The Silent Compass

J. Willard Gibbs and the Vectorial Foundation of Thermodynamics

The three-hundred-page treatise that appeared in two installments between 1875 and 1878 in the obscure Transactions of the Connecticut Academy of Arts and Sciences contained not a single experimental measurement, laboratory drawing, or steam-table lookup. Its author, Josiah Willard Gibbs, had been appointed Professor of Mathematical Physics at Yale College in 1871 at age thirty-two. Because Yale's trustees saw no practical financial value in purely theoretical physics, they paid Gibbs zero salary for his first nine years on the faculty. Living on a modest inheritance in the New Haven house half a block from campus where he had grown up, Gibbs worked alone in a third-floor study in the Old Sloane Laboratory until Johns Hopkins University tried to recruit him in 1880 for $3,000 a year, finally prompting Yale to offer him $2,000 to stay.

Before Gibbs published On the Equilibrium of Heterogeneous Substances, nineteenth-century thermodynamics was an engineering manual for steam cylinders. Rudolf Clausius, William Rankine, and Lord Kelvin had formulated the first and second laws of thermodynamics to calculate how much mechanical piston work could be extracted when heat flowed through a single pure fluid—water and steam—inside a closed boiler. Chemists, metallurgists, and geologists had no equivalent law. When a chemist mixed four substances in an open beaker at room temperature and atmospheric pressure, or when a steelmaker melted iron, carbon, and manganese in a blast furnace, nobody could calculate from first principles whether the mixture would react violently, separate into distinct liquid and solid phases, or remain unchanged.

Gibbs realized that the entropy formulation of Clausius—which states that the total entropy $S$ of an isolated universe increases toward a maximum—was mathematically awkward for laboratory chemistry because open chemical vessels are not thermally isolated; they exchange heat and volume freely with the surrounding air at constant temperature $T$ and pressure $P$. To bring chemical matter inside the thermodynamic calculus, Gibbs first represented the state of any fluid geometrically as a three-dimensional surface whose coordinates were volume $V$, entropy $S$, and internal energy $U$. On this surface, the slope along the volume axis gives negative pressure ($-P$), and the slope along the entropy axis gives absolute temperature ($T$).

Gibbs then took the decisive step. He added to the differential energy equation a new set of variables $n_1, n_2, \dots, n_C$ representing the masses or mole numbers of every distinct chemical species in the vessel, and defined the partial derivative of energy with respect to each species as its chemical potential $\mu_i$. Using a Legendre transformation to switch the independent variables from entropy and volume to temperature and pressure, he defined the thermodynamic potential now called the Gibbs free energy ($G = U + PV - TS$). Just as a heavy stone rolls downhill to minimize gravitational potential energy, any multi-component chemical system at constant temperature and pressure evolves spontaneously until its Gibbs free energy reaches a minimum and the chemical potential $\mu_i$ of each species is identical across every coexisting phase.

Counting the number of independent intensive variables ($T$, $P$, and the $C - 1$ concentration ratios inside each phase) and subtracting the number of equilibrium equations equating the chemical potentials across all $P$ phases (solid, liquid, vapor, or crystal polymorphs), Gibbs derived the universal Phase Rule in four lines of elementary algebra:

Here $F$ is the variance, or number of degrees of freedom—the number of physical variables an experimenter can alter independently without causing a phase to appear or vanish. For pure water ($C = 1$) boiling into steam ($P = 2$), $F = 1 - 2 + 2 = 1$, meaning temperature alone fixes the vapor pressure. When ice, liquid water, and water vapor coexist together ($P = 3$), $F = 1 - 3 + 2 = 0$, proving geometrically that the triple point is completely invariant in nature.

Because virtually no American physicist could read the Connecticut journal, Gibbs mailed printed offprints to Europe. Months before his death in 1879, James Clerk Maxwell hand-carved three plaster-of-Paris sculptures of Gibbs's thermodynamic energy-entropy-volume surface at the Cavendish Laboratory and mailed one across the Atlantic to New Haven, where it still rests in Yale's physics department.