Second Law of Thermodynamics

Multiple equivalent statements

The second law has many equivalent formulations:

1. Clausius: No process is possible whose sole result is the transfer of heat from a colder body to a hotter body.
2. Kelvin: No process is possible whose sole result is the complete conversion of heat into work.
3. Entropy form: The total entropy of an isolated system can never decrease over time. ΔS_universe ≥ 0.
4. Time's arrow: Time has a thermodynamic direction — from low to high entropy.

These are mathematically equivalent — accepting any one implies all the others.

Entropy

Entropy S is a measure of the disorder (or number of microstates) of a system. Defined as:

dS = dQ_reversible / T  (Clausius definition)
S = k_B · ln(W)  (Boltzmann definition; W = number of microstates)

The 2nd law says ΔS_universe ≥ 0 — for any process in an isolated system, total entropy increases (or stays constant for reversible processes).

Carnot efficiency limit

For a heat engine operating between hot reservoir T_h and cold reservoir T_c:

η_max = η_Carnot = 1 − T_c / T_h

(Both temperatures in Kelvin.) Real engines always have η < η_Carnot due to irreversibilities.

System	T_h	T_c	η_Carnot	Real η
Steam engine	~500 K	~300 K	40%	~25%
Internal combustion (gasoline)	~2200 K	~600 K	73%	~25%
Coal power plant	~800 K	~310 K	61%	~40%
Combined-cycle gas	~1750 K	~310 K	82%	~60%
Solar thermal (high T)	~700 K	~310 K	56%	~30%
Geothermal	~400 K	~280 K	30%	~10-15%

Real-world consequences

Phenomenon	Why 2nd law explains it
Coffee cools to room temperature	Entropy increases as heat spreads from hot to cold
Egg breaks but never unbreaks	Broken state is much more probable (more microstates)
Cars need fuel	Engine can't be 100% efficient; some energy "wasted" as heat
Refrigerators need power	Forcing heat from cold to hot requires work input
Universe expanding to "heat death"	Long-term entropy increase
Stars eventually die	Nuclear fuel exhausted, all energy radiated as low-T photons
Why people age	Living systems fight local entropy decrease but lose over decades

JavaScript — entropy and Carnot calculations

// Carnot efficiency
function carnotEfficiency(T_hot, T_cold) {
  return 1 - T_cold / T_hot;  // K
}

console.log(`Steam engine (500K/300K): ${(carnotEfficiency(500, 300) * 100).toFixed(0)}%`); // 40%
console.log(`Solar (700K/310K): ${(carnotEfficiency(700, 310) * 100).toFixed(0)}%`); // 56%

// Refrigerator coefficient of performance
function carnotCOP_refrigerator(T_cold, T_hot) {
  return T_cold / (T_hot - T_cold);
}

console.log(`Fridge (277K interior, 297K outside): COP = ${carnotCOP_refrigerator(277, 297).toFixed(1)}`);  // ~14

// Entropy change for heat transfer
function entropyChange_heatTransfer(Q, T_hot, T_cold) {
  // Heat Q goes from T_hot to T_cold
  // ΔS_hot = -Q/T_hot, ΔS_cold = +Q/T_cold
  // Total ΔS = Q · (1/T_cold - 1/T_hot) ≥ 0 always
  return Q * (1/T_cold - 1/T_hot);
}

console.log(`Heat 1000 J from 500K to 300K: ΔS = ${entropyChange_heatTransfer(1000, 500, 300).toFixed(2)} J/K`);
// Always positive

// Boltzmann entropy
function boltzmannEntropy(numberOfMicrostates) {
  const k_B = 1.380649e-23;  // J/K
  return k_B * Math.log(numberOfMicrostates);
}

// Compare ordered (1 microstate) vs disordered (10^23 microstates)
console.log(`Single arrangement: S = 0`);
console.log(`10^23 arrangements: S = ${boltzmannEntropy(1e23).toExponential(2)} J/K`);

// Engine efficiency from heat in/out
function engineEfficiencyFromHeats(Q_hot, Q_cold) {
  return (Q_hot - Q_cold) / Q_hot;
}

// Compare to Carnot
function vsCarnot(eta_real, T_hot, T_cold) {
  const eta_carnot = carnotEfficiency(T_hot, T_cold);
  return { real: eta_real, carnot: eta_carnot, efficiency_of_efficiency: eta_real / eta_carnot };
}

console.log(vsCarnot(0.30, 800, 300));  // Real 30% / Carnot 62.5% = 48% of Carnot

Where the second law shows up

• Engineering — heat engines and refrigerators. Sets fundamental efficiency limits. Modern power plants approach Carnot; real refrigerators have known thermodynamic costs.
• Chemistry. Reactions proceed in direction of decreased Gibbs free energy (linked to entropy + enthalpy).
• Astrophysics. Stellar evolution, black hole thermodynamics (Bekenstein-Hawking entropy), heat death of universe.
• Biology. Living organisms reduce local entropy at the cost of larger global entropy increase. Metabolism is essentially thermodynamic.
• Information theory. Shannon entropy is mathematically analogous to thermodynamic entropy. Erasing information has thermodynamic cost (Landauer's principle).
• Statistical mechanics. Boltzmann's S = k·ln(W) connects micro to macro, derives the 2nd law.
• Computing. Reversible computation could in principle have lower energy use; Landauer's bound is ultimate physical limit.

Common mistakes

• Misapplying "entropy always increases". Only for ISOLATED systems. Subsystems can decrease entropy at the cost of increasing it elsewhere (refrigerators, life, manufacturing).
• Treating Carnot as an upper bound for ALL engines. It IS the upper bound for engines operating between two temperatures. Multi-temperature reservoirs (e.g., fuel cells using chemical energy directly) can have higher efficiency.
• Calling it "anti-evolution" or "anti-life". Life isn't a violation. Earth receives low-entropy sunlight and radiates high-entropy heat — entropy budget is huge for life to thrive within.
• Confusing entropy with disorder (loosely). Disorder is a vague metaphor. Entropy is precisely defined — number of microstates, or dQ_rev/T. Be careful with "disorder" in advanced contexts.
• Forgetting Kelvin temperatures. Carnot efficiency uses absolute temperatures (K). Don't use Celsius — the formula breaks.
• Treating reversibility as "easy to reverse." Thermodynamically reversible means entropy doesn't increase — extremely slow processes only. Even quasi-reversible engines have small but nonzero entropy generation.