MENU
Advanced Materials Intelligence
  1. Home
  2. Flexible Perovskite Solar Cells Explained | Perovskite Solar Cells

Flexible Perovskite Solar Cells Explained | Perovskite Solar Cells

Back to Technology, Institution & Company Guides

TECHNOLOGY EXPLAINER

Flexible Perovskite Solar Cells
— light weight, paid for with limits on temperature and bending

These are perovskite solar cells built on bendable substrates such as plastic film. Japan's national strategy puts the weight of film-type cells at one-tenth that of typical silicon solar panels and aims to install them on places such as roofs that cannot bear much load. On film, however, high-temperature processes are ruled out, films crack when bent, and moisture gets through more easily.

Built from primary sources: the Next-Generation Solar Cell Strategy of Japan's Ministry of Economy, Trade and Industry (METI, November 2024), peer-reviewed papers (Nature, Nature Materials, Nature Communications, Science Advances, Nano-Micro Letters and others) and official LONGi announcements / Last updated September 2026

Conceptual image of a single thin, plain dark film lying in a gentle arc against a dark background, with soft light flowing across its surface
AI-generated concept image. An impression of a solar cell that is thin, light and bendable. It does not represent a real product, structure, electrode or colour.
What this article covers
  1. What a flexible solar cell is, in three points
  2. Why bother — light weight and where it can go
  3. Where efficiency stands — single-junction cells and silicon tandems
  4. The process-temperature ceiling — what temperatures film can take
  5. A materials engineer's view (1): achieving dense films at low temperature through materials
  6. Bending and strain — where films crack
  7. A materials engineer's view (2): adhesion and stress relief are what stop cracking
  8. Our calculation: power per unit weight
  9. Strengths and open problems
  10. Glossary / References / Claim-to-source audit
How claims are labelled in this article

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 a flexible solar cell is, in three points

A flexible perovskite solar cell (film-type) is a perovskite solar cell built on a bendable substrate such as plastic film instead of glass. Because the perovskite layer can be made from solution at low temperature, it can also be laid on plastics.

  • Why it attracts attention: METI's Next-Generation Solar Cell Strategy (November 2024), Japan's national plan for perovskite and other new solar cells, describes film-type cells as manufacturable at low temperature, light and flexible, and suited to a wide variety of installation formsSourced
  • How light: the strategy puts the weight of perovskite solar cells at one-tenth of typical silicon solar panels and the volume at one-twentieth, and uses 1.5 kg/m² for film-type cells in its estimatesSourced
  • What makes it hard: a review in Nano-Micro Letters (Xu et al., 2022) says flexible substrates need the flexibility to survive repeated bending, high transparency, thermal stability to withstand annealing in later deposition steps, and hermeticity to keep moisture outSourced
The single most important line in this article

Making a cell flexible is not a matter of swapping glass for film. It means redesigning the process temperature, mechanical properties and barrier performance of every layer to suit the film. Neither the 500 °C firing used on glass nor FTO deposited above 350 °C can be used on filmSourced. Bend it, and the transparent electrode cracksSourced.

2. Why bother — light weight and where it can go

The Next-Generation Solar Cell Strategy describes perovskite solar cells as light and flexible, which allows them to be installed in places where installation has been difficult until now, such as building wallsSourced. On that basis it sets out the following policies.

ItemWhat the Next-Generation Solar Cell Strategy (November 2024) says
Deployment targetAims for deployment of about 20 GW by 2040 (about 40 GW or more if costs fall substantially)
Generation costEstablish, through the Green Innovation (GI) Fund, technology capable of ¥20/kWh by 2025 and ¥14/kWh by 2030. By 2040, aim for a generation cost at which the technology can stand on its own (¥10/kWh to ¥14/kWh or below)
Production capacityAim to build GW-scale production capacity before 2030
Public sectorInclude in the Government Action Plan (the plan for cutting greenhouse gas emissions from the government's own operations) priority installation on roofs with low load-bearing capacity and other places where conventional solar cells were difficult to install
Weight and volumeOne-tenth (weight) and one-twentieth (volume) of typical silicon solar panels. Film-type estimated at 1.5 kg/m²
Assumptions for demand estimatesRoofs and walls set at 150 W/m² (15% efficiency), windows at 75 W/m² (7.5%)

The deployment, cost and production targets are Not yet confirmed (government targets and outlooks). Weight, volume and the demand-estimate assumptions are conditions set by the strategy for its estimates. All from METI's Next-Generation Solar Cell Strategy [Ref. 1].

The strategy also states that for film-type cells, know-how in materials processing and forming and in manufacturing processes will determine competitivenessSourced. In other words, it is a field where expertise in materials and processes translates directly into competitiveness (our commentary).

3. Where efficiency stands — single-junction cells and silicon tandems

CategoryEfficiencyArea and conditionsReport
Single-junction cell (flexible)Certified 25.55%Area not given in the text this article checked. Best value in the paper 25.88%Zhong et al. (Science Advances, 2026)
Single-junction module (flexible)21.77% / 19.23%5 cm × 5 cm / 10 cm × 10 cm. Values reported in the paperZhong et al. (2026)
Single-junction module (flexible)22.22%57.6 cm². Value reported in the paperTang et al. (Nature Communications, 2026)
Silicon tandem (flexible)Certified 33.6%Open-circuit voltage 2.015 V. Area not given in the abstractWang et al. (Nature, 2025)
Silicon tandem (flexible)Certified 33.4% / 29.8%1 cm² / 260 cm². Uses 60 µm thick siliconFang et al. / LONGi (Nature, 2025)
Silicon tandem (flexible)Certified 29.88%Aperture area 1.04 cm² (steady-state 29.2%)Sun et al. (Nature Communications, 2025)

All Sourced (Zhong et al. [Ref. 7], Tang et al. [Ref. 8], Wang et al. [Ref. 10], Fang et al. [Ref. 11], Sun et al. [Ref. 9]). Certifying bodies, areas and measurement conditions differ from paper to paper; unlike the efficiency tables, this is not a list compiled to a uniform standard.

In its own announcement, LONGi says the 29.8% on 260 cm² was certified by Fraunhofer ISE and the 1 cm²-scale value by the US NRELSourced. The company names applications such as space and vehicles, fields that call for light weight, flexibility and high outputNot yet confirmed.

4. The process-temperature ceiling — what temperatures film can take

Process temperatures and change points on film (literature values side by side) Bar length = temperature (°C), drawn at 1 °C = 0.7 px. Conditions differ between sources NiO spray pyrolysis (on glass) FTO deposition ITO on PEN: resistance rises ITO on PET: resistance rises SnO2 nanoparticle ETL Perovskite (low-temp. sequential) IZO deposition 500 °C over 350 °C 235 °C over 180 °C 150 °C 110 °C 100 °C Note: NiO 500 °C and perovskite 110 °C: Chen et al. [Ref. 4]; FTO and IZO: Xu et al. [Ref. 2]; ITO on PEN/PET: Chavan et al. [Ref. 3]; SnO2: Jiang et al. [Ref. 5]. The red dashed line marks the change point for ITO on PET (180 °C). Note: the PET/PEN values are not the films' heat limits but the temperatures at which ITO sheet resistance rose. Note: the choice of these seven items is this article's own; this is not a comprehensive comparison.
Fig. 1 Concept diagram (vector drawing). Each temperature follows Chen et al. [Ref. 4], the review by Xu et al. [Ref. 2], the review by Chavan et al. [Ref. 3] and Jiang et al. [Ref. 5]. Setting the seven items side by side is this article's own framing; equipment, duration, atmosphere and other conditions differ between sources. The PET and PEN values are the temperatures at which the sheet resistance of ITO on them rose, not the heat limits of the films themselves.

What Fig. 1 shows is that many of the high-temperature steps routinely used on glass cannot be used on film (our commentary).

  • According to Xu et al., FTO is deposited at substrate temperatures above 350 °C and cannot be used on polymer substrates, whereas IZO can be made at 100 °C and ITO, AZO and IWO at room temperatureSourced
  • The review by Chavan et al. cites reports that the sheet resistance of ITO on PET rose above 180 °C and ITO on PEN at 235 °CSourced
  • Fu et al. (Nature Communications, 2015) point out that conventional high-efficiency cells used TiO2 made at about 500 °C as the electron transport layer, which is incompatible with flexible cells on plasticSourced

5. A materials engineer's view (1): achieving dense films at low temperature through materials

Why this matters for materials engineers: making a good film without firing it

On glass, you can simply fire an oxide to crystallise and densify it. On film you cannot. What helps here is the idea of taking a material that is already crystalline and turning it into a film at low temperature.

Jiang et al. (Nature Energy, 2016) used SnO2 nanoparticles made by low-temperature solution processing as the electron transport layer, obtained a certified efficiency of 19.9 ± 0.6% in cells that can be made at 150 °C, and describe the method as suited to large-scale productionSourced. Crystallisation is moved forward to the stage where the particles are made (synthesis), so that film formation becomes just laying the particles down and drying them: a case where nanoparticle dispersion technology applies directly (our commentary).

The same pattern appears with transparent electrodes. Xu et al. say ITO made by room-temperature DC sputtering is amorphous, with high sheet resistance, low transmittance, a rough surface and insufficient flexibility, and cite an example in which ITO made by plasma-arc ion plating formed a smooth film on PET with a sheet resistance of 15.75 Ω/□ and a transmittance of 85.88%, and another in which annealing up to 300 °C on colourless polyimide drove crystallisation and dopant activation, giving lower sheet resistance than on PETSourced.

So there are two options: raise the temperature ceiling with a more heat-resistant film (such as colourless polyimide), or devise deposition methods and materials that make good films under a low ceiling. The first is the territory of film makers; the second belongs to makers of inks, sputtering targets and deposition equipment (our commentary).

6. Bending and strain — where films crack

Xu et al. give as a benchmark for flexible perovskite solar cells retaining about 90% of initial efficiency after 1,000 bends at a 4 mm radius, and say cells using ITO struggle to approach it, because repeated bending cracks the ITO film and raises its sheet resistanceSourced. In one example, reducing the ITO thickness from about 400 nm to 160 nm gave good mechanical durability even at a 4 mm bending radiusSourced.

How much the surface stretches when bent (our calculation) Surface strain ε ≈ thickness t / (2 × bend radius R). A simple formula assuming bending about mid-thickness The outer surface stretches The inner surface compresses Yellow dashes: neutral plane (no strain) Thinner means less stretch at the same radius Worked examples (our calculation) (1) 60 µm thick, 40 mm radius about 0.075% Matches the thin silicon and bending test of Fang et al. (2) 3 µm thick, 1 mm radius (assumed) about 0.15% Hypothetical case for the 3 µm cell of Kaltenbrunner et al. (3) 125 µm thick, 4 mm radius (assumed) about 1.6% 125 µm is our assumption; 4 mm is the benchmark cited by Xu et al. Note: 60 µm and 40 mm follow Fang et al. [Ref. 11], 3 µm Kaltenbrunner et al. [Ref. 6], and 4 mm Xu et al. [Ref. 2]. Note: the 1 mm radius in (2) and 125 µm in (3) are our assumptions. All strains are our calculation, not published values. Note: in a real stack, differing elastic moduli shift the neutral plane, so this simple formula does not hold exactly. Note: the curvature and thickness in the figure are schematic.
Fig. 2 Drawing that includes our calculation (vector drawing). The 60 µm thickness and radius of about 40 mm follow Fang et al. [Ref. 11], the 3 µm thickness Kaltenbrunner et al. [Ref. 6], and the 4 mm radius Xu et al. [Ref. 2]. The strain values (about 0.075%, 0.15% and 1.6%) were calculated by this article and are not published figures. The 1 mm radius in (2) and the 125 µm thickness in (3) are our own assumptions.
Our calculation: surface stretch set by thickness and radius
  • Formula (assumption): if the mid-thickness plane is the neutral plane that neither stretches nor compresses, surface strain ε ≈ t ÷ (2R)
  • (1) t = 60 µm, R = 40 mm: 60×10⁻⁶ ÷ (2 × 0.040) = 7.5×10⁻⁴ = about 0.075%Our calculation
  • (2) t = 3 µm, R = 1 mm (assumed): 3×10⁻⁶ ÷ (2 × 0.001) = 1.5×10⁻³ = about 0.15%Our calculation
  • (3) t = 125 µm (assumed), R = 4 mm: 125×10⁻⁶ ÷ (2 × 0.004) = 1.56×10⁻² = about 1.6%Our calculation

What this tells us: “bendable” covers a wide range: a 40-fold difference in thickness means a 40-fold difference in surface stretch at the same radius. For a brittle oxide film (such as ITO) sitting on the surface of a thick substrate, a small bending radius can be a harsh condition (our commentary). Assumptions and limits: in a real stack each layer has a different elastic modulus and the neutral plane shifts. The strain at which ITO cracks was not checked in this article's references.

7. A materials engineer's view (2): adhesion and stress relief are what stop cracking

Conceptual image of a thin dark film curving gently against a dark background, with a soft streak of light running along its edge
Fig. 3 AI-generated concept image. An impression of a thin film that neither cracks nor peels when bent. It does not represent a real product, layer structure or bending radius.
Why this matters for materials engineers: deliberately adding a loose layer to soak up strain energy

A 2025 Nature paper (Fang et al., with LONGi and others) identifies the challenge for flexible silicon tandems as mechanical stress from repeated environmental stresses causing interfacial delamination and degradationSourced. The answer is a two-layer buffer. A “loose” SnOx layer, made by adjusting the ALD purge time, dissipates strain energy, while a dense SnOx layer maintains reliable electrical contact. The buffer also softens ion bombardment during later sputteringSourced. As a result, the cells kept more than 97% of initial efficiency after 43,000 bends at a radius of curvature of about 40 mm, and about 97% after 250 thermal cycles between −40 and 85 °CSourced.

This is a design familiar to adhesive and composite engineers. Join two hard layers directly and stress concentrates at the interface until it peels; put a soft layer between them and the stress is relieved: the same thinking as a primer in a coating system, a stress-relief layer in an adhesive joint, or an interlayer in a multilayer film (our commentary). The difference is that here the layer is an oxide a few tens of nanometres thick that must also conduct electricity.

Another route is to strengthen adhesion itself. A 2025 Nature Communications paper (Li et al.) took its cue from the underwater adhesion of mussels, using a multi-branched polymer rich in dopamine groups to bind the perovskite grains and the substrate from many directions. It reports 24.43% for flexible cells, with 94.1% of initial efficiency kept after 10,000 bends at a 3 mm radius and 65% humiditySourced. Adhesion that holds even in wet conditions is squarely a research theme for adhesive chemists (our commentary).

There are also advances on the transparent-electrode side. Wang et al. (Nature, 2025) report that using in situ annealed IZO as the front transparent electrode markedly improved both optoelectronic and mechanical properties, with 91% of initial efficiency kept after 5,000 bends at a 17.6 mm radius and 90% after 1,000 hours of damp-heat testingSourced.

8. Our calculation: power per unit weight

One yardstick for the value of flexible solar cells is power per unit weight (W/g). Here are three figures from primary sources put into the same units.

Power per weight: cells and modules differ by orders of magnitude Units are W/g. The left two are values reported in papers; the right one is our conversion from the strategy's assumptions 23 W/g Ultrathin 3 µm cell 12% stabilised efficiency (2015) Reported by Kaltenbrunner et al. 1.77 W/g Tandem on 60 µm silicon Flexible (2025) Reported by Fang et al. (maximum) ~0.1 W/g Film-type module 150 W/m² / 1.5 kg/m² Our calculation (strategy assumptions) 23 / 0.1 = about 230 times. Most of the weight is the packaging, not the active layers Substrate, barrier film, encapsulant and wiring set module weight (this article's reading) Note: 23 W/g is reported by Kaltenbrunner et al. [Ref. 6] and 1.77 W/g by Fang et al. [Ref. 11]. Note: about 0.1 W/g is this article's conversion from the strategy's assumptions [Ref. 1] (150 W/m², 1.5 kg/m²). Note: the three differ in subject (cell or module) and conditions. The weight breakdown was not checked in the references. Note: the reading that the packaging sets the weight is this article's commentary.
Fig. 4 Drawing that includes our calculation (vector drawing). 23 W/g is reported by Kaltenbrunner et al. [Ref. 6] and 1.77 W/g by Fang et al. [Ref. 11]. About 0.1 W/g and about 230 times were calculated by this article from the assumptions in METI's strategy [Ref. 1] and are not published figures. Cells and modules are different things, so this is not a direct performance comparison.
Our calculation: putting the three figures on the same footing
  • Film-type module: the strategy's assumption of 150 W/m² (15% efficiency) ÷ 1.5 kg/m² = 100 W/kg = about 0.1 W/gOur calculation
  • Weight of a typical silicon solar panel (back-calculated from the strategy's ratio): 1.5 kg/m² × 10 = about 15 kg/m²Our calculation (a rough figure back-calculated from the strategy's “one-tenth”, not a measured value)
  • Weight of a 60 µm silicon wafer alone: taking the density of silicon as 2.33 g/cm³ (our assumption), 60 µm × 1 m² = 60 cm³ → about 140 g/m²Our calculation
  • At 29.8% efficiency on this wafer (1,000 W/m² incident), 298 W ÷ 140 g = about 2.1 W/g. That is close to the reported maximum of 1.77 W/g, suggesting most of the cell's weight is the silicon waferOur calculation

Assumptions and limits: 150 W/m² and 1.5 kg/m² are conditions the strategy set for its demand estimates and similar calculations, not measurements of a particular product. The area and configuration behind the 1.77 W/g of Fang et al. cannot be determined from the abstract this article checked.

What this calculation shows is that the lightness of a module is set not by the active layers but by the materials that package them (our commentary). The hermeticity to keep moisture out named by Xu et al. has to be achieved with film rather than glassSourced. Barrier films and encapsulant design are covered in our explainer on encapsulation and barrier layers.

9. Strengths and open problems

Strengths and challenges of flexible perovskite solar cells (this article's framing) Strengths Challenges One-tenth the weight of silicon (strategy estimate) Fits low-load roofs and curved surfaces Can be made by low-temperature solution processing Certified above 25% reported, single junction Certified above 33% reported, Si tandem 23 W/g reported for a bare cell No high-temperature firing or deposition Oxide films such as ITO crack when bent Interface delamination, stress concentration Film alone must keep moisture out Few efficiency lists with uniform criteria Deployment and cost targets unconfirmed Note: items follow the strategy [Ref. 1], the papers [Refs. 2 to 11] and our calculations (Sections 6 and 8). Note: the split into two groups is this article's own and is not an established industry assessment.
Fig. 5 Concept diagram (vector drawing). Each item follows METI's strategy [Ref. 1], the papers [Refs. 2 to 11] and this article's calculations (Sections 6 and 8). Sorting them into strengths and challenges is this article's own framing, not an established industry assessment.

(1) Deployment and cost goals are still just goals

About 20 GW by 2040, ¥14/kWh by 2030, GW-scale production capacity before 2030: all are government targetsNot yet confirmed. At the time of our research (September 2026), progress towards them could not be confirmed in primary sources. How generation cost is assessed is covered in our explainer on the levelised cost of electricity.

(2) Long-term durability

The durability figures in this article (10,000 and 43,000 bends, 1,000 hours of damp heat and so on) are all from accelerated laboratory testsSourced. How many years film-type cells last outdoors could not be confirmed in the primary sources within the scope of this articleNot yet confirmed.

(3) How to compare efficiencies

Efficiencies of flexible cells and modules come with different certifying bodies, areas and measurement conditions from paper to paper, and there is no list compiled to a uniform standard, like the efficiency tables, in which to compare them side by side (our commentary). Performance metrics are covered in our explainer on conversion efficiency and performance metrics.

The article in summary
  • Film-type cells weigh one-tenth as much as silicon (estimated at 1.5 kg/m²), and installation on roofs with low load-bearing capacity is a stated goalSourced
  • Deployment and cost targets such as about 20 GW by 2040 are still only targetsNot yet confirmed
  • On film, neither FTO deposited above 350 °C nor 500 °C firing can be used. The sheet resistance of ITO on PET has been reported to rise above 180 °CSourced
  • Surface stretch is set by thickness ÷ (2 × radius): about 1.6% for 125 µm at a 4 mm radiusOur calculation
  • A “loose” buffer layer relieves strain, and mussel-inspired adhesion copes with humidity: adhesion and stress-relief technology is paying offSourced
  • 23 W/g for a bare cell, about 0.1 W/g for a module. What sets the weight is the packagingOur calculation

10. Glossary

Flexible solar cell (film-type)
A solar cell built on a bendable substrate such as plastic film. Light, and can be mounted on curved surfaces.
PET / PEN
Polyethylene terephthalate / polyethylene naphthalate. Transparent plastic film substrates.
Colourless polyimide (CPI)
A transparent polyimide film with high heat resistance. Withstands higher temperatures than PET.
Bending radius (radius of curvature)
The radius of the arc when bent. The smaller it is, the sharper the bend.
Surface strain
The fraction by which a surface stretches or compresses when bent. Proportional to thickness and inversely proportional to bending radius.
Neutral plane
The plane that neither stretches nor compresses when bent. In a simple plate, the mid-thickness.
Power per weight (W/g)
A solar cell's output divided by its weight. A measure of lightness.
Delamination
Separation at the interface between stacked layers, caused by stress concentration.
Buffer layer (stress-relief layer)
A thin layer inserted between layers to ease stress or damage from later steps.
Dopamine group
The catechol structure found in mussel adhesive proteins. Sticks even to wet surfaces.
Ion plating
A deposition method in which material ionised by a plasma is driven onto the substrate.
GI Fund
Japan's Green Innovation Fund, a government fund programme that supports technology development.

11. References (primary sources)

  1. Ministry of Economy, Trade and Industry (METI), Public-Private Council on Expanding Deployment and Strengthening Industrial Competitiveness of Next-Generation Solar Cells “Next-Generation Solar Cell Strategy”, November 2024 (PDF, in Japanese) https://www.meti.go.jp/shingikai/energy_environment/perovskite_solar_cell/pdf/20241128_1.pdf
  2. Xu, Y. et al. “Recent Progress of Electrode Materials for Flexible Perovskite Solar Cells”, Nano-Micro Letters 14, 117 (2022, open access) https://doi.org/10.1007/s40820-022-00859-9
  3. Chavan, G.T. et al. “A Brief Review of Transparent Conducting Oxides (TCO): The Influence of Different Deposition Techniques on the Efficiency of Solar Cells”, Nanomaterials 13, 1226 (2023, open access) https://doi.org/10.3390/nano13071226
  4. Chen, M. et al. “Low-temperature sequential deposition for efficient inverted perovskite solar cells”, Nature Communications 16, 5746 (2025, open access) https://doi.org/10.1038/s41467-025-61144-y
  5. Jiang, Q. et al. “Enhanced electron extraction using SnO2 for high-efficiency planar-structure HC(NH2)2PbI3-based perovskite solar cells”, Nature Energy 2, 16177 (2016) https://doi.org/10.1038/nenergy.2016.177
  6. Kaltenbrunner, M. et al. “Flexible high power-per-weight perovskite solar cells with chromium oxide–metal contacts for improved stability in air”, Nature Materials 14, 1032–1039 (2015) https://doi.org/10.1038/nmat4388
  7. Zhong, Y. et al. “Homogenizing out-of-plane strain distribution for high-performance flexible perovskite photovoltaics”, Science Advances 12, eaec3238 (2026, open access) https://doi.org/10.1126/sciadv.aec3238
  8. Tang, Q. et al. “Buried-interface homogenization by asymmetric polymeric self-assembled layers powers efficient, durable flexible perovskite photovoltaics”, Nature Communications 17 (2026, open access) https://doi.org/10.1038/s41467-026-73276-w
  9. Sun, Y. et al. “Flexible perovskite/silicon monolithic tandem solar cells approaching 30% efficiency”, Nature Communications 16, 5733 (2025, open access) https://doi.org/10.1038/s41467-025-61081-w
  10. Wang, S. et al. “Flexible perovskite/silicon tandem solar cells with 33.6% efficiency”, Nature 649, 59 (published online 2025) https://doi.org/10.1038/s41586-025-09849-4
  11. Fang, Z. et al. (LONGi and others) “Flexible perovskite/silicon tandem solar cell with a dual-buffer layer”, Nature 649, 65 (published online 2025) https://doi.org/10.1038/s41586-025-09835-w
  12. LONGi “Nature Consecutively Publishes LONGi's Breakthroughs in HIBC and Flexible Silicon-Based Tandem Solar Cells”, 13 November 2025 https://www.longi.com/en/news/breakthroughs-in-hibc-and-flexible-silicon-based-tandem-solar-cells-nature/
  13. Li, Z. et al. “Boosting mechanical durability under high humidity by bioinspired multisite polymer for high-efficiency flexible perovskite solar cells”, Nature Communications 16, 1771 (2025, open access) https://doi.org/10.1038/s41467-025-57102-3
  14. Fu, F. et al. “Low-temperature-processed efficient semi-transparent planar perovskite solar cells for bifacial and tandem applications”, Nature Communications 6, 8932 (2015, open access) https://doi.org/10.1038/ncomms9932

12. Claim-to-source audit

Claim in the textBasisLabel
That film-type cells can be manufactured at low temperature, are light and flexible, and suit a wide variety of installation forms. The statement that they are light and flexible, allowing installation in places where it has been difficult, such as building walls. Weight one-tenth and volume one-twentieth of typical silicon solar panels, with film-type estimated at 1.5 kg/m². Inclusion in the Government Action Plan of priority installation on roofs with low load-bearing capacity. Demand estimates assuming 150 W/m² (15% efficiency) for roofs and walls and 75 W/m² for windows. That for film-type cells, know-how in materials processing and forming and in manufacturing processes determines competitivenessMETI, Next-Generation Solar Cell Strategy (November 2024). Reference 1 https://www.meti.go.jp/shingikai/energy_environment/perovskite_solar_cell/pdf/20241128_1.pdfSourced
The deployment target of about 20 GW by 2040 (about 40 GW or more with substantial cost reductions). Establishing technology for ¥20/kWh by 2025 and ¥14/kWh by 2030, and the target of ¥10 to ¥14/kWh or below by 2040. Building GW-scale production capacity before 2030Government targets and outlooks, not results. Reference 1 https://www.meti.go.jp/shingikai/energy_environment/perovskite_solar_cell/pdf/20241128_1.pdfNot yet confirmed
The properties required of flexible substrates (flexibility, transparency, thermal stability, hermeticity). That FTO is deposited above 350 °C and cannot be used on polymer substrates, while IZO can be made at 100 °C and ITO, AZO and IWO at room temperature. The benchmark of about 90% after 1,000 bends at a 4 mm radius and the difficulty with ITO, cracking and rising sheet resistance. The example of thinning ITO from about 400 nm to 160 nm for good durability at a 4 mm radius. The inferior properties of ITO made by room-temperature DC sputtering. Ion-plated ITO on PET at 15.75 Ω/□ and 85.88%. The example of annealing up to 300 °C on colourless polyimide driving crystallisation and dopant activation, with lower sheet resistance than on PETXu et al. (Nano-Micro Letters, 2022). Reference 2 https://doi.org/10.1007/s40820-022-00859-9Sourced
Reports that the sheet resistance of ITO on PET rose above 180 °C and of ITO on PEN at 235 °CChavan et al. (Nanomaterials, 2023). Reference 3 https://doi.org/10.3390/nano13071226Sourced
NiO made on glass by spray pyrolysis at 500 °C, and the perovskite made by low-temperature sequential deposition at 110 °CChen et al. (Nature Communications, 2025). Reference 4 https://doi.org/10.1038/s41467-025-61144-ySourced
SnO2 nanoparticles from low-temperature solution processing as the electron transport layer, fabrication at 150 °C, a certified efficiency of 19.9 ± 0.6%, and suitability for large-scale productionJiang et al. (Nature Energy, 2016). Reference 5 https://doi.org/10.1038/nenergy.2016.177Sourced
An ultrathin flexible cell 3 µm thick, 12% stabilised efficiency, and power per weight of 23 W/gKaltenbrunner et al. (Nature Materials, 2015). Reference 6 https://doi.org/10.1038/nmat4388Sourced
Best flexible cell at 25.88%, certified 25.55%, and 21.77% and 19.23% for 5 cm and 10 cm square modulesZhong et al. (Science Advances, 2026). Reference 7 https://doi.org/10.1126/sciadv.aec3238Sourced
22.22% for a 57.6 cm² flexible moduleTang et al. (Nature Communications, 2026). Reference 8 https://doi.org/10.1038/s41467-026-73276-wSourced
A flexible silicon tandem certified at 29.88% (steady-state 29.2%, aperture area 1.04 cm²)Sun et al. (Nature Communications, 2025). Reference 9 https://doi.org/10.1038/s41467-025-61081-wSourced
A flexible silicon tandem certified at 33.6% with an open-circuit voltage of 2.015 V. 91% after 5,000 bends at a 17.6 mm radius and 90% after 1,000 hours of damp-heat testing. Improved optoelectronic and mechanical properties from an in situ annealed IZO front electrodeWang et al. (Nature, 2025). Reference 10 https://doi.org/10.1038/s41586-025-09849-4Sourced
The challenge of interfacial delamination and degradation from mechanical stress. A two-layer buffer in which a “loose” SnOx layer made by adjusting the ALD purge time dissipates strain energy and a dense SnOx layer maintains electrical contact. Softening of ion bombardment during sputtering. Silicon 60 µm thick, certified 33.4% on 1 cm² and certified 29.8% on 260 cm², and a maximum of 1.77 W/g. More than 97% after 43,000 bends at a radius of curvature of about 40 mm, and about 97% after 250 cycles between −40 and 85 °CFang et al. (Nature, 2025). Reference 11 https://doi.org/10.1038/s41586-025-09835-wSourced
That the 29.8% on 260 cm² was certified by Fraunhofer ISE and the small-area value by NREL. That space and vehicle applications were named (the applications are the company's outlook)LONGi news release (13 November 2025). Reference 12 https://www.longi.com/en/news/breakthroughs-in-hibc-and-flexible-silicon-based-tandem-solar-cells-nature/Sourced
A multi-branched polymer with dopamine groups inspired by mussel underwater adhesion, 24.43% for flexible cells, and 94.1% after 10,000 bends at a 3 mm radius and 65% humidityLi et al. (Nature Communications, 2025). Reference 13 https://doi.org/10.1038/s41467-025-57102-3Sourced
That conventional high-efficiency cells used TiO2 made at about 500 °C, which is incompatible with flexible cells on plasticFu et al. (Nature Communications, 2015). Reference 14 https://doi.org/10.1038/ncomms9932Sourced
Surface strains (about 0.075%, 0.15% and 1.6%). About 0.1 W/g for film-type modules, about 15 kg/m² for silicon (back-calculated from one-tenth), about 140 g/m² and about 2.1 W/g for a 60 µm wafer, and 23 ÷ 0.1 = about 230 times. The proportional scale in Fig. 1Our calculation. ε ≈ t/(2R) is a simple formula assuming the neutral plane at mid-thickness. The 1 mm radius, 125 µm thickness and silicon density of 2.33 g/cm³ are this article's assumptions. 150 W/m² and 1.5 kg/m² are the strategy's estimation conditions, not measurements. Cells and modules are different subjectsOur calculation
Long-term outdoor durability of film-type cells, progress towards the deployment and cost targets, and the strain at which ITO cracksNot stated because they could not be confirmed in the primary sources within the scope of this article (commentary by this article)Not yet confirmed
Framing flexibility as a redesign of every layer. The reading of nanoparticles as moving crystallisation forward to synthesis. The framing of two options, heat-resistant film or low-temperature deposition. Reading the “loose” buffer layer as the same idea as primers, stress-relief layers and interlayers. Treating wet adhesion as a research theme for adhesive chemists. The reading that the packaging sets module weight. The point that few efficiency lists use uniform criteria. The two-way split into strengths and challengesThis article's own framing and commentary based on published content. Not views expressed by the institutions or researchersCommentary
That Figs. 1, 2, 4 and 5 are explanatory drawings, and that the hero image and Fig. 3 are AI-generated imagesA note by this article (commentary)Commentary

Last updated 25 September 2026. Sources are limited to primary material (METI's strategy document, peer-reviewed papers and official company announcements). No market estimates from research firms are used. Every efficiency value is given with whether it is certified and, where it could be confirmed, its area. The cell areas for the certified 25.55% of Zhong et al. and the certified 33.6% of Wang et al., the basis for the 1.77 W/g of Fang et al., the strain at which ITO cracks, and the long-term outdoor durability of film-type cells are not stated, because they could not be confirmed in published primary sources. All figures are for explanation. Figs. 1, 2, 4 and 5 are vector drawings, and the hero image and Fig. 3 are AI-generated images; none of them shows a real cross-section, micrograph or physical product.

🌐 Japanese