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

Perovskite Basics Explained | Perovskite Solar Cells

Back to Technology, Institution & Company Guides

TECHNOLOGY EXPLAINER

Perovskite Basics
— how a semiconductor you can dissolve and coat actually works

The light-absorbing layer of a perovskite solar cell is a crystal defined by how three kinds of ion arrange themselves: ABX3. Japan's Ministry of Economy, Trade and Industry (METI) describes the technology as “a Japan-born technology developed by domestic researchers”. After covering the crystal structure, the composition and the tolerance factor, this article explains in materials terms why such a thin film can absorb nearly all the light, and why it can be made at around 150 °C.

Built from primary sources: METI, NLR (formerly NREL), the solar cell efficiency tables that began in Progress in Photovoltaics (now published in Joule), and peer-reviewed papers / Last updated September 2026

Conceptual image of a plain glass plate resting on a dark surface, with a smooth dark brown thin film spread evenly across it
AI-generated concept image. An impression of the idea that a coated and dried solution turns into a black semiconductor film. It does not represent any real product, or the actual colour, thickness or surface condition of a film.
What this article covers
  1. What a perovskite is, in three points
  2. The ABX3 crystal structure: a framework of octahedra and the gaps between them
  3. What fills the three sites: MA, FA and Cs; lead; iodine and bromine
  4. Our calculation: working out the tolerance factor
  5. A materials engineer's view (1): the tolerance factor is only a rough entry test
  6. Why such a thin film can absorb nearly all the light
  7. Why it can be made by coating
  8. A materials engineer's view (2): easy to make is the flip side of easy to break
  9. Where efficiency stands: always read the number together with the area
  10. Strengths, weaknesses and open issues
  11. 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 or target with no confirmed track record
Structural readings and materials-design interpretations are marked separately as Commentary.

1. What a perovskite is, in three points

Strictly speaking, “perovskite” is not the name of a particular substance but the name of a type of crystal structure. What solar cells use is one family within that type, known as metal halide perovskites.

  • What it is made of: METI's strategy document describes perovskite solar cells as “the general term for solar cells whose power-generating layer uses a material in which three kinds of ion (typically A: organic ammonium, B: lead, X: iodine) are arranged in the ABX3 perovskite crystal structure”Sourced
  • Where it started: in 2009 the group of Tsutomu Miyasaka used CH3NH3PbI3 in a dye-sensitised photoelectrochemical cell and reported a conversion efficiency of 3.8%Sourced. METI calls it “a Japan-born technology developed by domestic researchers”Sourced
  • How far it has come: single-junction perovskite cells measured by an independent laboratory have reached 28.0% on a small 0.05 cm² cell and 26.9% on a cell of about 1 cm²Sourced (details in Section 9)
The single most important line in this article

A perovskite is an ionic crystal that also happens to be a very good semiconductor. Because it is an ionic crystal, it can be made from solution at low temperature. As a semiconductor, it absorbs light strongly and shrugs off many defects. And for the same reason, its ions move under heat, moisture and light, which makes it fragile. This article shows that all three grow from the same root (our commentary).

2. The ABX3 crystal structure: a framework of octahedra and the gaps between them

The perovskite structure consists of a framework of BX6 octahedra — six X (halide) ions around a central B (metal) ion — joined at their corners, plus a large cation A that sits in the gaps. Bartel and colleagues define the perovskite structure as ABX3 compounds in which a network of corner-sharing BX6 octahedra surrounds a larger A-site cationSourced.

The ABX3 perovskite framework (schematic projected onto 2D) Diamonds = BX6 octahedra seen from above. Neighbours share their corner (X) ions A A sits in the gap framed by four octahedra A site: a large monovalent cation Methylammonium (MA), formamidinium (FA), Cs B site: a divalent metal ion Lead (Pb²⁺) is standard; tin (Sn²⁺) is also studied X site: a halide ion Mainly iodide (I⁻), mixed with Br⁻ or Cl⁻ Charge balance: A(+1) + B(+2) + 3 × X(−1) = 0 Note: framework definition from Bartel et al. [Ref. 5]; typical ions per site from METI [Ref. 1] and papers [Refs. 3, 7, 8]. Note: real crystals are 3D, and MAPbI3 is tetragonal, not cubic, at room temperature (Whitfield et al. [Ref. 4]).
Fig. 1 Concept diagram (vector drawing). The definition of the framework follows Bartel et al. [Ref. 5]; the typical ions for each site follow METI [Ref. 1] and the individual papers [Refs. 3, 7 and 8]. Circle sizes and spacing do not accurately represent ratios of ionic radii. The drawing is a projection of a three-dimensional structure onto a plane.

MAPbI3 is not cubic at room temperature

In 1978 Weber reported that CH3NH3PbX3 (X = Cl, Br, I) adopts a cubic perovskite structure, with lattice constants of 5.68 Å for Cl, 5.92 Å for Br and 6.27 Å for ISourced. That, however, describes the high-temperature phase. Neutron and synchrotron diffraction by Whitfield and colleagues showed that MAPbI3 transforms from cubic to tetragonal at around 330 K and to orthorhombic at around 160 K, both being first-order transitionsSourced.

Our calculation: the transition temperature in Celsius

330 K − 273 = about 57 °COur calculation. A solar panel outdoors in full sun can reach that sort of temperature. Having a structural phase transition inside the operating temperature range is a property worth keeping in mind when MAPbI3 is used as the absorber (our commentary). How the transition affects performance and lifetime depends on composition and device structure, and this article does not assess it.

3. What fills the three sites: MA, FA and Cs; lead; iodine and bromine

Designing a composition comes down to choosing what goes into each of the three sites. The table below gives the ionic radii used in this article and the properties the papers report for each composition.

SiteIonIonic radius (Å)Properties reported in the papers
AMethylammonium (MA⁺, CH3NH3⁺)2.16MA-based bandgaps are about 1.55 eV or higher (Eperon et al.)
AFormamidinium (FA⁺, HC(NH2)2⁺)2.53FA-based bandgaps can be tuned from 1.48 to 2.23 eV via the halide (Eperon et al.). At room temperature FAPbI3 is more stable as a yellow hexagonal phase (δH) (Li et al.)
ACaesium (Cs⁺)1.88At room temperature CsPbI3 settles into an orthorhombic non-perovskite phase (δO) (Li et al.)
BLead (Pb²⁺)1.19 (Shannon) / 1.03 (revised value for iodides)The standard B-site metal
XIodide (I⁻)2.20Cubic lattice constant of MAPbI3: 6.27 Å (Weber)
XBromide (Br⁻)1.96Cubic lattice constant of MAPbBr3: 5.92 Å (Weber)
XChloride (Cl⁻)1.85Cubic lattice constant of MAPbCl3: 5.68 Å (Weber)

All Sourced. Radii for MA⁺, FA⁺, Pb²⁺ (both values) and the halides are from Travis et al. [Ref. 6] and for Cs⁺ from Bartel et al. [Ref. 5]; properties are from Eperon et al. [Ref. 8], Li et al. [Ref. 7] and Weber [Ref. 3]. The “radius” of an organic molecular ion is an effective value that treats it as a sphere.

Moving down the X site from I to Br to Cl shrinks the lattice and widens the optical bandgap. Stoumpos and colleagues reported that four iodides combining MA or FA with Sn or Pb are direct-gap semiconductors with bandgaps spread between 1.25 and 1.75 eVSourced. Bandgap design itself is covered in detail in our explainer on bandgap and composition design.

4. Our calculation: working out the tolerance factor

The classic way to estimate, from ionic radii alone, whether A fits neatly into its gap is the Goldschmidt tolerance factor, t.

Definition of the tolerance factor

t = (rA + rX) ÷ [√2 × (rB + rX)]

t = 1 is a perfect fit. Travis and colleagues explain that perovskites generally form for 0.8 ≤ t ≤ 1; above 1 the A ion is too large, and below 0.8 it is too small, so other structures tend to formSourced. Li and colleagues likewise state that 0.8 to 1.0 favours the cubic perovskiteSourced.

Our calculation: t for three lead iodide perovskites

Using the radii in the table aboveOur calculation:

  • With the Shannon radius of 1.19 Å for Pb²⁺: the denominator is √2 × (1.19 + 2.20) = 4.794. MAPbI3 = 4.36 ÷ 4.794 = 0.91; FAPbI3 = 4.73 ÷ 4.794 = 0.99; CsPbI3 = 4.08 ÷ 4.794 = 0.85
  • With the revised iodide value of 1.03 Å from Travis et al.: the denominator is √2 × (1.03 + 2.20) = 4.568. MAPbI3 = 0.95; FAPbI3 = 1.04; CsPbI3 = 0.89

Assumptions and limits: the organic cations are given effective radii that treat them as spheres. The choice of radius table moves t by about 0.05. t is a guide to how readily the structure forms, not a guarantee of phase stability.

Tolerance factor t of lead iodide perovskites (our calculation) Band = the 0.8 to 1.0 range in which perovskites are said to form readily 0.750.800.85 0.900.951.00 1.051.10 0.8 ≤ t ≤ 1.0 Pb radius 1.19 Å Cs 0.85 MA 0.91 FA 0.99 Pb radius 1.03 Å Cs 0.89 MA 0.95 FA 1.04 Note: radii from Travis et al. [Ref. 6] (MA, FA, Pb, I) and Bartel et al. [Ref. 5] (Cs). t is our calculation, not a published value. Note: in reality FAPbI3 and CsPbI3 are more stable in non-perovskite phases at room temperature (Li et al. [Ref. 7]).
Fig. 2 Drawing that includes our calculation (vector drawing). Ionic radii follow Travis et al. [Ref. 6] and Bartel et al. [Ref. 5]; the 0.8 to 1.0 range follows Travis et al. [Ref. 6] and Li et al. [Ref. 7]. The values of t were calculated by this article and are not published figures. Two Pb radii are used to show that the result shifts with the choice of radius table.

5. A materials engineer's view (1): the tolerance factor is only a rough entry test

A materials engineer's view: inside the range, yet yellow at room temperature

Fig. 2 shows that with the Shannon radius for Pb, MA, FA and Cs all fall inside the 0.8 to 1.0 rangeOur calculation. In practice, though, as Li and colleagues report, FAPbI3 is more stable at room temperature as a yellow hexagonal phase (δH), and CsPbI3 settles into an orthorhombic non-perovskite phase (δO)Sourced.

This is no accident. Testing 576 ABX3 compounds, Bartel and colleagues found that the Goldschmidt t classified 74% of them correctly overall, but only 33% of the iodidesSourced. Travis and colleagues also showed that t cannot accurately predict the stability of the 32 known inorganic iodide perovskitesSourced.

So what worked? Li and colleagues mixed the too-large FA with the too-small Cs to tune an “effective t”, raising the stability of the black phase of FA0.85Cs0.15PbI3 and also improving its stability under high humiditySourced.

This is solid-solution design, exactly as practised for alloys and ceramics. If a single composition will not hold its phase, mix in elements of different size until the average fits. That many of today's high-efficiency cells use “multi-cation” compositions mixing MA, FA and Cs is an extension of the same idea (our commentary). The flip side is that a mixed composition needs separate assurance that it stays mixed. The separation of halides under illumination is covered in our explainer on ion migration and degradation.

6. Why such a thin film can absorb nearly all the light

METI's strategy document compares silicon cells with (film-type) perovskite cells and gives light absorption coefficients of “~10⁴/cm” for silicon and “~10⁵/cm” for perovskiteSourced. De Wolf and colleagues measured the absorption spectrum of MAPbI3 thin films and found a high absorption coefficient and a particularly sharp absorption edge. The exponential absorption tail just below the bandgap (the Urbach energy) was a small 15 meV, which they take as a sign of a well-ordered microstructureSourced.

Our calculation: how much a 400 nm film absorbs

Using the Beer–Lambert law, the fraction absorbed in a single pass is A = 1 − exp(−α × d)Our calculation.

  • Assumption: take α at the orders of magnitude in METI's comparison (perovskite 1×10⁵ cm⁻¹, silicon 1×10⁴ cm⁻¹) as representative values
  • Assumption: take the thickness d as 400 nm (= 4×10⁻⁵ cm), the absorber thickness METI uses in its own calculations
  • Perovskite: α × d = 10⁵ × 4×10⁻⁵ = 4, so A = 1 − e⁻⁴ = about 98%
  • To reach the same α × d with a silicon-like α needs d = 4 ÷ 10⁴ = 4 µm (ten times thicker). At 400 nm, A = 1 − e⁻⁰·⁴ = about 33%

Assumptions and limits: real absorption coefficients change by orders of magnitude with wavelength, and silicon's falls further at long wavelengths. Reflection, the second pass after the back electrode and light trapping are all ignored. The calculation exists only to show that a tenfold difference in α means a tenfold difference in the thickness needed.

Film thickness and single-pass absorptance (our calculation) A = 1 − exp(−αd). Reflection and light trapping are not considered 100%75%50% 25%0% 00.51.0 1.52.0 µm Film thickness α = 10⁵ cm⁻¹: about 98% α = 10⁴ cm⁻¹: about 33% 400 nm Red: perovskite-level α Blue: silicon-level α Note: orders of α from METI's comparison table [Ref. 1]. Curves and absorptances are our calculation, not measurements.
Fig. 3 Drawing that includes our calculation (vector drawing). The orders of magnitude of the absorption coefficients (10⁵ cm⁻¹ and 10⁴ cm⁻¹) and the 400 nm thickness follow METI [Ref. 1]. The curves and absorptances (about 98% and about 33%) were calculated by this article and are not measured spectra. Real absorption coefficients vary widely with wavelength.

A thin film is enough because carriers travel far

Absorbing light is not enough on its own: if the electrons and holes it creates recombine before reaching the electrodes, no current flows. Stranks and colleagues reported that in a mixed halide (CH3NH3PbI3−xClx) the electron and hole diffusion lengths exceed 1 µm, an order of magnitude longer than the absorption depth, whereas in MAPbI3 they are about 100 nmSourced.

As background, Yin and colleagues showed from first-principles calculations that the dominant intrinsic defects in MAPbI3 create only shallow levels, attributing this to the antibonding coupling between the lead lone pair (6s) and iodine p orbitals, together with strong ionicitySourced. This “defect tolerance” is thought to be one reason why even polycrystalline, solution-coated films perform well. Defects do not disappear, however, and the treatment of surfaces and grain boundaries still decides efficiency; see our explainer on passivation.

7. Why it can be made by coating

A perovskite crystallises when halide salts such as lead iodide (PbI2) and methylammonium iodide (MAI) are dissolved in a solvent, coated and heated. In 2012 two groups reported solid-state cells on porous scaffolds, improving stability over the earlier liquid-electrolyte cellsSourced: Miyasaka (Toin University of Yokohama) and Snaith (University of Oxford) with colleagues reached up to 10.9% on a porous alumina scaffold (Lee et al.), and a group including Park of Sungkyunkwan University in Korea and Grätzel reached 9.7% on porous titania (Kim et al.). METI likewise states that “research and development accelerated after 2012, when researchers in the UK and Japan jointly developed a solid-state perovskite solar cell with improved stability”Sourced.

From solution to crystalline film: the coating process (our summary) Top = process flow / bottom = processing temperatures compared by METI 1 Precursor ink Dissolve halide salts such as PbI2 and MAI in a polar solvent 2 Coating Spread thinly on a substrate (spin, slot-die and so on) 3 Intermediate Pass through a phase that holds solvent, evening out nucleation 4 Heat, crystallise The solvent leaves, giving a polycrystalline perovskite film Processing temperature (METI comparison table) Silicon 1,400 °C or more Perovskite 150 °C Note: step 3 follows Jeon et al. [Ref. 15] (routing through a DMSO-containing phase). Steps 1, 2 and 4 are a general outline. Note: temperatures from METI's comparison table [Ref. 1]. Bar length is proportional to temperature (our drawing).
Fig. 4 Concept diagram (vector drawing). The role of the intermediate phase follows Jeon et al. [Ref. 15]; the processing temperatures (silicon 1,400 °C or more, perovskite 150 °C) follow METI's comparison table [Ref. 1]. The division into four steps is this article's own framing and does not show any particular company's manufacturing process.

The difficulty with coating lies in controlling where nuclei form and how crystals grow during the brief moment the solvent is leaving. Jeon and colleagues coated from a mixed solvent of γ-butyrolactone and DMSO, then dripped toluene onto the film. This routed the film through an MAI–PbI2–DMSO intermediate phase and produced an extremely uniform, dense layer, giving a cell with a certified efficiency of 16.2%Sourced. Coating methods themselves are covered in detail in our explainers on solution coating and on roll-to-roll processing and scale-up.

Conceptual image, seen from a low angle, of a thin layer of pale yellow transparent liquid on a plain glass plate turning into a dark brown-black film from one edge
Fig. 5 AI-generated concept image. An impression of solution coating, in which a pale solution turns into a black crystalline film as it dries. It does not represent real coating equipment, or the actual colour of a solution or appearance of a film.

8. A materials engineer's view (2): easy to make is the flip side of easy to break

A materials engineer's view: a crystal that assembles at 150 °C can come apart near 150 °C

METI's comparison table gives processing temperatures of “1,400 °C or more” for silicon and “150 °C” for perovskite, and production times of “3 days or more” for silicon against a target of “about 1 day” for perovskiteSourced.

That is a large advantage in equipment cost and production throughput. Seen from the materials side, though, it means something else as well. A crystal that assembles from its raw salts at low temperature and in a short time has relatively weak bonds, and its constituent ions move easily (our commentary).

  • The advantage for film making: it crystallises from solution at atmospheric pressure and low temperature
  • The other face of the same property: heat, moisture, light and electric fields can move ions through the lattice and change its composition
  • What it implies for design: both “making the material tougher” (composition, additives, interface treatment) and “protecting it from outside” (encapsulation, barriers) have to be considered

De Wolf and colleagues showed that deliberately letting moisture in strongly reduced absorption at photon energies below 2.4 eV and changed the compositionSourced. In other words, even the excellent light absorption is lost once moisture gets in. The degradation mechanisms are covered in our explainer on ion migration and degradation, and protection in our explainer on encapsulation and barrier layers.

9. Where efficiency stands: always read the number together with the area

The efficiency of a perovskite solar cell means entirely different things depending on how large the device was and who measured it. NLR (the National Laboratory of the Rockies, formerly the US National Renewable Energy Laboratory, NREL) includes only efficiencies confirmed by independent, recognised test centres in its Best Research-Cell Efficiency ChartSourced.

MeasuredEfficiencyAreaMade byMeasured by
May 201314.1%0.209 cm²EPFLNewport
March 201622.1%0.0946 cm²KRICT/UNISTNewport
July 201925.2%0.0937 cm²KRICT/MITNewport
March 202326.0%0.0746 cm²ISCASJET
July 202527.3%0.1065 cm²Soochow Univ. and othersNPVM
February 202628.0%0.051 cm²Hainan Univ.NPVM

All Sourced (main single-junction perovskite rows extracted from NLR's research-cell efficiency data table [Ref. 16]). The 3.8% of 2009 [Ref. 2] and the 9.7% and 10.9% of 2012 [Refs. 13 and 14] are values reported in papers and do not appear in this data table.

The “Solar cell efficiency tables (Version 68)” by Green and colleagues (July 2026, published in Joule), on the other hand, separate the tables that treat devices of 1 cm² or more as “records” from those listing smaller devices. According to that edition, a 1.017 cm² perovskite cell reached 26.9%, a 19.48 cm² minimodule 23.9% and a 756.0 cm² submodule 22.9%Sourced. For the 28.0% cell, the tables note, in effect, that it is the first to match the best silicon efficiency but is more than 3,000 times smaller, and that its area is too small for it to be classed as an outright recordSourced.

The larger the area, the lower the efficiency (single-junction perovskite) Bar height = efficiency measured by a third party; figures below = device area 28.0% 26.9% 23.9% 22.9% Cell (small area) 0.05 cm² Cell 1.017 cm² Minimodule 19.48 cm² Submodule 756.0 cm² Note: efficiency and area from Green et al., efficiency tables Version 68 [Ref. 17]. Bar height is proportional to efficiency. Note: 756.0 ÷ 0.05 = about 15,000x (our calculation). Four devices of different makers and designs, not one scaled up.
Fig. 6 Drawing that includes our calculation (vector drawing). Efficiencies and areas are published values from Green et al. [Ref. 17]. The area ratio (about 15,000 times) was calculated by this article. The four devices differ in maker and structure, so the chart does not show how much efficiency falls when one and the same cell is made larger.
Our calculation: getting a feel for the difference in area
  • 756.0 cm² submodule ÷ 0.05 cm² small cell = about 15,000 timesOur calculation
  • Difference in efficiency: 28.0 − 22.9 = 5.1 percentage pointsOur calculation

As area grows, non-uniform thickness and composition, pinholes and losses in the interconnects between cells start to matter (our commentary). Whenever you see “perovskite efficiency of X%”, the first practical step is to check the area and who measured it. Measurement methods are covered in our explainer on conversion efficiency and performance metrics.

10. Strengths, weaknesses and open issues

AspectStrength (confirmed in primary sources)Weakness or issue
Light absorptionAbsorption coefficient ~10⁵/cm (METI); sharp absorption edge (De Wolf et al.)Absorption falls when moisture gets in (De Wolf et al.)
Charge transportDiffusion length over 1 µm in the mixed halide (Stranks et al.)About 100 nm in MAPbI3 alone (Stranks et al.)
DefectsDominant intrinsic defects form shallow levels (Yin et al.)Surfaces and grain boundaries still need treatment
ManufacturingProcessing temperature of 150 °C (METI); crystallises from solutionNucleation and crystal growth are hard to control
Phase stabilityMixing on the A site can stabilise the black phase (Li et al.)On their own, FAPbI3 and CsPbI3 are more stable in non-perovskite phases at room temperature (Li et al.); MAPbI3 has a phase transition at about 330 K (Whitfield et al.)
Efficiency28.0% on a small-area cell (efficiency tables Version 68)Efficiency falls as area increases

Each item is Sourced (source in brackets). Sorting them into strengths and weaknesses is this article's own framing.

(1) Long-term outdoor durability is not assessed in this article

For film-type cells, METI's strategy document states that “further technology development to improve durability remains necessary to bring down the cost of generation”Sourced. How many years a module lasts outdoors depends not only on the materials but on encapsulation and installation conditions. This article gives no figure for service life. See our explainer on outdoor testing and durability assessment.

(2) It contains lead

Standard compositions use lead on the B site. METI's material states that “the lead contained, at around 0.5 g/m², must be properly treated and recovered”Sourced. The handling of lead and the supply of iodine are covered in our explainer on lead and iodine, and lead-free compositions in our explainer on tin-based perovskites.

(3) The deployment target is a target

METI's Next-Generation Solar Cell Strategy (Japan's national roadmap for perovskite and other new solar cells) aims for deployment of about 20 GW by 2040Not yet confirmed, but this is a policy goal, not an achievement.

The article in summary
  • “Perovskite” names a type of crystal structure: a framework of BX6 octahedra with A in the gapsSourced
  • Calculated tolerance factors for MA, FA and Cs all lie between 0.85 and 1.04Our calculation, yet for iodides the prediction is right only 33% of the timeSourced
  • Mixing FA and Cs to match an “effective t” — solid-solution design — stabilised the black phaseSourced
  • With an absorption coefficient ten times higher, 400 nm absorbs about 98%Our calculation
  • The processing temperature is 150 °C (silicon: 1,400 °C or more)Sourced. That ease of manufacture is the other face of how easily its ions move (our commentary)
  • Read efficiency together with area. 28.0% at 0.05 cm², 22.9% at 756 cm²Sourced

11. Glossary

Perovskite structure
A crystal structure type of composition ABX3, in which A sits in the gaps of a framework of corner-sharing BX6 octahedra.
Metal halide perovskite
A perovskite whose X site holds a halogen such as iodine, bromine or chlorine. Used as the absorber in solar cells.
MA and FA
Methylammonium (CH3NH3⁺) and formamidinium (HC(NH2)2⁺), organic cations that occupy the A site.
Tolerance factor (t)
An index that estimates from ionic radii whether a perovskite structure can form. The closer to 1, the better the fit.
δ phase (yellow phase)
The non-perovskite phase that FAPbI3 and CsPbI3 tend to adopt at room temperature. It absorbs little light.
Solid solution
A state in which several components are mixed within one crystal. Mixing on the A site is an example.
Absorption coefficient (α)
How strongly light is attenuated as it travels through a material. 1/α is roughly the absorption depth.
Urbach energy
The breadth of the absorption tail just below the bandgap. The smaller it is, the less disordered the crystal.
Diffusion length
The average distance a photogenerated electron or hole can travel before recombining.
Shallow level
A defect level at an energy close to a band edge. Such levels rarely act as recombination centres.
Intermediate phase
A crystal that incorporates solvent molecules. Routing a coated film through one makes a uniform film easier to obtain.
Certified efficiency
An efficiency measured under standard conditions by an independent test laboratory, as distinct from an in-house measurement.

12. References (primary sources)

  1. Ministry of Economy, Trade and Industry (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
  2. Kojima, Teshima, Shirai, Miyasaka “Organometal Halide Perovskites as Visible-Light Sensitizers for Photovoltaic Cells”, J. Am. Chem. Soc. 131, 6050 (2009) https://doi.org/10.1021/ja809598r
  3. Weber “CH3NH3PbX3, ein Pb(II)-System mit kubischer Perowskitstruktur”, Z. Naturforsch. B 33, 1443 (1978) https://doi.org/10.1515/znb-1978-1214
  4. Whitfield et al. “Structures, Phase Transitions and Tricritical Behavior of the Hybrid Perovskite Methyl Ammonium Lead Iodide”, Sci. Rep. 6, 35685 (2016) https://doi.org/10.1038/srep35685
  5. Bartel et al. “New tolerance factor to predict the stability of perovskite oxides and halides”, Sci. Adv. 5, eaav0693 (2019) https://doi.org/10.1126/sciadv.aav0693
  6. Travis et al. “On the application of the tolerance factor to inorganic and hybrid halide perovskites: a revised system”, Chem. Sci. 7, 4548 (2016) https://doi.org/10.1039/C5SC04845A
  7. Li et al. (NREL) “Stabilizing Perovskite Structures by Tuning Tolerance Factor: Formation of Formamidinium and Cesium Lead Iodide Solid-State Alloys”, Chem. Mater. 28, 284 (2016) https://doi.org/10.1021/acs.chemmater.5b04107
  8. Eperon et al. “Formamidinium lead trihalide: a broadly tunable perovskite for efficient planar heterojunction solar cells”, Energy Environ. Sci. 7, 982 (2014) https://doi.org/10.1039/C3EE43822H
  9. Stoumpos, Malliakas, Kanatzidis “Semiconducting Tin and Lead Iodide Perovskites with Organic Cations: Phase Transitions, High Mobilities, and Near-Infrared Photoluminescent Properties”, Inorg. Chem. 52, 9019 (2013) https://doi.org/10.1021/ic401215x
  10. De Wolf et al. “Organometallic Halide Perovskites: Sharp Optical Absorption Edge and Its Relation to Photovoltaic Performance”, J. Phys. Chem. Lett. 5, 1035 (2014) https://doi.org/10.1021/jz500279b
  11. Stranks et al. “Electron-Hole Diffusion Lengths Exceeding 1 Micrometer in an Organometal Trihalide Perovskite Absorber”, Science 342, 341 (2013) https://doi.org/10.1126/science.1243982
  12. Yin, Shi, Yan “Unusual defect physics in CH3NH3PbI3 perovskite solar cell absorber”, Appl. Phys. Lett. 104, 063903 (2014) https://doi.org/10.1063/1.4864778
  13. Kim et al. “Lead Iodide Perovskite Sensitized All-Solid-State Submicron Thin Film Mesoscopic Solar Cell with Efficiency Exceeding 9%”, Sci. Rep. 2, 591 (2012) https://doi.org/10.1038/srep00591
  14. Lee, Teuscher, Miyasaka, Murakami, Snaith “Efficient Hybrid Solar Cells Based on Meso-Superstructured Organometal Halide Perovskites”, Science 338, 643 (2012) https://doi.org/10.1126/science.1228604
  15. Jeon et al. “Solvent engineering for high-performance inorganic–organic hybrid perovskite solar cells”, Nat. Mater. 13, 897 (2014) https://doi.org/10.1038/nmat4014
  16. NLR (National Laboratory of the Rockies, formerly NREL) “Best Research-Cell Efficiency Chart” and research-cell efficiency data table https://www.nlr.gov/pv/cell-efficiency
  17. Green et al. “Solar cell efficiency tables (Version 68)”, Joule 10, 102494 (2026) https://doi.org/10.1016/j.joule.2026.102494

13. Claim-to-source audit

Claim in the textBasisLabel
That perovskite solar cells are “the general term for solar cells whose power-generating layer uses a material in which three kinds of ion (typically A: organic ammonium, B: lead, X: iodine) are arranged in the ABX3 perovskite crystal structure, and a Japan-born technology developed by domestic researchers”. That “research and development accelerated after 2012, when researchers in the UK and Japan jointly developed a solid-state perovskite solar cell with improved stability”. Light absorption coefficients of “~10⁴/cm” for silicon and “~10⁵/cm” for perovskite. Processing temperatures of “1,400 °C or more” for silicon and “150 °C” for perovskite. Production times of “3 days or more” for silicon and a target of “about 1 day” for perovskite. That a 400 nm absorber thickness is used in METI's calculations. That film-type cells still need “technology development to improve durability”. That “the lead contained, at around 0.5 g/m², must be properly treated and recovered”METI, Next-Generation Solar Cell Strategy (November 2024), pages 11, 12, 17, 52 and 54. Reference 1 https://www.meti.go.jp/shingikai/energy_environment/perovskite_solar_cell/pdf/20241128_1.pdfSourced
The aim of deploying about 20 GW by 2040METI, Next-Generation Solar Cell Strategy, page 32. A policy goal, not an achievement. Reference 1 https://www.meti.go.jp/shingikai/energy_environment/perovskite_solar_cell/pdf/20241128_1.pdfNot yet confirmed
That a photoelectrochemical cell using CH3NH3PbI3 achieved a conversion efficiency of 3.8% (2009, Kojima, Miyasaka et al.)J. Am. Chem. Soc. paper (abstract). Reference 2 https://doi.org/10.1021/ja809598rSourced
That CH3NH3PbX3 (X = Cl, Br, I) adopts a cubic perovskite structure with lattice constants of 5.68 Å (Cl), 5.92 Å (Br) and 6.27 Å (I)Z. Naturforsch. B paper (abstract). Reference 3 https://doi.org/10.1515/znb-1978-1214Sourced
That MAPbI3 transforms from cubic to tetragonal at around 330 K and to orthorhombic at around 160 K, both first-order transitionsSci. Rep. paper (abstract). Reference 4 https://doi.org/10.1038/srep35685Sourced
The definition of the perovskite structure (a network of corner-sharing BX6 octahedra surrounding A). The formula for the tolerance factor. That t classified 74% correctly overall, 51% of chlorides, 56% of bromides and 33% of iodides. That 576 ABX3 compounds were used in the test. The Cs⁺ radius of 1.88 ÅSci. Adv. paper (full text). Reference 5 https://doi.org/10.1126/sciadv.aav0693Sourced
That perovskites generally form for 0.8 ≤ t ≤ 1, with A too large for t > 1 and too small for t < 0.8. MA⁺ 2.16 Å, FA⁺ 2.53 Å, the Shannon radius of Pb²⁺ of 1.19 Å and the revised iodide value of 1.03 Å, I⁻ 2.20 Å, Br⁻ 1.96 Å, Cl⁻ 1.85 Å. That t cannot accurately predict the stability of the 32 known inorganic iodide perovskitesChem. Sci. paper (full text and Table 1). Reference 6 https://doi.org/10.1039/C5SC04845ASourced
That 0.8 to 1.0 favours the cubic perovskite. That FAPbI3, with a large t, is more stable at room temperature as the hexagonal δH phase (yellow phase), and CsPbI3, with a small t, stabilises at room temperature in the orthorhombic δO phase. That alloying FA and Cs tunes an effective t and raises the stability of the α phase. That FA0.85Cs0.15PbI3 films showed improved stability in a high-humidity environmentChem. Mater. paper (abstract). Reference 7 https://doi.org/10.1021/acs.chemmater.5b04107Sourced
That MA-based bandgaps are about 1.55 eV or higher. That FA-based bandgaps can be tuned between 1.48 and 2.23 eVEnergy Environ. Sci. paper (abstract). Reference 8 https://doi.org/10.1039/C3EE43822HSourced
That four iodides combining MA or FA with Sn or Pb are direct-gap semiconductors with bandgaps between 1.25 and 1.75 eVInorg. Chem. paper (abstract). Reference 9 https://doi.org/10.1021/ic401215xSourced
That MAPbI3 thin films show a high absorption coefficient and a sharp absorption edge, with an Urbach energy of 15 meV. That after moisture ingress absorption below 2.4 eV fell strongly and the composition changedJ. Phys. Chem. Lett. paper (abstract). Reference 10 https://doi.org/10.1021/jz500279bSourced
That in the mixed halide (CH3NH3PbI3−xClx) the diffusion length exceeds 1 µm, an order of magnitude longer than the absorption depth. That in MAPbI3 it is about 100 nmScience paper (abstract). Reference 11 https://doi.org/10.1126/science.1243982Sourced
That the dominant intrinsic defects in MAPbI3 create only shallow levels, owing to the antibonding coupling between the lead lone-pair s orbital and iodine p orbitals and to strong ionicityAppl. Phys. Lett. paper (abstract). Reference 12 https://doi.org/10.1063/1.4864778Sourced
That a solid-state mesoscopic cell achieved a conversion efficiency of 9.7% with better stability than liquid-electrolyte cells. That the authors include Park of Sungkyunkwan University and GrätzelSci. Rep. paper (abstract). Reference 13 https://doi.org/10.1038/srep00591Sourced
That a solid-state cell using porous alumina achieved up to 10.9%. That the authors include Tsutomu Miyasaka of Toin University of Yokohama and Snaith of the University of OxfordScience paper (bibliographic record and summary). Reference 14 https://doi.org/10.1126/science.1228604Sourced
That coating from a mixed solvent of γ-butyrolactone and DMSO, followed by dripping toluene, routed the film through an MAI–PbI2–DMSO intermediate phase to give a uniform, dense film. The certified efficiency of 16.2%Nat. Mater. paper (abstract). Reference 15 https://doi.org/10.1038/nmat4014Sourced
That NLR lists only efficiencies confirmed by independent, recognised test centres. The single-junction perovskite rows (May 2013, 14.1%, 0.209 cm², EPFL, Newport; March 2016, 22.1%; July 2019, 25.2%; March 2023, 26.0%; July 2025, 27.3%; February 2026, 28.0%, 0.051 cm², Hainan U, NPVM). The renaming of NREL as NLRNLR “Best Research-Cell Efficiency Chart” page and research-cell efficiency data table. Reference 16 https://www.nlr.gov/pv/cell-efficiencySourced
That a 0.05 cm² perovskite cell reached 28.0%, matching the best silicon efficiency on a cell more than 3,000 times smaller, and is too small in area to be classed as a record. That government research programmes generally use areas of 1 cm² or more. The 1.017 cm² cell at 26.9%, the 19.48 cm² minimodule at 23.9% and the 756.0 cm² submodule at 22.9%“Solar cell efficiency tables (Version 68)”, Joule. Reference 17 https://doi.org/10.1016/j.joule.2026.102494Sourced
The conversion of 330 K to about 57 °C. The tolerance factors (Shannon radius: MA 0.91, FA 0.99, Cs 0.85; revised radius: MA 0.95, FA 1.04, Cs 0.89). The single-pass absorptance of a 400 nm film (about 98% for α = 10⁵ cm⁻¹, about 33% for 10⁴ cm⁻¹) and the 4 µm needed for the same absorption. The area ratio 756.0 ÷ 0.05 = about 15,000 times, and the 5.1-point difference in efficiencyOur calculation. Treating organic cations as spheres with effective radii, taking the absorption coefficients as order-of-magnitude representative values, and ignoring reflection and light trapping are assumptions set by this article. Includes the numerical parts of Figs. 2, 3 and 6Our calculation
The reading that ease of manufacture and ease of ion movement are two sides of the same coin. Framing the tolerance factor in terms of solid-solution design. The list of reasons why efficiency falls at large area. The sorting into strengths and weaknesses. The significance of a phase transition within the operating temperature rangeOur summary and commentary based on published content. Not views expressed by the institutions or authorsCommentary
Outdoor service life, and the effect of the phase transition on lifetimeWithin the scope of this article, no primary source generalisable across compositions and structures could be confirmed, so no figures are given (our note)Commentary
That Figs. 1 to 4 and Fig. 6 are explanatory drawings, not real crystal images, manufacturing processes or measured data. That the hero image and Fig. 5 are AI-generated imagesOur noteCommentary

Last updated 25 September 2026. Sources are limited to primary material (a government strategy document, NLR's efficiency data and peer-reviewed papers). Because the article includes structural readings and materials-design interpretations, those are marked as Commentary and kept separate from sourced fact. Outdoor service life and the effect of the phase transition on lifetime are not stated, because no generalisable primary source could be confirmed. The content of papers is limited to what could be confirmed in the abstract or the openly available full text. All figures are explanatory concept graphics. Figs. 1 to 4 and Fig. 6 are vector drawings; the hero image and Fig. 5 are AI-generated images, and none of them shows a real crystal, micrograph or physical product.

🌐 Japanese