Rayleigh-Taylor Instability

• Rayleigh–Taylor instability is the interfacial instability occurring when a heavier fluid overlies a lighter fluid under acceleration, leading to energy release and mixing.
• It manifests through bubble, spike, and mushroom-like structures, with evolution influenced by factors such as stratification, magnetic fields, and surface tension.
• Studies reveal that RTI transitions from exponential linear growth to turbulent nonlinear mixing, with the mixing region's evolution governed by complex stabilizing and destabilizing mechanisms.

to=arxiv_search.search 񎔊ppjson {"query":"Rayleigh-Taylor instability", "max_results": 10, "sort_by":"relevance"} to=arxiv_search.search qq天天中彩票json {"query":"(Wykes et al., 2012) OR (Geng et al., 2024) OR (Porth et al., 2014) OR (Bret, 2011) Rayleigh-Taylor instability", "max_results": 10, "sort_by":"relevance"} Rayleigh–Taylor instability (RTI) is the interfacial instability of an accelerated density inversion. In its canonical form, a heavier fluid overlies a lighter fluid in a gravitational field, so a small corrugation of the interface lowers potential energy by allowing the heavy fluid to sink and the light fluid to rise. The resulting evolution produces bubbles, spikes, mushroom-like structures, and a mixing region whose growth may continue, saturate, or be arrested depending on stratification, geometry, forcing history, and the presence of additional restoring stresses such as surface tension, magnetic tension, elasticity, or radiative support (Bret, 2011, Wykes et al., 2012, Geng et al., 2024).

1. Classical configuration and linear theory

The classical RT configuration consists of two fluids separated by a horizontal interface, with the upper fluid denser than the lower one. In the incompressible planar problem, a small vertical displacement produces a pressure imbalance proportional to $(\rho_1-\rho_2)gz$; the displaced mass scales with $\rho_1+\rho_2$, yielding the standard linear growth law

$$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$

where $k$ is the perturbation wavenumber and $A_t$ is the Atwood number. The same criterion may be expressed as $\omega^2=gk(\rho_2-\rho_1)/(\rho_2+\rho_1)$, so instability corresponds to the heavy-over-light arrangement (Bret, 2011, Jacquet et al., 2011).

This linear picture immediately explains two recurrent features of RTI. First, the relevant “gravity” can be a true gravitational field or any effective acceleration directed from the heavy side toward the light side. Second, in the absence of stabilizing physics, shorter wavelengths grow faster because the linear rate scales as $k^{1/2}$ in the classical incompressible problem (Porth et al., 2014). That scaling is modified, but not conceptually replaced, when compressibility, radiation, magnetic fields, elasticity, or quantum interfacial tension are introduced.

Finite depth and genuinely three-dimensional structure alter the dispersion relation without changing the basic mechanism. For a fully 3D finite-depth inviscid problem, arbitrary small interface disturbances can be decomposed into Fourier modes, each growing with

$$\alpha=\sqrt{\frac{(D-1)\xi}{D\coth(\xi H_2)+\coth(\xi H_1)}},$$

where $D=\rho_2/\rho_1$, $\xi=\sqrt{m^2+n^2}$, and $\rho_1+\rho_2$0 are the lower and upper layer depths. Early-time evolution is therefore modewise exponential, while the full interface is a superposition of independently amplified 3D harmonics (Walters et al., 2019).

2. Nonlinear development, mixing, and confinement

Once perturbations leave the linear regime, the interface develops the familiar RT morphology: light-fluid bubbles rise, heavy-fluid spikes or fingers descend, and roll-up at the tips generates mushroom-shaped structures. In astrophysical MHD simulations of the Crab Nebula, the nonlinear sequence is described as linear growth, bubble-and-finger formation, merging or inverse cascade, fragmentation or replenishment of smaller structures, and turbulent interaction that smears out idealized mushroom caps (Porth et al., 2014). In a binary quantum fluid, the same sequence appears after initial sinusoidal corrugations, reinforcing that the bubble–spike–mushroom pattern is not restricted to classical laboratory liquids (Geng et al., 2024).

A central quantity in experiments is the thickness of the mixing region. In an unconfined fresh/salt-water tank with two layers of uniform density, removal of a stainless steel barrier triggers an unstable interface whose mixing region grows with time, accelerates, and ultimately fills the tank. When the same unstable layer is placed between two stable stratifications, the mixing region again accelerates initially, but the stable buoyancy gradients slow the growth and bring it to a halt before the tank is filled (Wykes et al., 2012). This establishes an important point: RTI does not imply unlimited domain-filling growth; surrounding stable stratification can arrest the instability.

Several theoretical constructions recover the characteristic quadratic growth of the mixed region. In a convex-integration formulation of the inhomogeneous incompressible Euler equations, explicit ultra-high-Atwood subsolutions produce a mixing zone

$\rho_1+\rho_2$1

with $\rho_1+\rho_2$2, so the vertical extent grows like $\rho_1+\rho_2$3 (Gebhard et al., 2020). A shell-model theory of RT turbulence reaches the same large-scale law, writing the mixing-layer width as $\rho_1+\rho_2$4 and reinterpreting the turbulent phase as a stochastic traveling wave in renormalized variables (Mailybaev, 2016). This suggests that the $\rho_1+\rho_2$5 law is robust across very different theoretical descriptions, although the microscopic interpretation differs substantially.

3. Geometry, forcing history, and spatiotemporal character

RTI is highly sensitive to the way acceleration is applied. For power-law forcing

$\rho_1+\rho_2$6

the early-time dynamics splits into two sub-regimes separated by the critical exponent $\rho_1+\rho_2$7. For $\rho_1+\rho_2$8, acceleration sets the time scale and the instability is RT-type; for $\rho_1+\rho_2$9, the evolution is controlled instead by the initial growth rate and becomes Richtmyer–Meshkov-like; at $$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$0, the outcome depends on the acceleration strength $$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$1 (Hill et al., 2019). This removes any assumption that “Rayleigh–Taylor” necessarily means constant acceleration.

Geometry can also change whether instability manifests locally. For a thin viscous film hanging beneath an inclined plane, the film remains temporally unstable for any $$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$2, yet the along-plane base flow can convect disturbances away before they form drops. The Briggs–Bers analysis yields a finite critical inclination through

$$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$3

so the relevant distinction is absolute versus convective instability rather than simple temporal growth (Brun et al., 2015). A common misconception is therefore that temporal instability alone guarantees local structure formation.

In spherical or radially accelerated settings, the same mechanism reappears with geometry-dependent mode selection. In pulsar-wind-nebula calculations, the spherical linear growth rate is written as

$$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$4

with $$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$5 the spherical harmonic degree and $$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$6 the nebular radius (Porth et al., 2014). In drop-impact cratering experiments, a dense drop spreads over an expanding crater wall; when the crater decelerates, the drop/pool interface becomes RT-unstable and produces mushroom-shaped plumes. There the effective radial deceleration can exceed ambient gravity by more than an order of magnitude, so the instability is set by crater dynamics rather than by static buoyancy (Lherm et al., 2020).

4. Stabilization and modification mechanisms

The simplest stabilizing mechanism is a restoring buoyancy gradient. The fresh/salt-water confinement experiment shows that stable stratification above and below an unstable layer slows the growth and halts the mixing region before it fills the tank (Wykes et al., 2012). More generally, any mechanism that adds a positive restoring term at the interface can suppress some or all of the unstable spectrum.

Magnetic tension provides the canonical example. For an interface threaded by a field parallel to the interface, the magnetized RT dispersion relation becomes

$$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$7

and perturbations aligned with the field are stabilized below the critical wavelength

$$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$8

Yet suppression is not automatic in realistic systems: long-time Crab Nebula simulations argue that magnetic dissipation and field randomization reduce the stabilizing role of magnetic tension, so prominent RT filaments remain compatible with observations (Porth et al., 2014). In partially ionized media, ions are tied directly to the field but neutrals are not; the two-fluid theory yields two unstable branches, one ion-like and one neutral-like, and increasing collision rate or ionization fraction reduces the growth rate without eliminating RTI (Shadmehri et al., 2013).

Radiation modifies RTI in two qualitatively distinct ways. In the optically thin, isothermal regime, radiation acts as part of an effective gravitational field, and instability occurs when

$$\delta^2=A_t g k,\qquad A_t=\frac{\rho_1-\rho_2}{\rho_1+\rho_2},$$9

In the optically thick, adiabatic regime, radiation instead modifies the equation of state of the gas-radiation mixture, so the instability becomes a Rayleigh–Taylor problem for a single effective fluid with total pressure $k$0 (Jacquet et al., 2011). Rotation has a different effect: for compressible rotating flows, the Coriolis force diminishes the growth rate of unstable modes but does not prevent instability or the resulting Hadamard ill-posedness of the linearized problem (Duan et al., 2012).

Elasticity can stabilize wavelengths that would be unstable in a homogeneous fluid model. For RTI with a foam beneath a heavier fluid, the elastic-regime growth rate is

$k$1

so perturbations with

$k$2

are stabilized (Bret et al., 31 Oct 2025). This directly contradicts the common inference from classical RTI that smaller scales must always grow faster: in an intact elastic foam, sufficiently short wavelengths are stable. Related quantum interfacial systems display the same structural idea. At the superfluid-solid $k$3He interface, the linear perturbation equation contains the restoring terms from surface stiffness and gravity together with an acceleration term and kinetic damping; for uniform acceleration, instability begins at

$k$4

and the threshold is independent of the surface stiffness (Burmistrov et al., 2010).

5. Quantum, granular, and other nonclassical realizations

A binary quantum fluid provides a direct quantum analogue of classical RTI. In an immiscible two-component Bose–Einstein condensate, a reversed magnetic-field gradient forces the two components together and destabilizes the interface. The linear interface mode

$k$5

obeys

$k$6

so $k$7 gives stable ripplons, whereas $k$8 makes long waves unstable with critical wavevector

$k$9

and maximum growth at $A_t$0. Nonlinearly, the system develops bubble-, spike-, and mushroom-like structures, while stabilization of the interface allows ripplon spectroscopy and matter-wave interferometry that converts interfacial counterflow into a vortex chain (Geng et al., 2024).

The superfluid-solid $A_t$1He interface offers a second quantum realization, but with crystallization waves rather than ordinary gravity-capillary waves. Its perturbation dynamics contains an effective inertia $A_t$2, a damping term $A_t$3, and an acceleration-dependent restoring coefficient. In the uniformly accelerated case, instability occurs only for a growing solid phase once $A_t$4, whereas in the impulsive Richtmyer–Meshkov limit the onset does not depend on the sign of acceleration and the initial amplification is linear in time rather than exponential (Burmistrov et al., 2010). This suggests that the RT mechanism survives even when the relevant interface physics is phase conversion rather than ordinary fluid advection.

Dry granular flows show that neither molecular fluids nor externally prepared density inversions are required. On a rough incline, size segregation can lift larger grains to the free surface; when those larger grains are also denser, the flow creates its own unstable density inversion. The resulting “self-induced” RTI generates descending plumes, counter-rotating rolls, and eventually a long-time state of alternating bands analogous to Rayleigh–Bénard convection cells (d'Ortona et al., 2020). Unlike ordinary fluids, which relax toward superposed stable layers, the granular system sustains recirculation because segregation continually regenerates the unstable density profile. A plausible implication is that RTI should be regarded more generally as an instability of accelerated effective density contrasts, not only of pre-imposed fluid layers.

6. Astrophysical, mathematical, and computational perspectives

Astrophysical RTI is often driven by acceleration or deceleration of interfaces rather than by static gravity alone. In the Crab Nebula, the contact discontinuity between the pulsar wind nebula and supernova ejecta is RT-unstable because a hot, low-density magnetized bubble accelerates denser ejecta outward; high-resolution axisymmetric AMR MHD simulations reproduce filamentary shells, irregular fingers, and a dominant filament separation influenced by nebular turbulence and inverse cascade rather than by a single linear wavelength (Porth et al., 2014). In relativistic gamma-ray-burst fireballs, the unstable interface is the contact discontinuity between shocked ejecta and shocked circumburst medium. Relativistic causality restricts growth to angular scales smaller than $A_t$5, the instability destroys the contact discontinuity, strengthens the reverse shock, and generates turbulent kinetic energy at the level $A_t$6, with the magnetic energy fraction inferred to be of order $A_t$7 through small-scale turbulent dynamo arguments (Duffell et al., 2013).

Structured relativistic jets add another internal-astrophysical variant. In two-component spine-sheath jets, reconfinement-induced oscillations generate an effective gravity that flips sign during collimation and de-collimation, so both heavy-spine-light-sheath and light-spine-heavy-sheath configurations can become RT/RM-unstable and mix without globally disrupting the jet (Toma et al., 2017). Impact cratering experiments extend the same mechanism to planetary differentiation: dense-drop impacts into a lighter pool produce spherical RT plumes around the crater wall, and the resulting mixing has been used to estimate metal-silicate equilibration during planetesimal impacts (Lherm et al., 2020).

RTI has also become a testing ground for modern mathematical fluid dynamics. By rewriting the inhomogeneous incompressible Euler equations as a differential inclusion and relaxing the constitutive constraints, one may construct infinitely many admissible weak solutions representing turbulent mixing in the classical heavy-over-light configuration. In the ultra-high Atwood regime, explicit subsolutions yield a mixing zone whose width is quadratic in time (Gebhard et al., 2020). Complementarily, shell-model work recasts the turbulent stage as a universal stochastic traveling wave in renormalized coordinates, with one limiting state corresponding to the initial discontinuity and the other to stationary small-scale turbulence (Mailybaev, 2016). These approaches do not replace laboratory or numerical studies, but they sharpen the distinction between interface instability as a deterministic linear mechanism and mixing as a potentially nonunique or stochastic nonlinear regime.

Across these settings, two recurring clarifications stand out. First, RTI is not synonymous with unrestricted growth: stable stratification, convective sweep, elasticity, or finite forcing history can halt or localize the instability (Wykes et al., 2012, Brun et al., 2015, Bret et al., 31 Oct 2025). Second, magnetic or rotational effects do not automatically suppress RTI in realistic systems; their efficacy depends on geometry, coupling, coherence, and the existence of modes or components that evade the restoring stress (Porth et al., 2014, Shadmehri et al., 2013, Duan et al., 2012). In that sense, the enduring significance of Rayleigh–Taylor instability lies not only in the classical heavy-over-light picture, but in the broad class of accelerated interfaces for which releasing potential energy competes with every available restoring mechanism.