Post

Parker Instability

Posted Updated
By Donghui Son 9 min read

Introduction

The Parker instability, introduced by E. N. Parker in 1966, is a magnetic-buoyancy instability of a gravitationally stratified plasma whose weight is supported partly by magnetic pressure. A small undulation of a horizontal magnetic field lets gas drain along the inclined field into magnetic valleys. The crest is then unloaded, becomes more buoyant, and can rise further unless magnetic tension restores the field to a horizontal state.

The same magnetic-buoyancy mechanism appears in models of galactic disks, solar flux emergence, and magnetized accretion-disk atmospheres.

This note uses a deliberately simplified model: uniform gravity, a non-rotating isothermal gas, a horizontal magnetic field, constant plasma beta, and no cosmic rays, shear, self-gravity, or diffusion. We derive the equilibrium, outline the linearized system, and quote model-specific stability scales from the classical linear analysis. These scales are not universal Parker-instability criteria.

Physical Setup

We assume:

  1. A stratified plasma in the upper half-space $z\geq 0$, with gravitational acceleration

    \[\mathbf{g}=-g\,\mathbf{e}_z,\]

    so the gravitational force density is $\mathbf{f}_g=\rho\mathbf{g}$.

  2. A static equilibrium, $\mathbf{v}_0=0$, with a horizontal magnetic field

    \[\mathbf{B}_0=B_0(z)\,\mathbf{e}_x.\]

  3. An isothermal closure imposed throughout the evolution,

    \[P_{\rm gas}=\rho C_s^2, \qquad C_s=\text{constant}.\]

    This assumption applies to both the background and its perturbations; an isothermal background alone would not determine the perturbation thermodynamics.

  4. A constant equilibrium plasma beta,

    \[\beta_{\rm pl} \equiv \frac{P_{\rm gas}}{P_{\rm mag}} = \frac{\rho_0C_s^2}{B_0^2/(8\pi)} =\text{constant}.\]

Define the magnetic-to-gas pressure ratio

\[\alpha \equiv \frac{P_{\rm mag}}{P_{\rm gas}} =\beta_{\rm pl}^{-1}.\]

Under these assumptions the Alfvén speed is exactly constant:

\[v_A^2 = \frac{B_0^2}{4\pi\rho_0} =2\alpha C_s^2 =\frac{2C_s^2}{\beta_{\rm pl}}.\]

Governing Ideal-MHD Equations

The model consists of:

  1. Continuity

    \[\frac{\partial\rho}{\partial t} +\nabla\cdot(\rho\mathbf{v})=0.\]

  2. Momentum conservation

    \[\rho\frac{d\mathbf{v}}{dt} = -\nabla\left(P_{\rm gas}+\frac{B^2}{8\pi}\right) +\frac{(\mathbf{B}\cdot\nabla)\mathbf{B}}{4\pi} +\rho\mathbf{g}.\]

  3. Induction

    \[\frac{\partial\mathbf{B}}{\partial t} =\nabla\times(\mathbf{v}\times\mathbf{B}).\]

  4. Solenoidal constraint

    \[\nabla\cdot\mathbf{B}=0.\]

  5. Isothermal closure

    \[P_{\rm gas}=\rho C_s^2.\]

The induction equation preserves $\nabla\cdot\mathbf{B}=0$ in the continuum if it is satisfied initially.

Equilibrium Configuration

For $\mathbf{v}_0=0$, force balance gives

\[0 = -\nabla\left(P_0+\frac{B_0^2}{8\pi}\right) +\frac{(\mathbf{B}_0\cdot\nabla)\mathbf{B}_0}{4\pi} +\rho_0\mathbf{g}.\]

Because $\mathbf{B}_0=B_0(z)\mathbf{e}_x$ has no $x$-dependence,

\[(\mathbf{B}_0\cdot\nabla)\mathbf{B}_0=0.\]

The vertical equilibrium is therefore

\[\frac{d}{dz} \left(P_0+\frac{B_0^2}{8\pi}\right) =-\rho_0g.\]

Define the gas-only and total-pressure scale heights as

\[H_g\equiv\frac{C_s^2}{g}, \qquad H_{\rm tot}\equiv(1+\alpha)H_g.\]

Using $P_0=\rho_0C_s^2$ and $B_0^2/(8\pi)=\alpha P_0$, we obtain

\[\rho_0(z) =\rho_0(0)\exp\left(-\frac{z}{H_{\rm tot}}\right),\]

and

\[B_0(z) =B_0(0)\exp\left(-\frac{z}{2H_{\rm tot}}\right).\]

Magnetic pressure therefore increases the density scale height from $H_g$ to $H_{\rm tot}$.

Linearization

Write each variable as an equilibrium value plus a small perturbation:

\[\begin{aligned} \rho &= \rho_0+\delta\rho,\\ \mathbf{v} &= \delta\mathbf{v},\\ P_{\rm gas} &= P_0+\delta P,\\ \mathbf{B} &= \mathbf{B}_0+\mathbf{b}, \end{aligned}\]

where $\mathbf{b}\equiv\delta\mathbf{B}$.

Continuity and isothermal closure

To first order,

\[\frac{\partial\delta\rho}{\partial t} +\nabla\cdot(\rho_0\delta\mathbf{v})=0.\]

Because this model explicitly imposes isothermal perturbations,

\[\delta P=C_s^2\delta\rho.\]

For adiabatic or non-isothermal perturbations, this relation must be replaced by an energy equation or another thermodynamic closure, and the stability boundary changes.

Momentum

The first-order Lorentz force is

\[\delta\mathbf{F}_L = \frac{1}{4\pi} \left[ (\nabla\times\mathbf{b})\times\mathbf{B}_0 +(\nabla\times\mathbf{B}_0)\times\mathbf{b} \right].\]

Using $\nabla\cdot\mathbf{B}_0=\nabla\cdot\mathbf{b}=0$,

\[\delta\mathbf{F}_L = \frac{1}{4\pi} \left[ (\mathbf{B}_0\cdot\nabla)\mathbf{b} +(\mathbf{b}\cdot\nabla)\mathbf{B}_0 -\nabla(\mathbf{B}_0\cdot\mathbf{b}) \right].\]

The first-order momentum equation is

\[\boxed{ \rho_0\frac{\partial\delta\mathbf{v}}{\partial t} = -\nabla\delta P -\nabla\left(\frac{\mathbf{B}_0\cdot\mathbf{b}}{4\pi}\right) +\frac{(\mathbf{B}_0\cdot\nabla)\mathbf{b}}{4\pi} +\frac{(\mathbf{b}\cdot\nabla)\mathbf{B}_0}{4\pi} +\delta\rho\,\mathbf{g} }.\]

The final term is the gravitational force associated with the density perturbation. For $\mathbf{g}=-g\mathbf{e}_z$, a negative $\delta\rho$ contributes an upward force.

Induction and magnetic divergence

The linear induction equation and constraint are

\[\frac{\partial\mathbf{b}}{\partial t} =\nabla\times(\delta\mathbf{v}\times\mathbf{B}_0), \qquad \nabla\cdot\mathbf{b}=0.\]

Normal Modes and Mode Geometry

Assume a horizontal Fourier dependence

\[\delta f(x,y,z,t) = \hat f(z) \exp\left[i\left(\omega t-k_xx-k_yy\right)\right].\]

With this time convention, a growing mode can be written as

\[\omega=\omega_R-i\gamma, \qquad \gamma=-\operatorname{Im}\omega>0.\]

Because $\mathbf{B}_0\parallel\mathbf{e}_x$, the parallel and transverse horizontal wavenumbers are

\[k_\parallel=k_x, \qquad k_\perp=k_y.\]

This distinction separates three related modes:

  • Undular mode: $k_x\neq0$ and $k_y=0$. Field lines bend in the $x$–$z$ plane, and gas can drain along them.
  • Interchange mode: $k_x=0$ and $k_y\neq0$. Flux tubes exchange position without the same parallel field-line bending penalty.
  • Mixed mode: $k_x\neq0$ and $k_y\neq0$. It combines undular drainage with transverse interchange structure.

Magnetic tension suppresses sufficiently short parallel wavelengths through terms proportional to $k_\parallel^2B_0^2$. It does not imply that every large transverse wavenumber is stabilized. In ideal three-dimensional models, mixed modes can instead favor very large $k_\perp$; additional non-ideal physics, finite boundaries, or numerical resolution can provide cutoffs. The energetic distinction between interchange and more general displacements goes back to Newcomb (1961), while the short-transverse-scale behavior is demonstrated in Kim et al. (1998) and Kim, Ryu, and Jones (2001).

For the normal modes above, the induction equation gives

\[\begin{aligned} i\omega\hat b_x &= i k_yB_0\hat v_y -\frac{d}{dz}(B_0\hat v_z),\\ i\omega\hat b_y &=-i k_xB_0\hat v_y,\\ i\omega\hat b_z &=-i k_xB_0\hat v_z. \end{aligned}\]

Substituting these relations into the continuity and momentum equations produces the linear eigenvalue problem. The exponential, constant-$\alpha$ equilibrium used here is a special case for which height dependence can be separated and an analytic dispersion relation can be obtained. More general $g(z)$, $\beta(z)$, thermal physics, rotation, or finite-boundary models generally require a numerical eigenvalue calculation.

Consistent vertical boundaries

The one-sided equilibrium above applies to $z\geq0$. One may impose a lower boundary at $z=0$ and decay or an upper boundary condition as $z\to+\infty$. A full disk symmetric about $z=0$ requires gravity to reverse direction across the midplane, producing a profile such as

\[\rho_0\propto\exp\left(-\frac{|z|}{H_{\rm tot}}\right)\]

for the idealized constant-magnitude gravity model. The eigenvalues depend on which vertical model and boundary conditions are selected.

Critical Wavelength and Growth Rate

For the present uniform-gravity, no-cosmic-ray, isothermal ($\gamma=1$) undular problem, Parker’s linear stability analysis gives the critical parallel wavelength

\[\boxed{ \lambda_{c,\parallel} = \frac{4\pi H_{\rm tot}}{\sqrt{1+2\alpha}} = \frac{4\pi H_g(1+\beta_{\rm pl}^{-1})} {\sqrt{1+2\beta_{\rm pl}^{-1}}} }.\]

Parallel wavelengths longer than this value are unstable in that particular model. For $\beta_{\rm pl}=1$,

\[\lambda_{c,\parallel} = \frac{4\pi}{\sqrt3}H_{\rm tot} \simeq7.26H_{\rm tot}.\]

The critical wavelength is not the fastest-growing wavelength. For the representative $\alpha=1$ model, the dispersion curves reproduced by Kim, Ryu, and Jones (2001) give approximately

\[\lambda_{\parallel,\max}\simeq12H_{\rm tot}, \qquad \gamma_{\max,\rm undular}\simeq0.34\frac{C_s}{H_{\rm tot}},\]

while the fastest ideal mixed mode has a similar parallel scale and a somewhat larger rate,

\[\lambda_{\parallel,\max}\simeq11H_{\rm tot}, \qquad \gamma_{\max,\rm mixed}\simeq0.41\frac{C_s}{H_{\rm tot}}.\]

The latter maximum occurs as the transverse wavelength tends toward zero in the idealized model. These numerical coefficients depend on the mode and assumptions; a crossing-time estimate such as $H/v_A$ supplies only a dimensional scale, not a universal growth time.

Intuitive Force Balance

An upward bend gives gravity a component along the inclined field, so gas drains from the crest into neighboring valleys. The crest loses mass and becomes more buoyant, while magnetic tension opposes the bend.

For a small sinusoidal displacement

\[z(x)=\Delta z\cos\left(\frac{2\pi x}{\lambda}\right),\]

the radius of curvature at the crest is

\[R_{\rm crest} \simeq \frac{\lambda^2}{4\pi^2\Delta z}.\]

The corresponding magnetic-tension force density has the scale

\[f_{\rm tension} \sim \frac{B^2}{4\pi R_{\rm crest}}.\]

A local force balance recovers the qualitative scaling that unstable parallel wavelengths must be several pressure scale heights long. It does not determine the exact coefficient reliably because the density perturbation, gas drainage, vertical eigenfunction, and field curvature are dynamically coupled. The coefficient in the preceding section comes from the full linear eigenvalue problem, not from this one-zone estimate.

Astrophysical Examples

  1. Solar Flux Emergence and Active Regions

    Magnetic buoyancy contributes to the rise of subsurface magnetic structures, but realistic flux emergence also involves convection, stable photospheric stratification, and interchange-type instability near the surface. The three-dimensional simulation by Toriumi and Yokoyama (2012) shows that a buoyant flux tube can spread beneath the photosphere before further magnetic-buoyancy instability builds magnetic domes and active-region structure. This is a more limited statement than attributing all coronal loops or sunspots to the simple uniform-gravity Parker model.

  2. Galactic-Center Molecular Loops

    Fukui et al. (2006) reported large molecular loops near the Galactic center and proposed magnetic flotation driven by the Parker instability as an explanation. In this interpretation, magnetic arches rise while gas tends to slide down the inclined field and accumulate near valleys or loop footpoints, as examined further by Torii et al. (2010); the molecular gas is not simply carried upward at unchanged loading. The observations are consistent with this interpretation but do not uniquely prove it. Three-dimensional calculations by Kim et al. (1998) and Kim, Ryu, and Jones (2001) also find density enhancements too small for Parker instability alone to form giant molecular clouds.

  3. Accretion Disks

    In some strongly magnetized, vertically stratified disks, Parker instability can lift azimuthal magnetic field and help create vertical field components, as in the simulations of Johansen and Levin (2008). MRI can then drive much of the disk-body turbulence and Maxwell stress responsible for angular-momentum transport, while Parker-unstable structures can dominate higher in the atmosphere, as found by Held, Mamatsashvili, and Pessah (2024). Parker instability, differential rotation, and MRI are therefore related but distinct pieces of the dynamics.

Scope and Limitations

The formulas in this note apply to a highly idealized equilibrium. Cosmic-ray pressure, rotation, shear, self-gravity, realistic vertical gravity, non-isothermal thermodynamics, viscosity, resistivity, and thermal conduction can all modify the instability threshold, fastest-growing wavelength, or growth rate. Solar, galactic, and accretion-disk applications therefore require models tailored to their own stratification and energy balance.

References

Study Log
plasma-physics instabilities magnetic-buoyancy astrophysics solar-physics intermediate
This post is licensed under CC BY 4.0 by the author.