Reliability Is a Scale-Up Decision, Not a Final Test

Reliability is not a certificate collected after a photovoltaic product is ready. It is a sequence of decisions that determines which architecture, bill of materials, process window and operating envelope are "allowed" to scale.

That distinction changes the role of testing. A test result should not merely confirm a product already frozen by schedule and capital commitments. It should remove—or expose—enough uncertainty to decide whether to lock a material, release a line, approve a substitution, sign a warranty or deploy into a specific environment.

Qualification is a gate, not a lifetime forecast

The public summary of IEC 61215-1:2021 is unusually clear about the boundary. The standard addresses design qualification and type approval for modules intended for long-term outdoor operation, while stating that useful service life depends on design, environment and operating conditions. It also states that test results are not a quantitative prediction of module lifetime.

Qualification therefore answers a bounded question: did the tested design survive defined sequences and criteria? It does not, by itself, demonstrate a thirty-year degradation trajectory, the behaviour of every production lot, or the validity of a later ‘equivalent’ material substitution. Treating it as a lifetime model converts a conformity result into a prediction it was not designed to make.

This is not an argument against qualification. It is an argument for using it at the correct decision gate—and for connecting the qualified sample to what the factory will actually build.

The evidence must follow the configuration into manufacturing

The public summary of IEC 62941:2019 makes the manufacturing link explicit. Design qualification depends on product and process design and on control of the materials and processes used to manufacture the module. Its objective is improved confidence in the ongoing consistency of certified modules.

That shifts the unit of reliability from a technology label to a controlled configuration. ‘TOPCon’, ‘HJT’, ‘bifacial’ or ‘glass-glass’ describes an architecture family. Reliability decisions attach to the cell process, glass, encapsulant, interconnect, junction box, adhesives, edge conditions, factory route and inspection system that create the saleable product.

Current UVID evidence provides a concrete example. Kiwa PVEL’s 2026 scorecard reports a 3.0% median UVID power loss for the TOPCon BOMs it tested and a 3% UVID failure share. In one published comparison, two BOMs using cells from the same supplier lost 14.1% and 3.6% after UVID120. Kiwa PVEL attributes the difference to the cell process recipe, particularly the passivation and anti-reflection stacks. These are test-provider results, not a statistically representative estimate for every TOPCon product, but they demonstrate why the process recipe and BOM cannot be collapsed into a generic technology category.

The test method itself also limits the conclusion. Kiwa PVEL describes UVID120 as a susceptibility screen using front-side UV exposure at controlled temperature and electrical condition. It is not designed to measure the full amount of UV-related degradation. Post-exposure dark storage and light soaking can change measured loss, so sequence, stabilisation and endpoint must be reported before the result is used in a warranty or energy-yield model.

Stability, qualification and lifetime prediction are different evidence products

• A stability test observes change under specified conditions and duration. Its value depends on complete reporting, reproducibility and a plausible connection between the applied stress and the mechanism of interest.

• Qualification applies standardised sequences and pass/fail criteria to a defined product. It supports comparison and release decisions, but it does not automatically transfer to a changed BOM, factory or environment.

• Lifetime prediction requires a validated relationship between stress, mechanism, acceleration and field exposure, together with uncertainty bounds and evidence that the model remains valid for the configuration being assessed.

The ISOS consensus statement for perovskite PV illustrates the first category well. It was developed to improve consistency, comparability and understanding of failure modes, including effects specific to perovskites such as reversible degradation and ion redistribution. The authors explicitly state that the protocols are not a replacement for existing qualification standards. Their value is methodological; their scope should not be expanded to every PV technology without justification.

Similarly, the IEA PVPS Task 13 report on degradation and failure modes shows that new module architectures introduce mechanism-specific questions—from UV-sensitive interfaces and encapsulation to thin glass, junction-box reliability and combined stresses. The report is a synthesis of evidence available in early 2025, not a universal acceptance test or a substitute for product-specific data.

A decision–evidence matrix for scale-up

Decision	Mechanism	Evidence	Main uncertainty	Advance criterion
Freeze a BOM	Known material and interface interactions under relevant stresses	BOM-specific accelerated tests, baseline characterisation and material traceability	Supplier or formulation changes after testing	Critical materials and revisions are locked; substitutions trigger defined requalification
Release pilot to pre-series	The process can hold the attributes that control degradation	Process windows, defect distributions, first-pass yield, metrology correlation and genealogy	Small samples may hide tails and drift	Stable distributions across representative lots, with excursions understood and contained
Approve volume ramp	Qualified design is reproduced by the actual line and suppliers	Quality plan, control charts, incoming-material controls, audit closure and change control	Ramp pressure can move the process outside the qualified envelope	No unresolved critical failure mode; deviations have owners, limits and response plans
Approve deployment	Test stresses represent the intended environment, installation and system interfaces	Site-specific stress review, installation controls, system compatibility and early field surveillance	Combined stresses and duty cycle may differ from laboratory sequences	Residual risks are quantified, monitored and allocated contractually
Extend warranty or bankability assumptions	Accelerated and field evidence support a bounded performance claim	Mechanism-based model, uncertainty range, field correlation and cohort traceability	Extrapolation beyond observed time and configuration	Claim is configuration-specific, auditable and linked to monitoring and corrective action

The criteria in this matrix are intentionally not universal numerical thresholds. A safe release limit depends on the failure mechanism, product architecture, application, consequence of failure and reversibility of the next commitment. The discipline is to define the criterion before the decision—not after an unfavourable result appears.

Reliability must close the loop before scale-up

Three operating rules follow.

• First, connect every test sample to a traceable configuration and manufacturing route. Without genealogy, a strong result may apply only to an unidentified prototype.

• Second, define change-control triggers around the variables that control the mechanism. A supplier change, thickness change, cure profile or cell-passivation revision may require targeted requalification even when the marketing name is unchanged.

• Third, combine accelerated testing with manufacturing distributions and field surveillance. Laboratory evidence identifies susceptibility; process data show whether the factory holds the relevant attributes; field data test whether the assumed mechanisms and acceleration remain credible.

This is where reliability becomes an industrialisation capability rather than a laboratory function. It shapes architecture selection, supplier qualification, metrology, process control, ramp release, procurement clauses, warranty structure and monitoring plans.

The practical question before every irreversible scale-up commitment is not simply ‘did it pass?’ It is: what configuration passed, which mechanism was challenged, what uncertainty remains, and what decision is this evidence strong enough to support?

Sources

• IEC 61215-1:2021 public product summary: https://webstore.iec.ch/en/publication/61345

• IEC 62941:2019 public product summary: https://webstore.iec.ch/en/publication/61932

• IEA PVPS Task 13, Degradation and Failure Modes in New Photovoltaic Cell and Module Technologies (2025): https://iea-pvps.org/key-topics/degradation-failure-modes-new-cell-module-technologies/

• Khenkin et al., Consensus statement for stability assessment and reporting for perovskite photovoltaics (2020): https://doi.org/10.1038/s41560-019-0529-5

• Kiwa PVEL, 2026 UVID results and test procedure: https://scorecard.pvel.com/uvid/