Entropy as a State Function

Melt an ice cube slowly on a warm plate, or crush it irreversibly between your palms. Either way, the entropy change of the water is exactly the same. This is not a coincidence. It reflects the fact that entropy is a state function. This is a property fixed entirely by the current state of a system. It has no memory of how that state was reached. This idea matters because it turns the abstract Clausius definition of entropy into a genuinely usable tool. If entropy only depended on the messy details of a real, irreversible process, we could never calculate it. Most real processes are irreversible, and their heat exchanges are hard to track. Because entropy is path-independent, we are free to invent any convenient reversible path between the same two states. We can then integrate along that path instead. This article proves that $dS = dq_{rev}/T$ defines a true state function. It shows why $\oint dS = 0$ around any cycle. It then demonstrates the practical payoff. This means replacing irreversible paths with reversible substitutes for real calculations.

Key takeaways

Entropy $S$ is a state function: $\Delta S$ depends only on initial and final states, never on the path taken.
The proof rests on the Clausius inequality applied to a full thermodynamic cycle, forcing $\oint dq_{rev}/T = 0$ .
Because $dS$ is an exact differential, $\Delta S$ can be computed along ANY convenient reversible path, even to describe an irreversible change.
For an irreversible process, you must construct a reversible path between the same endpoints — you cannot integrate $dq/T$ along the irreversible path itself.
Path-independence is what lets tabulated standard entropies $S^\circ$ exist and be added and subtracted like any other state function.
Key idea: entropy's path-independence converts an abstract inequality into a calculable, tabulable state function.

The core insight

Because $\oint dq_{rev}/T = 0$ for every reversible cycle, the quantity $dq_{rev}/T$ is a mathematically exact differential; integrating it between two states always gives the same value regardless of the connecting path, which is precisely the defining property of a state function, so entropy can be assigned to a state itself rather than to a process.

The physical picture

A state function is any property whose value depends only on the current pressure, temperature, and composition of a system — not on its history. Internal energy $U$ and enthalpy $H$ are familiar examples: you can heat a gas then compress it, or compress it then heat it, and reach the same final $U$ either way. Work $w$ and heat $q$ are the opposite kind of quantity, path functions, because the amount of work done depends on exactly how a process is carried out, even between fixed endpoints. The question for entropy is which category it belongs to. The Clausius inequality states that for any process,

dS \geq \frac{dq}{T}

with equality holding only for a reversible process. This inequality alone does not prove $S$ is a state function; it only defines a lower bound built from a path-dependent quantity, heat. The proof that entropy is nonetheless path-independent requires examining what happens when a system is taken around a closed cycle using only reversible steps, returning exactly to its starting state. Picture a system carried from state A to state B by reversible path 1, then back from B to A by a different reversible path 2. Because the whole loop is reversible, the Clausius inequality applies with equality at every point, and integrating around the full closed loop must give zero — a consequence of Carnot's theorem for reversible cycles operating between varying temperatures. If the integral of $dq_{rev}/T$ around every closed reversible loop vanishes, then the integral from A to B cannot depend on which reversible path was used: any two reversible routes between the same two states must give the same value, since together they would otherwise form a nonzero loop integral.

Figure

Two reversible paths between the same states give the same ∫dq_rev/T

The second, equally important consequence is practical. Suppose the real process from A to B is irreversible — a free expansion, a spontaneous mixing, an exothermic reaction running to completion. Because $S$ is a state function, $\Delta S$ for that irreversible change equals $\Delta S$ for any reversible path connecting the same two endpoints. You are never obligated to compute $dq/T$ along the actual, irreversible trajectory — indeed you generally cannot, because $dq_{irrev}/T$ is not even equal to $dS$ pointwise. Instead you design a convenient reversible substitute: isothermal reversible expansion, reversible heating at constant pressure, reversible phase change at the transition temperature, and integrate $dq_{rev}/T$ along that imagined path.

Figure

Irreversible process replaced by a reversible substitute path for the calculation

The core relations

dS \equiv \frac{dq_{rev}}{T}

where: $dq_{rev}$ is the heat exchanged along a reversible path connecting infinitesimally close states, $T$ is the absolute temperature at which that exchange occurs.

\oint \frac{dq_{rev}}{T} = 0

where: the loop integral runs over any complete reversible cycle, the vanishing value shows $dq_{rev}/T$ is an exact differential.

\Delta S = \int_{A}^{B} \frac{dq_{rev}}{T} \quad \text{(any reversible path from A to B)}

where: $\Delta S = S_B - S_A$ is path-independent even though $q$ itself is path-dependent.

\Delta S_{irrev, A\to B} = \Delta S_{rev, A\to B}

where: the left side is the actual entropy change of the system for a real irreversible process, computed by substituting any convenient reversible path with the same endpoints.

Derivation AProving dq_rev/T is exact: entropy is a state function

Goal. Show that the loop integral of $dq_{rev}/T$ vanishes for an arbitrary reversible cycle, and that this forces $\Delta S$ between any two states to be independent of the reversible path chosen.

Step 1: Decompose an arbitrary reversible cycle into Carnot cycles. Any smooth reversible cycle on a $P$ – $V$ diagram can be approximated arbitrarily well by a zigzag mesh of thin Carnot cycles: adiabats stitched together by many short isotherms at closely spaced temperatures. As the mesh is refined, the zigzag boundary converges to the true cycle, and the interior adiabatic legs cancel in pairs since each is traversed once in each direction by neighboring sub-cycles.

Step 2: Apply the Carnot result to each strip. For a single Carnot engine operating between hot reservoir $T_h$ and cold reservoir $T_c$ , absorbing heat $q_h$ and expelling $q_c$ (defined here as a positive magnitude), the Carnot efficiency theorem gives

\frac{q_h}{T_h} = \frac{q_c}{T_c}

so that, using the sign convention where heat expelled by the system is negative,

\frac{q_h}{T_h} + \frac{-q_c}{T_c} = 0

Each thin strip of the mesh obeys exactly this relation.

Step 3: Sum over all strips. Summing the two-term relation from Step 2 over every strip in the mesh gives a telescoping sum equal to zero, because every interior heat exchange cancels and only the boundary survives:

\sum_i \frac{q_i}{T_i} = 0

Step 4: Take the continuum limit. As the number of strips tends to infinity and each becomes infinitesimally thin, the sum becomes an integral over the actual reversible cycle:

\oint \frac{dq_{rev}}{T} = 0

This holds for every reversible cycle, of any shape, not just Carnot cycles, because the mesh construction works for an arbitrary closed curve.

Step 5: Convert the vanishing loop integral into path-independence. Consider two distinct reversible paths, path 1 and path 2, both connecting state A to state B. Form a cycle by going A → B along path 1, then B → A along the reverse of path 2. This composite loop is fully reversible, so by Step 4,

\int_{A \to B, \text{path 1}} \frac{dq_{rev}}{T} + \int_{B \to A, \text{path 2 reversed}} \frac{dq_{rev}}{T} = 0

Reversing the direction of traversal of path 2 flips the sign of its integral, giving

\int_{A \to B, \text{path 1}} \frac{dq_{rev}}{T} = \int_{A \to B, \text{path 2}} \frac{dq_{rev}}{T}

Step 6: Conclude. Since path 1 and path 2 were arbitrary reversible paths, the integral of $dq_{rev}/T$ from A to B is the same number regardless of which reversible path is used. This value can therefore be assigned to the states themselves, defining a function $S$ with

\boxed{\Delta S = S_B - S_A = \int_A^B \frac{dq_{rev}}{T}, \text{ independent of path}}

Worked examples

Example 1 — Reversible isothermal expansion of an ideal gas by two different reversible routes. Take 1.00 mol of ideal gas at 298 K, expanding reversibly and isothermally from 10.0 L to 20.0 L. Confirm that two different reversible calculation routes give the same $\Delta S$ . Route A — direct isothermal integration. For an ideal gas at constant $T$ , $dU = 0$ , so $dq_{rev} = -dw_{rev} = P\,dV$ . Then

\Delta S = \int_{V_1}^{V_2} \frac{P\,dV}{T} = \int_{V_1}^{V_2} \frac{nR}{V}\,dV = nR \ln\frac{V_2}{V_1}

\Delta S = (1.00\ \text{mol})(8.314\ \text{J K}^{-1}\text{mol}^{-1})\ln\left(\frac{20.0}{10.0}\right) = 8.314 \times 0.6931 = 5.76\ \text{J K}^{-1}

Route B — via a state function shortcut, $\Delta S = nC_{V,m}\ln(T_2/T_1) + nR\ln(V_2/V_1)$ . Since this is isothermal, $T_2 = T_1$ and $\ln(T_2/T_1) = 0$ , leaving the identical result,

\Delta S = 0 + (1.00)(8.314)\ln(2.00) = 5.76\ \text{J K}^{-1}

Both routes, one a direct mechanical-work integral and the other a general two-variable entropy formula built from tabulated heat capacities, agree exactly: $\Delta S = 5.76\ \text{J K}^{-1}$ . This is the expected sign: expansion increases the number of accessible microstates, so $\Delta S > 0$ , consistent with the qualitative picture of entropy and disorder.

Example 2 — Irreversible free expansion, calculated via a reversible substitute path. Now let the same 1.00 mol of ideal gas undergo a free expansion (Joule expansion) from 10.0 L to 20.0 L at 298 K: the gas expands into a vacuum inside a rigid, insulated container. Experimentally, $q = 0$ and $w = 0$ , so naively one might guess $\Delta S = q/T = 0$ . This is wrong, and seeing why illustrates the whole point of path-independence.

Step 1. Because $S$ is a state function, $\Delta S$ for this irreversible process must equal $\Delta S$ for any reversible process connecting the same initial state (1.00 mol, 10.0 L, 298 K) to the same final state (1.00 mol, 20.0 L, 298 K). The actual path (free expansion) is irreversible and cannot be used to compute $\Delta S$ via $\int dq/T$ , since $dq_{irrev} \neq dq_{rev}$ pointwise.

Step 2. Choose the convenient reversible isothermal expansion between the same two states — exactly the path already integrated in Example 1, since the endpoints coincide (same $T$ , because internal energy of an ideal gas depends only on $T$ , and free expansion of an ideal gas causes no temperature change).

Step 3. Import the result directly:

\Delta S_{system} = nR\ln\frac{V_2}{V_1} = 5.76\ \text{J K}^{-1}

Step 4. Check consistency with the Clausius inequality. For the real irreversible process, $q_{actual} = 0$ , so

\Delta S_{system} \geq \frac{q_{actual}}{T} = \frac{0}{298} = 0

Indeed $5.76\ \text{J K}^{-1} \geq 0$ , confirming the inequality holds strictly, as expected for an irreversible process. The surroundings exchange no heat either ( $q_{surr}=0$ , insulated rigid container), so $\Delta S_{surr} = 0$ and the total entropy change of the universe is

\Delta S_{univ} = \Delta S_{sys} + \Delta S_{surr} = 5.76 + 0 = 5.76\ \text{J K}^{-1} > 0

consistent with spontaneity of free expansion — a clean numeric confirmation that the state-function property, not the literal heat flow, is what gives the correct entropy change.

Limiting cases & common traps

Situation	Correct treatment
Process is irreversible	Never integrate $dq_{irrev}/T$ directly; build a reversible path with the same endpoints
Cyclic process, any kind	$\Delta S_{system} = 0$ exactly, since initial and final states coincide
Adiabatic and reversible	$dq_{rev} = 0$ at every step, so $\Delta S = 0$ (isentropic)
Adiabatic but irreversible	$q = 0$ does not imply $\Delta S = 0$ ; $\Delta S > 0$ is typical
Comparing two reversible paths, same endpoints	Must give identical $\Delta S$ ; if not, an error was made

The chemist behind it

Rudolf Clausius introduced the word "entropy" in 1865, deliberately choosing a Greek root meaning "transformation" to echo "energy." He had already shown, building on Sadi Carnot's earlier analysis of ideal heat engines, that the quantity $\oint dq_{rev}/T$ vanishes for any reversible cycle. Clausius recognized that this vanishing loop integral was the mathematical signature of a state function, and he used it to formulate his own version of the Second Law: entropy of an isolated system can never decrease. His insight — extracting a path-independent property from an inherently path-dependent quantity, heat — is exactly the derivation given above, and it remains the standard route through which every modern thermodynamics course introduces entropy.

Origins · 1 minRudolf Clausius and how this was discovered

Rudolf Clausius

1822-1888 · German physicist who coined "entropy" and proved its path-independence from the Carnot cycle

Ask the tutor: How do we extend this reversible-path method to calculate entropy changes when temperature also changes, using heat capacities?

Why can't we just integrate dq/T along the real, irreversible path?

Because for an irreversible process the actual heat exchanged, $q_{irrev}$ , does not satisfy $dq_{irrev}/T = dS$ at each step; the Clausius inequality tells us only that $dS \geq dq_{irrev}/T$ , with the inequality strict except at equilibrium. Entropy is defined through $dq_{rev}/T$ , the heat exchanged along a reversible path, so calculating $\Delta S$ requires replacing the real process with a reversible one sharing the same endpoints, then integrating along that instead.

Does path-independence mean entropy changes are always easy to compute?

It means a convenient reversible path always exists in principle, but choosing it well still takes practice. Common convenient choices include reversible isothermal steps, reversible constant-pressure heating using heat capacities, and reversible phase transitions at the transition temperature. The skill lies in decomposing a complicated real change into a sequence of simple reversible steps with identical overall endpoints.

How does this connect to tabulated standard molar entropies, S degree?

Tabulated values such as $S^\circ_m$ for water or oxygen gas only make sense because entropy is a state function: each value refers to one fixed state (298 K, 1 bar, specified phase) rather than to any process. Reaction entropies are then obtained by simple subtraction, $\Delta S^\circ_{rxn} = \sum \nu_i S^\circ_{m,i}(\text{products}) - \sum \nu_i S^\circ_{m,i}(\text{reactants})$ , exactly as with enthalpies of formation.

Is internal energy path-independent for the same reason as entropy?

Both rely on the same mathematical idea — an exact differential — but the physical justification differs. Internal energy is path-independent by the First Law itself, since $dU = dq + dw$ combines two path-dependent quantities into an exact differential essentially by definition and experimental confirmation. Entropy's path-independence instead requires the extra physical content of the Second Law, specifically the Carnot cycle result, which is a nontrivial theorem rather than a direct definition.

What happens if I mistakenly use an irreversible path's heat in the entropy formula?

You will generally get a value smaller than the true $\Delta S$ , or even the wrong sign, because $q_{irrev}$ is not equal to $q_{rev}$ between the same states. In the free-expansion example, using $q_{irrev}=0$ incorrectly suggests $\Delta S = 0$ , contradicting the true answer of $+5.76\ \text{J K}^{-1}$ . Always check whether the specified path is reversible before plugging heat values into $dS = dq/T$ .