# Tunnel Oxide Passivated Contact (TOPCON) for Solar Cells Topic radar · 1,747 documents · 2021-01-01–2026-10-28 · updated 2026-10-28 Canonical: https://slate.greyb.com/radar/tunnel-oxide-passivated-contact-topcon-for-solar-cells/innovations ## In one minute ### A one-nanometre oxide that conducts The cell rests on a contradiction resolved at atomic scale. A silicon oxide layer a few atoms thick should insulate. Here it carries one carrier type and blocks the other, and the whole advantage of the architecture follows from that. Two mechanisms do the work. Where the oxide is thin and uniform, below roughly 1.5 nm, carriers tunnel straight through. Above that, conduction concentrates in pinholes, nanoscale regions where the oxide is thinner or interrupted [1]. Working contacts show pinhole densities of 10^6 to 10^8 per cm² and still deliver contact resistivity around 10 mΩ·cm², with saturation current densities below 10 fA/cm² [2]. Whether pinholes are damage or design is unsettled, and the answer changes what a fab should measure. Modelling from Jülich found defect-assisted channels through the oxide explain transport better than pure tunnelling [3]. Delft found that moderate pinhole densities improved passivation for both polarities [4]. Molecular dynamics found the opposite risk at high hydrogen concentration, where pinhole nucleation barriers fall and dangling bonds increase [5]. A filing from October 2026 claims a two-layer oxide in which the outer layer is deliberately given a higher pinhole density than the inner one [6]. That is an engineered defect profile, claimed as a product. 1. [10.1063/1.5081832](https://doi.org/10.1063/1.5081832) 2. [10.1063/1.4986924](https://doi.org/10.1063/1.4986924) 3. [10.1088/0256-307x/38/3/036301](https://doi.org/10.1088/0256-307x/38/3/036301) 4. [10.1016/j.solmat.2023.112200](https://doi.org/10.1016/j.solmat.2023.112200) 5. [10.1021/acsaem.4c00171](https://doi.org/10.1021/acsaem.4c00171) 6. [CN122952747A](/?paper=CN122952747A&utm_source=radar&utm_medium=dashboard) ### The cell became a commodity, so the sale moved to the module Every TOPCon launch at India's REI Expo in late October was a module. Avaada showed a 745 W glass-to-glass panel at 23.98% module efficiency, with 132 half-cells and 18 busbars. Emmvee launched a quarter-cut design at 640 to 650 W, 25 to 30 W above a half-cut module of the same size. Saatvik and Solex both claimed 670 W and 24.8% with 264 quarter-cut cells. Shakti Pumps, a pump maker, entered with a 620 W 16-busbar panel and ALMM papers submitted. Sunbond wrapped its frame in glass-reinforced polymer so it will not pit in salt air. ### Six cell makers hold the invention, and the map is short Of the 1,203 patents published on this topic since 2021, 1,042 came from the Chinese office. WIPO accounts for 43, the United States 40, Japan 15, and Australia and Europe 12 each. India accounts for 9. ### The contact became the industry's standard cell, made two ways Most crystalline silicon cells now use this structure. The industry roadmap has it holding the lead through the mid-2030s, at roughly 70 to 80% of production in 2025 depending on the forecaster. Roadmap mass-production efficiency is 25.6% for 2026, rising to 26.4% by 2036. Chinese leaders report 26.1 to 26.5% at cell level and yields near 97%, against 25.1 to 25.3% for cells measured after module assembly. ### A cell line is sold as a machine, a recipe and a patent licence in one package New capacity is bought as a bundle, and which bundle depends on who is buying. ### The most prolific recent filer is working on the oxide interface In patents published from June 2025 onward, Aiko's legal names account for 64, Trina 43, Tongwei 27, Dmegc 21, Das Solar 19, Jietai 17 and Jinko 17. ### The invention concentrates in the stack, and the stack became a process window Of the 1,203 patents published on this topic, 855 sit in passivation and deposition, more than every other area together. That concentration is the clearest thing the patent record says: the field is deepening in one step. ### India's protected price is what makes those orders signable The buying wave answers a rule, not demand. From 1 June 2026 India required domestic cells for government-backed projects, at a point when approved cell capacity was roughly 31 GW against 193 GW of module capacity. Approved cell capacity passed 35,467 MW by August, with 3,709 MW added in one month. ## Overview TOPCon has stopped being an invention and become an industry. The contact is standard, both deposition routes are catalogued, and listed cell efficiencies sit in a band narrow enough that no buyer chooses on the cell's architecture. Three things remain genuinely contested: who holds the process window inside the furnace, who may sell at an administered price, and whether the same one-nanometre oxide survives a module or ends up as a component inside a perovskite tandem. ## Research ### Removing the polysilicon where it is not needed is now the main engineering line The doped layer absorbs light that should reach the wafer, so it should be thin. It has to survive silver paste firing, so it should be thick. The way out that the industry chose is to keep it thick only under the metal. ### Contacting a thin stack is a laser step, and the metal above it is changing The metallization problem follows from the stack. Firing has to open the dielectric hard enough to make ohmic contact and gently enough to leave a 1.5 nm oxide and 100 nm of polysilicon intact. Laser-enhanced contact firing has become the standard answer, and it changed what the emitter can be: sheet resistance of 300 to 500 Ω/sq is now reachable against roughly 300 in ordinary production. ### Packaging decides what the contact is worth, and tandem decides its next use The passivation advantage is fragile in the module. Lamination, soldering heat and damp encapsulant all act on the same few nanometres, and the measured cost of that transfer is the largest number in the downstream work. A certified tandem cell has reached 34.82%, while a shingled tandem module built to avoid soldering heat reached 25.6%, about level with a good single-junction TOPCon module already on sale. ### What has not been built yet Four gaps matter for anyone reading the build state rather than the filings. ## Market ### The US floor is a separate market, and the licence is the product Section 232 minimum import prices take effect on 4 December 2026 at 0.22/W for cells and 0.38/W for modules. Chinese TOPCon cells were trading near 0.0463/W and modules near 0.107/W in late October. On those numbers the cell floor sits at roughly five times the traded price. ### Silver is the one cell-level cost that visibly changed what buyers ordered The production announcements in metallization are about metal, and the biggest is an 8 GW retrofit. Jolywood has begun mass production of cells using silver on the front and market-bought aluminium paste on the rear, above 27%, needing only extra laser equipment and compatible with existing lines. Websol cut silver 20% in the last financial year and targets another 10% while converting a PERC line. ### Reliability became a purchase condition before it became a product claim Kiwa PVEL's 2025 scorecard put the share of manufacturers with at least one test failure at 83%, up from 66%. Munich Re made the UV60 stress test mandatory for its warranty partners in mid-2024, on the ground that the standard test applies 15 kWh/m² of ultraviolet while degradation appears near 60. A draft IEC technical specification is being written around that distinction, and Trina's researchers have proposed splitting the loss into reversible and irreversible forms. ### Tandem is being sold as an add-on glass, with 2028 as the date Caelux signed 3 GW with INA Solar and 5 GW with Navitas on the same model: deposit perovskite on a glass sheet and lay it over a finished TOPCon module, targeting 28% to 29%. The cell line is untouched. Commercial production is targeted for 2028, after technology validation, manufacturing readiness and ALMM approval. ### Two things have not reached the market, and both are the next question Tandem has no gigawatt line. Every deal in this record is an agreement with a validation phase and a 2028 production date. ## Strategy ### Suppliers appear only where a step needs a machine or a paste Metallization and patterning are where names from outside the cell business turn up. In metallization, Jiangsu Riyu holds 8 documents, Qiming Guangda 7, DKEM 6 and Foshan Ruina 6. In patterning, Dr Laser holds 5 recent filings and LAPLACE 2. ### The effort has moved outward to the module, and the patterning wave is already filed The composition of the patent record has shifted more than its size. Module integration and encapsulation publications went 3 in 2021, 8 in 2024 and 28 in 2025, then 44 in the ten months to 28 October 2026. Passivation and deposition went from 37 to 263 over the same span. ### The market record is growing faster than the invention record 315 news articles were published in the ten months to 28 October 2026, against 130 in all of 2025 and 52 in 2024. Patents ran 348 in 2025 and 343 in the same ten months of 2026. ### The deposition leaders are already filing for tandem The companies holding the stack work are the same ones claiming the next architecture. Trina reports 758 perovskite patent applications and a 29.2% industrial tandem module certified at 907 W. Jinko reports more than 700 N-type TOPCon patents and a 34.82% tandem cell. ### The firms building the new capacity barely appear as filers Emmvee, Waaree, SAEL, Premier, Vikram and ReNew run through the market documents. None of them appears as a filer in the patents collected here, and the Indian office accounts for 9 documents against China's 1,042. ### Two limits on reading the trend Publication lags mean the 2026 patent counts are incomplete for anything filed in the last eighteen months, which understates the most recent shift and overstates the fall in patterning. ## Areas - Midstream: Passivation & Deposition · 1,060 documents - Commercial: Market, Legal & Capacity · 288 documents - Downstream: Metallization & Grid · 183 documents - Upstream: Wafer & Tunnel Oxide · 63 documents - Midstream: Cell Patterning & Contact Formation · 57 documents - Downstream: Module Integration & Encapsulation · 92 documents