Not all cells are equal. India's solar sector is mid-transition between three architectures — and that transition is itself the investment vector.
Three years ago, PERC was the consensus architecture — proven, bankable, cheap. By 2024, that calculus broke. TOPCon ate the volume tier. HJT is eating the premium tier. Whoever is still running PERC at scale in FY27 is not just efficiency-disadvantaged — they are tender-disadvantaged.
The cell architecture beneath a solar module is the physics engine that sets every downstream number — module wattage, project yield, tender competitiveness. PERC was that engine for a decade. It has reached its ceiling. TOPCon is replacing it at volume. HJT is next.
PERC's hard physics limit is rear-surface recombination — minority carriers reach the back contact and are lost before contributing current. The theoretical ceiling is 24.5%; commercial lines run at 22–23%. No engineering path exists beyond it within the PERC architecture. TOPCon solves this with a 1.5 nm tunnel oxide that eliminates rear recombination entirely, delivering 24–26% efficiency at capex now within 20–30% of PERC. HJT goes further — amorphous silicon on both surfaces, 26–27.5% efficiency, best temperature coefficient of any silicon technology — but requires an entirely new line and consumes more than double the silver per watt. The staircase below shows why this transition is irreversible.
Passivated Emitter Rear Cell. P-type silicon with Al₂O₃ rear passivation. Was the dominant global architecture through 2022. Hard ceiling at 24.5% — minority carriers at the rear contact recombine faster than any passivation layer can suppress. Commercial lines sit at 22–23%, near that limit. No path higher exists within the architecture.
No engineering path exists beyond this limit within the PERC architecture. ALMM List-II exclusion from June 2026 makes this a tender risk, not just a technology gap.
Tunnel Oxide Passivated Contact. A 1.5 nm SiOₓ tunnel oxide on the rear surface stops recombination at the quantum level — electrons pass through, recombination cannot. N-type silicon base gives lower bulk recombination than PERC's p-type. The result: +2–3 pp efficiency and a temperature coefficient of −0.29%/°C that adds 8–12% more annual yield at Indian ambient temperatures. Convertible from PERC lines at moderate capex (~₹1–2 Cr/GW upgrade) — which is why all five major players are here now.
Dominant for FY27 tenders. Waaree (15.5 GW target), Adani (10 GW), Tata Power Solar (4.8 GW ALMM), Premier (10 GW) are all running this transition.
Heterojunction Technology. Amorphous silicon (a-Si:H) deposited on both surfaces of an n-type crystalline wafer at <200°C — no high-temp diffusion, no PERC line reuse. Both surfaces passivated simultaneously to near-perfect quality. Best-in-class temperature coefficient (−0.24%/°C), bifaciality above 90%, and <0.5% first-year LID. The constraint: 190 mg/W silver versus PERC's 85 mg/W — a cost exposure that copper electroplating targets to reduce to 140 mg/W by FY27.
Reliance Jamnagar targets 10 GW+ skipping PERC entirely. 2 GW operational today; 8–10 GW by FY28 is the stated target across all players.
Perovskite-silicon tandem puts two sub-cells in series: perovskite on top (1.7 eV bandgap, blue/green photons) and silicon below (1.1 eV, infrared). A recombination junction connects them. Together they harvest spectrum that single-junction silicon cannot touch — hence 33.9% at lab scale. Three things block commercial use: 25-year stability under UV/moisture, deposition uniformity at module scale, and lead (RoHS). None resolved yet.
Watch signal: credible 25-year warranty with TÜV/UL/BV certification. No manufacturer has issued one yet. When they do, tandem is commercially ready.
Every number in this table encodes a trade-off. Understanding what each parameter means for project economics — not just for the cell — is the analytical work this table is designed to do.
Read every row of the table below as a trade-off. PERC is cheapest — fully depreciated lines, lowest silver at 85 mg/W — but its efficiency ceiling is exhausted and it now competes on cost alone. TOPCon's −0.29%/°C temperature coefficient versus PERC's −0.35%/°C translates to 8–12% more annual energy yield in Rajasthan temperatures — that difference, not the capex, is why it's winning tenders. HJT's −0.24%/°C is the best of any silicon technology; its 90%+ bifaciality is unmatched. The single risk: silver intensity at 190 mg/W means HJT margins are directly exposed to commodity prices. Copper electroplating reduces this — but not before FY27 at Indian scale.
| Parameter | PERC P-type · Legacy |
TOPCon Volume N-type · Dominant transition |
HJT Premium N-type · Efficiency leader |
|---|---|---|---|
| Cell Efficiency (Commercial) | 22–23% | 24–26% | 26–27.5% |
| Theoretical Efficiency Limit | ~24.5% (largely tapped) | ~28–29% | ~29–30%+ |
| Temp. Coefficient (Pmax) | −0.35% / °C | −0.29% / °C | −0.24% / °C ← best |
| Bifaciality Factor | ~65–70% | 80–85% | 90–95% ← best |
| Silver Intensity | ~85 mg/W | ~110 mg/W | ~190 mg/W (→ ~140 with Cu plating) |
| Process Temperature | > 800 °C | > 800 °C | < 200 °C — no PERC line reuse |
| Capex / GW (India est.) | ₹3–4 Cr | ₹4–6 Cr | ₹7–9 Cr |
| Module Power (M10 wafer) | ~410–435 W | ~440–470 W | ~460–495 W |
| Degradation (Yr 1 LID) | ~1.5–2.5% (LID risk) | < 1% (n-type) | < 0.5% (minimal LID) |
| India Status (FY26) | Operational / declining | Rapid ramp — 15+ GW planned | Early stage · ~2 GW op. · 8+ GW planned |
At 190 mg/W, one GW of HJT production consumes 190 tonnes of silver annually. At ₹80,000/kg spot, the HJT–PERC silver delta adds ₹0.84 Cr/MW in raw material cost alone. Copper electroplating can close 55% of this gap — but no Indian manufacturer has demonstrated it in volume yet. Until FY27, silver is the number every HJT capex model must stress-test.
TOPCon compressed from ₹9 Cr/GW to ₹4 Cr/GW in three years. HJT started at ₹13 Cr/GW and is tracking the same deflation curve — two years behind. The gap converges at FY27–28, driven by Chinese equipment deflation, copper metallisation, and Jamnagar scale.
Five names. Three architectures. One June 2026 compliance deadline. Here is where each player stands.
ALMM List-II from June 2026 requires domestically manufactured n-type cells for government tenders. Architecture is no longer a technical footnote — it is tender eligibility. The five names below are not moving in parallel. Reliance is the outlier: the only one going HJT-direct, skipping PERC and TOPCon entirely.
Perovskite-silicon tandem holds the theoretical key to breaking 30% at commercial cost. The physics is settled. The engineering, the stability, and the supply chain are not — yet.
Tandem stacks a perovskite top cell (1.7 eV bandgap, blue-green photons) on a silicon bottom cell (1.1 eV, infrared) — harvesting spectrum that single-junction silicon cannot reach. The lab record is 33.9% (LONGi/KAUST, 2024). Three things block commercialisation: stability under UV and moisture over 25 years, scale from 1 cm² lab cells to 2 m² modules, and lead content for RoHS-restricted markets. None is solved yet.
Silver runs through every architecture. First Indian manufacturer to demonstrate Ag-free HJT in volume production cuts ₹0.84 Cr/MW from cell BOM and resets the HJT-vs-TOPCon cost comparison entirely. Watch copper electroplating adoption — it is the process inflection that changes the economics before tandem ever arrives.
HJT–PERC silver delta today. First Indian line running Ag-free at scale halves this gap and resets the architecture comparison.
Reliance HJT hitting 3 GW+ operational by FY27 closes the architecture gap faster than any competitor can respond to.
No manufacturer has issued a credible 25-year tandem degradation warranty. When one does — with TÜV/UL/BV certification — tandem is real. Until then: lab data only.
Domestic n-type cell compliance mandatory for all government tenders from this date. PERC-only manufacturers face exclusion from India's largest procurement channel.
The cell technology race in India is no longer theoretical — it is being settled by capex commitments made right now, in FY25–FY26 board rooms. Whoever is still running PERC at scale in FY27 is not just efficiency-disadvantaged; they are tender-disadvantaged, because the market is repricing on watts per rupee, not just rupees per watt. The architecture you bet on today is the margin structure you inherit in 2028. There is no neutral position in this transition — only a question of whether you chose your side early enough to build the cost curve that makes the premium defensible.
Front Wave Research LLP publishes institutional-grade equity research, sector strategy, and policy-grounded investment intelligence. Follow us on the channel that suits you to receive each note as it goes live — or message us directly on WhatsApp.