Inversion Temperature and Gas Liquefaction

Turn the knob on a cylinder of compressed nitrogen and let the gas hiss out through a small nozzle. Sometimes the escaping gas feels cold; under different starting conditions, it can actually warm up. This split personality is the Joule-Thomson effect, and the temperature that separates the two behaviors is called the inversion temperature. The inversion temperature is not a curiosity. It is the reason industrial gas liquefaction plants work at all, and the reason some gases simply refuse to cool by throttling near room temperature. Understanding it means understanding how real, non-ideal intermolecular forces leave fingerprints on a macroscopic property. This article builds on the Joule-Thomson coefficient. You will see why $\mu_{JT}$ changes sign. Then you will derive the inversion temperature from the van der Waals equation. Finally you will apply the result to real liquefaction cycles like Linde refrigeration.

Key takeaways

The Joule-Thomson coefficient $\mu_{JT}$ can be positive (cooling) or negative (heating) on throttling.
The inversion temperature $T_i$ is where $\mu_{JT} = 0$ ; it depends on pressure, forming an inversion curve.
Below $T_i$ , attractive forces dominate and expansion cools the gas; above $T_i$ , size effects dominate and expansion heats it.
For a van der Waals gas, the maximum inversion temperature is $T_i = 2a/(Rb)$ at low pressure.
Liquefaction (Linde process) requires starting the gas below its inversion temperature, often after precooling.
Key idea: Real-gas cooling on throttling is a competition between attractive forces (which favor cooling) and molecular size (which favors heating), and the inversion temperature marks the tipping point.

The core insight

The Joule-Thomson coefficient changes sign because throttling forces gas molecules to do work against two competing effects of non-ideality — pulling apart against attractive forces (which extracts internal energy and cools the gas) and rearranging finite-sized molecules (which can release energy and warm the gas) — and the inversion temperature is the exact point where these two contributions cancel.

The physical picture

In an ideal gas, throttling changes nothing thermally: enthalpy is conserved in a Joule-Thomson expansion, and for an ideal gas enthalpy depends only on temperature, so $T$ stays fixed. Real gases are different because their enthalpy depends on both $T$ and $p$ through intermolecular forces. Picture the gas molecules approaching the throttle, a porous plug or fine valve, with the pressure dropping sharply from $p_1$ to $p_2$ . As the average intermolecular separation increases, two things happen simultaneously. First, molecules must pull away from each other against attractive (van der Waals) forces; this "unsticking" costs potential energy, which is drawn from the kinetic energy pool, cooling the gas. Second, because real molecules have finite volume, the reduction in crowding effectively lets the gas do less "negative work" reclaiming volume than an ideal gas would — an effect that tends to warm the gas. At low temperature, molecules move slowly and spend more time close together, so attractive forces dominate: throttling cools the gas ( $\mu_{JT} > 0$ ). At high temperature, molecules move fast, attractive forces barely matter, and the finite-size (repulsive, or "excluded volume") effect dominates: throttling heats the gas ( $\mu_{JT} < 0$ ). The crossover temperature, at fixed pressure, is the inversion temperature $T_i$ .

Figure

Joule-Thomson coefficient sign change across the inversion temperature

The inversion temperature itself is not a single number for a gas — it depends on pressure, tracing out an inversion curve in the $p$ – $T$ plane. Inside this curve, throttling always cools; outside it, throttling heats or the effect vanishes.

Figure

Inversion curve in the pressure-temperature plane for a van der Waals gas

The core relations

\mu_{JT} = \left(\frac{\partial T}{\partial p}\right)_H

where: $\mu_{JT}$ is the Joule-Thomson coefficient (K/atm), positive means cooling on expansion ( $dp<0$ ).

\mu_{JT} = -\frac{1}{C_p}\left(\frac{\partial H}{\partial p}\right)_T

where: $C_p$ is the constant-pressure heat capacity; this links $\mu_{JT}$ to the pressure-dependence of enthalpy.

T_i = \frac{2a}{Rb}\quad(\text{low-pressure limit, van der Waals gas})

where: $a$ measures attractive-force strength, $b$ the excluded molar volume, $R$ the gas constant.

p_i = 3\left(\frac{2a}{Rb T_i'}-1\right)^{1/2}\!\!\left(\frac{RT_i'}{b}\right) - \text{(full inversion curve, reduced form)}

where: this traces the complete inversion curve; the simple $T_i = 2a/Rb$ result is its low-pressure limit.

Derivation AInversion temperature of a van der Waals gas

Goal. Find the temperature $T_i$ at which $\mu_{JT} = 0$ for a van der Waals gas, in the low-pressure limit.

Step 1. Start from the general thermodynamic expression for $\mu_{JT}$ . Using the definition of enthalpy and a Maxwell relation, one can show generally that

\mu_{JT} = \frac{1}{C_p}\left[T\left(\frac{\partial V}{\partial T}\right)_p - V\right]

This is exact for any gas; the bracket vanishes identically for an ideal gas, confirming $\mu_{JT}=0$ there.

Step 2. Write the van der Waals equation of state to first order in $b$ and $a$ . The van der Waals equation is

\left(p + \frac{a}{V_m^2}\right)(V_m - b) = RT

Solving for $V_m$ exactly is messy, so we expand for a dilute (low pressure) gas, keeping terms linear in $a$ and $b$ :

V_m \approx \frac{RT}{p} + b - \frac{a}{RT}

This approximation is the standard "leading real-gas correction" form, valid when $a/(V_mRT)$ and $b/V_m$ are both small.

Step 3. Differentiate $V_m$ with respect to $T$ at constant $p$ .

\left(\frac{\partial V_m}{\partial T}\right)_p = \frac{R}{p} + \frac{a}{RT^2}

Step 4. Substitute into the bracket of Step 1.

T\left(\frac{\partial V_m}{\partial T}\right)_p - V_m = T\left(\frac{R}{p} + \frac{a}{RT^2}\right) - \left(\frac{RT}{p} + b - \frac{a}{RT}\right)

The $RT/p$ terms cancel exactly, leaving

T\left(\frac{\partial V_m}{\partial T}\right)_p - V_m = \frac{a}{RT} + \frac{a}{RT} - b = \frac{2a}{RT} - b

Step 5. Set the bracket to zero — this defines the inversion condition. $\mu_{JT} = 0$ requires the bracket itself to vanish (since $C_p \neq 0$ ):

\frac{2a}{RT_i} - b = 0

Step 6. Solve for $T_i$ .

\boxed{T_i = \frac{2a}{Rb}}

This is the low-pressure (leading-order) inversion temperature. Above $T_i$ , the bracket is negative ( $2a/RT < b$ ), so $\mu_{JT}<0$ and throttling heats the gas; below $T_i$ , $\mu_{JT}>0$ and throttling cools it — exactly the physical picture of attraction (the $a$ term) competing with excluded volume (the $b$ term).

Worked examples

Example 1 — Computing $T_i$ for nitrogen from van der Waals constants. Nitrogen's van der Waals constants are $a = 1.408\ \text{L}^2\,\text{atm}\,\text{mol}^{-2}$ and $b = 0.03913\ \text{L mol}^{-1}$ (standard tabulated values). Apply the boxed result:

T_i = \frac{2a}{Rb} = \frac{2 \times 1.408\ \text{L}^2\,\text{atm}\,\text{mol}^{-2}}{(0.08206\ \text{L atm K}^{-1}\text{mol}^{-1})(0.03913\ \text{L mol}^{-1})}

Compute numerator: $2 \times 1.408 = 2.816\ \text{L}^2\,\text{atm}\,\text{mol}^{-2}$ . Compute denominator: $0.08206 \times 0.03913 = 3.211\times10^{-3}\ \text{L}^2\,\text{atm}\,\text{K}^{-1}\text{mol}^{-2}$ .

T_i = \frac{2.816}{3.211\times10^{-3}}\ \text{K} = 877\ \text{K}

Sanity check: the experimental inversion temperature of nitrogen (upper branch, low pressure) is about 621 K. The van der Waals estimate (877 K) overshoots by roughly 40%, which is typical — the linear expansion of $V_m$ in Step 2 is only a leading-order approximation, and the true inversion curve bends over at higher pressure. The qualitative conclusion is nonetheless right: nitrogen at room temperature (298 K) is far below both estimates of $T_i$ , so nitrogen cools on throttling at room temperature — consistent with everyday cryogenic practice.

Example 2 — Designing a Linde liquefaction step: how much precooling is needed for hydrogen? Hydrogen has $a = 0.2476\ \text{L}^2\,\text{atm}\,\text{mol}^{-2}$ and $b = 0.02661\ \text{L mol}^{-1}$ . Estimate its inversion temperature:

T_i = \frac{2(0.2476)}{(0.08206)(0.02661)} = \frac{0.4952}{2.183\times10^{-3}} = 227\ \text{K}

(The experimental value is about 202 K — again the van der Waals estimate overshoots, but by a smaller margin here.) Suppose a Linde plant feed enters at 298 K (room temperature, easily achieved with ordinary refrigeration) and 150 atm, then throttles to 1 atm. Since $298\ \text{K} > T_i \approx 202\text{–}227\ \text{K}$ , this gas is on the wrong side of the inversion curve: throttling would warm it slightly ( $\mu_{JT}<0$ ), not cool it — hydrogen famously cannot be liquefied by naive Joule-Thomson throttling from room temperature. To fix this, the gas must first be precooled below $T_i$ . Using liquid nitrogen (boiling point 77 K under 1 atm) as a precoolant brings hydrogen to roughly 77–100 K, safely below its ~202 K inversion temperature. Only then does throttling produce net cooling, and repeated cycling (the Linde/Hampson cascade) can drive the gas to its 20 K boiling point for final liquefaction. Quantitative check using a typical low-temperature $\mu_{JT}$ for hydrogen near 100 K, roughly $+0.5$ to $+1.0\ \text{K atm}^{-1}$ (tabulated), a throttle from 150 atm to 1 atm ( $\Delta p = -149$ atm) gives:

\Delta T \approx \mu_{JT}\,\Delta p \approx (0.7\ \text{K atm}^{-1})(-149\ \text{atm}) \approx -104\ \text{K}

Starting from 100 K, a single throttle stage could in principle approach $\sim$ −4 K, which is unphysical as a one-shot estimate (the coefficient itself varies strongly along the path), but it confirms the right order of magnitude and direction: with precooling below $T_i$ , throttling delivers tens to over a hundred kelvin of cooling per stage, enough after a few cascaded stages to reach hydrogen's 20 K liquefaction point.

Limiting cases & common traps

Situation	Behavior	Why
Ideal gas	$\mu_{JT} = 0$ always	No intermolecular forces, $H$ depends only on $T$
$T \ll T_i$	$\mu_{JT} > 0$ , strong cooling	Attractive forces ( $a$ -term) dominate
$T \gg T_i$	$\mu_{JT} < 0$ , heating	Excluded volume ( $b$ -term) dominates
Very high $p$	Inversion curve turns back, cooling region shrinks	Gas is already dense; approximation in Step 2 breaks down
Helium, hydrogen at 298 K	Warm on throttling	Room temperature exceeds their (low) $T_i$

The chemist behind it

The Joule-Thomson effect is named for James Prescott Joule and William Thomson (Lord Kelvin), who in 1852 refined Joule's earlier free-expansion experiments into the porous-plug throttling experiment that could detect the small enthalpy-driven temperature changes real gases show. Their careful measurements on air quantified deviations from ideal-gas behavior years before a molecular picture of intermolecular attraction was available, and their apparatus directly inspired the industrial liquefaction cascades built by Carl von Linde and William Hampson in the 1890s.

Origins · 1 minWilliam Thomson, 1st Baron Kelvin and how this was discovered

William Thomson, 1st Baron Kelvin

1824–1907 · Co-discovered the throttling effect and established the absolute temperature scale used to define it

Ask the tutor: How does the full van der Waals inversion curve bend and eventually close at high pressure, and why does that set an upper pressure limit for cooling by throttling?

What is the difference between Joule expansion and Joule-Thomson expansion?

Joule (free) expansion is into a vacuum with no external pressure resisting the gas, and no work is done; for an ideal gas, temperature does not change, and even for real gases the effect is tiny and hard to measure. Joule-Thomson expansion is throttling through a porous plug or valve, at constant enthalpy, against a finite downstream pressure. The Joule-Thomson version reveals second-order derivatives of the equation of state cleanly, which is why Joule and Thomson redesigned the experiment this way in 1852.

Can the Joule-Thomson coefficient be zero at a pressure other than the inversion pressure predicted by the simple formula?

Yes. The simple $T_i = 2a/(Rb)$ result is only the low-pressure limit of a full inversion curve, which is a closed loop in the $p$ – $T$ plane for the van der Waals equation. At each pressure there is generally both an upper and lower inversion temperature; the curve closes at some maximum pressure above which no inversion temperature exists at all, and throttling always heats the gas regardless of starting temperature.

Why does helium need such extreme precooling to liquefy by throttling?

Helium's inversion temperature is unusually low, only about 40–51 K depending on pressure, because its van der Waals $a$ constant is tiny — helium atoms interact only weakly (small, tightly-bound, low-polarizability atoms). Room-temperature helium is far above this $T_i$ , so throttling would warm it. Liquefaction plants must precool helium gas with liquid hydrogen or liquid nitrogen stages before Joule-Thomson throttling can finally cool it to its 4.2 K boiling point.

Does a positive Joule-Thomson coefficient guarantee successful liquefaction?

No — it guarantees cooling on each throttle pass, but liquefaction also requires enough cumulative cooling to reach the boiling point, good heat exchanger design to recycle the cold returning gas, and management of the fraction of gas that liquefies versus recirculates. Real Linde cycles use counter-current heat exchangers so the cold, unliquefied gas returning from the throttle precools the incoming high-pressure feed, making each subsequent pass more effective.