The glass beneath the glass: why 'dual glass' stopped describing one thing.
"Dual glass" sounds like one thing: a panel sealed in glass on both sides instead of plastic. It isn't. The label now stretches from a format with a 26-year field record to thinned-out builds the research links to breakage and moisture damage. Four things decide whether the promise holds, and the sticker names none of them.
The original case was structural symmetry, not efficiency, and one manufacturer has held it since 1998.
The 2mm-plus-2mm glass-glass format was not invented for mass-market economics. SolarWatt, the Dresden manufacturer, introduced glass-glass modules in 1998 and spent two decades making a specific structural argument for it. Two glass panes of equal thickness, front and rear, place the cell stack on what materials engineers call the neutral axis: the geometric centre of the laminate, where bending stress approaches zero. Apply a load (wind, snow, handling) and the upper glass enters compression while the lower glass enters tension, but the cells in between see minimal stress. That is the opposite of a glass-backsheet module, whose asymmetric construction sits the cells nearer the tensile zone and raises the risk of microcracking under load. A PV-Tech lamination paper confirms the mechanics: two equal glass panes place near-zero net bending force on the cells, and replacing the polymer backsheet with a second pane of glass cuts moisture penetration by orders of magnitude, from days-to-weeks equilibrium in a glass-backsheet module to years in a 2mm glass-glass one. SolarWatt has carried a 30-year product and performance warranty on the format since 1998, a track record almost no other manufacturer can claim.
The terms changedThe format was adopted at scale for bifacial yield, then thinned under cost and weight pressure.
What SolarWatt engineered for durability, the industry adopted for a different reason: bifacial energy yield. Bifacial cells generate from both surfaces and need a transparent rear, glass is transparent and polymer backsheets are not, and the economics of utility-scale bifacial drove mass adoption of glass-glass from roughly 2018. But as modules grew past 3m², manufacturers under cost and weight pressure began thinning the glass: 2mm-plus-2mm gave way to asymmetric 2mm-plus-1.6mm, then symmetric 1.6mm-plus-1.6mm, and more recently 1.2mm rear glass in some lines. The structural consequences of thinning are not linear, which is the part the shared label hides.
The physics of thinningHalving the glass thickness does not halve the stiffness. It removes roughly half at 1.6mm and nearly four-fifths at 1.2mm.
The stiffness of a glass pane scales with the cube of its thickness. Moving from 2mm to 1.6mm is not a 20% reduction, it is close to 49% (1.6³ ÷ 2.0³ = 0.512). Deflection under load then scales with the fourth power of span length, so large formats with thinner glass deflect dramatically more than small formats with thick glass.
Independent testing confirms the direction. A 2025 Royal Society of Chemistry enhanced mechanical-load study found equal-thickness glass-glass modules protect encapsulated cells better than asymmetric or single-glass builds, and concluded outright that "the recent industry trend towards modules with thinner glass panes is not suitable for application in harsh climates." Fraunhofer's 2024 mechanical-stability analysis went further and more counterintuitively: it found surface compressive stress in 1.6mm panes was actually higher than in 2.0mm panes, because tempering thin glass uniformly is harder to execute, and tied double-glazed designs at 2.0mm and thinner to field failures observed within months of installation once weather, faulty installation and design error were excluded. A separate Fraunhofer finding reported by pv magazine noted the glass-breakage cases under investigation all involved 2mm dual-glass modules. DNV's principal solar engineer framed the mechanism plainly: once the glass is bearing significant load, the failure mode is brittle fracture, "you could easily pass a few modules in a load test but then have something like 1% fail in the field."
Tempered versus heat-strengthenedThe consumer description of dual glass as uniformly "stronger" breaks down at the treatment step.
Standard single-glass panels use 3.2mm fully tempered front glass, surface compression above 90 MPa, four to five times stronger than annealed glass, dicing into small harmless fragments when it breaks. Dual-glass panels at 2.0mm use heat-strengthened glass instead, a slower cooling process reaching only 24 to 69 MPa surface compression, stronger than annealed but not as impact-resistant as full tempering, and breaking into larger, sharper fragments. The reason is physical: full tempering becomes hard to execute below roughly 2mm, where the glass is too thin to sustain the rapid uniform cooling it requires, and thinner glass more commonly requires heat-strengthening. RETC laboratory testing has "consistently observed that heat-strengthened solar glass is more prone to breakage than fully tempered solar glass." The practical consequence, per Kiwa PVEL and VDE Americas: a 3.2mm glass-backsheet package is roughly twice as resistant to hail damage as a typical 2.0mm dual-glass package, and a 1.6mm package is weaker still. For high-hail sites, that is a real trade-off the format label conceals.
The encapsulant problemGlass thickness is one variable. The film between the glass and the cells is the one that decides the sealed module's chemistry.
The dominant encapsulant historically is EVA (ethylene vinyl acetate): cheap, decades-proven, easy to process, and a serviceable choice in a glass-backsheet module. In a glass-glass module it becomes a long-term liability, through a specific mechanism. EVA degrades under heat, UV and moisture by deacetylation, releasing acetic acid. In a glass-backsheet module that acid diffuses out through the permeable backsheet. In a near-hermetic glass-glass module it cannot escape, it accumulates, and it attacks the silver-paste metallisation and solder joints, raising series resistance and producing measurable power loss. NREL's glass-glass focus-group report documents exactly this in historical G/G EVA modules: higher browning, delamination and interconnect corrosion than equivalent glass-backsheet modules, with degradation rates reaching −2.88%/year in some early field populations. POE and TPO encapsulants carry no vinyl acetate groups, so they generate no acetic acid, and their lower moisture-transmission rates slow ingress through the only remaining pathways: the module edges and junction-box penetrations.
The advantage is real but not automatic. A PV-Tech lamination paper records a glass-glass EVA module losing 19.5% of power after 7,000 hours of damp heat, against −0.4% for the equivalent glass-glass TPO module: the acetic-acid trap in laboratory form. And a matched-BOM comparison (same front glass and cells, different rear) reported by SurgePV showed the glass-glass version degrading 1.12% in damp heat where the glass-backsheet version degraded 2.04%, confirming the systematic advantage when encapsulant and lamination quality are controlled, and implying its reversal when they are not.
Not all POE is equal"POE" on a datasheet is not, by itself, a guarantee. One marketed high-reliability formulation failed catastrophically in testing.
The 2025 Kiwa PVEL scorecard cut through the POE-is-safe assumption: one commercially marketed "high-reliability" POE encapsulant produced roughly 55% relative power loss after 1,000 hours of damp heat, worse than EVA's ~11% in the same configuration, traced to the antioxidant and UV-stabiliser chemistry in the formulation combined with residual soldering flux. There is also a widely-used middle path, EPE, a co-extruded EVA/POE/EVA film (Hangzhou First's EP304 is one example): its POE core provides the primary moisture and PID barrier while the outer EVA layers still generate some acetic acid on degradation. EPE passes the same extended damp-heat thresholds as monolayer POE in volume TOPCon production, but its moisture-transmission rate is two to four times higher. The point for a buyer is that the three-letter code on the datasheet (EVA, EPE, POE, TPO) is a more reliable signal than the format label, and that some datasheets list "POE/EVA," meaning the manufacturer reserves the right to use either depending on the production run, so the specific module cannot be known from the datasheet alone. EPFL research makes one fair qualification explicit: properly stabilised EVA in a glass-glass format may perform adequately for a 25-year life in temperate, moderate-humidity climates. It is in hot-humid and coastal environments, with sustained high relative humidity, that the EVA/glass-glass interaction becomes genuinely problematic and POE or TPO becomes the defensible specification.
Where the risk actually sitsThe failure hierarchy is specific, and the highest-risk configuration is also one of the most common on price.
Drawing the research together, from highest risk to lowest: large-format (over 3m²) glass-glass with 1.6mm or thinner heat-strengthened glass, EVA encapsulant and inadequate edge sealing is the cluster that has produced spontaneous field breakage, delamination and corrosion losses. Below it: any glass-glass format with EVA in a hot-humid or coastal climate. Then: glass-glass with an inferior POE formulation, unverified by third-party damp-heat data. Then: asymmetric 2mm-plus-1.6mm POE, which loses some of the neutral-axis symmetry and hail margin but is not a failed design. Highest-performing, and best-supported by the literature: symmetric 2mm-plus-2mm with POE or TPO, well laminated, adequately edge-sealed, the configuration whose 30-year warranties are consistent with the materials science behind them.
Three products through the same lensThe framework is not theoretical. It separates real products currently on the market.
Three cases span the range. None is a failed design, and all three outperform glass-backsheet equivalents on moisture resistance and thermal cycling. The differences are in specification certainty, depth of evidence, and length of structural commitment.
| Variable | SolarWatt M 5.0 | DualSun Flash GG TOPCon | Trina Vertex S+ NEG9R.25 |
|---|---|---|---|
| Glass | 2mm + 2mm symmetric | 2mm + 2mm symmetric | 1.6mm + 1.6mm symmetric |
| Treatment | Heat-strengthened | Heat-strengthened | Heat-strengthened |
| Encapsulant | POE (datasheet-named) | EPE / EVA-POE-EVA (BOM-verified; datasheet silent) | "POE/EVA" (batch-variable) |
| Product warranty | 30 years | 25 to 30 years (activation-conditional) | 25 years |
| Annual degradation cap | ≤0.31%/yr | Not separately stated | ≤0.4%/yr |
| Field track record | 26 years glass-glass | Active since ~2012 | No glass-glass history pre-2018 |
| Independent damp-heat data | DH2000-equivalent (Solitek) | DH1000 (TÜV Rheinland) | DH1000 (field comparison) |
Specifications from manufacturer datasheets, warranty documents and, for DualSun, a signed July 2025 bill of materials identifying the encapsulant as Hangzhou First EP304 (an EPE film), where the datasheet lists none. Edge-sealing specification is not publicly disclosed for any of the three. Sources: SolarWatt, DualSun and Trina product documentation; Kiwa PVEL 2025 Scorecard; Hangzhou First EP304 sheet.
SolarWatt's advantage is specification transparency and track record: the encapsulant is named on the datasheet, the warranty is unconditional rather than activation-gated, and 26 years of glass-glass manufacturing is the only long-run field-evidence base in the residential segment. The trade-offs are price and weight (24.8kg). DualSun is architecturally equivalent on glass thickness, having made 2mm-plus-2mm a stated brand position rather than a cost variable, with strong hail certification; its gaps are encapsulant disclosure on the primary datasheet (the FAQ states a POE preference, the BOM shows EPE) and the absence of public extended damp-heat data. Trina's Vertex S+ is the honest picture of where market volume sits: lighter, cheaper, a genuine glass-glass module, but with a batch-variable "POE/EVA" listing and structurally thinner 1.6mm glass, whose ~2.0m² area sits below the over-3m² threshold that most amplifies the thinning risk. In moderate, non-coastal conditions the practical reliability gap over 25 years may be small; in hot-humid coastal sites, the specification differences become consequential.
What buyers can actually askFour questions the datasheet answers and the marketing line does not.
"Is this a dual-glass panel?" is insufficiently specific. The questions the research supports are: What is the glass thickness on each surface, and is it symmetric? A 2mm-plus-2mm build is structurally different from 2mm-plus-1.6mm or 1.6mm-plus-1.6mm, and all three are sold as "dual glass." Is the front glass fully tempered or heat-strengthened? At 2mm it is almost always heat-strengthened, with lower impact resistance than 3.2mm tempered single glass, a material point for high-hail sites. Which encapsulant, on which layer, EVA, EPE, POE or TPO? EVA in a sealed glass-glass module generates acetic acid that cannot escape; for a 25-to-30-year life in hot or humid conditions, POE or TPO is the defensible specification, confirmed on the datasheet rather than claimed in marketing. Has the specific bill of materials passed extended (DH2000) damp-heat testing? PVEL's scorecard is the most credible third-party reliability data available, and a module tested only to standard IEC 61215 (DH1000) has not met the same bar. The marketing description rarely specifies all four. The datasheet, third-party test results and warranty terms, read together, usually do.
How this feeds the panel score: review.solar/methodology. Term definitions: review.solar/glossary.
Sources16 references
- SolarWatt: Panel vision M 5.0 datasheet (2mm/2mm, POE)
- SolarWatt: 30-year product and performance warranty conditions
- PV-Tech: Lamination and encapsulation materials for glass-glass module design
- Royal Society of Chemistry: Enhanced mechanical-load testing of glass-glass modules (2025)
- Fraunhofer: Mechanical stability of double-glass modules, PV-Symposium 2024
- pv magazine: Weekend read: temper tantrum (spontaneous glass breakage, 2024)
- Solar Builder: Solar module glass is spontaneously breaking in the field (2024)
- Kiwa PVEL: 2025 damp-heat scorecard (glass-glass vs glass-backsheet)
- PV Tech: Kiwa PVEL 2025 scorecard, glass-breakage findings
- NREL: Glass/glass focus group report (EVA degradation in sealed modules)
- Progress in Photovoltaics: Interfacial degradation in bifacial glass modules (EVA vs POE, PID)
- SurgePV: Glass-glass vs glass-backsheet, matched-BOM damp-heat comparison
- PV Europe / Solitek: POE-encapsulated modules particularly durable (2024)
- DualSun: Dual-glass FAQ (2mm/2mm position, encapsulant preference)
- TaiyangNews: Hangzhou First EP304/EP308 EPE (EVA-POE-EVA) films
- VDE Americas: Hail resiliency: 3.2mm glass-backsheet vs 2.0mm dual-glass
Read our ownership, funding and editorial disclosures.
Stay across the benchmark.
New scorecards, score changes and market signals, published as they happen.