Fugacity & Activity

The patch that saved the ideal equations

Thermodynamics has one beautiful equation for how a substance's escaping tendency — its chemical potential μ — responds to pressure. For an ideal gas it reads

μ = μ° + RT·ln(P/P°)

It is clean, logarithmic, and exact — for a gas that does not exist. Real molecules attract and repel each other, so a real gas at high pressure has a different escaping tendency than this equation predicts. In 1901 Gilbert N. Lewis refused to abandon the equation. Instead he asked: what number would I have to substitute for P to make the equation true again? He called that number the fugacity, f — from the Latin fugere, "to flee," because it measures the tendency of molecules to escape a phase.

μ = μ° + RT·ln(f/f°)        (exact, for any real gas)

This is the whole trick. The equation keeps its elegant form; all the messy non-ideality is bundled into a single quantity. Fugacity is an effective pressure — the pressure a real gas would seem to have if it behaved ideally with respect to chemical potential. For an ideal gas f = P exactly. For a real gas the two differ, and the ratio carries a name: the fugacity coefficient,

φ = f / P

φ is the dimensionless deviation factor. φ < 1 means molecules escape less readily than their pressure suggests (attractions winning); φ > 1 means they escape more readily (repulsions winning at high compression). As P → 0 every gas becomes ideal, so φ → 1 and f → P. The standard state f° is conventionally chosen as exactly 1 bar (the ideal gas at 1 bar), which makes the logarithm's argument dimensionless.

Activity: the same idea for liquids and solutions

Gases are not the only non-ideal systems. A solute in solution, a component in a liquid mixture, a metal in an alloy — none of them behave as the simple "concentration goes in the log" picture suggests. Lewis's fix generalizes immediately. For condensed phases we define the activity a as a dimensionless effective concentration, again chosen so the chemical-potential equation stays exact:

μ = μ° + RT·ln(a)        where   a = f / f°

Activity is fugacity, simply normalized by the fugacity in a chosen reference state. That normalization strips the units and pins a = 1 for a pure substance in its standard state. The deviation from ideal-solution behaviour is captured by the activity coefficient γ:

a = γ · x        (mole-fraction scale, Raoult reference)
a = γ · (m/m°)   (molality scale, used for ionic solutes)
a = γ · (c/c°)   (molarity scale)

Here x is mole fraction, m is molality, c is molarity, and the °-superscripted values are reference points (1 mol/kg or 1 mol/L). An ideal solution has γ = 1 at all compositions; a real solution has γ that drifts away from 1 as interactions between unlike molecules differ from interactions between like molecules. The pairing is exact and worth memorizing:

GAS:       fugacity f  →  coefficient φ = f/P  →  ideal when φ = 1
SOLUTION:  activity a  →  coefficient γ = a/x  →  ideal when γ = 1

Where fugacity comes from: the compressibility factor

Fugacity is not measured directly; it is computed from how badly a gas violates PV = nRT. The yardstick is the compressibility factor Z = PV/(nRT), which equals 1 for an ideal gas. Integrating its departure from 1 gives the fugacity coefficient:

ln φ = ∫₀ᴾ (Z − 1)/P · dP        ⇒        f = φ·P

The logic is intuitive: where Z < 1 (attractions dominate, gas is more compressible than ideal) the integral is negative, φ < 1, and f < P. Where Z > 1 (repulsions dominate, gas resists compression) the integral turns positive, φ > 1, and f > P. For nitrogen at 273 K the data give:

P = 1 atm     Z ≈ 0.9998   φ ≈ 1.000   f ≈ 1.00 atm
P = 100 atm   Z ≈ 0.985    φ ≈ 0.97    f ≈ 97 atm
P = 400 atm   Z ≈ 1.26     φ ≈ 1.05    f ≈ 420 atm
P = 1000 atm  Z ≈ 2.0      φ ≈ 1.84    f ≈ 1840 atm

Up to about a hundred atmospheres nitrogen's attractions win and fugacity lags pressure, bottoming out near φ ≈ 0.97. Beyond ~150–200 atm the molecules are packed tightly enough that hard-core repulsion takes over, Z and φ climb back above 1, and the fugacity overtakes the pressure — by 1000 atm f is nearly twice P. Hydrogen and helium, whose attractions are weak, have φ > 1 over essentially the whole pressure range even at room temperature.

Any equation of state plugs straight into the integral. The van der Waals equation gives a closed-form ln φ; the Redlich–Kwong (1949) and Peng–Robinson (1976) equations are the workhorses of process engineering precisely because they reproduce real φ values to a few percent across the conditions inside a distillation column or a high-pressure reactor.

Activity in solutions: positive and negative deviations

For a volatile component in a liquid mixture, activity is what you read off the vapor. Raoult's law says an ideal component's partial pressure is P_i = x_i·P_i*, where P_i* is the pure-component vapor pressure. The activity is defined so that the real partial pressure obeys

P_i = a_i · P_i*  = γ_i · x_i · P_i*

So γ_i is literally the factor by which a component's real vapor pressure exceeds (γ > 1) or falls short of (γ < 1) the Raoult prediction.

• Positive deviation (γ > 1). Unlike molecules dislike each other; they escape more readily than in an ideal mix. Ethanol–water is the classic case — strong like–like hydrogen bonding, weaker cross-interaction. The excess escaping tendency produces a minimum-boiling azeotrope at 95.6 wt% ethanol (78.2 °C), the wall that ordinary distillation cannot cross.
• Negative deviation (γ < 1). Unlike molecules attract more strongly than like ones, suppressing escape. Acetone–chloroform forms a hydrogen bond between the chloroform H and the acetone carbonyl, giving γ < 1 and a maximum-boiling azeotrope at 64.5 °C.

Two thermodynamic anchors fix the endpoints. By Raoult's law the solvent (the major component) has γ → 1 as x → 1. By Henry's law the dilute solute has a constant activity coefficient γ → γ^∞ (its infinite-dilution value) as x → 0, where the relevant reference is the Henry constant rather than the pure-component vapor pressure.

Ionic solutions and the Debye–Hückel limit

Electrolytes are the most spectacularly non-ideal solutions because long-range Coulomb forces reach across many solvent molecules. Each ion drags an oppositely charged "ionic atmosphere" that lowers its escaping tendency, so γ falls below 1 even in fairly dilute solution. The Debye–Hückel limiting law (1923) predicts the mean ionic activity coefficient γ± from the ionic strength I alone:

log₁₀ γ± = −A·|z₊z₋|·√I          (A = 0.509 for water at 25 °C)
I = ½ · Σ cᵢ·zᵢ²

For 0.01 M NaCl, I = 0.01 and the law gives γ± ≈ 0.89 — already a 10 % correction. By 0.1 M the limiting law (γ± ≈ 0.69) already overshoots the measured γ± ≈ 0.78, the first sign that finite ion size matters. In seawater (I ≈ 0.7 M) the simple limiting law breaks down badly and extended forms (Davies equation, Pitzer model) are needed, but the message holds: ignore activity coefficients in concentrated ionic media and your equilibrium predictions are wrong by tens of percent. The famous pH scale is defined in terms of hydrogen-ion activity, pH = −log₁₀ a(H⁺), not concentration — which is why a glass electrode reads activity directly.

Why equilibrium constants are built from activities

The link that makes all this matter is the master equation of chemical equilibrium:

ΔG° = −RT·ln K

This K is a true constant — fixed by the standard Gibbs energy of reaction and therefore by temperature alone. But because it descends from chemical potentials, the species in it must appear as activities, not concentrations or pressures. For a gas reaction such as the Haber–Bosch synthesis

N₂(g) + 3 H₂(g)  ⇌  2 NH₃(g)

K  =  (f_NH₃)² / [ f_N₂ · (f_H₂)³ ]  =  Kφ · Kp

where Kp uses partial pressures and Kφ = φ_NH₃² / (φ_N₂·φ_H₂³) collects the fugacity coefficients. Ammonia is made at ~200 atm and 450 °C precisely the regime where φ values stray from 1; engineers who used Kp instead of the true activity-based K would predict the wrong equilibrium yield by several percent — and at industrial scale that is millions of tonnes. For a solution reaction the same structure holds with activities:

K = Π aᵢ^νᵢ = Kc · Kγ        where  Kγ = Π γᵢ^νᵢ

In dilute solution Kγ ≈ 1 and the activity-based K collapses to the familiar Kc taught in introductory courses. That collapse is exactly why introductory chemistry can get away with concentrations — it is silently working in the limit where every γ = 1.

Fugacity vs activity vs the raw quantities

	Pressure / Concentration	Fugacity (f)	Activity (a)
Applies to	Both, mechanically	Gases (and gas in mixtures)	Liquids, solids, solutes
Units	atm / bar / mol·L⁻¹	Pressure units (atm/bar)	Dimensionless
Deviation factor	—	φ = f/P	γ = a/x (or a/m, a/c)
Ideal-limit value	—	φ = 1 (P → 0)	γ = 1 (pure or x → 0)
Standard state	—	Ideal gas at 1 bar	Pure / unit molality, γ=1
Goes into	Mechanical balances	μ, Kf for gas equilibria	μ, Ka, K_sp, electrode potentials
Computed from	Direct measurement	∫(Z−1)/P dP, equation of state	Vapor pressure, EMF, Debye–Hückel

The unifying truth is in the bottom-left-to-right diagonal: fugacity and activity are the quantities you must use everywhere a chemical potential appears. Pressure and concentration are what you use everywhere a force or amount appears. The two are equal only in the ideal limit.

Where fugacity and activity decide outcomes

• Carbon-capture and gas pipelines. Supercritical CO₂ injected underground at 100–200 bar has φ ≈ 0.3–0.5; treating it as ideal would mis-size compressors and mis-predict how much dissolves into brine. Solubility models for sequestration run on fugacity, not pressure.
• The lever rule of distillation. Activity coefficients set the position of azeotropes. The ethanol–water azeotrope at 95.6 wt% (γ-driven) is why "absolute" ethanol needs benzene entrainment or molecular sieves, not just a taller column.
• Ocean and blood chemistry. The solubility of calcium carbonate in seawater, and therefore coral reef survival, is governed by ion activities. The free-ion activity coefficients of the divalent ions are small — γ(Ca²⁺) ≈ 0.2, and CO₃²⁻ is so heavily ion-paired with Na⁺ and Mg²⁺ that its free activity coefficient is only ≈ 0.04 — so the apparent (concentration-based) solubility product of calcite in seawater exceeds the thermodynamic one by about an order of magnitude. Treat ions as their molarities and you grossly over-predict the saturation state Ω that determines whether shells dissolve.
• Metallurgy. In steelmaking, the activity of carbon dissolved in molten iron (not its mass fraction) controls whether it reduces an oxide or precipitates as graphite; activity coefficients in liquid alloys can exceed 10.
• Battery and corrosion potentials. The Nernst equation E = E° − (RT/nF)·ln Q uses activities in Q. A lithium-ion cell's measured voltage tracks the activity of Li⁺ in the electrolyte, which is far from its molarity at the concentrations inside a real cell.

Common misconceptions and pitfalls

• Thinking fugacity is "just a fudge factor." It is a rigorously defined thermodynamic state function with units of pressure, derived from PVT data through ln φ = ∫(Z−1)/P dP. It is no more arbitrary than enthalpy.
• Confusing fugacity with partial pressure. Partial pressure is mechanical and measurable; fugacity is the chemical-potential-corrected version, f_i = φ_i·y_i·P. They coincide only for ideal gas mixtures.
• Forgetting activity is dimensionless. Because a = γ·(c/c°), the standard concentration c° (1 mol/L) divides out the units. Writing ln(c) with units inside the log is the most common dimensional error in equilibrium problems.
• Using concentrations in K for concentrated or ionic systems. A Kc computed from molarities in 1 M salt can be wrong by a factor of 2–3. The thermodynamic K uses activities; only in dilute solution does Kc approximate it.
• Assuming γ < 1 always. Ionic γ± usually falls below 1 at low I but rises back above 1 at high concentration as ion–solvent competition and finite ion size take over — NaCl's γ± bottoms out near 0.66 at ~1 M and climbs past 1 above ~6 M.
• Mixing standard states. Activity is meaningless without naming the reference: Raoult (pure liquid, γ → 1 as x → 1) versus Henry (infinite dilution, γ → 1 as x → 0). The same solution has different activity coefficients depending on which convention you adopt.