Physical Chemistry
Joule-Thomson Effect
Real-gas expansion through a porous plug at constant H — most gases cool below their inversion temperature, heat above
The Joule-Thomson effect is the temperature change a real gas undergoes when expanded irreversibly through a porous plug or throttle valve at constant enthalpy. The coefficient μ_JT = (∂T/∂P)_H is positive when expansion cools the gas and negative when it warms — the line μ_JT = 0 traces the inversion curve in (T, P) space. James Prescott Joule and William Thomson (Lord Kelvin) characterized it with porous-plug experiments between 1852 and 1862. Inversion temperatures span an enormous range: N₂ ≈ 621 K, O₂ ≈ 764 K, CO₂ ≈ 1500 K, but H₂ ≈ 205 K and He ≈ 40 K — so hydrogen and helium warm on JT expansion at room temperature, the precise reason early helium liquefiers had to pre-cool below 40 K before throttling. Carl von Linde commercialized JT cooling for air liquefaction in 1895.
- Coefficientμ_JT = (∂T/∂P)_H
- ProcessIsenthalpic (ΔH = 0)
- N₂ inversion T~621 K
- H₂ inversion T~205 K
- He inversion T~40 K
- DiscoveredJoule & Thomson 1852–62
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
Why the Joule-Thomson effect matters
- Industrial cryogenics. Carl von Linde's 1895 air liquefaction plant in Munich used Joule-Thomson cooling and counter-flow heat exchange to make liquid air at industrial scale. The same Linde-Hampson cycle is still the conceptual basis of every nitrogen, oxygen, and argon liquefier producing 100,000+ tons per year of cryogenic gases worldwide.
- Helium and hydrogen liquefaction. Because T_inv for helium is ≈ 40 K and for hydrogen ≈ 205 K, JT cooling alone won't reach their boiling points starting from room temperature. Modern He liquefiers use a turbine expansion (Claude cycle) to first cool to below 40 K, then JT-throttle to 4.2 K. This bootstrap mechanism is exactly how Heike Kamerlingh Onnes liquefied helium in 1908.
- Refrigeration cycle design. Vapor-compression cycles rely on JT throttling at the expansion valve. Refrigerants like R-134a, ammonia, and CO₂ all have positive μ_JT in the relevant operating range. The cooling capacity per kg of refrigerant is set by the latent heat at evaporator pressure plus the JT temperature drop at the throttle.
- Natural gas processing. JT valves in pipeline pressure-reduction stations cool natural gas, condensing heavier hydrocarbons (NGLs — natural gas liquids) for separation and sale. Without controlled JT cooling, hydrate formation and pipeline freeze-up are major operational hazards.
- CO₂ fire extinguishers and dry-ice production. CO₂ has T_inv ≈ 1500 K and μ_JT ≈ 1.1 K/atm at room temperature — large enough that releasing pressurized liquid CO₂ from 60 atm produces a stream cold enough (-78°C) to snow solid dry ice directly out of the nozzle.
- Test of real-gas thermodynamics. The 19th-century porous-plug experiments by Joule and Kelvin gave the first quantitative evidence that real gases deviate from PV = nRT. The fact that hydrogen warmed where nitrogen cooled was one of the founding experimental clues for the kinetic theory of gases.
- Atmospheric and geophysical phenomena. Adiabatic expansion of rising air parcels is closer to isentropic than isenthalpic, but JT effects matter at high altitude where compressibility deviates from ideal. Volcanic gas decompression and geyser steam emission also involve transient JT temperature changes.
Common misconceptions
- JT expansion always cools. Only below the inversion temperature. Above T_inv the same throttling warms the gas. Hydrogen at 300 K is famously warmed by JT expansion — exploited as a precaution against liquid-hydrogen leaks (the warming reduces ignition risk in the immediate vicinity, though embrittlement of metals remains a hazard).
- JT expansion is reversible. No — throttling through a porous plug or valve is highly irreversible: entropy is generated. The constant H is a property of the irreversible isolated process between two thermally-insulated reservoirs, not a statement that the process is quasi-static.
- Ideal gases also exhibit JT cooling. They do not. For an ideal gas, H depends only on T, so (∂T/∂P)_H = 0 — no temperature change with pressure at constant H. JT effect is purely a real-gas, intermolecular-interaction phenomenon.
- Inversion temperature is a single constant per gas. The inversion curve in (T, P) space is a closed loop. There's a low-T branch and a high-T branch at any given pressure; outside the loop, μ_JT < 0 (warming). The "inversion temperature" usually quoted is the maximum at zero pressure (≈ 866 K for N₂ in vdW, 621 K experimentally).
- JT is the same as adiabatic expansion in a turbine. Turbine (Claude) expansion is isentropic with shaft work extracted, cooling much more efficiently than JT throttling and giving cooling for any gas regardless of inversion temperature. The trade-off is mechanical complexity. Modern liquefiers combine both: turbine for the bulk of cooling, JT for the final phase change.
- JT is the same as Joule expansion (free expansion). Joule expansion is at constant U into a vacuum; JT is at constant H through a plug into a finite back-pressure region. The two coefficients (∂T/∂V)_U and (∂T/∂P)_H are different physical quantities, though both vanish for an ideal gas.
Derivation and the inversion curve
Consider gas at (P₁, T₁) flowing steadily through a porous plug into a region at lower pressure P₂. The walls are thermally insulating, so heat exchange is zero, and no shaft work is done. The work done by the upstream gas on the plug is P₁V₁ (per mole), while the downstream gas does work P₂V₂. Conservation of energy gives U₂ − U₁ = P₁V₁ − P₂V₂, equivalently H₂ = H₁. The process is isenthalpic. The temperature change ΔT for an isenthalpic pressure drop dP is set by the Joule-Thomson coefficient μ_JT ≡ (∂T/∂P)_H. Using a Maxwell relation, μ_JT = (1/C_p) [T(∂V/∂T)_P − V] = (V/C_p)(αT − 1), where α is the volume expansion coefficient. The inversion temperature is the locus where αT = 1.
For an ideal gas, V = nRT/P, so (∂V/∂T)_P = nR/P = V/T, and μ_JT = 0 — temperature is unchanged by throttling. For a real gas, attractive forces dominate at lower temperatures and (∂V/∂T)_P at constant P slightly exceeds V/T, giving μ_JT > 0 and cooling. At very high temperatures repulsive interactions dominate and μ_JT < 0, giving warming. From the van der Waals equation an approximate expression at low pressure is μ_JT = (1/C_p)[2a/(RT) − b], so the inversion temperature is T_inv = 2a/(Rb). This gives 866 K for N₂ (experiment 621 K), 224 K for H₂ (experiment 205 K), and 35 K for He (experiment 40 K) — qualitatively correct, off by 30% in absolute value.
The full inversion curve in the (T, P) plane is a closed loop. For nitrogen the curve passes through about T = 621 K at zero pressure, rises to a maximum near T = 600 K at P ≈ 350 bar, and closes at T ≈ 110 K at low pressure on the cold side. Inside the loop μ_JT > 0 (cooling); outside, μ_JT < 0 (warming). Refrigeration designers must operate inside this loop. The Linde process for air takes advantage of the fact that nitrogen at 300 K and 200 atm is well within the inversion loop, with μ_JT ≈ 0.2 K/atm — a single throttle drops T by tens of kelvins per pass.
Inversion temperatures and JT coefficients at low pressure
| Gas | T_inv (K, low P) | μ_JT at 300 K, 1 atm (K/atm) | Behavior at room T | Notes |
|---|---|---|---|---|
| Carbon dioxide (CO₂) | ~1500 | +1.11 | Strongly cools | Used in dry-ice production |
| Oxygen (O₂) | ~764 | +0.31 | Cools | Linde air-separation |
| Argon (Ar) | ~723 | +0.43 | Cools | Inert cryogen |
| Nitrogen (N₂) | ~621 | +0.27 | Cools | Linde process workhorse |
| Methane (CH₄) | ~939 | +0.44 | Cools | LNG production |
| Air (mixture) | ~603 | +0.20 | Cools | Linde 1895 commercialization |
| Hydrogen (H₂) | ~205 | −0.03 | Warms slightly | Must be pre-cooled below 205 K |
| Helium-4 (He) | ~40 | −0.06 | Warms | Must be pre-cooled below 40 K (Onnes 1908) |
| Helium-3 (³He) | ~36 | −0.07 | Warms | Used in dilution refrigeration |
Throttling vs alternative refrigeration / liquefaction methods
| Method | Process type | Cooling per pass (typical) | Pros | Cons |
|---|---|---|---|---|
| Joule-Thomson throttle (Linde) | Isenthalpic, irreversible | 10–50 K for air at 200 atm | No moving parts at low T; simple | Won't cool He, H₂ at room T; modest efficiency |
| Adiabatic turbine (Claude) | Isentropic, near-reversible | 50–100 K per stage | Cools any gas, high efficiency | Mechanical complexity at low T |
| Free expansion (Joule) | Isochoric, irreversible | ~0 (ideal); tiny for real | Conceptually simple | No useful cooling in practice |
| Stirling cycle cooler | Reciprocating regenerator | Continuous; reaches 40 K | Reaches cryogenic without LN₂; closed cycle | Vibration; moving seals at low T |
| Pulse-tube cryocooler | Acoustic, no cold moving parts | Continuous; reaches 4 K | Long-lived, low vibration | Lower efficiency than Stirling |
| Adiabatic demagnetization | Magnetic refrigeration | To < 1 mK | Reaches sub-K temperatures | Single-shot; needs pre-cooling to ~1 K |
| ³He–⁴He dilution | Mixing entropy | Continuous to ~10 mK | Reaches mK continuously | Very expensive ³He |
Applications
- Cryogenic Linde process for air liquefaction. Carl von Linde's 1895 plant used Joule-Thomson cooling and counter-flow heat exchange to liquefy air commercially. Modern derivative plants (Air Liquide, Linde, Air Products) produce O₂, N₂, Ar at the multi-million-tonne scale for steelmaking, semiconductor fabs, and medical use.
- Lord Kelvin and James Joule porous-plug experiments (1852–1862). The original quantitative measurements of μ_JT for various gases established that real gases differ from ideal in measurable, reproducible ways. The data fed directly into the kinetic theory developed by Maxwell and Boltzmann in the same decade.
- Helium liquefaction (Heike Kamerlingh Onnes, 1908). Because helium has T_inv ≈ 40 K, Onnes pre-cooled the gas with liquid hydrogen (which itself had to be pre-cooled with liquid air) before JT throttling could finally liquefy it at 4.2 K. This experiment opened the door to superconductivity, discovered three years later in mercury.
- Natural gas processing and LNG. JT valves cool compressed natural gas to condense heavier components (ethane, propane, butane) for separation. Liquefied natural gas (LNG) at −162°C uses cascade cooling with multiple refrigerants, finishing with JT throttling to atmospheric pressure.
- Cryogenic surgery and medical refrigeration. Cryosurgical probes use JT cooling of high-pressure argon or nitrogen to reach −150°C at the probe tip, freezing tumor tissue. MRI scanner liquid-helium top-up, biological sample storage, and vaccine cold chains all rely on JT-based cryogenic infrastructure.
Frequently asked questions
Why is the expansion isenthalpic?
In the porous-plug experiment, gas at high pressure P₁ is forced through a fibrous plug into a region at lower pressure P₂. The walls are thermally insulated so q = 0, and no shaft work is done. The work done on the gas pushed in is P₁V₁; the gas pushed out does work P₂V₂. By the first law ΔU = −Δ(PV), so U₂ + P₂V₂ = U₁ + P₁V₁ — that is, H₂ = H₁. Throttling is isenthalpic. The temperature change is therefore (∂T/∂P)_H along this constant-H path, which is the Joule-Thomson coefficient. For an ideal gas H depends only on T, so dT = 0; for a real gas H depends on P as well, and dT can be either positive or negative.
Why do hydrogen and helium warm on JT expansion at room temperature?
The JT coefficient sign depends on whether attractive or repulsive intermolecular interactions dominate at the working temperature. Below the inversion temperature, attraction dominates: as molecules separate during expansion, work is done against these attractive forces, drawing energy from the gas's kinetic store and cooling it. Above the inversion temperature, the relevant interactions become effectively repulsive (in the sense that ∂(PV/RT)/∂T at constant P > 0), and expansion warms. Hydrogen has T_inv ≈ 205 K and helium ≈ 40 K — both well below room temperature (300 K), so JT expansion at 300 K warms them. To liquefy hydrogen Carl von Linde first cooled it below 205 K with liquid air, then JT-throttled. To liquefy helium Heike Kamerlingh Onnes (1908) first cooled below 40 K with liquid hydrogen, then JT-throttled.
What is the Linde process?
The Linde cycle (Carl von Linde, 1895) liquefies air using only Joule-Thomson cooling and counter-flow heat exchange. Compressed air at 50–200 atm is precooled in a heat exchanger by the cold low-pressure return stream, then throttled through a JT valve. A small fraction of the cooled gas liquefies; the rest returns through the heat exchanger, cooling the next batch of incoming high-pressure gas. Continuous operation builds up liquid air at the bottom of the separator. The cycle gives about 70 K of cooling per pass for nitrogen near its inversion temperature, and bootstraps from room temperature to the boiling point of liquid air (77 K) in a few hours. Modern industrial plants combine Linde with Claude expansion (turbine doing work) for efficiency.
How is the JT coefficient related to van der Waals constants?
From the van der Waals equation, an approximate expression at low pressure is μ_JT = (1/C_p)[2a/(RT) − b], where a and b are the van der Waals constants and C_p is the molar heat capacity at constant pressure. Setting μ_JT = 0 gives the inversion temperature T_inv = 2a/(Rb), independent of pressure to leading order. For nitrogen with a = 1.39 L²·atm/mol² and b = 0.0391 L/mol, T_inv ≈ 866 K (low-pressure limit; experiment gives ≈ 621 K because van der Waals overestimates a/b). For helium with a = 0.034 and b = 0.0237, T_inv ≈ 35 K — close to the measured 40 K. The 2a/(Rb) prediction works qualitatively for all gases and explains the wildly different inversion temperatures of light versus heavy molecules.
Why does CO₂ produce dry ice when discharged from a fire extinguisher?
Inside a CO₂ fire extinguisher, CO₂ is stored at about 60 atm at room temperature, partially as liquid and partially as vapor in equilibrium. When discharged, two effects combine. First, the liquid flashes to vapor through the nozzle — a phase change that absorbs latent heat. Second, the cold vapor undergoes Joule-Thomson cooling further: CO₂ has an exceptionally high inversion temperature (≈ 1500 K), so JT cooling is large at room temperature, with μ_JT ≈ 1.1 K/atm. Combined, the discharge stream cools to about −78°C (the sublimation point of solid CO₂ at 1 atm), and a substantial fraction freezes directly into solid CO₂ snow. Industrial dry-ice production uses high-pressure liquid CO₂ throttled through nozzles for the same reason.
What's the difference between Joule expansion and Joule-Thomson expansion?
Joule expansion (free expansion into a vacuum) is at constant internal energy U: no work is done because the gas expands against zero external pressure. The Joule coefficient (∂T/∂V)_U is zero for an ideal gas (Joule's 1845 experiment) and very small for real gases — the famous failed experiment. Joule-Thomson is constant enthalpy H, with the gas pushing through a plug into a finite-pressure region. Both probe departures from ideal behavior, but JT couples to dH/dP rather than dU/dV. Crucially, the JT effect is much larger than Joule's effect for typical real gases (∂T/∂P)_H ≈ 1 K/atm versus (∂T/∂V)_U being too small to measure with mid-19th-century calorimeters — which is why Joule-Thomson became the basis of practical refrigeration while Joule expansion remained a thermodynamic curiosity.