You are on page 1of 11

PERSPECTIVE

www.solar-rrl.com

Silicon Heterojunction Solar Cells and p-type Crystalline


Silicon Wafers: A Historical Perspective
Bruno Vicari Stefani,* Matthew Wright, Anastasia Soeriyadi, Daniel Chen,
Moonyong Kim, Brendan Wright, Dmitriy Andronikov, Ilya Nyapshaev,
Sergey Abolmasov, Gregory Wilson, and Brett Hallam

c-Si and it can be easily doped (either p-


The first reports of both boron–oxygen (BO)-related light-induced degradation or n-type)[1] allowing the fabrication of elec-
(BO-LID) and amorphous/crystalline silicon heterojunction (SHJ) solar cell fab- tronic heterojunctions.[2]
rication date back to the early 1970s. However, the complete development of the The first reports of a-Si:H/c-Si hetero-
junctions date back to the early 1970s,
“modern” SHJ structure took place well before BO defect stabilization processes
when junctions were prepared by deposit-
were developed. Due to the susceptibility of p-type Czochralski (Cz)-grown silicon ing a-Si:H layers on both sides of a p-type
to BO-LID, such wafers were deemed unsuitable for SHJ solar cells. In addition to c-Si wafer[3] and were found to passivate
stability issues, lower charge carrier lifetimes due to contamination and challenges c-Si surfaces exceptionally well.[4] At that
with surface passivation posed barriers to the adoption of p-type wafers in SHJ time, it was only recently discovered that
applications. Herein, these three key challenges are discussed in detail. Kinetic homojunction solar cells fabricated with
p-type c-Si wafers were susceptible to a
modeling and experimental results reveal the severe impact of BO-LID in p-type
novel light-induced degradation (LID)
SHJ solar cells and provide possible explanations as to why earlier attempts using mechanism.[5] In their report, Fischer
p-type wafers might have failed. The role of gettering and advanced hydrogenation and Pschunder observed a 3–5%rel reduc-
in stabilizing BO defects in SHJ solar cells is demonstrated experimentally. Finally, tion in the conversion efficiency of these
a summary of the effective surface recombination velocities reported in the lit- cells during a 12 hour light-soaking test,
erature for hydrogenated intrinsic amorphous silicon passivation of p- and n-type which was found to be caused by a reduc-
tion in charge carrier lifetime.[5]
crystalline silicon wafers is presented. Based on these findings, the potential of
During the following decade, the first
p-type wafers to enable a next-generation of high-efficiency solar cells featuring a-Si:H/c-Si SHJ solar cell was fabricated
carrier-selective contacts is discussed. by Sanyo.[6] This cell was formed by
depositing a thin layer of B-doped a-Si:H
(a-Si:H(p)) on the front side of an n-type
c-Si wafer via plasma-enhanced chemical
A silicon heterojunction (SHJ) solar cell is formed by a crystalline vapor deposition (PECVD). A transparent conductive oxide
silicon (c-Si) wafer sandwiched between two wide bandgap (TCO) layer was used to form the front contact and aluminum
layers, which serve as carrier-selective contacts. For c-Si SHJ (Al) was deposited directly onto the c-Si to form the back contact.
solar cells, hydrogenated amorphous silicon (a-Si:H) films are This relatively simple structure (illustrated in Figure 1) yielded a
particularly interesting materials to form these carrier-selective conversion efficiency of 12.3% and a modest FF of 74%. In the
contacts. This is because the bandgap of a-Si:H is larger than quest to improve FF, an intrinsic a-Si:H (a-Si:H(i)) layer was

B. Vicari Stefani, G. Wilson M. Wright


CSIRO Energy Department of Materials
Newcastle Energy Centre University of Oxford
10 Murray Dwyer Circuit, Mayfield West, NSW 2304, Australia Oxford OX1 3PH, UK
E-mail: bruno.vicaristefani@csiro.au
A. Soeriyadi, M. Kim, B. Wright, B. Hallam
The ORCID identification number(s) for the author(s) of this article School of Photovoltaics and Renewable Energy Engineering
can be found under https://doi.org/10.1002/solr.202200449. UNSW
Sydney, NSW 2052, Australia
© 2022 Commonwealth Scientific and Industrial Research Organisation.
Solar RRL published by Wiley-VCH GmbH. This is an open access article D. Chen
under the terms of the Creative Commons Attribution-NonCommercial- Sundrive Solar Pty Ltd
NoDerivs License, which permits use and distribution in any medium, Kirrawee, NSW 2232, Australia
provided the original work is properly cited, the use is non-commercial
and no modifications or adaptations are made. D. Andronikov, I. Nyapshaev, S. Abolmasov
R&D Center of Thin Film Technologies in Energetics Hevel Solar
DOI: 10.1002/solr.202200449 St. Petersburg 194064, Russia

Sol. RRL 2022, 2200449 2200449 (1 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

Figure 1. Evolution of the silicon heterojunction (SHJ) solar cell structure over the years.

deposited underneath the front emitter.[7] While successful solar cell technology based on BO-free n-type wafers. In the fol-
in improving the FF, the most exciting gains were observed lowing year (2000), the latest chapter in the history of the classical
in VOC and conversion efficiency of 14.5% was recorded. SHJ solar cell development took place. An a-Si:H(i) layer was also
Meanwhile, little was known about the defect responsible for included between the c-Si wafer and the a-Si:H(n) at the back.
the LID phenomenon affecting p-type silicon wafers. Several This allowed a significant improvement in the rear-side passiv-
theories were proposed to explain it, which attributed the ation of the cell, while still maintaining the BSF properties.
observed degradation to complexes of: lattice defects with Finally, a TCO layer was used to improve the lateral conduction
silver,[8] vacancies with gold,[9] and iron with boron.[10] of carriers and to serve as an anti-reflection coating (ARC) on
The early 1990s marked another major step in the develop- both sides of the solar cell. This was followed by metallization
ment of SHJ solar cells. Textured c-Si wafers were used and for contact formation also on both sides of the solar cell. It is
an additional phosphorus-doped (P-doped) a-Si:H (a-Si:H(n)) not clear which materials were used for both TCO and
layer was formed underneath the back contact to provide a back metallization as this structure (known as Heterojunction with
surface field (BSF), significantly increasing the SHJ solar cell Intrinsic Thin-layer, HIT) was patented by Sanyo. Nevertheless,
conversion efficiency to 18.1%.[7] In parallel, the understanding these improvements resulted in a conversion efficiency of 20.1%
of the LID phenomenon in p-type silicon was also advancing. and a remarkably high VOC > 700 mV at that time.[20]
However, it was not until the mid–late 1990s that the involve- This VOC value was at the same level as the record VOC for a
ment of pairs of boron (introduced as the dopant atoms) and oxy- single-junction silicon solar cell, which was 706 mV (UNSW pas-
gen (introduced in large concentrations during crystal growth via sivated emitter with rear locally diffused cell).[21] In their report,
the Czochralski (Cz) process) in the degradation mechanism was Taguchi et al. noted that their novel SHJ structure was not
discovered.[11] In the mid–late 1990s, only a few key characteris- affected by BO defects, since it was based on n-type Cz-Si wafers.
tics of the defect were known, including the reversible nature Later in 2000, Sakata et al. (Sanyo) reported a record VOC of
during annealing at temperatures above  200 °C and the ability 719 mV for a large-area (100 cm2) SHJ solar cell, which corre-
to trigger degradation both via illumination and current injection sponded to a conversion efficiency of 20.7%.[22]
in the dark.[11,12] From a stability point of view, the use of such wafers for solar
It is important to highlight that at that point in time, p-type cell fabrication was certainly advantageous. Switching to other
float-zone silicon (FZ-Si) wafers and Cz-Si wafers doped with silicon wafer types (like n-type) was the only practical solution
gallium (p-type) or phosphorus (n-type) were already known to avoid the negative impact of BO defects on solar cell perfor-
not to be affected by this defect, which will hereafter be referred mance, as pathways for BO defect stabilization were still far from
to as the boron–oxygen (BO) defect. On a side note, it was later being developed. Interestingly, to the authors’ knowledge, there
(2009) discovered that gallium- or phosphorus-doped Cz-Si are no reports of experimentation involving other wafer types
wafers can be susceptible to BO defects if they are co-doped with (such as p-type Cz-Si) for SHJ applications dating back to
large concentrations of boron.[13–15] 1990–2000. Either other wafer types were not considered and
During the late 1990s, electronic properties of the defect were tested, or the results from such attempts were judged not rele-
first reported and the relationship between the concentrations of vant to be reported. Still, in 2000 homojunction solar cells based
boron and interstitial oxygen and the BO defect were further on p-type B-doped FZ-Si and magnetically confined Cz-Si wafers
explored.[16,17] In 1999, Glunz et al. reported conversion effi- (which were also not susceptible to BO defects) were approaching
ciency losses of up to 1.7%abs on p-type Cz-Si homojunction solar record conversion efficiencies of 25% with VOC values also above
cells with a starting conversion efficiency of 21.0%.[18] In their 700 mV.[23]
report, solar cells processed in parallel with p-type gallium-doped In the early 2000s, the understanding of the metastable BO
(Ga-doped) Cz-Si wafers exhibited a record stable conversion defect was greatly improved. Work from Bothe et al. indicated
efficiency of 22.5%. Yet, a patent application regarding gallium that the defect was composed of one substitutional boron atom
doping for silicon wafers was filed in that same year, creating a and two interstitial oxygen atoms.[24] The kinetics of BO-related
barrier to the use of gallium as the dopant agent for p-silicon c-Si LID were being explored[25,26] and the first wafer-level mitigation
wafer manufacturing.[19] strategies were developed.[27–30] These defect engineering strate-
While key properties of the BO defect were being unraveled, gies were based on thermal processing, including phosphorus
Sanyo was progressing with the development of their novel SHJ diffusion, thermal oxidation, and rapid thermal processing.

Sol. RRL 2022, 2200449 2200449 (2 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

This was particularly interesting because these types of processes reduces when the stabilization process temperature increases.[50]
were already naturally incorporated in homojunction solar cell Wilking et al. demonstrated an increase in the residual active BO
fabrication sequences (e.g., during emitter formation, surface defect concentration for illuminated annealing temperatures
passivation, and firing). Yet, this purely thermal route was not above 250 °C[51] (for an illumination intensity equivalent to
sufficient to completely suppress the formation of BO defects 2.7 suns). Interestingly, SHJ solar cell fabrication commonly
and was only able to reduce the total defect concentration by a utilizes a temperature ceiling of 250 °C to avoid potential loss
factor of 3.[29] It is important to highlight that these benefits of hydrogen from the amorphous silicon films.[2] In our recent
are not naturally observed in the standard SHJ fabrication work, we verified that higher illuminated annealing tempera-
sequence due to the low-temperature regime used.[31] The tures accelerate the conversion efficiency gains in SHJ solar
low-temperature processing of SHJ solar cells is commonly per- cells.[52] Hence, realizing illuminated annealing of SHJ solar cells
ceived as a benefit, as high-temperature processes have the at the maximum tolerable temperature facilitates shorter process-
potential to impact silicon’s bulk lifetime negatively.[32,33] ing times (important for high-throughput processing in indus-
In 2006, the complete stabilization of BO defects in a finished trial environments). As such, a similar temperature regime of
homojunction solar cell was first demonstrated by Herguth 200–300 °C can be used for BO-LID stabilization and illumi-
et al.[34] In their study, cells that were pre-degraded (i.e., BO nated annealing of SHJ solar cells. However, each process’s
defects were formed under illumination) were annealed under illumination intensity and duration might vary.
illumination until the VOC was completely recovered. This pro- Here, it is important to highlight the most important point of
cess is often referred to as either regeneration, hydrogen passiv- this perspective article. The first instance of BO defect stabiliza-
ation, permanent deactivation, or stabilization. The latter will be tion in literature dates to 2006. By that time, the development of
used in this article as it reflects the key feature of the transition, the SHJ solar cell structure was already well defined. In fact,
which is stability under illumination and temperature.[35] While industrial SHJ solar cells commercialized today are very similar
the processing conditions used to stabilize BO defects were not to the device designed by Sanyo in the 1990s. Despite the junc-
disclosed by Herguth et al. in their 2006 article, the impact of tion being shifted to the rear of the device to reduce restrictions
temperature and carrier injection are now well understood.[35,36] on the front TCO optoelectrical properties[53] and the use of alter-
Furthermore, it is now well understood that the presence of native carrier-selective contacts (micro- and nano-crystalline
hydrogen in the c-Si bulk is required for the effective stabilization silicon[54–56]) and TCO materials[57,58] by some researchers and
of BO defects.[37–39] Despite the first instance of BO defect sta- manufacturers, the foundation of the SHJ structure is essentially
bilization dating back to 2006, it was not until the following the same. This includes the absorber material, which has always
decade that the development of accelerated stabilization pro- been an n-type Cz-Si wafer. It is, therefore, possible to speculate
cesses suitable for commercial production became available. that the (relatively) poor understanding of BO-related LID back
The development of fast-stabilization processes for BO defects then was the major barrier preventing to use of p-type c-Si wafers
was based on excess carrier generation via high-intensity in the development of the SHJ technology. The lack of under-
illumination at elevated temperatures. This allows the control standing about BO defect stabilization during the development
of hydrogen charge states to maximize hydrogen diffusivity of SHJ technology may be the main reason why the development
and passivate BO defects.[40] These approaches allowed the process was restricted to n-type c-Si wafers, and if BO defect
complete stabilization of BO-LID in <10 s.[41,42] stabilization processes were readily available at the time, the tech-
Interestingly, Wright et al. recently reported defect-engineering nology could have been developed based on p-type c-Si wafers.
processes based on high-intensity illumination at elevated temper- In their 2012 review article (to which the reader is strongly
atures to impact industrial n-type SHJ solar cells’ conversion encouraged to refer), De Wolf et al. outlined the three main rea-
efficiency positively.[43] The conversion efficiency gains during sons why n-type Cz-Si wafers were the material of choice for SHJ
illuminated annealing of SHJ solar cells are thought to be related solar cell fabrication.[2] In the following sections, these reasons
to an improvement in both surface passivation and charge trans- will be directly addressed in the context of the significant
port under illumination, which results in increased VOC and FF advances in understanding the impacts and solutions of defects
values.[43–46] A pilot study with n-type SHJ solar cells sourced from in p-type wafers.
six manufacturers found illuminated annealing to increase the
conversion efficiency of cells sourced from all manufacturers,
albeit the magnitude of the improvement was variable.[47] The 1. The Negative Impact of Metallic Impurities on
same approach was also shown to improve the conversion effi- Cell Performance
ciency of p-type SHJ solar cells based on p-type silicon wafers
not affected by BO defects, such as Ga-doped Cz-Si[48] and The first disadvantage associated with p-type silicon was
multicrystalline silicon.[49] Together, these results illustrate how attributed to most transition metal point defects having larger
the knowledge acquired developing stabilization processes for electron-than hole-capture cross-sections, which results in lower
BO defects in homojunction silicon solar cells also benefited lifetimes for p-type silicon for the same impurity concentra-
the SHJ technology. tion.[59] Yet, while p-type silicon is more sensitive to contamina-
This is particularly interesting as similar illuminated anneal- tion by interstitial iron (Fei),[60] more recent work from Schmidt
ing processes can improve the conversion efficiency of both et al. concluded that n-type silicon is more susceptible to contam-
homo and heterojunction solar cells despite treating different ination by other metallic impurities, including copper, cobalt,
defects. This is related to the temperature requirements of each chromium, and nickel.[61] Nevertheless, small concentrations
process. The completeness of the BO defect stabilization reaction of Fei can result in significant conversion efficiency losses in

Sol. RRL 2022, 2200449 2200449 (3 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

p-type silicon-based solar cells.[59] For this reason, minimizing temperature. Since it is recombination-active, this state limits the
the detrimental impact of metallic impurities is essential to performance of p-type Cz silicon solar cells. The degraded state
enable high VOC values. There are two main pathways to reduce lifetime can be completely recovered by a low-temperature
the concentration of recombination-active metallic impurities in annealing in the dark.[11] This is referred to as the annealed state,
silicon solar cells. The first approach is to improve the initial where BO defects are recombination-inactive but unstable under
wafer quality. Continuous improvements by wafer manufac- carrier injection. When subjected to carrier injection under
turers have recently enabled stable VOC values above 730 mV elevated temperatures, BO defects can be transferred to the
on SHJ solar cells fabricated with non-treated p-type Cz-Si stabilized state. In this state, the defects are recombination-
wafers.[48,62] The second approach is utilizing dedicated defect inactive and stable under carrier injection. The total concentra-
engineering treatments, such as gettering,[63] which is naturally tion of BO defects is equal to the sum of the concentration of
incorporated in the production of p-type passivated emitter and defects in the annealed, degraded, and stabilized states.
rear cell (PERC) solar cells.[31] A pre-fabrication gettering treat- However, the concentration of BO defect precursors can also
ment (i.e., applied to the wafers prior to solar cell fabrication) be thermally modulated.[69,70] Based on this key characteristic
has elevated the average conversion efficiency of a large batch of the defect, a revised model incorporating the existence of a
of p-type Cz-Si SHJ solar cells to the same level as their n-type fourth state has been proposed, which is referred to as a
counterparts.[62] Pre-fabrication gettering treatments have also reservoir.[36,70] The transitions between all states are depicted
been shown to increase the conversion efficiency of SHJ solar in Figure 2.
cells fabricated with a range of silicon wafer types, including BO defects form easily under illumination for any light capable
n-type Cz-Si wafers commonly used in industry.[49,64–66] It is of generating excess carriers in the bulk and under carrier injection
important to note that these results (and in fact most of the in the dark.[12,71] The lifetime decay due to BO defects is observed
results on p-type SHJ solar cells) were achieved without to occur in two stages. This is attributed to the formation of fast-
any optimization of the SHJ structure to accommodate and slow-recombination centers (FRC and SRC, respectively) and it
p-type c-Si wafers, and further optimization will likely lead to is commonly referred to as fast and slow degradation.[35,72] The two
performance improvements.[67] stages take place on a different time scale.[25] In commercial-grade
p-type B-doped Cz-Si (with [B] 1  1016 cm3), the fast degrada-
2. Susceptibility of p-type c-Si to LID tion occurs within the first few minutes, followed by a slower deg-
radation, which occurs over several hours at room temperature.[35]
As previously discussed, (uncompensated) n-type c-Si wafers When these c-Si wafers are used to fabricate PERC solar
display a significant intrinsic advantage over their p-type counter- cells, a fraction of the total BO defect precursor concentration
parts due to the inherent susceptibility of p-type B-doped Cz-Si to inherently present in the wafers is thermally annihilated during
BO-LID. BO defects are meta-stable, and the transitions between high-temperature solar cell processing.[30] Processes such as phos-
the different defect states can be explained through a multiple phorus diffusion for emitter formation,[27] thermal oxidation,[30]
states model. The first model featured three distinct defect states and firing[70] are effective in thermally annihilating a fraction of
and was proposed by Herguth et al.[34] and is sufficient to explain BO defects present in the silicon wafers. Therefore, the extent
the behavior of the BO defects under operational conditions.[68] of BO-LID experienced by cells that undergo such processes is
In the degraded state, the lifetime is limited by the recombina- greatly reduced, in comparison to non-thermally processed silicon
tion-active BO defects and is stable under illumination at room wafers. Yet, commercial PERC solar cells with starting VOC of

Figure 2. Schematic illustrating the boron–oxygen (BO) defect states and the possible state transitions.

Sol. RRL 2022, 2200449 2200449 (4 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

670 mV can experience losses of 20 mV due to the formation of 700 mV in devices with a starting VOC of 724 mV. This significant
BO defects. From an industrial perspective, Fertig et al. reported loss in VOC under low illumination conditions illustrates the
conversion efficiency losses of up to 5%rel related to BO-LID in inherent susceptibility to BO-LID of SHJ solar cells fabricated
PERC solar cells.[73] with commercial-grade p-type Cz-Si wafers.
In the context of SHJ solar cells, the impact of BO-LID on solar In the following paragraphs, we perform kinetic modeling to
cell performance is exacerbated by the excellent surface passiv- complement these experimental results and investigate the
ation provided by the a-Si:H films, with conversion efficiency impact of boron-dopant concentration, which determines ρ, on
losses over 14%rel reported.[74] Additionally, the absence of ther- the impact of BO-LID in p-type SHJ solar cells. The modeling
mal annihilation of BO defect precursors during fabrication can provide insights into the possible outcomes of past and
means that finished p-type SHJ solar cells will have a larger con- future attempts to fabricate SHJ solar cells with commercial-
centration of BO defect precursors than PERC solar cells. grade p-type Cz-Si wafers.
Here, we experimentally demonstrate the impact of these two In short, the modeling is based on the single defect theory[77]
factors (i.e., excellent surface passivation and absence of BO and a Python-based ordinary differential equation solver is
defect annihilation) in p-type SHJ solar cells. For this purpose, used for the kinetic modeling simulation. Degradation rates
Hevel Solar fabricated SHJ solar cells using commercial-grade based on the values reported by Bothe et al. and Glunz et al.
p-type Cz-Si wafers with base resistivity (ρ) of 0.8 Ω cm that at 300 K for NA = 2  1016 cm3 were used: 1.66  101 s1
did not undergo any additional defect-engineering process. and 1.05  104 s1 for fast and slow degradation, respec-
Figure 3 shows the VOC as a function of a time under the illu- tively.[25,29] Defect annihilation rates were assumed to be zero
mination of a p-type SHJ solar cell directly after dark annealing since these are negligible at 300 K.[25] The initial concentration
(200 °C for 600 s). The VOC was monitored in situ using a Sinton ratio of the precursor states for fast and slow degradation based
ELITE Suns-VOC tool.[75] The sample was kept immobile during on the values reported by Kim et al. for a dark annealing condi-
testing to avoid potential damage to the passivation quality tion of 200 °C for 600 s was used.[78] The lifetime limitation due
caused by sample handling. We note that the VOC reported to a two-level defect from Murphy et al. and adapted by Niewelt
for stability testing in this section were extracted from Suns- et al. to account for the unknown carrier capture cross sections of
VOC measurements at an illumination intensity equivalent to BO defects is used together with the parameterization from
0.9 suns due to a limitation of the tool. Hallam et al.[79–81] The changes in the injection-dependent effec-
Surprisingly, a VOC drop of 30 mV after just 10 s is observed tive lifetime were then modeled to determine the changes in VOC
under an illumination intensity equivalent to 0.1 suns for the as a function of BO-related degradation. Four different lifetime
solar cells with a starting VOC of 724 mV. Since at this illumina- components were used to model the effective lifetime during
tion intensity Δn << NA, the rate of degradation is determined degradation: the first corresponding to the lifetime limitation
by NA (i.e., by the boron doping concentration), and would be due to BO defects obtained via the kinetic modeling; the second
similar even under milder illumination intensities at a given tem- representing intrinsic recombination in silicon;[82] the third rep-
perature.[71] It is expected that all BO defects are recombination resenting surface-related recombination;[83] the fourth represent-
inactive but unstable at the end of the SHJ solar cell fabrication, ing all other SRH-related recombination mechanisms present in
since the conditions used for the curing process (dark annealing the sample. Only the first-lifetime component was varied as a
at >200 °C for >20 min) are sufficient to transfer BO defects to function of BO defect formation based on the kinetic modeling,
the unstable dark-annealed state.[76] Consequently, 10 s of acci- while the last three lifetime components were fixed. The excess
dental exposure to light between curing and the characterization carrier concentration () was determined by multiplying the effec-
of the finished solar cell is sufficient to yield poor VOC well below tive lifetime by the generation rate. by The VOC at each given
stage was calculated based on the following equation
ΔnN A þ Δn
V OC ¼ kT=q ln (1)
ni2

where ni is the intrinsic carrier concentration, q is the elementary


charge, NA is the acceptor dopant concentration, k is the
Boltzmann constant, and T is the temperature. The model dis-
plays good agreement with the experimental data (see Figure 3)
and is then used to estimate the different degradation scenarios
for p-type SHJ solar cells fabricated with c-Si wafers of different
NA, based on the known relationship between defect generation
rates (Rdeg), total defect concentration (Nt), substitutional boron
concentration ([BS]) and NA[35]

N t ∝ ½BS  ∝ N A (2)

Figure 3. Open-circuit voltage (VOC) measured (orange circles) and Rdeg ∝ ρillum 2 ∝ N A 2 (3)
modeled (blue line) of a p-type SHJ solar cell as a function of time under
illumination (0.1 suns, 25 °C). VOC was extracted at 0.9 suns due to a limi- It is assumed that all other wafer properties are the same
tation of the tool. (e.g., [Oi]), Rdeg and Nt change exclusively due to changes in

Sol. RRL 2022, 2200449 2200449 (5 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

NA (i.e., Δn << NA), and only BO defects are limiting the bulk in stabilizing BO defects in p-type SHJ solar cells. For this experi-
lifetime of the samples. It is important to note here that knowing ment, Hevel Solar fabricated p-type SHJ solar cells with p-type
the exact [Oi] is not required for the modeling. That is possible Cz-Si wafers that underwent a pre-fabrication phosphorus diffu-
because the Nt for each ρ (∝ NA) is calculated relative to the exper- sion process typically used in PERC solar cells (herein referred to
imentally determined Nt (for NA  2  1016 cm3) based on the as gettered), and control p-type Cz-Si wafers that did not undergo
relationship expressed in Equation (2). The surface passivation any fabrication processes. Both control and gettered wafers had a
quality was represented by a total J0e of 6 fA cm2 and the wafer resistivity of 0.8 Ω cm.
thickness was set to 140 μm. The ρ for uncompensated p-type Cz- Figure 5 shows the VOC as a function of time during light-
Si varied from 0.5 to 14 Ω cm. The resulting 1 sun VOC as a func- soaking for SHJ solar cells fabricated with commercial-grade
tion of time for the different ρ (∝ NA) scenarios is depicted in p-type Cz-Si wafers that underwent different defect-engineering
Figure 4. processes prior to light-soaking. Despite the control and gettered
The modeling results provide an insight into the expected VOC samples displaying similar starting VOC (measured after
losses experienced by p-type SHJ solar cells if c-Si wafers of dif- dark annealing) of 724 and 725 mV, respectively, gettering
ferent ρ are used. If c-Si wafers commonly used for the fabrica- significantly reduces the extent of degradation. After 10 s of
tion of PERC solar cells are used (ρ  0.5–1.8 Ω cm), VOC losses light-soaking, a VOC drop of 2.4 mV was recorded for the get-
between 43 and 2 mV could take place under just 10 s under illu- tered sample, while the VOC losses in the control sample
mination at room temperature for cells with starting VOC of exceeded 30 mV as previously discussed. After complete forma-
731–737 mV. This would correspond to VOC losses of tion of BO defects (i.e., when VOC stabilizes, in this case after
101–72 mV on a long-time scale when BO defects are fully light soaking for >105 s) the total VOC degradation is reduced
formed. However, this can be greatly reduced if c-Si wafers with from 100 (control) to 60 mV (gettered). Yet, this level of
lower NA are used. In the resistivity range (8–14 Ω cm) reported degradation is unacceptable for manufacturing.
by Descoeudres et al. for non-gettered c-Si wafers, the total VOC As previously discussed, bulk hydrogen is required for the
losses would be reduced to 22–11 mV.[62] Additionally, the stabilization of BO defects on finished devices. This poses a
degradation kinetics are slower for lower NA values, meaning challenge to the stabilization of BO defects in p-type SHJ solar
accidental light exposure will affect such cells even less than cells since hydrogenation via SiNX:H deposition and firing does
higher doping concentrations. not take place in the SHJ fabrication sequence. However, recent
This result helps to understand the impact of BO defects in the results showed that the SHJ fabrication sequence can introduce
early work on SHJ solar cells fabricated with p-type B-doped similar amounts of hydrogen into the bulk of the wafer,[84,85]
Cz-Si wafers. If commercial-grade c-Si wafers were used to fab- resulting in stable VOC over 734 mV on p-type Cz-Si wafers after
ricate SHJ solar cells, the severity of BO-related degradation a post-fabrication advanced hydrogenation process (AHP).[74]
would impact the conclusions regarding the potential of such Here, we experimentally demonstrate the stabilization of BO
wafers of achieving the VOC required by SHJ solar cells, if care defects in SHJ solar cells fabricated with commercial-grade
was not taken to prevent any unwanted light-exposure prior p-type Cz-Si wafers. More specifically, the complementary nature
(and during) solar cell characterization. of gettering and AHP is evidenced. During AHP, illumination
Fortunately, the continuous efforts in solving BO defects in generates excess carriers that are used to manipulate the hydro-
gen already existing in the solar cell.[40] A continuous-wave laser
PERC solar cells over the past two decades provide us with
with a wavelength of 980 nm was used to illuminate the whole
the knowledge required to defect engineer p-type SHJ solar cells
solar cell area simultaneously for a period of 30 s with an
and fabricate devices with high and stable VOC s. Here, we exper-
imentally show the role of gettering and advanced hydrogenation

Figure 4. Modeled 1 sun VOC of SHJ solar cells fabricated with p-type Cz-Si Figure 5. VOC as a function of time under illumination (0.1 suns, 25 °C) of
wafers of varied ρ (∝ NA), as a function of time as BO defects are gener- p-type SHJ solar cells that underwent different defect engineering pro-
ated under carrier-injection (for Δn << NA). The numbers inline indicate cesses. The open-circuit voltage (VOC) was extracted at an illumination
the corresponding ρ value. intensity of 0.9 suns due to a limitation of the tool.

Sol. RRL 2022, 2200449 2200449 (6 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

illumination intensity of 55 kW m2 (equivalent to 100 suns in Nevertheless, the results presented in this section illustrate
terms of illumination intensity at 980 nm). During the process, how BO-LID can be mitigated and p-type SHJ solar cells can
solar cells were heated on a hot plate with a temperature of be made stable based on knowledge acquired over the past
200 °C. During AHP, the solar cell’s temperature was monitored two decades of research on BO-LID in homojunction solar cells.
with an infrared thermometer, and a peak temperature of 255 °C It can be speculated that the great sensitivity of SHJ solar
was recorded at 30 s under illumination. cells fabricated with commercial-grade p-type Cz-Si wafers to
The effective BO defect concentration (denoted N*BO) was BO-LID could have impacted the conclusions surrounding the
calculated to quantify the effectiveness of the AHP process in performance of such cells in comparison to their n-type counter-
stabilizing BO defects. The N*BO, in units of an inverse lifetime, parts. Perhaps, if the BO defect stabilization process were already
has served as a convention for BO defect studies[35] as it allows well-established at the time when the core of the SHJ structure
the assessment of N*BO from minority carrier effective lifetime was developed, p-type Cz-Si wafers could have been potential
measurements. The N*BO after LS and after the AHP þ LS was candidates for the absorber material.
calculated for each sample using Equation (4)
1 1 1 3. Surface Passivation of p-type c-Si Wafers
N BO ¼ ¼  (4)
τBO τdegraded τannealed
The third drawback of using p-type c-Si wafers in SHJ solar cells
where τBO is the effective lifetime attributed to BO defects, outlined by De Wolf et al. relates to surface passivation.
τannealed is the background effective lifetime after DA and Particularly, it relates to the fact that surface states (e.g., silicon
τdegraded is the effective lifetime after LS.[35] The effective life- dangling bonds) feature a large electron-to-hole capture cross-
times were extracted from the injection-dependent lifetime section ratio. For this reason, the chemical passivation of p-type
curves obtained from the Suns-VOC measurements at a fixed- c-Si wafers can be more challenging to achieve in comparison to
injection level of Δn = 1.4  1014 cm3 to minimize the impact n-type c-Si wafers.[91,92] This means that for the same passivation
of small concentrations of Feiþ that could be present in the sam- scheme, the effective surface recombination velocity (Seff ) mea-
ples.[86] It is important to note that a fixed-injection level is used sured on a p-type wafer would be higher than that measured on
when determining N*BO to account for any additional lifetime an n-type wafer (for a similar doping concentration). Improving
change due to a reduction in the injection level at a fixed bias.[87] surface passivation quality (e.g., reducing Seff ) has been an ongo-
The percentage of passivated BO defects in each sample was ing effort across all c-Si-based technologies.[93] We have surveyed
calculated by taking the ratio between the N*BO on samples after the literature and identified Seff values reported for p- and n-type
LS and after the AHP þ LS. c-Si wafers passivated with a-Si:H(i) films, with the aim of under-
The AHP treatment alone increased the VOC of the previously standing the evolution and status of surface passivation quality
degraded sample to 729 mV. Upon subsequent stability testing, a for a-Si:H(i)/c-Si interfaces. A nonexhaustive list of the best Seff
loss of VOC of 26 mV is observed. This corresponded to the pas- values reported by several authors for a minority carrier density
sivation of 96.7% of BO defects, and a final VOC of 703 mV. Δn = 1  1015 cm3 as a function of the time of publication is dis-
However, the stable VOC was greatly improved with the addition played in Figure 6, where results on p-type c-Si wafers are rep-
of gettering. Directly after AHP, the gettered sample exhibited a resented by blue markers and on n-type c-Si wafers by orange
VOC of 731 mV. After light-soaking, a VOC loss of only 3 mV is markers. Figure 6––and the accompanying articles––provide
observed, which translated into a final stable VOC of 728 mV. This interesting insights into the evolution of the a-Si:H/c-Si technol-
corresponded to the stabilization of 97.2% of BO defects and ogy. Initially, it appears that the passivation of p-type c-Si wafers
illustrates the complementary nature of gettering and hydrogena- with a-Si:H(i) films was investigated as a potential approach for
tion via AHP in stabilizing the BO defects in p-type SHJ solar the rear passivation in homojunction devices.[94] This was moti-
cells. It is important to note that a similar fraction of the vated by the lower temperature regime required for surface pas-
stabilized BO defects, despite the more significant VOC losses sivation, which reduces both the energy budget and the risk of
between control and gettered samples, is caused by an initial bulk silicon contamination.[95] Additionally, a-Si:H(i) passivation
BO defect concentration reduction due to gettering. The similar has an intrinsic advantage over SiNX:H for the passivation of
fraction of stabilized defects via AHP in both control and gettered p-type surfaces as it avoids parasitic shunting effects caused
cells indicates that the final stable VOC is highly dependent on the by the fixed positive charge of SiNX:H films.[96]
concentration of remaining defects. The Seff values reported over the past two decades show an
A perhaps more elegant––and timely––solution to the stability improvement in the surface passivation quality of a-Si:H(i) for
issues associated with p-type SHJ solar cells would be switching both p- and n-type c-Si wafers over time. However, to the authors’
to p-type Cz-Si wafers doped with gallium, which are predicted to knowledge, there are no recent reports of Seff values for a-Si:H(i)
completely replace boron as the dopant agent for p-type c-Si passivation of p-type c-Si wafers, despite of the impressive Seff of
wafers by 2025.[88] We have recently (July 2021) demonstrated 1 cm s1 reported by Gloger et al. back in 2011.[97] At that point in
a stable VOC of 734 mV (and conversion efficiency of 22.6%) time, the Seff values reported for p-type c-Si wafers were at the
in a p-type Ga-doped SHJ solar cell fabricated in an industrial same level as the Seff values reported for their n-type counter-
environment.[48] More recently (March 2022), two groups parts. The reason for the apparent reduction in the research out-
independently reported conversion efficiencies of 24.47%[89] put related to a-Si:H(i) passivation of p-type c-Si wafers is unclear.
and 25.47%[90] for p-type Ga-doped SHJ solar cells, highlighting Yet, it can be speculated that this might be related to the
the potential of this approach. increased interest on Al2O3 passivation of p-type c-Si wafers,

Sol. RRL 2022, 2200449 2200449 (7 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

Figure 6. Nonexhaustive list of the lowest effective surface recombination velocity (Seff ) values of a-Si:H(i) passivated c-Si wafers reported by several
authors (for a minority carrier density Δn = 1  1015 cm3) as a function of the time of publication. Results on p-type wafers are represented by blue
markers and on n-type wafers by red markers. Open/closed markers represent results on float-zone silicon (FZ-Si)/ Czochralski (Cz)-Si wafers, respec-
tively. This figure is reproduced with data from.[92,94,95,97–105]

which was rapidly increasing in the early 2010s. Al2O3 was found
to be particularly interesting for the passivation of p-type surfaces
due to the built-in negative charge, which provided excellent
field-effect passivation.[106] A search on Scopus for article titles
including the key words “Al2O3” and “silicon” reveals a sharp
increase in the number of publications between 2012 and
2014 (from 10 to 35 publications per year). In contrast, the
continuous development and optimization of a-Si:H(i)
passivation of n-type c-Si wafers have enabled Seff values as
low as 0.3 cm s1 in 2021.[98]
In p-type SHJ solar cells, surface-related recombination
affecting the low-injection range is known to severely limit
the FF of such devices.[74,92,107] This was illustrated by
Descoeudres et al. in 2013, where SHJ solar cells fabricated with
n- and p-type FZ-Si wafers with conversion efficiencies >21%
were compared.[107] In their report, despite both p- and n-type Figure 7. Minority carrier effective lifetime as a function of injection-level
for p- and n-type wafers passivated with inþ and ipþ stacks reported by
devices achieving comparable peak-VOC values the conversion
Descoeudres et al. in 2013 and 2020. The data displayed in this figure
efficiency of the champion p-type SHJ solar cell (21.38%) was
was extracted from.[62,107]
limited by a 1.3%abs lower FF when compared to the champion
n-type device (22.14%). The reduced FF was found to be related
to a reduced minority carrier lifetime in the low-injection range can maintain a high level of passivation upon deposition of
rather than resistive losses. This is illustrated in Figure 7, where the a-Si:H(p) layer were made. Here, it is essential to mention
the injection-dependent minority carrier lifetimes of p- and that n-type SHJ solar cells can also be susceptible to surface-
n-type Fz-Si wafers passivated with inþ and ipþ stacks are dis- related recombination, reducing the lifetime at low injection.
played. It was noted by Descoeudres et al. that the reduction Adachi et al. demonstrated that improving the surface passiv-
in minority carrier lifetime at the low-injection range could ation quality of the a-Si layers can minimize such recombina-
not be explained by bulk recombination, as high-quality FZ-Si tion.[109] These results indicate that overcoming the surface
wafers were used in their experiment. Moreover, they noted that passivation barrier preventing the adoption of p-type c-Si wafers
longer low-injection lifetimes are observed when similar p-type in SHJ solar cells is possible. These results also highlight low Seff
FZ-Si wafers are passivated with Al2O3,[108] which provides fur- values can be achieved via a-Si:H(i) passivation of p-type c-Si
ther evidence that the lower lifetimes at low-injection are related wafers and illustrate that continuous efforts in optimizing such
to recombination at the a-Si:H/c-Si interface. a structure can lead to further reductions in Seff.
However, Descoeudres et al. have recently demonstrated that
this low-injection minority carrier lifetime limiting the perfor-
mance of p-type c-Si can be suppressed. Although no specific 4. Final Thoughts
method was disclosed, the authors noted that improvements
in the silane plasma conditions, hydrogenation of the a-Si: In this perspective article, we first reviewed the history of SHJ
H(i)/c-Si interface, and efforts in ensuring the a-Si:H(i) layer solar cells and BO defects. This helped us to draw a timeline

Sol. RRL 2022, 2200449 2200449 (8 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

of events, which highlighted that the first instance of BO defect be more expensive than the p-type wafers used in PERC cells.
stabilization took place well after the core of the SHJ technology However, our previous work shows that commercial-grade p-type
was already developed. This historical review was followed by a Ga-doped Cz-Si wafers can yield comparable VOC values to ref-
discussion surrounding the three key barriers preventing the use erence SHJ solar cells fabricated with n-type Cz-Si wafers.[48]
of p-type c-Si wafers in SHJ solar cells. It was shown that the Further work is required to evaluate the impact of variations
continuous improvements in wafer lifetime, stability, and in wafer quality across different p-type Cz-Si ingots (and ingot
surface passivation quality of p-type c-Si have enabled stable positions) on the performance of p-type SHJ solar cells.
VOC values compatible with the SHJ technology.[48,62,74] Nevertheless, the historical perspective presented in this arti-
These improvements have recently enabled a record conver- cle illustrates how the lack of timely solutions for a single defect
sion efficiency of 25.47% (VOC of 747.6 mV) for a full-area p-type can result in a material being considered unsuitable for novel
SHJ solar cell fabricated with a Ga-doped Cz-Si wafer, with a applications. This is particularly important as the industry rapidly
remarkably high FF of 83.80%.[90] Descoeudres et al. presented shifts from boron to gallium as the main dopant agent for p-type
promising results with p- and n-type SHJ solar cells processed in silicon. It is of great importance to identify the defect(s) respon-
parallel with FZ-Si wafers. The p-type SHJ solar cells exhibited a sible for lifetime instabilities recently observed in solar cells fab-
peak conversion efficiency of 23.76%, while their n-type counter- ricated with p-type Ga-doped c-Si wafers,[119–122] as these wafers
parts achieved 24.21%. This small gap (0.45%abs) in conversion have the potential to enable a next-generation of high-efficiency
efficiency is interesting as it is close to the target conversion effi- silicon solar cells based on p-type c-Si wafers.
ciency required to enable the switch from n- to p-type wafers in
SHJ solar cell manufacturing. According to the techno-economic
analysis by Chang et al., based on a wafer premium of 8%, the Acknowledgements
conversion efficiency of p-type SHJ solar cells needs to be This work was supported by the Australian Government through the
within 0.4%abs of n-type SHJ solar cells for the transition to Australian Centre for Advanced Photovoltaics (ACAP) and the
be cost-effective.[110] It is also important to emphasize another Australian Renewable Energy Agency (ARENA) (2017/RND005). The views
key result of the work from Descoeudres et al., which was noted expressed herein are not necessarily the views of the Australian
earlier in this article. The conversion efficiency of a large batch of Government, and the Australian Government does not accept responsi-
SHJ solar cells fabricated with p-type (B-doped) and n-type Cz-Si bility for any information or advice contained herein. B.V.S acknowledges
the CSIRO Research Office for his post-doctoral fellowship. M.K. acknowl-
wafers was at the same level, with the only difference that the
edges the support of ACAP for his post-doctoral fellowship.
p-type wafers underwent a pre-fabrication gettering treatment.[62]
The conversion efficiencies achieved by p-type c-Si wafers in
other carrier-selective contact technologies, such as TOPCon Conflict of Interest
solar cells are even more impressive. In such devices, p-type
c-Si wafers have enabled record conversion efficiencies of The authors declare no conflict of interest.
26.0% and 26.1% by Richter et al. and Haase et al., respectively,
which are currently higher than the record conversion efficiency
for n-type TOPCon solar cells (25.80%).[111,112] Keywords
The continuous improvements in wafer quality, hydrogena-
tion of bulk and surface defects, and surface passivation heterojunction, passivation, p-type, silicon, stability
approaches could bring the performance of p-type devices to a Received: May 20, 2022
point, where other intrinsic and processing-related advantages Revised: July 24, 2022
of p-type doping of silicon might play a major role in the wafer Published online:
choice for next-generation high-efficiency solar cells featuring
carrier-selective contacts. These advantages include the higher
diffusivity of minority carriers, the ease of forming n-type doped [1] W. E. Spear, P. G. L. Comber, Solid State Commun. 1975, 17, 1193.
layers via phosphorus diffusion, and the better segregation coef- [2] S. De Wolf, A. Descoeudres, Z. C. Holman, C. Ballif, Green 2012, 2, 7.
ficient for boron (close to unity),[113] which improves ingot utili- [3] W. Fuhs, K. Niemann, J. Stuke, in Tetrahedrally Bond. Amorph.
zation. The former enables a higher theoretical limit for the Semicond. Int. Conf., ASCE, Yorktown Heights, NY 1974, pp. 345–350.
conversion efficiency of p-type solar cells.[114] The latter relates [4] J. I. Pankove, M. L. Tarng, Appl. Phys. Lett. 1979, 34, 156.
to material usage and is particularly important due to the high [5] H. Fischer, W. Pschunder, in Proc. 10th IEEE Photovolt. Spec. Conf.,
sensitivity of SHJ solar cells (and other cell structures where IEEE, New York 1973, pp. 404–411.
surface-related recombination is minimized) to changes in bulk [6] M. Taguchi, M. Tanaka, T. Matsuyama, T. Matsuoka, S. Tsuda,
S. Nakano, Y. Kishi, Y. Kuwano, in Tech. Dig. PVSEC-5, PVSEC,
lifetime, which can be significantly impacted by doping varia-
Kyoto, Japan 1990.
tions across the ingot.[115–117] In addition to these intrinsic advan-
[7] M. Tanaka, M. Taguchi, T. Matsuyama, T. Sawada, S. Tsuda,
tages of p-type c-Si wafers, the adoption of such wafers would S. Nakano, H. Hanafusa, Y. Kuwano, Jpn. J. Appl. Phys. 1992, 31, 3518.
further benefit from the cheaper wafer cost, since p-type Cz-Si [8] V. G. Weizer, H. W. Brandhorst, J. D. Broder, R. E. Hart,
wafers are currently 8–10% cheaper than their n-type counter- J. H. Lamneck, J. Appl. Phys. 1979, 50, 4443.
parts (for the same wafer size/thickness).[118] Yet, the SHJ [9] K. Graff, H. Pieper, Phys. Status Solidi A 1978, 49, 137.
solar cell’s strict lifetime requirements also apply to p-type [10] J. H. Reiss, R. R. King, K. W. Mitchell, Appl. Phys. Lett. 1996, 68, 3302.
wafers. As such, p-type wafers with a high initial lifetime are [11] J. Schmidt, A. G. Aberle, R. Hezel, in Conf. Rec. Twenty Sixth IEEE
required to fabricate high-efficiency SHJ solar cells, which might Photovolt. Spec. Conf. - 1997, IEEE, Anaheim, CA 1997, pp. 13–18.

Sol. RRL 2022, 2200449 2200449 (9 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

[12] S. W. Glunz, E. Schäffer, S. Rein, in Proc. 3rd World Conf. Photovolt. [43] M. Wright, M. Kim, P. Dexiang, X. Xin, Z. Wenbin, B. Wright,
Energy Convers. 2003, IEEE, Piscataway, NJ 2003, pp. 919–922. B. Hallam, in AIP Conf. Proc., AIP Publishing, Leuven, Belgium
[13] D. Macdonald, F. Rougieux, A. Cuevas, B. Lim, J. Schmidt, 2019, p. 110006.
M. Di Sabatino, L. J. Geerligs, J. Appl. Phys. 2009, 105, 093704. [44] E. Kobayashi, S. De Wolf, J. Levrat, G. Christmann, A. Descoeudres,
[14] T. Schutz-Kuchly, J. Veirman, S. Dubois, D. R. Heslinga, Appl. Phys. S. Nicolay, M. Despeisse, Y. Watabe, C. Ballif, Appl. Phys. Lett. 2016,
Lett. 2010, 96, 093505. 109, 153503.
[15] M. Forster, E. Fourmond, F. E. Rougieux, A. Cuevas, R. Gotoh, [45] E. Kobayashi, S. De Wolf, J. Levrat, A. Descoeudres, M. Despeisse,
K. Fujiwara, S. Uda, M. Lemiti, Appl. Phys. Lett. 2012, 100, F.-J. Haug, C. Ballif, Sol. Energy Mater. Sol. Cells 2017, 173, 43.
042110. [46] J. Cattin, L.-L. Senaud, J. Haschke, B. Paviet-Salomon, M. Despeisse,
[16] J. Schmidt, A. Cuevas, J. Appl. Phys. 1999, 86, 3175. C. Ballif, M. Boccard, IEEE J. Photovoltaics 2021, 11, 1.
[17] S. W. Glunz, S. Rein, W. Warta, J. Knobloch, W. Wettling, in Proc. [47] M. Wright, A. Soeriyadi, B. Wright, D. Andronikov, I. Nyapshaev,
2nd World Conf. Photovolt. Sol. Energy Convers., Vienna 1998, S. Abolmasov, A. Abramov, B. Hallam, IEEE J. Photovoltaics 2022,
pp. 1343–1346. 12, 267.
[18] S. W. Glunz, S. Rein, J. Knobloch, W. Wettling, T. Abe, Prog. [48] B. Vicari Stefani, M. Kim, M. Wright, A. Soeriyadi, D. Andronikov,
Photovoltaics Res. Appl. 1999, 7, 463. I. Nyapshaev, S. Abolmasov, K. Emtsev, A. Abramov, B. Hallam,
[19] T. Abe, T. Hirasawa, K. Tokunaga, T. Igarashi, M. Yamaguchi, Sol. RRL 2021, 5, 2100406.
US6815605B1 2004. [49] D. Chen, M. Kim, J. Shi, B. Vicari Stefani, Z. Yu, S. Liu, R. Einhaus,
[20] M. Taguchi, Prog. Photovoltaics Res. Appl. 2000, 8, 503. S. Wenham, Z. Holman, B. Hallam, Prog. Photovoltaics Res. Appl.
[21] M. A. Green, K. Emery, D. L. King, S. Igari, Prog. Photovoltaics Res. 2019, 29, 1165.
Appl. 2000, 8, 187. [50] A. Herguth, G. Hahn, J. Appl. Phys. 2010, 108, 114509.
[22] H. Sakata, T. Nakai, T. Baba, M. Taguchi, S. Tsuge, K. Uchihashi, [51] S. Wilking, C. Beckh, S. Ebert, A. Herguth, G. Hahn, Sol. Energy Mater.
S. Kiyama, in Conf. Rec. Twenty-Eighth IEEE Photovolt. Spec. Conf. - Sol. Cells 2014, 131, 2.
2000 Cat No00CH37036, IEEE, Anchorage, AK 2000, pp. 7–12. [52] M. Wright, A. Soeriyadi, M. Kim, B. Wright, B. Vicari Stefani,
[23] J. Zhao, A. Wang, M. A. Green, Prog. Photovoltaics Res. Appl. 1999, D. Andronikov, I. Nyapshaev, S. Abolmasov, A. Abramov,
7, 471. R. S. Bonilla, B. Hallam, Sol. Energy Mater. Sol. Cells Under revision.
[24] J. Schmidt, K. Bothe, Phys. Rev. B 2004, 69, 024107. [53] M. Bivour, S. Schröer, M. Hermle, S. W. Glunz, Sol. Energy Mater. Sol.
[25] K. Bothe, J. Schmidt, J. Appl. Phys. 2006, 99, 013701. Cells 2014, 122, 120.
[26] K. Bothe, R. Hezel, J. Schmidt, Appl. Phys. Lett. 2003, 83, 1125. [54] A. N. Fioretti, M. Boccard, R. Monnard, C. Ballif, IEEE J. Photovoltaics
[27] K. Bothe, J. Schmidt, R. Hezel, in Conf. Rec. Twenty-Ninth IEEE 2019, 9, 1158.
Photovolt. Spec. Conf. 2002, IEEE, New Orleans, LA 2002, pp. 194–197. [55] G. Nogay, J. P. Seif, Y. Riesen, A. Tomasi, Q. Jeangros, N. Wyrsch,
[28] K. Bothe, R. Sinton, J. Schmidt, Prog. Photovoltaics Res. Appl. 2005, 13, F.-J. Haug, S. De Wolf, C. Ballif, IEEE J. Photovoltaics 2016, 6, 1654.
287. [56] J. P. Seif, A. Descoeudres, G. Nogay, S. Hanni, S. M. de Nicolas,
[29] S. W. Glunz, S. Rein, J. Y. Lee, W. Warta, J. Appl. Phys. 2001, 90, 2397. N. Holm, J. Geissbuhler, A. Hessler-Wyser, M. Duchamp,
[30] S. W. Glunz, S. Rein, W. Warta, J. Knobloch, W. Wettling, Sol. Energy R. E. Dunin-Borkowski, M. Ledinsky, S. De Wolf, C. Ballif, IEEE J.
Mater. Sol. Cells 2001, 65, 219. Photovoltaics 2016, 6, 1132.
[31] B. Hallam, D. Chen, M. Kim, B. Stefani, B. Hoex, M. Abbott, [57] L.-L. Senaud, G. Christmann, A. Descoeudres, J. Geissbuhler,
S. Wenham, Phys. Status Solidi A 2017, 214, 1700305. L. Barraud, N. Badel, C. Allebe, S. Nicolay, M. Despeisse,
[32] N. E. Grant, V. P. Markevich, J. Mullins, A. R. Peaker, F. Rougieux, B. Paviet-Salomon, C. Ballif, IEEE J. Photovoltaics 2019, 9, 1217.
D. Macdonald, Phys. Status Solidi RRL 2016, 10, 443. [58] Z. Wu, W. Duan, A. Lambertz, D. Qiu, M. Pomaska, Z. Yao, U. Rau,
[33] M. Kim, P. Hamer, H. Li, D. Payne, S. Wenham, M. Abbott, B. Hallam, L. Zhang, Z. Liu, K. Ding, Appl. Surf. Sci. 2021, 542, 148749.
Front. Energy 2017, 11, 32. [59] J. E. Cotter, J. H. Guo, P. J. Cousins, M. D. Abbott, F. W. Chen,
[34] A. Herguth, G. Schubert, M. Kaes, G. Hahn, in 2006 IEEE 4th K. C. Fisher, IEEE Trans. Electron Devices 2006, 53, 1893.
World Conf. Photovolt. Energy Conf., IEEE, Waikoloa, HI 2006, [60] D. Macdonald, L. J. Geerligs, Appl. Phys. Lett. 2004, 85, 4061.
pp. 940–943. [61] J. Schmidt, B. Lim, D. Walter, K. Bothe, S. Gatz, T. Dullweber,
[35] T. Niewelt, J. Schon, W. Warta, S. W. Glunz, M. C. Schubert, IEEE J. P. P. Altermatt, IEEE J. Photovoltaics 2013, 3, 114.
Photovoltaics 2017, 7, 383. [62] A. Descoeudres, J. Horzel, B. Paviet-Salomon, L. Senaud,
[36] A. Herguth, B. Hallam, in AIP Conf. Proc., Vol 1999, AIP Publishing G. Christmann, J. Geissbühler, P. Wyss, N. Badel, J. Schüttauf,
LLC, Lausanne, Switzerland, 2018, p. 130006. J. Zhao, C. Allebé, A. Faes, S. Nicolay, C. Ballif, M. Despeisse,
[37] N. Nampalli, B. Hallam, C. Chan, M. Abbott, S. Wenham, Appl. Phys. Prog. Photovoltaics Res. Appl. 2020, 28, 569.
Lett. 2015, 106, 173501. [63] A. Liu, S. P. Phang, D. Macdonald, Sol. Energy Mater. Sol. Cells 2022,
[38] S. Wilking, S. Ebert, A. Herguth, G. Hahn, J. Appl. Phys. 2013, 114, 234, 111447.
194512. [64] F. Jay, D. Muñoz, T. Desrues, E. Pihan, V. Amaral de Oliveira,
[39] L. Helmich, D. C. Walter, R. Falster, V. V. Voronkov, J. Schmidt, Sol. N. Enjalbert, A. Jouini, Sol. Energy Mater. Sol. Cells 2014, 130, 690.
Energy Mater. Sol. Cells 2021, 232, 111340. [65] R. Basnet, W. Weigand, Z. J. Yu, C. Sun, S. P. Phang, H. C. Sio,
[40] B. J. Hallam, P. G. Hamer, A. M. Ciesla née Wenham, C. E. Chan, F. E. Rougieux, Z. C. Holman, D. Macdonald, Sol. Energy Mater.
B. Vicari Stefani, S. Wenham, Prog. Photovoltaics Res. Appl. 2020, Sol. Cells 2020, 205, 110287.
28, 1217. [66] M. Wright, B. Vicari Stefani, A. Soeriyadi, R. Basnet, C. Sun,
[41] P. Hamer, B. Hallam, M. Abbott, S. Wenham, Phys. Status Solidi RRL W. Weigand, Z. Yu, Z. Holman, D. Macdonald, B. J. Hallam, Phys.
2015, 9, 297. Status Solidi RRL 2021, 15, 2100170.
[42] B. J. Hallam, P. G. Hamer, S. Wang, L. Song, N. Nampalli, [67] D. L. Batzner, D. Habermann, L. Andreetta, W. Frammelsberger,
M. D. Abbott, C. E. Chan, D. Lu, A. M. Wenham, L. Mai, R. Kramer, D. Lachenal, B. Legradic, J. Meixenberger, P. Papet,
N. Borojevic, A. Li, D. Chen, M. Y. Kim, A. Azmi, S. Wenham, B. Strahm, G. Wahli, in Proc. EU PVSEC 2015 Hambg., EU PVSEC,
Energy Procedia 2015, 77, 799. Hamburg, Germany 2015, pp. 788–790.

Sol. RRL 2022, 2200449 2200449 (10 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH
www.advancedsciencenews.com www.solar-rrl.com

[68] M. Kim, PhD Thesis, Univeristy of New South Wales, 2020. [97] S. Gloger, N. Brinkmann, B. Terheiden, Energy Procedia 2011, 8, 666.
[69] D. C. Walter, R. Falster, V. V. Voronkov, J. Schmidt, Sol. Energy Mater. [98] S. Manzoor, M. Bertoni, in 2021 IEEE 48th Photovolt. Spec. Conf.
Sol. Cells 2017, 173, 33. PVSC, IEEE, Fort Lauderdale, FL 2021, pp. 1286–1288.
[70] N. Nampalli, Sol. Energy Mater. Sol. Cells 2017, 173, 12. [99] S. Bernardini, M. I. Bertoni, Phys. Status Solidi A 2018, 216,
[71] H. Hashigami, Y. Itakura, T. Saitoh, J. Appl. Phys. 2003, 93, 4240. 1800705.
[72] V. V. Voronkov, R. Falster, J. Appl. Phys. 2010, 107, 053509. [100] A. Neumüller, O. Sergeev, S. J. Heise, S. Bereznev, O. Volobujeva,
[73] F. Fertig, R. Lantzsch, A. Mohr, M. Schaper, M. Bartzsch, D. Wissen, J. F. L. Salas, M. Vehse, C. Agert, Nano Energy 2018, 43, 228.
F. Kersten, A. Mette, S. Peters, A. Eidner, J. Cieslak, K. Duncker, [101] X. Zhang, A. Cuevas, B. Demaurex, S. De Wolf, Energy Procedia 2014,
M. Junghänel, E. Jarzembowski, M. Kauert, B. , D. Meißner, 55, 865.
B. Reiche, S. Geißler, S. Hörnlein, C. Klenke, L. Niebergall, [102] X. Zhang, S. Hargreaves, Y. Wan, A. Cuevas, Phys. Status Solidi RRL
A. Schönmann, A. Weihrauch, F. Stenzel, A. Hofmann, 2014, 8, 231.
T. Rudolph, A. Schwabedissen, M. Gundermann, M. Fischer, [103] H. Wu, H. T. Nguyen, L. Black, A. Liu, R. Liu, W. Chen, D. Kang,
et al., Energy Procedia 2017, 124, 338. W. Yang, D. Macdonald, IEEE J. Photovoltaics 2020, 10, 1307.
[74] B. Vicari Stefani, A. Soeriyadi, M. Wright, D. Chen, M. Kim, Y. Zhang, [104] J. Mitchell, D. Macdonald, A. Cuevas, J. Cornish, Proc. 42nd Annual
B. Hallam, Sol. RRL 2020, 4, 2000134. Conf. of the Australian and New Zealand Solar Energy Society, Perth,
[75] M. Vaqueiro Contreras, A. Teymouri, A. Mahboubi Soufiani, C. Chan, Western Australia, 2004.
A. Ciesla, H. Wilterdink, B. Hallam, in Proc. 38th Eur. Photovolt. Sol. [105] J.-W. A. Schüttauf, K. H. M. van der Werf, I. M. Kielen,
Energy Conf., Lisbon 2021, pp. 250–252. W. G. J. H. M. van Sark, J. K. Rath, R. E. I. Schropp, Appl. Phys.
[76] A. Herguth, G. Schubert, M. Kaes, G. Hahn, Prog. Photovoltaics Res. Lett. 2011, 98, 153514.
Appl. 2008, 16, 135. [106] B. Hoex, S. B. S. Heil, E. Langereis, M. C. M. van de Sanden,
[77] B. Hallam, M. Abbott, T. Naerland, S. Wenham, Phys. Status Solidi W. M. M. Kessels, Appl. Phys. Lett. 2006, 89, 042112.
RRL 2016, 10, 520. [107] A. Descoeudres, Z. C. Holman, L. Barraud, S. Morel, S. D. Wolf,
[78] M. Kim, M. Abbott, N. Nampalli, S. Wenham, B. Stefani, B. Hallam, J. C. Ballif, IEEE J. Photovoltaics 2013, 3, 7.
Appl. Phys. 2017, 121, 053106. [108] B. Hoex, J. Schmidt, P. Pohl, M. C. M. van de Sanden,
[79] J. D. Murphy, K. Bothe, R. Krain, V. V. Voronkov, R. J. Falster, J. Appl. W. M. M. Kessels, J. Appl. Phys. 2008, 104, 044903.
Phys. 2012, 111, 113709. [109] D. Adachi, J. L. Hernández, K. Yamamoto, Appl. Phys. Lett. 2015,
[80] T. Niewelt, S. Mägdefessel, M. C. Schubert, J. Appl. Phys. 2016, 120, 107, 233506.
085705. [110] N. L. Chang, M. Wright, R. Egan, B. Hallam, Cell Rep. Phys. Sci. 2020,
[81] B. Hallam, M. Kim, M. Abbott, N. Nampalli, T. Nærland, B. Stefani, 1, 100069.
S. Wenham, Sol. Energy Mater. Sol. Cells 2017, 173, 25. [111] F. Haase, C. Hollemann, S. Schäfer, A. Merkle, M. Rienäcker,
[82] A. Richter, M. Hermle, S. W. Glunz, IEEE J. Photovoltaics 2013, J. Krügener, R. Brendel, R. Peibst, Sol. Energy Mater. Sol. Cells
3, 1184. 2018, 186, 184.
[83] A. Cuevas, D. Macdonald, Sol. Energy 2004, 76, 255. [112] A. Richter, R. Müller, J. Benick, F. Feldmann, B. Steinhauser,
[84] C. Sun, D. Chen, W. Weigand, R. Basnet, S. P. Phang, B. Hallam, C. Reichel, A. Fell, M. Bivour, M. Hermle, S. W. Glunz, Nat.
Z. C. Holman, D. Macdonald, Appl. Phys. Lett. 2018, 113, 152105. Energy 2021, 6, 429.
[85] C. Sun, W. Weigand, J. Shi, Z. Yu, R. Basnet, S. P. Phang, [113] P. S. Ravishankar, L. P. Hunt, R. W. Francis, J. Electrochem. Soc.
Z. C. Holman, D. Macdonald, Appl. Phys. Lett. 2019, 115, 252103. 1984, 131, 872.
[86] D. H. Macdonald, L. J. Geerligs, A. Azzizi, J. Appl. Phys. 2004, [114] T. Niewelt, B. Steinhauser, A. Richter, B. Veith-Wolf, A. Fell,
95, 1021. B. Hammann, N. E. Grant, L. Black, J. Tan, A. Youssef,
[87] S. Rein, W. Warta, S. W. Glunz, G. P. Willeke, in Proc. 17th Eur. J. D. Murphy, J. Schmidt, M. C. Schubert, S. W. Glunz, Sol.
Photovolt. Sol. Energy Conf., Munich, Germany 2001, pp. 1555–1560. Energy Mater. Sol. Cells 2022, 235, 111467.
[88] ITRPV, International Roadmap For Photovoltaics, 12th ed., VDMA, [115] J. Veirman, R. Varache, M. Albaric, A. Danel, B. Guo, N. Fu,
Frankfurt, Germany 2021. Y. C. Wang, Sol. Energy Mater. Sol. Cells 2021, 228, 111128.
[89] CEA INES, RECORD heterojunction cell with P-type silicon 2022. [116] A. Srinivasa, S. Herasimenka, A. Augusto, S. Bowden, in 2020 47th
[90] LONGi Solar, Longi achieves 25.47% efficiency for gallium-doped IEEE Photovolt. Spec. Conf. PVSC, IEEE, Calgary, AB, 2020,
p-type heterojunction solar cell 2022. pp. 2238.
[91] Yibin Bai, J. E. Phillips, A. M. Barnett, IEEE Trans. Electron Devices [117] F. Meng, J. Liu, L. Shen, J. Shi, A. Han, L. Zhang, Y. Liu, J. Yu,
1998, 45, 1784. J. Zhang, R. Zhou, Z. Liu, Front. Energy 2017, 11, 78.
[92] S. Olibet, E. Vallat-Sauvain, C. Ballif, Phys. Rev. B 2007, 76, 035326. [118] Y. Wang, 4th Int. Workshop on Silicon Heterojunction Solar Cells:
[93] R. S. Bonilla, B. Hoex, P. Hamer, P. R. Wilshaw, Phys. Status Solidi A Science and Industry Technology 2021, Oral presentation.
2017, 214, 1700293. [119] N. E. Grant, J. R. Scowcroft, A. I. Pointon, M. Al-Amin,
[94] M. Schaper, J. Schmidt, H. Plagwitz, R. Brendel, Prog. Photovoltaics P. P. Altermatt, J. D. Murphy, Sol. Energy Mater. Sol. Cells 2020,
Res. Appl. 2005, 13, 381. 206, 110299.
[95] S. Dauwe, J. Schmidt, R. Hezel, in Conf. Rec. Twenty-Ninth IEEE [120] N. E. Grant, P. P. Altermatt, T. Niewelt, R. Post, W. Kwapil,
Photovolt. Spec. Conf. 2002, IEEE, New Orleans, LA 2002, M. C. Schubert, J. D. Murphy, Sol. RRL 2021, 5, 2000754.
pp. 1246–1249. [121] R. Post, T. Niewelt, W. Kwapil, M. Schubert, IEEE J. Photovoltaics
[96] G. Dingemans, M. M. Mandoc, S. Bordihn, M. C. M. van de Sanden, 2021, 12, 1.
W. M. M. Kessels, Appl. Phys. Lett. 2011, 98, 222102. [122] M. Winter, D. Walter, J. Schmidt, IEEE J. Photovoltaics 2021, 11, 1.

Sol. RRL 2022, 2200449 2200449 (11 of 11) © 2022 Commonwealth Scientific and Industrial Research Organisation. Solar RRL published by Wiley-VCH GmbH

You might also like