Photovoltaics: Turning Light Into Current, and the Ceiling on Doing So
Photovoltaics is the most-deployed generating technology of the last decade and the one whose price collapse caught nearly every forecaster out. The physics is a century old and has barely changed. What changed was manufacturing, and understanding why the cell has a hard ceiling explains both why progress looked slow for decades and why the cost fell instead.
What Happens Inside the Cell
Light arrives as photons, and a photon with enough energy can knock an electron out of its bond in a semiconductor, leaving behind a positively charged vacancy called a hole. That much is the photoelectric effect, and on its own it produces nothing useful: the electron and hole drift about and recombine within microseconds, releasing the energy again as heat.
The photovoltaic cell adds the piece that makes it an energy source. Two layers of silicon are doped differently - one with a slight surplus of electrons, one with a slight deficit - and where they meet, charge redistributes until a permanent internal electric field forms across the junction. This is the p-n junction, and the field is typically a fraction of a volt across a region under a micrometre thick.
When a photon frees a pair inside or near that region, the field pushes the electron one way and the hole the other before they can recombine. Separated charges at the two contacts constitute a voltage, and connecting a circuit lets the electrons return the long way round, doing work as they go.
This is direct conversion in the strict sense, with no heat stage. The thermal efficiency limits that cap a coal or nuclear plant at 33 to 45 percent do not apply. A photovoltaic cell has its own ceiling, and it comes from somewhere else entirely.
The Shockley-Queisser Limit
In 1961 William Shockley and Hans-Joachim Queisser asked what the best possible single-junction cell could do, given only the spectrum of sunlight and the laws of thermodynamics. The answer was about 33 percent, and no single-junction cell of any material has ever exceeded it.
The reasoning rests on the band gap - the minimum photon energy a given semiconductor needs to free an electron. Silicon's is about 1.1 electronvolts. Every photon with less energy passes straight through the cell and contributes nothing at all; in sunlight that is roughly 20 percent of the incoming energy, most of it infrared.
Every photon with more energy frees exactly one electron, no more, and the surplus energy is immediately lost as heat within the crystal. A blue photon carries nearly three electronvolts; the cell keeps 1.1 of them and discards the rest. That thermalisation loss accounts for another 30 percent or so.
The two losses pull in opposite directions, which is what makes the limit a limit. Lower the band gap to catch more infrared and you increase thermalisation waste. Raise it to reduce waste and you lose more photons entirely. Somewhere near 1.3 electronvolts the two balance, and silicon at 1.1 happens to sit close enough that its abundance and maturity outweigh the difference.
Multi-junction cells escape by stacking layers with different band gaps, each taking a slice of the spectrum. They exceed 47 percent under concentrated light and cost far too much for anything but satellites and concentrator systems. Perovskite layers on top of silicon are the current attempt to do the same thing cheaply.
Why the Price Collapsed
Cell efficiency has improved slowly and steadily: commercial silicon modules went from around 15 percent in 2010 to 22 to 24 percent today. That is a meaningful gain and it is not what changed the industry.
What changed was cost per watt, which fell by more than 99 percent between 1975 and the present, and by roughly 90 percent in the 2010s alone. The mechanism was manufacturing scale and the learning curve that comes with it: each doubling of cumulative production has historically brought a price reduction of around 20 percent, a relationship that has held for four decades across three continents of production.
Practically, that meant larger wafers, thinner slices, less silver in the contacts, diamond-wire sawing that wastes less material, and factories running at gigawatt rather than megawatt scale. None of it is glamorous physics. All of it is why photovoltaics is now the cheapest source of new electricity in most of the world.
The consequence for the technology's future is that efficiency gains matter less than they used to. When modules were expensive, a percentage point of efficiency justified considerable cost. Now that modules are cheap, the balance of system - mounting, wiring, inverters, labour, permitting - dominates the installed price, and those costs fall with efficiency only indirectly.
What Photovoltaics Does and Does Not Solve
A solar cell produces power the moment light strikes it and stops the moment light stops. There is no inertia, no fuel store, no ramp. Capacity factors accordingly run from about 10 percent in northern Europe to 25 percent in desert regions - meaning a one-megawatt array produces an average of 100 to 250 kilowatts across a year.
This is a property of the resource, not a defect of the equipment, and the same distinction applies to wind. The engineering response is either to store the surplus, which is why energy storage and photovoltaics have grown together, or to pair it with generation that runs regardless of weather.
The daily mismatch is the more tractable half. Solar peaks at midday and demand peaks in the evening, a gap of a few hours that batteries handle well and that has made four-hour storage the standard grid product. The seasonal mismatch is harder: in northern latitudes December output can be a tenth of June output, and no battery fleet bridges a season.
That residual is why a grid does not consist of photovoltaics alone, and why research into sources that generate continuously - from geothermal and nuclear fission to ambient-energy approaches still in the laboratory - addresses a different part of the problem rather than competing for the same one.
Frequently asked questions
What is the maximum efficiency of a solar panel?
For a single-junction silicon cell, about 33 percent, set by the Shockley-Queisser limit. Commercial modules reach 22 to 24 percent and laboratory silicon cells about 27. Multi-junction cells stack different materials to exceed 47 percent under concentrated light, at a cost that restricts them to satellites and specialised systems.
Why can't a solar cell use all the sunlight?
Two unavoidable losses. Photons with less energy than the band gap pass through without freeing an electron - about 20 percent of sunlight for silicon. Photons with more energy free one electron each and waste the surplus as heat - roughly another 30 percent. Changing the band gap reduces one loss and increases the other.
Is a solar cell the same as the photoelectric effect?
The photoelectric effect is the underlying step: a photon frees an electron. A photovoltaic cell adds a p-n junction whose internal electric field separates the freed electron from its hole before they recombine, which is what turns the effect into a usable voltage.
Why did solar get so cheap so fast?
Manufacturing scale rather than physics. Module cost per watt has fallen roughly 20 percent with each doubling of cumulative production, a relationship that has held for four decades. Efficiency improved too, from about 15 percent to 23 percent since 2010, but the price collapse came from factories, wafer handling and materials use.
How long does a solar panel take to repay the energy used to make it?
Between about six months and two years depending on location and manufacturing energy mix, against a service life of 25 to 30 years with output typically above 80 percent of original at year 25. Silicon purification is the most energy-intensive step, which is why the payback is shorter where the factory grid is cleaner.
