TOPCon Voc drift after UV preconditioning is mainly linked to the front Al₂O₃/SiNₓ passivation stack. UV can activate defects at the silicon surface, change hydrogen bonding and movement, and alter electrical charge stored in Al₂O₃. The measured result may keep changing even after the UV lamp is turned off. In one commercial-module study, dark-state changes had fitted time constants of 45 and 345 hours. Part of the loss then recovered under light with time constants of only 36 seconds and 5 minutes.[4]
There is also a big difference from one TOPCon product to another. An IEA PVPS review reports approximately 0.5%-8% power loss across 14 TOPCon types after 60 kWh/m² UV exposure, with a median of about 3%. Another dataset covering 28 TOPCon module types ranged from about 0% to 10% at the same nominal dose.[1] In practice, the cell design, UV spectrum, temperature, electrical state, encapsulation and post-test stabilization matter far more than the label “TOPCon” by itself.
Front Passivation Takes Most of the UV Stress
In a conventional front-boron-emitter TOPCon cell, UV does not hit the silicon directly. It first passes through the module glass and encapsulant, then reaches a front structure similar to:
SiNₓ → Al₂O₃ → boron-doped emitter → n-type silicon
The rear normally contains:
silicon → thin SiOₓ tunnel oxide → phosphorus-doped poly-Si → rear dielectric → metal
TOPCon gets its name from the tunnel-oxide passivated contact, but that does not mean the rear contact is always the main UV-sensitive area. Tests on industrial TOPCon precursor structures found much stronger degradation at the front. The rear changed much less because its relatively thick poly-Si layer blocked much of the short-wavelength UV before it could reach the buried tunnel-oxide interface.[2]
In that study, radiation below roughly 370 nm was strongly absorbed by the rear structure. This number should not be treated as a fixed protection limit for every TOPCon cell. Poly-Si thickness and the rest of the layer stack change how much UV reaches the interface.
That is why the exact cell structure matters when comparing products. Current TNC cell formats and TOPCon cell designs, for example, vary in wafer size and product configuration. A UV result should be tied to the structure that was actually tested.
A 10 mV Voc Drop Is Already Significant
Voc is the highest voltage a cell produces when no external current is being drawn. The easiest way to understand the loss is to look at what happens inside the silicon: sunlight creates electrons and holes, and the cell needs to keep them apart. If they recombine too quickly, voltage drops.
For scale, take a cell with an initial Voc of 720 mV:
- 5 mV loss: Voc falls to 715 mV, a reduction of about 0.7%.
- 10 mV loss: Voc falls to 710 mV, a reduction of about 1.4%.
- 20 mV loss: Voc falls to 700 mV, a reduction of about 2.8%.
So a change of only a few millivolts is not trivial in a high-efficiency cell.
Researchers often track J0, or saturation current density, to see whether recombination is getting worse. In simple terms, when J0 rises, Voc usually falls.
Direct UV testing of non-encapsulated industrial and laboratory cells found some UV-sensitive TOPCon groups with more than 7% relative efficiency loss, with Voc providing the largest contribution.[5]
Those were bare-cell tests, not complete module tests. Glass and encapsulant change the UV spectrum reaching the cell, so the 7% figure should not be read as a normal field-loss value for finished commercial modules.
A 3% Pmax Loss Is Not a 3% Voc Loss
Many UVID reports lead with maximum-power loss. That is useful, but it does not tell you exactly what happened to Voc.
Pmax = Voc × Isc × FF
- Voc: open-circuit voltage;
- Isc: short-circuit current;
- FF: fill factor.
If a module loses 3% of Pmax, the cause can be lower Voc, lower current, lower fill factor or several effects at once.
Take a 600 W module as a simple example. A 3% Pmax loss brings it down to 582 W. That 18 W difference says nothing by itself about how many millivolts were lost at cell level.
For anyone comparing suppliers or test reports, a Pmax number alone is not enough. Voc, Isc and FF should also be compared before and after the test.
Hydrogen Moves After Short-Wavelength UV
Hydrogen is not an unwanted impurity in this case. It is deliberately used in silicon cells to neutralize electrically active defects. Hydrogenated SiNₓ is an important source, and firing moves some of that hydrogen toward the silicon/passivation interface.
The shorter the UV wavelength, the more energy each photon carries:
- 365 nm: about 3.40 eV per photon;
- 340 nm: about 3.65 eV;
- 315 nm: about 3.94 eV.
A 315 nm photon therefore carries about 16% more energy than a 365 nm photon.
That matters because Si-H bonds sit in an energy range that short-wavelength UV can disturb. Once a bond breaks, a defect that was previously passivated can become active again. The released hydrogen can also move elsewhere in the passivation stack.[2]
So “UV removes hydrogen” is not quite right. The better description is that UV changes where the hydrogen sits and how it is bonded.
ToF-SIMS measurements support this. They found increased hydrogen concentration near the AlOₓ/p⁺-Si interface after UV exposure, while the same samples also showed higher surface recombination.[2]
The hydrogen movement itself is measured. What researchers are still working out is which exact hydrogen states account for each part of the Voc loss.
Al₂O₃ Charge Can Hide Early Damage
Al₂O₃ protects the silicon surface in two different ways:
- Chemical passivation reduces active defects at the silicon interface.
- Electrical passivation uses negative charge in Al₂O₃ to keep electrons away from the defective surface.
UV can make one worse while temporarily making the other look better.
Hydrogen-related bond breaking can increase interface defects. At the same time, UV can place extra negative charge into traps inside Al₂O₃. That charge helps keep electrons away from the surface, partly hiding the damage that has already happened underneath.
A detailed mechanism study found evidence for this combination of chemical passivation loss and temporary field-effect improvement.[3]
For test interpretation, the practical result is straightforward: the first Voc measurement after UV exposure can look better than the underlying interface condition really is.
Dark Storage Can Keep Lowering Voc for Days
After UV exposure, some of the charge trapped in Al₂O₃ can leave those traps while the module sits in darkness. Once that happens, the surface gets less electrical protection, more carriers reach active defects, and recombination rises.
A 2026 commercial-module study found additional dark degradation after a 60 kWh/m² exposure at 340 nm. The tested modules already had about 2.3%-3.2% non-recovered UVID power loss before the extra dark-state change appeared.[4]
The measured dark-storage response was fitted with two time constants:
- 45 hours — about 1.9 days;
- 345 hours — about 14.4 days.
These values belong to the tested modules; they are not fixed constants for every TOPCon product.
For procurement or reliability review, the more useful point is timing. A module measured one day after UVID may not be in the same electrical state as the same module measured two weeks later. If a test report does not state the dark-storage period, comparisons can be misleading.
Light Can Recover Part of the Loss in Minutes
The same study found that light-driven recovery was far quicker than the dark-state change:
- 36 seconds for the faster fitted component;
- 5 minutes for the slower fitted component.
The dark response took tens to hundreds of hours. The recovery took seconds to minutes.[4]
After 520 hours, or about 21.7 days, of dark storage, individual cells showed EL intensity changes ranging from approximately +6% to -70%. After illumination, they returned to within about ±4% of their initial post-UVID EL state.[4]
That -70% figure needs context. EL is a very sensitive diagnostic technique. A 70% EL change does not mean the module lost 70% of its power.
Recovered Voc Does Not Mean a Fully Restored Interface
A module can recover electrically without returning to exactly the same microscopic material state it had before the test.
Controlled studies of TOPCon passivation structures found electrical recovery while FTIR measurements still detected changes in SiNₓ and in the thin oxide at the silicon interface.[3]
That leaves two separate questions for reliability work:
- electrical recovery: did Voc or Pmax come back?
- material recovery: did the passivation stack return to its original microscopic condition?
The first can happen quickly. The second may not be complete even when the power result looks much better.
UV-B Drives the Damage Faster
UV-A covers roughly 315-400 nm. UV-B covers about 280-315 nm.
Because UV-B is shorter in wavelength, each photon carries more energy. In TOPCon tests, UV-A and UV-B could reach a similar final level of front-passivation degradation, but UV-B got there significantly faster.[2]
That is why “60 kWh/m²” does not fully describe a UV test.
Two chambers can deliver the same total dose while using different wavelength distributions. The chamber with more short-wavelength UV exposes the passivation stack to higher-energy photons.
A useful report should therefore state both:
- the accumulated UV dose in kWh/m²;
- the spectral distribution or wavelength range used to deliver it.
Hot Tests Can Shift Raw Voc by About 8%
Temperature can change Voc even when no permanent degradation has occurred.
Current Tongwei TNC module specifications list a Voc temperature coefficient of approximately -0.24%/°C. The same product data also lists a Pmax coefficient of about -0.28%/°C. TNC electrical and temperature specifications provide the product-level values.
If module temperature rises from 25°C to 60°C:
35°C × -0.24%/°C ≈ -8.4%
A raw Voc reading at 60°C can therefore be roughly 8.4% lower than at 25°C simply because the module is hotter.
That 8.4% is not UVID.
Formal PV measurements correct I-V results for temperature and irradiance. IEC 60891 defines procedures for these corrections.[8]
For anyone comparing test reports, the practical check is simple: make sure the voltage numbers have been measured or corrected under comparable temperature conditions.
1.3 nm More Al₂O₃ Cut Loss by 2.39 Points in One Study
A 2025 study compared TOPCon cells using different front Al₂O₃ thicknesses.
Increasing Al₂O₃ from approximately 2.8 nm to 4.1 nm reduced module Pmax loss after the study's UV treatment from 3.24% to 0.85%.[6]
The change was:
- Al₂O₃ thickness: +1.3 nm;
- relative thickness increase: about 46%;
- Pmax-loss reduction: 2.39 percentage points;
- relative reduction in measured loss: about 74%.
That is a large performance difference from a very small physical change in layer thickness.
It does not mean 4.1 nm is the right answer for every TOPCon design. Al₂O₃ quality, SiNₓ properties, interface cleaning, firing temperature and hydrogen distribution all affect the final result.
Electrical Bias Shifted Pmax Loss by 2.3 Points
A module's electrical state during UV exposure changes the number of electrons and holes inside the cells. That can affect defect reactions and charge trapping.
In a Fraunhofer ISE study of commercial TOPCon modules, mean measured PMPP loss changed from:
- -4.2% under short circuit;
- to -1.9% under open circuit.
The difference was about 2.3 percentage points in the tested sample set.[7]
This should not be turned into a general rule that open circuit always improves UVID by 2.3 points. It shows that electrical state can materially affect the result.
That matters in real projects because operating modules spend much of a sunny day near maximum power point, not continuously at short circuit.
Encapsulation Changes the Cell-Level UV Dose
A UV chamber controls the radiation hitting the outside of the module. The cell only sees the part that makes it through the materials above it.
A useful simplified relationship is:
cell-level UV ≈ incident UV × glass transmission × encapsulant transmission × cell-coating transmission
Each of those factors changes with wavelength.
A UV-blocking encapsulant removes part of the high-energy UV before it reaches the Al₂O₃/SiNₓ stack. A more UV-transparent encapsulant lets more short-wavelength light reach the cell.
The IEA PVPS Task 13 reliability report also identifies front-side optical materials as part of the UVID problem because reflecting or absorbing UV before it reaches sensitive interfaces can reduce degradation.[1]
For module selection, this means a bare-cell test should not be applied directly to a finished module with a different optical stack. Different module constructions use different glass, backsheet or double-glass configurations, so the tested bill of materials needs to match the product being evaluated.
Process Control Changes the Starting Passivation State
Two TOPCon cells can start with similar efficiency and still behave very differently under UV.
The reason often sits in the production process:
- surface cleaning: affects the number of defects present before Al₂O₃ deposition;
- Al₂O₃ deposition: affects thickness, charge and interface quality;
- SiNₓ deposition: affects hydrogen content and optical properties;
- firing: controls hydrogen release and redistribution;
- boron-emitter processing: affects surface doping and recombination;
- uniformity: determines whether every cell in a module starts with similar passivation quality.
The 2026 module metastability study showed this at cell level. Cells with high UVID also showed injection-dependent carrier lifetimes and significant variation within individual cells.[4]
For manufacturers, this is why initial efficiency cannot be the only acceptance metric. Cell-process development has to be checked against finished-module reliability. Tongwei's PV R&D and testing facilities include a TNC cell pilot line, module R&D center and PV testing center, allowing process changes to be checked beyond initial cell efficiency alone.
Product Spread Reaches 0%-10% at 60 kWh/m²
The commercial data shows just how wide the spread can be.
The 2025 IEA PVPS Task 13 report summarizes one dataset containing 14 TOPCon types tested at 60 kWh/m²:
- lowest reported power loss: about 0.5%;
- highest: about 8%;
- median: about 3%.
The most degraded product in that dataset lost about 16 times as much power as the least degraded product.
The same IEA report cites another dataset containing 28 TOPCon module types, with power loss ranging from approximately 0% to 10% after the same nominal 60 kWh/m² dose.[1]
For B2B comparison, this is more useful than asking for a single “typical TOPCon UVID percentage.” The product, cell process, module BOM and test condition need to be known before the number means much.
15, 60 and 120 kWh/m² Are Not Equivalent Tests
IEC 61215-2 is a design-qualification standard. It checks whether a PV module can withstand defined stresses; it is not a direct simulation of 25 or 30 years in the field.[9]
Fraunhofer ISE states that module-certification UV preconditioning uses a dose of 15 kWh/m². Higher-dose testing is used when engineers want to make UV-sensitive failure mechanisms easier to see.[10]
- 15 kWh/m²: certification-scale UV preconditioning;
- 60 kWh/m²: 4 times that dose;
- 120 kWh/m²: 8 times that dose.
Those multipliers describe laboratory energy dose only. They do not mean 60 kWh/m² equals four times the module lifetime or 120 kWh/m² equals eight times the lifetime.
The IEA PVPS review estimates that 60 kWh/m² can correspond to roughly 1-2 years of outdoor UV exposure depending on location.[1] Climate, altitude, spectrum, module orientation and optical BOM all affect that relationship.
Use the Stabilized Value, Not the Worst Dark Value
For a commercial decision, the lowest number seen after a long dark-storage period can be misleading if the module has not been stabilized afterward.
Fraunhofer ISE found that after 60 kWh/m² UV exposure followed by sunlight stabilization:
- some tested module types showed hardly any significant remaining degradation;
- other module types still showed losses of up to about 5%.
Fraunhofer described 60 kWh/m² as roughly comparable to one year of UV exposure in Germany.[11]
A separate 2026 outdoor study followed TOPCon modules with different laboratory UVID sensitivity for five months. The average energy-yield difference between the more UV-sensitive M1 group and the more stable M3 group was about 0.17%. After light soaking, the difference in power degradation between the groups was approximately 0.09%.[12]
Five months is not enough to prove 30-year reliability. What it does show is that an unstabilized dark-state result can look worse than the short-term outdoor result.
Read a UVID Report in This Order
Before comparing two TOPCon UVID reports, check the test conditions in this order:
- UV dose: 15, 60, 120 kWh/m² or another value;
- UV spectrum: especially the amount of short-wavelength UV;
- module temperature: during exposure and electrical measurement;
- electrical state: short circuit, open circuit or another operating condition;
- cell structure: front passivation and rear poly-Si design;
- module BOM: glass, encapsulant and cell combination;
- dark-storage time: hours, days or weeks;
- stabilization: whether controlled light exposure was used before the final measurement;
- electrical metrics: Voc, Isc, FF and Pmax separately;
- sample count: the number of modules and production lots represented.
A broader module reliability comparison also needs to separate UVID from other failure mechanisms such as PID, corrosion, thermal cycling and mechanical damage. A lower post-UV Pmax alone does not tell you which failure mode caused the loss.
What the Current Evidence Cannot Prove
There is still no reliable universal conversion such as “60 kWh/m² equals exactly X years of TOPCon degradation.”
Three limits matter most:
- Long-term damage versus recovery: outdoor modules receive UV damage and recovery-producing light every day, followed by darkness every night. The decades-long balance is not established for every TOPCon design.
- Recovered output versus recovered material: Voc can recover even when microscopic changes remain in the passivation stack.[3]
- Product transferability: data from one Al₂O₃/SiNₓ recipe, encapsulant or rear poly-Si thickness cannot automatically be applied to a different TOPCon product.
The reported 0%-10% spread across commercial TOPCon types is already enough to show that product-level testing is more useful than assuming all TOPCon modules behave the same way.[1]
Finally
TOPCon Voc drift should be judged from the stabilized test state, not from one isolated post-UV number. UV can weaken front passivation through hydrogen and interface changes, while Al₂O₃ charge can temporarily hide part of that loss. In one module study, dark-state changes took about 45 and 345 hours, while recovery occurred in 36 seconds and 5 minutes. At 60 kWh/m², reported commercial-product losses span roughly 0%-10%. For a meaningful comparison, the report should state the UV spectrum, temperature, electrical condition, exact BOM, dark-storage duration and stabilized Voc/Pmax.[4][1]