7. Solar Power

Every module so far has ended at the same place: something spins a shaft, the shaft spins magnets, Faraday does the rest. Coal, gas, nuclear, hydro — all of them.

This one breaks the pattern completely. A solar panel has no moving parts at all. No boiler, no turbine, no shaft, no magnet. Light falls on a slab of silicon and electricity comes out of a wire. It is the strangest and, in the last fifteen years, the most consequential machine in this course.

Light knocking electrons loose

Module 2 described metals as a lattice of atoms in a sea of loose electrons — that is why they conduct. Insulators are the opposite: every electron is locked to its atom.

Silicon sits in between. It is a semiconductor: its electrons are bound, but only just. Give one a specific minimum kick of energy and it breaks free and can move.

That threshold is called the band gap, and for silicon it is about 1.1 electronvolts — which happens, wonderfully, to sit right in the middle of the energy carried by visible sunlight. Photons of light arriving at the silicon each carry a packet of energy; those carrying more than 1.1 eV can knock an electron loose.

But a loose electron by itself is useless — it would just wander and settle back. It needs a direction. That is what makes a solar cell more than a slab of silicon.

Manufacturers deliberately contaminate the silicon — “doping” — in two different ways, creating a layer with spare electrons (n-type) above a layer hungry for them (p-type). At the boundary, a permanent internal electric field forms: the p–n junction. It acts as a one-way street. Any electron freed near that junction gets swept firmly to one side.

Now the electrons have nowhere to go except out through the metal contact on the front, around whatever circuit you have connected, and back in at the rear. That journey is the current. This is the photovoltaic effect: sunlight in, direct current out, silently, with nothing moving but electrons.

From cell to grid

One cell produces about half a volt — useless alone. So:

• Cells are wired in series into a panel (typically 60–72 cells, a few hundred watts).
• Panels wire into strings, strings into arrays, building up to useful voltages.
• The array’s DC power reaches an inverter, which converts it to 50 Hz AC, synchronised precisely with the grid’s own rhythm.
• A transformer steps it up, and it joins the network like any other generator.

The inverter deserves more respect than it usually gets. It is not merely a DC-to-AC box; it must match the grid’s frequency and phase continuously, disconnect instantly if the grid fails (so it cannot electrocute a lineman working on a “dead” line), and increasingly help stabilise the grid. Module 10 will show why a grid full of inverters and empty of spinning turbines is a genuinely new engineering problem.

The number missing from the label

Now our hook question, and the single most misunderstood idea in renewable energy.

A “100 MW” rating is a power rating — Module 1’s tap. It means: in full sunshine, at the right angle, at reasonable temperature, this park delivers 100 MW.

But the sun sets. Clouds pass. Mornings and evenings are oblique. Averaged across a whole year, an Indian solar park delivers roughly 17–20% of its rated power. That ratio is the capacity factor.

Plant type	Typical Indian capacity factor
Solar PV	~17–20%
Wind	~20%
Coal	~60–70%
Nuclear	~80%

These are rough fleet averages around FY 2024-25, not promises: a well-sited new park beats an old one in a cloudier state, and every row moves a little from year to year. Divide Module 8’s wind figures — roughly 83 TWh from about 50 GW — and you land a little under the wind row, for the same reason the solar arithmetic below comes out low.

So a 100 MW solar park produces roughly what a 30 MW coal plant would. This is not a scandal or a hidden defect — it is simply what the label means, and every serious analyst works in energy (units generated), not capacity (megawatts installed).

It also resolves the puzzle from Module 3: coal is under half of India’s installed capacity but about 73% of its generation. Capacity counts what could run; generation counts what did.

India’s solar capacity reached about 105 GW and generated roughly 144 TWh — about 8% of national electricity — in FY 2024-25. Both numbers are true; only the second one keeps the lights on.

Divide one by the other and you get about 16% — below the range above. That is not a contradiction, it is a warning about the arithmetic: 105 GW is where the fleet finished the year, and India was building solar all the way through it, so most of those terawatt-hours came from a noticeably smaller fleet. Dividing a full year’s generation by year-end capacity always understates the capacity factor of anything growing this fast.

The price collapse

Something genuinely remarkable happened in the 2010s, and India is one of its clearest demonstrations.

Around 2010, solar electricity in India was contracted at well over ₹10 per unit — several times the cost of coal power, and only viable with policy support. In May 2017, an auction for 500 MW at Bhadla Solar Park in Rajasthan discovered a tariff of ₹2.44 per unit, fixed for 25 years with no escalation. Bhadla, in the Thar desert, is now one of the world’s largest solar parks at about 2,245 MW.

That is roughly a fivefold fall in about seven years — for the same product. It has few parallels in industrial history, and it happened not through one breakthrough but through the grinding compound effect of manufacturing scale, cheaper finance, better module efficiency, and auction competition. Module 12 explains the mechanism, called a learning curve, properly.

The consequence: solar is now frequently the cheapest source of new electricity in India by a clear margin. Which is exactly why the interesting problems have moved from cost to timing.

The duck

Here is the defining problem of the solar age. Move the slider and watch it appear.

total demand net load (non-solar plants) solar output

Solar on the grid: 100 units

Solar share of the day

32%

Belly (lowest net load)

12:00

Evening ramp needed

14%/h

With no solar, net load simply follows demand. Add solar and the middle of the day gets eaten away — the belly. But the evening peak does not move at all, because at 7 pm, when Indian demand is at its highest, solar output is exactly zero.

The result is the neck of the duck: every evening, the rest of the fleet must climb from its midday minimum back to full output. The whole climb — the gap between the two dots on the chart above, which sit at about noon and about 7 pm — takes several hours, but its punishing stretch is squeezed into about two or three hours. That squeeze is the figure grid operators actually watch, defined as the largest change in net load across any three-hour window. And the more solar you add, the steeper it gets.

This is why the interesting conversation about solar is no longer about panels. It is about:

• Flexibility. Which plants can climb that fast? Hydro (Module 5) and gas (Module 4) — which is why India’s idle gas fleet is such a frustrating asset to own.
• Storage. Move midday surplus to the evening. This is Module 11, and it is why battery costs matter as much as panel costs now.
• Curtailment. Push the slider far enough and solar output exceeds demand at midday. That surplus, with nowhere to go, is simply wasted.

Rooftops, and PM Surya Ghar

Solar comes in two very different forms. Utility-scale parks like Bhadla are cheapest per unit — bulk land, bulk procurement. Rooftop solar costs more per unit but competes against the retail tariff you pay, not the wholesale price, which changes the arithmetic entirely.

Net metering makes rooftop work: your meter runs backwards when you export surplus, and you are billed on the net. Effectively, the grid becomes your battery — which works beautifully while rooftop solar is a small share, and becomes a harder policy question as it grows, since those exports arrive precisely when the grid already has too much.

PM Surya Ghar: Muft Bijli Yojana, launched on 13 February 2024, is the largest domestic rooftop programme ever attempted. It targets one crore households — a crore is ten million, and a lakh is one hundred thousand; both are standard units in Indian counting — by March 2027, with a central subsidy of ₹30,000 per kW for the first 2 kW and ₹18,000 per kW beyond, capped at ₹78,000, aiming to give participating homes up to 300 free units a month.

Its real significance is structural. Every rooftop turns a consumer into a producer, pushing generation to the very edge of the grid, where the system was designed only to deliver. Module 10 examines what that does to a network built to flow one way.

The other solar

One footnote that restores the course’s central image. Concentrated solar power takes a completely different route: fields of mirrors focus sunlight onto a receiver, heating a fluid — often molten salt — to hundreds of degrees. That heat boils water, the steam spins a turbine, and the turbine spins a magnet.

CSP is a thermal power station whose furnace is the sun. It is more expensive than PV and has largely lost the cost race, but it retains one genuine advantage: heat is far easier and cheaper to store than electricity. A CSP plant with a tank of hot salt can keep generating for hours after sunset — attacking the duck curve at its neck. India has experimented with CSP but has overwhelmingly built PV, following the economics.

🔥Why hot panels perform worse, and what's coming after siliconAdvanced · optional

The heat paradox. Solar panels produce less power as they get hotter — typically around 0.3–0.45% less for every degree Celsius above their 25 °C test condition. On a Rajasthan afternoon, panel temperature can reach 60–70 °C, costing perhaps 15% of output.

Why? Rising temperature raises the energy of electrons already in the material, which narrows the effective band gap and — more importantly — sharply increases recombination, where freed electrons fall back before reaching the contacts. Current rises very slightly; voltage falls considerably more. Power, being their product, falls.

The counterintuitive consequence: a cold bright day beats a hot bright one. Panels in Ladakh’s cold, high-altitude sunshine can out-perform the same panels in a hotter desert.

The materials race.

• Monocrystalline silicon — one continuous crystal, ~20–23% efficient in commercial modules. Now dominant as manufacturing costs fell.

• Polycrystalline — cheaper, many small crystals, slightly less efficient. Largely displaced by mono.

• Thin film (CdTe, CIGS) — a microns-thin layer on glass. Less efficient, cheap, performs better in heat and diffuse light.

• Perovskites — the genuine excitement. A class of crystals that can be printed from solution at low temperature, with laboratory efficiencies that climbed faster than any previous PV technology. Their trick is a tunable band gap, which allows tandem cells: a perovskite layer capturing high-energy blue photons stacked on silicon capturing the red ones. Tandems have already passed 30% in the laboratory, beating silicon’s practical ceiling.

  The unsolved problem is durability. Perovskites degrade with moisture, heat and UV — and a panel must survive 25 years on a roof, not six months in a laboratory. Encapsulation is improving steadily. If it is solved, module efficiency jumps by half again.

🔥🔥Photons, band gaps and the Shockley–Queisser limitDeep · optional

Photon energy. Light arrives in quanta of energy E = hf = hc/λ, where h is Planck’s constant and λ the wavelength. In convenient units:

E (eV) ≈ 1240 / λ (nm)

Silicon’s band gap of 1.12 eV therefore corresponds to λ ≈ 1,110 nm, in the near infrared. Any photon with a longer wavelength passes straight through and does nothing. That alone discards about a fifth of the sun’s energy.

The double loss. Now consider a blue photon at 450 nm, carrying 2.76 eV. It frees an electron — but only 1.12 eV can be captured as electrical potential. The surplus 1.64 eV becomes heat within picoseconds, as the electron “thermalises” down to the band edge. So:

• Photons below the gap: transmitted, entirely lost.
• Photons above the gap: partly wasted as heat, and the bigger the excess the worse.

These two losses pull in opposite directions. A small band gap captures more photons but wastes more of each; a large gap captures fewer but uses each better. There is an optimum, and around 1.1–1.4 eV is close to it — which is why silicon, chosen originally for manufacturing reasons, turned out to be near-ideal anyway.

The Shockley–Queisser limit. In 1961 William Shockley and Hans Queisser worked out the maximum efficiency of a single-junction cell under standard sunlight, accounting for these losses plus unavoidable radiative recombination. The answer is about 33%, peaking near a 1.34 eV gap. Silicon’s practical laboratory ceiling is around 29%, and the best commercial modules are in the low twenties.

This is a real physical bound, in the same family as Carnot’s — but with an escape route Carnot does not offer. The limit applies per junction. Stack cells with different band gaps so each handles the slice of the spectrum it converts best, and the ceiling rises: about 42% for two junctions, higher still for more. That is precisely the logic behind perovskite-on-silicon tandems, and why they are the most closely watched technology in solar.

What comes next

Solar broke the spinning-magnet rule, crashed in price faster than anyone forecast, and handed grid operators a duck-shaped problem they are still solving. Module 8 returns to rotation — but with the most counterintuitive physics in Part B, since a wind turbine’s blades do not get pushed by the wind at all. They fly.