What are the main components of solar cell?

Solar cells feature monocrystalline Si wafers (180μm thick), a p-n junction from phosphorus/boron doping, a 75nm SiNₓ anti-reflective coating to maximize light absorption, and silver electrodes for current extraction.

Internal Components of Solar Cells

Global photovoltaic installations exceeded 400GW in 2023, with the efficiency of mainstream PERC cells reaching over 23.5%. This achievement relies on the precise coordination of internal components at the millimeter or even micrometer level. Take silicon wafers, for example. The mainstream now uses 182mm×182mm monocrystalline silicon wafers, with thicknesses reduced to 130-150 micrometers (slightly thicker than a human hair). Thinner wafers save material but are more fragile, while thicker ones have poorer light absorption.

Another example is the PN junction. Its depth directly affects carrier separation efficiency. The PN junction depth of conventional Aluminum Back Surface Field (Al-BSF) cells is about 0.3-0.5 micrometers, while the combination of the tunnel oxide layer and polysilicon layer in TOPCon cells acts like a "high-speed channel" for the PN junction, allowing electrons to move faster, directly boosting efficiency by 1-2 percentage points.

Semiconductor Substrate

In 2023, global new PV installations exceeded 400GW, over 90% of which used monocrystalline silicon cells. The thickness of a mainstream 182mm×182mm silicon wafer has been reduced from 200μm in 2015 to 130-150μm—a 35% reduction in thickness, but requiring a 20% increase in cost for thinning process technology.

This is because the silicon wafer is the "raw material foundation" of the cell: its length and width determine how many wafers can be cut from an ingot (182mm ingot yields 54 wafers, 210mm yields 96 wafers), thickness affects light absorption and breakage rate (a 10μm decrease increases breakage rate by 0.3%), and minority carrier lifetime directly determines efficiency (a wafer with 500μs lifetime results in 23% cell efficiency; 800μs can reach 23.8%). Even the size of the surface pyramid texture is strictly controlled: 1-5μm pyramids can increase the optical path length by 4 times, reducing reflectivity from 30% to below 10%.

Wafer "Dimensions": Size, Thickness, and Cutting Cost

The mainstream wafer sizes in the market are 182mm (M10) and 210mm (G12), accounting for over 90%. Size selection isn't arbitrary: a 182mm wafer has a power of about 7.7W (at 23.5% efficiency), while a 210mm wafer can reach 10.1W (at 23.2% efficiency). The latter has an 8% lower cost per watt, but requires cutting machines with wider blades (blade width increases from 1.5mm to 2.5mm), and cutting loss increases from 0.5% to 0.8%. Thickness is even more nuanced: a 130μm wafer yields 58 slices per kilogram; a 150μm wafer only yields 48 slices. But if thickness goes below 120μm, the breakage rate soars from 0.5% to 2% (an extra 1500 broken wafers per 100,000 wafers, costing over 200,000 RMB). Therefore, manufacturers currently target 130-150μm: using diamond wire cutting (wire diameter 40μm, compared to 120μm for traditional slurry cutting), reducing cutting loss to 0.3%, while controlling oxygen content during crystal pulling (≤1.5 ppma) to make wafers more crack-resistant.

Wafer "Quality": Purity, Minority Carrier Lifetime, and Resistivity

Crystal pulling uses electronic-grade polysilicon produced by the Siemens method (9N purity, i.e., 99.9999999%), but melting silicon absorbs oxygen—if the oxygen content exceeds 3 ppm (3×10¹⁵ oxygen atoms per cubic meter), "oxygen precipitates" form in the wafer, acting as recombination centers (an extra 10¹³ traps per cm²), causing minority carrier lifetime to drop from 800μs to 500μs. Thus, top-tier wafers control oxygen content at 1-1.5 ppm, stabilizing minority carrier lifetime at 700-1,000μs (laboratory-grade can reach 1500μs).

Resistivity is a "balancing act": too low (<1 Ω·cm), the wafer conducts too well, carriers leak before reaching the electrodes (leakage current density >1 fA/cm²); too high (3 Ω·cm), carrier migration slows, reducing collection efficiency. The mainstream choice is 1-2 Ω·cm for N-type silicon (more electrons) or 1.5-3 Ω·cm for P-type silicon (more holes). For example, P-type silicon for PERC cells: at 1.8 Ω·cm resistivity, the fill factor can reach 79%; below 1.5 Ω·cm, the fill factor drops to 77%, directly reducing efficiency by 0.3%.

Surface Treatment: The Dual Challenge of "Texturing" and "Passivation"

Freshly cut wafers have a surface damage layer (2-5μm deep) full of dangling bonds (10¹² recombination centers per cm²). First, acid etching/polishing removes the damage layer (HF+HNO₃ mixture, etch rate 0.5μm/min), then alkaline solution texturing—NaOH concentration 2%, temperature 80°C, etch for 3 minutes, growing 1-3μm pyramid structures. This structure reduces direct light reflectivity from 30% to below 10%, effectively "frosting" the wafer surface, making light travel several passes before being absorbed.

But non-uniform texturing is problematic: if pyramid size varies by over 2μm, local reflectivity can rebound to 15%, corresponding to a 0.2% cell efficiency loss. Thus, "black silicon" technology is now used, creating nanoscale pyramids (50-200nm) via plasma etching, reducing reflectivity to <3%, but requiring an additional equipment investment of 5 million RMB/GW.

The final step is rear surface passivation: conventional Al-BSF cells use aluminum paste sintering to form a p + layer on the rear (doping concentration 1×10¹⁹ cm⁻³), reducing the rear surface recombination velocity from 1×10⁴ cm/s to 1×10³ cm/s; TOPCon cells are more aggressive, depositing 10nm silicon oxide + 150nm polysilicon on the rear, directly dropping the recombination velocity to 1×10² cm/s, increasing efficiency by 1%.

Grade Differences: Impact of A/B/C Grade Wafers on Efficiency

Wafers are graded A/B/C upon shipment. Grade A: minority carrier lifetime ≥800μs, resistivity 1.5-2.5 Ω·cm, oxygen content ≤1.5 ppma, surface free of scratches—resulting cell efficiency ≥23.5%. Grade B: minority carrier lifetime 500-800μs, resistivity 1-3 Ω·cm, oxygen content 1.5-3 ppma, may have minor scratches—efficiency around 23%, priced 0.05 RMB cheaper per wafer. Grade C: minority carrier lifetime <500μs, oxygen content >3 ppma, obvious defects—efficiency below 22%, mostly used for low-end modules. In 2023, Grade A accounted for 65%, Grade B 25%, Grade C 10%—choosing Grade A costs an extra 0.05 RMB/wafer, but cell efficiency is 0.5% higher, earning an extra 0.1 RMB/W at the module level. Manufacturers have calculated this clearly.

PN Junction

In a 23.5% efficient PERC cell, 70% of the photoelectric conversion contribution comes from the PN junction—this interface formed by P-type and N-type semiconductors "holding hands" silently sorts 10²¹ photogenerated carriers daily within 1500GW of global modules. It's like a parcel sorting center: electrons go to the N-region, holes to the P-region; any mistake (like recombination) means losing a unit of electricity.

The PN junction in high-end cells has boron doping concentration precisely set at 5×10¹⁶ cm⁻³ (500 trillion boron atoms per cm³), phosphorus doping at 1×10¹⁶ cm⁻³, junction depth 0.3-0.5μm (1/200th of a hair's width), built-in electric field strength 10³ V/cm—a 0.1 difference in these numbers reduces efficiency by 0.2%.

How is the PN Junction Formed? A Two-Step "Doping + Diffusion"

The essence of a PN junction is the boundary between the P and N regions. First, make the N-type silicon wafer: place the silicon ingot in a diffusion furnace, introduce phosphorus source (POCl₃), at 900°C phosphorus atoms "drill" into the silicon lattice, replacing original silicon atoms—after 1 hour of diffusion, surface phosphorus concentration surges to 1×10²¹ cm⁻³, forming the N+ layer. Then, create the P-region: immerse the entire wafer in borosilicate glass (BSG) solution, anneal at 850°C, boron atoms diffuse from the surface inward, stopping at the "balance of power" between the N and P regions.

Now, ion implantation is mainstream instead of diffusion: boron ions bombard the wafer surface with 50keV energy, directly "pinning" them 0.3μm deep, concentration 1×10¹⁷ cm⁻³, then anneal to activate. The benefit is more uniform doping, junction depth error shrinks from ±0.1μm to ±0.02μm, minority carrier lifetime loss reduces from 20% to 5%. But the equipment is expensive—an ion implanter costs 20 million RMB, 5 times that of a diffusion furnace.

Doping Concentration: More or Less?

Too much boron (1×10¹⁸ cm⁻³), too many carriers in the P-region, electrons get "traffic jammed" (increased recombination) when moving to the N-region; too much phosphorus (5×10¹⁹ cm⁻³), excess electrons in the N-region, holes have difficulty passing. For PERC cells, boron concentration in the P-region is controlled at 5×10¹⁶ cm⁻³, phosphorus in the N-region at 1×10¹⁷ cm⁻³—this combination is the lab-tested "golden ratio": the built-in field is just right for sorting carriers without causing excessive resistance (contact resistance 70 mΩ·cm²).

TOPCon cells are even more precise: adding a tunnel oxide layer (1-2nm SiO₂) + polysilicon layer (150nm) on the rear of the N-type wafer acts like an "insulating gate" for the PN junction. The polysilicon layer is doped with phosphorus to 1×10²⁰ cm⁻³ (10 times that of a conventional N-region), but the oxide layer only allows electrons to tunnel through (electrons smaller than 1nm can pass, holes cannot), thus increasing electron collection efficiency from 95% to 98%, directly boosting efficiency by 1%.

Junction Depth: Shallow or Deep?

Al-BSF cells have a depth of 0.3-0.5μm, PERC uses laser SE (Selective Emitter) to reduce depth to 0.2-0.4μm—why shallower? Because a shallow junction is closer to the surface, photogenerated carriers don't have to "travel far" to be sorted, reducing recombination loss. But too shallow (0.2μm) is bad: boron and phosphorus diffusion can "cross-contaminate", phosphorus enters the P-region, boron enters the N-region, forming a "dead layer" (carrier lifetime <100μs), actually reducing efficiency by 0.3%.

This is now solved with double diffusion: first, a shallow junction (0.2μm) ensures sorting efficiency, then supplemental doping in non-gridline areas (junction depth 0.5μm) reduces contact resistance. This technique increases cell fill factor from 78.5% to 79.5%, boosting efficiency by 0.4%.

Selective Emitter: Creating a "Fast Lane" for the PN Junction

In the PN junction area directly under the fine gridlines, heavy doping is intentionally applied (boron concentration 1×10¹⁸ cm⁻³, 10 times the conventional area), effectively building a "highway" for electrons—when gridlines collect current, contact resistance drops from 100 mΩ·cm² to 70 mΩ·cm², each gridline can collect an extra 2mA current. But the heavy doping area cannot be too large: if it exceeds 10% of the wafer area, it causes light absorption loss (the heavy doping layer absorbs more light), actually reducing efficiency by 0.1%. Therefore, laser grooving is used now, applying heavy doping only under the gridlines in a 10μm wide area, controlling the area ratio to 5%, resulting in a net efficiency gain of 0.5%.

Defect Control: "Leaks" in the PN Junction

A conventional PN junction has 10¹⁰ recombination centers per cm², causing a 0.5% efficiency loss; after hydrogen passivation (introducing H₂ during diffusion), recombination centers drop to 10⁹ per cm², efficiency increases by 0.3%. TOPCon's polysilicon layer is cleaner: using LPCVD (Low-Pressure Chemical Vapor Deposition) for coating, particle count <10 per cm² (compared to 100 for conventional PECVD), recombination centers are an order of magnitude lower, directly adding 0.2% efficiency.

For a cell's PN junction, from doping concentration to junction depth, from selective emitter to defect control, every parameter is crucial for efficiency. The 0.3μm junction depth, 1×10¹⁸ cm⁻³ heavy doping, 10⁹ per cm² recombination centers—these "small numbers" hidden within the PN junction are the key code for the PV industry's annual 0.5% efficiency increase.

Anti-Reflection Coating

In 2023, global PV module efficiency was 3% higher than ten years ago, with the anti-reflection coating contributing 0.8% of that share—this film thinner than a hair (75nm thick, roughly equivalent to 20 layers of graphene stacked) reduces silicon wafer surface reflectivity from 30% to <5%. In other words, for every 100 photons hitting the silicon wafer, previously 30 bounced off directly, now fewer than 5 do, the remaining 95 can enter the silicon to generate electricity.

This coating also doubles as a "cleaner": hydrogen atoms in the coating can neutralize dangling bonds on the silicon surface (10¹² recombination traps per cm²), increasing minority carrier lifetime from 200μs to 500μs, effectively "paving a smoother road" for carriers.

ARC "Material Choice": Why is SiNx the Top Choice?

Currently, 90% of cells use SiNx (silicon nitride) for the ARC, not because it's cheap (deposition cost 0.05 RMB per square meter), but because its optical and passivation properties are so balanced. Compared to other materials: TiO₂ (titanium dioxide) refractive index 2.3, good anti-reflection but weak passivation (hydrogen content only 1/3 of SiNx); Al₂O₃ (aluminum oxide) refractive index 1.6, poor anti-reflection (reflectivity can't be reduced below 8%), but can supply negative charges to the silicon—so now high-end cells use a "SiNx+Al₂O₃" stack, with SiNx for anti-reflection upfront and Al₂O₃ for passivation in the rear, boosting efficiency another 0.2%.

SiNx is prepared using PECVD (Plasma-Enhanced Chemical Vapor Deposition), chamber temperature 300-400°C (200°C lower than LPCVD, saving energy), silane (SiH₄) and ammonia (NH₃) introduced in a 1:5 ratio, depositing a uniform film on the wafer surface in 15 minutes, thickness deviation ±5nm (industry standard ±10nm)—this precision ensures consistent reflectivity for each cell, reducing module efficiency dispersion from 2% to 1%.

Thickness: The "Goldilocks Zone" at 75nm

The ARC thickness isn't arbitrary; it must be at the quarter-wavelength point. Taking mainstream 1100nm infrared light as an example, the wavelength in silicon is 1100nm / 3.5 (silicon's refractive index) ≈ 314nm, quarter-wavelength is 78.5nm—so SiNx film thickness is set at 75-85nm.

But infrared light alone isn't enough; the solar spectrum spans 300nm to 1200nm. Tests show: 75nm thick SiNx has 5.2% reflectivity at 400nm violet light, 4.1% at 1100nm IR, comprehensive reflectivity 4.5%; if thickness reduces to 65nm, violet reflectivity rises to 7%, IR drops to 3.8%, comprehensive reflectivity 5.1%—efficiency directly decreases by 0.15%. Therefore, ellipsometers are used for online thickness monitoring, error controlled within ±3nm, ensuring full-spectrum reflectivity remains below 5%.

Refractive Index: Matching Silicon's "Optical Code"

SiNx's refractive index is a "parameter tuning game". Too low (<2.0), the film is too "thin", poor anti-reflection; too high (>2.2), the film is too "dense", causes secondary reflection. Actual measurement: at refractive index 2.0, reflectivity for 1100nm light is 4.1%; at 2.1, it's 3.8%; at 2.2, it's 4.5%—the optimal range is 2.0-2.1. How to adjust? By changing the silane-to-ammonia ratio: more ammonia, higher nitrogen content, lower refractive index (ammonia proportion increasing from 50% to 70%, refractive index drops from 2.2 to 2.0). High-end equipment can precisely control gas flow (error <1 sccm), stabilizing refractive index at 2.05±0.03.

Bonus Skill: Passivation is More Important than Anti-Reflection

After cutting, the wafer surface is full of dangling bonds (10¹² per cm²), like countless tiny hooks grabbing carriers. Hydrogen atoms in the SiNx film (containing 5×10²² per m²) penetrate the silicon surface, neutralizing these dangling bonds—after treatment, the dangling bond count drops to 10¹⁰ per cm², minority carrier lifetime increases from 200μs to 500μs.

TOPCon cells are more aggressive: first deposit 10nm SiO₂ under the SiNx, forming a "SiO₂/SiNx" stack. SiO₂ itself has strong passivation ability (higher hydrogen content), reducing recombination velocity from 1×10⁴ cm/s to 1×10³ cm/s; SiNx adds another reduction, down to 1×10² cm/s—this stack increases cell open-circuit voltage from 680mV to 695mV, boosting efficiency by 1%.

Cost Calculation: The Secret to Saving 0.01 RMB per Watt

Don't be fooled by the thinness; the ARC cost isn't negligible. A single PECVD machine can coat 5000 wafers per hour, depreciation cost 0.01 RMB/wafer; silane + ammonia consumption 0.003 RMB/wafer; electricity cost 0.002 RMB/wafer—total 0.015 RMB/wafer. But the benefits are: 25% reduction in reflectivity (from 30% to 5%), improved silicon utilization, silicon cost per watt decreases by 0.02 RMB; increased minority carrier lifetime, cell efficiency rises 0.3%, module sells for 0.03 RMB more per watt. Calculated, the ARC's return on investment is 2:1, payback within a year—this "anti-glare film" truly delivers big results with small investment.

The science of the ARC, from material selection to thickness calibration, refractive index adjustment to passivation function, every parameter contends with light. The 75nm thickness, 2.05 refractive index, 5% reflectivity—these "small numbers" hidden in the film are the key to PV modules generating 0.5% more electricity annually.

Current Collection and Transport Components

The electricity generated by a module must be "transported out" by the current collection and transport components—the front electrodes collect photogenerated carriers, tabbing wires connect the cells, and the rear electrodes consolidate the current. Harsh data: traditional 2-busbar modules lose 5% power due to electrode shading; a 1°C deviation in tabbing wire soldering temperature increases contact resistance by 0.5mΩ, increasing annual module degradation by 0.2%.

More tangible is cost: silver paste accounts for 10%-15% of module cost (about 0.18 RMB/W for a 182mm module), but increasing busbars from 2 to 9 can earn an extra 0.1 RMB/W in generation revenue.

Front Electrode

Traditional 2-busbar modules have 3% gridline shading (equivalent to losing 3% of sunlight per square meter daily), but the current collection path is long, resistive loss 0.8mΩ·cm²; now mainstream 9-busbar, single gridline width reduced from 0.4mm to 0.15mm, shading reduced to 1.2% (recovering 2% generation), resistive loss <0.3mΩ·cm² (current flows 3 times faster).

Silver paste accounts for 10%-15% of module cost (about 0.18 RMB/W for a 182mm module), but 9-busbar yields 8% more annual generation than 2-busbar (1kW earns 40kWh more, 25 years earns 1000 RMB more).

Silver Paste Printing: From 2 Busbars to 9 Busbars, the Ultimate Challenge of Fine Line Width

90% of front electrodes use silver paste printing (silver content 85%-90%, about 800g silver per kg paste).

· Bus bar Number Evolution: Early 2 busbars, line width 0.4mm, shading 3% (blocking 3% sunlight), but current had to travel farther from cell edge to busbar, resistive loss 0.8mΩ·cm². Now 9 busbars, single line width 0.15mm (human hair diameter 50μm, gridline thinner than hair), total shading 1.2% (recovering 2% generation), resistive loss <0.3mΩ·cm² (current transmission 3x faster). Measured on a 182mm module: 9-busbar yields 8% more annual generation than 2-busbar (1kW module earns 0.08kWh more daily, 25 years earns 750 RMB more).

· Fine Line Process Limit: Fine line width has been pushed to 25μm (0.025mm), using laser engraving instead of traditional screen printing. Screen printing line width deviation ±5μm (prone to broken lines), laser engraving deviation <2μm (almost no breaks). One factory test: broken line rate dropped from 0.1% (printing) to 0.05% (laser), Grade A yield increased from 97% to 99%, saving 0.02 RMB/W cost (saving 2 million RMB per GW).

· Silver Paste Formula Upgrade: Low-silver paste (75%-80% Ag) is 15% cheaper than traditional paste, but post-sintering conductivity is 10% lower. One factory used low-silver paste + 9-busbar design, silver paste cost reduced 12%, generation only decreased 0.3% (net saving 0.09 RMB/W), more cost-effective than using expensive paste alone.

Laser Engraving vs. Traditional Printing: Power Difference Behind 2μm Line Width Deviation

Front electrode formation uses two processes: screen printing (old tech) and laser engraving (new tech), the difference lies in precision.

· Screen Printing: Uses nylon screen to scrape paste, line width limited by mesh count (300 mesh ≈ 0.3mm line width), edges are rough (roughness Ra=5μm). Burrs can pierce the ARC (MgF₂ layer on glass), increasing local reflectivity by 0.5%, reducing module power by 0.2%.

· Laser Engraving: Uses UV laser (355nm wavelength) to ablate paste, line width precise to 25μm, smooth edges (Ra=1μm). One factory comparison: laser-engraved module microcrack rate 0.05% (printed 0.2%), because laser heat-affected zone is small (only 10μm, printed 50μm), less thermal damage to cells.

· Cost Calculation: Laser equipment is expensive (5 million RMB per unit), but screen printing requires frequent screen changes (every 100,000 wafers, 2 hours downtime each time). For 1GW capacity, laser line saves 80 hours annual downtime, produces 800,000 more wafers, payback period only 2 years.

Silver Paste Usage: Saving 60mg Paste, How Much More Profit?

9-busbar design not only reduces shading but also saves paste—silver paste consumption for a 182mm module drops from 180mg/cell for 2-busbar to 120mg/cell for 9-busbar (saving 1/3).

· Silver Price Cost: Silver price ~6000 RMB/kg, 120mg/cell saves 0.036 RMB per cell compared to 180mg/cell (saves 36 million RMB per GW).

· Generation Gain: 9-busbar has 1.8% lower shading, each module collects 0.018kWh more daily (based on 4kWh daily for 1kW module), over 25 years earns 0.018×365×25=164kWh more per module, at 0.5 RMB/kWh, earns 82 RMB more per module. 1GW modules (2 million cells) earns 164 million RMB more, far exceeding saved paste cost.

· Industry Status: Leading manufacturers mass-produce 12-busbar (single line width 0.1mm, shading 1%, paste consumption 100mg/cell), generation gain increases another 0.5%, but process difficulty is high (laser precision required <15μm), yield only 95% (9-busbar yield 98%).

The "Love-Hate Relationship" Between Gridlines and ARC: Poor Contact Ruins Everything

The contact between the front electrode gridlines and the glass ARC (MgF₂ layer) determines if current can flow "smoothly" into the gridlines.

· Contact Resistance: Ideal contact resistance <5mΩ·cm². One factory increased ARC thickness from 100nm to 120nm, gridline-to-coating contact area decreased 15%, contact resistance increased to 8mΩ·cm², module power decreased 0.4% (1kW loses 20kWh annually, 25 years loses 500 RMB).

· Coating Defects: If ARC has pinholes (diameter >1μm), silver paste can seep in, corroding Na⁺ (sodium ions) in the glass, causing delamination. One test: ARC with pinhole rate >0.1% had 5% delamination after 10 years (qualified <1%), repair cost 0.1 RMB per watt.

Rear Electrode

If the aluminum paste sintering temperature is 10°C lower, the Al-Si alloy layer (BSF) thickness decreases by 3μm (normal 5-8μm), carrier recombination current increases by 1mA/cm², efficiency drops directly by 0.3%; increasing aluminum layer thickness from 20μm to 30μm reduces sheet resistance from 15mΩ/□ to 8mΩ/□, module power increases 0.5% (1kW earns 25kWh more, 25 years earns 625 RMB more). More tangible is cost: aluminum paste accounts for 3%-5% of module cost (about 0.05 RMB/W for 182mm module), but this 0.05 RMB buys 25 years of stable current transmission.

Aluminum Paste Sintering: 10°C Lower Temperature, Efficiency Drops 0.3%

Aluminum paste is the "material" for the rear electrode, 90% aluminum powder (particle size 1-5μm), 10% glass frit (corrodes ARC during sintering). Sintering is key—using belt furnace, temperature 700-800°C, time 0.5-2 seconds, melts glass frit to corrode glass layer, aluminum reacts with silicon forming Al-Si alloy (BSF).

· Cost of Temperature Deviation: Setpoint 750°C, actual 10°C lower (740°C), glass frit doesn't melt completely, corroded ARC thickness decreases 20% (normal 0.5μm, becomes 0.4μm). BSF layer thickness drops from 6μm to 3μm, recombination current increases from 0.8mA/cm² to 1.5mA/cm², efficiency drops 0.5% (1kW module generates 0.005kWh less daily, 25 years loses 438 RMB). One factory had ±5°C furnace fluctuation, batch module efficiency variation 0.8%, customer return rate 15%.

· Sintering Belt Speed: Increasing speed from 2.5m/min to 3m/min (shorter sintering time), Al-Si alloy layer thickness decreases from 6μm to 4μm, recombination current increases 0.7mA/cm², efficiency drops 0.4%. But slow speed (2m/min) causes furnace clogging, capacity drops 20%. Compromise speed 2.8m/min, efficiency loss 0.2%, capacity maintained 95%, optimal overall.

· Glass Frit Content: Reducing frit from 10% to 8%, weaker corrosion, thinner BSF (4μm), efficiency drops 0.3%; but too much frit (12%), can pierce ARC, causing poor contact between Al BSF and wafer, contact resistance increases 2mΩ/□. One factory test: 10% frit gives contact resistance 8mΩ/□ (ideal <10mΩ/□), optimal efficiency.

Aluminum Layer Thickness: 20μm vs 30μm, Difference in Power and Lifespan

Post-sintering aluminum thickness directly affects conductivity and mechanical strength. 20μm is the lower limit, 30μm is the upper limit.

· 20μm Aluminum Layer: Sheet resistance 15mΩ/□, slow current transmission. Module at 85°C/85%RH for 1000 hours, aluminum oxidation rate 10% (alumina has high resistance), contact resistance increases to 20mΩ/□, power drops 1% (1kW loses 50kWh, 25 years loses 1250 RMB).

· 30μm Aluminum Layer: Sheet resistance 8mΩ/□, fast current transmission, oxidation rate only 3% (10-year contact resistance 9mΩ/□), high power retention. One module with 30μm Al layer had 2% higher power after 10 years than 20μm, earning 0.08 RMB/W more (1GW earns 8 million RMB more).

· Thickness Deviation: Thickness deviation >2μm (e.g., local 28μm, local 32μm), thin areas have higher sheet resistance (12mΩ/□), current collection causes heating, hot spot temperature exceeds 150°C (normal <120°C), accelerates EVA yellowing. One factory controlled deviation <1μm, hot spot failure rate dropped from 0.5% to 0.1%.

PERC Rear Silver Paste: 5x More Expensive, Why Use It?

Traditional BSF cells use pure aluminum BSF, but PERC cells add 2-4 silver gridlines on the rear—expensive (silver paste 5x costlier than aluminum paste), but earns more efficiency.

· Reduce Recombination Current: PERC cells have a passivation layer (Al₂O₃/SiNₓ) on the rear, silver gridlines penetrate it to collect edge carriers. Measured: adding 2 silver gridlines reduces recombination current from 1.2mA/cm² to 0.6mA/cm², efficiency increases 0.8% (1kW earns 40kWh more, 25 years earns 1000 RMB more).

· Silver Paste Usage: Each silver gridline width 0.2mm, thickness 0.1mm, usage 15mg/cell (5x costlier than Al paste). But generation gain covers cost—each cell earns 0.05 RMB more (1GW earns 10 million RMB more), more cost-effective than pure Al BSF.

· Soldering Reliability: Silver gridlines are softer than aluminum layer, prone to deformation during soldering, pull strength drops from 1.5N/mm to 1N/mm (industry standard ≥1N/mm). One factory used low-temperature solder ribbon (soldering temp 200°C), maintained pull strength 1.2N/mm, desoldering rate <0.1%.

The "Embrace" Between Al BSF and Wafer: Poor Contact Ruins Everything

The contact quality between the Al-Si alloy layer (BSF) and the wafer determines if current can "get out".

· Contact Area: BSF-to-wafer contact area should be >95% (measured by SEM). One factory printed paste too thick (25μm), sintering caused BSF bumps, contact area dropped to 90%, recombination current increased 0.5mA/cm², efficiency dropped 0.2%.

· Interface Defects: Microcracks in BSF (width >0.1μm) absorb moisture, causing aluminum corrosion (thickness decreases 2μm after 10 years), contact resistance increases 3mΩ/□. One test: modules with interface defect rate >0.5% had 3% power drop after 10 years, repair cost 0.15 RMB per watt.

Tabbing Wire (Ribbon)

A 1°C deviation in soldering temperature increases the cold solder joint rate from 0.1% to 5% (cold joint resistance is 10mΩ higher, 30% probability of hot spot burnout within 3 years); increasing tabbing wire thickness from 0.1mm to 0.3mm reduces resistance by 30%, but cell microcrack rate increases from 0.2% to 1% (5% power loss per cell). More tangible is cost: tabbing wire accounts for 3%-5% of module cost (about 0.05 RMB/W for 182mm module), but this 0.05 RMB determines if the module generates steadily for 25 years or fails early.

Ribbon Material: Copper Core Tin-Plated, How Much Does 0.01mm Thickness Difference Matter?

Ribbon is 95% copper core tin-plated (copper 90%, tin 10%). Copper conducts well (resistivity 1.7μΩ·cm), tin melts easily (232°C) for soldering. Material details directly affect performance:

· Copper Purity: Oxygen-free copper (99.97% purity) has 5% lower resistance than electrolytic copper (99.95%), but costs 10% more. One leading factory used OFC ribbon, module resistive loss decreased from 1.2% to 1.1%, annual generation increased 0.3% (1kW earns 15kWh more, 25 years earns 375 RMB more).

· Tin Plating Thickness: Tin layer thickness 0.02mm (total 0.12mm) vs 0.01mm (total 0.11mm) provides better tin flow during soldering, cold solder rate drops from 0.3% to 0.1%. But too thick (0.03mm, total 0.13mm), ribbon flexibility decreases, cell microcrack rate increases from 0.2% to 0.8% (3% power loss per cell).

· Surface Treatment: Ribbon surface with "micro-roughening" (Ra=0.5μm) vs smooth surface (Ra=0.1μm) has 20% better tin adhesion during soldering, contact resistance lower by 1mΩ. One factory test: micro-roughened ribbon module contact resistance increased only 3% after 10 years (smooth increased 8%), higher power retention by 5%.

Soldering Temperature: 220°C vs 260°C, 5x Difference in Cold Solder Rate

· Temperature Window: Optimal soldering temperature 245±5°C (tin melts at 232°C). Temperature 240°C (5°C low), tin not fully melted, cold solder rate increases from 0.1% to 5% (30% probability of hot spot within 3 years at cold joint); temperature 255°C (10°C high), tin oxidizes forming SnO₂ (high resistivity), contact resistance increases 2mΩ, module power drops 0.5% (1kW loses 25kWh, 25 years loses 625 RMB).

· Furnace Temperature Uniformity: Furnace temperature field deviation >±3°C (e.g., local 240°C, local 250°C), cold solder rate fluctuates 3%-8%. One factory upgraded furnace temperature control (deviation <±1°C), cold solder rate stabilized at 0.05%, customer complaints dropped 70%.

· Soldering Speed: Increasing belt speed from 1.5m/min to 2m/min (shorter soldering time), tin doesn't melt fully, cold solder rate increases from 0.2% to 1%. But too slow (1m/min) reduces capacity 30%. Compromise 1.8m/min, cold solder rate 0.1%, capacity maintained 90%.

Flux Residue: 0.5mg/cm² vs 0.2mg/cm², 20% Resistance Difference After 10 Years

Flux (rosin-based) is the "lubricant" during soldering, but excess residue absorbs moisture and causes corrosion.

· Residue Standard: Industry requires ≤0.5mg/cm². One factory used ordinary flux, residue 0.6mg/cm², after 10 years ribbon corrosion rate 15% (thickness reduced 0.03mm), resistance increased 20% (contact resistance from 5mΩ to 6mΩ), module power dropped 1% (1kW loses 50kWh, 25 years loses 1250 RMB).

· Low-Residue Flux: Using modified rosin (solid content reduced from 50% to 30%), residue 0.2mg/cm², after 10 years corrosion rate only 3% (thickness reduced 0.006mm), resistance increased 5% (contact resistance 5.25mΩ). One module factory switched, 10-year power retention 1.5% higher, repair cost saved 0.08 RMB per watt.

· Cleaning Process: Plasma cleaning after soldering (power 300W, time 10 seconds) removes 90% residue. One factory didn't clean, after 10 years the delamination rate was 8% (cleaned <2%), direct loss 0.1 RMB per watt.

Ribbon Width: 1mm vs 2mm, Balancing Current Transmission and Microcracks

Ribbon width affects resistance and stress on cells. 1mm is the economic line, 2mm is the safety line.

· 1mm Wide Ribbon: Resistance 30% lower than 2mm (faster current transmission), but less flexible, stress concentration on bent cells, microcrack rate increases from 0.2% to 1% (5% power loss per cell). A distributed PV plant using 1mm ribbon had annual microcrack repair cost 0.05 RMB per watt.

· 2mm Wide Ribbon: Resistance 30% higher, but more flexible, microcrack rate 0.2% (same as conventional). A utility-scale plant used 2mm ribbon, although resistance 0.3% higher (annual loss 0.15 RMB/W), microcrack repair cost saved 0.05 RMB/W, optimal overall.

· Ribbon Spacing: Reducing spacing from 1.2mm to 1.0mm (saves material), but heat concentration during soldering, local cell temperature 10°C higher (normal 250°C, becomes 260°C), microcrack rate increases 0.5%. One factory adjusted spacing to 1.1mm, temperature uniformity improved, microcrack rate returned to 0.2%.

Ribbon-to-Cell "Solder Joint Strength": Insufficient Pull Strength Ruins Everything

Solder joint pull strength (≥1N/mm) is the ribbon's "grip"; if it doesn't hold, desoldering occurs.

· Pull Test: Industry standard ≥1N/mm. One factory had tin plating thickness 0.01mm (total 0.11mm), pull strength only 0.8N/mm (fails), module desoldering rate during transport vibration 2% (each desoldered ribbon causes 0.5% power loss).

· Aged Pull Strength: Ribbon aged 5000 hours (UV+damp heat), pull strength retention ≥80% (≥0.8N/mm). A module using ribbon with aged strength 0.7N/mm had a 15% desoldering rate after 10 years, repair cost 0.1 RMB per watt.

Encapsulation and Protection Structure

Solar cells themselves are precise semiconductor devices; without encapsulation, exposure to the outdoors for 3 months can lead to failure due to moisture ingress, UV degradation, or mechanical impact—industry data shows unencapsulated monocrystalline silicon cells degrade over 30% in power within 200 hours in Hainan's hot-humid environment, becoming basically useless after half a year.

Systematic encapsulation extends module lifespan to over 25 years, which is the basis for the 25-year power generation warranty of PV power plants. Currently, encapsulation accounts for 10%-15% of total module cost (e.g., about 0.3-0.4 RMB/W for a 182mm monocrystalline silicon module), but directly affects power retention rate<strong (first-year degradation ≤2%, subsequent annual degradation ≤0.45%) and levelized cost of electricity (LCOE) (reducing LCOE by 0.05-0.08 RMB/kWh).

Glass Cover

Industry test data: modules using ordinary float glass lose 5% transmittance in the first year (ultra-clear patterned glass only loses 1.2%), the generation gap widens to 8% after 5 years; in hail tests, 3mm ordinary glass has a 37% probability of being pierced by 25mm hail (ultra-clear patterned glass only 2%).

Seemingly spending an extra 0.1 RMB/W for ultra-clear glass earns an extra 0.15 RMB/W in generation revenue over 25 years due to higher transmittance and slower degradation. Glass isn't just "transparent plastic"; it's a fine calculation based on every 1% transmittance, every hail impact.

Why Choose Ultra-Clear Patterned Glass? Transmittance is Squeezed Out

95% of module front-side glass uses ultra-clear patterned glass (not ordinary float glass). What's the difference? Ordinary glass contains 0.2%-0.3% iron (appears greenish), transmittance 88%-89%; ultra-clear glass contains <0.015% iron (almost colorless), base transmittance is 91%-93%. That's step one.

The glass surface is etched with micron-scale prismatic textures (height 5-10μm, spacing 20-30μm), like countless tiny mirrors reflecting light. When sunlight hits at a 60 angle of incidence (common solar altitude angle), 3%-5% of the light that would normally be reflected by the glass surface is scattered by the texture at different angles, with 70% of the reflected light re-entering the glass, ultimately increasing transmittance to 93%-95% (2%-3% higher than smooth ultra-clear glass).

Example: A 182mm module with smooth ultra-clear glass generates 1200kWh annually; after switching to patterned glass, transmittance increases 2%, annual generation rises to 1224kWh, earning 6000kWh more over 25 years (at 0.5 RMB/kWh, earning 3000 RMB more).

3.2mm or 2.0mm? The Trade-off Between Durability and Cost in Thickness

Mainstream glass thicknesses are 3.2mm (single-glass modules) and 2.0mm (double-glass modules). Choice depends on application:

· 3.2mm Glass: Stronger impact resistance. IEC 61215 hail test requires withstanding 25mm diameter hail at 23m/s (83km/h) without breaking—3.2mm glass can withstand, 2.0mm glass has a 15% probability of cracking (test data). Weight: 3.2mm is 12kg/m², 50% heavier than 2.0mm (8kg/m²), transport cost increases 0.02 RMB/W (extra 2 million RMB for 1GW modules).

· 2.0mm Glass: Cost-saving + lightweight. Double-glass modules using 2.0mm glass reduce total weight from 28kg/m² to 20kg/m², racking cost decreases 10% (saves 0.03 RMB/W). But impact resistance is weaker, use cautiously in hail-prone areas—one plant using 2.0mm glass had a 3% breakage rate in the first year due to hail, repair cost 0.1 RMB per watt.

UV Aging 1000 Hours, How Much Transmittance is Lost? Weather ability is Key for 25 Years

Glass fears not wind and rain, but ultraviolet (UV) light (wavelength 280-400nm). The SiO₂ and Al₂O₃ in ultra-clear glass absorb UV, but still 3%-5% UV penetrates to the encapsulant layer.

Test data: Ordinary ultra-clear glass after 1000 hours UV aging, transmittance drops from 93% to 89% (4% drop); glass with a UV-cut coating (thickness 0.1μm, contains TiO₂ nanoparticles) only drops to 91% (2% drop). Don't underestimate this 2%—it corresponds to 1% less annual generation, 0.08 RMB/W less earnings over 25 years (1GW plant loses 80 million RMB).

More critical is damp heat aging: at 85°C/85%RH for 1000 hours, the adhesive strength between glass and encapsulant decreases. Glass pretreated with silane coupling agent (surface grafted with amino groups) maintains 95% adhesive strength (untreated only 70%), delamination risk reduced 80% (delamination causes module power to drop 10% within 3 months).

Anti-Reflection Coating is Not Mystical, Nanometer Layer Earns More

The anti-reflection coating on the glass surface (about 100nm thick) is typically MgF₂ or SiO₂ deposited by magnetron sputtering. Simple principle: two reflected light beams (air-coating, coating-glass) are out of phase, canceling each other, reducing reflectivity from 3% to 0.5%.

Measured: Single-layer MgF₂ coating transmittance 93.5%, double-layer (MgF₂+SiO₂) can reach 94%. Corresponding module annual generation: single-layer increases 1.5%, double-layer increases 2%—1kW module earns 7-10kWh more annually (175-250kWh more over 25 years, at 0.5 RMB/kWh, earns 87-125 RMB more).

Coating uniformity is crucial: thickness deviation over 5nm (e.g., local 95nm) causes reflectivity to rebound to 1.5%, entire module power deviation 0.3%-0.5% (1GW plant loses 3-5MW generation).

Adhesion to Encapsulant: Poor Bonding Ruins Everything

The bond strength between glass and encapsulant (EVA/POE) relies on the glass's soda-lime silicate layer (thickness 0.5-1μm). Production requires acid etching cleaning (hydrofluoric + sulfuric acid mixture, 60°C, 3 minutes) to remove organics and dust, exposing the clean silicate layer.

One factory reduced cleaning time by 1 minute, adhesion strength dropped from 50N/cm to 35N/cm (industry standard ≥40N/cm). After 1000 hours at 85°C/85%RH, delamination area reached 12% (qualified <2%), moisture ingress at delaminated areas caused cell degradation over 15% within 5 years.

Encapsulant Film

Inside a module, the encapsulant film is less than 0.5mm thick, yet accounts for 10%-15% of module cost (e.g., EVA about 0.15 RMB/W, POE about 0.2 RMB/W for a 182mm single-glass module). It acts like glue bonding glass, cells, and backsheet into a whole, but its value lies in water vapor transmission rate (WVTR) (determines cell corrosion), anti-PID capability (affects power degradation), and lamination compatibility (determines delamination).

Harsh test data: Double-glass modules using EVA lose 8% power after 5 years (POE only 3%); 1°C deviation in lamination temperature increases bubble residue rate by 0.3%, these bubbles cause delamination within 3 years, power dropping 5%.

EVA vs. POE, The Difference is More Than the Name

Two mainstream encapsulants: EVA (Ethylene-Vinyl Acetate Copolymer) and POE (Polyolefin Elastomer). The choice directly determines the module's 25-year degradation curve.

· Water Vapor Transmission Rate (WVTR): EVA is a "sieve", 0.8-1.2 g/m²·day (lab data); POE is a "sealed jar", 0.08-0.1 g/m²·day (an order of magnitude difference). In southern plants with humidity >85%RH, modules with EVA have internal moisture content reaching 5g/m² after 5 years (POE only 0.5g/m²), silver electrodes corrode, efficiency drops 1.5% (POE only 0.3%).

· Anti-PID Degradation: PID is Potential Induced Degradation; under high voltage (1000V), acetate ions in EVA migrate, corroding the cell surface. EVA double-glass modules show 2%-3% PID degradation (1000h test), POE can suppress it to <0.5%. A northwest plant using EVA double-glass lost 3% first year due to PID, 10% cumulative over 5 years; after switching to POE, only 3% loss over 5 years, earning 0.1 RMB/W more.

· Cost Calculation: EVA is cheaper, 12-15 RMB/m²; POE is 30%-40% more expensive (18-20 RMB/m²). But POE modules have 1%-2% higher power retention over 25 years; for 1GW, that's 20 million RMB more earnings (at 0.5 RMB/W premium). Single-glass modules can use EVA cost-effectively (lower PID risk), double-glass must use POE (higher edge moisture ingress risk).

Lamination Isn't Just Heat Pressing; Slight Parameter Deviation Fails Everything

The encapsulant must "fuse" with glass and backsheet, relying entirely on lamination (temperature 145-150°C, vacuum <10 mbar, time 18-22 minutes). A 0.5°C or 1 mbar deviation causes problems:

· Temperature 0.5°C Low: Encapsulant crosslinking degree drops from 85% to 80% (standard ≥80%), adhesion decreases 15%. After 1000h damp heat (85°C/85%RH), delamination area increases from 1% to 5%, moisture ingress at delamination causes cell degradation over 10% within 3 years.

· Vacuum Exceeds 1 mbar: Bubble residue rate increases from 0.2% to 0.7%. One factory had vacuum pump failure, vacuum 12 mbar, defective module rate jumped from 0.3% to 1.2%; these bubbles cause local delamination within 5 years, power loss 5%-8%.

· Time Short 2 Minutes: Encapsulant flow insufficient, edge areas not fully filled. Tests show 18 min vs 20 min lamination increases edge delamination probability 3x, module mechanical load test (5400Pa) pass rate drops from 98% to 92%.

Moisture is the Encapsulant's Nemesis, Ingress Ruins Cells

The encapsulant's task is moisture barrier, as moisture is a "slow poison" for cells:

· Moisture ingress reacts with silver gridlines forming Ag₂O, increasing contact resistance 20%, module power drops 1%.

· Simultaneously, moisture dissolves acetic acid from EVA (releasing 0.5g/m²), acetic acid accumulates on the cell surface, corroding the ARC (MgF₂), transmittance drops 0.5%, annual generation decreases 2%.

Encapsulant Thickness Not Arbitrary; Thin is Brittle, Thick is Expensive

Common thicknesses: 0.4mm, 0.5mm, 0.6mm. Choice depends on application:

· 0.4mm Thin Film: Lower cost (10% cheaper than 0.5mm), but weaker impact resistance. Hail test (25mm hail, 23m/s): thin film module microcrack rate 1.5% (0.5mm: 0.8%, 0.6mm: 0.3%). Using 0.4mm in hail-prone northwest incurs annual repair cost 0.05 RMB/W.

· 0.6mm Thick Film: Better impact resistance, but poorer flowability. Lamination requires higher temperature (150°C) or longer time (22 min), otherwise edge filling is incomplete. One factory used 0.6mm film without extending lamination time, delamination rate increased from 0.5% to 2%, module failed 5 years early.

Encapsulant Not Sticking to Glass/Backsheet Leads to Eventual Delamination

Adhesion strength is the encapsulant's "grip". Industry standard ≥40N/cm (force per cm width).

· Production requires surface treatment for glass/backsheet: glass plasma cleaning (500W, 30s) removes organics; backsheet corona treatment (800W, 1m/min) increases surface energy. Inadequate treatment reduces adhesion to 30N/cm (fails), after 1000h damp heat, delamination area >10%, module scrapped.

· Encapsulant aging also affects adhesion: EVA aged 5000h, adhesion drops from 50N/cm to 35N/cm (POE maintains 45N/cm). A plant using EVA modules had 20% delamination after 10 years, repair cost 0.2 RMB/W.

Backsheet

It accounts for 8%-12% of module cost (e.g., about 0.1 RMB/W for a 182mm single-glass module using TPT backsheet), but must withstand 25 years of UV exposure, southern humidity, coastal salt spray. Harsh data: Modules with non-fluorinated backsheets have 3x higher moisture ingress after 5 years than fluorinated ones, power degrades 2% faster; in salt spray tests, inferior backsheets develop pits after 1000 hours, causing short circuits.

Why Choose Fluorinated Backsheet? UV Resistance is a Hard Requirement

Backsheets are divided into fluorinated (TPT/TPE) and non-fluorinated (PET). 90% of plants choose fluorinated for UV resistance.

· Material Structure: TPT is a "PVF outer layer + PET core layer + PVF inner layer" sandwich. PVF contains ≥50% fluorine (fluorine atom 19x heavier than hydrogen, stronger chemical bonds). UV light (280-400nm) hits, PVF absorbs over 90% of the energy, preventing penetration to the PET layer.

· UV Aging Test: Industry standard is 5000 hours UV aging (equivalent to 25 years outdoors). Fluorinated backsheet yellowing index <5 (yellowing reduces reflectivity), power degradation <3%; non-fluorinated backsheet yellowing index >10, power degradation 8%. A northwest plant using non-fluorinated backsheet had 5% lower efficiency after 5 years than fluorinated, losing 0.15 RMB/W annually.

· Weatherability Comparison: Fluorinated backsheet WVTR <0.1 g/m²·day (almost impermeable); non-fluorinated 0.5-1 g/m²·day (like leaking half a cup of water daily per m²). Southern plant humidity 85%RH, non-fluorinated module internal moisture reaches 15g/m² after 10 years (fluorinated only 2g/m²), silver electrodes corrode through, direct failure.

Moisture Barrier Rate More Important Than Thickness

Common backsheet thicknesses: 0.25mm, 0.3mm, 0.35mm. But barrier ability depends on material, not thickness.

· Fluorinated Backsheet: 0.3mm thick TPT, WVTR 0.08 g/m²·day (lab limit). A module using 0.3mm TPT has total internal moisture only 2g/m² over 25 years, silver gridlines begin slight oxidation after 25 years.

· Non-Fluorinated Backsheet: 0.35mm thick PET, WVTR 0.8 g/m²·day (10x fluorinated). Over 25 years, internal moisture 16g/m², silver electrodes corrode forming cracks within 5 years, contact resistance increases 30%, power drops 2%.

· Actual Cost: A southern plant using non-fluorinated backsheet had 15% delamination after 10 years, repair cost 0.1 RMB/W; after switching to fluorinated, delamination <1%, saved 2 million RMB over 25 years.

Poor Adhesion to Encapsulant Leads to Delamination in Damp Heat

The bond strength between backsheet and encapsulant (EVA/POE) determines if "delamination" occurs.

· Peel Strength Standard: Industry requires ≥40N/cm. Tests show plasma cleaning (500W, 30s) of back sheet surface achieves 50N/cm adhesion; without cleaning, only 30N/cm (fails).

· Performance in Damp Heat: One factory cut corners, no back sheet cleaning, adhesion 30N/cm. Module at 85°C/85%RH for 1000 hours, delamination area 12% (qualified <2%). Moisture ingress at delamination caused cell degradation over 15% within 3 years, plant retired 5 years early.

· Aged Adhesion: Fluorinated backsheet aged 5000 hours, adhesion retention 95% (still 45N/cm); non-fluorinated only 70% (28N/cm). A plant using non-fluorinated back sheet had 20% delamination after 10 years, repair cost 0.2 RMB/W.

Salt Spray Corrosion: The Lifeline for Coastal Power Plants

Coastal areas have high salt spray concentration (chloride ion 0.1-0.3 mg/m³). Backsheet corrosion resistance directly determines lifespan.

· Salt Spray Test Data: TPT back sheet after 1000 hours salt spray (equivalent to 3 years coastal): corrosion depth <10μm (standard ≤20μm); non-fluorinated backsheet corrosion depth 30μm, surface pitting, moisture and chloride ions ingress directly.

· Real Case: A coastal plant in Guangdong used non-fluorinated back sheet, after 5 years module edge delamination 40%, power degradation 12%; after switching to TPT, after 10 years delamination <5%, degradation only 4%.

· Repair Cost: Delaminated modules need back sheet replacement, labor+material cost 0.3 RMB/W. A 1MW plant has 1000 modules, replacement costs 300,000 RMB, enough to buy 100 new modules.