells.¹⁹⁴ ITRPV projected that the Cu plating technology could reduce the silver consumption of TOPCon electrodes from 20 mg/W of the screen-printed method to 0.6 mg/W.

It is understood that the most straightforward pathways to improve the current cell technology is to reduce recombination losses due to metal contacts. High-quality p⁺ poly-Si passivating contacts allows this reduction by replacing direct metal-silicon contacts under the fingers. It enables many solar cell architectures with high-efficiency potential, some of which are illustrated in Fig. 12. Direct integration of p⁺ poly-Si contacts into the current TOPCon structure involves replacing the heavily doped boron-diffused p⁺⁺ Si beneath the front fingers with p⁺ poly-Si contacts [Fig. 12(a)] which will help to improve the V_oc. This configuration is believed to increase efficiency potential of the TOPCon cell to 26.5%. Moreover, p⁺ poly-Si contacts can also be incorporated under the front fingers in p-type wafers for the rear-junction (RJ) TOPCon solar cells [Fig. 12(b)]. This technology will be highly sensitive on the quality of p-type wafers and has efficiency potential to achieve 27% due to added advantage of fill factor improvement. Interdigitated back contact (IBC) solar cells have demonstrated great potential by incorporating both doped poly-Si contacts on the rear side of the cells⁹⁰ [Fig. 12(c)]. The biggest challenges for this cell structure compared to a both-sided-contacted cell structure is the added complexity, with doping and contacts of both polarities on rear-side. Additionally, the utilization of p⁺ poly-Si contacts as a top junction of bottom Si cells has been demonstrated in novel high-efficiency perovskite/Si tandem solar cells.^195,−197 This structure enables direct contact of electron transport layers, such as titanium oxide (TiO_x) in n-i-p perovskite top cells, eliminating the need for additional recombination layers. Optoelectronic modeling of this design has shown tandem cells efficiency potential exceeding 32%.¹⁹⁶ Clearly, the incorporation of high-quality p⁺ poly-Si contacts offers numerous potential advantages for upgrading current Si-PV technologies. However, in most cases, achieving this improvement requires the localization of the doped poly-Si contacts. In this context, it is essential to note that research into the localization of doped poly-Si contacts is still in its nascent stage. Several strategies for poly-Si localization have been proposed, such as photolithography,¹⁹⁸ reactive ion etching,¹⁹⁹ laser processing,²⁰⁰ shadow mask deposition,^201,202 and chemical etch back.^203,204 Shadow mask deposition can be used in combination with a directional deposition method, such as sputtering. Alternatively, the common patterning processes of chemical etch back, and laser processing maybe used when a masking layer is applied on the poly-Si layers, capable of withstanding subsequent chemical etching or laser ablation.

The advancement of passivating contact solar cell technologies, particularly with doped poly-Si contacts, marks a significant step in the evolution of the PV industry. While n⁺ poly-Si contacts have achieved a rapid pace of progress, the comparative lack of progress in p⁺ poly-Si contacts performance requires focused research and development efforts. The fundamental limitations of p⁺ poly contacts, including segregation of boron in the interfacial oxide and the resulting boron-induced damage have hindered its efficacy compared to n⁺ poly-Si contacts. To address these challenges, alternative dopants like Ga have been explored to reduce interface damage, while alternative interfacial layers such as SiO_xN_y have shown promise in improving passivation quality. The adoption of multilayer configurations and pre-annealing techniques also offers potential solutions to optimize p⁺ poly-Si contact performance. Meanwhile, the metallization of p⁺ poly-Si contacts, particularly using screen-printed and fire-through methods, requires tailored pastes that can create shallow, dense, and small metal crystallites. Therefore, to unlock the full potential of doped poly-Si contacts based passivating contact solar cells, addressing p⁺ poly-Si contacts challenges is essential. Collaborative efforts between academia and industry offer promise in overcoming these limitations and advancing efficient and commercially viable fully passivating contact solar cells, thereby improving solar cell technology.

This work has been supported by the Australian Renewable Energy Agency (ARENA) through the Australian Centre for Advanced Photovoltaics (ACAP) and Project Nos. 2022/TRAC004 and 2022/TRAC005.

The authors have no conflicts to disclose.

Rabin Basnet: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Di Yan: Data curation (equal); Methodology (equal); Writing – review & editing (equal). Di Kang: Data curation (equal); Investigation (equal); Writing – review & editing (equal). Mohamed M. Shehata: Investigation (equal); Visualization (equal); Writing – review & editing (equal). Sieu Pheng Phang: Investigation (equal); Writing – review & editing (equal). Thien Truong: Investigation (equal); Writing – review & editing (equal). James Bullock: Methodology (equal); Writing – review & editing (equal). Heping Shen: Project administration (equal); Writing – review & editing (equal). Daniel Macdonald: Conceptualization (equal); Funding acquisition (equal); Methodology (equal); Resources (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding authors upon reasonable request.