The Mentos-and-Soda Geyser: Nucleation on Demand

It was never a chemical reaction

The single most important fact about this demonstration is what is not happening: no new substance is being made. People reach for chemistry — “the acid in the soda reacts with the mint,” “the candy dissolves and releases gas” — but careful studies (most famously Tonya Coffey’s 2008 analysis in the American Journal of Physics) show the mechanism is purely physical. Carbonated soda is water holding a large amount of dissolved carbon dioxide, CO₂(aq). The gas that erupts is exactly that CO₂ coming back out of solution. The reversible process is just a phase change:

CO₂(aq) ⇌ CO₂(g)

There is a slow, minor acid–base equilibrium in any carbonated water — a small fraction of CO₂ forms carbonic acid, H₂CO₃ ⇌ H⁺ + HCO₃⁻ — but that chemistry is not what drives the geyser and it is not accelerated by the candy. The Mentos does not generate gas. It releases gas that was already there, waiting. Understanding the geyser therefore means understanding why the CO₂ was trapped in the first place, and what the candy does to set it free all at once.

Supersaturation: energy held in a metastable solution

Soda is bottled by forcing CO₂ into the liquid under pressure. How much dissolves is set by Henry’s law: the concentration of dissolved gas is proportional to its partial pressure above the liquid, [CO₂] = k_H·p_CO₂. For CO₂ in water at 25°C, k_H ≈ 0.034 mol·L⁻¹·atm⁻¹. Bottlers carbonate to roughly 3–4 “volumes” of CO₂, which corresponds to a head pressure of about 2–4 atm and a dissolved concentration near 0.1–0.15 mol/L (about 5–8 grams of CO₂ per liter).

The instant you crack the cap, the pressure above the liquid falls to atmospheric, where the partial pressure of CO₂ is only about 4×10⁻⁴ atm. Henry’s law now says the liquid should hold only ~1×10⁻⁵ mol/L — roughly ten-thousand times less. The drink is wildly supersaturated: thermodynamically it “wants” to dump almost all its CO₂, yet an open, undisturbed soda fizzes only lazily for an hour. That gap between what thermodynamics demands and what actually happens is a kinetic problem. The energy to make a fountain is already stored in the bottle; the question is purely how fast the system can find a path to release it.

The nucleation barrier: why a new bubble is so expensive

To leave solution, CO₂ molecules must gather into a bubble — and creating a bubble means creating new gas–liquid surface, which costs energy set by the liquid’s surface tension γ (about 72 mN/m for water). A tiny embryonic bubble is dominated by this surface cost; only once it grows past a critical radius does the volume’s pressure gain win. Classical nucleation theory gives the height of this energy hill for a bubble forming in the bulk liquid (homogeneous nucleation):

ΔG* = 16πγ³ / (3ΔP²)

where ΔP is the supersaturation pressure driving the bubble outward. Two features matter enormously. First, a small bubble has a huge internal Laplace pressure, ΔP = 2γ/r; a 10-nanometer embryo would need hundreds of atmospheres inside to survive, far more than the few atmospheres the soda can supply. Second, the barrier scales as γ³, and the nucleation rate depends on it exponentially, J ∝ exp(−ΔG*/k_BT). Plug in real numbers and the homogeneous rate is effectively zero — a clean glass of soda would take an impractically long time to spontaneously boil off its gas. The CO₂ is trapped not by a lid but by a wall of surface energy.

Heterogeneous nucleation: how a rough surface cheats the barrier

Real sodas fizz because bubbles almost never form in the bulk — they form on surfaces. A microscopic crack, scratch, dust mote, or cloth fiber traps a pocket of gas. That pocket already is a curved gas–liquid interface with a large radius, so the system skips the punishing tiny-embryo stage entirely. This is heterogeneous nucleation, and it slashes the barrier by a geometric factor f(θ) that depends on the contact angle θ of the liquid on the solid:

ΔG*_het = f(θ) · ΔG*_hom,   f(θ) = (2 + cosθ)(1 − cosθ)² / 4

For a poorly wetted, rough surface f(θ) can drop to a small fraction — and because the barrier is inside an exponential, even a modest reduction multiplies the nucleation rate by many orders of magnitude. This is the entire secret of the Mentos. Its candy shell looks smooth, but under a scanning electron microscope the surface is a cratered, pitted moonscape carrying thousands of gas-trapping crevices per candy. Drop one in and you do not add one nucleation site — you add a whole galaxy of them at once. Each crevice launches a continuous stream of bubbles, and the total exsolution rate skyrockets. Sand, rock salt, or a scratched surface do the same thing, just less spectacularly; the Mentos is simply an unusually rough, dense, conveniently sized delivery system for nucleation sites.

The surfactant assist: lowering γ

Rough texture explains most of the effect, but the Mentos has a second trick. Its coating contains gum arabic (a natural surfactant/emulsifier) along with sugars, and diet formulations of the soda contain aspartame and preservatives that also behave as surfactants. Surfactants lower the liquid’s surface tension γ — and remember the barrier scales as γ³. Coffey’s measurements found that Mentos and aspartame meaningfully reduced the drink’s surface tension. The leverage is real: cutting γ by just 20% (say from 72 to ~58 mN/m) shrinks ΔG* by roughly (0.80)³ ≈ 0.51 — about half. Because nucleation rate rides on exp(−ΔG*/k_BT), halving the barrier is not a 2× speed-up but a colossal one. Lower surface tension also lets bubbles grow, detach, and coalesce more easily instead of clinging to the candy. So the Mentos attacks the problem from two directions at once: it multiplies the number of nucleation sites through its roughness, and it lowers the cost of every bubble through its surfactant coating.

From fizz to fountain: buoyancy, entrainment, and the neck

A high nucleation rate alone would just make the soda foam. Turning foam into a geyser is a fluid-dynamics finish. Each nucleated bubble grows as more dissolved CO₂ diffuses in, then rises under buoyancy. A dense Mentos sinks quickly, so it seeds bubbles all the way down the column; those bubbles then have the full height of the bottle to grow, rise, and merge. As they ascend they entrain the surrounding liquid, dragging water upward with them — a two-phase plume that is far less dense than the soda around it. All of this is funneled toward the one narrow exit: the bottle neck. Rapidly expanding foam, forced through a small opening, converts into a fast vertical jet, exactly as a nozzle converts pressure into velocity. The result is a column that erupts in about 1–3 seconds and can reach several meters. Practiced demonstrators (the “EepyBird” duo made hundreds of synchronized Diet Coke fountains famous) use the geometry deliberately — drop tubes that release many candies simultaneously to maximize the instantaneous bubble load through the neck.

Tuning the geyser — and the myths

Because the mechanism is physical, you can predict how to make it bigger by changing the physics:

• Temperature. Gas solubility falls as the liquid warms (dissolving CO₂ is exothermic, ΔH_sol ≈ −19 kJ/mol, so by Le Chatelier warmth drives it out). Warm soda is more supersaturated and erupts far more vigorously than chilled soda — one of the strongest levers in the whole demonstration.
• Diet vs. regular. Diet cola is the classic choice. Partly it is practical — no sticky sugar residue — but there is real chemistry too: aspartame and benzoate lower surface tension, whereas sugar raises it and increases viscosity, so diet soda gives a cleaner, slightly taller fountain.
• The candy. Anything rough works — salt, sand, a scratched marble — but a Mentos combines high roughness, useful density, and a surfactant coating in one object.

Now the myths. It is not an acid–base reaction — adding a base does little; the mint flavor is irrelevant (the pitted texture, not the taste, matters); and the candy is not “eaten” by the soda. As a safety note, the hazard is pressure and mess, not toxicity: this is a controlled outdoor demonstration, and the value in understanding it is precisely knowing that a shaken or warm carbonated bottle stores real mechanical energy that any nucleation trigger can release suddenly.

Why a Mentos beats a smooth object dropped into the same soda — the candy is optimized for nucleation on almost every count

Property	Mentos candy	Smooth object (marble, glass bead)
Surface texture	Extremely rough, pitted (SEM shows craters)	Polished, few defects
Nucleation-site density	Thousands of active crevices	Only a handful of scratches
Surface coating	Gum arabic + sugars lower surface tension	None; no surfactant
Density / behavior	Dense — sinks fast through the column	May sink but seeds only near bottom
Result	Massive, near-instant foam geyser	A slow trickle of bubbles at best