How Solar Cells Turn Light Into Electricity
A photon hits silicon. An electron jumps. Current flows. The physics of photovoltaics is quantum mechanics meets semiconductor engineering — and it's the fastest-growing energy technology on Earth.
Table of Contents
Sand Into Electricity
Here’s the short version of how a solar cell works: a photon from the Sun hits a piece of silicon. The photon’s energy kicks an electron out of its bond, creating a free electron and a free “hole” (the vacancy left behind). An electric field built into the silicon pushes the electron one way and the hole the other. Connect a wire between the two sides, and current flows. That’s it. Photon in, electron out.
The long version involves quantum mechanics, semiconductor band theory, p-n junction physics, and about 70 years of engineering refinement. But the core idea is simple: light carries energy in discrete packets (photons), and silicon can convert that energy into electrical current with no moving parts, no combustion, no noise, and no emissions.
The fact that we can do this at all is because of the photoelectric effect — the quantum mechanical process that Einstein explained in 1905 (and won his only Nobel Prize for, in 1921). Photons are not continuous waves of energy. They’re particles, each carrying a specific energy E = hf, where h is Planck’s constant and f is the frequency. If that energy exceeds the threshold needed to liberate an electron from a material, the electron escapes. Below the threshold, nothing happens — no matter how intense the light.
The Band Gap: Silicon’s Sweet Spot
In a semiconductor like silicon, electrons exist in energy bands. The lower band — the valence band — is full of electrons locked in chemical bonds. The upper band — the conduction band — is empty (at zero temperature) and represents states where electrons are free to move and carry current. Between them is the band gap — an energy range where no electron states exist.
For silicon, the band gap is 1.12 electron volts (eV). A photon with at least 1.12 eV of energy — corresponding to a wavelength of about 1,100 nm (near infrared) — can excite an electron from the valence band to the conduction band, creating a free electron-hole pair.
The solar spectrum peaks at about 500 nm (2.5 eV) and contains photons from about 300 nm (4.1 eV, ultraviolet) to 2,500 nm (0.5 eV, infrared). Silicon absorbs everything below 1,100 nm — which includes visible light and some UV — but is transparent to longer infrared wavelengths. About 20% of the solar energy reaching Earth’s surface is carried by photons that silicon simply can’t absorb. This is one of the unavoidable losses.
The other unavoidable loss is thermalisation. A photon with 3 eV of energy hitting silicon generates an electron with 3 − 1.12 = 1.88 eV of excess kinetic energy. This excess energy is lost as heat within picoseconds — the “hot” electron quickly relaxes to the band edge through collisions with the crystal lattice. Only 1.12 eV of the original 3 eV is captured as useful electrical energy. Higher-energy photons are partially wasted.
These two losses — transparency below the band gap and thermalisation above it — are the reason a single-junction solar cell can’t convert more than about 33% of sunlight to electricity. This is the Shockley-Queisser limit, calculated in 1961, and it’s one of the most important numbers in solar energy.
The p-n Junction: Building a One-Way Gate
Absorbing photons and creating free electrons is only half the battle. You also need to collect those electrons before they recombine with holes (losing their energy). This is the job of the p-n junction — the heart of every solar cell.
Take a silicon wafer and dope one side with phosphorus (adding extra electrons — n-type) and the other side with boron (creating electron vacancies, or holes — p-type). At the boundary, electrons from the n-side diffuse into the p-side, and holes from the p-side diffuse into the n-side. This charge movement creates a thin region — the depletion zone — with a built-in electric field pointing from n to p.
This built-in field is the key. When a photon creates an electron-hole pair near the junction, the field sweeps the electron toward the n-side and the hole toward the p-side. The charges are separated. Connect a wire from the n-side to the p-side (through a load — a light bulb, a battery, a power grid) and the electrons flow through the external circuit, doing useful work before recombining with holes on the other side.
No battery. No generator. No spinning turbine. Just photons creating electron-hole pairs and a built-in electric field pushing them apart. The voltage produced by a silicon solar cell is about 0.5–0.7 V per cell — determined by the band gap and the built-in potential of the junction. To get useful voltages, cells are wired in series (a typical residential panel has 60–72 cells producing 30–40 V).
Real-World Efficiency: Where the Losses Go
The Shockley-Queisser limit is 33.7% for silicon. The best laboratory silicon cells reach about 26.8% (the current world record, by LONGi in 2024). Commercial panels are typically 20–24%. Where does the rest go?
Beyond the fundamental thermalisation and transparency losses, there are practical losses at every stage.
Reflection: Silicon is shiny. Untreated, it reflects about 35% of incident light. Anti-reflection coatings (typically silicon nitride, giving panels their characteristic blue-black colour) reduce this to 1–3%. Textured surfaces further reduce reflection by causing light to bounce multiple times.
Recombination: Not every photogenerated electron-hole pair reaches the junction. Some recombine within the silicon bulk (if the crystal has defects), at the surfaces (where dangling bonds trap carriers), or in the metal contacts. High-quality monocrystalline silicon and surface passivation techniques minimise these losses.
Resistance: Current flowing through the silicon, metal contacts, and interconnects encounters electrical resistance, which converts some power to waste heat (I²R losses). The fingers of metal on the cell’s front surface are a compromise — wider fingers reduce resistance but shade more active area.
Temperature: Solar cells lose about 0.3–0.5% of their power for every degree Celsius above 25 °C. On a hot roof at 60 °C, a panel produces about 10–15% less than its rated power. This is because higher temperature increases the thermal energy of charge carriers, increasing recombination and slightly decreasing the voltage. Desert installations get lots of sun but also lots of heat — the two effects partially cancel.
Multi-Junction Cells: Breaking the Single-Junction Limit
The 33% barrier applies to a single band gap. But what if you stack multiple semiconductors with different band gaps?
A triple-junction cell might use gallium indium phosphide (GaInP, bandgap 1.86 eV) on top, gallium arsenide (GaAs, 1.42 eV) in the middle, and germanium (Ge, 0.67 eV) on the bottom. High-energy photons are absorbed by the top layer, medium-energy photons pass through to the middle layer, and low-energy photons reach the bottom. Each layer converts its slice of the spectrum near the optimal efficiency for its band gap.
Laboratory multi-junction cells have exceeded 47% efficiency under concentrated sunlight. They’re expensive — used mainly in space (where power per kilogram matters more than cost per watt) and in concentrator systems (where lenses focus sunlight onto small, expensive cells).
For terrestrial applications, perovskite-silicon tandem cells are the current frontier. A perovskite layer (band gap ~1.7 eV) is deposited on top of a conventional silicon cell (1.12 eV). The perovskite absorbs blue and green light efficiently; the silicon absorbs the red and infrared that passes through. Lab tandems have hit 33.9% — already above the single-junction limit — and they can potentially be manufactured at costs comparable to silicon alone. If the durability challenges are solved (perovskites degrade in moisture), this technology could redefine residential solar within a decade.
The Price Curve
I’d be leaving out the most important part of the solar story if I didn’t mention cost. In 1976, crystalline silicon solar cells cost about $106 per watt. In 2024, the price was about $0.20 per watt — a factor of 500 decrease. This is Swanson’s Law: for every doubling of cumulative manufactured volume, the price drops by about 20%.
The physics hasn’t changed. The band gap is the same. The Shockley-Queisser limit is the same. What changed is manufacturing scale, process optimisation, and supply chain maturity — the same learning-curve effects that made computing cheap. Silicon is the second most abundant element in Earth’s crust. The raw material was never the bottleneck.
As of 2025, solar is the cheapest source of electricity in most of the world, measured by levelised cost of energy (LCOE). The remaining challenges are not about the physics of the cell — they’re about storage (what happens when the Sun goes down), grid integration, and the energy transition infrastructure needed to deploy solar at the terawatt scale.
Quantum Mechanics on Your Roof
There’s a certain poetry to solar energy that I think gets lost in discussions about capacity factors and grid parity. Every solar cell on every roof is a device that converts quantum mechanical events — individual photons liberating individual electrons across a band gap determined by the quantum mechanics of silicon’s crystal structure — into macroscopic electrical power.
Einstein explained the photoelectric effect. Shockley and colleagues invented the p-n junction. Thousands of engineers optimised the manufacturing. And now you can buy a device, made from refined sand, that sits on your roof and silently converts starlight into electricity for 30 years. The physics is exactly what Einstein described in 1905 — photons carrying discrete energy packets, interacting with matter one quantum at a time. Just scaled up by a factor of 10²² and priced at $0.20 per watt.
Frequently Asked Questions
What limits the efficiency of solar cells?
The fundamental limit for a single-junction silicon solar cell is about 33% — the Shockley-Queisser limit. Two main losses cause this. First, photons with energy below the bandgap (1.12 eV for silicon, corresponding to infrared wavelengths longer than 1,100 nm) pass right through without being absorbed — they're wasted. Second, photons with energy above the bandgap generate electrons with excess kinetic energy, which is quickly lost as heat (thermalisation) rather than being converted to electricity. Only photon energy exactly equal to the bandgap is fully converted. For the solar spectrum, silicon's bandgap is near-optimal for single-junction efficiency, but roughly two-thirds of incident solar energy is still lost. Multi-junction cells using stacked semiconductors with different bandgaps can capture more of the spectrum, reaching over 47% in laboratory settings.
How long do solar panels last?
Modern crystalline silicon solar panels are warranted to produce at least 80% of their initial power after 25 years, and many perform well beyond this. Degradation is typically 0.3–0.5% per year, caused by UV exposure, thermal cycling, and moisture ingress that slowly degrade the encapsulant, anti-reflection coating, and metal contacts. The silicon cells themselves are extremely durable — silicon doesn't rust, corrode, or wear out mechanically. The weakest links are the interconnects (thin metal ribbons connecting cells), the encapsulant (EVA plastic that can yellow), and the junction box. Panels installed in the 1980s are still producing useful power. The practical lifespan is likely 30–40 years for well-made panels.
Do solar cells work on cloudy days?
Yes, but less effectively. On an overcast day, solar irradiance drops to roughly 10–25% of full sunshine, and solar cell output drops proportionally. The cells still work because they respond to diffuse light (photons scattered by clouds), not just direct sunlight. Even on a very cloudy day, the irradiance is typically 50–150 W/m² compared to about 1,000 W/m² in full sun. Modern solar installations in famously cloudy countries like Germany, the UK, and the Netherlands produce significant annual energy because they benefit from long summer days, even if individual cloudy days produce less. The annual energy yield depends more on total irradiance over the year than on the number of sunny days.
What's the difference between monocrystalline and polycrystalline panels?
Monocrystalline panels are made from a single silicon crystal, grown slowly using the Czochralski process. The uniform crystal structure means fewer grain boundaries (which scatter charge carriers and reduce efficiency), giving higher efficiencies — typically 20–24%. They're more expensive to produce. Polycrystalline panels are made from silicon melted and cast into blocks, then sliced. The resulting wafer contains many small crystal grains with boundaries between them. These boundaries reduce efficiency to about 16–20%, but the manufacturing is cheaper and faster. In practice, the price difference has narrowed so much that monocrystalline cells now dominate the market — the modest efficiency advantage justifies the small cost premium, especially since higher efficiency means fewer panels needed for the same power output.