MENU
Advanced Materials Intelligence
  1. Home
  2. Roll-to-Roll and Scale-Up Explained | Perovskite Solar Cells

Roll-to-Roll and Scale-Up Explained | Perovskite Solar Cells

Back to Technology, Institution & Company Guides

TECHNOLOGY EXPLAINER

Roll-to-Roll Processing and Scale-Up
— why efficiency falls as the area grows

The highest perovskite solar cell efficiencies come from small cells of less than 1 cm². To make a module, a large film has to be cut into narrow strips and connected in series, and those cuts, together with the resistance of the electrodes, eat into efficiency. This article sorts out P1 to P3 scribing, the relationship between area and efficiency, and where roll-to-roll (R2R) production stands today, from a materials engineer's point of view.

Built from primary sources: peer-reviewed papers (NREL, the University of Rome Tor Vergata, CSIRO and others), Japan's Next-Generation Solar Cell Strategy from METI, and published material from Sekisui Chemical / Last updated September 2026

Conceptual image of a thin, dark continuous film running between plain metal rollers against a dark background
AI-generated concept image. An impression of roll-to-roll production, in which a rolled substrate is fed through and processed continuously. It does not represent any real production equipment, product, or film colour or structure.
What this article covers
  1. What scale-up means, in three points
  2. Not one big sheet but narrow strips in series: P1, P2 and P3 scribes
  3. Active area and aperture area: the efficiency figure depends on how area is defined
  4. Four reasons efficiency falls as area grows
  5. Our calculation: there is an optimum strip width
  6. A materials engineer's view (1): the sheet resistance of the transparent electrode sets the strip width
  7. A materials engineer's view (2): P2 is a contact-resistance problem
  8. Roll-to-roll: a demonstration with every step done on rolls
  9. Our calculation: annealing time sets the length of the line
  10. Pilot and volume-production plans in Japan
  11. Open problems, and what this article could not confirm
  12. Glossary / References / Claim-to-source audit
How claims are labelled in this article

Sourced = stated in published material or a peer-reviewed paper (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 or process interpretations are marked separately as Commentary.

1. What scale-up means, in three points

  • No single large cell: the NREL review says that making one cell on a large substrate is impractical, because resistive losses grow when current has to travel a long way through the transparent electrode. Instead, a large cell is divided into small sub-cells that are connected in series to form a moduleSourced
  • The interconnects do not generate power: the series interconnections take up roughly 3 to 10% of the module area and produce nothing, which is described as a major source of module lossesSourced
  • R2R means making it on rolls: a flexible substrate is unwound from a roll, coated, dried and processed continuously, then wound up again. In 2024 a group led by CSIRO (Australia's Commonwealth Scientific and Industrial Research Organisation) reported the first series-connected modules made entirely on R2R printing equipment, in ambient airSourced
The single most important line in this article

Efficiency falls at large area not only because of uneven coating. Design-determined losses are always added on top: area that does not generate power, needed to cut the film into strips, and the resistance of the transparent electrode. A group including the University of Rome Tor Vergata splits the loss from cell to module into three parts, geometric loss from the interconnect area, resistive loss in the transparent electrode, and resistive loss at the P2 contact, and then assesses the contribution of defects as the gap between calculation and measurementSourced.

2. Not one big sheet but narrow strips in series: P1, P2 and P3 scribes

As with CdTe, CIGS and other thin-film solar cells, perovskite modules are made by depositing each layer over the whole substrate and then cutting grooves with a laser or mechanical blade to separate the cells. There are three kinds of groove, called P1, P2 and P3Sourced.

Series connection by P1, P2 and P3 scribes (schematic cross-section) Bottom up: substrate / TCO / active layers (transport + perovskite) / top electrode; light enters via the substrate Substrate (glass or film) P1 P2 P3 Top electrode Active layers TCO Cell A Cell B Red dashes = current: A top electrode → P2 → B TCO Inactive: P1 to P3 Note: P1 splits the TCO, P2 links top and bottom electrodes, P3 splits the top electrode [References 1 and 3]. Note: layer thicknesses, groove widths and spacing are all schematic and are not real proportions.
Fig. 1 Conceptual diagram (vector drawing). The roles of P1, P2 and P3 follow the NREL review [Reference 1] and Di Giacomo et al. [Reference 3]. Layer thicknesses, groove widths and spacing are schematic, not real proportions. The actual layer stack (which electrode light enters through, and which transport layers are used) varies between products and studies.
GrooveWhat it cutsPurposeExamples of reported dimensions
P1The bottom transparent electrode (TCO)Insulates the bottom electrodes of neighbouring cellsA single laser line 15 µm wide gave adequate insulation
P2Every layer above the transparent electrode (leaving the electrode itself)Lets the top electrode, applied afterwards, touch the next cell's transparent electrode, connecting them in seriesWidths of 60 to 300 µm studied, with parallel lines overlapped at 10 µm spacing to build up width
P3The top electrode (and the layers below it)Separates the top electrodes of neighbouring cellsGood insulation at 90 to 150 µm. Harder than P1, because the layers under the top electrode are easily damaged by laser heat

All Sourced (Di Giacomo et al. 2020 [Reference 3]; an example using an ultraviolet nanosecond laser on an inverted (p-i-n) structure). The inactive zone between P1 and P3 consists of the three grooves plus two safety margins.

The NREL review describes laser scribing as low-cost, fast and precise and well suited to volume production, noting that an all-laser process has demonstrated modules with a geometric fill factor (GFF) of 94%. At the same time it points out that perovskites are more heat-sensitive than other thin-film materials, so material near the groove may degrade and ablated material may redepositSourced.

Another approach is to print each layer only where it is needed instead of cutting grooves. But each layer then has to be aligned, and the resolution is lower than a laser's, so the inactive area tends to growSourced.

3. Active area and aperture area: the efficiency figure depends on how area is defined

A module from above: active area, aperture area and GFF (schematic) blue = generating cells (active area) / thin grey = P1 to P3 interconnects (inactive) / heavy outline = aperture area 1 2 3 4 5 Example with five cells in series (the number is arbitrary) Defining the area Active area: sum of the generating cells Aperture: lit area incl. interconnects GFF = active area ÷ aperture area Aperture eff. = active-area eff. × GFF Example: 15% on active area, GFF 80% → 12% on aperture area Note: the definitions and the example (15% with GFF 80% gives 12%) follow the NREL review [Reference 1]. Note: cell count and widths are schematic, not real proportions (real interconnects are much narrower).
Fig. 2 Conceptual diagram (vector drawing). The area definitions and worked example follow the NREL review [Reference 1]. The interconnects are drawn wide for legibility and are not to scale.

The NREL review notes that for conventional solar cells it is established practice to report efficiency on the aperture area (the designated illumination area), which includes all the interconnects. Some perovskite module efficiencies, however, are reported on an "active area" that excludes the interconnects, and the review cautions that such figures cannot be compared properly unless the GFF is given alongsideSourced.

ReportAreaActive-area efficiencyAperture-area efficiencyGFF
Rakocevic et al. 20174 cm²—up to 15.3%up to 94%
Yang et al. 2018 (NREL)aperture about 10.36 cm²17.9%15.6% (stabilised)about 87.3%
Di Giacomo et al. 2020active 10.2 cm²15.9%14.5%90.9%
Nikbakht et al. 2025aperture 156 cm² (active 150.4 cm²)17.68%Not found in the relevant parts of the abstract or text96.4%

All Sourced ([References 4, 5, 3 and 6]). All are measurements within the papers; the abstracts do not say whether they were certified by a third party. Setting them side by side is this article's own arrangement, and the results differ in structure, materials and measurement conditions.

4. Four reasons efficiency falls as area grows

Nikbakht et al. (2025) plotted module efficiencies from the literature against active area and concluded that in general, the larger the area, the lower the efficiency. As reasons they cite variation in efficiency from imperfect film coverage and uniformity, poorer charge extraction from a higher defect density in the perovskite layer, and higher series resistance from the high sheet resistance of the transparent electrode, explaining that large cells are split into small sub-cells connected in series to limit resistive lossesSourced.

From cell to module: four factors that erode efficiency (our framing) 1 Interconnect area 2 TCO resistance 3 P2 contact resistance 4 Defects, unevenness The P1 to P3 zone is inactive The narrower the strips, the larger its share Hits aperture efficiency Current flows sideways through the TCO, with a Joule loss along the way Grows with width squared Residual layers or debris at the top electrode-TCO contact add resistance Depends on laser settings Poor coverage, pinholes Higher defect density Shunts, weak cells Chance of hitting one rises with area 1 to 3 can largely be calculated from the design / 4 is about film quality and yield Note: 1 to 3 follow Di Giacomo et al. [Reference 3]; 4 follows Nikbakht et al. [Reference 6] and NREL [Reference 1]. Note: "grows with width squared" follows the Di Giacomo equation; "rises with area" is our commentary. Note: setting out four factors is our framing and does not show their relative size.
Fig. 3 Conceptual diagram (vector drawing). The classification of losses 1 to 3 follows Di Giacomo et al. [Reference 3]; the factors in 4 follow Nikbakht et al. [Reference 6] and the NREL review [Reference 1]. Box size does not indicate the size of the loss. The reading that the chance of hitting a defect rises with area is this article's commentary.

On point 4, the NREL review says that when evaluating scale-up, it is important to locate shunts, weak diodes and high-resistance regions with PL and EL (luminescence) imaging and lock-in thermography. A shunted sub-cell appears dark in EL, and a shunt shows up bright in thermography because of Joule heatingSourced. Nikbakht et al. also point out that as area grows, the number of connected cells increases, series resistance rises and the fill factor falls, and that defects arising from P2 processing can also matter in large cellsSourced.

In practice, the CSIRO-led R2R work reached a best of 15.5% for small-area cells but 11.0% for a series-connected large-area module (active area about 50 cm²)Sourced. Nikbakht et al. report that scaling from large-area cells to modules up to 150 times larger kept the loss in efficiency below 19%Sourced.

5. Our calculation: there is an optimum strip width

Di Giacomo et al. (2020) used a thin-film module model and expressed the losses as followsSourced. Here w_active is the width of the generating part of a cell, w_int the width of the interconnect from P1 to P3, R_SH the sheet resistance of the transparent electrode, and J_MPP and V_MPP the current density and voltage at the maximum power point.

  • Geometric loss from the interconnect area: f_G = w_int ÷ (w_int + w_active)
  • Resistive loss in the transparent electrode: f_TCO = (J_MPP ÷ V_MPP) × (R_SH ÷ 3) × w_active³ ÷ (w_active + w_int)

For its worked example the paper assumes R_SH = 15 Ω/□, J_MPP = 19 mA/cm² and V_MPP = 850 mVSourced. Using those equations and assumptions, we calculate the losses as the cell width changes.

Our calculation: cell width and losses (aperture-area basis)

AssumptionsOur calculation

  • The equations and the values of R_SH, J_MPP and V_MPP are those of Di Giacomo et al. [Reference 3]
  • Two interconnect widths w_int: 0.5 mm, as in that paper's prototype, and the 0.18 mm P1-to-P3 spacing of Nikbakht et al. [Reference 6]
  • The two losses are simply added as an approximation; P2 contact resistance and defect losses are not included

Outline of the calculation: J_MPP ÷ V_MPP × R_SH ÷ 3 = 190 A/m² ÷ 0.85 V × 15 Ω ÷ 3, or about 1,118 /m² (0.001118 /mm²). Multiply this by the cube of the cell width and divide by (cell width + interconnect width).

Cell widthInterconnect area loss
(w_int 0.5 mm)		Transparent electrode resistive lossTotal
(w_int 0.5 mm)		Total
(w_int 0.18 mm)
3 mm14.3%0.9%15.1%6.6%
5 mm9.1%2.5%11.6%6.2%
7 mm6.7%5.1%11.8%7.8%
10 mm4.8%10.6%15.4%12.7%
15 mm3.2%24.3%27.6%26.0%
  • With a 0.5 mm interconnect: the loss is smallest at a cell width of about 5.8 mm, for a total of about 11.4%
  • With a 0.18 mm interconnect: the minimum is at a cell width of about 4.2 mm, for a total of about 6.0%

With a 5 mm cell and a 0.5 mm interconnect, the GFF is 5 ÷ 5.5 = 90.9%, matching the GFF of the Di Giacomo prototype module (90.9%). The paper expects a cell-to-module loss of about 3% at these dimensionsSourced. That figure is on an active-area basis, and is consistent with the resistive loss in our calculation (about 2.8% on an active-area basis).

Assumptions and limits: adding the losses together is an approximation, and P2 contact resistance, defects and shunts, current collection at the electrode edges and differences in illumination are not included. The calculation is meant to show the structure of the design problem, that strips which are too narrow or too wide both cost you; it does not predict the efficiency of any particular module.

Cell width and losses (our calculation, 0.5 mm interconnect) R_SH = 15 Ω/□, J_MPP = 19 mA/cm², V_MPP = 0.85 V (the assumptions of Di Giacomo et al. [Reference 3]) 0 4 8 12 16 20 mm Width of the generating part of the cell, w_active 0% 10% 20% 30% 40% Minimum: 11.4% at 5.8 mm Total TCO resistance Interconnect area How to read it Narrower strips: interconnect share grows Wider strips: TCO resistive loss grows with width squared → an optimum width exists Note: the curves are our calculation on the assumptions above, not published values; P2 contact, defects and shunts are excluded.
Fig. 4 Drawing including our calculation (vector drawing). The loss equations and assumed values (15 Ω/□, 19 mA/cm², 0.85 V) follow Di Giacomo et al. [Reference 3]. The curves and the minimum (about 5.8 mm, about 11.4%) are values calculated by this article, not published values. The approximation of simply adding the losses is used.

6. A materials engineer's view (1): the sheet resistance of the transparent electrode sets the strip width

Why this matters for materials engineers: "transparent" and "conductive" are hard to have at once

Di Giacomo et al. write that a transparent conductive oxide (TCO) cannot be made thick without losing transparency, so its sheet resistance is limited to roughly 7 to 60 Ω/□Sourced. They estimate that using the 7 Ω/□ readily obtained with fluorine-doped tin oxide (FTO) would allow wider cells and bring the minimum aperture-area loss down to 4.6%, but add that a more conductive transparent electrode is less transparent, so overall efficiency may not improve, and that going below 7 Ω/□ requires adding a current-collecting grid, which in turn has to be balanced against the grid's shadingSourced.

In the equation in Section 5, the resistive loss in the transparent electrode is proportional to sheet resistance × the square of the cell width. So halving the sheet resistance of the transparent electrode allows cells about 1.4 times (√2 times) wider for the same loss, cutting the number of interconnectsOur calculation.

The module layout is worked backwards from the properties of a single material, the transparent electrode (the trade-off between sheet resistance and transmittance). Improve one deposition condition or dopant in the transparent conductive film, and the cell width, the number of scribes, the laser processing time and the GFF all change in a chain. This is a field where improving a material directly changes the production design (our commentary). Transparent electrode materials themselves are covered in detail in our explainer on transparent conductive films.

7. A materials engineer's view (2): P2 is a contact-resistance problem

Why this matters for materials engineers: a 10 nm film left at the bottom of a groove can make or break a module

Yang and colleagues at NREL (2018) made modules with a spray-coated TiO2 electron transport layer and blade-coated perovskite and hole transport layers. What they found was that the TiO2 layer left in the sub-cell interconnect (P2) strongly affected module performance. Thinning the TiO2 changed the interconnect contact from Schottky-diode-like behaviour to ohmic behaviour, and 10 nm of TiO2 improved mainly the fill factorSourced.

  • A four-cell series module gave a stabilised efficiency of 15.6% on an aperture area of about 10.36 cm², 17.9% on an active-area basis, and a GFF of about 87.3%Sourced
  • Di Giacomo et al. assessed the quality of the P2 contact by its transfer length (L_T), calculating that with an L_T of 0.06 mm, any P2 wider than 10 µm keeps the resistive loss below 0.1%Sourced
  • The same paper did not judge P2 optimisation by its appearance under an optical microscope alone, but combined energy-dispersive X-ray spectroscopy (EDX) with transfer-length measurementsSourced

So P2 is both a machining problem of cutting a groove and a contact-resistance problem of how to join an interface between dissimilar materials. The selectivity of the laser in removing the upper layers cleanly without damaging the layer beneath, the electrical character of the thin film left at the bottom of the groove (rectifying or ohmic), and the redeposition of ablated debris all change a great deal with materials-side design: layer thickness, light absorption, thermal conductivity and interface chemistry (our commentary). The choice of charge transport layers is covered in our explainer on charge transport layers and cell architecture.

8. Roll-to-roll: a demonstration with every step done on rolls

Conceptual image of a flat film being fed forward from a roll of thin film wound on a plain core, against a dark background
Fig. 5 AI-generated concept image. An impression of a flexible substrate being fed from a roll for processing. It does not represent any real product, equipment, or film structure or colour.

In 2024 a group led by CSIRO (Australia's Commonwealth Scientific and Industrial Research Organisation) reported in Nature Communications perovskite solar cell modules made entirely by R2R. The authors describe it as the first demonstration of modules of series-connected cells made solely with industrial R2R printing equipment, in ambient airSourced.

ItemWhat the paper states
Substrate and layer stackPET / transparent electrode / SnO2 / FA0.45MA0.55PbI3 / HTAB / P3HT / carbon / Ag
Eliminating vacuum stepsReplaced the costly vacuum-evaporated metal electrode with a printed carbon electrode
Perovskite coatingSlot-die, web speed 0.3 m/min, coating width 13 mm. A 10 cm wide air knife placed about 10 cm behind the coating head blew nitrogen onto the film
Condition screening1,600 cells made under 20 sets of conditions were evaluated in one batch, optimising a wide parameter space in a short time
EfficiencyBest 15.5% for small-area cells; 11.0% for a series-connected large-area module (active area about 50 cm²)
Cost outlookAssuming production of 1 million m² a year, a projection of about US$0.7/W in Australia (the paper's own cost model)Not yet confirmed

Everything except the cost outlook is Sourced (Weerasinghe et al. 2024 [Reference 7]). The US$0.7/W figure is a projection from the authors' cost model, not an actual manufacturing cost.

Earlier, in 2018, Dou et al. (ACS Energy Letters) used high-speed in-situ X-ray diffraction to confirm that perovskite films printed at room temperature in air show high crystallinity and orientation within 1 second of coating, and reported up to 19.6% for cells with every layer blade-coated, 17.3% for slot-die-coated cells, and 14.1% for cells on flexible glass substrates with some steps done by R2R slot-die coatingSourced.

The NREL review points out that the major constraint in R2R is annealing (heat-treatment) time. Annealing the perovskite layer usually takes 5 to 120 minutes, and long times would make an R2R line impractically longSourced. The next section puts numbers on what that means.

The flow of roll-to-roll production (schematic) The film runs from the left roll to the right roll at constant speed. Each section's length = speed × time needed Unwind Rewind Coating Slot-die etc. Drying and annealing This tends to get long Next layer Coat or print Patterning Stripes, processing Web direction Section length = web speed × dwell time: one slow step makes the whole line longer or slower Note: the step order is one example; the CSIRO-led work used separate R2R machines and steps per layer [Reference 7]. Note: that annealing constrains the length of an R2R line follows the NREL review [Reference 1].
Fig. 6 Conceptual diagram (vector drawing). That annealing time constrains an R2R line follows the NREL review [Reference 1]. The order of steps and their relative lengths are schematic and do not show any particular production line. In the CSIRO-led demonstration [Reference 7], separate R2R machines and steps were used for each layer.

9. Our calculation: annealing time sets the length of the line

Our calculation: the length needed for a drying and annealing oven

Assumptions (set by this article)Our calculation

  • Oven length = web speed × dwell time (the time spent passing through the oven), ignoring festooned or serpentine web paths
  • Dwell times are the 5 to 120 minutes given in the NREL review, plus the 1 minute of the shortened example
Web speed1 min5 min30 min120 min
0.3 m/min (the CSIRO-led perovskite coating)0.3 m1.5 m9 m36 m
3 m/min (= 180 m/h)3 m15 m90 m360 m
10 m/min10 m50 m300 m1,200 m

180 m/h is a blade-coating speed reported in research, and the Park and Zhu review notes that on a 1 m wide line it would correspond to a production capacity of 236 MW a yearSourced (also discussed in our explainer on solution coating). Even if you can coat at that speed, an anneal of 30 minutes needs a 90 m oven. Cut the dwell time to 1 minute and 3 m is enough.

Assumptions and limits: real equipment may fold the web back and forth inside the oven, or shorten heating times with infrared or light. The calculation is meant to show that the faster you coat, the more the time taken by downstream steps drives line length; it does not give the dimensions of any particular equipment.

The NREL review cites an example where adjusting the precursor chemistry cut annealing from over 10 minutes to under 1 minute, and another where near-infrared heating shortened a 45-minute treatment to 2.5 seconds without a major loss of performance. It also notes that early attempts to cut the time to 1.5 milliseconds with a xenon flash gave non-uniform films and poor performanceSourced.

10. Pilot and volume-production plans in Japan

Japan's Next-Generation Solar Cell Strategy (November 2024), compiled by a public-private council convened by METI (the Ministry of Economy, Trade and Industry), divides perovskite solar cells into film-type (flexible), glass-type and tandem. It states that in Japan some companies are scheduled to begin commercialisation from FY2025, and that for film-type cells, Japan leads the world technically in durability and in making larger panelsSourced. The strategy also says that much of a product's competitiveness comes from manufacturing-process know-how, "complex material processing and forming, and control of temperature and humidity, that are not embodied in the manufacturing equipment" (our translation)Sourced.

As a concrete film-type example, Sekisui Chemical has published its own development and volume-production plans. What follows is the company's own announcement.

DateWhat Sekisui Chemical has announced
January 2025 (business briefing slides)On development status: "R2R at 30 cm width, moving toward establishing 1 m-wide manufacturing technology" (our translation); conversion efficiency of 15% achieved (aiming for 20%); durability equivalent to 10 years achieved (aiming for 20 years). The challenges named are improving production yield and developing installation and construction methods
26 December 2024The board resolved to move to volume production, aiming for a 100 MW production line in operation in 2027 with total investment of ¥90 billion (including subsidies, covering the building purchase and the 100 MW production equipment). Equipment operation is scheduled to start on 1 April 2027. The company aims to build a GW-scale production line by 2030
Same announcementSelected for METI's GX Supply Chain Support Program (a subsidy scheme for building domestic supply chains for decarbonisation, or green transformation). Total eligible expenditure ¥314.5 billion, total subsidy ¥157.25 billion, subsidy rate one half, production capacity 1 GW-scale
27 March 2026Announced the launch of the "SOLAFIL" business. Manufacturing technology established on existing equipment and preparations completed for commercialising products for metal roofs. Limited production volumes on existing equipment in FY2026 (year to March 2027)

All Sourced from Sekisui Chemical's announcements [References 9, 10 and 11]. However, the measurement conditions for the 15% efficiency (cell or module, area, third-party certification or not) and the method behind "equivalent to 10 years" are not stated in the material. The 100 MW line and GW-scale production are plansNot yet confirmed.

11. Open problems, and what this article could not confirm

(1) We could not consult the latest certified efficiency tables

The "Solar cell efficiency tables", which list certified efficiencies by area category (the latest being Version 68, July 2026), could not be accessed in full from this article's working environment, so the latest certified records by area category are not given here. All efficiency figures in this article are values stated in individual peer-reviewed papers. How to read certified efficiencies is covered in our explainer on conversion efficiency and performance metrics.

(2) Scribing and coating conditions on production lines

Sekisui Chemical lists "precision reactive coating of the four power-generating layers (nanometre level)" and "fine processing technology (50 to 100 µm)" among its differentiating technologies (our translation)Sourced, but the specific coating method, scribing method and cell width could not be confirmed in the published material this article consulted, so they are not stated.

(3) R2R efficiency still trails sheet-to-sheet processing

The CSIRO-led fully R2R module (active area about 50 cm²) reached 11.0%Sourced, below both the small-area cells in the same paper (15.5%) and the reported values for modules made by sheet-to-sheet processing on glass substrates. How much efficiency can be kept with large-area, high-speed, fully R2R production remains an open questionNot yet confirmed.

(4) Outdoor trials of large panels have only just begun

Nikbakht et al. connected 156 cm² modules into a 0.73 m² panel, encapsulated it by hot vacuum lamination with a commercial thermoplastic film, obtained up to 12.0% under outdoor conditions, and evaluated it at a test site on Crete, GreeceSourced. The paper also mentions earlier cases in which 1 m²-class panels did not work after fabrication, and in which efficiency was not achieved because of problems in the lamination of the encapsulantSourced. Encapsulation and lamination are covered in our explainer on encapsulation and barrier layers, and outdoor evaluation in our explainer on outdoor testing and durability assessment.

The article in summary
  • At large area, no single big cell is made; the film is cut into strip cells that are connected in seriesSourced
  • P1 splits the transparent electrode, P2 connects the top and bottom electrodes, and P3 splits the top electrode. The interconnects do not generate powerSourced
  • Efficiency changes depending on whether it is quoted on active area or aperture area. Comparisons need the GFF alongsideSourced
  • The losses are interconnect area, transparent electrode resistance, P2 contact resistance and defects. On the paper's assumptions, the loss is smallest (about 11.4%) at a cell width of about 5.8 mmOur calculation
  • Fully R2R modules were demonstrated in 2024, reaching 11.0% on an active area of about 50 cm²Sourced
  • Sekisui Chemical plans to move from 30 cm-wide R2R to 1 m width, with a 100 MW line operating in 2027Not yet confirmed

12. Glossary

Roll-to-roll (R2R)
A production method in which a rolled flexible substrate is unwound, processed continuously and wound up again.
Sheet-to-sheet
A production method in which substrates are processed one at a time. Common with glass substrates.
Sub-cell
One of the individual strip-shaped cells that make up a module.
P1, P2 and P3 scribes
The three kinds of groove used for monolithic series connection: separating the bottom electrode, connecting top and bottom electrodes, and separating the top electrode.
Active area
The part of a module made up of cells that actually generate power.
Aperture area
The illuminated area, including the interconnects. Conventional solar cells report efficiency on this area.
GFF (geometric fill factor)
Active area ÷ aperture area. The narrower the interconnects, the larger it is.
Sheet resistance (Ω/□)
The in-plane electrical resistance of a thin film, expressed as a value independent of the size of the square.
Transparent electrode (TCO)
An oxide film that conducts electricity while letting light through, such as ITO or FTO.
Transfer length (L_T)
The distance over which current fully transfers into the electrode at a contact. The shorter it is, the lower the contact resistance.
Fill factor (FF)
A measure of how "square" the current-voltage curve is. It falls as series resistance rises.
Annealing
Heating a coated film to drive crystallisation and repair defects.

13. References (primary sources)

  1. Li Z, Klein TR, Kim DH, Yang M, Berry JJ, van Hest MFAM, Zhu K (NREL) “Scalable fabrication of perovskite solar cells”, Nature Reviews Materials 3, 18017 (2018) (DOE Public Access accepted manuscript) https://www.osti.gov/servlets/purl/1430821
  2. Park NG, Zhu K “Scalable fabrication and coating methods for perovskite solar cells and solar modules”, Nature Reviews Materials 5, 333–350 (2020) (DOE Public Access accepted manuscript) https://www.osti.gov/servlets/purl/1605079
  3. Di Giacomo F, Castriotta LA et al. “Upscaling Inverted Perovskite Solar Cells: Optimization of Laser Scribing for Highly Efficient Mini-Modules”, Micromachines 11, 1127 (2020) (full text on PMC) https://pmc.ncbi.nlm.nih.gov/articles/PMC7767295/
  4. Rakocevic L et al. “Interconnection Optimization for Highly Efficient Perovskite Modules”, IEEE Journal of Photovoltaics 7, 404–408 (2017) https://doi.org/10.1109/jphotov.2016.2626144
  5. Yang M et al. (NREL) “Highly Efficient Perovskite Solar Modules by Scalable Fabrication and Interconnection Optimization”, ACS Energy Letters 3, 322–328 (2018) https://doi.org/10.1021/acsenergylett.7b01221
  6. Nikbakht H et al. “Upscaling Perovskite Photovoltaics: from 156 cm² Modules to 0.73 m² Panels”, Advanced Science 12, 2416316 (2025) (full text on PMC) https://pmc.ncbi.nlm.nih.gov/articles/PMC12165116/
  7. Weerasinghe HC et al. (CSIRO and others) “The first demonstration of entirely roll-to-roll fabricated perovskite solar cell modules under ambient room conditions”, Nature Communications 15, 1656 (2024) (full text on PMC) https://pmc.ncbi.nlm.nih.gov/articles/PMC10933357/
  8. Dou B et al. “Roll-to-Roll Printing of Perovskite Solar Cells”, ACS Energy Letters 3, 2558–2565 (2018) https://doi.org/10.1021/acsenergylett.8b01556
  9. Sekisui Chemical “Notice regarding volume production of perovskite solar cells”, 26 December 2024 (PDF, in Japanese) https://www.sekisui.co.jp/news/2024/__icsFiles/afieldfile/2024/12/26/241226.pdf
  10. Sekisui Chemical “Perovskite solar cell business briefing”, presentation slides, 7 January 2025 (PDF, in Japanese) https://www.sekisui.co.jp/ir/event/other/__icsFiles/afieldfile/2025/01/07/20250107PVK.pdf
  11. Sekisui Chemical and Sekisui Solar Film “Notice of the launch of the film-type perovskite solar cell business, SOLAFIL”, 27 March 2026 (in Japanese) https://www.sekisui.co.jp/news/2026/1450530_42699.html
  12. Public-Private Council for Expanding the Deployment of Next-Generation Solar Cells and Strengthening Industrial Competitiveness (METI) “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

14. Claim-to-source audit

Claim in the textBasisLabel
That making a single cell on a large substrate is impractical because of resistive loss in the transparent electrode, so it is split into sub-cells connected in series. That the interconnects take up about 3 to 10% of module area and are a major source of loss. The definition of GFF. That efficiency is conventionally reported on the aperture area (designated illumination area), that active-area reports need the GFF alongside, and the example of 15% with GFF 80% giving 12%. The roles of P1 to P3. That laser scribing is low-cost, fast and precise, with an all-laser example at GFF 94%. That perovskites are heat-sensitive, risking degradation and redeposition near the groove. That partial deposition by printing needs alignment and tends to lower the GFF. That PL, EL and lock-in thermography locate shunts and high-resistance regions. That annealing of 5 to 120 minutes would make an R2R line impractically long; the reduction from over 10 minutes to under 1 minute; near-infrared shortening 45 minutes to 2.5 seconds; and the early xenon-flash attempt at 1.5 ms giving non-uniform films and poor performanceReference 1 https://www.osti.gov/servlets/purl/1430821Sourced
The blade-coating speed of 180 m/h reported in research, and the statement that it would correspond to a production capacity of 236 MW a year on a 1 m wide lineReference 2 https://www.osti.gov/servlets/purl/1605079Sourced
The definitions of P1 to P3, and that the inactive zone consists of three grooves and two safety margins. P1 insulating at 15 µm; P2 widths of 60 to 300 µm with parallel lines at 10 µm spacing; P3 at 90 to 150 µm and harder than P1. The equations splitting loss into geometric, TCO and P2 parts, with defects assessed as the gap from calculation. The assumptions R_SH 15 Ω/□, J_MPP 19 mA/cm² and V_MPP 850 mV. A cell-to-module loss of about 3% at a 5 mm cell and 500 µm interconnect. Active-area 15.9%, aperture-area 14.5%, GFF 90.9%, 10.2 cm². That TCO sheet resistance is limited to about 7 to 60 Ω/□; a minimum loss of 4.6% at 7 Ω/□; that lower transparency may mean no gain in overall efficiency; that below 7 Ω/□ a grid is needed. That with L_T 0.06 mm, a P2 of 10 µm or more keeps loss below 0.1%. The combined use of EDX and transfer-length measurementsReference 3 https://pmc.ncbi.nlm.nih.gov/articles/PMC7767295/Sourced
Up to 15.3% aperture-area efficiency and up to 94% GFF for 4 cm² modules. Optimisation of the P2 interconnectionReference 4 https://doi.org/10.1109/jphotov.2016.2626144Sourced
Modules made with spray-coated TiO2 and blade-coated perovskite and hole transport layers. That TiO2 left in the interconnect strongly affects performance, and that thinning it changes the contact from Schottky to ohmic. FF improvement at 10 nm. A four-cell module with a stabilised 15.6% on an aperture of about 10.36 cm², 17.9% on active area, and GFF about 87.3%Reference 5 https://doi.org/10.1021/acsenergylett.7b01221Sourced
The trend of lower efficiency at larger area and its causes (coverage and uniformity, defect density, series resistance from the TCE sheet resistance). 24 cells, cell width 0.5 cm, aperture 156 cm², active 150.4 cm², P1-to-P3 spacing 180 µm, GFF 96.4%. Active-area 17.68% (Spiro-OMeTAD) and 16.32% (PTAA). A drop of under 19% at 150 times the area. That more cells raise series resistance and lower FF, and defects from P2. The 0.73 m² panel encapsulated by hot vacuum lamination with thermoplastic film, up to 12.0% outdoors, evaluated on Crete. Earlier failures of 1 m²-class panelsReference 6 https://pmc.ncbi.nlm.nih.gov/articles/PMC12165116/Sourced
The first demonstration of series-connected modules made entirely on industrial R2R printing equipment in ambient air. The layer stack, the switch to a carbon electrode, slot-die at 0.3 m/min and 13 mm width, the air knife, evaluation of 1,600 cells under 20 conditions, small-area 15.5% and module (active about 50 cm²) 11.0%. That the authors' cost model projects about US$0.7/W at 1 million m² a yearReference 7 https://pmc.ncbi.nlm.nih.gov/articles/PMC10933357/Sourced
That films printed at room temperature in air showed high crystallinity and orientation within 1 second. All-blade 19.6%, slot-die 17.3%, and 14.1% for flexible-glass cells made partly by R2RReference 8 https://doi.org/10.1021/acsenergylett.8b01556Sourced
The resolution to move to volume production, the 100 MW line in 2027, total investment of ¥90 billion, operation scheduled for 1 April 2027, GW-scale by 2030, and under the GX Supply Chain Support Program eligible expenditure of ¥314.5 billion, a subsidy of ¥157.25 billion, a subsidy rate of one half and 1 GW-scale capacityReference 9 https://www.sekisui.co.jp/news/2024/__icsFiles/afieldfile/2024/12/26/241226.pdfSourced
"R2R at 30 cm width, moving toward establishing 1 m-wide manufacturing technology", 15% conversion efficiency achieved (aiming for 20%), durability equivalent to 10 years achieved (aiming for 20 years), and the challenges of yield and installation methods. "Precision reactive coating of the four power-generating layers (nanometre level)" and "fine processing technology (50 to 100 µm)" (our translations)Reference 10 https://www.sekisui.co.jp/ir/event/other/__icsFiles/afieldfile/2025/01/07/20250107PVK.pdfSourced
The launch of the SOLAFIL business, manufacturing technology established on existing equipment and preparations completed for metal-roof products, limited production in FY2026, and top priority given to starting up the 100 MW line in FY2027Reference 11 https://www.sekisui.co.jp/news/2026/1450530_42699.htmlSourced
The division into film-type, glass-type and tandem; commercialisation by some companies scheduled from FY2025; the view that film-type leads the world in durability and larger panels; and that competitiveness depends heavily on manufacturing-process know-howReference 12 https://www.meti.go.jp/shingikai/energy_environment/perovskite_solar_cell/pdf/20241128_1.pdfSourced
Operation of the 100 MW line, a GW-scale production line, how much efficiency R2R can retain, and the CSIRO-led cost projectionPlans and projections, not actual results. As of this article's research (September 2026), no primary source showing the 100 MW line in operation had been confirmed (commentary)Not yet confirmed
The table and curves of cell width against loss (minimum at about 5.8 mm and about 11.4%; about 4.2 mm and about 6.0% with a 0.18 mm interconnect). The check of GFF 5 ÷ 5.5 = 90.9%. Cell width √2 times at half the sheet resistance. The table of oven length = web speed × dwell timeOur calculation. The equations and assumed values are from Reference 3, the 0.18 mm interconnect from Reference 6, the dwell times from Reference 1, and the speeds from References 7 and 2. Simply adding the losses and excluding P2 contact and defects are this article's assumptionsOur calculation
The certified records by area category in the latest efficiency tables (Version 68). The measurement conditions for Sekisui Chemical's 15% and the evaluation method for "equivalent to 10 years". The coating method, scribing method and cell width on production linesNot stated because the full text could not be accessed from this article's working environment, or because the published material does not give them (commentary)Commentary
Framing the losses as four factors, and the reading that the chance of hitting a defect rises with area. The reading that improving a material directly changes the production design. Framing P2 as a contact-resistance problemThis article's own framing and commentary based on published content. Not views expressed by the institutions or authorsCommentary
That Figs. 1, 2, 3, 4 and 6 are explanatory drawings rather than real cross-sections, equipment or measured data, and that the hero image and Fig. 5 are AI-generated imagesOur note (commentary)Commentary

Last updated 25 September 2026. Sources are limited to primary material (peer-reviewed papers, NREL accepted manuscripts published through DOE Public Access, METI's strategy document and Sekisui Chemical's published material). Because the article includes structural readings and materials and process interpretations, those are marked as Commentary and kept separate from sourced fact. The certified records by area category in the latest efficiency tables (Version 68), the measurement conditions for Sekisui Chemical's efficiency and its durability evaluation method, and the specific process conditions on production lines are not stated because this article could not confirm them. The efficiencies quoted here differ from paper to paper in area definition, certification and measurement method, and cannot be compared directly. All figures are explanatory concept graphics. Figs. 1, 2, 3, 4 and 6 are vector drawings; the hero image and Fig. 5 are AI-generated images, and none of them shows a real cross-section photograph, equipment or physical product.

🌐 Japanese