Heterojunction solar cell

SHJ has the highest efficiency amongst crystalline silicon solar cells in both laboratory (world record efficiency)[2][21][23] and commercial production (average efficiency). In 2023, the average efficiency for commercial SHJ cells was 25.0%, compared with 24.9% for n-type TOPCon and 23.3% for p-type PERC.[26] The high efficiency is owed mostly to very high open-circuit voltages—consistently over 700 mV—as a result of excellent surface passivation. Since 2023, SHJ bottom cells in Perovskite tandems also hold the highest 2-junction cell efficiency at 33.2%.[23]

Bifaciality refers to the ability of a solar cell to accept light from the front or rear surface. The collection of light from the rear surface can significantly improve energy yields in deployed solar arrays.[27] SHJ cells can be manufactured with a conductive ARC on both sides, allowing a bifaciality factor above 90%, compared to ~70% for PERC cells with rear grid.[28] Bifacial solar cells are expected to significantly increase their market share over monofacial cells to 85% by 2032.[3]

The temperature coefficient refers to how the output power of a solar module changes with temperature. Typically, solar modules see a reduction in output power and efficiency at elevated temperatures. From lab testing and supplier datasheet surveys, modules fabricated with SHJ cells consistently measure an equal or lower temperature coefficient (ie. the decrease in efficiency is less severe) compared with Al-BSF, PERC, PERT and hybrid PERT/rear-heterojunction solar cells. This applies to a range of parameters, including open-circuit voltage, maximum power point power, short circuit current and fill factor.[29] The temperature sensitivity of solar cells has been inversely correlated to high open-circuit voltages compared to the absorber band gap potential,[30] as noted by Martin Green in 1982; "As the open-circuit voltage of silicon solar cells continues to improve, one resulting advantage, not widely appreciated, is reduced temperature sensitivity of device performance".[31] Thus the low temperature sensitivity of SHJ cells has been attributed to high ${\displaystyle V_{OC}}$ from well passivated contacts.[32]

SHJ production lines fundamentally do not use high temperature equipment such as diffusion or metal paste curing furnaces,[17] and on average have a lower power consumption per watt of fabricated cells. According to China PV Industry Development Roadmap, in 2022, the average electricity consumption of n-type Heterojunction cell lines was 47,000 kWh/MW, whereas p-type PERC production lines consumed about 53,000 kWh/MW and for n-type TOPCon, about 56,000 kWh/MW. It is estimated that by 2030, the power consumption of n-type Heterojunction, p-type PERC and n-type TOPCon cell production lines will drop to 34,000 kWh/MW, 35,000 kWh/MW and 42,000 kWh/MW respectively.[26]

Crystalline silicon wafers used in solar cells typically have a thickness between 130–180 μm. The mass of consumed silicon wafer comprises a significant proportion of the cost of the solar module, and as such reducing the wafer thickness has potential to achieve significant cost reduction. However, fewer photons are absorbed in thinner silicon. To compensate for this, as long as surface recombination is effectively suppressed, thinner wafers can maintain—or even improve upon—very high open-circuit voltages.[33][34] That is, the open-circuit voltage is increased to compensate for losses in short-circuit current. As SHJ cells have excellent surface passivation, reduction in their wafer thickness is more feasible than with other crystalline silicon solar cell technologies.[17][28] As such, high efficiencies have been reported over a large range of wafer thicknesses, with the minimum on the order of 50 μm.[35]

Although high efficiency SHJ cells can be manufactured using a p-type silicon substrate, the low temperature constraint on SHJ production makes the process of gettering (management of contamination defects) impossible and bulk hydrogenation cannot reliably passivate excessive defects. For the same concentration of contaminant transition metal defects, n-type wafers have a higher minority carrier lifetime due to the smaller capture cross section of holes (the minority charge carrier) compared to electrons. Similarly, the capture cross section ratio of electrons to holes is large for surface states (eg. silicon dangling bonds) and therefore well passivated surfaces are easier to achieve on n-type wafers.[18][32] For these reasons, n-type wafers are strongly preferred for manufacturing, as inconvenient steps for improving bulk lifetimes are cut out and the risk of developing light-induced degradation is reduced.[36] However, the cost of n-type wafers is usually cited to be about 8–10% higher than p-type.[36][37]

One of the first steps in manufacturing crystalline silicon solar cells includes texturing and cleaning the surface of the silicon wafer substrate. For monocrystalline wafers, this involves an anisotropic wet chemical etch using a mixture of an alkaline solution (usually potassium hydroxide or metal ion-free tetramethylammonium hydroxide) and an organic wetting agent (traditionally isopropyl alcohol, but now proprietary additives are used). The etch forms the light-trapping pyramidal texture that improves the output current of the finished solar cell. Due to stringent requirements for surface cleanliness for SHJ compared to PERC, the texturing and cleaning process is relatively more complex and consumes more chemicals. Some of these surface treatment steps include RCA cleaning, sulfuric acid/peroxide mixtures to remove organics, removal of metal ions using hydrochloric acid, and nitric acid oxidative cleaning and etch backs.[38] Recent developments in research has found that oxidative cleaning with ozonated water may help improve process efficiency and reduce waste, with the possibility of completely replacing RCA cleaning whilst maintaining the same surface quality.[38][39]

The substrate, in which electron-hole pairs are formed, is usually n-type monocrystalline silicon doped with phosphorus. In industrial production of high efficiency SHJ solar cells, high quality n-type Czochralski silicon is required because the low-temperature process cannot provide the benefits of gettering and bulk hydrogenation.[10][43] Photons absorbed outside the substrate do not contribute to photocurrent and constitute losses in quantum efficiency.

Intrinsic amorphous silicon is deposited onto both sides of the substrate using PECVD from a mixture of silane (SiH₄) and hydrogen (H₂), forming the heterojunction and passivating the surface. The buffer layer must be sufficiently thick to provide adequate passivation, however must be thin enough to not significantly impede carrier transport or absorb light. The buffer layer is on the order of 1–10 nm thick. Despite similarities between the buffer layer structure and Metal–Insulator–Semiconductor (MIS) solar cells, SHJ do not necessarily rely on quantum tunnelling for carrier transport through the low-conductivity buffer layer; carrier diffusion is also an important transport mechanism.[18][44]

The selective contacts (also referred to as the "window layers") are then similarly formed by deposition of the p- and n-type highly doped amorphous silicon layers.[45][46] Examples of dopant gases include phosphine (PH₃) for n-type and trimethylborane (B(CH₃)₃) or diborane (B₂H₆) for p-type.[47] Due to its defective nature, doped amorphous silicon (as opposed to intrinsic) cannot provide passivation to crystalline silicon; similarly epitaxial growth of any such a-Si layer causes severe detriment to passivation quality and cell efficiency and must be prevented during deposition.[48]

Recent developments in SHJ efficiency have been made by deposition of n-type nanocrystalline silicon oxide (nc-SiO_x:H) films instead of n-type amorphous silicon for the electron contact. The material commonly referred to as "nanocrystalline silicon oxide" is actually a two-phase material composed of nanoscale silicon crystals embedded in an amorphous silicon oxide matrix. The silicon oxide has a higher band gap and is more optically transparent than amorphous silicon, whereas the columnar nanocrystals enhance vertical carrier transport, thus leading to increased short circuit current density ${\displaystyle J_{SC}}$.[40] The material band gap can be tuned with varying levels of carbon dioxide during PECVD.[49] The replacement of amorphous silicon with nanocrystalline silicon/silicon oxide has already been integrated by some manufacturers on n-type, with p-type (hole contact) to follow in the near future.[21]

The dual purpose antireflection coating (ARC) and carrier transport layer, usually composed of Indium tin oxide (ITO), is sputtered onto both sides over the selective contacts. Indium tin oxide is a transparent conducting oxide (TCO) which enhances lateral conductivity of the contact surfaces without significantly impeding light transmission. This is necessary because the amorphous layers have a relatively high resistance despite their high doping levels, and so the TCO allows carriers to be transported from the selective contact to the metal electrodes.

For destructive interference antireflection properties, the TCO is deposited to the thickness required for optimum light capture at the peak of the solar spectrum (around 550 nm  ). The optimum thickness for a single-layer ARC is given by;

${\displaystyle d={\frac {\lambda }{4\eta }}}$

where ${\displaystyle d}$ is the layer thickness, ${\displaystyle \lambda }$ is the desired wavelength of minimum reflection and ${\displaystyle \eta }$ is the material's refractive index.

Depending on the refractive index of the ITO (typically ~0.9),[50] the optimum layer thickness is usually 70-80 nm. Due to thin-film interference, the ITO (a dull grey-black ceramic material) appears a vibrant blue colour at this thickness.

Due to the scarcity of indium, alternative TCOs such as aluminium-doped zinc oxide (AZO) are being researched for use in SHJ cells.[51] AZO has a much higher chemical sensitivity than ITO, which presents challenges for certain metallisation methods that require etching, such as nickel seed layer etch-backs[52] and typically has a poorer interface contact to both p- and n-type amorphous layers.[53] AZO may have long-term stability issues when cells are used in modules, which may require capping layers such as SiO_x.[54][55]

Enhancement of the optical and electronic properties of indium oxide based TCOs has been achieved through co-doping with cerium and hydrogen, which results in high electron mobility. Such films can be grown at temperatures sufficiently low to be compatible with the heat-sensitive SHJ production process.[56][55] Indium oxide doped with cerium oxide, tantalum oxide and titanium oxide have also resulted in favourable electronic properties. The process is tunable through introduction of water vapour into the sputtering chamber[49] in which hydroxyl radicals in the plasma are believed to terminate oxygen vacancies in the TCO film, leading to enhanced electron mobility and lower sheet resistance, however stability and contact resistance must be considered when using this method in SHJ cells.[57]

Through evaporation, a double-antireflection coating of magnesium fluoride (MgF₂)[58] or aluminium oxide (Al₂O₃)[50] may be used to further reduce surface reflections, however this step is not currently employed in industrial production. AZO capping layers such as SiO_x can also act as a double AR coating.[54]

The TCO layer for SHJ cells should ideally have a high work function[59] (ie. the energy difference between the Fermi level and the Vacuum level) to prevent formation of a parasitic Schottky barrier at the interface between the TCO and the p-type amorphous layer.[60][61][62] This can be partially alleviated by increasing the doping of the p-type layer, which decreases the barrier width and improves open-circuit voltage (${\displaystyle V_{OC}}$) and fill factor (${\displaystyle FF}$). However increased doping increases junction recombination, diminishing ${\displaystyle V_{OC}}$ gains. Depositing a higher work function TCO such as tungsten oxide (WO_x) or tuning the deposition parameters of ITO can also reduce the barrier height; typically the latter is used due to the preferable optical properties of ITO.[53]

A SEM-EDS image of a Cu/Sn-plated heterojunction solar cell. The colours are indicative of elements present.

Heterojunction solar cells are typically metallised (ie. fabrication of the metal contacts) in two distinct methods. Screen-printing of silver paste is common in industry as is with traditional solar cells, with a market share of over 98%.[3] However low-temperature silver paste is required for SHJ cells. These suffer major drawbacks including low grid conductivity and high silver consumption,[63][64] volatile production costs[17] or poor adhesion to the front surface.[9][63] Despite their significantly higher cost,[52] the resistivity of low-temperature silver pastes has been estimated to be 4–6 times higher than standard silver paste.[18] To compensate for lowered conductivity, low-temperature silver pastes also consume more silver than conventional silver pastes,[63] however silver consumption is trending downward as the development of screen-printing technology reduces finger linewidths.[65] Improvements in the composition of low-temperature pastes are expected to further reduce silver consumption, such as through screen-printable silver-coated copper paste. Such pastes perform comparably to conventional low-temperature pastes, with up to 30% reduction in silver consumption.[66] Silver-coated copper pastes are becoming an increasingly dominant metallisation technology amongst Chinese SHJ manufacturers into 2030, with 50% market share expected from 2024–2025.[26]

Silver nanoparticle ink can be deposited onto a SHJ solar cell using inkjet printing, or through contact deposition with a hollow glass capillary. Inkjet deposition has been reported to reduce silver consumption from 200 mg per cell to less than 10 mg per cell compared with traditional silver paste screen printing. Further reductions are possible with capillary deposition (known as "FlexTrail" as the capillary is flexible and trails across the wafer surface) leading to as little as 3 mg of silver deposited.[66] Such a large reduction in silver has implications for the grid design to compensate for lower conduction, namely using a busbar-less design.

An alternative to printed electrodes uses electroplated copper. The conductivity of electroplated copper is similar to that of bulk copper.[67] This has potential to increase the SHJ cell current density through decreasing grid resistance. Improved feature geometry can also be achieved. However industrial production is challenging as electroplating requires selective patterning using a sacrificial inkjet-printed or photolithographically-derived mask.[64][68] As a result, electroplated SHJ cells are not currently manufactured commercially. Copper plated directly to the ITO also suffers from adhesion issues. Therefore it is usually necessary to first deposit a thin (~1μm) seed layer of nickel through sputtering or electrodeposition.[52][63][69][70]

SHJ temperature sensitivity has further implications for cell interconnection when manufacturing SHJ-based solar panels. High temperatures involved in soldering must be carefully controlled to avoid degradation of the cell passivation. Low temperature pastes have also suffered from weak adhesion to interconnecting wires or ribbons, which have consequences for module durability. Optimisation of these pastes and infrared soldering parameters, as well as careful selection of solder alloys, has led to increased success of interconnection processes on standard industrial equipment.[71]

Heterojunction–Perovskite tandem structures have been fabricated, with some research groups reporting a power conversion efficiency exceeding the 29.43% Shockley–Queisser limit for crystalline silicon. This feat has been achieved in both monolithic and 4-terminal cell configurations.[72][23] In such devices, in order to reduce thermalisation losses, the wide bandgap Perovskite top cell absorbs high energy photons whilst the SHJ bottom cell absorbs lower energy photons. In a bifacial configuration, the bottom cell can also accept light from the rear surface.

In 2017, tandem solar cells using a SHJ bottom cell and Group III–V semiconductor top cells were fabricated with power conversion efficiencies of 32.8% and 35.9% for 2- and 3-junction non-monolithic stacks respectively.[73]

In April 2023, the efficiency record for SHJ tandems was set at 33.2% by KAUST Solar Center using a Perovskite top cell in a monolithic configuration.[23] This is the highest efficiency recorded for a 2-junction solar cell.

Aside from the typical c-Si/a-Si:H structure, various groups have successfully produced passivated contact silicon heterojunction solar cells using novel semiconducting materials, such as between c-Si/SiO_x,[47] c-Si/MoO_x[74][75] and c-Si/poly-Si or c-Si/SiO_x/poly-Si (POLO; polycrystalline silicon on oxide).[76][77] Hybrid inorganic–organic heterojunction solar cells have been produced using n-type silicon coated with polyaniline emeraldine base.[78] Heterojunction solar cells have also been produced on multicrystalline silicon absorber substrates.[79]

Band gap energies of semiconductors commonly used in heterostructures

Material	Band gap energy; Eg (eV)	Notes	Reference
c-Si	~1.12	Typical figure measured at 298 K	[1]
a-Si:H	~1.7	Compared to c-Si, the wider band gap is attributable mostly to the high (~10% in SHJ solar cells) hydrogen content of amorphous silicon.[80] The band gap energy is affected by the crystalline fraction and hydrogen content of the amorphous network, and is dependent on the method in which the thin film is prepared. A higher ratio of H2:SiH4 during deposition increases the band gap energy.[81]	[82]
SiOx:H	~1.4–3.3	Band gap increases as oxygen content ${\displaystyle x}$ increases where ${\displaystyle 0<x<2}$. A higher ratio of CO2:SiH4 during deposition increases the band gap energy.[83]	[84]
MoOx	~3		[75]

Heterojunction solar cells are compatible with IBC technology, ie. the cell metallisation is entirely on the back surface. A Heterojunction IBC cell is often abbreviated to HBC. A HBC structure has several advantages over conventional SHJ cells; the major advantage is the elimination of shading from the front grid, which improves light capture and hence short circuit current density ${\displaystyle J_{SC}}$. Compared to PERC, conventional SHJ cells often suffer from poor ${\displaystyle J_{SC}}$ with values rarely exceeding 40 mA/cm², as some light is parasitically absorbed in the front amorphous silicon layers due to its high absorption coefficient.[40] By removing the need for the front metal contact, as well as the front amorphous silicon contact, ${\displaystyle J_{SC}}$ can be recovered. As such, HBC cells have potential for high efficiencies; notably a long-standing world record heterojunction cell employed a HBC structure, at 26.7% efficiency fabricated by Kaneka with a ${\displaystyle J_{SC}}$ of 42.65 mA/cm².[20][85] Despite HBC's high efficiency, double-sided cells are mainstream in industrial production due to their relatively simple manufacturing process.[49] However, HBC cells may find specialised applications such as in vehicle-integrated PV systems where there are significant area constraints.[86]

HBC cells are fabricated by localised doping of the rear side, in an alternating pattern of p- and n-type areas in an interdigitated pattern. The front side does not require a specific doping profile.[87]

Defects are sites at which charge carriers can inadvertently become "trapped", making them more likely to recombine through the Shockley-Read-Hall method (SRH Recombination). They are most likely to exist at interfaces (surface recombination), at crystal grain boundaries and dislocations, or at impurities. To prevent losses in efficiency, defects must be passivated (ie. become chemically and electrically neutral). Generally this occurs through bonding of the defect interface with interstitial hydrogen. In SHJ cells, hydrogenated intrinsic amorphous silicon is very effective at passivating defects existing at the absorber surface.

Understanding the behaviour of defects, and how they interact with hydrogen over time and in manufacturing processes, is crucial for maintaining the stability and performance of SHJ solar cells.

The behaviour of light-sensitive defect passivation in amorphous silicon networks has been a topic of study since the discovery of the Staebler–Wronski effect in 1977.[88] Staebler and Wronski found a gradual decrease in photoconductivity and dark conductivity of amorphous silicon thin films upon exposure to light for several hours. This effect is reversible upon dark annealing at temperatures above 150 °C and is a common example of reversible Light-induced Degradation (LID) in hydrogenated amorphous silicon devices. The introduction of new band gap states, causing a decrease in the carrier lifetime, was proposed to be the mechanism behind the degradation. Subsequent studies have explored the role of hydrogen migration and metastable hydrogen-trapping defects in the Staebler–Wronski effect.[89]

Amongst many variables, the kinetics and extent of the Staebler–Wronski effect is dependent on crystallite grain size in the thin film[90] and the light soaking illuminance.[91]

Some amorphous silicon devices can also observe the opposite effect through LID, such as the increase in ${\displaystyle V_{OC}}$ observed in amorphous silicon solar cells[92][93] and notably SHJ solar cells[94] upon light soaking. Kobayashi, et al. (2016) proposes that this is due to the shifting of the Fermi level of the intrinsic buffer layer closer to the band edges when in contact with the doped amorphous silicon selective contacts,[94] noting that a similar reversal of the Staebler–Wronski effect was observed by Scuto et al. (2015) when hydrogenated a-Si photovoltaic devices were light-soaked under reverse bias.[95]

Deliberate annealing of heterojunction cells in an industrial post-processing step can improve lifetimes and decrease surface recombination velocity. It has been suggested that thermal annealing causes interstitial hydrogen to diffuse closer to the heterointerface, allowing greater saturation of dangling bond defects.[96] Such a process may be enhanced using illumination during annealing, however this can cause degradation before the improvement in carrier lifetimes is achieved, and thus requires careful optimisation in a commercial setting.[97] Illuminated annealing at high temperatures is instrumental in the Advanced Hydrogenation Process (AHP), an inline technique for defect mitigation developed by UNSW.

The Boron–Oxygen complex LID defect is a pervasive problem with the efficiency and stability of cheap p-type wafers and a major reason that n-type is preferred for SHJ substrates. Stabilising wafers against B–O LID using the Advanced Hydrogenation Process has had variable success and reliability issues.[37] Therefore gallium has been proposed as an economically feasible alternative p-type dopant for use in SHJ absorbers.[98][36] Gallium doped cells have potential for higher stability and lower defect density than boron, with research groups achieving ${\displaystyle V_{OC}}$ exceeding 730 mV on gallium-doped p-type SHJ.[37]

Solar modules are exposed to various stressors when deployed in outdoor installations, including moisture, thermal cycling and ultraviolet light. Solar modules may be expected to be in service for decades, and these factors can reduce module lifespan if unaccounted for. The mechanisms of degradation include efficiency loss in the cell itself from cracking, gradual corrosion or defect activation; delamination of the module layers; UV degradation of the cell or lamination; encapsulant embrittlement or discolouration; and failure of the metal conductors (fingers, busbars and tabbing).[99]

Potential-induced degradation (PID) refers to degradation caused by high voltage stress in solar modules. It is one of the primary mechanisms of solar module degradation.[100] Strings of modules in series can accumulate up to 1000 V in a photovoltaic system, and such a potential difference can be present over a small distance between the solar cells and a grounded module frame, causing leakage currents. PID is primarily an electrochemical process causing corrosion[101] and ion migration[102] in a solar module and cells, facilitated by moisture ingress and surface contamination.[103][104] This leads to reduction in the efficiency and lifespan of a PV system.

PID has been observed in all types of crystalline silicon solar cells, as well as thin-film solar cells, CIGS cells and CdTe cells. In research, PID can be replicated in accelerated aging tests by applying high bias voltages to a sample module, especially in an environmental chamber. In SHJ cells, PID is mostly characterised by the reduction in ${\displaystyle J_{SC}}$ caused by optical losses, and unlike the PID observed in other module technologies, the PID is mostly irreversible in SHJ modules with only a small recovery from applying the opposite bias. This indicates that some component of the PID occurs through a different mechanism in SHJ modules. It has been suggested that optical losses are caused by indium metal precipitating in the TCO. Degraded modules have also measured high concentrations of sodium ions deeper in the cell, which is consistent with PID caused from negative bias.[100]

Encapsulants are thermoplastic materials used to encase solar cells in modules for stability. In the lamination process, the cells are sandwiched between the encapsulant film and it is melted. Traditionally, the cheap copolymer Ethylene-vinyl acetate (EVA) has been used in crystalline silicon modules as encapsulant.[105] After long duration exposure to moisture, EVA can hydrolyse and leech acetic acid[106] with the potential to corrode the metal terminals[107] or surface[108] of a solar cell.

Non-bifacial modules are composed of a textured glass front and UV-stabilised polymer (commonly polyvinyl fluoride) backsheet, whereas bifacial modules are more likely to be glass–glass.[28] The polymer backsheet, despite being more permeable to moisture ingress than glass–glass modules (which facilitates hydrolysis of EVA), is allegedly "breathable" to acetic acid and does not allow it to build up. As SHJ-based modules are more likely to be bifacial glass–glass, the risk of acetic acid buildup is claimed to be greater;[105] however manufacturers have found the impermeability of glass–glass modules is generally sufficient to prevent EVA degradation, allowing modules to pass accelerated aging tests. Some studies have also found that glass–glass construction reduces the extent of degradation in EVA-encapsulated modules against glass–backsheet.[109]

Additionally, ITO used in SHJ cells may be susceptible to acetic acid etching, causing ${\displaystyle V_{OC}}$ loss.[108][110] Despite the higher cost, acetate-free encapsulants such as polyolefin elastomers (POE) show reduced degradation after damp-heat testing in comparison to EVA.[105][109]

Encapsulant-free module designs have also been developed with potential for reduced long term degradation and CO₂ footprint. However reflection losses may arise from the lack of optical coupling between the front glass and the cell that encapsulant provides.[111]

POE has higher resistance to water ingress compared to EVA, and hence prevents PID and other moisture-related issues. However, the lamination time is longer, and the adhesion between POE and the cell or glass is inferior to EVA. Therefore, POE is increasingly used as the centre layer in a three-layer coextruded polymer encapsulant with EVA, known as EPE (EVA–POE–EVA) which entails the benefits of both polymers.[112][113]