See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/253309918

X-Winds Theory and Observations

Article · May 2000

CITATIONS READS

248 185

4 authors, including:

Frank Shu Hsien Shang

University of California, San Diego Academia Sinica
255 PUBLICATIONS   20,665 CITATIONS    109 PUBLICATIONS   3,609 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Star-Disk Interactions View project

Origins of the Solar System View project

All content following this page was uploaded by Hsien Shang on 02 September 2014.

The user has requested enhancement of the downloaded file.

X-WINDS: THEORY AND OBSERVATIONS

FRANK H. SHU
University of California at Berkeley

JOAN R. NAJITA
Space Telescope Science Institute

HSIEN SHANG
University of California at Berkeley
and
ZHI-YUN LI
University of Virginia at Charlottesville

We review the theory of x-winds in young stellar objects (YSOs), and we com-
pare its predictions with a variety of astronomical observations. Such flows arise
magnetocentrifugally from accretion disks when their inner edges interact with
strongly magnetized central stars. X-winds collimate logarithmically slowly into
jets, and their interactions with the surrounding molecular cloud cores of YSOs
yield bipolar molecular outflows.

I. INTRODUCTION
.
Strong magnetic fields can considerably enhance the mass loss Mw of ther-
mally driven winds from the surfaces of rapidly rotating stars (Mestel
1968). Even if the resultant. flow is quite cold, Hartmann and MacGregor
(1982) demonstrated that Mw could have almost arbitrarily large values,
dependent only on the ratio of the azimuthal and radial field strengths at the
position where the gas is injected onto open field lines near the equator of a
protostar that rotates near breakup. Shu et al. (1988) assigned the cause for
the protostar to spin at breakup to a circumstellar disk that abuts. against
the surface of the central object and accretes onto it at a high rate MD . They
also replaced Hartmann and MacGregor’s (1982) arbitrary angle of injec-
tion for gas velocity and magnetic field direction with the requirement that
in steady state, the wind
. mass. loss rate must be a definite fraction f of the
disk accretion rate Mw ⳱ f MD (see below).
Blandford and Payne (1982) advanced an influential self-similar
model of centrifugally driven winds from the surfaces of magnetized

[789]
790 F. H. SHU ET AL.

accretion disks. Pudritz and Norman (1983) applied these pure disk wind
models to bipolar outflows, and Ko¨ nigl (1989) investigated how the wind
might smoothly join a pattern of accretion flow inside the disk. Heyvaerts
and Norman (1989) studied how the winds might collimate asymptotically
into jets, while Uchida and Shibata (1985) and Lovelace et al. (1991) ad-
vocated alternative driving mechanisms in which magnetic pressure gra-
dients play a bigger role in the acceleration of a disk-blown wind.
Motivated by the problem of binary X-ray sources, a parallel line of re-
search developed concerning how magnetized stars accrete from surround-
ing disks. Ghosh and Lamb (1978) used order-of-magnitude arguments to
show that a strongly magnetized star would truncate the surrounding ac-
cretion disk at a larger radius than the stellar radius R多 and divert the equa-
torial flow along closed field-line funnels toward the polar caps. Although
Ghosh and Lamb thought that this inflow would spin the central object up
faster than if the accretion disk had extended right up to the stellar surface,
Ko¨ nigl (1991) made the surprising and insightful suggestion that the pro-
cess might torque down the star and account for the relatively slow rate of
spin of observed T Tauri stars. Observational support for magnetospheric
accretion was subsequently marshaled by Edwards et al. (1994) and Hart-
mann et al. (1994).
Arons, McKee, and Pudritz (Arons 1986) proposed that a centrifu-
gally driven outflow accompanies the funnel inflow (see also Camenzind
1990). Independently, Basri (unpublished) arrived observationally at the
same suggestion by extending the synthesis work by Bertout et al. (1988)
on ultraviolet excesses in T Tauri stars. Shu et al. (1994a) put these ideas
together into a concrete proposal that generalized the earlier x-wind model
of Shu et al. (1988). A related proposal, the so-called “magnetic propeller”
(e.g., Li and Wickramasinghe 1997; Lovelace et al. 1999), has been in-
voked recently to explain the spindown of the cataclysmic variable AE
Aquarii (Wynn et al. 1997) and the Rossi X-ray Timing Explorer (RXTE)
observations of X-ray pulsars GX 1Ⳮ4 and GRO J1744-28 (Cui 1997). A
quick and somewhat oversimplified summary might be that x-wind theory
adds the possibility of outflow to magnetospheric accretion models, while
the magnetic propeller idea adds the possibility of time dependence.

II. GENERALIZED X-WIND MODEL

In the generalized x-wind model, if the star has mass M多 and magnetic
dipole moment ␮多 , the gas disk is truncated at an inner radius,
1/7
Rx ⳱ ⌽dx ⫺4/7
冢 ␮多4
.
GM多 MD2 冣 (1)

A disk of solids may extend inward of Rx to the evaporation radius of

calcium-aluminum silicates and oxides (see Shang et al. 1997 and Meyer
 X-WINDS: THEORY AND OBSERVATIONS 791

et al. 1997). In equation (1), ⌽dx is a dimensionless number of order unity

that measures the amount of magnetic dipole flux that has been pushed
by the disk accretion flow to the inner edge of the disk. In the closed
dead-zone model of Najita and . Shu
. (1994) and Ostriker and Shu (1995),
⌽dx ⳱ 2␤ w f 1/2 , with f ⳱ Mw /MD and ␤ w defined as the streamline-
averaged ratio of magnetic field to wind mass flux. This expression for
the coefficient ⌽dx holds when magnetic flux is swept toward Rx from
both larger and smaller radii (see below), and the magnetic flux trapped in
a small neighborhood of Rx is 32 times larger than the pure dipole value.
For a Keplerian disk, the inner edge of the disk rotates at angular speed
1/2

冢 冣
GM多
⍀x ⳱ (2)
Rx3

To satisfy mass and angular .momentum

. balance, the disk accretion divides
.
at Rx int.o a wind fraction, Mw ⳱ f MD , and a funnel-flow fraction, M多 ⳱
(1 ⫺ f )MD , where
1⫺J多 ⫺␶
f ⳱ (3)
Jw ⫺J多

In equation (3) J 多 and J w are, respectively, specific angular momenta

nondimensionalized in units of Rx2 ⍀ x and averaged over funnel and wind
streamlines, and ␶ is the negative of the viscous torque
. of the disk, ⫺᐀ ,
acting on its inner edge and measured in units of MD Rx2 ⍀ x . In the ap-
proximation that the embedded stellar fields force the x-region to be only
weakly differentially rotating (Shu et al. 1994a,b), ␶ is small compared to
unity.
In steady state, the star is regulated to corotate with the inner disk
edge,

⍀多 ⳱ ⍀x (4)

If corotation did not hold, the system would react to reduce the discrep-
ancy between ⍀ 多 and ⍀ x , where ⍀ x is given by equation (2) with Rx
determined by how much magnetic field there is to maintain a certain av-
erage standoff distance for the stellar magnetosphere [see equation (1)].
For example, suppose the star turns faster than the inner edge of the disk,
⍀ 多 ⬎ ⍀ x . The field lines attached to both would then continuously wrap
into ever tighter trailing spirals, with the field lines adjacent to the star
tugging it backward in the sense of rotation. This tug decreases the star’s
angular rate of rotation ⍀ 多 to more nearly equal the rate ⍀ x . Conversely,
imagine that ⍀ 多 ⬍ ⍀ x . With the star turning more slowly than the inner
edge of the disk, the field lines attached to both would continuously wrap
into ever tighter leading spirals, with the field lines adjacent to the star
tugging it forward in the sense of rotation. This tug increases the star’s
792 F. H. SHU ET AL.

angular rate of rotation ⍀ 多 , again to more nearly equal the rate ⍀ x . In true
steady state, ⍀ 多 ⳱ ⍀ x , and the funnel flow field lines acquire just enough
of a trailing spiral pattern (but without continuously wrapping up) that
the excess of material angular momentum brought toward the star by the
inflowing gas is transferred outward by magnetic torques to the footpoints
of the magnetic field in the disk (Shu et al. 1994a,b).
The x-wind gas gains angular momentum, and the funnel gas loses
angular momentum, at the expense of the matter at the footpoint of the
field in, respectively, the outer and inner parts of the x-region. As a conse-
quence, this matter, and the field lines across which it is diffusing, pinch
toward the middle of the x-region. In reality, Shu et al. (1997) point out
that the idealized steady state of exact corotation probably cannot be main-
tained because of dissipative effects. Two surfaces of null poloidal field
lines (labeled as “helmet streamer” and “reconnection ring” in Fig. 1) me-
diate the topological behavior of dipolelike field lines of the star, opened
field lines of the x-wind, and trapped field lines of the funnel inflow em-
anating from the x-region. Across each of the null surfaces, which begin
or end on “Y-points” (called “kink points” by Ostriker and Shu 1995), the
poloidal magnetic field suffers a sharp reversal of direction. By Ampere’s `
law, large electric currents must flow out of the plane of the figure along
the null surfaces. Nonzero electrical resistivity would lead to the dissi-
pation of these currents and to the reconnection of the oppositely directed
field lines (see, e.g., Biskamp 1993). The resultant reduction of the trapped
r
me

Coronal
trea

Coronal
Wind Wind
et S
m
Hel

Helmet
Dome
Soft
X-Rays
Funnel Flow

X-Wind
Soft Reconnection
Dipole X-Rays Ring
Star Field
X-Region Accretion Disk

Figure 1. Schematic drawing of the x-wind model. Coronal winds from the star
and the disk may help the x-wind to open field lines surrounding the helmet
streamer, but this aspect of the configuration is not central to the model.
 X-WINDS: THEORY AND OBSERVATIONS 793

magnetic flux in the x-region as the fan of field lines presses into the evac-
uated region of the annihilated fields would change the numerical value of
the coefficient ⌽dx in equation (1).
When Rx changes, the angular speed ⍀ x of the footpoint of magnetic
field lines in the x-region will vary according to equation (2). However,
the considerable inertia of the star prevents its angular velocity ⍀ 多 from
changing on the timescale of magnetic reconnection at the null surfaces
of the magnetosphere. The resulting shear when ⍀ 多 苷 ⍀ x will stretch
and amplify field lines attached to both the star and the disk. The poloidal
field will bulge outward from the increased magnetic pressure, insert-
ing more magnetic flux into the fan of field lines emanating from the x-
region [see the simulations of Linker and Mikic´ (1995) and Hayashi et
al. (1996)]. Dynamo action inside the star would presumably replace the
upward-rising dipolelike poloidal fields. This dynamo action would be en-
hanced by the wrapping of field lines between the star and disk. Averaged
over long times, we envisage a secular balance, with enough dynamo-
generated poloidal field being inserted into the x-region to balance the rate
of field dissipation at the null surfaces.

III. THE CENTRAL MACHINE: CHECKS

WITH OBSERVATIONS
In lieu of spatially resolved interferometric images (which will become
available within the next two decades), the picture just presented of the
central machine that drives collimated outflows from young stellar objects
(YSOs) is subject to many different kinds of spectroscopic checks. Here
we comment only on four crucial observations: (a) the demonstration that
rapidly rotating disks do extend almost to the surfaces of YSOs of sun-
like masses, (b) the demonstration that the central stars are magnetized
to the extent necessary to truncate the disks at an inner radius Rx com-
patible with locking stellar rotation rates to their observed low values, (c)
the demonstration that the final accretion process onto the central stars is
magnetically channeled, and (d) the demonstration that such funnel flows
are accompanied by outflows that have the requisite geometry and phase
relationship implied by x-wind theory.
A. Existence of Rapidly Rotating Inner Disks
The best evidence to date for the existence of inner disks comes from high-
resolution spectroscopy of CO overtone emission in young stars (see the
chapter by Najita et al., this volume). Through detailed modeling of CO
overtone emission it is possible to constrain the outer and, in particular,
inner radii for the emission. For example, in the case of the Class I source
WL16, the overtone emission is found to arise from disk radii between 5
and 30 R䉺 . The outer cutoff of 30 R䉺 probably arises from lack of exci-
tation and has little other physical significance. The inner cutoff of 5 R䉺
794 F. H. SHU ET AL.

may arise either because of CO dissociation or because of a real truncation

of the disk at an inner edge.
The latter interpretation is compatible with the emitting area required
to explain the strength of the Brackett ␥ line at line center. The dramatic
agreement between the Br␥ profile for this source (Najita et al. 1996) and
model Br␥ spectra for accreting magnetospheres (Muzerolle et al. 1998;
see also section III.C below) is suggestive evidence in favor of the claim
that WL16 has a strong stellar magnetosphere that truncates its surround-
ing disk, and through which disk matter accretes onto the stellar surface
(Fig. 1). From such evidence, we conclude that one-half of the two crucial
ingredients needed for x-wind theory to hold, a rapidly rotating, gaseous,
truncated inner disk, does exist in at least one well-studied source found
in nature.

B. Strong Magnetization of the Central Stars

If one eliminates Rx from equations (1), (2), and (4), one gets the magnetic
dipole moment of the star,
.
␮多 ⳱ ⌽dx (GM多 )5/6 MD1/2 ⍀ 多⫺7/6 (5)
required to enforce, in steady state, .a given stellar spin rate ⍀ 多 in a YSO
of mass M多 and. disk accretion rate MD . We should not expect equation (5)
to hold when MD drops to very low rates, because the remaining reservoir
of mass and angular momentum coming through the disk may not then
suffice to overcome the inertia of the central star and continue to regulate
its spin period. Typical conditions for a classical. T Tauri star might be
marginal for the purpose at hand: M多 ⳱ 0.5 M䉺 , MD ⳱ 3 ⫻ 10⫺8 M䉺 yr⫺1 ,
and ⍀ 多 ⳱ 2␲ /(8 dy) (Bertout 1989; Edwards et al. 1993; Gullbring et al.
1998). With ␤ w ⬃ 1 and f ⬃ 13 , so that ⌽dx ⬃ 2/ 冪3, and an equatorial
field strength B多 given ideally by B多 ⳱ ␮多 /R多3 , where R多 ⳱ 2 R䉺 is a
typical stellar radius, equation (5) now implies B多 ⳱ 2.1 kG, a measurable
value.
Johns-Krull et al. (1999) have determined the Zeeman broadening of
magnetically sensitive lines in the classical T Tauri star BP Tauri. [Sev-
eral other sources show similarly large fields (C. M. Johns-Krull, personal
communication, 1998); see also the early work of Basri et al. 1992.] Johns-
Krull and coworkers (1999) have compared their results with the predicted
relation (5) and find good agreement. These quantitative measures of the
stellar magnetic field buttress the inference from X-ray flaring activity that
T Tauri stars are strongly magnetized (Montmerle et al. 1993). One of the
more remarkable recent developments in this field is the discovery that
even Class I sources are strong X-ray emitters (see the chapter in this vol-
ume by Glassgold et al.), indicating that stellar magnetic activity begins
early in young stars. We conclude that the other half of the two crucial in-
gredients needed for x-wind theory to hold, a strong stellar magnetic field,
also does exist for several sunlike YSOs found in nature.
 X-WINDS: THEORY AND OBSERVATIONS 795

C. Magnetically Channeled Accretion onto the Central Star

Since the work of Ghosh and Lamb (1978), workers in the field of X-ray
binaries have concurred that if the central object is sufficiently strongly
magnetized, the last stages of mass accretion from a surrounding disk may
be channeled magnetically onto the central (neutron) star rather than con-
tinue to the star’s surface by viscous inspiraling. Only in the case of YSOs,
however, do we have direct spectroscopic evidence of this process in the
shape of predicted line profiles.
It now appears that the bulk of the hydrogen line emission in young
stars arises not in outflowing gas, as was long believed, but in inflowing
gas located within several stellar radii of the stellar surface. Outflow sig-
natures are more often seen in H␣ and Na I, and inflow signatures are
commonly seen in the higher Balmer and Brackett ␥ lines (Edwards et al.
1994; Najita et al. 1996). The comparison of hydrogen line profiles with
models of the kinematics and excitation conditions expected in magneto-
spheric accretion flows supports the idea that the emission originates in
hot gas inflowing along stellar magnetic field lines (Muzerolle et al. 1998;
see also the chapter in this volume by Najita et al.).
The reader should not conclude from this description, however, that
the funnel flow in x-wind theory is identical to the magnetically channeled
infall envisioned by Ghosh and Lamb (1978; see also the comments of Shu
et al. 1994a and Ostriker and Shu 1995). In Ghosh and Lamb’s (1978)
theory, all stellar field lines are closed, and the braking torques exerted by
field lines that thread distant radii in the disk balance, in steady state, the
accelerating torques exerted by field lines that thread nearby radii in the
disk. The “steady state” in this picture is intrinsically statistical and violent
in nature (and therefore not computed in any detail), because it involves
rapid field wrapping and magnetic reconnection. Because higher latitudes
on the star are tugged more slowly by field lines that thread the disk more
distantly than field lines from lower latitudes, such a picture would pre-
sumably result in a central object that has considerable differential rotation
and no well-defined stellar period, contrary to many observations of the ro-
tation properties of classical T Tauri stars (e.g., Vogel and Kuhi 1981; Ry-
dgren and Vrba 1983; Walter et al. 1988; Bouvier et al. 1993; Johns-Krull
1996).
In x-wind theory, closed stellar field lines that thread a nearly uni-
formly rotating, small, x-region of the disk (funnel-flow field lines) trans-
fer in steady state the excess angular momentum contained in the inflowing
matter, not to the star but to the inner portions of the x-region. This ex-
cess angular momentum is then removed from the outer portions of the x-
region by the opened field lines of the YSO outflow. [Angular-momentum
transport from the inner portions to the outer portions of the x-region is
probably accomplished by the magnetorotational instability (Balbus and
Hawley 1991) and has not yet been examined in any detail.] In pure x-wind
theory, the opened field lines came originally from the star (see Fig. 1), and
796 F. H. SHU ET AL.

therefore the magnetocentrifugally driven outflow bears definite geometric

and phase relationships to the magnetocentrifugally driven inflow.
D. Geometric and Phase Relationships
As mentioned above, outflow and inflow absorption components are of-
ten observed simultaneously and are now believed to arise in the cooler
wind and in accretion footpoints, respectively. Periodograms of the hy-
drogen lines in T Tauri stars can be used to probe the dynamical rela-
tionship between the wind, the funnel flow, and the rotation period of
the star. Synoptic monitoring by Johns and Basri (1995b) of the T Tauri
star SU Aurigae revealed inflow and outflow components to be variable
at the photometric (rotation) period of the star but separated in phase by
180⬚ . Johns-Krull and Hatzes (1997) have since found another system
with similar properties. If the funnel flow and collimating wind emanate
in these objects from the surrounding accretion disks, then the source re-
gions for the flowing gas corotate with the central stars, in agreement with
the prediction that ⍀ 多 ⳱ ⍀ x . Moreover, the phase lag of 180⬚ is consis-
tent with an origin for both components in the funnel flow and x-wind
induced by a tilted magnetic dipole, where field lines in one part of the
magnetosphere naturally bend inward, promoting accretion, and field lines
180⬚ away bend outward, promoting mass loss (see Fig. 15 of Johns and
Basri 1995b). Unfortunately, the majority of objects studied by Johns and
Basri (1995a) do not show such periodicity, indicating that even if ⍀ 多
does equal ⍀ x as a long-term average, ⍀ 多 may not generally equal ⍀ x
instantaneously.

IV. OUTLINE OF MATHEMATICAL FORMULATION

A mathematical formulation of the steady-state problem when ⍀ 多 ⳱ ⍀ x
was given in outline by Shu et al. (1988) and in detail by Shu et al. (1994b),
who n. ondimensionalized the governing equations by introducing Rx , ⍀ x ,
and Mw /4␲Rx3 ⍀ x , respectively, as the units of length, time, and density.
Assuming axial symmetry and time independence in a frame that corotates
with ⍀ x , we may then introduce cylindrical coordinates (␼, ␸, z ) and a
streamfunction ␺ (␼, z ) that allows the satisfaction of the equation of con-
tinuity, ⵜ ⭈ (␳ u) ⳱ 0, in the meridional plane:
1 ⭸␺ 1 ⭸␺
␳u␼ ⳱ , ␳uz ⳱ ⫺ (6)
␼ ⭸z ␼ ⭸␼
Field freezing in the corotating frame, B ⫻ u = 0, implies that the magnetic
field is proportional to the mass flux, B ⳱ ␤␳ u, where ␤ is a scalar. The
condition of no magnetic monopoles, ⵜ ⭈ B ⳱ 0, now requires that ␤ be
conserved on streamlines, i.e.,
␤ ⳱ ␤ (␺ ) (7)
 X-WINDS: THEORY AND OBSERVATIONS 797

Similarly, the conservation of total specific angular momentum requires

that the amount carried by matter in the laboratory frame, ␼ (␼ ⍀ x Ⳮ u␸ ),
where ⍀ x is replaced by 1 in dimensionless equations, plus the amount
carried by Maxwell torques, ⫺␼B␸ B, per unit mass flux, ␳ u, equals a
function of ␺ alone:
␼ [(␼ Ⳮ u␸ ) ⫺ ␤ 2 ␳u␸ ] ⳱ J (␺ ) (8)

Finally, conservation of specific “energy” in the corotating frame for a

gas with dimensionless isothermal sound speed ⑀ ⬅ ax /Rx ⍀ x , results in
Bernoulli’s theorem:
1 2
兩u兩 Ⳮ ᐂeff Ⳮ ⑀ 2 ln ␳ ⳱ H (␺ ) (9)
2
Up to an arbitrary constant, defined so that ᐂeff ⳱ 0 at the x-point ␼ ⳱ 1
and z ⳱ 0, ᐂeff is the dimensionless gravitational potential plus centrifugal
potential (measured in units of GM多 /Rx ⳱ ⍀ 2x Rx2 ):

3 1 ␼2
ᐂeff ⳱ ⫺ ⫺ (10)
2 (␼ Ⳮ z )
2 2 1/2 2
Equations (7)–(9), with ␤ , J , and H arbitrary functions of ␺ , repre-
sent formal integrations of the governing set of equations. The re-
maining equation for the transfield momentum balance, the so-called
Grad-Shafranov equation, cannot be integrated analytically and reads

ⵜ ⭈ (Ꮽⵜ␺ ) ⳱ ᏽ (11)

where Ꮽ is the Alfven

´ discriminant,

␤ 2␳ ⫺ 1
Ꮽ⬅ (12)
␼ 2␳
and ᏽ is a source function for the internal collimation (or decollimation)
of the flow:

ᏽ⳱␳ 冋␼
u␸
J ⬘(␺ ) Ⳮ ␳ 兩u兩2 ␤␤ ⬘(␺ ) ⫺ H ⬘(␺ ) 册 (13)

with primes denoting differentiation with respect to the argument ␺ .

The quantity Ꮽ ⳱ 0, i.e., ␤ 2 ␳ ⳱ 1, when the square of the flow
´ speed, 兩B兩2 /␳ . For sub-Alfvenic
speed 兩u兩2 equals the square of the Alfven ´
flow, Ꮽ ⬎ 0; for super-Alfvenic
´ flow, Ꮽ ⬍ 0. If Ꮽ were freely specifiable
(which it is not), equation (11) would resemble the time-independent heat
conduction equation. What is spreading in the meridional plane of our
problem, however, is not heat, but streamlines.
798 F. H. SHU ET AL.

V. FIXING THE FREE FUNCTIONS

When combined with Bernoulli’s equation (9), the Grad-Shafranov
equation (11) has three possible critical surfaces associated with it, corre-
sponding to slow magnetohydrodynamic (MHD), Alfven, ´ and fast MHD
crossings (Weber and Davis 1967). Thus, when applied to the problem of
the x-wind, equation (11) is a second-order partial differential equation
of elliptic type interior to the fast surface and of hyperbolic type exterior
to this surface (see Heinemann and Olbert 1978 and Sakurai 1985). The
unknown functions ␤ (␺ ), J (␺ ), and H (␺ ) are to be determined self-
consistently so that the three crossings of the critical surfaces are made
smoothly. This does not fix all three functions uniquely, because the loci
of the critical surfaces in (␼, z )-space are not known in advance. It turns
out that we can choose one of the loci freely. Alternatively, we can freely
specify one of the functions ␤ , J , or H . The method of Najita and Shu
(1994) fixes in advance the locus of the Alfven ´ surface and determines
all other quantities self-consistently from this parameterization. Shang
and Shu (1998) have invented a simplified procedure in which the func-
tion ␤ (␺ ) is specified in advance; then J (␺ ), H (␺ ), and the loci of the
´ and fast surfaces are found as part of the overall solution
slow, Alfven,
(including force balance with the opened field lines of the star and dead
zone). The mathematical procedure of choosing an appropriate ␤ (␺ ) is a
substitute for the physical problem of how to load matter onto field lines in
the x-region where the ideal MHD approximation of field freezing breaks
down (Shu et al. 1994a,b). . .
Apart from a trivial replacement of M多 for Mw , the funnel flow be-
haves somewhat differently from the x-wind. The funnel flow has a slow
MHD crossing but probably no Alfven ´ or fast MHD crossings. (In steady
state the star and disk can communicate with each other along closed field
lines by means of the latter two signals.) As a consequence, both ␤ (␺ )
and J (␺ ) can be freely specified for the funnel flow, although J (␺ ) must
ultimately be made self-consistent with the physical assumption that no
spinup or spindown of the star occurs in steady state (that is, that ⍀ 多 re-
mains equal to ⍀ x as a long-term average). In these circumstances, when
the star is small, Ostriker and Shu (1995) show that J (␺ ) is likely to be
nearly zero on every funnel-flow streamline ␺ = constant. In other words,
J 多 ⬇ 0, and any excess angular momentum brought to the star by the mat-
ter inflow is transferred back to the disk by the magnetic torques of a trail-
ing spiral pattern of funnel field lines.

VI. THE COLD LIMIT

The overall problem is mathematically tractable, because the parameter
⑀ ⬅ ax /Rx ⍀ x is much smaller than unity (typically, ⑀ ⬇ 0.03). This leads
to many simplifications; in particular, to the use of matched asymptotic
expansions for solving and connecting different parts of the flow. For ex-
 X-WINDS: THEORY AND OBSERVATIONS 799

ample, we may show that the slow MHD crossing must be made by matter-
carrying streamlines within a fractional distance ⑀ of Rx (unity in our
nondimensionalization), and that to order ⑀ 2 , H (␺ ) may be approximated
as zero. In the limit ⑀ y 0, the gas that becomes the x-wind (or the funnel
flow) emerges with linearly increasing velocities in a fan of streamlines
from the x-region as if from a single point.
In this approximation, the function ␤ (␺ ) cannot be chosen completely
arbitrarily if the magnetic field, mass flux, and mass density do not diverge
on the uppermost streamline ␺ y 1 as the x-wind leaves the x-region. For
modeling purposes, Shang and Shu (1998) adopt the following distribution
of magnetic field to mass flux:
␤ (␺ ) ⳱ ␤0 (1 ⫺ ␺ )⫺1/3 (14)
where ␤0 is a numerical constant related to the mean value of ␤ averaged
over streamlines:
1
冮
3
␤⬅ ␤ (␺ )d␺ ⳱ ␤0 (15)
0 2

The reader should not worry that equation (14) implies ␤ y ⬁ as ␺ y

1. This singular behavior merely reflects the fact that the magnetic field
B ⳱ ␤␳ u is nonzero on the last x-wind streamline, where, by definition,
␳ must become vanishingly small while u remains finite.
In what follows, we choose ␤0 so that f ⬇ 13 or 14 , with equation (3)
implying that J w ⬇ 3 or 4 if J 多 ⬇ 0 and ␶ ⬇ 0. In other words, when
the stellar magnetic field is strong enough to truncate the disk via a bal-
anced x-wind and funnel flow, and when one-third to one-fourth of the
matter drifting toward the inner edge of the disk becomes entangled in the
one-third of the field lines in the x-region that have a proper outward ori-
entation to launch an outflow, then the streamline-averaged location of the
Alfven´ surface is a factor 冪J w ⳱ 冪3, or 2 larger than the launch radius
Rx . The conclusion that J w ⬇ 3 or 4 is important when we compare the
predicted terminal velocities of stellar jets, (2J w ⫺ 3)1/2 Rx ⍀ x averaged
over all streamlines, with measured values (see below).

VII. ASYMPTOTIC COLLIMATION INTO JETS

A. Streamline Shape
Given the form ␤ (␺ ) from equation (14), the loci of streamlines at large
distances from the origin may be recovered in spherical polar coordinates
[r ⳱ (␼ 2Ⳮz 2 )1/2 , ␪ ⳱ arctan(␼ /z )] from the asymptotic analysis of Shu
et al. (1995):
2␤
r⳱ cosh[F (C, 1)] sin ␪ ⳱ sech[F (C, ␺ )] (16)
C
800 F. H. SHU ET AL.

where different values of the dimensionless current C correspond to dif-

ferent locations on any streamline ␺ ⳱ constant, and where F (C, ␺ ) is the
integral function
␺
␤ (␺ )d␺
冮
1
F (C, ␺ ) ⬅ (17)
C 0 [2J (␺ ) ⫺ 3 ⫺ 2C␤ (␺ )]1/2

Notice that r y ⬁ when C y 0 and vice versa.

The factor 2␤ in equation (16) is replaced by 6␤ [see the discussion of
Shang and Shu (1998)] if the dead zone is completely opened, as in Fig. 1,
rather than completely closed, as assumed in the calculations of Ostriker
and Shu (1995) and Shu et al. (1995). In the former case, the hoop stresses
of the toroidally wrapped fields on the uppermost streamlines of the x-
wind are balanced by the magnetic pressure of a bundle of longitudinal
field lines, coming from the star, that carries the same dimensionless mag-
netic flux, 2␲␤ , as their opened counterparts in the x-wind. In the latter
case, there are additional longitudinal field lines roughly parallel and an-
tiparallel to the polar axis carrying oppositely directed flux ⫾2␲␤ from
the opened field lines of the dead zone. In reality, the system probably
alternates between states in which outbursts, similar to coronal mass ejec-
tions, completely open the field lines of the dead zone and states in which
finite resistivity helps to reconnect the oppositely directed open lines and
produce a completely closed configuration for the dead zone.

B. Density and Velocity Fields

The density and velocity fields associated with equation (16) are obtained
from
C
␳⳱ vw ⳱ (2J ⫺ 3 ⫺ 2C␤ )1/2 (18)
␤␼ 2

where ␼ ⬅ r sin ␪ and vw is the wind speed in an inertial frame. The solid
and short dashed lines in Fig. 2 show, respectively, isodensity contours and
flow streamlines for a case ␤0 ⳱ 1 computed on four different scales by
the simplified approximate procedure discussed by Shang and Shu (1998).
For the model, J w ⳱ 3.73, obtainable from the average ␼ 2 of the location
´ surface marked by the inner set of long dashes in Fig. 2 [see
of the Alfven
equation (8) when ␤ 2 ␳ ⳱ 1]. With J 多 and ␶ assumed to be zero, equations
(3) and (15) imply f ⳱ 1/J w ⳱ 0.268 and ␤ w ⳱ 23 . The associated value
of ⌽dx ⳱ 2␤ w f 1/2 ⳱ 1.55.
⫺8
. The outerm3ost density contour in the fourth panel is ␳ ⳱ 10 in units
of Mw /4␲ ⍀ x Rx . The Alfven´ and fast surfaces formally asymptote to in-
finity as the upper streamline ␺ ⳱ 1 is approached, where the density
becomes vanishingly small. Because numerical computations are difficult
in this limit, the actual uppermost streamline displayed is ␺ ⳱ 0.98 rather
 X-WINDS: THEORY AND OBSERVATIONS 801

Figure 2. Isodensity contours (solid curves) and streamlines (dotted curves) for
a cold x-wind with ␤ (␺ ) ⳱ ␤0 (1 ⫺ ␺ )⫺1/3 , where ␤0 ⳱ 1. Isodensity contours
are spaced logarithmically in intervals of ⌬ log10 ␳ ⳱ 0.5, and streamlines are
spaced so that successive dotted lines contain an additional 10% of the total
mass loss in the upper hemisphere of the flow. The loci of the Alfven´ and fast
surfaces are marked by dashed lines. The empty space inside the uppermost
streamline, ␼ ⱕ ␼1 , is filled with open field lines from the central star that
asymptotically have the field strength, Bz ⳱ 2␤ /␼12 .

than ␺ ⳱ 1. Although streamlines collimate logarithmically slowly, iso-

density contours become cylindrically stratified fairly quickly, roughly as
␳ ⬀ ␼ ⫺2 [cf. equation (18)]. Since the radiative emission of forbidden
lines is highly biased toward regions of moderately high density, the flow
will appear more collimated than it actually is (see Shang et al. 1998).
The dimensional units of length and velocity, Rx and ⍀ x Rx , are ⬃0.06
AU and ⬃100 km s⫺1 in typical application; thus, at a distance of ⬃600
AU, about 50% of the streamlines are flowing at speeds ⬃200 km s⫺1
802 F. H. SHU ET AL.

within an angle ␪ ⬃ 4.5⬚ of the polar axis, while another 50% exit in
a wide-angle wind. In contrast, high-spatial-resolution imaging with the
Hubble Space Telescope shows emission-line jets from young stars ap-
pearing to collimate perfectly within tens of AU of the star (e.g., Burrows
et al. 1996). If our explanation is correct, the effect is partly an optical illu-
sion arising from the strong cylindrical density stratification of the x-wind
flow (see Fig. 2 of Shang et al. 1998).
The relative lack of streamline collimation at tens of AU scales, as
opposed to density collimation, provides potentially a key discriminating
test of the model. As seen in Fig. 3, the model predicts that forbidden
lines arising in the x-wind should have larger velocity widths at the base
of the flow, where a typical line of sight encounters flow velocities at a
variety of angles, than farther up the length of the jet, where the flow vec-
tors are better oriented along a single direction as the x-wind continues to
collimate. Long-slit spectroscopy by Reipurth and Heathcote (1991) and
Bacciotti et al. (1996) for several jets shows exactly this behavior. Un-
fortunately, the light seen near the base of the flow in embedded sources
probably arrives by scattering from surrounding dust particles, so the ob-
served effect appears to occur at a larger physical scale than predicted by
the theory. Whether the basic kinematics has been affected by observing
the phenomenon via “mirrors” remains to be ascertained, perhaps by spec-
tropolarimetry. Better source candidates for a cleaner test of our prediction
are the “naked jets” now found in evolved H II regions (B. Reipurth, per-
sonal communication, 1998).
The conventional interpretation of line profiles at the base of the flow
being wider than in the jet proper invokes a mixing of fast and slow winds
(e.g., Kwan and Tademaru 1988; Hirth et al. 1997). Such mixing of fast
and slow material might occur because of turbulent entrainment as the
lowermost x-wind streamlines interact with a flared accretion disk (Li and
Shu 1996a). An x-wind could also interact with a slow wind driven by
photoevaporation of a nebular disk (e.g., Shu et al. 1993). In all cases,
however, as long as the source of the slower-moving material lies in a flat-
tened distribution, appreciable mixing is easily understood only if the fast
flow has a substantial equatorially directed component on the size scale of
the inferred accretion disks.

VIII. BIPOLAR MOLECULAR OUTFLOWS

The phenomenon of bipolar molecular outflows (Snell et al. 1980) caught
theorists by surprise, because infall rather than outflow had been expected
to define the process of star formation. Since 1980, more than 200 molec-
ular outflows have been observed (see, e.g., the reviews of Lada 1985;
Fukui et al. 1993; Wu et al. 1996; Cabrit et al. 1997). By now, this ubiq-
uitous phenomenon has become an integral part of the standard paradigm
for star formation (Shu et al. 1987). Here, we limit our discussion to recent
developments since the publication of Protostars and Planets III.
 X-WINDS: THEORY AND OBSERVATIONS 803

300

200
velocity (km/s)

100

–100

–200

–300
–100 0 100 200 300 400 500

300

200
velocity (km/s)

100

–100

–200

–300
–100 0 100 200 300 400 500

300

200
velocity (km/s)

100

–100

–200

–300
–100 0 100 200 300 400 500
position (AU)

Figure 3. Position-velocity diagrams for [S II] ␭ 6731 emission when the syn-
thetic spectrum is taken with a long slit placed along the length of the jet but
displaced laterally by 1.5Rx with respect to its central axis. The different figures
correspond to inclination angles i ⳱ 90⬚ (top), 60⬚ (middle), and 30⬚ (bottom).
The range of projected terminal velocities seen in the models compares well
with observed values. By the same token, the fact that measured terminal ve-
locities in stellar jets have a limited range of values implies that the streamlines
cannot be launched from a wide variety of disk radii of differing centrifugal
speeds, as is implicit in many disk-wind models.
804 F. H. SHU ET AL.

IX. THE AMBIENT MEDIUM INTO WHICH

OUTFLOWS PROPAGATE

Low-mass stars form in the dense cores of molecular clouds (Myers 1995).
These cores have a typical size ⬃0.1 pc, a number density ⬃3 ⫻104 cm⫺3 ,
a mass ranging from a small fraction of a solar mass to about 10 M䉺 , and
an axial ratio for flattening of typically 2:1. For isolated cores, the last
fact implies that agents other than isotropic thermal or turbulent pressures
help to support cores against their self-gravity, although it is not yet clear
observationally whether the true shapes are oblate, prolate, or triaxial (P. C.
Myers, personal communication, 1998). Observed cloud rotation rates are
generally too small to account for the observed flattening (Goodman et al.
1993). This leaves magnetic fields, which are believed for other reasons
to play a crucial role in contemporary star formation.
In one scenario, the weakening of magnetic support in the central part
of a molecular cloud by ambipolar diffusion leads to the continued con-
traction of a cloud core with ever-growing central concentration (Nakano
1979; Lizano and Shu 1989; Basu and Mouschovias 1994). Li and Shu
(1996b) referred to the cloud configuration when the central isothermal
concentration first becomes formally infinite as the “pivotal” state. This
state separates the nearly quasistatic phase of core evolution from the
fully dynamic phase of protostellar accretion (see the first two stages de-
picted in Fig. 11 of Shu et al. 1987). The numerical simulations indicate
that the pivotal states have several simplifying properties, which moti-
vated Li and Shu (1996b) to approximate them as scale-free equilibria
with power law radial dependences for the density and magnetic flux
function:

a2 4␲a 2 r
␳ (r, ␪ ) ⳱ R (␪ ) ⌽(r, ␪ ) ⳱ ␾ (␪ ) (19)
2␲Gr 2 G 1/2

In equation (19), we have adopted a spherical polar coordinate system

(r, ␪, ␸ ), and a is the isothermal sound speed of the cloud, while R (␪ ) and
␾ (␪ ) are the dimensionless angular distribution functions for the density
and magnetic flux, given by force balance along and across field lines. The
resulting differential equations and boundary conditions yield a linear se-
quence of possible solutions, characterized by a single dimensionless free
parameter, H0 , which represents the fractional overdensity supported by
the magnetic field above that supported by thermal pressure. Comparison
with the typical degree of elongation of observed cores suggests that the
overdensity factor H0 ⬇ 0.5 ⫺ 1. In Fig. 4, we plot isodensity contours
and field lines for the case H0 = 0.5. The presence of very low-density
regions at the poles of the density toroid has profound implications for the
shape and kinematics of bipolar molecular outflows. (See also Torrelles et
al. 1983, 1994.)
 X-WINDS: THEORY AND OBSERVATIONS 805

Figure 4. Isodensity contours (heavy curves) and magnetic field lines (light
curves) in a meridional plane for a molecular cloud core in the pivotal state
with an overdensity factor of H0 ⳱ 0.5.

X. MECHANISM FOR MOMENTUM TRANSFER

Observations of one of the earliest types of young stellar objects, the so-
called “Class 0” sources, indicate that a high-speed wind is turned on early
in the main accretion phase of star formation (Bontemps et al. 1996). The-
oretically, the wind should create two wind-blown bubbles, one on each
side of the accretion disk, which continues to feed the central stellar ob-
ject through the equatorial region. The bubbles are bound by the swept-up
ambient core material and are expected to become elongated in the direc-
tion along the rotation/magnetic axis, where the wind momentum flux is
expected to be largest and the ambient density lowest. To determine the
shape and kinematics of the swept-up material (i.e., the molecular out-
flow), one needs to know how momentum is transferred from the wind to
the ambient medium. Generally speaking, the ambient medium is swept
up by a forward shock that runs ahead of the wind bubble. At the same
time, the free wind runs into a reverse shock. Between the shocked am-
bient medium and the shocked wind material lies a contact discontinuity,
across which pressure balance must be maintained.
Depending on the cooling timescale, the shocked material could be
either radiative or adiabatic. For an ambient medium shocked with a rela-
tively low shock speed of order 10 km s⫺1 and a relatively high preshock
density of order 103 cm⫺3 , the cooling timescale is always much shorter
than the time to sweep up a mass comparable to the wind mass (Koo and
McKee 1992). The short cooling timescale leads to a radiative shocked
806 F. H. SHU ET AL.

ambient medium, which forms a thin shell around the wind bubble. For a
wind speed of order 300 km s⫺1 or less, the shocked wind material will be-
come radiative as well (Koo and McKee 1992). With these assumptions,
the free protostellar wind is bound by two thin layers of shocked mate-
rial. The two layers tend to slide relative to each other, exciting Kelvin-
Helmholtz instabilities at the interface. Full development of fluid insta-
bilities may lead to a well-mixed shell of two different types of shocked
material, creating an essentially ballistic putty. Each segment of mixed ra-
diative shell absorbs all of the wind momentum imparted to it. This local
conservation of vector momentum forms the basis of a simple theory for
molecular outflows (Shu et al. 1991; see also Wilkin 1997). A more realis-
tic treatment, where the shell is not treated as spatially thin because of the
“cushioning” effect of the embedded magnetic fields, will be more com-
plicated but deserves investigation [see Li and Shu’s (1996a) treatment of
the related problem of the interaction of a wide-angle wind with a flared
disk].

XI. A SIMPLE THIN-SHELL THEORY

In the simple theory developed by Shu et al. (1991), wind momentum flux
per steradian of the form
.
Mw vw
K (r, ␪ ) ⳱ P (␪ ) (20)
4␲
propagates into a static molecular clo.ud core of density given by the first
of equations (19). In equation (20), Mw and vw are the mass loss rate and
average speed of the wind, and P (␪ ) is a normalized angular distribution
function. The model makes an implicit assumption that the directions of
the spin vector ⍀ x of the accretion disk, which defines the axis of the bipo-
lar outflow, and the ambient magnetic field B, which defines the axis of the
molecular toroid, are aligned. Presumably this alignment arises because
magnetic braking in the prior epoch of molecular cloud core formation was
more efficient at reducing components of precollapse ⍀ perpendicular to
B than at reducing the component parallel to B (see, e.g., Mouschovias
and Paleologou 1980).
If we assume the same polar axis for the functions P (␪ ) and R (␪ ),
momentum conservation in every direction ␪ yields the expansion velocity
of the swept-up shell as

冋 ␪␪ 册
. 1/2

冢 冣
1/2
Mw P( )
vs ⳱ . (avw )1/2 (21)
2Ma R( )
.
where Ma ⬅ a 3 /G is a measure of the mass accretion rate onto the cen-
tral protostellar object (Shu 1977). Note that the shell expansion speed is
independent of radius r , a special property of the r ⫺2 density profile of
 X-WINDS: THEORY AND OBSERVATIONS 807

the ambient medium. Aside from factors of order unity, equation (21) pre-
dicts a characteristic bipolar molecular outflow speed that is the geometric
mean of the wind speed and the cloud sound speed. Since the wind speed
vw is typically a few hundred km s⫺1 (Edwards et al. 1994) and the sound
speed a is of the order of a fraction of a km s⫺1 , we expect a characteris-
tic molecular outflow speed of order 10 km s⫺1 , as observed (Fukui et al.
1993). The simple model also naturally accounts for an otherwise myste-
rious “Hubble” expansion, i.e., a flow velocity proportional to the distance
from the central object that is often observed in molecular outflow lobes
(e.g., Lada and Fich 1996).
At the time of their proposal, the functional forms of P (␪ ) and R (␪ )
were not known to Shu et al. (1991). Masson and Chernin (1992) therefore
criticized their model on the grounds that “reasonable” choices for P (␪ )
and R (␪ ) yielded masses for the swept-up material that should increase
with increasing line-of-sight velocities v rather than decline as a power
law, dm /dv ⬀ v ␣ , with the exponent ␣ close to ⫺1.8, as determined em-
pirically for several well-observed outflows. Masson and Chernin (1992)
argued that such a steep decrease of mass with velocity would be diffi-
cult to reproduce with any model of outflow cavities carved by wide-angle
winds, in essence because the portion of the lobe that travels fastest also
travels farthest and is therefore likely to sweep up the greatest amount of
matter, not the least amount. This argument presupposes that the main ␪
dependence enters in P (␪ ) and not in R (␪ ), and that neither angular dis-
tribution is extreme.
The assumption that P is a function only of ␪ neglects the continuing
collimation of the x-wind, which occurs logarithmically slowly according
to the asymptotic analysis of section VII. Under the approximation that
C can be treated as a constant, rather than a quantity that varies logarith-
mically with r , and that the dependences of ␤ (␺ ) and J (␺ ) on ␺ can be
ignored for the streamlines (␺ not near 1) that interact most strongly with
the ambient cloud, equation (18) implies that P (␪ ) ⬀ sin⫺2 ␪ . This ex-
treme behavior for P (␪ ), very large near the poles ␪ ⳱ 0 and ␲ , is paired
with an equally extreme behavior for R (␪ ), vanishingly small near the
poles. For the particular case of H0 ⳱ 0.5 shown in Fig. 4, we plot the re-
sulting lobe shape in Fig. 5a and the mass-velocity distribution in Fig. 5b.
Both properties are in reasonably good agreement with observations, lead-
ing to a refutation of the primary argument that bipolar molecular outflows
must be driven by highly collimated jets, despite the objection that such
jets have difficulty accounting for the relatively poor collimation of most
molecular outflow lobes (Bachiller 1996; Cabrit et al. 1997).

XII. STABILITY OF PARSEC-LONG JETS

The small amount of ambient matter swept to high velocities, which
solves the original objection of Masson and Chernin (1992) to the bipolar
808 F. H. SHU ET AL.

outflow model of Shu et al. (1991), arises because the heaviest part of
the x-wind encounters only the low-density material of the poles of the
molecular toroid [where R (␪ ) ideally becomes vanishingly small]. As a
consequence, it is easy for the x-wind to break out of the molecular cloud
core in these directions (see Fig. 5) and, thus, to account for the parsec-long
jets discovered by Bally and Devine (1997).
Because the x-wind is launched within a few stellar radii of the star,
even logarithmically slow collimation yields highly directed jets at parsec
distances (⬃3 ⫻106 Rx ). For r ⳱ 3 ⫻ 106 , equation (16) implies that an
undeflected ␺ ⳱ 0.50 streamline is collimated within an angle ␪ ⳱ 0.9⬚
of the polar axis for the example illustrated in Fig. 3. In other words, at a
distance ⬃1 pc from the central source, 50% of the undisturbed outflow
is encased within a cylinder of radius ⬃0.015 pc. At very large distances
from the central source the difference between x-winds and pure stellar
jets narrows considerably.
It has often been asked how jets can maintain their integrity for such
long paths. The glib answer provided by all proponents of magnetocen-
trifugal driving is that the flow is self-collimated. Self-collimation occurs

6
log10(dm/dv)

3
–1.5 –1 –0.5 0 0.5 1
log10(|v|)

a b

Figure 5. (a) Shape of swept-up lobes in the meridional plane for a wind power ⬀
sin⫺2 ␪ and a core density distribution ⬀ R (␪ ) of a singular model with H0 ⳱
0.5. The dashed line indicates schematically the trajectory of the dense parts of
the x-wind seen as optical jets that easily burst through the rarefied polar regions
of the surrounding molecular toroid. (b) The logarithm of the swept-up mass
at each logarithmic interval of the line-of-sight velocity for inclination angles
␽ ⳱ 80⬚ , 45⬚ (middle line), and 10⬚ with respect to the axis of the bipolar flow.
The case ␽ ⳱ 80⬚ has a nearly discontinuous step due to the waist of the two
lobes in (a).
 X-WINDS: THEORY AND OBSERVATIONS 809

by the hoop stresses of the toroidal magnetic fields that grow in strength
relative to the poloidal components as the flow distance from the source
increases. However, such magnetic confinement schemes have long been
known to the fusion community to be unstable with respect to kinking and
sausaging (see Fig. 10.6 in Jackson 1975; see also Eichler 1993). A solu-
tion to the problem has also long been known and is the principle behind
working Tokamaks: the introduction of dynamically important levels of
longitudinal magnetic fields along the length of the plasma column to sta-
bilize the kinking and sausaging motions (see Figures 10.7 and 10.8 of
Jackson 1975). Such dynamically important levels of longitudinal fields
along the central flow axis are exactly what fill the hollow core of the jet
in the x-wind model of Fig. 2.

Acknowledgments This research is funded by a grant from the Na-

tional Science Foundation and by the NASA Astrophysics Theory Pro-
gram, which supports a joint Center for Star Formation Studies at NASA/
Ames Research Center, the University of California at Berkeley, and the
University of California at Santa Cruz.

REFERENCES

Arons, J. 1986. Centrifugally driven winds from X-ray pulsars. In Plasma Pene-
tration into Magnetospheres, ed. N. Kylafis, J. Papamastorakis, and J. Ven-
tura (Iraklion: Crete University Press), pp. 115–124.
Bacciotti, F., Hirth, G. A., and Natta, A. 1996. The optical jet of RW Aurigae:
Excitation temperature and ionization state from long-slit spectra. Astron. As-
trophys. 310:309–314.
Bachiller, R. 1996. Bipolar molecular outflows from young stars and protostars.
Ann. Rev. Astron. Astrophys. 34:111–154.
Balbus, S., and Hawley, J. 1991. A powerful local shear instability in weakly mag-
netized disks. I. Linear analysis. Astrophys. J. 376:214–233.
Bally, J., and Devine, D. 1997. Giant Herbig-Haro flows. In Herbig-Haro Flows
and the Birth of Low Mass Stars, Proc. IAU Symp. 182, ed. B. Reipurth and
C. Bertout (Dordrecht: Kluwer), pp. 29–38.
Basri, G., Marcy, G. W., and Valenti, J. A. 1992. Limits on the magnetic flux of
pre-main-sequence stars. Astrophys. J. 390:622–633.
Basu, S., and Mouschovias, T. 1994. Magnetic braking, ambipolar diffusion, and
the formation of cloud cores and protostars. I. Axisymmetric solutions. As-
trophys. J. 432:720–741.
Bertout, C. 1989. T Tauri stars—Wild as dust. Ann. Rev. Astron. Astrophys.
27:351–395.
Bertout, C., Basri, G., and Bouvier, J. 1988. Accretion disks around T Tauri stars.
Astrophys. J. 330:350–373.
Biskamp, D. 1993. Nonlinear Magnetohydrodynamics (Cambridge: Cambridge
University Press).
Blandford, R. D., and Payne, D. G. 1982. Hydromagnetic flows from accretion
discs and the production of radio jets. Mon. Not. Roy. Astron. Soc. 199:883–
903.
810 F. H. SHU ET AL.

Bontemps, S., Andre, P., Terebey, S., and Cabrit, S. 1996. Evolution of outflow
activity around low-mass embedded young stellar objects. Astron. Astrophys.
311:858–872.
Bouvier, J., Cabrit, S., Fernandez, M., Martin, E. L., and Matthews, J. M. 1993.
Coyotes-I—The photometric variability and rotational evolution of T Tauri
stars. Astron. Astrophys. 272:176–206.
Burrows, C. J., Stapelfeldt, K. R., Watson, A. M., Krist, J. E., Ballester, G. E.,
Clarke, J. T., Crisp, D., Gallagher, J. S., III, Griffiths, R. E., Hester, J. J.,
Hoessel, J. G., Holtzman, J. A., Mould, J. R., Scowen, P. A., Trauger, J. T.,
and Westphal, J. A. 1996. Hubble Space Telescope observations of the disk
and jet of HH 30. Astrophys. J. 473:437–451.
Cabrit, S., Raga, A. C., and Gueth, F. 1997. Models of bipolar molecular outflows.
In Herbig-Haro Flows and the Birth of Low Mass Stars, Proc. IAU Symp.
182, ed. B. Reipurth and C. Bertout (Dordrecht: Kluwer), pp. 163–180.
Camenzind, M. 1990. Magnetized disk-winds and the origin of bipolar outflows.
In Reviews in Modern Astronomy, Vol. 3, ed. G. Klare (Berlin: Springer), pp.
234–265.
Cui, W. 1997. Evidence for “propeller” effects in X-ray pulsars GX 1+4 and GRO
J1744-28. Astrophys. J. Lett. 482:L163–L166.
Edwards, S., Strom, S. E., Hartigan, P., Strom, K. M., Hillenbrand, L. A., Herbst,
W., Attridge, J., Merreil, K. M., Probst, R., and Gatley, I. 1993. Angular mo-
mentum regulation in low-mass young stars surrounded by accretion disks.
Astron. J. 106:372–382.
Edwards, S., Hartigan, P., Ghandour, L., and Andrulis, C. 1994. Spectroscopic
evidence of magnetic accretion in classical T Tauri stars. Astrophys. J.
108:1056–1070.
Eichler, D. 1993. Magnetic confinement of jets. Astrophys. J. 419:111–116.
Fukui, Y., Iwata, T., Mizuno, A., Bally, J., and Lane, A. 1993. Molecular out-
flows. In Protostars and Planets III, ed. E. H. Levy and J. I. Lunine (Tucson:
University of Arizona Press), pp. 603–639.
Ghosh, P., and Lamb, F. K. 1978. Disk accretion by magnetic neutron stars. As-
trophys. J. Lett. 223:L83–L87.
Goodman, A. A., Benson, P. J., Fuller, G. A., and Myers, P. C. 1993. Dense cores
in dark clouds. VIII. Velocity gradients. Astrophys. J. 406:528–547.
Gullbring, E., Hartmann, L., Briceno, C., and Calvet, N. 1998. Disk accretion rates
for T Tauri stars. Astrophys. J. 492:323–341.
Hartmann, L., and MacGregor, K. B. 1982. Protostellar mass and angular momen-
tum loss. Astrophys. J. 259:180–192.
Hartmann, L., Hewitt, R., and Calvet, N. 1994. Magnetospheric accretion mod-
els for T Tauri stars. 1: Balmer line profiles without rotation. Astrophys. J.
426:669–687.
Hayashi, M. R., Shibata, K., and Matsumoto, R. 1996. X-ray flares and mass out-
flows driven by magnetic interaction between a protostar and its surrounding
disk. Astrophys. J. Lett. 468:L37–L40.
Heinemann, M., and Olbert, S. 1978. Axisymmetric ideal MHD stellar wind flow.
J. Geophys. Res. 83:2457–2460.
Heyvaerts, J., and Norman, C. 1989. The collimation of magnetized winds. Astro-
phys. J. 347:1055–1081.
Hirth, G. A., Mundt, R., and Solf, J. 1997. Spatial and kinematic properties of
the forbidden emission line region of T Tauri stars. Astron. Astrophys. Suppl.
126:437–496.
Jackson, J. D. 1975. Classical Electrodynamics (New York: Wiley).
Johns, C. M., and Basri, G. 1995a. Hamilton echelle spectra of young stars. II.
Time series analysis of H(alpha) variations. Astron. J. 109:2800–2816.
 X-WINDS: THEORY AND OBSERVATIONS 811

Johns, C. M., and Basri, G. 1995b. The line profile variability of SU Aurigae.
Astrophys. J. 449:341–364.
Johns-Krull, C. M. 1996. Limits on the differential rotation of T Tauri stars. As-
trophys. J. 306:803–810.
Johns-Krull, C. M., and Hatzes, A. P. 1997. The classical T Tauri star Sz 68: The
Doppler imaging and evidence of magnetospheric accretion. Astrophys. J.
487:896–915.
Johns-Krull, C. M., Valenti, J. A., and Koresko, C. 1999. Measuring the magnetic
field on the classical T Tauri star BP Tauri. Astrophys. J. 516:900–915.
Ko¨ nigl, A. 1989. Self-similar models of magnetized accretion disks. Astrophys. J.
342:208–223.
Ko¨ nigl, A. 1991. Disk accretion onto magnetic T Tauri stars. Astrophys. J. Lett.
370:L39–L43.
Koo, B.-C., and McKee, C. F. 1992. Dynamics of wind bubbles and superbubbles.
I. Slow winds and fast winds. Astrophys. J. 388:93–126.
Kwan, J., and Tademaru, E. 1988. Jets from T Tauri stars. Spectroscopic evidence
and collimation mechanism. Astrophys. J. Lett. 332:L41–L44.
Lada, C. J. 1985. Cold outflows, energetic winds, and enigmatic jets around young
stellar objects. Ann. Rev. Astron. Astrophys. 23:267–317.
Lada, C. J., and Fich, M. 1996. The structure and energetics of a highly collimated
bipolar outflow: NGC 2264G. Astrophys. J. 459:638–652.
Li, J., and Wickramasinghe, D. T. 1997. Disc accretion onto magnetic stars: Slow
rotator and propeller. In Accretion Phenomena and Related Outflows, IAU
Colloq. 163, ASP Conf. Ser. 121, ed. D. T. Wickramasinghe, L. Ferrario, and
G. V. Bicknell (San Francisco: Astronomical Society of the Pacific), pp. 241–
250.
Li, Z. Y., and Shu, F. H. 1996a. Interaction of wide-angle MHD winds with flared
disks. Astrophys. J. 468:261–268.
Li, Z. Y., and Shu, F. H. 1996b. Magnetized singular isothermal toroids. Astrophys.
J. 472:211–224.
Linker, J. A., and Mikic,´ Z. 1995. Disruption of a helmet streamer by photospheric
shear. Astrophys. J. Lett. 438:L45–L48.
Lizano, S., and Shu, F. H. 1989. Molecular cloud cores and bimodal star formation.
Astrophys. J. 342:834–854.
Lovelace, R. V. E., Berk, H. L., and Contopoulos, J. 1991. Magnetically driven
jets and winds. Astrophys. J. 379:696–705.
Lovelace, R. V. E., Romanova, M. M., and Bisnovatyi-Kogan, G. S. 1999. Mag-
netic propeller outflows. Astrophys. J. 514:368–372.
Masson, C., and Chernin, L. 1992. Properties of swept-up molecular outflows.
Astrophys. J. Lett. 387:L47–L50.
Mestel, L. 1968. Magnetic braking by a stellar wind—I. Mon. Not. Roy. Astron.
Soc. 138:359–391.
Meyer, M. R., Calvet, N., and Hillenbrand, L. A. 1997. Intrinsic near-infrared ex-
cesses of T Tauri stars: Understanding the classical T Tauri star locus. Astron.
J. 114:288–300.
Montmerle, T., Feigelson, E., Bouvier, J., and Andre, P. 1993. Magnetic fields, ac-
tivity and circumstellar material around young stellar objects. Protostars and
Planets III, ed. E. H. Levy and J. I. Lunine (Tucson: University of Arizona
Press) pp. 689–717.
Mouschovias, T. C., and Paleologou, E. V. 1980. Magnetic braking of an aligned
rotator during star formation. Astrophys. J. 237:877–899.
Muzerolle, J., Calvet, N., and Hartmann, L. 1998. Magnetospheric accretion mod-
els for the hydrogen emission lines of T Tauri stars. Astrophys. J. 492:743–
753.
812 F. H. SHU ET AL.

Myers, P. C. 1995. Star-forming molecular clouds. In Molecular Clouds and Star

Formation, ed. C. Yuan and J. You (Singapore: World Scientific), pp. 47–96.
Najita, J. R., and Shu, F. H. 1994. Magnetocentrifugally driven flows from young
stars and disks. III: Numerical solutions of the sub-Alfvenic region. Astro-
phys. J. 429:808–825.
Najita, J. R., Carr, J. S., and Tokunaga, A. T. 1996. High-resolution spectroscopy
of Br gamma emission in young stellar objects. Astrophys. J. 456:292–299.
Nakano, T. 1979. Quasistatic contraction of magnetic protostars due to mag-
netic flux leakage. I. Formulation and an example. Pub. Astron. Soc. Japan
31:697–712.
Ostriker, E. C., and Shu, F. H. 1995. Magnetocentrifugally driven flows from
young stars and disks. IV. The accretional funnel and dead zone. Astrophys.
J. 447:813–828.
Pudritz, R. E., and Norman, C. A. 1983. Centrifugally driven winds from con-
tracting molecular disks. Astrophys. J. 274:677–697.
Reipurth, B., and Heathcote, S. 1991. The jet and energy source of HH 46/47.
Astron. J. 246:511–534.
Rydgren, A. E., and Vrba, F. J. 1983. Periodic light variations in four pre-main-
sequence K stars. Astrophys. J. 267:191–198.
Sakurai, T. 1985. Magnetic stellar winds: A 2-D generalization of the Weber-
Davis model. Astron. Astrophys. 152:121–129.
Shang, H., and Shu, F. H. 1998. X-wind made simple. Astrophys. J., in preparation.
Shang, H., Shu, F., Lee, T., and Glassgold, A. E. 1997. Protostellar cosmic rays and
extinct radioactivities. In Low Mass Star Formation from Infall to Outflow,
Poster Proc. IAU Symp. No. 182, ed. F. Malbet and A. Castets (Grenoble:
Laboratorie d’Astrophysique), pp. 312–315.
Shang, H., Shu, F. H., and Glassgold, A. E. 1998. Synthetic images and long-slit
spectra of protostellar jets. Astrophys. J. Lett. 493:L91–L94.
Shu, F. H. 1977. Self-similar collapse of isothermal spheres and star formation.
Astrophys. J. 214:488–497.
Shu, F. H., Adams, F. C., and Lizano, S. 1987. Star formation in molecular clouds:
Observation and theory. Ann. Rev. Astron. Astrophys. 25:23–81.
Shu, F. H., Lizano, S., Ruden, S., and Najita, J. 1988. Mass loss from rapidly
rotating magnetic protostars. Astrophys. J. Lett. 328:L19–L23.
Shu, F. H., Ruden, S. P., Lada, C. J., and Lizano, S. 1991. Star formation and the
nature of bipolar outflows. Astrophys. J. Lett. 370:L31–L34.
Shu, F. H., Johnstone, D., and Hollenbach, D. 1993. Photoevaporation of the solar
nebula and the formation of the giant planets. Icarus 106:92–101.
Shu, F., Najita, J., Ostriker, E., Wilkin, F., Ruden, S., and Lizano, S. 1994a. Mag-
netocentrifugally driven flows from young stars and disks. I. A generalized
model. Astrophys. J. 429:781–796.
Shu, F. H., Najita, J., Ruden, S. P., and Lizano, S. 1994b. Magnetocentrifugally
driven flows from young stars and disks. II. Formulation of the dynamical
problem. Astrophys. J. 429:797–807.
Shu, F. H., Najita, J., Ostriker, E. C., and Shang, H. 1995. Magnetocentrifugally
driven flows from young stars and disks. V. Asymptotic collimation into jets.
Astrophys. J. Lett. 455:L155–L158.
Shu, F. H., Shang, H., Glassgold, A. E., and Lee, T. 1997. X-rays and fluctuating
x-winds. Science 277:1475–1479.
Snell, R. L., Loren, R. B., and Plambeck, R. L. 1980. Observations of CO
in L1551—evidence for stellar wind driven shocks. Astrophys. J. Lett.
239:L17–L22.
Torrelles, J. M., Rodriguez, L. F., Canto, J., Carral, P., Marcaide, J., Moran, J. M.,
and Ho, P. T. P. 1983. Are interstellar toroids the focusing agent of the bipolar
molecular outflow? Astrophys. J. 217:214–230.
 X-WINDS: THEORY AND OBSERVATIONS 813

Torrelles, J. M., Gomez, J. F., Ho, P. T. P., Rodriguez, L. F., Anglada, G., and
Canto, J. 1994. The puzzling distribution of the high-density molecular gas
in HH 1-2: A contracting interstellar toroid? Astrophys. J. 435:290–312.
Uchida, Y., and Shibata, K. 1985. Magnetodynamical acceleration of CO and op-
tical bipolar flows from the region of star formation. Pub. Astron. Soc. Japan
37:515–535.
Vogel, S. N., and Kuhi, L. V. 1981. Rotational velocities of pre–main sequence
stars. Astrophys. J. 245:960–976.
Walter, F. M., Brown, A., Mathieu, R. D., Myers, P. C., and Vrba, F. J. 1988. X-ray
sources in regions of star formation. III. Naked T Tauri stars associated with
the Taurus-Auriga complex. Astron. J. 96:297–325.
Weber, E. J., and Davis, L. 1967. The angular momentum of the solar wind. As-
trophys. J. 148:217–227.
Wilkin, F. P. 1997. Bow shocks in non-uniform medium. In Low Mass Star For-
mation from Infall to Outflow, Poster Proc. IAU Symp. 182, ed. F. Malbet
and A. Castets (Grenoble: Laboratorie d’Astrophysique), p. 190.
Wu, Y., Huang, M., and He, J. 1996. A catalogue of high velocity molecular out-
flows. Astron. Astrophys. Suppl. 115:283–284.
Wynn, G. A., King, A. R., and Horne, K. 1997. A magnetic propeller in the cata-
clysmic variable AE Aquarii. Mon. Not. Roy. Astron. Soc. 286:436–446.
View publication stats