TECHNOLOGY EXPLAINER
All-Perovskite Tandems
— two kinds of perovskite stacked, with no silicon at all
In an all-perovskite tandem, both the top and the bottom cell are made of perovskite. The bottom cell uses a perovskite that mixes tin and lead, which reaches out into the infrared. A cell of about 1 cm² has been independently measured at 28.2% (initial efficiency). The key question is how to protect tin, which oxidises all too easily.
- What an all-perovskite tandem is, in three points
- Pairing the bandgaps — 1.8 eV and 1.2 eV
- Where efficiency stands — the distance from small cells to modules
- The layer stack — two cells coated one on top of the other on a single sheet of glass
- The narrow-bandgap side: the tin-lead bottom cell
- A materials engineer's view (1): undoing tin oxidation with metallic tin
- The wide-bandgap side: why voltage gets left on the table
- A materials engineer's view (2): coating twice with the same kind of solvent
- How it compares with the silicon tandem
- Strengths and open problems
- Glossary / References / Claim-to-source audit
Sourced = stated in published material (link given)
Our calculation = a value this article derived from assumptions it states
Not yet confirmed = a plan, target or outlook with no confirmed track record
Structural readings and materials-design interpretations are marked separately as Commentary.
Every efficiency figure is given with whether it is certified, whether it is a cell or a module, and its area.
1. What an all-perovskite tandem is, in three points
An all-perovskite tandem is a tandem solar cell that stacks a perovskite with a wide bandgap (wide-gap) on top of one with a narrow bandgap (narrow-gap). Because it uses no silicon, the whole device can be made from thin films alone.
- Why bother: a review in Energy & Environmental Science (EES; Lim et al., 2024) lists the advantages of perovskites as including a bandgap that can be tuned through composition, low cost from solution processing and similar methods, and ease of scaling combined with light weightSourced
- How far it has come: Version 68 of the efficiency tables lists 28.2% for a cell of about 1 cm² (1.038 cm²; measured by JET, initial efficiency) and 22.8% for a module of about 810 cm² (measured by Fraunhofer ISE, initial performance)Sourced
- What makes it hard: the same review names as challenges oxidation of tin (Sn²⁺) on the narrow-gap side, voltage loss on the wide-gap side, the lack of standardisation in recombination layers, and the need to deposit every layer with near-perfect uniformitySourced
The bottom cell of an all-perovskite tandem is a semiconductor built largely around a metal ion that oxidises easily. The tin needs to stay divalent (Sn²⁺), but on contact with oxygen it tends to become tetravalent (Sn⁴⁺). A 2019 Nature Energy paper attacked this problem with a classic reaction from inorganic chemistry: adding metallic tin to turn Sn⁴⁺ back into Sn²⁺Sourced.
2. Pairing the bandgaps — 1.8 eV and 1.2 eV
In a tandem, the top cell handles the short wavelengths (blue to red) and the bottom cell the long wavelengths (red to near infrared). The review by Lim et al. says that the ideal top cell for pairing with silicon or CIGS (bandgap about 1.1 eV) is close to 1.7 eV, but that when pairing with a narrow-gap perovskite of about 1.2 to 1.3 eV, around 1.8 eV is preferableSourced.
According to the same review, 1.22 eV is widely used on the narrow-gap side, and a two-terminal pairing of a 1.82 eV top cell with a 1.22 eV bottom cell has a theoretical efficiency of 43%Not yet confirmed. Among actual reports, pairings of narrow gaps of 1.21 to 1.26 eV with wide gaps of 1.73 to 1.78 eV are summarised as reaching efficiencies of 26.0 to 28.5%Sourced.
- 1240 ÷ 1.77 = about 701 nm (absorption edge of the wide-gap top cell)Our calculation
- 1240 ÷ 1.22 = about 1016 nm (absorption edge of the tin-lead bottom cell)Our calculation
- 1240 ÷ 1.12 = about 1107 nm (absorption edge of crystalline silicon)Our calculation
What this tells us: a tin-lead bottom cell stops absorbing about 90 nm short of where silicon does. That costs current, but the larger bandgap raises the voltage (Section 9 compares measured values). Assumptions and limits: the edge conversion is an approximation and does not include the real absorption spectrum or the effect of film thickness.
3. Where efficiency stands — the distance from small cells to modules
| Category | Efficiency | Area | Test centre (date) | Made by |
|---|---|---|---|---|
| Cell (very small area) | 30.1 ± 0.8% | 0.0493 cm² (da) | JET (October 2023) | Nanjing University / Renshine, two-terminal |
| Cell (about 1 cm²) | 28.2 ± 0.5% | 1.038 cm² (da) | JET (December 2022) | Nanjing University / Renshine, two-terminal |
| Minimodule | 26.2 ± 0.6% | 64.84 cm² (da) | JET (September 2024) | Renshine / Nanjing University, 14 cells |
| Module | 22.8 ± 0.8% | 809.7 cm² (da) | Fraunhofer ISE (April 2025) | Renshine Solar |
All Sourced (efficiency tables Version 68 [Ref. 1]). All carry the note initial efficiency / initial performance. Because its cell area is small, the 30.1% result appears not in the main record table but among the “Notable exceptions”.
In addition, a Nanjing University group reported in Science in 2024 an all-perovskite tandem module whose tin-lead film was made by blade coating (spreading the solution with a blade), with a certified efficiency of 24.5% on an aperture area of 20.25 cm²Sourced. A 2024 Nature paper also points out that certified efficiencies of 1 cm²-scale cells lag behind those of small cells of about 0.05 cm², and that one reason is non-uniformity of the wide-gap subcell over larger areasSourced.
- 0.0493 cm² → 1.038 cm² (about 21 times the area): 30.1 − 28.2 = 1.9 pointsOur calculation
- → 64.84 cm² minimodule: 30.1 − 26.2 = 3.9 pointsOur calculation
- → 809.7 cm² module: 30.1 − 22.8 = 7.3 pointsOur calculation
- Checking the efficiency: the 1.038 cm² cell gives Voc 2.159 V × Jsc 16.59 mA/cm² × FF 0.789 = 28.26 mW/cm² → 28.3% (in close agreement with the table's 28.2%)Our calculation
Assumptions and limits: neither the measurement dates (2022 to 2025) nor the area definitions are consistent, so not all of the gap is an area effect. For perovskite-silicon tandems, the gap from a cell of about 1 cm² to a module of about 1.7 m² was 5.8 points (see our explainer on perovskite-silicon tandems). All-perovskite tandems already show a gap of more than 7 points at about 810 cm².
4. The layer stack — two cells coated one on top of the other on a single sheet of glass
A typical all-perovskite tandem lets light in through glass coated with a transparent electrode (a superstrate configuration). The wide-gap cell is built on the glass, followed by a recombination layer and then the narrow-gap cell on top. Hooijer et al. summarise that all-perovskite tandems generally use a p-i-n configuration, with recombination layers made of stacks such as SnO2/Au or ITO/PEDOT:PSSSourced.
The crucial point is that after the lower layers (the wide-gap cell) are made, a second perovskite is coated from solution on top of them. For the common configuration in which the wide-gap side is made first and the narrow-gap side added afterwards, Lim et al. point to a stability drawback: the oxygen-sensitive tin-lead layer ends up as the outermost layerSourced. Conversely, they note that in a “substrate” configuration, where the narrow-gap side is made first, the tin-lead layer is buried inside and protected from oxygen, and films such as PET, PEN and polyimide, or metal foils, can also serve as the substrateSourced.
5. The narrow-bandgap side: the tin-lead bottom cell
Replacing part of the lead (Pb) with tin (Sn) lowers the perovskite bandgap to 1.2 to 1.4 eV, in the near infraredSourced. In 2016 Eperon et al. made a 1.2 eV perovskite with the composition FA0.75Cs0.25Sn0.5Pb0.5I3 and reported 14.8% as a single junction, 17.0% in a two-terminal tandem with a wide-gap perovskite (open-circuit voltage above 1.65 V), and 20.3% in a four-terminal tandemSourced.
(1) Mixing them gives a lower bandgap than either end member
The review by Lim et al. explains that tin-lead mixed perovskites show “bandgap bowing” that does not follow Vegard's law, and that a Pb0.25Sn0.75 alloy has a smaller bandgap than the pure tin compoundSourced. A mixture whose properties do not fall between those of its end members is a surprisingly important behaviour in materials design (our commentary). For how composition relates to properties, see our explainers on bandgap and composition design and on tin-based perovskites.
(2) Too thin and it gives too little current; too thick and the charges cannot get out
A 2022 Nature paper (Lin et al.) says that a large photocurrent in a tandem needs a thick tin-lead subcell, but that this is difficult because tin-lead perovskites have short carrier diffusion lengthsSourced. By passivating the surface with an ammonium-based cation, the group doubled the diffusion length to over 5 µm, achieved an absorber about 1.2 µm thick, and reported an all-perovskite tandem with a certified efficiency of 26.4%Sourced. The encapsulated tandem kept more than 90% of its initial performance after 600 hours of operation at the maximum power point under 1 sunSourced.
6. A materials engineer's view (1): undoing tin oxidation with metallic tin
To reduce tin vacancies in tin-lead perovskites, Lin et al. used a comproportionation reaction in which metallic tin turns Sn⁴⁺ (the oxidation product of Sn²⁺) back into Sn²⁺Sourced. As a result the diffusion length reached 3 µm in the best material, and a 1.22 eV cell achieved 21.1% efficiencySourced.
The elegance of the method is that the reducing agent is the same element. Add a different reducing agent, and its oxidised form stays in the film as an impurity. With metallic tin, whatever reacts simply becomes the Sn²⁺ you wanted, adding no by-products (our commentary).
The enemies of stability are in the neighbourhood too. Hooijer et al. point out that PEDOT:PSS, widely used as a hole transport layer, is acidic and hygroscopic, and accelerates tin oxidation in narrow-gap perovskites, severely damaging stabilitySourced. So in a tin-lead cell, the movement of acid, moisture and oxygen through every adjoining layer, not only the absorber, becomes part of the design. Materials engineers who have handled easily oxidised metal powders or acid-sensitive substrates will recognise the way of thinking (our commentary).
7. The wide-bandgap side: why voltage gets left on the table
The usual way to reach the wide bandgap of the top cell (about 1.8 eV) is to replace part of the iodine (I) with bromine (Br). But the review by Lim et al. explains that mixed I-Br perovskites undergo light-induced phase segregation, separating under illumination into I-rich and Br-rich regions, which leads to open-circuit voltage losses by creating traps and shortening lifetimesSourced. The cause is ion migration, covered in detail in our explainer on ion migration and degradation.
The same review cites a case in which a 1.77 eV FA-Cs cell reached only 1.03 V measured, against a theoretical open-circuit voltage of 1.49 VSourced. By contrast, Wang et al. in 2024 formed a thin two-dimensional perovskite layer at the upper interface with the electron transport layer (C60), achieving an open-circuit voltage of 1.35 V in a 1.77 eV cell and 20.5% efficiency on about 1 cm², and reported 28.5% (certified 28.2%) for a 1.05 cm² tandem made by stacking it with a narrow-gap cellSourced.
- Against a bandgap of 1.77 eV, the Wang et al. cell has an open-circuit voltage of 1.35 V. The difference is 1.77 − 1.35 = 0.42 VOur calculation
- In the example cited by the review (1.03 V measured): 1.77 − 1.03 = 0.74 VOur calculation
Assumptions and limits: the gap between bandgap and open-circuit voltage is never zero, even in an ideal cell (according to the review, the theoretical open-circuit voltage at 1.77 eV is 1.49 V). The difference of about 0.3 V between these two 1.77 eV examples is given only as a rough measure of how much careful interface engineering has recovered.
8. A materials engineer's view (2): coating twice with the same kind of solvent
An all-perovskite tandem is a structure in which perovskite is coated from solution on top of perovskite. Hooijer et al. point out that because neighbouring subcells use similar solvents, layers can dissolve or degrade chemicallySourced.
The answer is a dense SnO2 buffer layer made by ALD (atomic layer deposition). In all-perovskite cells, a dense SnO2 layer about 10 to 20 nm thick, made at 100 °C or below, is commonly used to stop solvent penetrating when the upper cell is madeSourced. However, the underlying C60 is hydrophobic and chemically inert, so nuclei form poorly in the first ALD cycles and the film tends to be non-uniform. Making the SnO2 thicker to compensate adds parasitic absorption, contact resistance and mechanically weak interfacesSourced.
Reported remedies include modifying the C60 surface with PEIE or with molecules terminated by carboxyl groups, and ozone treatment partway through ALD, to improve the coverage and density of the SnO2Sourced.
Putting an inorganic barrier film evenly on top of a hydrophobic organic film. This has the same shape as problems materials engineers have solved with primers and nucleation treatments, when putting a gas barrier on a plastic film or plating onto resin. The difference is that the allowed thickness is a dozen or so nanometres, and the layer must also let light through (our commentary).
Hooijer et al. also list as a challenge that concentrated perovskite precursor solutions create micrometre-scale surface wrinkles, which the thin recombination layer on top cannot smooth outSourced. Levelling of the wet film, drying rate, stress: the fundamentals of coating feed straight into efficiency and yield (our commentary). Coating processes are covered in our explainers on solution-based film deposition and on roll-to-roll processing and scale-up.
9. How it compares with the silicon tandem
Here are the two tandem cells of about 1 cm² listed in Version 68 of the efficiency tables, side by side.
| Item | Perovskite-silicon | All-perovskite |
|---|---|---|
| Efficiency | 35.2 ± 1.1% | 28.2 ± 0.5% |
| Area | 0.9994 cm² (da) | 1.038 cm² (da) |
| Open-circuit voltage, Voc | 1.998 V | 2.159 V |
| Short-circuit current density, Jsc | 20.55 mA/cm² | 16.59 mA/cm² |
| Fill factor, FF | 85.2% | 78.9% |
| Test centre (date) | ESTI (February 2026) | JET (December 2022) |
All Sourced (efficiency tables Version 68 [Ref. 1]). Both are two-terminal and both are initial efficiencies.
- Voltage: 2.159 − 1.998 = +0.16 V, higher for the all-perovskite tandemOur calculation
- Current: 16.59 ÷ 20.55 = about 0.81, so the all-perovskite tandem delivers about 20% lessOur calculation
- Fill factor: a gap of 85.2 − 78.9 = 6.3 pointsOur calculation
What this tells us: because the bottom cell's bandgap is larger (1.12 eV → about 1.2 to 1.3 eV), the voltage rises and the current falls (the difference in absorption edge in Section 2). Beyond that, most of the efficiency gap comes from fill factor and current. Assumptions and limits: the two cells differ in maker and in measurement date (about three years apart).
10. Strengths and open problems
(1) Long-term durability is still measured in hundreds of hours
The durability data in this article are on the scale of hundreds of hours: more than 90% of initial performance after 600 hours (Lin et al., 2022, encapsulated)Sourced and 90% after 463 hours (Lin et al., 2019)Sourced. How many years these cells last outdoors could not be confirmed in the primary sources within the scope of this article. Outdoor test methods are covered in our explainer on outdoor testing and durability assessment.
(2) Theoretical efficiencies are only calculated values
The figures of 43% for a two-terminal 1.82 eV / 1.22 eV pairing, and 46% (two-terminal) and 46.1% (four-terminal) when the bandgaps are optimised over 0.5 to 2.5 eV, are theoretical upper limits cited in the reviewNot yet confirmed.
(3) It uses both lead and tin
Tin-lead bottom cells contain lead. How lead is handled is covered in our explainer on lead and iodine.
(4) Timing of mass production is outside this article
The efficiency tables include a Renshine Solar module, but at the time of our research (September 2026), full-scale mass production of all-perovskite tandems could not be confirmed in independent primary sources.
- An all-perovskite tandem stacks a perovskite of about 1.8 eV on one of about 1.2 to 1.3 eVSourced
- 28.2% on about 1 cm² and 22.8% for a module of about 810 cm² (efficiency tables, initial efficiency)Sourced. The gap from small-area cells widens to more than 7 pointsOur calculation
- Tin in the bottom cell oxidises easily, and a method using metallic tin to turn Sn⁴⁺ back into Sn²⁺ has been reportedSourced
- The wide-gap side tends to lose voltage through light-induced phase segregation, and careful interface engineering is winning it backSourced
- Coating the upper layer without dissolving the one below needs a dense protective layer a dozen or so nanometres thick. Nucleating it on a hydrophobic organic film is a materials problem (commentary)
- Higher voltage than a silicon tandem, but about 20% less currentOur calculation
11. Glossary
- All-perovskite tandem
- A tandem solar cell in which both the top and the bottom cell are made of perovskite.
- Wide gap / narrow gap
- Semiconductors with a wide (about 1.8 eV) or narrow (about 1.2 to 1.3 eV) bandgap.
- Tin-lead perovskite
- A composition in which part of the lead on the B site is replaced by tin. Absorbs out into the near infrared.
- Bandgap bowing
- The bandgap of a mixture falling below a straight-line interpolation between the end members.
- Comproportionation
- A reaction in which two species of the same element in different oxidation states react to give an intermediate oxidation state.
- Tin vacancy
- A site in the crystal where tin is missing. A cause of poorer performance in tin-based perovskites.
- Diffusion length
- The average distance that light-generated charge can travel before it recombines.
- Light-induced phase segregation
- Mixed I-Br perovskites separating under illumination into I-rich and Br-rich regions.
- Superstrate / substrate configuration
- Light enters through the substrate / light enters from the side opposite the substrate.
- ALD (atomic layer deposition)
- A deposition method that builds a film one atomic layer at a time by alternating precursor gases. Dense, with excellent step coverage.
- PEDOT:PSS
- A composite of a conducting polymer and polystyrene sulfonic acid. Used as a hole transport layer. Acidic.
- Blade coating
- A coating method that spreads solution over the substrate with a blade. Suited to scale-up.
- JET
- Japan Electrical Safety & Environment Technology Laboratories, one of the Japanese bodies that measure solar cell efficiency as an independent third party.
12. References (primary sources)
- Green, M.A. et al. “Solar cell efficiency tables: Version 68”, Joule 10, 102494 (2026) https://doi.org/10.1016/j.joule.2026.102494
- Eperon, G.E. et al. “Perovskite-perovskite tandem photovoltaics with optimized band gaps”, Science 354, 861–865 (2016) https://doi.org/10.1126/science.aaf9717
- Lin, R. et al. “Monolithic all-perovskite tandem solar cells with 24.8% efficiency exploiting comproportionation to suppress Sn(II) oxidation in precursor ink”, Nature Energy 4, 864–873 (2019) https://doi.org/10.1038/s41560-019-0466-3
- Lin, R. et al. “All-perovskite tandem solar cells with improved grain surface passivation”, Nature 603, 73–78 (2022) https://doi.org/10.1038/s41586-021-04372-8
- Wang, Y. et al. “Homogenized contact in all-perovskite tandems using tailored 2D perovskite”, Nature 635, 867–873 (2024) https://doi.org/10.1038/s41586-024-08158-6
- Hooijer, R. et al. “Why more junctions do not yet deliver: interconnection challenges in perovskite multijunction solar cells”, Energy & Environmental Science 19, 4582–4596 (2026, open access) https://doi.org/10.1039/d6ee01631f
- Lim, J. et al. “All-perovskite tandem solar cells: from fundamentals to technological progress”, Energy & Environmental Science 17, 4390–4425 (2024, open access) https://doi.org/10.1039/d3ee03638c
- Gao, H. et al. “Homogeneous crystallization and buried interface passivation for perovskite tandem solar modules”, Science 383, 855–859 (2024) https://doi.org/10.1126/science.adj6088
13. Claim-to-source audit
| Claim in the text | Basis | Label |
|---|---|---|
| The all-perovskite tandem entries in Version 68 of the efficiency tables: cell at 28.2 ± 0.5% (1.038 cm², da, JET, December 2022, Nanjing University / Renshine, two-terminal, Voc 2.159 V, Jsc 16.59 mA/cm², FF 78.9%); minimodule at 26.2 ± 0.6% (64.84 cm², JET, September 2024, 14 cells); the “notable exception” cell at 30.1 ± 0.8% (0.0493 cm², JET, October 2023); module at 22.8 ± 0.8% (809.7 cm², FhG-ISE, April 2025, Renshine Solar). The note “initial efficiency / initial performance”. The perovskite-silicon tandem at 35.2 ± 1.1% (0.9994 cm², ESTI, February 2026, Voc 1.998 V, Jsc 20.55 mA/cm², FF 85.2%) | Efficiency tables Version 68 (Joule, 2026). Reference 1 https://doi.org/10.1016/j.joule.2026.102494 | Sourced |
| A 1.2 eV perovskite of FA0.75Cs0.25Sn0.5Pb0.5I3; 14.8% as a single junction, 17.0% in a two-terminal tandem (Voc above 1.65 V) and 20.3% in a four-terminal tandem | Eperon et al. (Science, 2016). Reference 2 https://doi.org/10.1126/science.aaf9717 | Sourced |
| That tin vacancies were reduced by a comproportionation reaction in which metallic tin turns Sn⁴⁺ back into Sn²⁺. Diffusion length of 3 µm (best material), 21.1% for a 1.22 eV cell, certified tandem efficiencies of 24.8% (0.049 cm²) and 22.1% (1.05 cm²), and 90% retained after 463 hours of operation at the maximum power point | Lin et al. (Nature Energy, 2019). Reference 3 https://doi.org/10.1038/s41560-019-0466-3 | Sourced |
| That high photocurrent needs a thick tin-lead subcell, which is difficult because of short diffusion lengths. Passivation with an ammonium-based cation doubling the diffusion length to over 5 µm, an absorber about 1.2 µm thick, and a certified efficiency of 26.4%. More than 90% of initial performance retained by the encapsulated tandem after 600 hours at the maximum power point under 1 sun | Lin et al. (Nature, 2022). Reference 4 https://doi.org/10.1038/s41586-021-04372-8 | Sourced |
| That certified efficiencies at the 1 cm² scale lag behind those of cells of about 0.05 cm², one reason being non-uniformity of the wide-gap cell over larger areas. A two-dimensional perovskite layer at the upper interface with C60, giving Voc of 1.35 V and 20.5% efficiency for a 1.77 eV cell (square-centimetre scale), and 28.5% (certified 28.2%) for a 1.05 cm² tandem | Wang et al. (Nature, 2024). Reference 5 https://doi.org/10.1038/s41586-024-08158-6 | Sourced |
| That all-perovskite tandems generally use a p-i-n configuration, with recombination layers of SnO2/Au or ITO/PEDOT:PSS. That PEDOT:PSS is acidic and hygroscopic and accelerates tin oxidation. Dissolution and degradation of layers from similar solvents. That dense SnO2 layers about 10 to 20 nm thick made at 100 °C or below prevent solvent penetration. That C60 is hydrophobic and inert, hindering ALD nucleation, and that thick SnO2 brings parasitic absorption, contact resistance and weak interfaces. Improvements from modification with PEIE or carboxyl-terminated molecules and from ozone treatment. Micrometre-scale wrinkles from concentrated precursor solutions | Hooijer et al. (Energy & Environmental Science, 2026). Reference 6 https://doi.org/10.1039/d6ee01631f | Sourced |
| The advantages of perovskites (tunable bandgap, low cost, ease of scaling and light weight) and the challenges (Sn²⁺ oxidation, voltage loss on the wide-gap side, non-standardised recombination layers, uniform deposition). An ideal top cell close to 1.7 eV when paired with Si or CIGS, and around 1.8 eV when paired with a 1.2 to 1.3 eV perovskite. That 1.22 eV is widely used. A theoretical 43% for 1.82/1.22 eV, and 46% (two-terminal) and 46.1% (four-terminal) with optimisation over 0.5 to 2.5 eV. Actual reports of 26.0 to 28.5% with narrow gaps of 1.21 to 1.26 eV and wide gaps of 1.73 to 1.78 eV. Tin-lead mixing giving 1.2 to 1.4 eV, bowing, and a lower bandgap for Pb0.25Sn0.75 than for the pure tin compound. Light-induced phase segregation in mixed I-Br compositions and ion migration. The example of 1.03 V measured against a theoretical Voc of 1.49 V at 1.77 eV. The drawback of the tin-lead layer being outermost in the superstrate configuration, and that the substrate configuration allows PET, PEN, polyimide, metal foil and the like as substrates. An example layer order (ITO / NiOx / wide gap / C60 / SnO2 / ITO / PEDOT:PSS / narrow gap / C60 / Ag). The silicon bandgap of 1.12 eV. That the ease of tin(II) oxidation is a major source of vacancy defects in tin-based narrow-gap layers | Lim et al. (Energy & Environmental Science, 2024). Reference 7 https://doi.org/10.1039/d3ee03638c | Sourced |
| An all-perovskite tandem module with a blade-coated tin-lead film, with a certified efficiency of 24.5% on an aperture area of 20.25 cm² | Gao et al. (Science, 2024). Reference 8 https://doi.org/10.1126/science.adj6088 | Sourced |
| Theoretical efficiencies of 43%, 46% and 46.1% | Theoretical upper limits cited in the review, not demonstrated values. Reference 7 https://doi.org/10.1039/d3ee03638c | Not yet confirmed |
| Absorption-edge wavelengths (about 701, 1016 and 1107 nm) and the gap of about 90 nm. Efficiency gaps by area (1.9, 3.9 and 7.3 points) and the area ratio of about 21. The check of 28.2% as Voc × Jsc × FF (28.26 mW/cm²). Voltage left on the table of 0.42 V and 0.74 V, and their difference of about 0.3 V. Differences from the silicon tandem (+0.16 V, current ratio about 0.81, FF gap of 6.3 points) | Our calculation. Absorption edges are approximated as 1240 divided by the bandgap. The cells compared differ in maker, area definition and measurement date. The silicon value of 1.12 eV is used as a general figure | Our calculation |
| Long-term outdoor durability, and full-scale mass production of all-perovskite tandems | Not stated because they could not be confirmed in the primary sources within the scope of this article (commentary by this article) | Not yet confirmed |
| The reading that metallic tin is a same-element reducing agent that adds no by-products. The framing that acid, moisture and oxygen in every adjoining layer become design targets. Treating nucleation of an inorganic barrier on a hydrophobic organic film as the same problem as primers and nucleation treatments. The reading that coating fundamentals affect efficiency and yield. The importance of mixture properties falling outside the range of the end members. The comparison of the two tandems and the sorting into strengths and challenges. The extent of the recombination-layer bracket in Fig. 2 | This article's own framing and commentary based on published content. Not views expressed by the institutions or researchers | Commentary |
| That Figs. 1, 2, 3, 5 and 6 are explanatory drawings; that the hero image and Fig. 4 are AI-generated images; and the way the equation in Fig. 3 is written | A note by this article (commentary). The equation in Fig. 3 is this article's rendering of the wording of the Lin et al. abstract | Commentary |
Last updated 25 September 2026. Sources are limited to primary material (the efficiency tables and peer-reviewed papers). Every efficiency value is given with whether it is certified, whether it is a cell or a module, its area and the test centre. Long-term outdoor durability and full-scale mass production of all-perovskite tandems are not stated, because they could not be confirmed in published primary sources. All figures are for explanation. Figs. 1, 2, 3, 5 and 6 are vector drawings, and the hero image and Fig. 4 are AI-generated images; none of them shows a real cross-section, micrograph or physical product.