The Carnot Cycle

What the Carnot cycle does

The Carnot cycle answers one deceptively simple question: given a hot source at temperature T_hot and a cold sink at temperature T_cold, what is the most work you can possibly squeeze out per unit of heat drawn from the hot side? Sadi Carnot proved in 1824 that there is a hard ceiling, and that a specific idealized engine reaches it. Every power plant, jet engine, and car engine ever built is quietly benchmarked against this number.

The trick is that the engine must be reversible: every step is run so slowly and so gently that it could be reversed by an infinitesimal change, and no entropy is generated anywhere in the universe. To keep heat flow reversible, heat may only cross a zero temperature difference. That single constraint forces the shape of the cycle — you can only exchange heat with a reservoir while sitting at exactly its temperature (an isotherm), and you can only move between the two temperatures without leaking heat across a gap (an adiabat).

    η_Carnot = W_net / Q_hot = 1 − Q_cold/Q_hot = 1 − T_cold/T_hot     (T in kelvin)

The final equality — that the ratio of heats equals the ratio of temperatures — is the deep result. It only holds for a reversible cycle, and it is what elevates absolute temperature from a thermometer reading to a fundamental quantity in the second law.

The four strokes, step by step

Picture an ideal gas trapped in a cylinder under a frictionless piston. One cycle is four consecutive processes; on a pressure-volume (P-V) diagram it traces a lens-shaped loop, and on a temperature-entropy (T-S) diagram it becomes a perfect rectangle.

1. Isothermal expansion at T_hot (A→B). Put the cylinder in contact with the hot reservoir. The gas absorbs heat Q_hot and expands, pushing the piston out. Because it stays at T_hot, its internal energy is unchanged (ΔU = 0 for an ideal gas at constant T), so all the absorbed heat becomes work: Q_hot = W_AB = nRT_hot ln(V_B/V_A). Entropy of the gas rises by +Q_hot/T_hot.
2. Adiabatic expansion (B→C). Insulate the cylinder — no heat in or out. The gas keeps expanding and does work, but now that work comes out of its internal energy, so it cools. It slides from T_hot down to T_cold. Because Q = 0 and the process is reversible, entropy stays constant (isentropic): TV^γ−1 = const.
3. Isothermal compression at T_cold (C→D). Put the cylinder on the cold reservoir. Compress the gas; it rejects heat Q_cold to the cold sink while staying at T_cold. Work is done on the gas: Q_cold = |W_CD| = nRT_cold ln(V_C/V_D). Entropy of the gas falls by −Q_cold/T_cold.
4. Adiabatic compression (D→A). Insulate again. Keep compressing; the work done on the gas raises its internal energy and it heats from T_cold back to T_hot, returning to the exact starting state. Isentropic again. The loop is closed.

    A ──isothermal expansion, +Q_hot──▶ B
    │                                   │
    adiabatic compression       adiabatic expansion
    (T_c → T_h)                  (T_h → T_c)
    │                                   │
    D ◀──isothermal compression, −Q_cold── C

The net work per cycle is the area enclosed by the loop on the P-V diagram, and equally the area of the rectangle T_hot−T_cold times ΔS on the T-S diagram: W_net = Q_hot − Q_cold = (T_hot − T_cold)·ΔS.

Why this is the maximum (Carnot's theorem)

Carnot's theorem has two parts, both provable from the second law alone: (1) no engine operating between two reservoirs can be more efficient than a reversible one, and (2) all reversible engines between the same two reservoirs have identical efficiency, regardless of the working substance — gas, steam, or a rubber band.

The proof is a beautiful reductio. Suppose a "super-engine" X were more efficient than a Carnot engine C running between the same reservoirs. Run C backwards as a refrigerator, driven by the work output of X. Because X produces more work per unit Q_hot, the pair would end up transferring net heat from the cold reservoir to the hot one with no external work input — a perpetual-motion machine of the second kind, forbidden by the Clausius statement of the second law. Contradiction. Therefore no engine can beat Carnot.

Part (2) — substance independence — is what lets us define the thermodynamic (Kelvin) temperature scale: T_cold/T_hot ≡ Q_cold/Q_hot for a reversible engine. Absolute temperature is defined by the heat ratios of a Carnot engine, which is why the efficiency formula contains temperatures at all.

Worked example: a steam power plant

A supercritical steam plant boils water with heat drawn at T_hot = 810 K (about 537 °C) and condenses its exhaust against a river at T_cold = 300 K (27 °C). What is the best efficiency thermodynamics allows?

    η_Carnot = 1 − T_cold/T_hot = 1 − 300/810 = 1 − 0.370 = 0.630  →  63.0%

    Heat in:   Q_hot  = 1000 MJ  (per unit time, say)
    Max work:  W_net  = 0.630 × 1000 MJ = 630 MJ
    Heat dumped: Q_cold = Q_hot − W_net = 370 MJ to the river

• The ceiling. 63% is the best any engine between 810 K and 300 K can do. A real supercritical plant achieves roughly 45%, losing the rest to friction, finite-rate heat transfer across real temperature gaps, and non-ideal steam.
• Why the cold side matters. Drop the condenser to 280 K (a colder river in winter) and the ceiling rises to 1 − 280/810 = 65.4%. Every kelvin you shave off T_cold is worth as much as heating the boiler further — a reason power plants love cold cooling water.
• The Celsius trap. If you mistakenly plug in Celsius (537 and 27), you get 1 − 27/537 = 95%, wildly wrong. The formula is only valid in kelvin because it comes from the ratio of absolute heats.

Carnot vs real engine cycles

	Carnot cycle (ideal)	Real cycles (Otto / Rankine / Brayton)
Processes	2 isotherms + 2 adiabats	Isochoric, isobaric, or polytropic strokes
Reversible?	Yes — quasi-static, ΔSuniv = 0	No — friction, turbulence, finite ΔT
Efficiency	1 − Tc/Th (the ceiling)	Always below the Carnot value
Heat transfer	Across zero temperature gap	Across a finite gap (irreversible)
Power output	Zero — infinitely slow steps	Finite and useful — the whole point
T-S diagram shape	Perfect rectangle	Rounded, lower area for same T range
Working fluid dependence	None — same η for any substance	Strong — steam vs air vs refrigerant matters
Practical use	Benchmark / theoretical limit	Every real engine ever built
Typical efficiency (810 K→300 K)	≈ 63%	≈ 35–45% (steam), ≈ 25–30% (gasoline)

Run it backwards: the ideal refrigerator

Reverse every arrow and the Carnot engine becomes the ideal refrigerator (or heat pump): put in work W, pull heat Q_cold from the cold space, and dump Q_hot = Q_cold + W into the warm room. Its figure of merit is the coefficient of performance, not efficiency:

    Fridge:     COP_cool = Q_cold / W = T_cold / (T_hot − T_cold)
    Heat pump:  COP_heat = Q_hot  / W = T_hot  / (T_hot − T_cold)

A kitchen fridge holding 277 K (4 °C) in a 298 K room has a Carnot COP of 277/(298−277) ≈ 13 — meaning in the ideal limit, one joule of electricity could move thirteen joules of heat. Real fridges reach 2–4 because their compressors and heat exchangers are irreversible. The same math shows why heat pumps beat resistive heaters: a heat pump's COP_heat is always greater than 1, so it delivers more heat than the electricity it consumes.

The entropy view: why the rectangle is exact

Plot the cycle with temperature on the vertical axis and entropy on the horizontal, and something clean happens. The two adiabats are vertical lines (ΔS = 0, constant entropy). The two isotherms are horizontal lines (constant T). The four together enclose a perfect rectangle of height (T_hot − T_cold) and width ΔS = Q_hot/T_hot.

    Over one full cycle (state function → returns to start):
      ΔS_gas   = +Q_hot/T_hot − Q_cold/T_cold = 0   (reversible ⇒ the two terms cancel)
      ΔS_univ  = 0                                    (no entropy generated anywhere)

    W_net = area of rectangle = (T_hot − T_cold) · ΔS

This is the geometric heart of the whole subject. Any irreversibility — a friction spike, a heat leak across a real temperature gap — rounds off a corner or shrinks the rectangle, generating ΔS_univ > 0 and stealing work. The Carnot cycle is simply the cycle that generates no entropy, and that is exactly why it is the most efficient one possible.

Who discovered it, and when

Nicolas Léonard Sadi Carnot published Réflexions sur la puissance motrice du feu (Reflections on the Motive Power of Fire) in 1824, at age 28, motivated by the practical failure of French steam engines to match British ones. Remarkably, he did it while still believing in the caloric theory — the idea that heat was a conserved fluid — yet reached the correct efficiency limit anyway. His single 118-page book was largely ignored for a decade.

Émile Clapeyron revived and formalized Carnot's argument on a P-V diagram in 1834. Then in the late 1840s and 1850s, Rudolf Clausius and William Thomson (Lord Kelvin) rebuilt Carnot's result on the correct foundation — energy conservation plus a new quantity Clausius named entropy in 1865. The Carnot cycle is thus the seed of the entire second law: Kelvin's absolute temperature scale and Clausius's entropy both fall directly out of it. Sadi Carnot died of cholera in 1832 at 36, never seeing his work vindicated.

Limitations and why no one builds one

• Zero power. Reversibility demands quasi-static, infinitely slow steps. A true Carnot engine would take forever to complete one stroke, delivering essentially zero power. Useful engines must run fast, and fast means irreversible.
• The Curzon-Ahlborn limit. If you optimize a heat engine for maximum power instead of maximum efficiency (allowing finite-rate heat transfer), the efficiency at peak power drops to η = 1 − √(T_cold/T_hot). For 810 K→300 K that is 39% — much closer to what real plants actually achieve than the 63% Carnot ceiling.
• Perfect isothermal heat exchange is impractical. Transferring large amounts of heat at a truly constant temperature requires an infinite heat exchanger or infinite time. Real cycles (Rankine, Otto, Brayton) deliberately use constant-volume or constant-pressure heating instead, sacrificing some ceiling efficiency for buildable hardware.
• It is a benchmark, not a blueprint. Engineers use the Carnot efficiency to grade a design (a plant at 45% out of a 63% ceiling is "72% of Carnot"), not to copy the cycle itself. Its value is as the yardstick that says how much room is left.