Gathers electrons
Many narrow metal fingers collect electrons from across the cell's surface.
FOUNDATIONS
Light, photons and photovoltaic bandgaps
Light is electromagnetic radiation. A solar cell responds to the energy carried by its photons, not simply to whether the light looks bright to the human eye.
Light can be described as an electromagnetic wave with wavelength λ and frequency ν, related by c = λν. It also arrives in discrete packets called photons. The energy of one photon is Eph = hν = hc/λ, so shorter-wavelength light has higher-energy photons. Sunlight at Earth's surface extends beyond the visible rainbow into ultraviolet and infrared wavelengths.
A semiconductor's bandgap Eg is the energy separation between its valence and conduction bands. A photon can generate an electron–hole pair only if it is absorbed and Eph ≥ Eg. A lower-energy photon normally passes through a simple absorber; a much higher-energy photon can be absorbed, but energy above the bandgap is rapidly lost as heat while the carrier relaxes towards the band edge.
Photon energy: Eph = hν = hc/λ
Absorption threshold: Eph ≥ Eg
Approximate cut-off: λg [nm] ≈ 1240 / Eg [eV]
The bandgap therefore sets a material's longest usable wavelength, but it does not alone determine how strongly or how deeply light is absorbed. Optical thickness, direct or indirect transitions, reflection and material quality also matter. Intensity controls how many photons arrive per unit area and time; wavelength controls the energy of each photon.
A single-junction cell cannot use every photon energy perfectly. A lower bandgap captures more of the spectrum but generally limits voltage, while a higher bandgap can support more voltage but misses longer wavelengths. Tandem and multijunction cells stack different bandgaps so high-energy photons are absorbed first and lower-energy photons can reach a narrower-bandgap cell beneath.
Scientific references: NASA electromagnetic-spectrum fundamentals, U.S. DOE photovoltaic-cell basics, and DOE wavelength and PV-loss overview.
What are photovoltaics?
Photovoltaics (PV) convert sunlight directly into electrical power.
Each photon carries energy E = hν. If an absorbed photon has at least the semiconductor bandgap energy (hν ≥ Eg), it can excite an electron from the valence band into the conduction band, leaving a mobile, positively charged “hole”. This process generates an electron–hole pair.
In a conventional crystalline-silicon cell, the junction between n-type and p-type silicon forms a charge-depleted region with a built-in electric field. Carriers generated in, or diffusing into, this region are separated: electrons are driven towards the n-type side and holes towards the p-type side. Carriers generated in the neutral bulk must reach the junction or a selective contact before they recombine. Metal contacts collect the separated carriers, creating a negative n terminal and a positive p terminal under illumination. With a load connected, electrons travel through the external circuit from n to p; conventional current is defined in the opposite direction.
Electrical power is P = IV. Reflection and transmission reduce absorption, excess photon energy is lost through thermalisation, and carrier recombination and electrical resistance reduce the available current and voltage. A PV cell converts energy—it does not store it—and differs from solar-thermal technology, which first converts sunlight into heat.
2 · 1955–1956
Single-junction gallium arsenide (GaAs)
GaAs was the next major high-performance junction photovoltaic investigated after the practical crystalline-silicon cell.
Interest in GaAs photovoltaics emerged almost immediately after the 1954 silicon breakthrough. Gremmelmaier reported a GaAs photocell in 1955, and RCA researchers described experimental GaAs p–n junction cells in 1956. The first devices were small and only a few percent efficient because suitable single crystals were difficult to grow and control.
GaAs has a direct bandgap of about 1.42 eV at room temperature and absorbs sunlight much more strongly than silicon, so its active absorber can be only a few micrometres thick. Later III–V epitaxy made it possible to add lattice-matched, wider-bandgap window and back-surface-field layers that reduce recombination while allowing light and useful charge to move in the intended directions.
These cells became especially important for spacecraft because they combine high efficiency with good resistance to particle radiation. Their specialised crystal substrates, epitaxial growth and processing remain expensive, so GaAs is used primarily where performance per unit area or mass matters more than minimum cost.
Historical and device references: NASA GaAs history, representative GaAs device stack, and NREL III–V research overview.
3 · 1958
Organic photovoltaics (OPV)
Organic photovoltaics introduced carbon-based semiconducting molecules as light absorbers and charge-transport materials.
David Kearns and Melvin Calvin reported the photovoltaic effect in laminated organic systems in 1958. Their magnesium-phthalocyanine device produced about 200 mV but extremely little power. It was primarily a scientific demonstration inspired by questions about light-driven charge separation in photosynthesis—not a practical power generator.
The field advanced when C. W. Tang's 1986 donor–acceptor bilayer showed that an interface between two organic semiconductors could separate tightly bound photoexcitations efficiently. In 1995, Yu and colleagues blended donor and acceptor materials into a bulk heterojunction, creating a nanoscale network with many charge-separating interfaces. That concept underpins most modern OPV cells.
Modern OPV can be lightweight, flexible, coloured, semitransparent and compatible with coating or printing. Its active layer absorbs light as molecular excitons; these reach a donor–acceptor interface, split into electrons and holes, and travel through selective layers to opposite electrodes. Controlling nanoscale morphology, non-radiative recombination and long-term stability remains central to development.
Historical and device references: Kearns and Calvin (1958), Tang's donor–acceptor bilayer (1986), bulk-heterojunction milestone (1995), and NREL OPV overview.
4 · 1963
Large-area Cu₂S/CdS thin-film photovoltaics
Cu₂S/CdS helped turn the thin-film heterojunction from a laboratory effect into a deliberately engineered, lightweight solar-energy converter.
The technology's roots predate the 1960s: photovoltaic Cu/CdS structures were reduced to practice in 1954–1955. The 1963 NASA-sponsored large-area programme is used here as the development milestone because it addressed controlled vacuum deposition, junction formation and useful device area rather than the first observation.
In the representative front-wall architecture, light passes through a thin SiO antireflection layer and a sparse metal grid into p-type Cu₂S. That narrow-bandgap layer is the principal absorber. It meets wider-bandgap n-type CdS, which acts as the window/body of the heterojunction. A zinc layer and copper foil complete the rear contact and mechanically support the very thin semiconductor stack.
Large-area cells reached roughly 4–8% efficiency by the mid-1960s and demonstrated low material use and high specific power. Their Achilles' heel was materials stability: copper migration and changes among copper-sulfide phases altered the junction and contact properties. The programme nevertheless established methods and questions that shaped later CdTe and chalcopyrite thin-film photovoltaics.
Historical and device references: 1963 NASA large-area thin-film study, 1966 CdS-cell history and performance review, and representative Cu foil/Zn/CdS/Cu₂S/grid/SiO structure.
5 · 1972
Cadmium telluride thin-film photovoltaics
The CdS/CdTe heterojunction established a practical all-thin-film route based on strongly absorbing polycrystalline compound semiconductors.
Cadmium compounds had shown photovoltaic behaviour earlier, but Dieter Bonnet and Helmut Rabenhorst reported the landmark all-thin-film p-CdTe/n-CdS heterojunction cell in 1972. Its roughly 6% efficiency demonstrated that useful junctions could be deposited as thin films rather than cut from high-quality single-crystal wafers.
CdTe has a direct bandgap near 1.45 eV and absorbs the solar spectrum strongly, so only a few micrometres of absorber are required. The classic superstrate design deposits a transparent conductor, thin n-type CdS window, p-type CdTe absorber and rear contact onto glass. Chloride activation, grain-boundary control and better contacts progressively improved voltage, current and manufacturability.
Modern commercial devices retain the glass-first thin-film concept but may replace or supplement CdS with wider-bandgap and alloyed front layers such as MgZnO and CdSeTe. CdTe became the leading commercial thin-film alternative to crystalline silicon, particularly for utility-scale modules. Long service life depends on stable back contacts, strong encapsulation and controlled manufacturing and recycling of cadmium-containing material.
Historical and device references: NREL CdTe history, classic glass/TCO/CdS/CdTe stack, DOE modern CdTe overview, and modern layer variants.
6 · 1976
Hydrogenated amorphous-silicon photovoltaics
Hydrogenated amorphous silicon made it possible to deposit a complete silicon solar cell as thin semiconductor films over large areas.
David Carlson and Christopher Wronski reported the first hydrogenated amorphous-silicon solar cell at RCA Laboratories in 1976. Their glow-discharge-deposited device was about one micrometre thick, used a p–i–n structure and reached 2.4% efficiency—modest performance, but a striking departure from sliced crystalline-silicon wafers.
Amorphous silicon lacks the repeating crystal lattice of wafer silicon. Hydrogen bonds to many otherwise electrically active dangling bonds, allowing useful doping and photoconductivity. Because carrier diffusion is limited by disorder, the p and n regions are kept very thin and a comparatively thick intrinsic absorber sits between them. The field across that intrinsic layer drives electrons and holes to opposite contacts.
Low-temperature plasma-enhanced chemical-vapour deposition enabled large-area glass modules, flexible stainless-steel products and small consumer devices. Textured transparent conductors and reflective rear contacts improved light trapping, while a-Si/a-SiGe multijunction stacks extended spectral absorption. A defining challenge is the Staebler–Wronski effect: illumination creates metastable defects and lowers initial performance, so products are characterised by stabilised rather than initial output.
Historical and device references: Carlson and Wronski (1976), NREL superstrate p–i–n structure, light-induced degradation research, and roll-to-roll multijunction manufacture.
7 · August 1976
Copper indium gallium diselenide photovoltaics
The first CuInSe₂ thin-film heterojunction opened the path to composition-tunable chalcopyrite absorbers now known as CIGS.
Lawrence Kazmerski, Frederick White and Gary Morgan reported thin-film p-CuInSe₂/n-CdS heterojunction cells in August 1976, with efficiencies of roughly 4–5% under their test illumination. The work showed that a ternary I–III–VI₂ compound could be deposited as a polycrystalline film and coupled to a wider-bandgap n-type partner.
Adding gallium produced Cu(In,Ga)Se₂—CIGS—and allowed researchers to tune and grade the absorber bandgap. A conventional substrate cell is built on soda-lime glass coated with molybdenum, followed by p-type CIGS, a very thin n-type buffer such as CdS, intrinsic zinc oxide, a conductive transparent oxide and a front grid. Sodium from the glass and later controlled alkali treatments became important tools for improving grain growth, carrier density and interface quality.
CIGS absorbs light strongly and can be deposited by co-evaporation or by reacting metal precursors with selenium. It can also be made on flexible metal or polymer foils. Its compositional freedom enables high efficiency but makes uniform control of copper, indium, gallium, selenium, alkalis, interfaces and large-area processing demanding.
Historical and device references: Kazmerski, White and Morgan (1976), NREL standard CIGS configuration, DOE CIGS processing overview, and representative device variants.
8 · 1988
Monolithic GaInP/GaAs tandem photovoltaics
A lattice-matched GaInP top cell, GaAs tunnel junction and GaAs bottom cell demonstrated how multiple junctions could share one epitaxial crystal stack and one external circuit.
NREL's programme began in 1984, obtained a foundational tandem patent in 1987 and reported a 21.8% two-junction GaInP₂/GaAs device in 1988. The milestone belongs to the 1980s even though rapid refinement continued: a closely related monolithic device reached 27.3% in 1990.
The wide-bandgap GaInP top junction absorbs higher-energy visible photons while transmitting lower-energy light to the 1.42 eV GaAs bottom junction. Their voltages add. A heavily doped GaAs tunnel diode forms a low-loss internal series connection, so the same current must pass through both subcells and spectral current matching is critical.
Lattice matching to GaAs allowed the layers to be grown coherently by metal-organic vapour-phase epitaxy, limiting dislocations that would otherwise accelerate recombination. This architecture became the foundation for higher-order III–V multijunction cells used where efficiency and specific power justify complex epitaxial growth, especially in space and concentrator systems.
Historical and device references: NREL's 1984–1988 programme retrospective, 1990 peer-reviewed monolithic device, and current-matched GaInP/GaAs tandem patent.
9 · 1991
Dye-sensitized solar cells
A molecular dye and a porous oxide separated light absorption from charge transport, creating a new kind of photoelectrochemical solar cell.
In 1991, Brian O’Regan and Michael Grätzel reported a cell built from an approximately 10 µm film of nanocrystalline TiO₂ coated with a light-absorbing ruthenium dye. Coupling the enormous internal surface area of that porous film to an iodide/triiodide electrolyte produced 7.1–7.9% efficiency under simulated sunlight and 12% under diffuse daylight in the original report.
The mechanism differs from a conventional semiconductor p–n junction. A dye molecule absorbs a photon and injects an excited electron into TiO₂. That electron moves through the oxide to the transparent front electrode and around the external circuit. The oxidised dye accepts an electron from the redox mediator, while the mediator is regenerated at a catalytic counter electrode, completing the cycle.
Dye-sensitized cells can be semitransparent, coloured and effective under diffuse or indoor illumination. Research has expanded from ruthenium complexes and iodide electrolytes to metal-free dyes, cobalt or copper redox couples and solid-state hole conductors. Key engineering challenges include electrolyte sealing, long-term chemical stability, voltage loss and interfacial recombination.
Historical and device references: O’Regan and Grätzel (1991), DOE dye-sensitized-cell principle, representative interfaces and redox chemistry, and the DOE sensitized-solar-cell roadmap.
10 · 1998
Quantum-dot photovoltaics
Semiconductor crystals only a few nanometres across made the absorber’s optical and electronic properties tunable through particle size.
In 1998, Arie Zaban, Olga Micic, Brian Gregg and Arthur Nozik reported the photosensitization of porous nanocrystalline TiO₂ with indium-phosphide quantum dots. Approximately 6.5 nm InP particles penetrated and adsorbed throughout the porous oxide, produced visible-light photoconductivity and injected electrons into TiO₂ in a photoelectrochemical cell containing a redox electrolyte and platinum counter electrode.
Quantum confinement is the defining idea. When a semiconductor crystal becomes comparable in size to the carrier’s characteristic length scale, its allowed energy states become discrete and its effective bandgap depends on particle size. A quantum dot can absorb a photon to create a bound electron–hole pair; interfacial energetics must then separate that pair before it recombines. In the original sensitized architecture, the electron enters TiO₂ and the electrolyte removes the remaining hole.
The field later branched into solid-state colloidal-quantum-dot films, first reported as primary thin-film absorbers in 2005, and into quantum-dot junctions, tandems and perovskite nanocrystals. Solution synthesis and size-tunable absorption—including into the infrared—are attractive, but surface traps, ligand-controlled transport, voltage deficit, oxidation, material toxicity and scalable uniformity remain central challenges.
Historical and device references: Zaban et al. (1998), NREL quantum-dot cell configurations, colloidal-quantum-dot architecture history, and the NREL/DOE review of modern CQD photovoltaics.
11 · 2009–2012
Metal-halide perovskite photovoltaics
A strongly absorbing, solution-processable semiconductor advanced from a short-lived liquid-junction sensitizer to an efficient solid-state solar cell in only three years.
In 2009, Akihiro Kojima, Kenjiro Teshima, Yasuo Shirai and Tsutomu Miyasaka used CH₃NH₃PbBr₃ and CH₃NH₃PbI₃ nanocrystals to sensitize mesoporous TiO₂. The iodide cell responded to wavelengths up to 800 nm and reached 3.8% efficiency, while the bromide cell produced 0.96 V. Both used a liquid electrolyte, which dissolved the perovskite and caused rapid degradation.
The decisive step came in 2012. Hui-Seon Kim, Nam-Gyu Park and collaborators replaced the liquid electrolyte with the solid molecular hole conductor spiro-OMeTAD. Their submicrometre mesoscopic CH₃NH₃PbI₃/TiO₂ cell reached 9.7% and was substantially more stable. In parallel, meso-superstructured cells showed that perovskite was not merely a surface dye: it could transport both electrons and holes and operate as the principal semiconductor absorber.
Modern cells place an ABX₃ metal-halide perovskite film between electron- and hole-selective contacts. Composition controls bandgap, crystallisation and stability, enabling lightweight coating and high-efficiency silicon tandems. Remaining challenges include moisture, heat and light degradation; ion migration; scalable defect control; lead containment and recycling; and proving long operational lifetimes in modules.
Historical and device references: Kojima et al. (2009), Kim et al. (2012), Lee et al. (2012), and representative modern n–i–p and p–i–n stacks.
12 · 2015
Perovskite–silicon tandem photovoltaics
A wide-bandgap perovskite top cell and a lower-bandgap silicon bottom cell divide the solar spectrum between two junctions, reducing the fundamental losses of either absorber alone.
In work first published online at the end of 2014 and appearing in 2015, Colin Bailie and colleagues made semitransparent perovskite cells with silver-nanowire electrodes and mechanically stacked them over silicon and CIGS cells. The four-terminal subcells operated independently and produced a net performance gain over the bottom cells, establishing perovskite as a practical, tuneable top-cell material.
In March 2015, Jonathan Mailoa and colleagues reported the first monolithic two-terminal perovskite/silicon multijunction. A silicon interband tunnel junction connected the subcells in series and allowed majority carriers to recombine internally. The 1 cm² device produced up to 1.65 V open-circuit voltage and 13.7% stable efficiency, although its perovskite top cell limited the series current.
The principle is spectral division: the wide-gap top junction converts higher-energy photons at high voltage, while longer-wavelength light passes into silicon. Two-terminal tandems need current matching, a transparent low-resistance interconnection and mutually compatible processing. Four-terminal tandems relax electrical matching but add transparent electrodes and optical loss. Durability, textured interfaces, large-area coating, lead management and low-cost interconnection remain central commercial challenges.
Historical and device references: Bailie et al. (2015), Mailoa et al. (2015), the NREL tandem-photovoltaics roadmap, and NREL hybrid-tandem research.
13 · 2016
All-perovskite tandem photovoltaics
Two compositionally different metal-halide perovskites divide the solar spectrum without relying on a crystalline-silicon or III–V bottom cell.
The enabling material advance came from mixed tin–lead perovskites. Replacing roughly half the lead with tin produces strong bandgap bowing, allowing a narrow gap near 1.2 eV. This made it possible to pair a low-gap perovskite rear junction with a bromide-containing, wide-gap lead-perovskite front junction near 1.8 eV.
In 2016, Giles Eperon and colleagues demonstrated the first efficient two- and four-terminal perovskite/perovskite tandems with deliberately matched bandgaps. Their monolithic two-terminal cell connected the subcells through a solvent-resistant SnO₂/ZTO/ITO recombination stack and reached 17.0% stabilized efficiency with 1.66 V open-circuit voltage. A mechanically stacked four-terminal version reached 20.3% stabilized efficiency.
All-perovskite tandems retain the thin-film, low-temperature and potentially lightweight character of perovskite PV while offering independent control of both absorber bandgaps. Their central challenges are suppressing Sn²⁺ oxidation and self-doping in the narrow-gap cell, preventing light-induced halide segregation in the wide-gap cell, forming a transparent low-loss interconnection, matching subcell currents and protecting both soft ionic absorbers during processing and operation.
Historical and device references: Eperon et al. (2016), its supporting device methods, the NREL review of mixed Sn–Pb perovskites, and the wide-bandgap tandem review.
14 · 2018
Perovskite/colloidal-quantum-dot tandem photovoltaics
A visible-light metal-halide perovskite front cell and a near-infrared lead-sulfide quantum-dot rear cell combine two solution-processable, bandgap-tunable absorber families.
Colloidal quantum dots offered something the first perovskite tandems could not easily provide: a rear-cell bandgap tuneable below 1.1 eV simply by changing nanocrystal size. PbS quantum dots can therefore absorb red and near-infrared photons transmitted through a roughly 1.55 eV perovskite front cell, while both absorbers can be deposited from inks at comparatively low temperatures.
In March 2018, Arfa Karani, Le Yang and colleagues reported the first prototype solution-processed monolithic perovskite/CQD tandem. Its ITO/TiO₂/MAPbI₃/PTAA front cell was joined to a ZnO/PbS-CQD rear cell through a MoOₓ/ultrathin-Ag recombination and solvent-barrier stack. The best efficiency in the prototype series was only 1.01%; a separate device produced 1.17 V open-circuit voltage, demonstrating subcell voltage addition and showing that the many sequentially coated interfaces could remain intact.
The work also highlighted radiative coupling: light emitted by recombination in the perovskite can be reabsorbed by the lower-gap CQDs. Modelling predicted a 43% ideal limit for 1.55/1.0 eV absorbers and a 29.7% device using then state-of-the-art subcells. Real devices remain constrained by current matching, PbS surface traps and ligand transport, parasitic interconnection absorption, solvent compatibility, oxidation and the stability of both lead-containing absorbers.
Historical and device references: the open-access Karani et al. (2018) article, the University of Cambridge repository copy, and a review of quantum-dot tandem cells.
15 · 2019
Flexible monolithic perovskite/CIGS tandem photovoltaics
A wide-gap perovskite top cell and a lower-gap Cu(In,Ga)Se₂ bottom cell divide the spectrum inside one lightweight, series-connected thin-film device.
Perovskite and CIGS are complementary thin-film absorbers. The perovskite converts much of the visible spectrum, while the compositionally tuneable CIGS junction uses longer wavelengths transmitted through the top cell. Earlier four-terminal demonstrations mechanically combined independently wired cells—including a 24.6% result announced in 2018—but retained duplicated electrodes and separate electrical outputs.
In 2019, Fan Fu and colleagues reported a proof-of-concept two-terminal monolithic tandem fabricated directly on a 30 µm polyimide foil. Its complete sequence was polyimide/Mo/CIGS/CdS/i-ZnO/AZO/PTAA/MAPbI₃/PCBM/ZnO nanoparticles/AZO/Ni-Al/MgF₂. The tandem sustained 13.2% efficiency and exceeded 1.75 V open-circuit voltage, showing voltage addition in a flexible stack assembled on a rough CIGS bottom cell.
The monolithic arrangement removes an electrode and connects the two junctions through the i-ZnO/AZO/PTAA recombination region. It also makes the subcells current-matched: the lower photocurrent limits the complete device. Conformal coverage over micrometre-scale CIGS texture, transparent-contact absorption, interfacial recombination, perovskite stability and mechanical fatigue therefore became central engineering problems for a lightweight tandem platform.
Historical and device references: the Fu et al. (2019) open preprint, the full device report and stack description, and imec's 2018 four-terminal tandem announcement.
16 · 2020
High-efficiency monolithic all-perovskite triple-junction photovoltaics
Three compositionally tuned perovskite junctions divide sunlight into high-, middle- and lower-energy spectral bands within one internally series-connected device.
The first monolithic all-perovskite triple-junction proof of concept appeared in February 2019. It combined approximately 1.94, 1.55 and 1.34 eV absorbers and demonstrated voltage addition up to 2.83 V, but reached only 6.7% efficiency because the rear junction severely limited current and the many solution-processed interfaces added optical and electrical losses.
In October 2020, Junke Wang and colleagues established the next practical architecture using a universal sequential two-step process for 1.73, 1.57 and 1.23 eV perovskites. Two repeated C₆₀/spatial-ALD SnO₂/approximately 1 nm Au/PEDOT:PSS interconnections protected completed subcells from later solvents while recombining adjacent charges. The monolithic device achieved 16.8% efficiency, 2.78 V open-circuit voltage and a fill factor of 0.81; stabilized output reached 16.9%.
Every series-connected subcell must carry the same current. The measured photocurrents were 8.2, 8.9 and 7.6 mA cm−2, so the 1.23 eV rear cell constrained the entire stack. Later research therefore focused on wider-gap top absorbers, current matching, suppressing light-induced halide segregation, thinner low-loss interconnections, tin oxidation control and long-term operational stability.
Historical and device references: the open-access Wang et al. (2020) triple-junction article, the McMeekin et al. (2019) first proof of concept, and a technical review of monolithic perovskite triple junctions.
17 · 2022
High-efficiency monolithic perovskite–organic tandem photovoltaics
A wide-bandgap perovskite front cell and a narrow-gap organic back cell divide the solar spectrum while retaining a thin-film, low-temperature device architecture.
Perovskite/polymer monolithic tandems were first demonstrated in 2015, and reproducible devices exceeded 20% in 2020. The decisive post-2020 advance arrived in April 2022, when Kai Oliver Brinkmann and colleagues combined a 1.85 eV FA₀.₈Cs₀.₂Pb(I₀.₅Br₀.₅)₃ perovskite with a 1.33 eV PM6:Y6:PC₆₁BM organic bulk heterojunction. The perovskite captures higher-energy visible light; transmitted red and near-infrared photons reach the organic subcell.
The central innovation was an approximately 1.5 nm, oxygen-deficient InOx film grown by atomic-layer deposition between SnOx and MoOx. It removed the roughly 0.6 eV transport barrier found at a direct SnOx/MoOx junction while avoiding the near-infrared absorption caused by an ultrathin metal interconnect. Electrons from the perovskite subcell can therefore recombine efficiently with holes from the organic subcell.
The two subcells each supplied an EQE-integrated current density of 14.1 mA cm−2. Their voltages added to 2.15 V, enabling an 80% fill factor and 24.0% stabilized efficiency; Fraunhofer ISE CalLab certified 23.1%. The architecture showed why complementary absorbers are not sufficient by themselves: optical transmission, interface selectivity, solvent protection and an ohmic recombination contact must all work together.
Historical and device references: the 2022 Nature article, its open author manuscript and methods, and the 2015 first full-solution perovskite/polymer tandem report.
18 · 2024
High-performance monolithic perovskite–perovskite–silicon triple-junction photovoltaics
Two compositionally different perovskite cells divide the visible spectrum before crystalline silicon collects the remaining near-infrared light.
Perovskite–perovskite–silicon triple junctions were first demonstrated as proof-of-concept devices in 2018. A 2022 monolithic device then exceeded 20% efficiency and reached 2.74 V. The next major architecture milestone, submitted in October 2023 and published in February 2024 by Hu and colleagues, improved the difficult middle cell and raised this technology to 24.4% efficiency.
The device combines an approximately 1.84 eV Cs₀.₂FA₀.₈Pb(I₀.₅Br₀.₅)₃ top absorber, an approximately 1.52 eV FAPbI₃ middle absorber and an approximately 1.1 eV crystalline-silicon bottom cell. Vacuum-assisted growth produced a roughly 650 nm FAPbI₃ film without the wrinkles, cracks and pinholes associated with conventional antisolvent quenching. The approximately 200 nm wide-gap top film was then added above it.
Two sputtered ITO recombination junctions connect the three subcells in series. Their measured EQE-integrated currents were 11.6, 11.0 and 15.9 mA cm−2, so the middle perovskite remained current-limiting. The champion reached 2.84 V open-circuit voltage, 11.6 mA cm−2 short-circuit current and a 0.74 fill factor; stabilized output was 24.1% after five minutes. Unencapsulated devices retained 96.6% of their initial efficiency after 1,081 hours of dark storage at 85 °C in nitrogen.
Historical and device references: the open-access Hu et al. triple-junction article, its complete device and fabrication PDF, and the 2022 device above 20% efficiency.
19 · 2025
Proof-of-concept monolithic all-perovskite quadruple-junction photovoltaics
Four compositionally tuned perovskite absorbers divide sunlight into progressively lower-energy spectral bands within one series-connected thin-film device.
Hu and colleagues reported this architecture online on 23 December 2024; its version of record appeared in February 2025 in the March 2025 issue of Nature. It is therefore shown as the 2025 milestone immediately after the earlier 2024 perovskite–perovskite–silicon triple junction.
The proof-of-concept device used approximately 2.26, 1.80, 1.55 and 1.26 eV perovskite absorbers with respective thicknesses of roughly 250, 370, 740 and 900 nm. The first two are lead-based wide-gap cells, the third is a middle-gap cell, and the rear tin–lead cell extends absorption into the near infrared.
Three transparent recombination junctions connect the four subcells monolithically. The first two use PCBM/PEI with approximately 20 nm ALD-SnOx and 10 nm IZO; the third uses C60/SnOx/IZO followed by PEDOT:PSS. The study's phenylalanine-hydrochloride precursor strategy improved the difficult 1.26 eV tin–lead film. Quadruple-junction devices reached 27.9% at 0.25 cm2, 27.4% at 1 cm2, and a best open-circuit voltage of 4.94 V. These remain laboratory proof-of-concept results rather than a commercial module architecture.
Historical and device references: the open-access Hu et al. Nature article, its archived full text and methods, and the Oxford research record and supplementary files.
20 · 2026
Memristor-integrated, self-protecting perovskite photovoltaics
A small memristive region is fabricated inside the solar-cell footprint so a shaded cell can automatically bypass damaging reverse current.
Mohammadi and colleagues introduced the Memsol concept in Nature on 18 March 2026. It follows the 2025 four-junction milestone in this chronology, but it is a different kind of advance: Memsol does not add another light-absorbing junction. It integrates an adaptive protection element into a perovskite cell.
Series-connected cells can force a shaded perovskite cell into reverse bias, causing rapid degradation. The Memsol uses area-selective deposition to form a local metal–insulator–perovskite switching region that shares the cell's perovskite absorber and outer electrodes. During ordinary illuminated operation it remains in a high-resistance off state, so the photovoltaic region delivers full output. Under harmful reverse bias it changes to a low-resistance on state and carries bypass current around the shaded photovoltaic region.
The demonstrated inverted cell uses ITO, NiOx/Me-4PACz, a mixed-cation lead-halide perovskite, PZ-2HI, C60/BCP and gold. The adjacent memristive region adds patterned silver, PL-177 and SiO isolation while retaining common TCO, perovskite and gold connections. The paper demonstrated autonomous protection in a nine-cell laboratory string. One reported device reached 22.5% efficiency, and the memristor's off-state current was seven orders of magnitude below the photocurrent.
Historical and device reference: the primary Mohammadi et al. Nature article, including its device cross-section and extended reverse-bias measurements.