Inside the factory, step by step
A solar panel starts as quartz sand and ends as a sheet of glass with a few grams of silver printed on silicon behind it. Between those two points sit six industrial stages, temperatures above 1,700°C, and a supply chain that is now a live political question in the UK.
Most explanations of solar panel manufacturing describe a product that has largely stopped being made. They walk through polycrystalline wafers, boron-doped p-type silicon and PERC cells, which was an accurate picture in about 2019. The panel going on a British roof in 2026 is built differently, and the differences change what it costs, how long it lasts and who made it.
- Six stages turn sand into a panel: quartz to metallurgical silicon, purification into polysilicon, ingot growth, wafer slicing, cell processing, and module assembly. Only the last two happen in the same factory as each other in most cases.
- By weight a panel is mostly glass. Around 76% glass, 10% polymer, 8% aluminium, 5% silicon, 1% copper and under 0.1% silver. The photovoltaic part is a small fraction of what you lift onto the roof.
- The cell technology changed almost overnight. TOPCon accounted for roughly 85% of global cell production at the start of 2026, with PERC down below 5%. Guides describing p-type PERC are describing a legacy product.
- Making the silicon is where the energy goes. Polysilicon and wafer production account for around 70% of a panel’s manufacturing energy. In UK sunlight a panel repays that energy in roughly 1.5 to 2.5 years, then runs for another 25.
- The panel is about a tenth of what you pay. A 440W module leaves the factory for around £38. The line on your UK quote is usually £180 to £220, because installation businesses are priced on the job, not the hardware.
What a solar panel is actually made of
Before the process makes sense, it helps to know the bill of materials. Lift a modern 440W module and you are holding roughly 22 kg, of which the great majority is a single sheet of toughened glass.
| Material | Share | What it does |
|---|---|---|
| Toughened glass | ~76% | Front cover, 2-3.2mm, low-iron for light transmission |
| Polymers | ~10% | EVA or POE encapsulant plus backsheet film |
| Aluminium | ~8% | Anodised frame and mounting flanges |
| Silicon | ~5% | The cells themselves, the only part that generates |
| Copper | ~1% | Interconnect ribbons, busbars, junction box wiring |
| Silver and other metals | under 0.1% | Screen-printed contacts. Tiny by mass, large by cost |
That last row is the one to remember. Silver is a rounding error by weight and one of the biggest cost battles in the industry, which is a theme we return to below.
- Polysilicon
- Silicon purified to at least 99.9999% (six nines, written 6N) and up to eleven nines. The feedstock for every crystalline panel.
- Ingot
- The solid cylinder or block of crystallised silicon grown from molten polysilicon before it is sliced.
- Wafer
- A slice of that ingot, now typically 130-150 microns thick, roughly the thickness of a sheet of paper.
- Doping
- Introducing controlled impurities, usually phosphorus or boron, so the silicon carries a positive or negative charge and forms a junction.
- Passivation
- Surface treatments that stop generated electrons recombining before they reach a contact. Almost all recent efficiency gains come from better passivation.
- PERC / TOPCon / HJT
- Three cell architectures. PERC is the p-type design that dominated until 2023. TOPCon adds a tunnel oxide and doped polysilicon layer on the rear. HJT sandwiches amorphous silicon films around a crystalline wafer.
- Lamination
- The vacuum and heat step that melts the encapsulant into a single sealed sandwich around the cells.
- Flash test
- A calibrated light pulse used to measure a finished panel’s output at standard test conditions before it is boxed.
From sand to solar-grade polysilicon
Panels begin as quartz, a rock made of silicon dioxide. Sand on a beach is the same compound, though commercial production uses mined high-purity quartzite rather than anything scooped off a shore.
Carbothermic reduction
Crushed quartzite goes into a submerged arc furnace with a carbon source, usually coal, charcoal and woodchip. At around 1,700 to 2,000°C the carbon strips the oxygen from the silicon dioxide, leaving metallurgical-grade silicon at roughly 98 to 99% purity and releasing carbon dioxide. This step alone takes about 14 to 16 kWh of electricity per kilogram of silicon produced.
Purification to six nines
Metallurgical-grade silicon is nowhere near clean enough. Around 90% of the world’s solar-grade material is then refined through the Siemens process: the silicon is converted into trichlorosilane gas, distilled, and deposited back onto heated silicon filaments as ultra-pure polysilicon rods. The rest comes mostly from fluidised bed reactors, which produce granular polysilicon at lower energy cost.
The Siemens route consumes roughly 60 to 120 kWh of electricity per kilogram. Because a kilowatt-peak of panels needs a few kilograms of polysilicon, this single stage represents about 170 kWh per kWp installed and roughly 40% of the entire manufacturing energy budget. It is also why polysilicon plants cluster where electricity is cheapest.
Around 60% of the electricity used in global solar manufacturing comes from coal, according to the International Energy Agency, well above coal’s share of world power generation. That is a geography problem rather than a technology problem: the furnaces sit in provinces where coal is the cheap option.
Ingots and wafers
Growing the crystal
Polysilicon chunks are melted at 1,414°C in a quartz crucible. In the Czochralski process a small seed crystal is dipped into the melt and drawn slowly upwards while rotating, pulling a single continuous crystal behind it. The result is a cylindrical monocrystalline ingot, typically 2 to 3 metres long and weighing several hundred kilograms.
Dopant is added to the melt at this point. For decades that meant boron, giving p-type silicon. The shift to TOPCon has flipped this: modern cells are built on phosphorus-doped n-type wafers, which are less prone to the light-induced degradation that used to cost p-type panels a percent or two of output in their first weeks on a roof.
The older alternative, casting molten silicon into a square block to make a multicrystalline ingot, is cheaper and produces the mottled blue panels of a decade ago. It has effectively disappeared from new production because the efficiency penalty is no longer worth the saving.
Squaring, slicing and cleaning
The round ingot is cropped and squared off into a pseudo-square profile so that finished cells tile efficiently across a panel with minimal wasted glass. It is then sliced by a diamond wire saw into wafers.
Two numbers matter here. Wafers have thinned from around 180 microns a decade ago to 130-150 microns today, and saw kerf has fallen to roughly 50 microns, so less silicon ends up as slurry. Wafer format has also standardised: the 182mm by 210mm rectangle is now the industry norm, which is why panels have grown from the old 1.65 metre 60-cell format to today’s 1.9 to 2.1 metre modules.
Sliced wafers are then etched to remove saw damage and textured, usually into a pyramid microstructure that traps light by making it bounce into the surface rather than off it.
Making the cell: the 2026 version
This is the stage where almost every published guide is now out of date. The transition from p-type PERC to n-type TOPCon happened faster than any previous technology change in the industry, largely because existing PERC lines could be upgraded rather than replaced. By the start of 2026 TOPCon accounted for close to 85% of global cell production while PERC had fallen below 5%.
- Wafer
- Boron-doped p-type
- Module efficiency
- 20-21.5%
- Silver use
- Lowest
- Annual degradation
- ~0.5%
- Manufacturing score
- 3.3 / 5
- Wafer
- Phosphorus-doped n-type
- Module efficiency
- 22-24.5%
- Silver use
- ~30% more than PERC
- Annual degradation
- ~0.4%
- Manufacturing score
- 4.5 / 5
- Wafer
- N-type with a-Si films
- Module efficiency
- 22.3-24%
- Silver use
- Highest, low-temp paste
- Annual degradation
- ~0.25%
- Manufacturing score
- 3.9 / 5
The cell line, step by step
An alkaline etch cuts pyramids into the wafer surface to trap incoming light, followed by a chemical clean to strip residues from the saw.
On an n-type wafer, boron is diffused into the front surface at high temperature to create the p-n junction. The old approach, diffusing phosphorus into a p-type wafer, is now the minority route.
The defining TOPCon step. An ultra-thin silicon oxide layer, roughly one to two nanometres, is grown on the rear, then capped with a doped polysilicon film. Together they let charge carriers through while blocking recombination at the metal contact.
Aluminium oxide and silicon nitride films are laid down by plasma deposition. The silicon nitride is what gives cells their dark blue colour and cuts front-surface reflection from about 30% to under 3%.
Silver paste is screen-printed into fine gridlines and fired in a belt furnace so the metal burns through the coating and bonds to the silicon. Line widths are now under 25 microns to shade as little of the cell as possible.
Every cell is flash-tested and binned by current output. Cells with mismatched output in the same string drag down the whole panel, so sorting is what protects the finished module’s rated power.
TOPCon uses roughly a third more silver per watt than PERC did, and silver is one of the few panel inputs whose price has not fallen with scale. This is why manufacturers are pushing silver-coated copper pastes, busbar-free cell designs and thinner gridlines. If panel prices stop falling over the next few years, silver is the most likely reason.
Assembling the module
Cell fabrication and module assembly are different businesses. A module plant buys finished cells, often from a separate company on a separate continent, and turns them into a laminated, framed product.
- Cutting. Cells are laser-scribed and split in half, or into thirds, to cut resistive losses. Almost all modern panels are half-cut, which is why the front looks split down the middle.
- Stringing. A tabber-stringer solders copper ribbons across cells to form strings, then lays strings side by side into a matrix.
- Layup. The sandwich is built glass first, then encapsulant, cells, more encapsulant, then a polymer backsheet or a second glass sheet for bifacial panels.
- Lamination. The stack goes into a vacuum laminator at around 150°C for ten to fifteen minutes. The encapsulant melts, flows around the cells and cross-links into a permanent seal.
- Trimming and framing. Excess encapsulant is cut back and an anodised aluminium frame is fitted with a silicone or butyl sealant.
- Junction box. Bonded to the rear and wired to the string leads, containing the bypass diodes that let current route around a shaded section.
- Curing. Adhesives and potting compound cure before the panel is tested.
Testing and certification
Every finished panel goes through a flash test under standard test conditions: 1,000 watts per square metre of simulated light, a cell temperature of 25°C and an air mass of 1.5. That is where the wattage printed on the label comes from, and it is why panels rarely hit their nameplate figure on a real roof, where the cells run far hotter than 25°C.
Alongside the flash test, panels are checked with electroluminescence imaging. A current is pushed through the cells in a dark chamber and the silicon emits faint infrared light. Cracks, broken fingers and dead cell regions show up as dark patches. It is the closest thing the industry has to an x-ray.
The standards that matter to a UK buyer
- IEC 61215. Design qualification and type approval, covering thermal cycling, damp heat, humidity freeze, hail impact and mechanical load.
- IEC 61730. Photovoltaic module safety qualification, parts 1 and 2.
- IEC 62804. Potential induced degradation testing, worth asking about for any large array.
- MCS certification. Required for the panel to be used in an MCS-registered UK installation, which is what gives access to Smart Export Guarantee tariffs.
- Factory audit reports. Independent site assessments from bodies such as TUV or Bureau Veritas, distinct from the product certificate.
A certificate belongs to a specific product code made at a specific factory. Manufacturers with several plants can and do ship the same model number from different sites. If a datasheet matters to you, ask which facility the batch came from.
What a panel actually costs to make
This is the part almost nobody publishes, and it explains a lot about solar quotes. Working from early 2026 cell and module pricing, here is where the money sits in a 440W TOPCon panel on its way to a British roof. Figures are indicative and move with polysilicon, silver and freight rates.
Two things follow from that table. First, comparing quotes on panel brand alone is close to meaningless, because the hardware is a small slice of the price. Second, the popular idea that panels have a big margin baked in gets the location wrong. Manufacturing margins across the Chinese module industry have been thin to negative through the recent overcapacity cycle. The money in a UK install is in labour, scaffolding, design, certification and warranty risk.
Where UK panels come from
There is no volume crystalline panel manufacturing in the UK. Sharp closed its Wrexham module plant in 2014, and while there is British activity in adjacent technologies, including perovskite cell research and hybrid solar-thermal products, the panels going on British roofs are imported.
China dominates every stage of the chain: polysilicon, wafers, cells and modules. A parliamentary debate in June 2026 put the share of UK panels coming from China at around 68%. Malaysia, Vietnam, Thailand and increasingly India and the United States account for most of the rest, though a good deal of that capacity is Chinese-owned and Chinese-supplied upstream.
The Xinjiang region has been estimated to supply a large share of the world’s solar-grade polysilicon, and campaigners and parliamentary committees have raised forced labour concerns about production there. China denies the allegations.
The UK response has been to amend the Great British Energy Act so that GB Energy cannot back projects with forced labour in the supply chain, with the industry-run Solar Stewardship Initiative used as the assurance mechanism. Coverage of the scheme is growing but not universal, and a June 2026 parliamentary exchange noted that several installers holding public contracts could not guarantee their chains were clear.
What a homeowner can practically ask
Ask the installer which manufacturer and which factory, not just the brand name.
Ask whether the manufacturing site holds Solar Stewardship Initiative ESG certification.
Ask for the IEC 61215 and 61730 certificates matching the exact model code quoted.
Check the manufacturer has a UK or EU entity that can honour a warranty claim.
Assume a European brand name means European manufacturing. Most is assembled in Asia.
Treat a 25-year warranty as meaningful without checking who underwrites it.
Pay a premium for a cell technology your installer cannot name.
Assume the cheapest quote uses the same panel batch as the dearest one.
Energy and carbon payback in UK conditions
The commonly quoted figure is that panels repay their manufacturing energy in four to eight months. That number comes from the International Energy Agency and reflects sunny installation sites. It does not describe a roof in Manchester.
The honest UK arithmetic runs like this. Embodied energy for a crystalline silicon module and its mounting works out at very roughly 1,500 to 2,200 kWh per kilowatt-peak installed. A south-facing UK roof at a decent pitch generates somewhere between 800 and 1,000 kWh per kilowatt-peak per year. Divide one by the other and the energy payback lands at about 1.5 to 2.5 years, with Scotland at the longer end and the south coast at the shorter.
IEA figure, reflecting high-irradiance installation sites
Same panels, British irradiance and pitch
That is still an excellent return. Against a 25 to 30 year working life, a UK panel spends over 90% of its existence in energy surplus. Carbon tells the same story: lifecycle emissions for crystalline silicon PV sit somewhere in the 20 to 45 grams of CO2 equivalent per kilowatt-hour range depending on where it was made, against roughly 490 for gas and 820 for coal.
Mottled blue 60-cell panels, around 265W and 16% module efficiency, on 180 micron wafers.
Rear passivation lifts efficiency past 19%. Uniform black cells replace the mottled look.
N-type wafers with a tunnel oxide rear contact. The decisive factor: existing PERC lines could be upgraded.
PERC below 5%. Typical residential panel: 430-460W, 108 half-cut cells, 22-23% module efficiency.
Laboratory tandem cells have passed 34%. Durability testing and 25-year warranties are the remaining barrier.
What goes wrong, and at which manufacturing stage
Warranty claims trace back to specific points on the production line. Knowing the mapping helps you read a datasheet, and helps an installer diagnose an underperforming array.
| Fault | Manufacturing origin | How it shows up |
|---|---|---|
| Microcracks | Stringing, lamination or handling | Gradual output loss, visible on EL scan |
| Snail trails | Encapsulant and paste chemistry | Dark lines on cell surface after 2-4 years |
| Potential induced degradation | Encapsulant and glass composition | Sharp string underperformance |
| Backsheet cracking | Backsheet film selection | Insulation faults, safety risk |
| Junction box failure | Solder joints and potting | Arcing, hot spots, dead strings |
| Delamination | Lamination time or temperature | Cloudy patches, moisture ingress |
| Light-induced degradation | Boron-doped p-type wafer | Largely designed out by n-type TOPCon |
Microcracks are the most common quality issue and most are introduced after the factory, during transport, handling or a boot on the glass at install. If panels are unloaded roughly on your driveway, that is a legitimate reason to ask for an electroluminescence scan before commissioning.
Recycling and end of life
UK panels fall under the Waste Electrical and Electronic Equipment regulations, so producers carry take-back obligations and panels should not go to landfill. Recycling recovers the aluminium frame and glass readily. The harder part is separating the laminate: the encapsulant is cross-linked by design, which is what makes panels last 30 years and also what makes them awkward to take apart. Silicon, silver and copper recovery is technically possible and commercially marginal at current volumes, though that changes as the first large wave of panels reaches retirement in the 2030s.
The case for and against buying on manufacturing quality
- N-type cells, for lower first-year and annual degradation
- A manufacturer with a UK or EU warranty entity
- Full IEC 61215 and 61730 certification on the exact model
- Independent factory audit or SSI ESG certification
- A low temperature coefficient if your roof runs hot
- A percentage point of module efficiency on an unconstrained roof
- Brand prestige without a named factory behind it
- Bifacial panels on a standard pitched tile roof
- Warranty length beyond the likely life of the supplier
- Marketing terms like “Neo” or “Pro” with no cell type stated
Frequently asked questions
What are solar panels made from?
By weight a crystalline silicon panel is roughly 76% toughened glass, 10% polymer encapsulant and backsheet, 8% aluminium frame, 5% silicon, 1% copper and under 0.1% silver and other metals. The silicon cells are the only part that generates electricity, and they make up a small fraction of the total mass.
Are solar panels really made from sand?
They start from quartz, which is silicon dioxide, the same compound as beach sand. Commercial production uses mined high-purity quartzite rather than sand from a beach, because impurity levels matter enormously. The rock is reduced in an arc furnace at 1,700 to 2,000°C, then purified further until it reaches at least 99.9999% silicon.
How long does it take to make a solar panel?
Module assembly takes about an hour from cells to boxed panel, with lamination the slowest step at ten to fifteen minutes. The full chain is much longer: polysilicon deposition runs for days, and a Czochralski ingot takes two to three days to pull. Counting every stage, silicon mined today reaches a finished panel in a matter of weeks.
Are solar panels made in the UK?
Not at volume. There is no large-scale crystalline module factory in the UK; Sharp’s Wrexham plant closed in 2014. British firms are active in perovskite research, hybrid solar-thermal products, inverters and mounting systems, and the government’s Solar Roadmap sets out an intention to grow domestic supply chain capacity, but the panels themselves are imported.
How much energy does it take to make a solar panel?
Roughly 1,500 to 2,200 kWh per kilowatt-peak, most of it in polysilicon and wafer production. In UK sunlight, where a system generates around 800 to 1,000 kWh per kilowatt-peak per year, that energy is repaid in about 1.5 to 2.5 years. Global headline figures of four to eight months assume sunnier installation sites.
What is the difference between PERC, TOPCon and HJT panels?
They are three cell architectures. PERC is the older p-type design with rear passivation. TOPCon uses an n-type wafer with a tunnel oxide and doped polysilicon rear contact, giving higher efficiency and lower degradation, and it now makes up around 85% of global production. HJT layers amorphous silicon films around a crystalline wafer for the best temperature behaviour, at higher cost and lower volume.
Why do solar panels contain silver?
Silver is the most conductive metal available, and it is screen-printed as a fine grid to carry electrons off the cell without shading too much of it. A modern panel contains only a few grams, but silver is a meaningful share of cell cost and TOPCon uses more of it than PERC did. Manufacturers are working on silver-coated copper and busbar-free designs to reduce it.
Can solar panels be recycled in the UK?
Yes. Panels are covered by the WEEE regulations, which place take-back obligations on producers, and they should not be sent to landfill. Aluminium frames and glass are recovered easily. Separating the laminated layers to recover silicon and silver is harder and currently marginal on cost, though volumes and processes are both improving.
Updated 21 August 2026 with current cell technology production shares, 2026 module and cell pricing, and the state of UK supply chain assurance under the Great British Energy Act amendment.
The process is settled. The questions worth asking are elsewhere
Turning quartz into a working panel has been an industrial process for forty years, and the physics has not moved much. What has moved is the cell architecture, which flipped from p-type PERC to n-type TOPCon in under three years, and the scrutiny applied to who mined, refined and assembled the parts.
For a UK buyer that has two practical consequences. Check the cell type, because n-type panels degrade more slowly and the price gap has effectively closed. And ask which factory built them, because a model number alone no longer tells you enough about either quality or provenance.
The panel is around a tenth of what you pay for a solar installation. Choose it carefully, then spend the rest of your attention on the installer, because that is where the other nine tenths of the money and nearly all of the risk sits.
METHODOLOGY: Process descriptions compiled from manufacturer technical documentation, IEC standards and published industry production data. Technology production shares reflect analyst figures for early 2026. Cost breakdown built from early 2026 FOB cell and module pricing plus typical UK distributor and installer margins, and is indicative rather than a quotation. Energy payback modelled at 1,500-2,200 kWh embodied energy per kWp against 800-1,000 kWh/kWp annual UK yield.
DISCLAIMER: Prices move with polysilicon, silver and freight markets. Supply chain and certification positions change; verify current status with manufacturers and the relevant certification bodies before purchase. Claims regarding labour practices in specific regions are contested and are reported here as the positions of the parties involved.
UPDATED: 21 August 2026.
