Editorial responsibility for this article rests with Wuhan Ooi Photoelectric Technology Co., Ltd. Written and reviewed by the Ooitech editorial team. See our editorial policy for how we write, where AI tools are used, and how to request a correction.
Perovskite in Space: What 100 Thermal Shock Cycles Cost
Table of Contents
Quick answer: A satellite in low earth orbit crosses into and out of the earth's shadow about fifteen times a day, and the temperature ramp that produces is roughly four times what the ground thermal-cycling standard allows. In a 2026 study, FAPbI3 perovskite cells with no bromide stabiliser converted to the inactive yellow phase during 100 accelerated cycles and kept about 62% of their starting efficiency, while cells with a few percent of bromide held the black phase and about 80%. The same cells flown to 35 km on a balloon separated the two compositions by an order of magnitude in short-circuit current.
Space is usually described as a radiation problem for solar cells, and for the cell technologies that fly today that description is fair. For perovskites the first obstacle arrives much sooner, and it is thermal. A satellite in low earth orbit spends roughly an hour in sunlight and half an hour in shadow, over and over, and each pass through the terminator is a heating ramp followed by a cooling ramp.
That cycling lands on a material that is not yet stable at room temperature. FAPbI3 sits close to a phase boundary: the black perovskite phase that absorbs light and carries charge can convert to a yellow phase that does neither, and temperature is what pushes the conversion along. The engineering question is therefore not whether a perovskite cell is efficient, but whether it can hold its phase through the thermal environment of an orbit.
A study published in Energy & Environmental Science in 2026 approached that question in two steps. It built a thermal shock programme around measured satellite data rather than around an existing laboratory convention, then took the same cells to 35 km on a high-altitude balloon to check the laboratory result against a near-space environment. This page covers what the two experiments found, where the loss actually occurs inside the cell, and what the result leaves unsettled.
Why the orbit, not the radiation, sets the schedule
Low earth orbit means 200 km to 2,000 km, where an orbit takes about 90 minutes. That yields roughly fifteen sunrise-and-eclipse cycles per day, every day, for the life of the satellite. Each cycle is a step change in the heat the cell receives, and the cell answers it with a temperature ramp.
The study starts from measurements taken on the NOAA-21 weather satellite rather than from assumptions. The measured heating rate is 6.77 °C/min and the measured cooling rate is 1.89 °C/min. The asymmetry is expected: heating comes from absorbed sunlight plus internal dissipation, while cooling depends on radiative loss to deep space, which is slower. Measured surface temperatures on the satellite span −90 °C to +80 °C, inside an ECSS qualification envelope of −175 °C to +125 °C.
Against that, the ground standard is mild by design. IEC 61215 limits the thermal-cycling ramp to 100 °C/h, which is about 1.67 °C/min, and it was written for modules that sit outdoors on the ground and see one cycle a day at most. The measured rate in orbit is about four times what the standard allows.
| Condition | Ramp rate | What it represents |
|---|---|---|
| IEC 61215 thermal cycling | 1.67 °C/min (100 °C/h) | The limit for a terrestrial module test |
| Measured heating, NOAA-21 | 6.77 °C/min | What the orbit does to a satellite surface |
| Measured cooling, NOAA-21 | 1.89 °C/min | Radiative loss to deep space is slower |
| Accelerated test in the study | 16 °C/min | 2.4× the measured rate, 9.6× the standard |
What was actually tested
The test article was an encapsulated perovskite cell with the architecture ITO / SnO2 / perovskite / PEAI / PTAA / Au. The absorber is FAPbI3, the composition that has produced the highest reported laboratory efficiencies for single-junction perovskite devices, above 27% in the literature. That efficiency figure is a laboratory result and not the subject of this study; what the study measures is how long the phase survives.
Stabilisation was done the way the field normally does it, by substituting a small amount of bromide. Five compositions were prepared, with MAPbBr3 at 0%, 1%, 3%, 5% and 7%, so that the effect of the additive could be read as a curve rather than as a single data point.
The thermal programme then applied −80 °C to +80 °C at 16 °C/min for 100 cycles. Two things about that choice matter. First, the window covers the measured satellite surface range with margin and stays inside the ECSS envelope. Second, the ramp is 2.4 times the measured heating rate, so the test is an accelerating stress rather than a recording of one orbit. The authors describe it that way: the point is to rank compositions in days, not to reproduce flight minute by minute.
| Parameter | Value used |
|---|---|
| Temperature window | −80 °C to +80 °C |
| Ramp rate | 16 °C/min |
| Cycles | 100 |
| Absorber compositions | MAPbBr3 at 0, 1, 3, 5 and 7 mol% |
| Measurement set | XRD, KPFM, surface photovoltage, J–V, EQE |
| Second environment | 35 km balloon flight, near-space spectrum and UV |
The black phase is the one that works
Perovskite absorbers are polymorphic, and the two phases that matter here look different and behave differently. The α phase is black and photoactive: it absorbs across the visible range and transports charge. The δ phase is yellow, has a different crystal structure, and does not do the job. In FAPbI3 the α phase is the one that is wanted and the δ phase is thermodynamically close by at room temperature, which is why stabilisation is a composition problem rather than a packaging problem.
A smaller halide is the usual lever. Bromide is a smaller ion than iodide, and a few percent of it incorporated into the lattice contracts the structure and moves the tolerance factor toward the range where the perovskite structure is stable. The study reads the outcome with X-ray diffraction after the 100 cycles, and the result is not monotonic. The 0% sample ends with a large amount of the δ phase. The 1%, 3% and 5% samples keep the α phase. The 7% sample keeps the phase as well but begins to show more PbI2, a decomposition product that indicates the additive has been pushed past its useful range.
That gives the shape of the answer: the useful stabiliser content is a window, not a direction. Below the window the phase converts under cycling, above it a second phase appears, and the practical question for a manufacturer is where the middle of the window sits for a given absorber and a given thermal profile.
Where the loss appears in the data
Efficiency retention after the 100 cycles separates the two ends of the range: about 62% for the 1% composition and about 80% for the 5% composition. Both numbers describe encapsulated cells measured after cycling, and both are reported values from the study rather than measurements taken here.
The current–voltage curves show where the difference comes from. The 1% sample loses short-circuit current and fill factor together, which is the signature of a cell that is losing carriers it has already generated. The 5% sample keeps more of its curve shape. EQE narrows the picture further: the 1% sample falls across the spectrum, while the 5% sample drops mainly between 700 nm and 800 nm, the long-wavelength end where light is absorbed deeper and the rear interface matters more. That pattern points at an interface problem, consistent with a difference in thermal expansion between the perovskite and the layers around it.
The scanning-probe measurements supply the mechanism. Surface-potential mapping shows the 1% sample developing a wider spread of potential between grains after cycling, which is what turns a grain boundary into a recombination centre rather than a conductive path. The 5% sample stays more uniform, and its surface photovoltage is about 1.5 times that of the 1% sample, meaning less non-radiative loss at the surface it was measured on.
The balloon flight at 35 km
Laboratory cycles answer a question about the material, not about the environment, so the study added a second experiment. Working with a university group, the team flew the same cells to 35 km on a high-altitude balloon and measured them on the way up. At that altitude the ambient pressure is about 2% of the ground value, air density about 1.5%, the ambient temperature around −40 °C, and the spectrum close to the AM0 profile used for space, with the ultraviolet content that goes with it. It is the closest thing to a space environment that can be reached without a launch, and it is still not space.
The two compositions separated again, and in the same direction as in the laboratory. The efficiency of the 1% sample declined as altitude increased; the 5% sample rose. The clearest number is the slope of short-circuit current with altitude: 0.00016 for the 1% composition against 0.00364 for the 5% composition, an order of magnitude apart. Extra irradiance cannot help a cell that is already losing carriers at its boundaries before they can be collected, which is what the surface-potential data had already indicated.
What this changes for a cell or module line
Perovskite module production is not yet a mature line process, and this result does not hand anyone a station specification. It does clarify four things for the people who will build those lines.
Composition moves from a paper value to an inspected value
Two percentage points of bromide separate a cell that keeps its phase through 100 cycles from one that does not. On a silicon line, the incoming checks that matter are wafer thickness, resistivity and paste viscosity. On a perovskite line, absorber composition and phase purity sit in the same category, and they need a measurement that reads phase rather than only stoichiometry.
A fast ramp is a usable screen
The value of the method is the ramp rate. Running 100 cycles at 16 °C/min takes days, which means a composition can be ranked before a campaign is committed. Silicon module manufacturing already works this way with damp heat and thermal cycling acceptance tests, where a long test stands in for years outdoors. Perovskite needs the equivalent, and this study is a template for what one might look like. See DH1000 damp heat testing and the wider list of PV module reliability tests for how those acceptance tests are defined on the silicon side.
Adding more of a good additive creates a new problem
The 7% sample shows PbI2 segregation, so the stabiliser has an optimum rather than a direction. The same pattern appears in encapsulation, where a higher UV-absorber loading protects the cell and costs optical transmission, and the answer is a balance point rather than a maximum. The comparison of EVA, POE and EPE encapsulant films covers that trade-off on the silicon side.
Phase damage is a mapping problem
The damage that matters is not uniform across a cell. It concentrates at grain boundaries and at interfaces, and it shows up as a spread in surface potential before it shows up as a drop in a coupon measurement. That makes imaging inspection the natural companion to the chamber test, the same way electroluminescence testing reads cracks and inactive regions on silicon modules. The equipment set used for IEC qualification testing is where that kind of station would sit in a line.
What the study does not settle
100 cycles is a short period in orbital terms. At roughly fifteen cycles a day, the test covers about a week of a satellite's life, and it says nothing about what the same cell does in year three. Everything that follows from that limitation is worth stating plainly.
- The experiment is thermal only. It does not include the ultraviolet dose, atomic oxygen, vacuum outgassing or charged-particle radiation that a satellite meets at the same time, and combined stress is not the same as one stress at a time.
- It covers one absorber composition family, one cell architecture and one encapsulation scheme. A different transport layer or a different barrier may move the numbers.
- 35 km is near space, not orbit. The balloon provides reduced pressure, low temperature and a space-like spectrum, but not the vacuum, the plasma environment or the multi-year time base.
- The laboratory programme and the balloon flight agree on direction. That agreement is evidence that the mechanism is real, and it is not the same as a service-life prediction.
- No data from an operating satellite is reported, because perovskite arrays are not flying in a configuration that would produce it.
FAQ
Why is the temperature swing harder in orbit than on the ground?
A terrestrial module sees one cycle a day and changes temperature slowly, under 1.67 °C/min in the standard test. A satellite in low earth orbit crosses the terminator about fifteen times a day and changes temperature at up to 6.77 °C/min on the measured record. The number of cycles and the rate both increase, and for a material that is metastable at room temperature, the rate is the part that matters.
Does adding bromide always help?
No. Between 1% and 5% MAPbBr3 the black phase survived 100 cycles; at 7% the phase also survived but PbI2 appeared. Below that range the phase converted. The additive works inside a window, and the position of the window is specific to the absorber and the thermal profile it will see.
What does the yellow phase actually cost?
The δ phase absorbs poorly in the visible range and does not transport charge the way the α phase does, so the conversion removes working material rather than merely changing its colour. In this study the composition that converted retained about 62% of its initial efficiency, and the electrical data show current and fill factor falling together.
Could a module maker run this test today?
The chamber requirements are ordinary: a wide temperature range and a controlled ramp. The demanding part is meeting 16 °C/min while holding uniformity across a sample, and then having the measurement set ready for XRD, surface potential and photovoltage rather than only a power reading. The study's conditions are a reasonable starting specification for anyone building such a screen.
Why does one composition gain current with altitude and the other not?
More irradiance produces more carriers in both cells. In the cell with damaged boundaries, a growing share of those carriers recombines before reaching the contacts, so the additional light is largely wasted. In the cell that kept its phase, the additional light converts to additional current, which is why its slope is an order of magnitude steeper.
Is silicon still the safer choice for space?
For a mission flying now, yes. Crystalline silicon and III-V cells have decades of flight history, quantifiable degradation in the range of a few percent over a qualification sequence, and established specifications. Perovskite offers higher efficiency per unit area and the prospect of flexible, lightweight arrays, and it has not demonstrated an orbital lifetime. Read the structure of mainstream space photovoltaics for what that flight history is built on.
Final thoughts
The useful part of this result is not the efficiency figure. It is that a composition decision measurably decides whether a perovskite cell survives a thermal environment, that the decision has an optimum rather than a direction, and that a fast ramp test can separate the candidates in days. Those three statements transfer to any perovskite product, whether the destination is a satellite or a rooftop.
The part to keep in proportion is the test itself. A hundred cycles is a week of orbital life, the chamber applies heat and cold without the rest of the space environment, and the balloon reaches 35 km rather than orbit. The result is a strong screening method and a direction of travel, not a service-life guarantee.
Related reading: Perovskite/silicon tandem cells and the 8 nm ITO interconnect · 26.2% certified all-perovskite tandem modules · Interface reactions in perovskite/silicon tandems · TOPCon, HJT and perovskite compared · Triple-junction GaAs cells for space · DH1000 damp heat test · IEC 61215 and IEC 61730 · PV module reliability tests · EL testing inline versus offline · Testing equipment for IEC certification
Data note: the temperature rates (6.77 °C/min heating, 1.89 °C/min cooling), the qualification ranges, the test conditions (−80 °C to +80 °C, 16 °C/min, 100 cycles), the diffraction and surface-potential results, the efficiency retentions (about 62% and 80%) and the current-versus-altitude slopes (0.00016 and 0.00364) are reported values from the 2026 study published in Energy & Environmental Science. Ooitech did not run these tests. The IEC 61215 ramp limit is quoted as a specification value, and the description of phase stabilisation is general materials reasoning rather than a measured result from the study.
Related: Space solar cover glass: why cerium, and how to read it applies the same environmental reasoning to a different cell technology in orbit.
Further reading and background data
Public reference material on this topic, published by the organisations that maintain the industry's technology roadmap and market data:
