Magnetorotational Instability

The puzzle the MRI solves

Accretion disks are everywhere a star, a black hole, or a young planet is being fed: protoplanetary disks around T Tauri stars, the disks in cataclysmic variables and X-ray binaries, and the colossal disks around supermassive black holes in active galactic nuclei. In every case the gas orbits in nearly circular, Keplerian paths, where the angular velocity falls off as Ω ∝ R^−3/2. For that gas to actually fall in and be accreted, it must lose angular momentum — and that requires friction inside the disk.

Here is the problem. The ordinary molecular viscosity of disk gas is almost nothing. Plug realistic numbers into the diffusion time and the gas would take on the order of 10¹² years to drift inward — far longer than the age of the universe. Yet we watch disks evolve in thousands to millions of years: dwarf novae erupt every few weeks, FU Orionis stars dump matter in decades, AGN flicker on human timescales. Something inside the disk must be vastly more efficient at moving angular momentum than molecular collisions. For decades that "something" was simply hidden inside the famous α-viscosity of Shakura and Sunyaev (1973): the effective viscosity was written as ν = α c_s H, with α an unknown fudge factor of order 0.01–0.1, and nobody knew what physical process α stood for.

The magnetorotational instability is that process. Balbus and Hawley showed in 1991 that a magnetized, differentially rotating disk is always linearly unstable as long as the angular velocity decreases outward — which it always does in a Keplerian disk. The instability does not need a strong field; an arbitrarily weak one will do. It grows on the orbital timescale and saturates as vigorous MHD turbulence whose Maxwell and Reynolds stresses transport angular momentum outward exactly as required.

The spring-and-bead picture

The mechanism is beautifully simple once you picture it. Imagine two parcels of gas at slightly different radii, both orbiting the central mass, and link them with a single magnetic field line. Because the field is "frozen" into the highly conducting plasma (flux freezing), that field line behaves like a spring connecting the two parcels.

The inner parcel orbits faster (Kepler's third law), so it races ahead, stretching the field line. The stretched line pulls back with magnetic tension. That tension acts as a brake on the inner parcel — it removes angular momentum from it, so the inner parcel drops to a smaller orbit. The same tension acts as a thrust on the outer parcel — it adds angular momentum, so the outer parcel climbs to a larger orbit. But moving them apart stretches the spring even more, which increases the tension, which separates them further still. That is a runaway: a positive feedback loop powered by the disk's differential rotation. The inner gas spirals in, the outer gas moves out, and angular momentum flows outward — exactly what accretion demands.

The counterintuitive part is that adding momentum to the outer parcel makes it move outward and slower, while braking the inner parcel makes it move inward and faster. In a gravitational orbit, taking angular momentum away makes you sink and speed up. So the field line keeps feeding off the velocity difference it just amplified. The instability shuts off only when the field has grown strong enough that magnetic tension can no longer be stretched the relevant distance inside the disk.

The numbers that matter

The MRI's behavior is governed by a handful of comparisons between length and time scales. The maximum growth rate is a striking ¾ Ω, meaning the instability e-folds in less than one orbital period — it is explosively fast in disk terms. The fastest-growing wavelength scales with the Alfvén speed v_A = B / √(4πρ): roughly λ ≈ 2π v_A / Ω. This sets the two ends of the field-strength window.

Field-strength regimes for the MRI in a disk

Field strength	Unstable wavelength	Outcome
Zero / non-magnetic	None	Hydrodynamically stable (Rayleigh criterion satisfied); no transport
Weak (vA ≪ cs)	Short, fits inside disk	MRI active; grows to turbulence, α ≈ 0.01–0.1
Moderate (vA ≲ cs)	Comparable to scale height H	Maximally efficient transport
Too strong (vA > cs·order unity)	Longer than H — doesn't fit	MRI suppressed; magnetic tension overstabilizes

Crucially, a purely hydrodynamic Keplerian disk is stable by the Rayleigh criterion — its specific angular momentum increases outward, which damps perturbations. The magic of the MRI is that the magnetic field destabilizes a flow that would otherwise be perfectly stable. The relevant criterion flips from "specific angular momentum increases outward" (hydro, stable) to "angular velocity decreases outward" (MHD, unstable). Since dΩ/dR < 0 holds in essentially every astrophysical disk, the MRI is nearly universal — wherever the gas is conducting enough to be coupled to a field.

How the MRI compares with other proposed transport mechanisms

Mechanism	Effective α	Where it works	Status
Molecular viscosity	~10⁻¹⁰	Everywhere, negligibly	Far too weak
MRI turbulence	0.01–0.1	Ionized, magnetized disks	Leading mechanism
Gravitational instability / spiral waves	up to ~1	Massive, self-gravitating disks	Important early, briefly
Disk winds (magnetocentrifugal)	n/a (removes L vertically)	Dead-zone disks, jet launching	Complementary to MRI
Convective / hydro turbulence	≲10⁻³, often inward	Generic disks	Too weak, wrong sign

Dead zones and where the MRI fails

The MRI's one requirement is that the field be well coupled to the gas — which demands a minimum level of ionization. In hot, well-ionized environments (the inner regions of X-ray binary disks, AGN, the surfaces of protoplanetary disks irradiated by stellar X-rays and cosmic rays) this is easily met. But in the cold, dense midplane of a protoplanetary disk, the ionization fraction can drop so low that three non-ideal MHD effects — Ohmic resistivity, the Hall effect, and ambipolar diffusion — let the field slip through the neutral gas and quench the instability. The result is a dead zone: a quiescent, low-turbulence region sandwiched between MRI-active surface layers.

Dead zones are not a footnote — they may be where planets form. Suppressing turbulence at the midplane allows dust to settle, drift, and concentrate (for instance via the streaming instability) into the dense clumps that become planetesimals. So the very failure of the MRI in one region may be a precondition for building planets, while in disk surface layers and in hot inner disks the MRI keeps the accretion machinery running. In these dead-zone disks, angular momentum may instead be carried off vertically by magnetized disk winds rather than radially by turbulence — an active area of research feeding directly into ALMA observations of disk structure and the interpretation of low measured turbulent velocities.

Evidence and where to look

• Numerical simulations. Local shearing-box and global MHD simulations since the mid-1990s reliably reproduce sustained turbulence with outward angular-momentum transport and α ≈ 0.01–0.1, matching the Shakura–Sunyaev prescription.
• Dwarf-nova outbursts. The thermal-viscous instability that drives these eruptions requires α to jump between roughly 0.01 (quiescence) and 0.1–0.2 (outburst) — values consistent with MRI turbulence in different ionization states.
• Protoplanetary disks. ALMA finds surprisingly low turbulent line-broadening at disk midplanes (some < 0.05 c_s), direct observational support for dead zones where the MRI is quenched.
• Laboratory analogs. Liquid-sodium and plasma Couette experiments (e.g. the Princeton MRI experiment) attempt to excite the instability in the lab — extremely hard because real fluids have far higher resistivity than astrophysical plasmas.

Why the MRI matters

• It powers accretion. Every luminous accretion source — quasars, X-ray binaries, protostars — relies on angular-momentum transport the MRI naturally provides.
• It explains α. It turns the Shakura–Sunyaev fudge factor into real physics.
• It is nearly universal. Any disk with dΩ/dR < 0 and enough ionization is MRI-unstable.
• It shapes planet formation. Its dead zones may be the cradles of planetesimals.
• It seeds jets and dynamos. MRI turbulence amplifies and orders magnetic fields, feeding disk winds and relativistic jets.