Magnetorotational Instability and Accretion Turbulence

Every accretion disk hides a paradox. Gas orbiting a star or black hole must lose angular momentum to spiral inward and release its gravitational energy — the release that lights up quasars, powers X-ray binaries, and builds planets. Yet the molecular viscosity of a real astrophysical plasma is pitifully weak: left to microscopic friction, a disk around a supermassive black hole would take far longer than the age of the Universe to drain. Worse, the flow looks stable. A Keplerian disk has specific angular momentum rising outward, so by Rayleigh's criterion it should not even be turbulent. For decades this was the central embarrassment of accretion theory: the disks plainly accrete, Shakura and Sunyaev fit the data with a fudge factor $\alpha$ , and nobody could say what $\alpha$ physically stood for. The resolution, published by Steven Balbus and John Hawley in 1991, was hiding in a result Velikhov and Chandrasekhar had derived thirty years earlier for magnetized Couette flow. Thread the disk with a field so weak it seems dynamically irrelevant, and the flow turns violently unstable. Two parcels on neighbouring orbits, tethered by a field line acting like a weak spring, exchange angular momentum through magnetic tension: the inner one is dragged back, sinks, and speeds up; the outer is dragged forward and climbs out. The feedback runs away on the orbital timescale. The instability cares only about the sign of the angular-velocity gradient — reversing Rayleigh's intuition — and it saturates into turbulence whose stresses are the physical meaning of $\alpha$ . This article develops the magnetorotational instability from its dispersion relation to the turbulence that feeds black holes.

Key takeaways

The magnetorotational instability (MRI) destabilizes any weakly magnetized, differentially rotating fluid in which the angular velocity decreases outward, $\mathrm{d}\Omega^2/\mathrm{d}R<0$ — the reverse of the Rayleigh criterion, which depends on $\mathrm{d}(R^2\Omega)^2/\mathrm{d}R$ .
A weak field line links fluid elements like a spring; its tension carries angular momentum outward and drives exponential growth on the orbital timescale.
The peak growth rate is $\gamma_{\max}=\tfrac34\Omega$ for a Keplerian disk, independent of field strength — about $4.7$ e-foldings per orbit no matter how weak the seed field.
The field sets not the growth rate but the wavelength of the fastest mode, $\lambda_{\max}\approx2\pi v_A/\Omega$ ; the MRI operates only while that fits inside the disk, $v_A\lesssim c_s$ .
Saturation is turbulence whose correlated Reynolds and Maxwell stresses transport angular momentum outward; their ratio to the pressure is the Shakura–Sunyaev $\alpha\sim0.01$ – $0.1$ .
Where ionization is too low for ideal MHD — cold protoplanetary-disk midplanes — resistive and ambipolar diffusion quench the MRI, leaving turbulence-poor "dead zones".
Key idea: A weak magnetic field threading a differentially rotating disk destabilizes it whenever $\mathrm{d}\Omega^2/\mathrm{d}R<0$ — the opposite of the hydrodynamic Rayleigh criterion — because magnetic tension couples fluid elements like springs, transporting angular momentum outward and driving turbulence that acts as the disk's effective viscosity.

Where this lives today

The MRI underlies almost every quantitative model of accretion, and its frontier is computational. Global general-relativistic MHD simulations that resolve and saturate the MRI are the engines behind the Event Horizon Telescope's models of M87 and Sagittarius A — the turbulent stress they produce sets the accretion rate, emission, and photon ring. Local "shearing-box" simulations measure $\alpha$ from first principles and its dependence on net flux, stratification, and radiation pressure, feeding calibrated viscosities into models of X-ray binaries and dwarf-nova outbursts. In protoplanetary disks the story is one of absence: ALMA line widths in HD 163296 and TW Hya imply very low midplane turbulence, consistent with non-ideal-MHD dead zones — pushing theorists toward magnetized disk winds as the dominant angular-momentum sink there. Liquid-metal Taylor-Couette experiments (Princeton MRI, PROMISE) chase the instability in the laboratory. How the MRI dynamo sustains large-scale field, how saturation scales to real Reynolds numbers, and where winds beat turbulence remain among the most active questions in accretion physics.

The physical picture

The mechanism needs no equations to grasp. Picture two beads on adjacent orbits joined by a weak elastic band — the field line. In a Keplerian disk the inner bead orbits faster, so it runs ahead and the band stretches, pulling its ends toward each other along the orbit: it tugs the leading inner bead backwards and the trailing outer bead forwards. But "backwards" means losing angular momentum, dropping the inner bead onto a smaller, faster orbit — so it pulls ahead harder; "forwards" means gaining angular momentum, lifting the outer bead to a larger, slower orbit — so it falls further behind. Each effect stretches the band more, and the configuration flies apart. That runaway is the instability, and its by-product is exactly what accretion needs: a steady outward flux of angular momentum carried by magnetic tension.

Figure

Two fluid elements on neighbouring Keplerian orbits, tethered by a magnetic field line acting as a weak spring. The inner element orbits faster, so tension drags it back (it loses angular momentum, L, and spirals inward) and pulls the outer element forward (it gains L and drifts outward) — a runaway whenever the angular velocity falls outward, dΩ²/dR negative

Why does a field-free flow stay stable while an arbitrarily weak field destabilizes it? Without a field, a displaced parcel conserves its angular momentum and performs a stable epicyclic oscillation — it swings back. The field line changes the bookkeeping: tension enforces near-corotation between the parcels, and for outward-decreasing $\Omega$ that torque is anti-restoring. Weak magnetism does not add a small correction to a stable oscillation; it converts the oscillation into exponential growth. The MRI is not a perturbation of hydrodynamics — it is a different regime. The most surprising fact is that the growth rate does not depend on the field strength. A weaker field is a "softer spring" that responds more slowly, but it also permits longer unstable wavelengths, and the two effects cancel so precisely that the fastest mode always grows at $\sim\Omega$ . The field sets only the wavelength of that mode, $\lambda_{\max}\approx2\pi v_A/\Omega$ . The second figure shows this: the peak is pinned at $\tfrac34\Omega$ while changing the field only slides the curve horizontally.

Figure

Growth rate of the axisymmetric MRI versus Alfvén frequency ω_A for a Keplerian disk with epicyclic frequency κ = Ω. Growth peaks at γ = ¾ Ω when ω_A² = 15⁄16 Ω² and vanishes at the marginal wavenumber ω_A² = 3 Ω². The peak height is independent of field strength — a stronger field only slides the curve to larger ω_A

Saturation is not captured by linear theory. As the modes grow they build coherent "channel flows" that are themselves unstable to secondary "parasitic" modes, which shred them into small-scale turbulence. That turbulence sustains correlated velocity and magnetic fluctuations whose stresses keep transporting angular momentum, self-regulating near a fixed fraction of the local pressure — the number Shakura and Sunyaev had introduced by hand two decades earlier.

The core equations

The starting point is the classical result the MRI overturns. Displace a ring outward at fixed $\ell=R^2\Omega$ and ask whether centrifugal support at the new radius exceeds what is required: the flow is stable when $\ell$ increases outward.

\frac{\mathrm{d}(R^2\Omega)^2}{\mathrm{d}R}>0 \qquad (1)

where: $R$ — cylindrical radius; $\Omega(R)$ — angular velocity; $R^2\Omega=\ell$ — specific angular momentum. Reading: stable if $\ell$ rises outward; Keplerian disks have $\ell\propto R^{1/2}$ , so they are Rayleigh-stable — the puzzle. Introduce even a whisper of field, and the relevant gradient becomes that of angular velocity, not angular momentum. This single sign flip is the entire conceptual content of the MRI.

\frac{\mathrm{d}\Omega^2}{\mathrm{d}R}<0 \qquad (2)

where: $\Omega^2$ — square of the angular velocity. Reading: a magnetized disk is unstable when $\Omega$ decreases outward. Keplerian rotation $\Omega\propto R^{-3/2}$ satisfies this handily, so nearly every astrophysical disk is MRI-unstable even while Rayleigh-stable. The quantitative engine is the local dispersion relation. Linearize ideal MHD about a differentially rotating background threaded by a vertical field $B_z$ , take axisymmetric perturbations $\propto e^{i(k_z z-\omega t)}$ , and eliminate the fluctuation amplitudes to obtain a quartic in $\omega$ .

\omega^4 - \omega^2\!\left(\kappa^2 + 2\,\omega_A^2\right) + \omega_A^2\!\left[\omega_A^2 + \frac{\mathrm{d}\Omega^2}{\mathrm{d}\ln R}\right] = 0 \qquad (3)

where: $\omega$ — complex mode frequency, $\mathrm{Im}(\omega)>0$ signals growth; $\kappa$ — epicyclic frequency (equation 4); $\omega_A\equiv k\cdot v_A=k_z v_{Az}$ — the mode's Alfvén frequency; $v_A=B/\sqrt{\mu_0\rho}$ — Alfvén speed. Reading: $\omega_A\to0$ collapses the quartic to a stable $\omega^2=\kappa^2$ ; finite $\omega_A$ can force $\omega^2<0$ . The epicyclic frequency is the restoring frequency for small radial excursions — the hydrodynamic anchor of the dispersion relation.

\kappa^2 = \frac{1}{R^3}\frac{\mathrm{d}(R^2\Omega)^2}{\mathrm{d}R} = 4\Omega^2 + 2R\,\Omega\,\frac{\mathrm{d}\Omega}{\mathrm{d}R} \qquad (4)

where: $\kappa$ — epicyclic frequency. Reading: Keplerian rotation gives $\kappa=\Omega$ , and $\kappa^2>0$ for any Rayleigh-stable disk — why hydrodynamics predicts stability and the field is needed to break it. Treating equation (3) as a quadratic in $X=\omega^2$ , a negative real root exists when the constant term is negative — the wavenumber window of unstable modes.

\omega_A^2 < -\frac{\mathrm{d}\Omega^2}{\mathrm{d}\ln R} \qquad (5)

where: $\omega_A^2=(k\cdot v_A)^2$ — squared Alfvén frequency. Reading: only modes with small enough $\omega_A$ — long enough wavelength — grow. For Keplerian rotation $\mathrm{d}\Omega^2/\mathrm{d}\ln R=-3\Omega^2$ , so the band is $\omega_A^2<3\Omega^2$ . Maximizing the growth over wavenumber gives the flagship result: a peak rate set entirely by the shear, with no reference to the field.

\gamma_{\max} = \frac12\left|\frac{\mathrm{d}\Omega}{\mathrm{d}\ln R}\right| = \frac34\,\Omega \quad(\text{Keplerian}) \qquad (6)

where: $\gamma_{\max}=\mathrm{Im}(\omega)_{\max}$ — fastest exponential growth rate. Reading: a Keplerian disturbance e-folds $\tfrac34\Omega$ times per unit time — about $4.7$ per orbit — regardless of the seed field. The MRI is the fastest local instability a disk possesses. The field's only job is to select which wavelength grows fastest, via the optimal $\omega_A^2=\tfrac{15}{16}\Omega^2$ .

\lambda_{\max} \approx 2\pi\,\frac{v_A}{\Omega} \qquad (7)

where: $\lambda_{\max}$ — vertical wavelength of the fastest mode; $v_A$ — Alfvén speed. Reading: a stronger field pushes the mode to longer wavelength. The MRI operates only while $\lambda_{\max}$ fits inside the disk, $\lambda_{\max}\lesssim H\approx c_s/\Omega$ , i.e. $v_A\lesssim c_s$ — the weak-field condition. Finally, the saturated turbulence delivers the sought-after transport, packaged as an effective viscosity in the Shakura–Sunyaev form.

\nu = \alpha\,c_s H,\qquad \alpha = \frac{\langle w_{R\phi}\rangle}{\langle P\rangle} \qquad (8)

where: $\nu$ — turbulent kinematic viscosity; $c_s$ — sound speed; $H\approx c_s/\Omega$ — scale height; $w_{R\phi}=\rho\,\delta v_R\,\delta v_\phi - B_R B_\phi/\mu_0$ — the $R\phi$ stress (Reynolds minus Maxwell); $P$ — total pressure; $\alpha$ — viscosity parameter. Reading: $\alpha$ is the volume-averaged stress-to-pressure ratio, measured in simulations to be $\sim0.01$ – $0.1$ — not a fudge factor.

Derivation AFrom the dispersion relation to the maximum growth rate

The goal is to extract the field-independent peak growth rate $\tfrac34\Omega$ directly from the Balbus–Hawley quartic. We treat the quartic as a quadratic in $X=\omega^2$ , locate the root that goes negative (negative $X$ means pure-imaginary $\omega$ , hence growth), specialize to Keplerian rotation, and maximize over the Alfvén frequency — showing explicitly that the field strength drops out of the peak rate. Take axisymmetric perturbations with $k=k_z\hat z$ , so $\omega_A=k_z v_{Az}$ .

Step 1. Rewrite equation (3) with $X\equiv\omega^2$ as $X^2 - X(\kappa^2+2\omega_A^2) + \omega_A^2\big(\omega_A^2+\mathrm{d}\Omega^2/\mathrm{d}\ln R\big)=0$ . A negative root $X<0$ makes $\omega=\pm i\sqrt{-X}$ : a mode growing as $e^{\gamma t}$ , $\gamma=\sqrt{-X}$ . Why: a quadratic in $\omega^2$ turns "does a mode grow" into the algebra of a negative root.

Step 2. The roots multiply to the constant term $\omega_A^2(\omega_A^2+\mathrm{d}\Omega^2/\mathrm{d}\ln R)$ ; since their sum $\kappa^2+2\omega_A^2$ is positive, a negative root exists exactly when the product is negative — the window $\omega_A^2<-\mathrm{d}\Omega^2/\mathrm{d}\ln R$ . Why: the product of roots is the cleanest instability test — no need to solve anything.

Step 3. Insert Keplerian $\kappa^2=\Omega^2$ , $\mathrm{d}\Omega^2/\mathrm{d}\ln R=-3\Omega^2$ . The unstable root is $X_-=\tfrac12\big[(\Omega^2+2\omega_A^2)-\sqrt{D}\big]$ with $D=(\Omega^2+2\omega_A^2)^2-4\omega_A^2(\omega_A^2-3\Omega^2)$ ; the $\omega_A^4$ terms cancel, leaving $D=\Omega^2(\Omega^2+16\,\omega_A^2)$ . Why: this cancellation is what ultimately renders the peak rate independent of $v_A$ .

Step 4. Then $\gamma^2=-X_-=\tfrac12\big[\Omega\sqrt{\Omega^2+16\,\omega_A^2}-(\Omega^2+2\omega_A^2)\big]$ . Setting $\mathrm{d}\gamma^2/\mathrm{d}\omega_A^2=0$ gives $8\Omega/\sqrt{\Omega^2+16\,\omega_A^2}=2$ , so $\sqrt{\Omega^2+16\,\omega_A^2}=4\Omega$ and $\omega_A^2=\tfrac{15}{16}\Omega^2$ . Why: the fastest mode dominates the exponential competition, so its location sets what the disk feels.

Step 5. Substitute back: $\gamma^2=\tfrac12\big[4\Omega^2-(\Omega^2+\tfrac{15}{8}\Omega^2)\big]=\tfrac{9}{16}\Omega^2$ , so $\gamma_{\max}=\tfrac34\Omega$ . The Alfvén speed appears nowhere; it survives only through $\omega_A=k_z v_{Az}$ , fixing the wavelength $\lambda_{\max}\approx2\pi v_A/\Omega$ . Why: the growth rate is a property of the shear, and the field merely chooses which wavelength cashes it in. *The discriminant reduces to $\Omega^2(\Omega^2+16\omega_A^2)$ , the optimum sits at $\omega_A^2=\tfrac{15}{16}\Omega^2$ , and $\gamma_{\max}=\tfrac34\Omega$ for Keplerian $\kappa=\Omega$ .*

Go deeper · Derivation B · 1 minAngular-momentum transport by magnetic tension — the spring model

Derivation BAngular-momentum transport by magnetic tension — the spring model

The aim is to recover the instability criterion and, crucially, the direction of angular-momentum transport from a minimal mechanical model stripped of the full MHD machinery. We join two elements on neighbouring orbits by a field line whose tension acts as a linear spring, write the azimuthal torque, track the radial response, and read off when the separation grows. The cartoon is crude but it captures the feedback and makes unmistakable that the net flux is outward — the requirement for accretion.

Step 1. Two elements at radii $R_1<R_2$ are connected by a field line of tension $\propto\omega_A^2$ , exerting equal and opposite azimuthal forces $F_{\text{tension}}\propto-\omega_A^2(\phi_1-\phi_2)$ that resist any twist between them. Why: magnetic tension behaves like a stretched spring, and all it can do azimuthally is oppose the shear that stretches it.

Step 2. Because $\Omega(R_1)>\Omega(R_2)$ in a disk with $\mathrm{d}\Omega/\mathrm{d}R<0$ , the inner element pulls ahead; the spring torques it backwards (removing $\ell$ ) and the outer element forwards (adding $\ell$ ). Why: the torque's sign is fixed by which element leads, and in any outward-decreasing- $\Omega$ disk it is always the inner one.

Step 3. Losing $\ell$ , the inner element drifts inward to a faster orbit and pulls ahead even more; gaining $\ell$ , the outer element drifts outward to a slower orbit and falls further behind. The separation grows without bound. Why: this is the positive feedback — the radial response amplifies the very shear driving the torque, making it an instability, not an oscillation.

Step 4. The feedback is destabilizing exactly when the inner element speeds up on moving inward, i.e. when $\mathrm{d}\Omega^2/\mathrm{d}R<0$ , recovering equation (2); throughout, $\ell$ passes from inner to outer, so the flux is outward. Why: outward transport is forced by the tension always opposing the shear — precisely what a disk needs to accrete. *The spring picture reproduces the Balbus–Hawley criterion $\mathrm{d}\Omega^2/\mathrm{d}R<0$ and the outward sign of the angular-momentum flux.*

Limiting cases

The MRI's regimes are organized by how the fastest wavelength compares to the disk thickness and how well the gas couples to the field.

Case	Result	Meaning
Weak-field (ideal) limit	$v_A\ll c_s,\ \lambda_{\max}\approx\dfrac{2\pi v_A}{\Omega}\ll H$	Fastest mode fits inside $H$ ; MRI at full strength, $\gamma_{\max}=\tfrac34\Omega$ — the regime of most ionized disks.
Strong-field cutoff	$\lambda_{\max}\gtrsim H\ \Leftrightarrow\ v_A\gtrsim c_s$	As $v_A$ nears $c_s$ the longest unstable wavelength no longer fits; magnetic pressure stabilizes the disk.
Rotation profile	$\dfrac{\mathrm{d}\Omega^2}{\mathrm{d}R}<0$	Needs $\Omega$ falling outward. Keplerian disks always qualify; rigid rotation ( $\Omega=\text{const}$ ) is MRI-stable.
Resistive / dead-zone limit	$\mathrm{Rm}=\dfrac{v_A^2}{\eta\,\Omega}\lesssim1$	In poorly ionized midplanes, diffusion damps the field faster than the mode grows; the MRI is quenched — a "dead zone".

Common trapConfusing the MRI criterion with the Rayleigh criterion

The two criteria depend on different gradients — the single most common source of confusion. Rayleigh stability is governed by the specific-angular-momentum gradient $\mathrm{d}(R^2\Omega)^2/\mathrm{d}R$ , which is positive for a Keplerian disk, so a field-free disk is stable. The MRI is governed by the angular-velocity gradient $\mathrm{d}\Omega^2/\mathrm{d}R$ , which is negative for the same disk, so the magnetized disk is unstable. A Keplerian disk is thus Rayleigh-stable and MRI-unstable at once — no contradiction, the criteria measure different things. A second trap: "the growth rate is field-independent" does not mean the field is irrelevant. It sets which wavelengths are unstable and whether the fastest fits inside the disk; remove it entirely ( $v_A=0$ ) and the instability vanishes.

Worked examples

1 · Growth rate and e-foldings per orbit (fully worked)

A parcel of gas orbits a $1\,M_\odot$ protostar at $R=1$ AU in a Keplerian disk threaded by a weak vertical field. Find the orbital frequency, the maximum MRI growth rate, the growth time, and the number of e-foldings the instability achieves in one orbit. Then state how the answer changes if the field is ten times weaker.

Step 1. Orbital frequency. With $GM_\odot=1.327\times10^{20}\ \mathrm{m^3\,s^{-2}}$ and $R=1.496\times10^{11}$ m, $R^3=3.35\times10^{33}\ \mathrm{m^3}$ , so $\Omega=\sqrt{GM/R^3}=\mathbf{1.99\times10^{-7}\ s^{-1}}$ and $P=2\pi/\Omega=1.0$ yr, as it must be at 1 AU.

Step 2. Maximum growth rate. $\gamma_{\max}=\tfrac34\Omega=\mathbf{1.49\times10^{-7}\ s^{-1}}$ , giving an e-folding time $\tau=1/\gamma_{\max}\approx0.078$ yr — under a month, versus billions of years for molecular viscosity in the same disk.

Step 3. E-foldings per orbit. $\gamma_{\max}P=\tfrac34\Omega\cdot(2\pi/\Omega)=\tfrac{3\pi}{2}=\mathbf{4.71}$ , so the amplitude grows by $e^{4.71}\approx111$ every orbit. The MRI amplifies by two orders of magnitude per orbit.

Step 4. Weaker field. Since $\gamma_{\max}=\tfrac34\Omega$ has no $v_A$ in it, a ten-times weaker field changes neither the rate nor the $4.71$ e-foldings — only the fastest wavelength $\lambda_{\max}\approx2\pi v_A/\Omega$ shrinks by ten. Growth is field-independent; only the wavelength scales with the field.

2 · Does the MRI fit inside the disk?

Express the fastest-growing wavelength as a fraction of the disk scale height in terms of the plasma parameter $\beta=P_{\text{gas}}/P_{\text{mag}}$ , and evaluate it for $\beta=10^3$ , $\beta=10$ , and $\beta=1$ . Which disks are MRI-active?

Step 1. Relate $v_A$ to $c_s$ : $\beta=\rho c_s^2/(B^2/2\mu_0)=2c_s^2/v_A^2$ , so $v_A/c_s=\sqrt{2/\beta}$ . A large $\beta$ (weak field) means a small Alfvén speed.

Step 2. With $\lambda_{\max}\approx2\pi v_A/\Omega$ and $H\approx c_s/\Omega$ , the $\Omega$ 's cancel: $\dfrac{\lambda_{\max}}{H}\approx\dfrac{2\pi v_A}{c_s}=2\pi\sqrt{\dfrac{2}{\beta}}$ . The MRI is supported when this is below about one.

Step 3. Evaluate. $\beta=10^3$ : $2\pi\sqrt{2/1000}=\mathbf{0.28}$ — a quarter of a scale height, MRI fully active. $\beta=10$ : $2\pi\sqrt{0.2}=\mathbf{2.8}$ — the mode overflows the disk, marginal. $\beta=1$ (equipartition): $2\pi\sqrt2=\mathbf{8.9}$ — no unstable wavelength fits, magnetically stabilized. Weakly magnetized disks ( $\beta\gg1$ ) are the MRI's natural home; equipartition fields shut it off.

Physics of the portrait

The magnetorotational instability is a rare case of a fundamental result discovered, forgotten, and rediscovered in a different context three decades later.

Origins · 1 minSteven Balbus and how this was discovered

Steven Balbus

b. 1953, American-British astrophysicist; with John Hawley (b. 1958) · magnetohydrodynamics of accretion disks

The mathematics was first written down for the laboratory, not for astrophysics. Evgeny Velikhov in 1959, and Subrahmanyan Chandrasekhar in his 1960 hydromagnetic-stability work, showed that magnetized Couette flow between differentially rotating cylinders is unstable when the angular velocity decreases outward — the exact criterion $\mathrm{d}\Omega^2/\mathrm{d}R<0$ . But their result was a curiosity of fluid dynamics; nobody connected it to accretion, and for thirty years disk theory limped along with an unexplained $\alpha$ . The decisive step came in 1991, when Steven Balbus and John Hawley, then at the University of Virginia, recognized this old instability as the missing engine of disk turbulence. Balbus supplied the linear analysis and the luminous "spring" intuition; Hawley, a pioneer of numerical MHD, followed it into the nonlinear regime with three-dimensional simulations that showed it saturating into the sustained turbulence disks needed. Their 1991 Astrophysical Journal paper and their 1998 Reviews of Modern Physics review reorganized the field: accretion, disk dynamos, and jet launching now all begin from the MRI. The two shared the 2013 Shaw Prize in Astronomy for turning a forgotten stability criterion into the foundation of high-energy astrophysics.

History

1959

— Velikhov finds magnetized Couette flow unstable when Ω decreases outward

1973

— Shakura and Sunyaev introduce the α-parameter for disk viscosity

1991

— Balbus and Hawley identify the instability as the origin of accretion-disk turbulence

1998

— the Balbus–Hawley Reviews of Modern Physics synthesis appears

2013

— Balbus and Hawley share the Shaw Prize in Astronomy

Going further

Texts. Balbus & Hawley (1991), ApJ 376, 214 — "A powerful local shear instability in weakly magnetized disks. I." · Balbus & Hawley (1998), Rev. Mod. Phys. 70, 1 — "Instability, turbulence, and enhanced transport in accretion disks" · Shakura & Sunyaev (1973), A&A 24, 337 — the original $\alpha$ -disk prescription · Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed. · Binney & Tremaine, Galactic Dynamics, 2nd ed. · Frank, King & Raine, Accretion Power in Astrophysics, 3rd ed. Key numerical checks: the quartic (3) reduces to a quadratic in $X=\omega^2$ with discriminant $\Omega^2(\Omega^2+16\omega_A^2)$ ; maximizing $-X_-$ for Keplerian $\kappa^2=\Omega^2$ , $\mathrm{d}\Omega^2/\mathrm{d}\ln R=-3\Omega^2$ gives $\gamma_{\max}=\tfrac34\Omega$ at $\omega_A^2=\tfrac{15}{16}\Omega^2$ ; the instability window is $\omega_A^2<3\Omega^2$ ; at 1 AU around $1\,M_\odot$ , $\Omega=1.99\times10^{-7}\ \mathrm{s^{-1}}$ and $\gamma_{\max}P=3\pi/2\approx4.71$ ; the fastest wavelength satisfies $\lambda_{\max}/H=2\pi\sqrt{2/\beta}$ ; and saturated simulations yield $\alpha\sim0.01$ – $0.1$ .

Prerequisites. Thin Accretion Disks and the Shakura–Sunyaev Model · Ideal Magnetohydrodynamics · Differential Rotation and Epicyclic Frequency. Leads to. Advection-Dominated Accretion Flows · Relativistic Jets and Launching Mechanisms · Eddington Luminosity and Accretion Limits.

Ask the tutor: Want to see why a milligauss field and a microgauss field grow at the very same rate but on different scales — walk through the $\omega_A^2=\tfrac{15}{16}\Omega^2$ maximization with me.

Why is a Keplerian disk unstable to the MRI when Rayleigh says it is stable?

Because the two criteria weigh different gradients. Rayleigh stability depends on the specific angular momentum $R^2\Omega$ , which increases outward for a Keplerian disk ( $\ell\propto R^{1/2}$ ), so a field-free disk is stable and just epicycles. The MRI depends instead on the angular velocity $\Omega$ , which decreases outward ( $\Omega\propto R^{-3/2}$ ): a field replaces conservation of each parcel's angular momentum with a tension enforcing near-corotation, and that torque is destabilizing precisely when $\Omega$ falls outward. So the same disk is Rayleigh-stable and MRI-unstable at once — no contradiction, just two different physics.

How can an infinitesimally weak field produce a growth rate as large as ¾ Ω?

Because the growth rate is a property of the shear, not the field. A weaker field is a softer spring that responds more slowly — which alone would slow growth — but it also allows longer unstable wavelengths, and the two effects cancel exactly. So the fastest mode always grows at $\tfrac34\Omega$ for a Keplerian disk, whether the field is a microgauss or a milligauss. The field is not idle: it sets the wavelength of that mode, $\lambda_{\max}\approx2\pi v_A/\Omega$ . Only when that wavelength exceeds the disk thickness does the MRI shut off.

What sets the strong-field cutoff of the MRI?

The requirement that the fastest wavelength fit inside the disk. With $\lambda_{\max}\approx2\pi v_A/\Omega$ and half-thickness $H\approx c_s/\Omega$ , fitting the mode means $v_A\lesssim c_s$ — magnetic pressure below gas pressure, plasma $\beta\gtrsim1$ . When $v_A\gtrsim c_s$ , every unstable wavelength is longer than the disk is thick and there is no room for it; tension then stabilizes rather than destabilizes. This is why the MRI is fundamentally a weak-field instability, even though its growth rate is field-independent.

What is the connection between the MRI and the α of the Shakura–Sunyaev model?

They are the same physics at two levels. Shakura and Sunyaev (1973) parameterized the disk viscosity as $\nu=\alpha c_s H$ , with $\alpha$ a number they could not derive — a stand-in for "whatever transports angular momentum". The MRI supplies the mechanism: its saturated turbulence produces correlated Reynolds and Maxwell stresses whose volume-averaged ratio to the pressure is $\alpha$ . Simulations measure $\alpha\sim0.01$ – $0.1$ , the Maxwell stress dominating. So $\alpha$ was never a fudge factor in principle — it is a measurable property of MRI turbulence, and computing it from first principles is a central achievement of the theory.

Does the MRI operate in every accretion disk?

No — it needs the gas well coupled to the field, requiring enough ionization for ideal MHD. In hot, well-ionized disks (around black holes, neutron stars, white dwarfs in outburst) the MRI is vigorous and is the accepted transport mechanism. But in the cold, dense midplanes of protoplanetary disks the ionization is tiny, and non-ideal effects — Ohmic and ambipolar diffusion, plus the Hall effect — damp it wherever the magnetic Reynolds number falls below unity. These "dead zones" have little turbulence, so mass piles up and dust settles, shaping where planetesimals form. Recent ALMA measurements support this and have revived magnetized disk winds as an alternative.

Is the MRI related to how jets are launched from disks?

Indirectly but importantly. The MRI's primary job is transport, but along the way it amplifies and reorganizes field, sustaining a dynamo that can build ordered, large-scale poloidal field threading the disk and corona. That field anchors the two leading jet-launching mechanisms: the magnetocentrifugal (Blandford–Payne) wind, which flings gas out along inclined field lines, and the Blandford–Znajek process, which taps a spinning black hole's rotation through field threading the horizon. So the MRI does not collimate a jet itself, but the turbulence and dynamo it drives supply the magnetic scaffolding jets require — linking the instability to the relativistic jets of blazars and radio galaxies.