Three weeks ago, a perovskite result landed that the field has been waiting for in one specific sense: it was measured at a size that starts to resemble a product. On October 7, researchers at UNSW Sydney, working with Chinese perovskite manufacturer UtmoLight, announced a certified stabilized power conversion efficiency of 23.5% for a perovskite submodule measuring 30 × 30 cm, with an aperture area of 676 cm². The team says the figure beats the previous benchmark for this size class by 0.6 percentage points. And it did so without the conventional nickel oxide layer that submodule makers have long treated as close to mandatory.

That combination — large area, high efficiency, and one fewer layer — is why this story belongs in the scaling arc this desk has been following all month. It is also why it needs to be read with the care its own paperwork invites. The university release calls the result a world record but does not name the certifying body, the previous record holder, or the measurement standard. A peer-reviewed paper published two months earlier reports a higher certified efficiency at a larger aperture area. And the publicly circulated test-report excerpt does not, on one trade analysis’s arithmetic, independently reconcile with the headline figure. None of that erases the result. All of it shapes how it should be described: a genuine large-area benchmark from a university–manufacturer partnership built to test scalability early, whose record claim is contested and whose documentation is thinner than the headline suggests.

Why 30 cm is the interesting number

Perovskite solar cells have spent a decade being brilliant at small sizes. Laboratory cells are typically around 1 cm², and at that scale the field has produced efficiency figures in the high twenties, including the 27.4% champion inverted cell from a Northwestern–DGIST team that this desk covered days ago. The problem has never been the film at fingertip scale. It has been what happens when you coat the same material across a sheet the size of a floor tile, interconnect dozens of cells in series, and ask the whole assembly to behave as one device.

Every step up in area is a tax. Coating uniformity degrades, and a single pinhole or shunt path can drag down a whole submodule. Interconnects eat active area — which is why the distinction between outer dimensions and aperture area matters. The UNSW–UtmoLight submodule is 30 × 30 cm on the outside, or 900 cm². The 23.5% figure is measured over the 676 cm² aperture area, the region delimited for illumination during testing. That is standard practice, not a trick — but it means the headline efficiency is not the efficiency of the full 30 cm square, and it cannot be directly compared with figures quoted on a total-area basis.

Against that background, 23.5% at this aperture is a serious number. The previous benchmark for the size class, on the team’s account, was 22.9%. Small-area cells still run ahead, as they always do, but the gap between a 1 cm² hero cell and a 676 cm² submodule has narrowed to a few points rather than a chasm. That is the honest sense in which this result matters: not as a product, not as a factory announcement, but as evidence that a university materials concept survives contact with industrial-scale processing.

The partnership structure is part of the point. UNSW’s School of Photovoltaic and Renewable Energy Engineering, led by Scientia Professor Xiaojing Hao with Dr Zhen Li and Dr Ziyue Feng, developed the materials innovation; UtmoLight, a China-based perovskite manufacturer, contributed large-area processing and module fabrication expertise. Hao framed the collaboration explicitly as a way to test scalability early in research and development rather than after lab records — to find out whether an idea can be made at size before years are spent perfecting it at 1 cm².

The missing layer, and why it mattered

The more technically interesting half of the announcement is not the efficiency figure. It is the layer that is not there.

Nickel oxide (NiOₓ) is the conventional hole-transport layer in many perovskite devices, and it is especially common at submodule scale, where its job is partly protective: it helps prevent electrical short circuits and ensures the device functions properly across a large area. The trouble, as UNSW itself notes, is that nickel oxide can react adversely with the perovskite material, contributing to instability — and depositing it adds another manufacturing step, with its own uniformity and yield costs, to every module that leaves the line.

The UNSW team says it eliminated the layer entirely through materials innovation and a different fabrication approach: instead of building the device layer by layer, with a discrete nickel oxide film deposited as a separate step, the hole-selective contact forms directly during fabrication. The announcement does not disclose the device stack, the materials substituted for NiOₓ, or the deposition sequence — no journal paper, DOI, or stack diagram accompanied the release, and this article will not invent one. What can be said is what the claim implies. A hole-selective contact extracts positive charge carriers from the perovskite while blocking electrons; if it can be made to self-form during fabrication, the device sheds a deposition step, a material, and a known reactivity risk at once. For a technology whose commercial case rests partly on low-cost, high-throughput manufacturing, removing a step is a cost and yield argument wearing a chemistry costume.

It is also, to be clear, an argument with an untested tail. A contact that forms during fabrication must form uniformly across 676 cm², reproducibly, batch after batch — and the announcement contains no yield data, no reproducibility statistics, and no stability data for the submodule. The team’s own acknowledgment, carried in the coverage, is that further work is needed on efficiency, reproducibility, and long-term stability before wider deployment. A hero submodule proves a concept can reach a number once. It does not prove a factory can reach it ten thousand times.

The record-keeping caveat, stated plainly

UNSW’s release describes the result as a certified, world-record efficiency. But three gaps in the record deserve to be on the table, because they determine what the claim can bear.

First, the certifier. UNSW’s release says “certified” without naming the certifying body, and pv magazine noted the same gap in its October 7 coverage. PV Tech, likewise, reported that the announcement named neither the previous record holder, the certifying laboratory, nor the measurement standard. UtmoLight’s own announcement, according to one trade analysis, names China’s National Center of Supervision and Inspection on Solar Photovoltaic Product Quality, and a test-report excerpt circulated by the company bears the name of the Fujian Metrology Institute, with the measurement carried out under standard test conditions using maximum power point tracking. That is a plausible certification trail — but it comes from the manufacturer’s announcement, not from the university release or a published certificate, and this desk has not independently verified the certificate itself. The responsible formulation is that the 23.5% is a reported certified stabilized efficiency, with the certifier named by the manufacturer rather than by UNSW.

Second, the competing claim. On August 12, 2026, a team led by Nanjing University with perovskite specialist Renshine Solar published a paper in Nature, “Lead carboxylates passivation for meter-scale perovskite solar modules,” reporting a certified 24.0% efficiency on an 810 cm² aperture module and 22.0% certified on a 0.72 m² module, with certification by TÜV SÜD in China and a pass through the IEC 61215 reliability sequence. That is a larger aperture area at a higher certified efficiency, published in a peer-reviewed journal, two months before the UNSW announcement. The two results are not perfectly commensurable — the Nature paper’s abstract does not spell out its measurement protocol in the same detail, and the UNSW figure is a stabilized maximum-power-point value, a deliberately hard-to-flatter protocol — but on the numbers as published, the “world record” framing does not survive unqualified. The UNSW result is best described as a benchmark for the 30 cm submodule class, and specifically as a benchmark achieved without nickel oxide.

Third, the arithmetic. The same trade analysis that traced the certifier notes that the test-report excerpt’s stated output — 15.68 W — corresponds to about 23.2% over a 676 cm² aperture under standard 1,000 W/m² illumination, not 23.5%, which would require roughly 15.89 W; and that the excerpt contains no information on area or measurement uncertainty. That does not prove the headline figure wrong — the excerpt may be partial, and a stabilized maximum-power-point average over the last 30 seconds of a 300-second test need not match a single output reading — but it does mean the publicly available paperwork does not independently confirm the headline. Until a full certificate is published, 23.5% should be quoted as the team’s reported certified figure, not as an independently confirmed one.

Where it sits in the scaling arc

Read that way, the UNSW result slots into October’s perovskite coverage as the middle entry in a trio of answers to the same question — can perovskite scale? — each answering a different part of it.

The incumbent-scale answer came from Trina, whose 907 W tandem module record, covered by this desk on October 28, showed what a mature manufacturer can already ship in module form. The crystal-growth answer came from Argonne’s eutectic-crystallization work, which attacked the scale-up wall from the materials side, engineering how crystals form so that large-area films start with fewer defects. The UNSW–UtmoLight result is the process-simplification answer: remove a troublesome layer, form the contact in situ, and prove the concept at a size where manufacturing realities start to bite. And the legal answer, covered October 30 in First Solar’s CubicPV license, is the reminder that scaling is also a patent problem — freedom to operate is a gate that opens in parallel with the technical ones.

There is also a useful contrast inside the UNSW story itself. The team’s next step is a full-size 2.8 m² module, at commercial module dimensions, targeted for the coming months at about 18–19% efficiency. That target is deliberately lower than the submodule figure — the team knows the area tax is real — and it is a step up from UtmoLight’s own prior full-size result, a 2.8 m² module rated 450 W at 16.1% full-area efficiency unveiled in November 2024. The honest way to read the roadmap is as a staircase: 23.5% at 676 cm² aperture now, 18–19% at 2.8 m² next, with reproducibility and stability data still to come.

What to watch

The 2.8 m² module is the first thing to watch, and the number to watch is not just its efficiency. It is whether the nickel-oxide-free contact survives the jump from 676 cm² aperture to a full commercial-size module, and what the team discloses about yield and uniformity when it gets there.

The second watch item is documentation. A published certificate naming the certifier, the measurement standard, and the area basis would let the 23.5% figure be quoted without qualifiers. A peer-reviewed paper describing the contact-formation mechanism would close the second gap: what replaces NiOₓ, and how the contact forms during fabrication, are the details other laboratories need to reproduce or refute the approach.

The third is stability. No damp-heat, thermal-cycling, or outdoor-exposure data accompanies the submodule, and the Nature paper’s IEC 61215 pass is the standard this result will implicitly be measured against. Removing a reactive layer should help stability in principle — one fewer interface at which the perovskite can degrade — but principle is not data.

And the fourth is the record book itself. If the field’s efficiency tables adopt the UNSW figure for the 30 cm class, the “world record” language will acquire the institutional backing it currently lacks. If they instead list the Nanjing–Renshine 24.0% at 810 cm² as the large-area benchmark, the UNSW result will stand as a process milestone rather than a record.

The bottom line

Three weeks on, the UNSW–UtmoLight submodule reads as a genuinely useful data point in perovskite’s hardest problem: a 676 cm² aperture device at a reported 23.5% certified stabilized efficiency, built without the nickel oxide layer that submodule makers have long relied on, by a partnership designed to test scalability early rather than late. The 2.8 m² module now in the pipeline, targeted at 18–19%, is the test that will tell whether this is a process or a hero device.

The caveats travel with the result and should not be detached from it. The certifier is named by the manufacturer, not by UNSW, and no full certificate is public. A peer-reviewed August paper reports a higher certified efficiency at a larger aperture area. No stability, yield, or reproducibility data accompanies the announcement. Frame it as a large-area benchmark in the 30 cm class, achieved by removing a layer rather than adding one — and let the full-size module, the certificate, and the stability data decide whether it becomes more than that.

No investment recommendation is made or implied anywhere in this article.

Sources and further reading

  • UNSW Sydney / TechXplore (October 7, 2026), “Researchers set large-scale perovskite photovoltaic efficiency world record” — the primary source for the 23.5% certified stabilized efficiency, the 676 cm² aperture area, the 0.6-point margin, the nickel oxide removal and directly formed hole-selective contact, the team (Xiaojing Hao, Zhen Li, Ziyue Feng), and the 2.8 m² module plan at 18–19%. The release does not name the certifying body, the previous record holder, or the measurement standard, and no journal paper or DOI accompanies it.
  • pv magazine (October 7, 2026), “UNSW researchers achieve 23.5% efficiency record for 676 cm2 perovskite PV submodule” — confirming the headline figures and noting that the certifying entity was not disclosed.
  • PV Tech (October 7–8, 2026), “UNSW, UtmoLight set 23.5% efficiency record for large-area perovskite solar submodule” — the team details, the unnamed previous holder, certifier, and standard, and the UtmoLight November 2024 baseline (2.8 m² module, 450 W, 16.1% full-area efficiency).
  • Xenospectrum (October 2026), “UNSW and UtmoLight Report 23.5% Efficiency for a 30cm Perovskite Solar Cell Without a Nickel Oxide Layer” — tracing the certifier to UtmoLight’s announcement (National Center of Supervision and Inspection on Solar Photovoltaic Product Quality; Fujian Metrology Institute on the report excerpt), the stabilized maximum-power-point protocol (average of the last 30 seconds of a 300-second test), the 15.68 W / ~23.2% arithmetic question, and the competing August 12, 2026 Nature paper.
  • pv magazine (August 17, 2026), “Chinese scientists achieve record-breaking 24.0% efficiency for meter-scale perovskite solar modules” — the Nanjing University–Renshine Solar result: 24.0% certified at 810 cm² aperture, 22.0% certified at 0.72 m² total area, TÜV SÜD certification, lead carboxylate passivation, IEC 61215 reliability pass, and the Nature paper “Lead carboxylates passivation for meter-scale perovskite solar modules.”
  • Excitonic.com, “Two Defect-Fixers That Don’t Fight Each Other Take Perovskite Cells to 27.4%” — the small-area contrast (0.049 cm² champion cell); and “First Solar Buys Its Way Past the Perovskite Patent Thicket” — the freedom-to-operate gate on the same scaling road.

A note on sourcing: no device stack, deposition sequence, yield, reproducibility, or stability data for the UNSW–UtmoLight submodule appears in the coverage consulted, and none is stated here. The 23.5% figure is reported as the team’s certified stabilized value, with the certification trail and the competing claim described as above, because the publicly available paperwork does not independently confirm it.