The ring the drop leaves behind

A drop of coffee dries on a table and leaves a dark ring with a pale middle. Almost everything that dries out of a drop does the same — spilled wine, salt water, a spot of blood on a slide, the ink from a leaking pen — and almost nothing about the substance matters. What decides the pattern is that the edge of the drop cannot move.

The consequences follow from that curve and from one geometric fact, and neither of them mentions what is dissolved in the drop.

The edge that cannot move #

A drop resting on a surface meets it at a contact angle set by the three surface tensions — liquid–vapour, solid–liquid and solid–vapour — and in a perfectly ideal situation that angle is a material constant.

Real surfaces are not ideal. They are rough at the scale of micrometres and chemically patchy at the scale of nanometres, and both defects pin the contact line: to move outward it must climb over an obstacle, to move inward it must let go of one. The result is that the angle can take a range of values without the line moving at all.

So a drying drop does not shrink in the way a puddle does. Its radius stays fixed and its height falls, its contact angle sliding down through the hysteresis range. It shrinks vertically and not horizontally, and it does that for the great majority of its life.

It is worth noticing that pinning is a defect effect and that the phenomenon therefore depends on the surface being imperfect. The same curvature that pins the edge is what pulls a column of water up a narrow tube, where the surface is not imperfect at all. On an atomically smooth, chemically uniform substrate the contact line would slide freely, the drop would retreat as it dried, and the deposit would be concentrated at the centre rather than the rim. The ring is a signature of ordinary surfaces, and the reason it is so nearly universal is that ordinary surfaces are what everything dries on. That is an unusual position for a physical effect to be in: the idealised case is the exception, and the messy case is the rule.

Why the rim dries fastest #

Evaporation from a drop into still air is limited by how fast the vapour diffuses away, not by how fast molecules leave the liquid. The vapour concentration is saturated at the surface and falls to the ambient value far away, and in between it satisfies Laplace’s equation.

Evaporation into still air is the steady version of a diffusion problem. The vapour concentration settles into the shape with no net accumulation anywhere — a solution of Laplace’s equation with the drop’s own surface as a boundary — and the rate at which the drop loses liquid at each point is the gradient of that solution there. Gradients crowd at edges, which is why a flat drop with a pinned contact line loses liquid fastest at its rim and not at its thickest part.

That is the same equation, with the same boundary condition, as the electrostatic potential around a conductor held at a fixed voltage — so the evaporation flux from a drop is the electric field at the surface of a charged lens of the same shape. And the field at the edge of a conductor with a sharp edge diverges, which is the reason lightning conductors are pointed.

The exact result is a power law in the distance from the contact line:

$$J(r) \propto \left(1 - \frac{r^2}{R^2}\right)^{-\lambda}, \qquad \lambda = \frac{\pi - 2\theta}{2\pi - 2\theta},$$

with $\theta$ the contact angle. For a flat drop, $\theta \to 0$ and $\lambda \to \tfrac12$: the flux goes as the inverse square root of the distance from the rim, which is integrable — the total evaporation is finite — but unbounded at the edge itself.

The sign of $\lambda$ is what makes the essay. It is positive for $\theta < 90°$, zero at exactly $90°$, and negative above it, so a drop that beads up rather than spreading dries slowest at its rim.

The electrostatic analogy is worth taking seriously rather than treating as a mnemonic, because it does real work. The whole problem of computing the evaporation from a drop of a given shape is the problem of computing the capacitance and surface-charge distribution of a conductor of that shape, and the latter has been studied for two centuries. The total evaporation rate of a drop is proportional to its “capacitance” in exactly this sense, which is why a flat drop of a given volume dries faster than a beaded one — it has more capacitance for the same volume, in the same way a flattened conductor holds more charge at a given voltage than a compact one.

The flow that has to exist #

Now put the two facts together. The drop loses most of its liquid at the rim. The rim cannot move inward. Therefore liquid must flow outward through the interior to replace what is lost — and the flow is not a detail of the drying but a requirement of mass conservation with a fixed boundary.

Its size follows from bookkeeping alone. If the drop’s shape stays a spherical cap and the radius is fixed, the height at each radius falls at a rate the cap’s geometry dictates, while the evaporation at each radius is the singular profile above. The difference between the two is the divergence of the radial flow, so the depth-averaged radial velocity at any radius is fixed by an integral of the mismatch — no dynamics, no viscosity, no pressure required.

What keeps the drop a spherical cap while all this happens is the pressure jump across its curved surface. Any local departure from uniform curvature produces a pressure difference that drives liquid back until the curvature is uniform again, and the restoring is fast compared with the drying — a drop reshapes itself in milliseconds and dries in minutes. So the shape is a constraint on the flow rather than a consequence of it, which is what lets the argument be made in that order.

The magnitude is not small. For a millimetre drop drying over ten minutes the depth-averaged outward speed near the rim is of the order of a micrometre per second, rising steeply as the contact line is approached and as the drop thins — and near the very end, when the cap has become a film a micrometre thick, the same volumetric flux through a much smaller cross-section makes the velocity rise sharply. The last few per cent of the drying carries a disproportionate share of the material to the edge, which is why rings are usually sharper than the smooth outward drift alone would predict.

Everything suspended in the liquid goes where the liquid goes. The particles are carried outward, arrive at the contact line and are stranded there as the liquid they arrived in evaporates, and the deposit accumulates into a ring whose width is set by how much material there is and how big the particles are.

Diffusion works against the ring, spreading particles back toward the centre, and it loses. For a micrometre particle in a millimetre drop drying over minutes the outward drift covers the radius many times over before diffusion covers it once — a competition decided by the ratio of two timescales, and for anything large enough to see under a microscope the flow wins comfortably. Shrink the particle to tens of nanometres and the ratio turns over, which is why the deposit from a nanoparticle suspension can be uniform where the same experiment with pollen is not.

The number that decides whether a ring forms at all #

The competition between the outward drift and diffusion deserves a number rather than an assertion, because the number is what separates the cases where a ring appears from the cases where it does not.

The comparison is a ratio of two rates: how fast the flow carries something across the drop, against how fast that thing diffuses back. For a particle of radius $a$ the diffusion coefficient is $k_BT/6\pi\eta a$, so the ratio is

$$\mathrm{Pe} = \frac{vR}{D} = \frac{6\pi\eta a\,vR}{k_BT},$$

and the particle size is in the numerator. A micrometre particle in a millimetre drop, carried at a micrometre a second, gives a ratio of a couple of thousand: the flow wins by three orders of magnitude and every particle ends up at the rim.

Now do the same for something dissolved rather than suspended. A small molecule has a diffusion coefficient a thousand times larger, of order $10^{-9}$ square metres per second, and the same arithmetic gives a ratio of about one. Transport and diffusion are then comparable, the outward drift is largely undone as it happens, and the solute is deposited across the whole drop rather than at its edge.

Which explains something that would otherwise look like an exception. A drop of coffee leaves a ring because coffee is a suspension of particles; a drop of salt water leaves a fairly even film with crystals scattered through it, and a drop of sugar solution leaves a glassy disc. The mechanism has not changed at all — the flow is identical — and what differs is whether the cargo can swim back against it.

It also says which way to move to suppress a ring without touching the flow. Smaller particles diffuse faster and ring less; a drop dried faster has a larger $v$ and rings more; and a drop of the same material made ten times smaller has a ratio a hundred times smaller, since both $v$ and $R$ fall with it. Rings are a phenomenon of millimetre drops and visible particles, which is exactly where they were first noticed.

What reverses it #

The one reliable way to destroy the ring is to make the surface tension vary with position, because a surface tension that varies produces a stress along the surface.

The tears of wine are the same mechanism running in the other geometry. Alcohol evaporates faster than water from the thin film that capillarity draws up the side of a glass, so the film’s surface tension rises, and the resulting gradient pulls liquid up from the bulk until a drop forms and runs back down. Nothing in that account is about the glass or about wine specifically — it needs only a mixture whose components evaporate at different rates and whose surface tensions differ, which is why the same tears form with any spirit and none with water.

Whether the recirculation wins depends on details that are genuinely delicate — how volatile each component is, how the drop cools, whether a surfactant adsorbs quickly enough to keep the surface tension uniform. That is why the practical rule “add a surfactant and the ring goes away” works sometimes and not always, and why controlling deposition from a drying drop is a manufacturing problem rather than a settled recipe.

There is a second, quite different reversal worth recording because it says something about what the ring really needs. Suspend particles that are not spherical — ellipsoids rather than spheres — and the ring largely disappears even with the outward flow unchanged. The particles still arrive at the rim; what changes is what happens when they get there. Elongated particles deform the free surface as they approach it and interact strongly with one another through that deformation, so instead of packing tightly at the contact line they jam into a loose open network and are left distributed across the whole drop. The transport is the same and the deposition is not, which separates two things the phrase “the coffee-ring effect” usually runs together.

Where it is a defect, and where it is a tool #

The pattern was a curiosity until it became a manufacturing problem, and the problem is what made it a subject.

Printing anything functional — a conducting track, a transistor’s channel, a coloured filter, a row of biological probes — increasingly means depositing a drop of a suspension and letting it dry. What the process wants is a uniform disc of material. What it gets, by default, is a ring: most of the solids at the perimeter, a depleted middle, and a feature whose electrical or optical properties are wrong everywhere. A printed silver track that is thick at its edges and thin along its centre conducts worse than either would suggest, and a printed pixel with a bright rim is a visible defect.

So a great deal of effort goes into the reversals above — surfactants to generate a recirculation, mixed solvents to drive one deliberately, heated substrates to control where the drop is coolest, non-spherical particles to stop them packing at the line. Each of them is an attempt to defeat a boundary condition rather than to change a material, and the fact that none of them is a settled recipe is a fair measure of how robust the effect is.

Run deliberately, the same transport is useful. A drop of solution containing long molecules, dried on a surface they stick to, delivers them all to the contact line and stretches them out along the direction of the flow as it goes — which is a way of laying down a long polymer straight and taut when nothing else will. And a drop dried in a controlled retreat, where the contact line is allowed to depin in steps rather than staying put, deposits a series of concentric rings whose spacing is set by the stepping and which can be made regular enough to be a grating.

The two uses share the same reading of the phenomenon. What a drying drop does is take everything suspended in it and deliver it to a line, and whether that is a defect or a technique depends only on whether a line was wanted.

How long a drop takes #

One consequence of the electrostatic analogy is worth extracting, because it gives the drying time without any of the detail.

The total rate at which a drop loses liquid is proportional to the “capacitance” of its shape, which for a body of a given form scales with its linear size — so the volume lost per second goes as $R$, while the volume to be lost goes as $R^3$. The drying time therefore goes as $R^2$.

That is a strong dependence and it is easy to check. A drop twice as wide takes four times as long to dry, not twice; a mist of ten-micrometre droplets evaporates ten thousand times faster than a millimetre drop of the same liquid, which is a fraction of a second against an hour. It is the reason a spray dries almost instantly and a spill does not, and the same square law governs the evaporation of fuel droplets in an engine and of water droplets in a cloud.

It also explains why the ring is a slow-drying phenomenon in practice. The outward velocity near the rim scales in a way that keeps the total transport roughly fixed, but the competition with diffusion above depends on the velocity, and a drop forced to dry quickly rings more sharply than one left in still air. Anyone who has tried to suppress a ring by drying faster has made it worse.

Where the model stops #

The flow calculation is kinematic and stops there. Everything above follows from mass conservation, a fixed radius and a spherical cap, and computes a depth-averaged velocity. It does not compute the flow’s profile across the depth, its stresses, or anything that requires solving the equations of motion — those are questions about flow, and flow belongs to the collection that owns it. Nothing here contains a drag, a Reynolds number or a flow regime.

The contact line is assumed pinned throughout. Real drops depin near the end, when the contact angle falls below the receding value, and the last stage of drying is a shrinking drop that deposits a smaller inner ring or a spot. Many real stains have both features, and the pinned model describes the first and larger of them.

Gravity is neglected. That requires the drop to be smaller than the capillary length — 2.7 mm for water — above which it flattens into a puddle with a nearly flat top and thin edges, and the spherical-cap geometry fails.

Only one component is assumed to evaporate, and at a rate independent of what is dissolved. A drying drop concentrates its solute as it loses liquid, and by the end the concentration near the rim can be high enough to change the liquid’s own properties — its viscosity, its vapour pressure and its surface tension all move. The equations here take all three as constants, which is right early in the drying and increasingly wrong late in it, when most of the deposition happens.

And the evaporation is assumed diffusion-limited into still air. A breath of air across the drop replaces diffusion with convection, changes the flux profile completely, and can move the fastest-evaporating region to the upwind side. Every result here is for a drop drying in a room where nothing is moving.

What the pictures cannot show #

The hero figure draws a flux profile, and the flux is what nobody can see. What is visible is the deposit, which is the time-integral of a transport of particles by a flow that is set by the flux — three steps removed from the drawing. A picture of the ring itself would be a photograph rather than a computed figure, and would show a pattern without showing anything about why it is there.

Nor can any figure here show the drop’s height falling, which is the other half of the geometry. The flux profile is drawn against radius at one instant, and the drying is a process in which the cap flattens continuously while its radius does not change. What makes the ring is the whole history, and every figure here is a snapshot of a rate.

Where the ladder goes next #

Surface tension has appeared on this ladder five times now — as a skin that is not a skin, as the pressure inside a bubble, as the instability that breaks a thread, as a field with a gradient, and now as a boundary that will not move. The rungs above it are surface rheology proper, where the interface has a viscosity of its own; the coupling between Marangoni and buoyant convection in a layer thick enough to have both; and evaporation from a mixture, where the composition of the surface changes as it dries and the tension follows it.

The habit worth carrying away is about where a pattern comes from. A boundary condition can produce structure that no property of the material would predict, and the coffee ring is the cleanest case: the same ring appears for particles of any material, any colour and almost any size, because none of those enters the argument. What produced it was an edge that could not retreat.