The second law of thermodynamics

The first law, ΔU = Q − W, is an equality, and it says nothing about direction. Let a 2.0 kg book fall 1.5 m onto a bench. It gives up mgh = 2.0 × 9.81 × 1.5 = 29 J, which ends up as thermal energy in the book and the bench. Now run the film backwards: the bench and book cool by 29 J — with a combined heat capacity near 800 J K⁻¹ that is a drop of 29/800 = 0.036 K, far too small to notice — and the book rises 1.5 m into your hand. Energy is conserved to the last joule in both films. One happens daily, the other has never been seen. That gap is not a detail the first law forgot to mention; it is a second, independent law, and what it constrains is direction rather than totals. Nothing in mechanics or in energy accounting implies it, so it has to be asserted from experience and then tested.

Kelvin and Planck: no cyclic device can have, as its sole effect, the absorption of heat from a single reservoir and its complete conversion into work. Three words carry the weight. Cyclic means the working substance returns to its starting state, so ΔU = 0 over the cycle and W = Q_H − |Q_C| exactly. Single reservoir means nothing colder is available to dump into. Sole effect means no permanent change is left anywhere else. Put together, they say |Q_C| > 0 strictly, so η = W/Q_H < 1 for every engine ever built or ever to be built — not because of friction or poor insulation, but because of this law. The stakes are large: the ocean holds about 1.4 × 10²¹ kg of water at roughly 4.0 kJ kg⁻¹ K⁻¹, so cooling it by one millikelvin would release 5.6 × 10²¹ J, some nine years of world energy use. A ship that ran on that alone is called a perpetual-motion machine of the second kind, and the engine statement is the claim that none exists.

Clausius: no cyclic device can have, as its sole effect, the transfer of heat from a colder body to a hotter one. Read carelessly, this says heat never flows from cold to hot — which your kitchen refutes every night. A fridge lifting 120 W out of a 277 K cabinet into a 295 K room while drawing 40 W of electrical work does exactly that, and delivers 160 W to the room. It survives the statement because the transfer is not its sole effect: 40 W of work has been consumed and the surroundings carry a permanent record of it. Strip the payment away and the transfer is forbidden. So the statement is not heat never goes from cold to hot; it is heat never goes from cold to hot for free. The same clause is what saves an air conditioner, a heat pump and every cryostat, and the size of the required payment is what the quantitative form of the law fixes.

Suppose someone breaks the engine statement, with a cycle that takes 400 J per cycle from a 500 K reservoir and delivers 400 J of work while rejecting nothing. Feed that work to an ordinary refrigerator running between 300 K and 500 K with a coefficient of performance of 1.0, comfortably inside the reversible limit of 300/(500 − 300) = 1.5. It lifts 400 J from the cold reservoir and delivers 400 + 400 = 800 J to the hot one. Draw one box around both machines. The hot reservoir gives up 400 J and receives 800 J, a net gain of 400 J; the cold reservoir loses 400 J; no work crosses the boundary at all. The box moves 400 J from cold to hot as its sole effect, which is precisely what Clausius forbids. The reverse coupling works too, so the two statements stand or fall together. They are not two laws that happen to agree; they are one law read through an engine or through a refrigerator.

A process is reversible if at every instant the system and its surroundings are only infinitesimally out of balance, so the whole path can be retraced leaving no residue anywhere. Nothing real qualifies. Heat crossing a finite temperature difference, friction and viscous drag, a gas expanding into vacuum, two gases mixing, current warming a resistor, a metal bent past its elastic limit — each of these has a direction you can spot on film, and each is irreversible. Reversibility demands infinite time, so it is a limit, not a procedure. Its value is that it is the boundary: every real process falls on one side of it, so the reversible device sets the record that real ones approach and never beat. That is the origin of the inequality ∮ dQ/T ≤ 0, whose equality is reserved for the reversible cycle, and it is why the next topic can turn the whole law into a single statement about entropy.

In practice nobody uses the second law to say impossible; they use it to say at most. Because the reversible cycle is the boundary case, no cycle between two reservoirs can beat η = 1 − T_C/T_H, and no refrigerator can beat K = T_C/(T_H − T_C). A supercritical steam plant taking steam at 873 K and condensing at 303 K has a ceiling of 1 − 303/873 = 0.653; a real net efficiency of 46% is therefore 70% of the best the law allows, which is a fair verdict on the engineering and a hard stop on the marketing. Both ceilings demand absolute temperatures: writing 1 − 30/600 for a plant quoted in degrees Celsius gives 0.95 and is nonsense. The habit worth forming is to audit any claim in two passes — first the first law, does the energy balance, then the second law, does the work or the cooling sit under the ceiling.