Your stability test is only as harsh as your lamp's UV — and most lamps fade

This is a sponsored article, by Lumartix

Plasma-lit climatic, altitude and thermal-shock chambers that hold the full AM1.5G or AM0 spectrum, UVA + UVB (+ UVC), for the entire duration of the ageing run.

Perovskite research has changed its unit of measurement. Efficiency records still make headlines, but the numbers that now decide whether a device gets published, funded or flown are T80 lifetimes, retention after thousands of hours under continuous illumination, and the ISOS protocol used to get there.

That shift puts an uncomfortable amount of weight on a single instrument: the lamp.

The two blind spots in accelerated ageing

LED simulators stop where the interesting chemistry starts. They are stable, efficient and have become the default for light soaking. But below roughly 350 nm they have little to offer, and UVB and UVC are simply absent from the spectrum. Every degradation pathway driven by high-energy photons stays invisible for the whole test.

Xenon and HMI lamps do emit UV. They just don’t keep emitting it. Electrode erosion, envelope solarisation and quartz devitrification attack the short-wavelength end first. After a few hundred operating hours, broadband irradiance can still read nominal on the sensor while the UV component has quietly collapsed. The test does not stay constant: it becomes gentler every week. The lifetime that gets reported is the lifetime of a device that never received the dose written in the protocol.

Over a 1,000-hour ISOS-L-2 run, that is not a rounding error. It is a systematically optimistic result, and it is not reproducible from one lab to the next.

Why the short wavelengths decide the outcome

UV is where thin-film photovoltaics actually fail. Photocatalytic activity at metal-oxide transport layers, UV-accelerated halide segregation and ion migration, photo-oxidation of organic transport layers and OPV blends, encapsulant yellowing and delamination: none of these are properly excited by a spectrum that stops at 380 nm, and all of them are under-excited by a lamp whose UV output has decayed.

For space photovoltaics the point is not arguable. Under AM0 there is no ozone layer to strip UVC, and a qualification campaign that omits it is not representative of orbit.

A light source with nothing to erode

Lumartix designs and builds its own plasma light engines in Switzerland. The bulb is electrodeless: a sealed envelope of selected molecules excited by a radio-frequency field, with no electrodes to wear out. There is no erosion mechanism, and therefore no progressive UV loss.

• Continuous spectrum from 280 nm to 2500 nm, with 100% spectral coverage over the range defined by IEC 60904-9 Ed. 3, including UVA, UVB and UVC
• Class AAA to IEC 60904-9 Ed. 3 (2020): temporal instability < 2%, non-uniformity < 2%, spectral match class A
• Bulb with a virtually unlimited lifespan; RF driver rated up to 40,000 h
• AM1.5G, AM0 or custom spectrum

Plasma lamp output against the AM1.5G reference spectrum. Image: Lumartix

No filter in the path, no drift in the spectrum

There is a second failure mode that gets less attention than lamp ageing, and it corrupts long runs just as effectively: the correction filter. Most simulators do not emit a solar spectrum. They emit a source spectrum and then shape it with dichroic or absorptive filters until it reaches AM1.5G or AM0. Those filters sit in the beam, absorb energy, run hot and degrade. As their transmission curve moves, the spectrum arriving at the sample moves with it, band by band, slowly enough that nobody notices until results stop reproducing between campaigns.

Lumartix bulbs are formulated to emit AM1.5G or AM0 directly. The spectrum comes out of the plasma chemistry, not out of a filter stack, which removes the one component in the optical path whose ageing can pull the spectrum away from the standard.

The longest documented run to date is at CEA Paris-Saclay, where a Lumartix lamp has passed 30,000 operating hours with no measurable spectral shift. The result was presented at ESEMA 2024, the CNRS International Workshop on Emerging Solar Energy Materials and Applications, held on Porquerolles island, France. Teams at CEA Paris-Saclay also use Lumartix systems in their perovskite work.

The practical consequence is simple: the UV dose you specify is the dose your sample receives, at hour 10 and at hour 5,000. Your ageing curve reflects the material, not the lamp.

Lumixo-CORE and Bulb M46: spectral stability and AM1.5G matching, measured out to 10,780 operating hours. Image: Lumartix

Light on, while everything else moves

Degradation in the field is coupled, so Lumartix puts the light inside the environmental chamber rather than beside it.

This covers ISOS-L, ISOS-T, ISOS-LT and ISOS-LC on one platform, plus IEC 61215-2 (MQT 02, 04, 06, 07, 19.1, 19.2), MIL-STD-810G and DIN 75220.

Roughly 30% below comparable systems

Chambers are built to European quality standards around European-branded components, CE and RoHS conform, and designed to be opened. Every spare part can be bought individually. There is no proprietary consumable and no lamp-replacement cycle built into the business model. At equal specification, Lumartix systems come in around 30% below comparable equipment. For a group building a stability platform out of a fixed grant, that gap is often a second chamber.

Lumartix equipment is already at work at CEA Paris-Saclay, EPFL, CSEM, KAUST, IIT Bombay, Université Paris-Saclay, UM6P, TÜV Rheinland, First Solar, Qcells and LG Electronics.

Meet us in Berlin

Lumartix will be at Perovskite Connect 2026, Berlin, 21–22 October. Bring your ageing protocol and we will go through the spectrum, the irradiance and the UV dose it actually requires, with you, on the spot.

Configure your system and request a quote: www.lumartix.com/machines
Download the product brochure or ask for a spectral irradiance report for your protocol: www.lumartix.com/contactus

Lumartix SA, Switzerland. Extreme conditions in your test lab.