Leidenfrost Effect

The intuition — a drop on its own exhaust

Flick a few drops of water onto a pan and listen. On a merely hot pan (say 120–150 °C) each drop spreads, hisses, and vanishes in a second or two — that's ordinary boiling, with the water touching the metal and ripping heat out of it fast.

Now crank the pan up past about 193 °C and try again. The drops do the opposite of what you'd expect from "hotter pan, faster evaporation." They pull into tight glossy beads, hover, and go shooting across the surface like tiny hovercraft. A single drop can survive for a full minute, darting around long after a cooler pan would have flashed it to steam.

That reversal is the Leidenfrost effect. The bottom of the drop boils so violently that it generates a continuous cushion of steam, and the drop never quite touches the metal again. It floats on a film of its own vapor — and because vapor is a lousy heat conductor, that same film starves the drop of the heat it needs to finish evaporating. The drop is riding on, and insulated by, its own exhaust.

How the vapor film forms and holds up the drop

Three things happen at once, and they reinforce each other into a stable steady state:

1. Flash boiling at the base. The metal is hundreds of degrees above water's boiling point. The instant the drop's underside gets close, a sheet of vapor erupts off it.
2. That vapor has to escape sideways. The gap between drop and metal is only ~0.1 mm. Vapor generated underneath must squeeze out radially through this thin slot.
3. Forcing gas through a thin gap builds pressure. This is lubrication pressure — the same physics that lets an air bearing or an air-hockey table float a puck. The overpressure under the drop, integrated over its base area, produces an upward force that exactly balances the drop's weight.

It's a self-regulating loop. If the drop sinks a little, the gap narrows, vapor escapes harder, pressure rises, and it gets pushed back up. If it floats too high, the gap widens, pressure drops, and gravity pulls it back down. The drop settles at the height where vapor support equals gravity.

Because there is no solid contact, there is almost no friction — that's why the drop skitters. And because the film is vapor, not liquid, heat now has to conduct through gas to reach the drop, which is the slow part.

The governing physics

Heat conduction across the vapor film. The heat flux into the drop is set by Fourier's law across the thin gas gap of thickness d:

q = k_vapor · (T_surface − T_boil) / d   [W/m²]

With k_vapor ≈ 0.025 W/m·K, ΔT ≈ 200 °C, and d ≈ 1×10⁻⁴ m, that's roughly 5×10⁴ W/m² — large in absolute terms, but the key point is the 1/d dependence: the film throttles heat far more than direct liquid–metal contact would.

The film is the bottleneck. Compare conductivities:

k_water_vapor ≈ 0.025 W/m·K
k_liquid_water ≈ 0.6   W/m·K     (~24× higher)
k_stainless    ≈ 15    W/m·K     (~600× higher)
k_copper       ≈ 400   W/m·K

Swapping direct metal contact for a vapor gap cuts the effective heat transfer coefficient enormously. That's the whole effect in one line.

Levitation balance. The vapor pressure built up under the drop supports its weight:

∫ p_excess dA  =  m · g  =  ρ_liquid · V · g

The film thickness d that the drop self-selects scales with the imposed superheat ΔT and the drop size R. For small drops, lubrication theory gives a film thickness that grows weakly with superheat, roughly d ∝ (ΔT)^1/4 for a given drop — hotter surface, slightly thicker (and more insulating) cushion.

Evaporation rate and lifetime. The drop loses mass at a rate set by the heat that makes it through the film and the latent heat of vaporization L (for water, L ≈ 2.26×10⁶ J/kg):

dm/dt = − (q · A_contact) / L

Because q is small (thanks to the insulating film), dm/dt is small, and the lifetime is long. For larger drops, classic analysis (Biance et al., 2003) gives lifetimes that scale with drop radius and grow as the surface gets hotter — the opposite of the naive "hotter = faster" intuition.

The boiling curve — four regimes

The Leidenfrost effect is one corner of a bigger map. Plot heat flux versus surface superheat (how far above boiling the surface is) and you get the classic boiling curve, first mapped by Shiro Nukiyama in 1934:

Regime	Surface superheat (water)	What happens	Heat transfer
Natural convection	0–5 °C above 100 °C	No bubbles; heat carried by warm rising liquid	Modest
Nucleate boiling	~5–30 °C	Bubbles nucleate, detach, stir the liquid violently	Highest — peaks at the "critical heat flux"
Transition boiling	~30–120 °C	Unstable mix of contact and vapor patches; flux drops as temperature rises	Falling (unstable)
Film boiling (Leidenfrost)	> ~120 °C superheat	Continuous vapor film; drop levitates	Low again — the insulating minimum

The Leidenfrost point is the minimum of this curve — the temperature where heat transfer bottoms out because the vapor blanket becomes continuous and stable. Counter-intuitively, between nucleate boiling and film boiling, making the surface hotter makes it transfer less heat.

Drop lifetime by surface temperature

Approximate lifetimes for a single ~30 µL water drop (a few mm across) placed on a polished metal plate. These are order-of-magnitude — real values depend heavily on surface finish — but the shape is the point:

Plate temperature	Regime	Drop lifetime	Behavior
110 °C	Gentle boiling	~20–40 s	Spreads, simmers in place
140 °C	Nucleate boiling	~2–5 s	Violent hiss, spits, vanishes fast
170 °C	Transition	~1–2 s	Shortest life — most efficient cooling
200 °C	Leidenfrost onset	~30–60 s	Beads up, starts to glide
300 °C	Full film boiling	~60–120 s	Glossy, mobile, long-lived

Notice the U-shape: lifetime is shortest in transition boiling (around 170 °C) and gets longer as the plate gets hotter past the Leidenfrost point. The most "violent-looking" pan is actually the deadliest to the drop; the calm, levitating drop on the hottest pan lives longest.

Where it shows up — from chefs to reactors

• The cook's "dancing water" test. Chefs flick water on a skillet: if it beads and skates, the pan is past ~190 °C and ready to sear. Beading = Leidenfrost.
• Why you can (briefly) survive touching hot metal. A wet finger tapped on a 200 °C surface, or a hand dipped in liquid nitrogen, gets a flash vapor film that insulates skin for a fraction of a second. People walk briefly over hot coals on the same principle (sweaty feet plus low coal conductivity). It is not safe to rely on — timing is everything.
• Quenching and metallurgy. When you quench a hot steel part in water or oil, a vapor blanket forms first (film boiling), then collapses into nucleate boiling. The vapor-blanket stage cools slowly and unevenly, which can warp or crack parts — heat-treaters fight it with agitation and additives.
• The boiling crisis in power plants. In nuclear-reactor fuel channels and boilers, engineers run near the critical heat flux for maximum cooling. Cross it and the surface flips into film boiling — the "departure from nucleate boiling" (DNB). Cooling collapses, the cladding temperature can jump by hundreds of degrees in seconds, and the metal can fail. Avoiding DNB is a central reactor-safety constraint.
• Liquid nitrogen and cryogenics. LN₂ poured on a warm floor beads and rolls because room temperature is hundreds of degrees above its −196 °C boiling point.
• Self-propelling drops. Surfaces micro-machined with asymmetric ratchet teeth rectify the escaping vapor, pushing Leidenfrost drops in one direction — even uphill — a route to droplet transport and heat-engine concepts with no moving parts.

Conditions and how to defeat it

The Leidenfrost transition is sensitive to the surface, not just the temperature:

• Surface texture matters enormously. Highly textured, microstructured, or superhydrophilic surfaces can raise the Leidenfrost temperature far above 193 °C — pillars and channels pierce the vapor film and re-wet the surface. This is actively engineered for spray cooling of electronics and reactor surfaces, where you want to stay out of film boiling.
• Liquid properties. Lower boiling point, lower latent heat, or different surface tension shifts the Leidenfrost point. Ethanol, acetone, and liquid nitrogen all film-boil at their own characteristic surface temperatures.
• Drop size and arrival. A drop dropped from a height can briefly punch through the film and contact the surface before re-levitating; a gently deposited drop floats immediately.
• Contamination. Dissolved salts and surfactants change wetting and can suppress or trigger the effect, which is one reason published Leidenfrost temperatures for "water" scatter so widely.

Common misconceptions and edge cases

• "The drop touches the pan." In steady film boiling it doesn't — that's the whole point. There's a continuous ~0.1 mm vapor gap. (During the brief approach or after a hard impact it can momentarily contact.)
• "Hotter pan = faster evaporation." Only up to the transition regime. Past the Leidenfrost point, hotter means a thicker, more insulating vapor film and a longer-lived drop.
• "It's the same as a drop floating on air." It's floating on vapor it is actively generating, not still air. Cut off the boiling (e.g. let the drop shrink to nothing) and the support vanishes with it.
• "It happens right at 100 °C." No — it needs a large superheat, typically ~90+ °C above boiling for water (≈193 °C surface). At 100–150 °C you get ordinary boiling, not levitation.
• "It only happens with water." Any liquid far below the surface temperature film-boils — nitrogen, ethanol, even molten metals on hotter surfaces.
• "It makes hot surfaces cool things better." The opposite — film boiling is the worst cooling regime, which is exactly why the boiling crisis is dangerous in reactors and boilers.