On a rooftop, nobody asks a solar panel to break a record for a few minutes. It is expected to keep generating electricity, summer after summer, despite rain, frost and scorching days. This is the test facing tandem cells that combine silicon and perovskites: their electrical potential is impressive, but their durability must inspire confidence. Looking ahead to September 2026, the real issue is therefore no longer simply capturing more sunlight. It is maintaining that advantage long enough to make the technology profitable and insurable.
This analysis draws on publicly documented advances through mid-2024. The outlook for September 2026 is presented as industrial scenarios, not as commercial results already achieved.
Two absorbers to make better use of light
Silicon dominates photovoltaics because it benefits from decades of improvements and enormous factories. But a single-junction cell does not harness every color in sunlight equally. Some photons are not absorbed; for others, part of their energy ends up as heat. Even when refined, this architecture faces physical limits.
A tandem divides the work. A perovskite layer, placed above the silicon, preferentially absorbs the highest-energy photons. The silicon captures some of the light that passes through. By adjusting the perovskite’s composition, researchers can tailor its bandgap, meaning the energy needed to free electrical charges.
The intended gain is tangible: more watts from the same area. On a roof with limited space, this can increase generation without adding square meters. At a solar farm, it can reduce certain requirements for mounting structures, wiring or land per installed watt. Provided the module’s additional cost does not swallow those savings.
Records are no guarantee
By 2023 and the first half of 2024, certified silicon-perovskite tandem cells had exceeded 33% efficiency in the laboratory. Companies such as LONGi, along with several European university teams, contributed to this progress. Oxford PV had also announced results for larger devices and was preparing to scale up manufacturing.
These milestones matter: they show that the combination works beyond a theoretical concept. But certified efficiency describes performance under defined conditions. It guarantees neither consistent factory production nor thirty years of operation. A record-breaking cell, a commercial module and a reliable solar farm are three different things.
The transition to a module adds losses and challenges: electrical connections, inactive areas, layer uniformity and mechanical protection. Comparing the record efficiency of a small tandem cell with that of a commercially available silicon panel can therefore give a misleading picture of the advantage on offer.
Moisture is only the first adversary
Photovoltaic perovskites are generally metal halides whose crystal structure provides excellent optoelectronic properties. Their fragility stems in part from their sensitivity to the environment. Water can promote their decomposition; oxygen, light and heat can also trigger or accelerate unwanted reactions.
Internal mechanisms add to these challenges. Ions can move under the influence of temperature or an electric field. In some compositions containing several halogens, illumination can cause phase segregation that disrupts performance. The interfaces between the perovskite and neighboring layers are also critical areas.
On a rooftop, these stresses accumulate. The panel heats up in the sun, cools at night and undergoes mechanical changes. A shaded area can impose unfavorable electrical conditions on certain cells. A material that is stable in a dry, controlled atmosphere is therefore not automatically stable in an exposed module.
Protection without excessive costs
The first response is to create a barrier. Glass, barrier films and edge seals must limit the ingress of water and oxygen. Seal quality is particularly important: excellent protection across the entire face of a panel can be compromised by a weakness at its edges.
The second response operates at the chemical level. Teams are modifying compositions, reducing crystal defects and stabilizing interfaces. Some treatments limit charge recombination while slowing degradation mechanisms. The challenge is to combine these benefits without making manufacturing too slow or too delicate.
Encapsulation cannot solve everything. It protects against the external environment, but does not necessarily eliminate internal instabilities under illumination and electrical bias. And every additional material brings trade-offs between cost, transparency, weight, recyclability and compatibility with industrial processes.
The real evidence comes from the field
To build confidence, manufacturers must accumulate complementary tests: damp heat, thermal cycling, ultraviolet exposure and electrical stress, followed by outdoor exposure in several climates. Photovoltaic qualification standards, notably IEC 61215, provide an essential framework. Passing them, however, does not in itself demonstrate a service life spanning several decades.
The decisive question is how representative the tests are. Accelerated stress must reproduce relevant mechanisms without masking those that emerge when several stresses combine. Protocols must also specify whether cells are operating at their maximum power point: a measurement after storage does not tell the same story as continuous operation.
For a buyer, the most useful data will be long-term datasets from modules representative of production, including the spread of results and failure analysis. A reassuring average is not enough if a proportion of panels ages much faster.
Manufacturing at scale, quickly and without defects
Depositing a uniform layer over a small area is one thing; doing so on large, textured silicon wafers is another. Thickness, crystallization and coverage must remain under control. A local defect can impair an entire device. Solution-based, vacuum and hybrid processes each involve their own trade-offs.
The challenge goes beyond deposition speed. Raw materials must be controlled, contamination prevented and defects detected early enough. Manufacturing yield—the proportion of products that can actually be sold—can matter as much as electrical efficiency. An excellent cell that is rejected too often remains a poor business proposition.
Tandem architectures also require choices. Two electrical terminals simplify some forms of integration, but require good current matching between the subcells. Four terminals offer greater electrical independence, at the cost of additional complexity. No solution wins on every front.
What next?
Looking ahead to September 2026, a credible scenario would be gradual adoption, starting where available space is expensive and an efficiency boost justifies a higher price. It will also be necessary to ensure secure containment of the lead present in the best-performing compositions and to plan for their end-of-life management. The decisive signal will not be another isolated record, but the convergence of three forms of evidence: modules that remain stable over the long term, repeatable manufacturing and warranties backed by robust data. Perovskites could then become more than a bright promise.
