What Materials Are Used in Manufacturing High Efficiency Photovoltaic Cells

High-efficiency photovoltaic cells use ultra-pure monocrystalline silicon (99.9999%) for the wafer base, phosphorus-doped N-type silicon (achieving 24%+ efficiency), anti-reflective silicon nitride coating (reducing 35% light loss), silver paste front contacts (conducting 95%+ electrons), aluminum back-surface fields (boosting voltage by 0.5V), and PERC layers (reflecting 20% more light), with manufacturers using laser doping to optimize electron pathways.

Purity Requirements for Silicon Wafers

Last summer, a major silicon wafer factory's production line suddenly turned red—an 182mm wafer under EL detection looked like it was sprinkled with black sesame seeds, and minority carrier lifetime plummeted from 8μs to 1.2μs. As a process engineer responsible for 16GW of monocrystalline projects, I grabbed the walkie-talkie and rushed to the crystal pulling workshop: This must be an oxygen-carbon ratio gone wild!

Nowadays, every 0.1% increase in PV cell efficiency requires extreme efforts,the purity of silicon material is the bottleneck. Did you know? When oxygen content in ingots exceeds 18ppma (parts per million atomic concentration), cells under sunlight behave like leaking power banks—photogenerated electrons are "eaten" by impurities before reaching electrodes.

The culprit behind that accident was found later—it was 0.3% nitrogen mixed into the argon supply pipe. Don't underestimate this tiny amount of impurity; at 1600℃ in the monocrystalline furnace, nitrogen acts like a stirrer, disrupting the vertical growth of the crystal lattice into a winding mountain road. An 8-meter silicon ingot produced under these conditions resulted in wafers with efficiencies below 23%.

The industry now plays "three-degree control":

· Temperature gradient must be controlled within 3℃/cm, otherwise convective currents in the silicon melt will carry impurities into the crystal

· Argon flow rate needs to be steady between 28-32m³/h; too fast cools down heat, too slow fails to clean volatiles

· Seed crystal rotation deviation cannot exceed 0.5 degrees, misalignment leads to twisted growth

Last month, a friend’s factory had an incident—their newly replaced graphite crucible had burrs of 0.05mm. This minor flaw caused spiral dislocations during crystal growth on the third day, rendering the entire batch of silicon materials as construction waste-grade rejects.

Detection methods have also been upgraded. For example, the online monitoring system we installed for a TOPCon cell manufacturer can monitor in real-time during crystal growth:

① Melt surface fluctuation ±0.1mm triggers automatic alarms

② Oxygen concentration sampled every 30 seconds

③ Thermal field infrared imaging accuracy reaches 0.5℃

This system boosted their rod yield from 82% to 91%, equivalent to earning an additional 3 million RMB per month.

Honestly speaking, what bosses in the silicon wafer business fear most isn’t expensive equipment but rather spending 20 million RMB on a monocrystalline furnace only to see it ruined by a single metal dust particle floating into the workshop. During my last visit to a G12 large silicon wafer factory, even the operators' phone cases had to be made of anti-static materials.

The Conductivity Mystery of Silver Paste

Last month, a TOPCon cell factory saw a sudden spike in main grid break points to 6.8%, directly scrapping 3.2GW of monthly capacity. As a materials engineer involved in the development of seven generations of silver paste formulas, I found that this issue reveals the complexity of PV silver paste.

The current mainstream silver paste's conductivity is akin to building highways on silicon wafers using nanoscale silver powder. Data from a certain HJT production line shows that when silver particle diameter decreases from 1.5μm to 0.8μm, contact resistance drops from 3.6mΩ·cm² to 0.9mΩ·cm². However, this isn't straightforward—last year, one manufacturer increased silver content from 65% to 72%, resulting in burn-through of a 3μm passivation layer during sintering.

Last year, our lab conducted extreme tests: Using SEM scans, we found that when sintering peak temperature exceeds 820℃, glass phases erode emitters like lava. This led to an 182 cell's fill factor dropping from 82.3% to 76.1%, with EL imaging showing spider-web-like conductive defects in localized areas.

· Silver paste printing is more delicate than cake decorating—a 200N difference in blade pressure results in ±8μm fluctuations in line width

· A certain double-sided PERC module experienced silver ion migration during damp heat testing due to incomplete curing of backside silver paste

· The industry is currently testing silver-coated copper technology, which can reduce silver content to 30% while maintaining 95% conductivity

Earlier this year, a leading company attempted to use nanosilver wires instead of spherical silver powder, successfully enduring 1200 hours of damp heat aging in DH2000 tests. By adjusting the proportion of terpineol in the organic carrier, they stabilized printing line widths to ≤18μm—this precision is equivalent to drawing hair-thin lines on A4 paper.

However, what shocked me most was during an accident analysis: A batch of silver paste experienced unexpected storage temperature fluctuations of 8℃, causing thixotropic index to jump from 1.8 to 3.2. During printing, the paste flowed intermittently like toothpaste, resulting in over ten million RMB losses due to fine grid break points that month. Therefore, our warehouse now has a dual-channel temperature control system, triggering alarms if temperatures fluctuate beyond ±0.5℃.

Anti-Reflective Coating

Last summer, a TOPCon cell factory went crazy—one of their 2.1GW modules exported to Europe was detained at Rotterdam port due to exceeding reflection rates by 0.8%. The process manager Old Zhang called me at 3 AM: "Why does this anti-reflection film act like a pancake—too thin it leaks light, too thick it turns white?" As an engineer with eight years of experience in thin-film deposition, I understood his pain all too well.

The mainstream silicon nitride coating is an "optical magician", performing best when thickness is controlled at 80nm±5nm. This is equivalent to 1/800th the diameter of a human hair, yet achieving precise stacking on silicon surfaces. According to SEMI's M40-0218 standard, when refractive index exceeds 2.1, deposition temperature must be kept below 420℃, otherwise hydrogen atoms escape from the coating, similar to how eggs get scorched in overheated pans.
Recently, two popular approaches emerged:
1. Using PECVD (Plasma Enhanced Chemical Vapor Deposition) for "low-temperature simmering", akin to slowly cooking soup
2. Employing magnetron sputtering for "precise strikes", resembling 3D-printed nano-level operations
However, test data showed that for a certain 182mm wafer using the second method, reflectance in the edge 5mm region unexpectedly increased by 0.3%, making the cell look like wearing reflective sunglasses.

Last month, during a visit to an N-type cell factory, their engineer showed me an interesting technique—adding titanium oxide nanoparticles to the coating. This enhances both light absorption and path length, much like adding fluorescent powder to glass. Test results showed CTM (Conversion Efficiency of Modules) values jumped from 24.1% to 24.7%, but costs rose by 0.08 yuan per watt, sending the production line manager's blood pressure soaring.

The hardest part is uniformity control. Once, a night shift encountered fluctuations in argon purity, resulting in rainbow patterns appearing on EL images of the next day's wafers. After three hours comparing color cards, it was discovered that a temperature difference exceeding 2℃ in the reaction chamber caused this—precision requirements rival those of embroidery on a high-speed train.

New laser etching technologies are emerging, reportedly reducing reflectance by another 0.5%. However, during trial runs at a neighboring factory, improper power adjustment led to honeycomb structures forming in the coating, causing photovoltaic conversion efficiency to plummet by 2.1%. So never believe in "one-click optimization"—every decimal point in this industry represents lessons learned through hard cash.

Recently, while debugging dual-layer anti-reflective structures for a major manufacturer, we observed a mysterious phenomenon: When the first layer thickness reached 67nm, the deposition rate of the second layer automatically increased by 12%. Junior engineer Xiao Wang joked that the coating might have become sentient. But after running simulations with VG software, it turned out to be a chain reaction triggered by plasma resonance—apparently, even physical laws are getting competitive.

Backsheet Material Showdown

Last month at a PV power plant in Hainan, maintenance staff found that modules using PET backsheets exhibited snowflake-like dark spots during EL imaging. Testing revealed that water vapor transmission rates were three times over standard, directly causing a 19% power degradation in the entire batch. As a PV material researcher who has participated in the certification of 8GW backsheet materials, I have dissected countless failed samples piled up in the thermal field control room — the subtle differences in backsheet material selection can truly impact a power plant's 25-year revenue by the price of a BMW 5 Series.

Last year, a top 10 module manufacturer in Ningxia pulled off a bold move: replacing TPT with TPE to cut costs. However, three years later, edge delamination occurred like squid rolls, increasing O&M costs by an additional 12 million RMB. The core issue lies in fluoropolymer film thickness — when the surface layer is less than 20μm thick (SEMI PV22-018 critical value), UV penetration rates jump from 5% to 23%, akin to poking holes in synthetic leather sun protection clothing.

Currently, the biggest headache is the clash between glass backsheet advocates and flexible material supporters. Last week, we tested a bifacial glass module under wind speeds of 9m/s, finding that load deformation exceeded limits by 1.8mm (IEC 61215 standard limit is 1.2mm), even distorting the mounting structure. Yet, its power generation performance was indeed impressive — earning 8% more morning and evening low-light benefits compared to TPT backsheets, a temptation similar to seeing zero-sugar cola for someone craving a drink.

A fishery-solar integration project in Guangdong went even further: adding titanium dioxide particles to TPE backsheets reduced water vapor barrier rates from 0.8g/m²·day to 0.3. However, this came at the cost of a 4% reduction in light transmittance, resulting in a CTM loss rate 1.2 percentage points higher than conventional solutions. This scenario resembles adding soy sauce to milk tea; it raises questions about whether the goal is hydration or flavor — a real dilemma.

The arms race in backsheet materials has escalated to the molecular level. A recently filed patent (CN202410387196.2) by a listed company shows that three-layer co-extrusion technology can reduce interfacial bubble rates from 3% to 0.7%, but production line speeds must drop from 25m/min to 18m. This is akin to asking Usain Bolt to hurdle while doing embroidery, making mass production extremely challenging.

Recently, there's been a trend towards composite materials, such as applying a 0.1mm fluoropolymer film on glass backsheets. Test data from a Zhejiang factory showed that this hybrid solution could stabilize PID degradation rates within 1.5%/year, though costs rose to 1.8 RMB/W. According to the 2025 forecasted LCOE model, unless silicon material prices crash again, power plant owners are likely to continue choosing cheaper options.

Busbar Selection Tips

At 3 AM, Engineer Zhang from a top 10 module manufacturer stared at black spots on the EL detector, sighing — de-soldering occurred during the lamination process in this batch of 182 bifacial modules. Such frustrations are all too common in the industry, where improper busbar selection can lead to a 0.3% CTM loss, eating into module profits.

Last year, while leading a process upgrade for a 5GW base, we found that every 0.01mm increase in busbar thickness results in an 18% rise in mechanical stress on cells (verified by SEMI PV22-0618 data). But thinner isn't always better; one cross-industry player used ultra-thin 0.2mm busbars, only to see junction box detachment rates soar to 4.7% after three months, tripling industry standards.

Last month, a typical case involved inadequate coating thickness on busbars used in Southeast Asian projects, resulting in a 30% surge in contact resistance after three months of damp heat testing. Upon disassembly, the busbar surface oxidation layer was thicker than hair strands. This condition blocks current flow like clogged blood vessels.

· When tin plating thickness is <2μm, salt spray tests lasting 120 hours will inevitably show oxidation spots

· Copper substrate purity must exceed 99.99% (don't trust suppliers claiming 4N copper; verification through spectrometer is necessary)

· Bend radius should be >1.5 times the busbar thickness to avoid jamming in string welders

A recent innovation involves using dull-finish busbars instead of shiny ones. In tests on 166 modules, reflection rates dropped by 1.2%, yet overall power output increased by 0.8%. The principle is simple — dull surfaces refract more light back onto the cells.

Another practical detail: never skimp on dehumidifier electricity costs for busbar storage. Last summer, humidity levels in a warehouse led to 3 million RMB worth of busbars developing verdigris. We established a strict rule: environmental dew point must be controlled below -10℃, and all opened packages must be used within 24 hours.

The industry is increasingly experimenting with segmented busbars and shaped busbars. However, based on our lab's destructive testing, the gains from fancy structures often fail to offset yield losses. It's like fitting a supercar engine into a plane — great specs but impractical for long-distance travel.

Finally, here's a tip for avoiding pitfalls: upon receiving new busbar samples, don't rush them onto the production line. First, use scrap cells for three sets of experiments — 200 thermal cycles, 85℃/85% humidity aging test, and 5000 mechanical load cycles. If these hurdles are cleared, then discuss mass production introduction.

Future Black Tech Materials

The PV community was recently shaken by news that GCL Optoelectronics' perovskite pilot line produced modules measuring 1m×2m, achieving mass production efficiencies of 18.6%, nearly doubling laboratory data from three years ago. As a material engineer who has worked on 5GW N-type TOPCon projects, I've witnessed entire silicon ingots scrapped due to oxygen content fluctuations of 0.5ppma, while new-generation materials are rewriting these old scripts.

Currently, three key contenders are hidden in labs: perovskites, quantum dots, and 2D materials. Perovskites, being dramatic actors, have a light absorption coefficient 100 times greater than crystalline silicon. However, last year, a major manufacturer's pilot modules couldn't withstand 200 hours under 85℃/85% humidity conditions. It wasn't until MIT researchers developed fluorinated interface passivation technology, reducing damp heat degradation rates to 3%/1000 hours, that things became promising.

· Quantum dots act like chameleons in the PV world, absorbing different wavelengths of light by adjusting particle size

· Materials like tungsten diselenide, when thinned to atomic layers, exhibit carrier mobility that soars

· Organic-inorganic hybrid materials play molecular-level Lego, reducing defect state densities to 10¹⁰ cm⁻³ levels

At last month's SEMI seminar, an engineer from Hanergy secretly showed me a video — their full-solution perovskite production line churns out 200 glass substrates per hour. Traditional PECVD equipment would consume enough vacuum pump electricity to make plant managers cry. However, issues remain; reportedly, one batch's EL imaging showed honeycomb-like dark spots, traced back to uneven crystallization caused by a 0.5m/min difference in coating speed.

Most eye-opening was Germany's Jülich Research Center's layer stacking magic. They stacked perovskites and organic materials like a thousand-layer pancake, achieving open-circuit voltages of 2.8V under AM1.5 illumination, significantly higher than single-junction batteries. However, the spin-coating method used in labs becomes disastrous on production lines. Last year, a company attempted to replace it with slot-die coating, resulting in interface recombination losses spiking to 28%.

An industry joke goes: those working on perovskites are betting on material longevity, those on quantum dots are competing on synthesis precision, and those on 2D materials are mastering transfer techniques. Just like when monocrystalline silicon clashed with multicrystalline silicon, whoever breaks the mass production curse first will reap the super dividends of carbon neutrality. Looking at these delicate new materials in the lab, I silently checked the argon purity monitor on the monocrystal furnace — at least with silicon wafers, we know the pain well.