Solar Power

In roughly fifteen years, solar photovoltaics (PV) went from a niche product to the fastest-growing source of electricity in history. World solar generation rose from about 250 TWh in 2015 to roughly 2,800 TWh in 2025, close to 9% of global electricity, and annual installations now exceed those of every other generating technology combined. China alone produces more than 40% of the world’s solar electricity, followed by the United States and India. Below: how the technology works, how the main cell types differ, and what the numbers say about cost, land, materials and emissions.

How solar PV works

A solar cell is a semiconductor diode, usually made of crystalline silicon. When a photon with enough energy strikes the silicon, it frees an electron and leaves behind a positively charged “hole”. This is the photovoltaic effect. An internal electric field at the p-n junction pushes electrons and holes in opposite directions, and metal contacts on the front and back collect them as direct current (DC). A single silicon cell produces about 0.6-0.7 V, so a module wires 60-144 cells (or half-cells) in series to reach a usable voltage. Bypass diodes limit the losses from partial shading.

Silicon’s band gap of 1.12 eV caps the theoretical efficiency of a single-junction cell at roughly 29-33% (the Shockley-Queisser limit, slightly lower for silicon once Auger recombination is included). Photons below the band gap pass straight through, and whatever extra energy the higher-energy photons carry ends up as heat. Real modules also lose energy to reflection, to resistance in the contacts and, above all, to temperature: crystalline silicon loses about 0.25-0.40% of its output for every degree Celsius above 25 °C, so a module on a hot roof falls short of its nameplate rating.

Modules are rated in watts-peak under Standard Test Conditions (1,000 W/m² irradiance, 25 °C cell temperature). Inverters turn the DC into grid-compatible AC. They also run MPPT algorithms, which keep adjusting the operating voltage to pull maximum power from the array as light and temperature change.

Main types of PV technology

Crystalline silicon cells

Crystalline silicon accounts for more than 95% of global module production. Within silicon, the industry has changed cell architecture several times in a decade:

• Polycrystalline (multi) silicon dominated until about 2017. It was cheaper to cast but less efficient (typically 16-18% modules) and has now almost vanished from new production.
• Monocrystalline PERC (Passivated Emitter and Rear Cell) adds a passivating dielectric layer to the rear of the cell, reflecting unabsorbed light back and reducing recombination. PERC carried the market from 2018 to 2023, with modules at roughly 20-22%.
• TOPCon (Tunnel Oxide Passivated Contact) uses an ultra-thin oxide and doped polysilicon layer to passivate the rear contact. Factories can make it on upgraded PERC lines, so from 2024 it became the main technology in new plants, with modules around 22-23.5%.
• Heterojunction (HJT) sandwiches a crystalline wafer between thin amorphous-silicon layers. HJT has the best temperature coefficient (about -0.24 to -0.26 %/°C), high bifaciality and module efficiencies of 22-24%. The downside: it needs different equipment and more silver, or copper-plated contacts.
• Back-contact (IBC) cells move all metal contacts to the rear, removing front-side shading and pushing commercial modules toward 23-25%.

Bifacial modules

A bifacial module uses a transparent back sheet or glass so the rear of the cells captures light reflected from the ground. It typically yields 5-15% more energy than a monofacial module, rising to 20% or more over bright surfaces such as sand, snow or white gravel (albedo of 0.3-0.8). For large ground-mounted plants, bifacial glass-glass modules on single-axis trackers are now the default.

Thin film

Thin-film modules deposit a few micrometres of light-absorbing material on glass. Cadmium telluride (CdTe), produced at scale mainly by First Solar, reaches commercial efficiencies of about 19-20% and has a lower carbon footprint than most silicon modules. CIGS (copper indium gallium selenide) reaches 15-18% in production; amorphous silicon (6-9%) has mostly vanished from power markets. Perovskite cells, especially perovskite-on-silicon tandems with laboratory efficiencies above 33%, are the next step everyone is watching. Nobody yet knows how well they hold up outdoors over decades.

Rooftop, utility-scale and trackers

Distributed rooftop systems (a few kW on homes up to several MW on warehouses) use no extra land, avoid transmission losses and pay back mostly through self-consumption at retail prices. Utility-scale plants of 50 MW to several GW get much lower costs per kWh from scale, the best orientation and solar trackers. Single-axis trackers, which rotate rows east to west through the day, raise annual yield by roughly 15-25% compared with fixed tilt at sunny, low-latitude sites and now equip the majority of new utility plants in the Americas, Middle East and Australia. Floating PV on reservoirs and agrivoltaics over crops are smaller but growing niches.

Inverters

Residential systems use string inverters, microinverters on each module, or string inverters with module-level power optimisers. Utility plants use either large central inverters (several MW each) or many distributed string inverters. Designers routinely install more DC module capacity than inverter AC capacity (a DC/AC ratio of 1.2-1.4). They accept some clipping at midday in return for more output in the mornings, evenings and winter. Inverters now also help run the grid: grid-forming inverters can provide synthetic inertia and voltage support previously supplied by spinning generators.

Efficiency, yield and capacity factor

Module efficiency decides how much area a system needs. How much energy it produces depends on the local sunshine. The standard estimate of annual output is:

Here H_poa is the annual irradiation on the plane of the array and PR is the performance ratio, typically 0.75-0.85 for a well-designed modern system. It covers temperature, inverter, wiring, soiling and mismatch losses. A 10 kWp system in Madrid receiving 2,000 kWh/m² on its tilted plane with a PR of 0.80 produces about 16,000 kWh per year: a specific yield of 1,600 kWh/kWp and a DC capacity factor of 18%. The solar panel calculator runs this hour by hour with real weather, and the solar potential map shows how the resource varies worldwide.

Solar’s capacity factor is low next to thermal plants because the sun sets every day and is weaker in winter. That variability is the main problem for grid operators. It causes the midday price dip known as the duck curve, and it is why solar increasingly comes paired with battery storage. The capacity factor calculator converts between capacity, hours and output.

Costs

No energy technology has fallen in cost as fast as PV. Module prices have dropped by roughly 20-25% for every doubling of cumulative production (the learning rate known as Swanson’s law), and in 2024 spot prices for standard modules fell below USD 0.10-0.12/W amid manufacturing overcapacity in China.

• IRENA (2024): the global weighted-average LCOE of newly commissioned utility-scale PV was about USD 0.044/kWh in 2023, down roughly 90% from USD 0.46/kWh in 2010. The global weighted-average total installed cost was around USD 750-800/kW, with India and China well below that and Japan and some European markets well above.
• Lazard LCOE+ (2024): unsubsidised US utility-scale PV at roughly USD 29-92/MWh; community and commercial systems in the mid-range; residential rooftop at about USD 177-282/MWh, reflecting much higher installed costs per watt.
• NREL ATB projects continued declines in utility PV capital costs through 2035, with a wide spread between conservative and advanced scenarios.

Two caveats. Solar LCOE is very sensitive to the discount rate: nearly all the cost is paid upfront, so financing alone can double the LCOE between a low-risk OECD market and a high-risk emerging one. And solar power loses value as its share grows, because every plant produces in the same hours. In markets with high penetration, midday prices are frequently negative and curtailment rises, so “solar plus storage” costs are increasingly the relevant benchmark. Use the LCOE calculator to test how CAPEX, yield and WACC interact, and the technology comparison to set solar against other sources.

Land, water and materials

Land. Utility-scale PV needs roughly 1-2.5 hectares per MW of capacity including spacing, access roads and inverters, depending on latitude, tracker use and module efficiency. Per unit of energy that is more land than nuclear or gas but less than bioenergy, and the ground isn’t sealed: sheep can graze, pollinator habitat can grow, and crops can sit under raised arrays (agrivoltaics). Rooftops, car parks and reservoirs need no extra land at all.

Water. In operation, PV uses very little water, mainly for occasional module cleaning: typically tens of litres per MWh, far below what thermal plants need for cooling. The water-hungry stage is making polysilicon and wafers.

Materials. A crystalline module is mostly glass (about 70% by weight), aluminium frame, polymer encapsulant, copper and silicon. The critical inputs are high-purity polysilicon and silver for the cell contacts. PV already takes a large and growing share of global silver demand, which is pushing manufacturers toward copper plating. CdTe uses tellurium, a by-product of copper refining with limited supply. The IEA’s 2022 supply-chain report found that China held more than 80% of global capacity at every manufacturing stage, and over 95% for wafers. That concentration has shaped industrial policy in the US, India and the EU.

Lifecycle CO2 emissions

PV emits nothing while it runs. Its lifecycle emissions come from mining, refining silicon (an electricity-intensive process), manufacturing, transport, installation and end-of-life treatment.

• IPCC AR5 WGIII Annex III (2014): utility-scale PV median 48 gCO2e/kWh (range 18-180); rooftop PV median 41 g (range 26-60). These figures reflect pre-2012 manufacturing and are now considered conservative.
• UNECE (2021): 8-83 gCO2e/kWh across PV technologies and regions, with thin-film CdTe and modules manufactured on low-carbon grids at the bottom, and polysilicon modules made with coal-heavy power and installed in low-irradiance locations at the top.

For comparison, the same assessments put coal at roughly 750-1,100 g and gas combined-cycle at roughly 400-500 g. Most of the emissions come from the electricity used in manufacturing, so a module’s footprint depends heavily on where it was made, and it shrinks as those grids decarbonise. Energy payback times are now about a year or less in sunny regions.

Safety and environmental impacts

PV has one of the lowest death rates per unit of energy of any source. The main occupational risk is falling off a roof during installation. Fire risk is low but real, usually from faulty DC connectors or installation errors, and many jurisdictions now require rapid-shutdown electronics so firefighters can de-energise rooftop arrays. Cadmium in CdTe modules is chemically bound and encapsulated in glass, with manufacturer take-back programmes.

The biggest long-term problem is what to do with old panels. IRENA and IEA PVPS projected in 2016 that cumulative PV waste could reach roughly 60-78 million tonnes by 2050. Glass, aluminium and copper are easy to recycle. Recovering high-purity silicon and silver at a profit is not there yet. The EU’s WEEE Directive already makes producers responsible for collection. Large ground-mounted plants can also fragment habitat if poorly sited, though well-managed sites often increase biodiversity relative to intensive farmland.

Outlook

Solar is now the largest source of new generating capacity worldwide, with annual additions of roughly 600 GW in 2024 and cumulative capacity passing 2 TW. The IEA’s scenarios consistently show solar becoming the largest source of electricity by installed capacity before 2030 and, in net-zero pathways, the largest by generation in the 2030s. The technology is heading toward n-type cells (TOPCon, HJT, back-contact), bifacial trackers as the utility standard, and perovskite-silicon tandems that could lift module efficiencies toward 30%.

The bottleneck is no longer cost but integration: grid connection queues, transmission, midday price cannibalisation and the need for flexibility. So solar’s growth now depends on battery storage, demand flexibility, and green hydrogen for absorbing surplus. For dispatchable solar from thermal storage, see concentrated solar power, and to size a home system with a battery try the solar battery calculator.