Fluid dynamics
Marangoni Convection: Flow Driven by Surface-Tension Gradients
A tear of wine climbing the inside of a glass, a soap film that never quite drains, a laser-melted weld pool that digs a hole instead of spreading flat — all are the same physics: Marangoni convection, bulk fluid flow dragged along an interface by gradients in surface tension σ. Because a liquid surface pulls hardest where σ is largest, any variation in temperature or solute concentration along the interface exerts a tangential stress τ = ∇ₛσ that sets the fluid in motion, with no gravity, no density difference, and no pump required.
Formally, it is a purely interfacial instability: a stagnant layer heated from below (or with a volatile solute) becomes unstable once the destabilizing thermocapillary stress overwhelms viscous drag and thermal diffusion. The controlling parameter is the Marangoni number Ma; for a thin film the classic linear analysis gives a critical value Ma_c ≈ 79.6.
- RegimeInterface-driven flow; gravity-independent (dominates below ~1 mm / in microgravity)
- Key relationTangential stress τ = ∇ₛσ = (dσ/dT)∇ₛT; balances μ ∂u/∂z at the surface
- Control parameterMa = −(dσ/dT)ΔT·L / (μα); critical Ma_c ≈ 79.6 (thin film, adiabatic)
- Named for / datedCarlo Marangoni (1865, 1871); J. Thomson (1855, tears of wine); Pearson (1958, instability)
- Characteristic scaledσ/dT ≈ −0.15 mN·m⁻¹·K⁻¹ for water; effect wins over buoyancy for L ≲ ℓ = √(σ/ρg) ≈ 2.7 mm
- Matters forWelding/crystal growth, coffee-ring & inkjet drying, thin films, microgravity fluid physics
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.
What it is and why it matters
Marangoni convection is bulk flow set in motion by gradients in surface tension along a fluid interface. Surface tension σ is an energy per unit area; a gradient ∇ₛσ along the surface is mechanically a tangential force per unit area pulling fluid toward regions of higher σ. Because σ depends on temperature and on the concentration of surfactants or solutes, any non-uniform heating or composition creates a shear stress that viscosity transmits into the bulk, driving convection cells or coherent streaming.
Its importance is disproportionate to its everyday invisibility. On sub-millimetre scales — thin films, small droplets, foam lamellae — and in microgravity, buoyancy is negligible and Marangoni stresses become the dominant mechanism of convective transport. It governs weld-pool and float-zone crystal-growth defects, the coffee-ring and reverse-ring patterns of drying droplets, the drainage and stability of soap films and foams, boiling heat transfer, and the wobble of pinned menisci. It is the archetypal example of interfacial hydrodynamics coupling thermodynamics of surfaces to Navier–Stokes flow.
The mechanism, step by step
Start with a stationary liquid layer, free surface on top, warmer below. Now perturb the surface temperature: a spot that is slightly warmer has, for a normal liquid, lower surface tension (dσ/dT < 0). Its cooler neighbours, with higher σ, pull the surface fluid away from the warm spot. This outward surface flow must be fed from below, so warm fluid is drawn up from the interior to the hot spot — reinforcing the temperature perturbation. That is a positive feedback: an instability.
What opposes it? Two dissipative processes. Thermal diffusion (diffusivity α) erases the temperature bump before it can drive flow; viscosity μ resists the surface shear. The instability grows only when the destabilizing thermocapillary drive exceeds both. The balance is set at the interface by the boundary condition that the jump in tangential viscous stress equals the surface-tension gradient: μ(∂u/∂z) = ∂σ/∂x = (dσ/dT)(∂T/∂x). This one equation — a stress balance at the surface rather than a body force in the bulk — is what distinguishes Marangoni from Rayleigh–Bénard convection.
The key criterion, numbers, and scales
Non-dimensionalizing that stress balance yields the Marangoni number:
Ma = −(dσ/dT) · ΔT · L / (μ α)
the ratio of thermocapillary drive to the product of viscous and thermal diffusion (μ×α). For a thin layer with a non-deformable free surface, an adiabatic interface, and a rigid conducting bottom, the linear-stability calculation of J. R. A. Pearson (1958) gives the critical value Ma_c ≈ 79.6 at dimensionless wavenumber k ≈ 1.99. Above Ma_c, hexagonal convection cells appear. (With a constant-flux lower boundary the threshold drops to ≈ 48; with a deformable interface a separate long-wave mode enters.)
Numbers: for water dσ/dT ≈ −0.15 mN·m⁻¹·K⁻¹, μ ≈ 1 mPa·s, α ≈ 1.4×10⁻⁷ m²/s. Buoyancy competes via the dynamic Bond number Bo_d = Ra/Ma = ρgβL²/(−dσ/dT); Marangoni wins when L is below the capillary length ℓ = √(σ/ρg) ≈ 2.7 mm for water. The relevant ratio of momentum to thermal diffusion is the Prandtl number Pr = ν/α.
How it is realized, measured, and observed
The canonical laboratory system is a thin liquid layer (silicone oil, ~1 mm) uniformly heated from below in a shallow dish. As the temperature difference rises past Ma_c, the flat surface spontaneously breaks into a striking pattern of hexagonal convection cells — the surface-tension-driven cousin of Bénard's original 1900 experiment, which for decades was misattributed to buoyancy until Block (1956, experiment) and Pearson (1958, theory) showed the free surface was essential.
Flow is visualized by seeding tracer particles and using PIV, by shadowgraph/schlieren for the thermal field, and by infrared thermography of the surface. A cleaner realization is the float-zone / liquid-bridge geometry: a column of fluid held between two rods at different temperatures, run in microgravity (Spacelab, ISS, and the JAXA/ESA "Marangoni Experiment" on Kibo) to remove buoyancy entirely. There the steady axisymmetric thermocapillary flow undergoes a sharp transition to oscillatory, traveling-wave convection above a critical Ma, a benchmark test of the theory. Everyday manifestations — tears of wine, the self-propelling camphor boat, and the coffee-ring reversal in drying droplets — are the same effect made visible.
Where it operates and how it differs from cousins
Marangoni convection reigns wherever a free or fluid–fluid interface carries a σ gradient and length scales are small or gravity is weak: thin films and coatings, foam and emulsion lamellae, small droplets and menisci, boiling bubbles, molten weld and solder pools, and float-zone crystal growth of silicon and oxides.
Two distinctions matter. Versus Rayleigh–Bénard convection: that is a bulk body force (buoyancy), scales as depth³ (Ra ∝ L³, Ra_c ≈ 1708), needs no interface, and dies in free fall; Marangoni is an interfacial stress, scales as depth (Ma ∝ L), and survives microgravity. In real experiments both act, superposing as Bénard–Marangoni convection. Versus solutocapillary flow: identical mathematics, but σ varies with composition rather than temperature (replace α by the solute diffusivity D, so the effective Schmidt-number analog Sc = ν/D ≫ Pr, making solutal effects far more persistent). Surfactant gradients further add Marangoni elasticity, which stiffens interfaces and stabilizes foams — the physical basis of Gibbs–Marangoni film stability.
Applications, open questions, and significance
Applications span manufacturing and biology. In laser and arc welding, the sign of dσ/dT (flipped positive by trace sulfur/oxygen surfactants) determines whether the pool flows inward for deep penetration or outward for shallow, wide beads — a multimillion-dollar process lever. In crystal growth, oscillatory Marangoni flow imprints striations that degrade semiconductor and optical crystals. In printing and coatings, thermal/solutal Marangoni stresses drive the coffee ring and its suppression, enabling uniform inkjet deposits and self-assembly. Marangoni propulsion powers surfactant-driven microswimmers and lab-on-chip droplet transport.
Open questions remain lively. The route to spatiotemporal chaos and pattern selection (hexagons vs. rolls vs. squares) near threshold is not fully classified; coupling to evaporation, phase change, and deformable/rupturing interfaces (dewetting, the long-wave Marangoni instability) is analytically hard. Microgravity experiments continue to probe the transition to oscillatory 3D flow and its dependence on Pr, aspect ratio, and interfacial deformation. As a paradigm, Marangoni convection remains the cleanest demonstration that surface thermodynamics can, by itself, organize a fluid into motion.
| Property | Marangoni (Bénard–Marangoni) | Rayleigh–Bénard (buoyancy) | Solutocapillary Marangoni |
|---|---|---|---|
| Driving force | Surface-tension gradient ∇ₛσ (interfacial stress) | Density gradient in a gravity field (body force) | Surface-tension gradient from solute ∇ₛc |
| Origin of σ change | Temperature: σ = σ(T), dσ/dT < 0 | — (no free-surface stress needed) | Composition: σ = σ(c) |
| Control number | Ma = −(dσ/dT)ΔT·L/(μα) | Ra = gβ ΔT L³/(να) | Ma_c = −(dσ/dc)Δc·L/(μD) |
| Critical value | Ma_c ≈ 79.6 (non-deformable, adiabatic film) | Ra_c ≈ 1708 (rigid–rigid) | Same form, replace α→D (Schmidt ≫ 1) |
| Scaling with depth | ∝ L (favored in thin films) | ∝ L³ (favored in deep layers) | ∝ L |
| Survives in microgravity? | Yes — needs only an interface | No — vanishes as g → 0 | Yes |
Frequently asked questions
What actually drives Marangoni flow — is it a body force?
No. Unlike buoyancy, there is no body force in the bulk. The drive is a tangential stress that lives entirely at the interface: τ = ∇ₛσ, the gradient of surface tension along the surface. Fluid is pulled toward regions of higher σ (lower temperature for a normal liquid), and viscosity transmits that surface shear into the bulk to set up convection.
Why does surface tension usually decrease with temperature?
For a normal pure liquid dσ/dT < 0 because raising the temperature increases molecular thermal motion and reduces the cohesive energy deficit at the surface, so σ falls and vanishes at the critical point. For water dσ/dT ≈ −0.15 mN·m⁻¹·K⁻¹. Some systems (certain aqueous alcohol mixtures, or metals with dissolved sulfur/oxygen) can locally reverse the sign, which flips the flow direction — crucial in welding.
What is the Marangoni number and its critical value?
Ma = −(dσ/dT)ΔT·L/(μα) is the ratio of thermocapillary drive to viscous and thermal diffusion. Below Ma_c the surface is stable; above it, convection cells form. For a thin non-deformable film with an adiabatic surface and rigid conducting bottom, Pearson's 1958 linear analysis gives Ma_c ≈ 79.6 at wavenumber k ≈ 1.99; a constant-flux bottom lowers it to about 48.
How is Marangoni convection different from Rayleigh–Bénard convection?
Rayleigh–Bénard is buoyancy-driven: a body force from density differences in gravity, governed by Ra ∝ L³ (Ra_c ≈ 1708), and it disappears in microgravity. Marangoni is interfacial-stress-driven, governed by Ma ∝ L (Ma_c ≈ 79.6), needs a free surface, and persists in free fall. Because Ma scales as L and Ra as L³, Marangoni dominates in thin layers (below the ~2.7 mm capillary length for water); both coexist as Bénard–Marangoni convection.
What are the famous everyday examples?
The 'tears of wine' (explained by James Thomson in 1855): alcohol evaporates faster near the meniscus, raising the local surface tension and pulling a film up the glass that then beads and runs back. The self-propelling camphor or soap boat, the coffee-ring stain and its Marangoni-driven reversal in drying droplets, and the persistence of soap films and foams (Gibbs–Marangoni elasticity) are all the same effect.
Why do microgravity experiments study the Marangoni effect?
Because on Earth buoyancy (Rayleigh) convection masks the interfacial effect except in very thin layers. In orbit (Spacelab, ISS Kibo Marangoni experiments) gravity is effectively removed, so a heated liquid bridge exhibits pure thermocapillary flow. Researchers measure the critical Marangoni number for the onset of steady-to-oscillatory transition as a function of Prandtl number and aspect ratio, providing clean benchmarks for the theory relevant to space-based crystal growth.