Bipolar Outflows: A Forming Star Firing Jets From Both Poles

The angular momentum problem a protostar has to solve

Dense molecular cloud cores rotate slowly, but they are enormous. Mapping velocity gradients in nearby cores, Alyssa Goodman and collaborators (1993) found typical specific angular momenta of j ≈ 10²¹ cm² s^-1 on scales of ~0.1 parsec. Now compare that with a star. Material sitting at the surface of the Sun rotating at break-up would have j = √(GM_☉R_☉) ≈ 3 × 10¹⁸ cm² s^-1; the actual Sun, spinning once a month, has j ≈ 10¹⁵ cm² s^-1.

So the gas that must become a star arrives carrying roughly a few hundred times more angular momentum than a star can hold even if it were spinning itself apart, and about a million times more than the Sun actually has. Collapse cannot simply proceed. Long before it reaches the star, infalling gas spins up (j is conserved, so v_φ ∝ 1/r) and settles into a rotationally supported accretion disc. Something then has to remove angular momentum from that disc, or accretion stalls and no star forms.

Magnetic braking against the envelope helps on large scales. But the decisive mechanism, operating from ~0.05 AU out to a few AU, is a magnetised wind: material is flung outward along field lines anchored in the rotating disc, carrying angular momentum away with it, and is then squeezed into two opposed jets. That is a bipolar outflow.

Magnetocentrifugal launching: a bead on a rotating wire

The canonical picture comes from Roger Blandford and David Payne's 1982 paper on hydromagnetic flows from accretion discs. Inside the disc the gas dominates the magnetic field; above the disc surface, where densities plummet, the field dominates the gas. In that magnetically dominated region the field line acts like a rigid wire, forced to corotate with its footpoint at the Keplerian angular velocity Ω₀, and gas is a bead threaded on it.

Write the effective potential in the corotating frame, Φ_eff = -GM/(r²+z²)^1/2 - ½Ω₀²r². Expanding around the footpoint gives the famous criterion: if the poloidal field line is inclined by more than 30° from the rotation axis — equivalently, makes an angle of less than 60° with the disc plane — then Φ_eff decreases outward along the wire, and gas is centrifugally slung out with no thermal push at all. This is the magnetocentrifugal, or 'cold', launch.

The bead corotates until the flow speed reaches the Alfvén speed, at the Alfvén radius r_A. Beyond that, inertia wins and the field is dragged backward into a tight spiral. Two consequences follow, and both are quantitative:

• Lever arm. Each gram of wind carries away specific angular momentum Ω₀r_A², not Ω₀r₀². To let the disc accrete, you only need Ṁ_wind/Ṁ_acc ≈ (r₀/r_A)². For a modest lever arm r_A ≈ 3r₀, that is ~0.1 — precisely the ratio measured in real young stellar objects.
• Terminal speed. Energy conservation along the streamline gives v_∞ ≈ v_K(r₀)√(2λ-3) with λ = (r_A/r₀)². Launch from r₀ = 0.1 AU around a 0.5 M_☉ protostar gives v_K ≈ 67 km/s, so λ = 9 predicts v_∞ ≈ 260 km/s. Observed protostellar jets move at 100-300 km/s. The jet speed is essentially the escape speed from where it was launched.

Collimation is a separate step and it is magnetic, not centrifugal. Past the Alfvén surface the toroidal component grows until B_φ ≫ B_p, and the resulting hoop stress, ~B_φ²/4πr, pinches the flow onto the axis. That is why the wide-angle wind at the base becomes a jet with an opening angle of only a few degrees by 50-100 AU.

Disc wind or X-wind? Where the launching happens

Everyone agrees on magnetocentrifugal launching; the argument is about the footprint. Two families dominate:

• Extended disc wind (Ralph Pudritz & Colin Norman 1983; Arieh Königl 1989; Jonathan Ferreira 1997). Field lines thread the disc over decades in radius, from ~0.1 AU out to several AU, and each annulus contributes a nested shell of wind. Predicts a radially stratified jet: fast on the axis, slower and more rotating at the edges.
• X-wind (Frank Shu, Joan Najita and colleagues, 1994). The stellar magnetosphere truncates the disc near the corotation radius, ~0.05 AU for a T Tauri star, and essentially all the wind is launched from that single narrow annulus, the 'X-point'. Predicts a jet with one launch radius and a tight relation between accretion and ejection.

Also on the table are pure stellar winds enhanced by accretion (Matt & Pudritz), which help spin the star down but cannot carry the disc's angular momentum, and the 'magnetic tower' of Yutaka Uchida and Kazunari Shibata (1985), in which torsional Alfvén waves inflate a growing helical magnetic bubble.

The evidence is mixed and probably means both are right in different places. ALMA measurements of rotation in the SiO jet of HH 212 (Chin-Fei Lee and collaborators, 2017) imply a launch radius of only ~0.05 AU — X-wind-like — while the surrounding rotating SO/SO₂ wind in the same source, and HST/STIS measurements of DG Tau (Francesca Bacciotti et al. 2002), imply footpoints out to 0.5-3 AU, which requires an extended disc wind.

From a thin jet to a huge lobe: entrainment and bow shocks

What single-dish radio telescopes actually detect is not the jet. It is a pair of enormous, slow, molecular CO lobes, and their mass is mostly ambient cloud gas that the jet has swept up. The bookkeeping is momentum-driven: to good approximation Ṁ_lobev_lobe ≈ Ṁ_jetv_jet. A jet losing 10^-7 M_☉/yr at 200 km/s therefore drives ~2 × 10^-6 M_☉/yr of molecular gas at 10 km/s — a mass loading of about 20 — and injects a force of ~2 × 10^-5 M_☉ km s^-1 yr^-1, the range measured for Class 0 sources. Its kinetic luminosity, ½Ṁv², is a few tenths of a solar luminosity.

Where the jet rams the cloud, it terminates in a bow shock. Shock velocities of 20-100 km/s heat gas to 10⁴-10⁵ K, which then cools radiatively through a ~10⁴ K zone glowing in Hα, [O I] 6300 Å and [S II]. That glowing patch is a Herbig-Haro object, and the [S II] 6716/6731 doublet ratio gives its electron density directly, typically 10²-10⁴ cm^-3. Shocks faster than ~25 km/s also sputter silicon off dust grains, which is why SiO emission is a jet tracer but is nearly absent from quiescent clouds.

Jets are not steady. They are strings of knots, internal working surfaces where fast material rams slower material ejected earlier. Knot spacings correspond to ejection intervals of years to decades, so an HH chain is a stratigraphic record of the protostar's accretion history — the same episodic behaviour that, at the extreme, produces FU Orionis outbursts. Outflow cavities also widen with age, from opening angles of ~20-30° in deeply embedded Class 0 sources to more than 90° by the Class II stage.

How outflows are actually observed and measured

The observational history has two independent roots. Sherburne Wesley Burnham noticed a small nebulosity beside T Tauri in 1890; George Herbig (1951) and Guillermo Haro (1952) independently catalogued such knots, and more than a thousand HH objects are known today. Separately, in 1980 Ronald Snell, Robert Loren and Richard Plambeck mapped high-velocity CO around L1551 IRS 5 in Taurus and found two oppositely directed lobes — the first recognised bipolar molecular outflow. Only later were the optical jets and the CO lobes understood as one phenomenon.

Modern measurement uses several complementary handles:

• Proper motions. HH knots move at 0.1-0.4 arcsec/yr. At 400 pc, 0.2 arcsec/yr is ~380 km/s. The Hubble Space Telescope's time-lapse movies of HH 47 in the Gum Nebula (Patrick Hartigan and colleagues, 2011, combining 1994, 1999 and 2008 epochs) show the flow visibly moving and shocks evolving.
• Edge-on geometry. HH 30 in Taurus (~140 pc) is the textbook image: an opaque, flared dust disc a few hundred AU across, seen exactly edge-on, with a jet emerging perpendicular from each face (Christopher Burrows et al., HST, 1996). Its jet also wiggles on a ~100 year cycle, betraying an unseen companion.
• Interferometric line kinematics. ALMA resolves the SiO, CO, SO and H₂O structure of HH 212 in Orion — a jet so symmetric that Hans Zinnecker, Mark McCaughrean and John Rayner (Nature, 1998) called it 'symmetrically pulsed' — down to tens of AU, including the small dusty disc and transverse velocity shifts interpreted as jet rotation.
• Shocked molecular hydrogen. JWST's 2023 NIRCam images of HH 211 in Perseus and HH 46/47 map H₂ 2.12 μm emission from 1000-3000 K shocked gas, revealing knots and wiggles in outflows still buried in their envelopes.
• Radio. The VLA detects thermal free-free emission from the ionised jet base, giving mass-loss rates that dust extinction cannot hide, and 22 GHz water masers give VLBI-precision motions within tens of AU of the star.

Why outflows matter: feedback on the star and the cloud

Outflows are not a side effect; they are a regulator with three measurable jobs.

They enable accretion. Removing angular momentum at the ~10% mass cost implied by the lever-arm argument is what allows disc material to spiral inward at all. Whether magnetised winds or internal turbulence (the magnetorotational instability) dominate that transport is one of the live debates in disc physics; ALMA measurements of very low turbulent line widths in several discs have pushed opinion toward wind-driven accretion.

They limit the final stellar mass. Christopher Matzner and Christopher McKee (2000) showed that an outflow escaping through the poles unbinds a substantial fraction of the parent core, leaving a core-to-star efficiency of roughly 25-70%, most often near a third. This is the standard explanation for why the observed core mass function has the shape of the stellar initial mass function shifted to higher masses by about a factor of three — equivalently, why the IMF is the core mass function scaled down by that efficiency.

They stir the cloud. A single low-mass protostar injects of order 1-10 M_☉ km/s of momentum over its embedded lifetime of a few × 10⁵ years. In dense clustered regions such as NGC 1333 that is enough to sustain the observed turbulence and puncture the cloud with cavities; on the scale of a whole giant molecular cloud the momentum budget falls short, and supernovae and ionising radiation take over.

Look-alikes, failure modes and open questions

Things that look like protostellar outflows but are not. Bipolar pre-planetary and planetary nebulae — the Egg Nebula, the Calabash Nebula — also show twin lobes and fast collimated jets, but they are shaped at the end of a star's life, most likely by a binary companion and a short-lived disc, not by ongoing protostellar accretion. Relativistic jets from active galactic nuclei share the magnetocentrifugal physics but add black-hole spin extraction (Blandford-Znajek) and Lorentz factors of tens. Photoevaporative disc winds are slow (~10 km/s), thermally driven and essentially uncollimated. And the spectacular Orion BN/KL 'explosive' outflow is not a disc wind: it is debris from a dynamical encounter that ejected several stars about 500 years ago.

Where the standard picture strains. Jet rotation is the key test of magnetocentrifugal launching, and it is genuinely hard: transverse velocity shifts of only 5-30 km/s can be mimicked by asymmetric shocks or precession, and RW Aurigae notoriously appeared to reverse its rotation sense between epochs. Whether jets carry all the required angular momentum, or only the fraction visible in the atomic jet with the rest hidden in a slower wide-angle wind, is unsettled. Ideal-MHD simulations of collapsing cores with realistic magnetisation (mass-to-flux ratios λ ≈ 2-3) suffer the magnetic braking catastrophe: braking is so efficient that no rotationally supported disc forms at all. Fixing it requires non-ideal effects — ambipolar diffusion, Ohmic dissipation and the Hall effect, the last of which makes disc size depend on the sign of the field relative to the rotation axis — plus turbulence and field-rotation misalignment.

Open questions worth watching: why current-driven kink instabilities do not shred real jets; whether the mechanism scales cleanly to high-mass protostars, as the rotating SiO wind from Orion Source I (Tomoya Hirota and colleagues, 2017) suggests; and exactly how episodic accretion, jet knots and FU Orionis outbursts are causally linked.

The nested layers of a protostellar bipolar outflow, from the rotation axis outward

Layer	Typical speed	Traced by	What it tells you
Collimated atomic / SiO jet	100-300 km/s	[S II] 6716/6731, [O I] 6300, [Fe II] 1.64 um, SiO with ALMA	Launch radius, mass-loss rate, jet rotation
Wide-angle molecular wind	~5-100 km/s	SO, SO2, H2O, CO high-velocity wings	Likely carrier of much of the extracted angular momentum
Entrained CO cavity / lobe	1-30 km/s	CO J=2-1 and 3-2 (ALMA, IRAM, JCMT)	Total momentum injected into the cloud
Bow shock / Herbig-Haro object	shock velocity 20-100 km/s	H-alpha, [S II], H2 2.12 um (HST, JWST, VLT)	Episodic ejection history, cooling-zone densities
Quiescent ambient core	~0.2-1 km/s (turbulent)	N2H+, NH3, dust continuum	The reservoir being disrupted and dispersed