Defect engineering of p‐type silicon heterojunction solar cells fabricated using commercial‐grade low‐lifetime silicon wafers

In this work, we integrate defect engineering methods of gettering and hydrogenation into silicon heterojunction solar cells fabricated using low‐lifetime commercial‐grade p‐type Czochralski‐grown monocrystalline and high‐performance multicrystalline wafers. We independently assess the impact of gettering on the removal of bulk impurities such as iron as well as the impact of hydrogenation on the passivation of grain boundaries and B‐O defects. Furthermore, we report for the first time the susceptibility of heterojunction devices to light‐ and elevated temperature–induced degradation and investigate the onset of such degradation during device fabrication. Lastly, we demonstrate solar cells with independently verified 1‐sun open‐circuit voltages of 707 and 702 mV on monocrystalline and multicrystalline silicon wafers, respectively, with a starting bulk minority‐carrier lifetime below 40 microseconds. These remarkably high open‐circuit voltages reveal the potential of inexpensive low‐lifetime p‐type silicon wafers for making devices with efficiencies without needing to shift towards n‐type substrates.


| INTRODUCTION
The silicon heterojunction (SHJ) solar cell owes its success to the excellent passivation quality provided by hydrogenated amorphous silicon (a-Si:H) films. 1 These films have enabled devices with opencircuit voltages (V OC ) over 750 mV and multiple efficiency demonstrations beyond 25%, [2][3][4] with the current world record of 26.7%. 5 In addition to the excellent surface passivation, SHJ cells have been attributed manufacturing advantages, where devices can be (and, in fact, must be) completely fabricated below 250 C. 6 Such temperature restrictions prevent both a degradation of the a-Si:H passivation layers and degradation of the bulk silicon material commonly associated with higher-temperature processing. 7,8 These lower temperatures also enable the simple fabrication of symmetrical bifacial structures and the use of thinner wafers without thermal-stress-related yield losses. 9 However, to take full advantage of the excellent efficiency capabilities of SHJ cells, highlifetime silicon wafers are required. Table 1 lists several recently fabricated SHJ devices on n-type and p-type silicon wafers as an extension to the technology summary presented by De Wolf et al. 6 Remarkably, although n-type SHJ technologies have continuously pushed the limits of efficiency beyond 25%, the performance of p-type SHJ has yet to see such breakthroughs and is still well behind the UNSW 25%-efficient, 706-mV p-type PERL solar cell fabricated over two decades ago, 22 and the more recent 26.1% (V OC of 737 mV) polycrystalline silicon on oxide (POLO) interdigitated back contact (IBC) cell by ISFH. 23 Nonetheless, significant progress has been made in 2018 with the fabrication of a 23.8%efficient device. 14 It is also interesting to note that there is little work in recent years in the area of multicrystalline silicon (mc-Si) solar cells using SHJ technologies. In contrast, optimisation of other passivation techniques such as those used in p-type passivated emitter and rear cell (PERC) solar cells and n-type tunnel oxide passivated contact (TOPCon) solar cells has enabled mc-Si devices with efficiencies reaching 22.04% (V OC of 672 mV) and 22.3% (V OC of 674 mV), respectively. 24,25 Although SHJ devices are more commonly fabricated on n-type substrates, p-type materials, in particular mc-Si, still offer a competitive cost advantage when it comes to ingot production. If these lower quality substrates can be transformed into high-efficiency devices, then there exists a chance for a low-cost and commercially competitive option. Fundamental studies are critical to understanding the potential of such materials and suitability for next-generation solar cells featuring carrierselective contacts.

| Defects and degradation in p-type silicon
A common pitfall of p-type silicon is the increased susceptibility of the material to recombination-active defects within the bulk. 26,27 Metallic impurities (including iron, copper, titanium, nickel, and cobalt) 26 29 Fe i and copper (Cu) concentrations are also commonly used as metrics to assess the efficiency of gettering processes. 30 This is due to their relatively simple detection methods and high diffusivity and therefore responsiveness to gettering processes. 31 On the other hand, studies on industrial PERC solar cells have shown that metals such as copper, nickel, and cobalt are more detrimental to device performance in n-type silicon than in p-type silicon for similar impurities concentrations. 26 Mc-Si wafers have both a larger concentration of metallic impurities and inherently T A B L E 1 Published J-V performance metrics of leading n-type and p-type silicon heterojunction solar cells from various research groups and SHJ manufacturers  Similarly, bulk passivation with hydrogen is extremely temperature dependent. Figure 1B  The migration of hydrogen is often limited by interactions with traps (particularly in mc-Si), especially at low temperatures, and therefore affected by the concentration of impurities and defects in the material, further reducing diffusivity. 85,86 For effective passivation in an SHJ solar cell, it is desirable for hydrogen to be dispersed throughout the bulk of the wafer prior to AHP.
One method of achieving this would be to use initial highertemperature processing steps, such as an equivalent conventional metallisation firing step, to introduce hydrogen into the bulk prior to SHJ fabrication. To obtain a hydrogen diffusion length of 100 μm in less than 10 seconds, a timescale compatible with high-throughput manufacturing, a temperature in excess of 500 C is required. Experimental data within the temperature range of 600 C to 800 C mea-

| P-type heterojunction solar cell fabrication
Prior to the fabrication of SHJ solar cell precursors, all wafers were again chemically cleaned. This three-step cleaning process consisted of (a) a Piranha clean for 10 minutes in a 4:1 ratio of 96% sulfuric acid (H 2 SO 4 ) and H 2 O 2 heated to 110 C followed by a 10-minute rinse in DIW, (b) a 10-minute RCA 2 clean, and (c) a 1-minute submersion in 10:1 buffered oxide etch (BOE) followed by a 10-minute rinse in DIW. Testing for B-O defect passivation and stability was carried out using a 0.02 kW/m 2 white-light LED for a period of 48 hours at room temperature (23 ± 2 C).

| Characterisation
Lifetime testing was performed using quasi-steady-state  Nonetheless, the enhancements from defect engineering are reflected in the respective increases in both the bulk minority-carrier lifetime ( Figure 3B) and the iV OC ( Figure 3C). By using both gettering and hydrogenation processes in conjunction with each other, we can achieve a sevenfold increase in τ bulk and a maximum iV OC measured above 700 mV.
The injection-dependent minority-carrier lifetime of subsequently fabricated SHJ precursors in each processing group are shown in Note. I-V measurements were performed in-house at UNSW. Note that the champion cell had sustained damage prior to independent verification such that the exact cell area and hence J SC could not be accurately determined. a Independently measured at SERIS.
T A B L E 2 Performance of best p-type Cz SHJ solar cells fabricated on substrates with various defect-engineering treatments, before and after AHP on the same cells  improvement of 30%, which is consistent with our observed enhancements albeit on p-type wafers. We also find that although the lowinjection, B-O-related SRH behaviour is no longer present after AHP and after stability testing, we continue to observe injection-dependent recombination activity that cannot be accurately explained using SRH defect parameters. This behaviour is similar to that observed in previous studies, 5,6,112 which was postulated to be due to surface-assisted minority-carrier recombination. Later analysis by Olibet et al 113 suggested a possible involvement of dangling bonds at the a-Si:H/c-Si interface which occurs more predominantly on p-type substrates than on n-type substrates. Similar low-injection recombination activity was observed by Adachi et al 4 on SHJ structures fabricated on n-type substrates; however, the reduction in low-injection lifetime is believed to be more pronounced on p-type substrates. 112 Regardless, this residual recombination provides potential for further defect passivation in future work.
A summary of the device performance of the best cell in each group before and after the AHP is presented in Table 2. We find that ΩÁcm 2 and 6.5 ΩÁcm 2 on the champion cell prior to and after AHP. We speculate that this may be due to a combination of resistive losses both in the a-Si:H films and the low temperature cured paste. We cannot, however, exclude other factors such as laser-edge-isolation which may also affect both the J SC and FF.

| P-type multicrystalline silicon
In agreement with the results on Cz wafers, gettering and hydrogenation provide similar benefits for mc-Si wafers. The PL images in to passivate the crystal defects sufficiently, or perhaps a redistribution of mobile impurities during firing. 114 The enhanced hydrogen passivation after gettering may be due to the reduction in bulk impurities and defect states thus reducing the need for significant quantities of hydrogen in order to passivate remaining recombination-active constituents. However, an equally likely explanation is that some impurities that may be effectively gettered may not be effectively hydrogen passivated.
As with the Cz wafers, performance improvements were observed on fabricated cells in response to the AHP. The performances of the best p-type mc-Si cells in each group before and after the AHP are presented in Table 3. An improvement in V OC of up to 44 mV was observed within the G + H + AHP group, with a maximum V OC of 701 mV (independently measured at SERIS to be 702 mV). Impressively, this is approximately 30 mV higher than the V OC for record efficiency devices on both n-type and p-type mc-Si. 24,115 However, we find an intriguing observation in the V OC behaviour prior to AHP that differs from the trends observed on SiN x :H passivated samples: on average, the G + H group demonstrated a V OC more than 20 mV lower than the G group. A difference in 26 mV is seen from the best cells in each group. We suspect that the observed lower V OC may have been influenced by LeTID, which will be discussed in Section 3.3.

| Hydrogen passivation of B-O defects in ptype Cz SHJs
Elimination of the B-O defect in p-type Cz silicon is essential for maintaining the performance of solar cells. In Section 3.1, we discussed the impacts of the AHP in enhancing bulk passivation within both Cz and mc-Si silicon wafers after hydrogenation. Figure 6 shows Further work will be required to determine which of these explanations may or may not be responsible for the behaviour in

| Hydrogen-related degradation in mc-Si SHJs
In Section 3.1, we suggested that hydrogenated SHJ solar cells may be susceptible to LeTID as an explanation for the lower-than-expected V OC s measured directly after fabrication. We observed this phenomenon only on hydrogenated wafers (groups H and G + H), in agreement with recent studies proposing hydrogen as a root cause for LeTID. 37,41,[118][119][120] Chan et al reported that LeTID can occur in the dark at elevated temperatures, and SHJ cells undergo such process steps. 36,41 In particular, the PECVD deposition of the a-Si:H layers and the curing of the silver paste front electrode are usually conducted at temperatures between 200 C and 250 C in the absence of illumination. From the work by Chan et al , 41 at these temperatures, the estimated time to reach maximum degradation of p-type mc-Si may be as quick as 10 minutes or as slow as 50 minutes, depending on the device structure and any thermal pretreatments.
In order to investigate possible LeTID during SHJ fabrication, mc-Si G + H sister wafers from the same batch as those in In order to verify whether the observed degradation was in fact an onset of LeTID, we carried out injection-dependent lifetime spectroscopy analysis on the samples in their most degraded state. Figure 7B shows the inverse lifetime of one sample after firing Thus, we report for the first time that SHJ solar cells may also be susceptible to the LeTID degradation phenomena commonly observed in PERC solar cells and that the fabrication process flow of SHJ devices may facilitate the defect formation process. An open question is why the same LeTID phenomenon was not seen in the gettered and hydrogenated Cz solar cells and why the V OC of that group was instead significantly higher than the other groups. Although Cz materials are also prone to LeTID, 36 this may be due to inherently lower LeTID-related precursor concentrations in Cz wafers. Unlike in Cz, the greater quantity of crystallographic defects and grain boundaries present in mc-Si may enhance hydrogen diffusivity via these alternate pathways into the bulk. 87 LeTID is also suspected to be caused by a secondary impurity (eg, metallic impurities, 123 vacancies, 37 and oxygen precipitates 28 ) that may bind to hydrogen to form a recombination active complex. This secondary impurity, the exact of which is still unknown, may be present in far lower concentrations within Cz silicon than in mc-Si. As a component of future work, LeTID in SHJ cells may be eliminated or prevented, firstly, through hydrogenation processes at lower temperatures that may be used to maintain effective hydrogen passivation without leading to subsequent degradation 124,125 and secondly, by treating LeTID prior to SHJ processing to limit formation during cell fabrication.

| CONCLUSION
In this work, we incorporated prefabrication gettering and a multistage hydrogenation process into SHJ solar cells fabricated on low-quality p-type substrates with initial τ bulk lower than 40 μs. Using this method, we demonstrated open-circuit voltages exceeding 700 mV for both mc-Si and Cz p-type SHJ solar cells. For mc-Si, the V OC is approximately 30 mV higher than that for current record efficiency devices.
Our study also identified, for the first time, the presence of LeTID in hydrogenated SHJ solar cells. The mitigation or suppression of this defect in such structures is an area that requires further investigation, particularly when using low-quality substrates that require bulk hydrogenation. The demonstration of such high V OC values in this work challenges the assertion that expensive, high-quality n-type wafers are needed for fabrication of SHJ solar cells. Although the efficiencies of our initial cells are relatively low, the inherent efficiency limitation due to the bulk material causing low terminal voltages has been overcome.
If the issues regarding J SC and FF can be overcome, low-lifetime p-type silicon wafers could be an alternative material for the fabrication of SHJ devices and perhaps, other passivated contact solar cells.