Content uploaded by Luis F. Miranda
Author content
All content in this area was uploaded by Luis F. Miranda on Jan 27, 2015
Content may be subject to copyright.
arXiv:astro-ph/0407030v1 1 Jul 2004
Scheduled for the October 2004 issue of The Astronomical Journal
Preprint typeset using L
A
T
E
X style emulateapj v. 11/12/01
MZ 3, A MULTIPOLAR NEBULA IN THE MAKING
Mart
´
ın A. Guerrero1,2,3, You-Hua Chu1, and Luis F. Miranda2
1Astronomy Department, University of Illinois at Urbana-Champaign, Urbana, IL 61801
2Instituto de Astrof´ısica de Andaluc´ıa (CSIC), Spain
mar@iaa.es, chu@astro.uiuc.edu, lfm@iaa.es
Scheduled for the October 2004 issue of The Astronomical Journal
ABSTRACT
The nebula Mz 3 has arguably the most complex bip olar morphology, consisting of three nested pairs of
bipolar lobes and an equatorial ellipse. Its three pairs of bipolar lobes share the same axis of symmetry,
but have very different opening angles and morphologies: the innermost pair of bipolar lobes shows closed
lobe morphology, while the other two have open lobes with cylindrical and conical shapes, respectively.
We have carried out high-dispersion spectroscopic observations of Mz 3, and detected distinct kinematic
properties among the different morphological components. The expansion characteristics of the two
outer pairs of lobes suggest that they originated in an explosive event, whereas the innermost pair of
lobes resulted from the interaction of a fast wind with the surrounding material. The equatorial ellipse
is associated with a fast equatorial outflow which is unique among bipolar nebulae. The dynamical ages
of the different structures in Mz 3 suggest episodic bipolar ejections, and the distinct morphologies and
kinematics among these different structures reveal fundamental changes in the system between these
episodic ejections.
Subject headings: ISM: kinematics and dynamics — planetary nebulae: individual (Mz 3)
1. introduction
Mz 3, the Ant Nebula, is perhaps one of the most stun-
ning bipolar nebulae. The Hubble Space Telescope (HST )
color image presented by the Hubble Heritage Program
(STScI-PRC01-05, PI: B. Balick, V. Icke, R. Sahai, and J.
T. Trauger) reveals a complex system of three nested pairs
of bipolar lobes. These bipolar lobes are roughly aligned
along the same axis of symmetry, but have vastly different
shapes, opening angles and detailed morphologies. In ad-
dition, a faint ellipse of emission aligned along the equator
of these bipolar lobes is seen. Not only is the morphology
of Mz 3 complex, but its nature is also uncertain. While
usually classified as a young planetary nebula (PN), Mz 3
has also been suggested to be a circumstellar nebula of a
symbiotic star, based on the high density of its core (Zhang
& Liu 2002), its near-IR colors (Schmeja & Kimeswenger
2001), and the spectrum of its central star.
Previous high-dispersion spectroscopic observations of
Mz 3 have detected several pairs of bipolar lobes, and their
kinematic properties led to the suggestion of episodic bipo-
lar ejections (L´opez & Meaburn 1983; Meaburn & Walsh
1985). More recently, Redman et al. (2000) reported the
discovery of fast, 500 km s−1, collimated outflows. De-
tailed modeling of the structure of Mz 3 has been ham-
pered by the limited detector sensitivity or sparse slit cov-
erage of these previous observations. Therefore, we have
carried out new long-slit, high-dispersion echelle observa-
tions of Mz 3, emphasizing particularly the morphological
features that have not been observed previously. These
echelle observations, combined with the high-resolution
HST narrow-band images and Chandra X-ray observation,
allow us to produce a complete spatio-kinematic model of
Mz 3, adequately representing the three pairs of bipolar
lobes and the equatorial ellipse. While our results confirm
the previous suggestion that the multipolar structure was
produced by episodic bipolar ejections (L´opez & Meaburn
1983; Meaburn & Walsh 1985), we are able to describe the
kinematic properties and determine the formation process
more precisely. This paper reports our new observations
and analysis of the physical structure of Mz 3.
2. observations
2.1. Archival HST Images
Narrow-band WFPC2 images of Mz 3 in the Hα, Hβ,
and [N ii]λ6583 emission lines were retrieved from the
HST archive (Proposal IDs 6502 and 9050, PI: Balick, and
Proposal ID 6856, PI: Trauger). The images we used in
this work are listed in Table 1 with their integration times,
filters, and the location of Mz 3 on the WFPC2 (PC or
WFC). These images were calibrated via the pipeline pro-
cedure, including the analog-to-digital correction, bias and
dark image subtraction, and flat-field correction. We re-
moved the cosmic rays and combined different exposures
obtained with the same filter using standard IRAF rou-
tines. The Hαand [N ii] images of Mz 3, displayed in Fig-
ure 1, are used to analyze the nebular morphology. The
Hαto Hβratio map of Mz 3, shown in Figure 2-left, is used
to investigate the distribution of intranebular extinction.
2.2. Archival Chandra X-ray Observations
The Advanced CCD Imaging Spectrometer (ACIS) on
board the Chandra X-Ray Observatory was used on 2002
October 23 to obtain a 40.8 ks exposure of Mz 3 (Obser-
vation ID: 2546; PI: Kastner). Mz 3 was positioned at
the nominal aim point of the ACIS-S array on the back-
illuminated S3 CCD. We retrieved the level 1 and level
2 processed data from the Chandra Data Center and fur-
3Visiting Astronomer, Cerro Tololo Inter-American Observatory, National Optical Astronomy Observatories, operated by the Association of
Universities for Research in Astronomy, Inc. (AURA) under a cooperative agreement with the National Science Foundation.
1
2 Guerrero et al.
ther processed the data using the Chandra X-Ray Cen-
ter software CIAO v3.0.2 and the Calibration Data Base
CALDB v2.25. The background count rate is consistent
with the quiescent background2for most of the observ-
ing time. Only two background “flares” of short duration
occurred. After excluding these high-background periods
from our analysis, the net exposure time was reduced to
39.6 ks. We used this dataset to extract an image in the
0.5-1.8 keV band at a resolution of ∼1.
′′0, and overplot-
ted the X-ray contours on the HST WFPC2 Hαimage
in Figure 2-right to illustrate the relative distribution of
X-ray-emitting gas and the ionized nebular material.
2.3. Echelle Observations
High-dispersion sp ectroscopic observations of Mz 3 were
obtained on 2002 June 23 and 24 using the echelle spec-
trograph on the CTIO 4m telescope. The spectrograph
was used in the long-slit mode to obtain single-order ob-
servations of the Hαand [N ii]λλ6548,6584 lines for an
unvignetted slit length of 3′. The 79 line mm−1echelle
grating and the long-focus red camera were used, result-
ing in a reciprocal dispersion of 3.4 ˚
A mm−1. The data
were recorded with the SITe 2K No. 6 CCD with a pixel
size of 24 µm. This configuration provides a spatial scale
of 0.
′′26 pixel−1and a sampling of 3.7 km s−1pixel−1along
the dispersion direction. The slit width was set to 0.
′′9, and
the resultant instrumental FWHM was 8 km s−1. The an-
gular resolution, determined by the seeing, was better than
1.
′′2.
The echelle observations were made with the slit ori-
ented along different position angles and placed at various
offsets from the central star, in order to sample the com-
plex morphological features of Mz 3. The slit positions and
exposure times of these observations are given in Table 2.
Although both Hαand [N ii] lines are available, only the
[N ii]λ6583 line is used to analyze the kinematics of Mz 3
because of its smaller thermal width. The echellograms
of the [N ii]λ6583 line are presented in Figure 3, where
an [N ii] image is also presented with the slit positions
overplotted.
3. results
The HST narrow-band images of Mz 3 reveal three pairs
of bipolar lobes and one elliptical feature along the equa-
tor of these lobes; they are marked in Figure 1 as BL1,
BL2, BL3, and EE, respectively. These features are also
detected in the echellograms and marked correspondingly
in Figure 3. In the following sections, we discuss de-
tailed morphologies and kinematics and propose spatio-
kinematical models for each of these structures in Mz 3.
3.1. BL1: the Hourglass-Shaped Inner Bipolar Lobes
The innermost pair of bipolar lobes (BL1), called In-
ner Bipolar Lobes (IBL) in Redman et al. (2000), have an
hourglass morphology, with a narrow waist along the east-
west direction (Fig. 1). The position-velocity diagram,
i.e., the long-slit echellogram, also shows a tilted hourglass
pattern (Fig. 3), which can be produced by two shells ex-
panding oppositely along the polar axis with the south
pole tilted toward us. The velocity difference between the
walls of each lobe is &100 km s−1, much larger than the
F W H M of the [N ii] line at the walls, 10–15 km s−1. The
sharp morphology and narrow [N ii] line shape at the walls
of the BL1 lobes indicate that the material originally resid-
ing in the lobes has been evacuated and compressed into
thin shells by a bipolar outflow. The bipolar expansion of
the lobes is a direct consequence of the bipolarity of the
outflow. The lateral expansion of the lobes, on the other
hand, may be driven by the thermal pressure of hot gas
shock-heated by the outflow impinging on the circumstel-
lar material. The hot gas in the central cavities of the BL1
lobes has been detected in X-rays (Fig. 2-right; Kastner et
al. 2003).
The HST images of the BL1 lobes reveal protrusions
from their polar caps indicative of blowouts. The south-
ern lobe shows a single protruding blister at its polar cap,
while the northern lobe shows multiple blister-like struc-
tures extending from the polar cap and converging into
a single blister at the end. The [N ii] echellogram cover-
ing these regions, shown in Figure 4, reveal gas motions
reflecting the blowout process. The blister at the cap of
the southern lobe shows a spindle-shaped [N ii] line that
broadens up to 100 km s−1at its leading edge, while the
multiple blister-like extensions of the northern lobe show
multiple velocity components and the convergent blister at
the end shows a bubble-like structure expanding rapidly
both laterally and radially. The [N ii] echellograms also
detect nebular knots outside the BL1 lobes along the axis
of symmetry, as marked in Figure 4. Exterior to the south-
ern lobe, a bright knot at ∼19′′ from the central star is de-
tected at roughly −30 km s−1from the systemic velocity
(vsys). Exterior to the northern lobe, a counterpart of the
southern knot is detected at ∼19′′ from the central star
with roughly +30 km s−1offset from vsys; in addition, a
fainter knot is detected at ∼25′′ from the central star with
a velocity offset of about +40 km s−1.
The different characteristics of the northern and south-
ern lobes of BL1 are probably caused by the detailed in-
teractions between the bipolar outflow and the dense cir-
cumstellar material. The circumstellar material has a high
concentration in the equatorial plane, as indicated by the
higher extinction around the waist of BL1. The sharp band
of obscuration over the northern BL1 lobe at ∼2.
′′5 north
of the central star suggests that dense equatorial material
is located in front of this lobe and therefore confirms the
orientation of BL1 implied from its kinematics. The vari-
ations in the local extinction, as derived from the Hα/Hβ
ratio map shown in Figure 2-left, support this hypothesis:
the Hα/Hβratio is higher on the northern lobe than on the
southern lobe, and therefore extinction towards the north-
ern lobe is higher, indicating larger amounts of intervening
material. In addition to the surrounding material that ob-
scures the northern BL1 lobe, the central star of Mz 3 is
embedded in a thick, extended shell detected through the
mid-infrared emission of dust (Quinn et al. 1996).
This Hα/Hβratio map discloses additional clues on
the distribution of absorbing material within Mz 3. The
Hα/Hβratio, i.e., the extinction, is especially enhanced at
the projected edge of the lobes and along the bright optical
filaments, suggesting that the expanding lobes carry large
amounts of dust and suffer from self absorption. In agree-
2Reported by M. Markevitch (2001), available at http://cxc.harvard.edu/contrib/maxim/bg/index.html.
The Multipolar Nebula Mz 3 3
ment with Smith’s (2003) conclusions based on the differ-
ent amounts of extinction derived from infrared H iand
[Fe ii] lines, we conclude that a significant fraction of the
extinction towards Mz 3 is lo cal rather than interstellar.
The lo cal nature of the extinction in Mz 3 and its nonuni-
form distribution affects the morphology of the diffuse X-
ray emission which is anticorrelated with the amount of
extinction (Fig. 2), as typically observed in other PNe
(Kastner et al. 2002).
To determine the dynamical age and inclination of the
polar axis for each of the BL1 lobes, the shell morphol-
ogy and position-velocity relation need to be analyzed and
modeled quantitatively. We have adopted a simple expres-
sion to approximate the radial expansion velocity of an
hourglass as a function of the latitude angle, θ(Solf &
Ulrich 1985):
v(θ) = ve+ (vp−ve)×sin(|θ|)γ,(1)
where veand vpare the expansion velocities at the equa-
tor and pole, respectively, and the exponent γsets the lobe
geometry. We have also assumed a homologous expansion
so that
r(θ) = ∆t×v(θ) (2)
where ∆tis the time since the lobe was formed.
Using the model outlined above, we have determined
veand vp, the age, the exponent γ, and the inclination
with respect to the sky and PA of the symmetry (polar)
axis of each of these bipolar lobes. The best fits for the
southern and northern BL1 lobes are shown in Figure 5
and the parameters of these fits are listed in Table 3. As
expected from the different morphological and kinematical
properties of the southern and northern lobes, the best-fit
parameters to each lobe are not exactly the same, though
both fits have similar inclination of the symmetry axis with
respect to the plane of the sky, 15◦–20◦, and kinematical
age3, (600±50)×(D
kpc ) yr, where Dis the distance in kpc
to Mz 3.
If the bright knots at the tip of the bipolar lobes share
their inclination angle, then the true de-projected veloc-
ity of these knots is in the range between 90 km s−1and
150 km s−1. For comparison, we have also included in
Tab. 3 the parameters of the best fit to the northern lobe
considering the extension and kinematics of the converg-
ing blister at its polar cap. The shorter kinematical age
of the northern lobe when its blister is considered may be
suggestive of acceleration of the gas motions caused by a
blowout process.
3.2. BL2: the Cylindrical Bipolar Lobes
The bipolar lobes of BL2, the Outer Bipolar Lobes 1
(OBL1) in Redman et al. (2000), have an almost rectangu-
lar morphology, with the PA’s of the western and eastern
edges having a difference as small as ∼5◦(Fig. 1). The
lobes have a width of ∼24′′ and a length up to ∼85′′ for
the northern BL2 lobe, i.e., the aspect ratio is 7:1. Their
edges are rather straight, bending inwards only at the lo-
cation where these lobes contact the inner BL1 lobes. The
detailed morphology of BL2 shows a complex system of
long filaments extending radially outwards. These fila-
ments originate from a collection of knots at the base of
the BL2 lobes that form a cavity-like structure just outside
the BL1 lobes (Fig. 4).
The [N ii] echellogram of the BL2 lobes along PA 8◦(i.e.,
roughly the BL2 symmetry axis) shows two velocity com-
ponents with a velocity gradient of ∼1.1 km s−1arcsec−1
(Fig. 2). The difference in velocity between these two com-
ponents, ∼110 km s−1, does not change significantly with
the position along the symmetry axis of these lobes. Along
the orthogonal direction, the echellograms at PA 98◦and
offset 14′′, 19′′ , and 26′′ show hollow position-velocity el-
lipses in BL2 (Fig. 2). Material in these lobes is thus
mostly confined in the thin walls of hollow cylinders. This
material cannot just flow along the walls of the cylinder,
as a cross section of such a cylinder would have exactly
the same observed velocity. Instead, the apparently con-
stant velocity-split implies that the section of the cylinder
expands with a constant, ∼55 km s−1, expansion velocity.
Meanwhile, the velocity along the walls increases with the
distance from the central star and must be faster than the
transversal velocity; otherwise the lobes will not show the
high aspect ratio, ∼7:1, that characterizes them.
The Hubble law-like expansion of the BL2 lobes sug-
gests that these lobes were made in a single, explosive
event. As the difference in velocity between the blue- and
red-shifted components at a given location of BL2 is the
same, ∼110 km s−1, the kinematical age of BL2 can easily
be derived assuming that the 24′′ width of BL2 are sim-
ply due to expansion along this direction. The kinematics
are well reproduced using a cylinder4tilted with respect
to the plane of the sky with fixed, 55 km s−1, expansion
velocity across its section and linearly increasing velocity
along the walls (Figure 6). The best fit model has an incli-
nation of 20◦±5◦against the sky plane, in agreement with
the previous value reported by Meaburn & Walsh (1985),
and a kinematical age of (1,000±100)×(D
kpc ) yr. At the
maximum distance of 85′′ from the central star of Mz 3,
the de-projected expansion velocity is ∼320 km s−1.
3.3. BL3: the Conical Lobes
We have named BL3 the pair of bipolar lobes with con-
ical shape, called the Outer Bipolar Lobes 2 (OBL2) by
Redman et al. (2000). These lobes have an opening an-
gle of ∼50◦and their limbs point directly to the central
star of Mz 3. In the images in Fig. 1, the conical lobes
BL3 are composed of multiple knots with long, radial tails
stretching out up to 60′′ from the central star of Mz 3. The
distribution of these knots and filaments is looser than this
of the filaments in BL2. Indeed, the knots and filaments
in BL3 look disconnected, more like individual streams of
material than as part of a contiguous structure.
Further information on the kinematics and structure of
BL3 can be derived from the echellograms at PA 43◦, 52◦,
and −28◦through the central star, and at PA 98◦and
offset 14′′, 19′′ , and 26′′ to the south of the central star
of Mz 3 (Figs. 3 and 7). In the echellograms at PA’s 43◦,
52◦, and −28◦, the knots and filaments composing BL3
appear as tilted straight features with different slopes on
3The distance to Mz 3 is highly uncertain. Hereafter we have chosen to show explicitly the dependence of the kinematical age on distance.
4Actually, these are not cylinders, as they are opening gradually with distance from the central star, but the angle of divergence, ∼5◦, is too
small to affect significantly the model fits.
4 Guerrero et al.
the position-velocity space. The structure and kinemat-
ics of BL3 in these echellograms is somehow confused by
that of BL2, but the echellograms at PA 98◦resolve un-
ambiguously BL3 from BL2. In these echellograms, the
velocities of the knots and filaments of BL3 are mostly
distributed, but not completely confined, along ellipses.
The radial velocities of these ellipses as well as the ve-
locity differences between their red- and blue-shifted sides
increase radially from the central star of Mz 3. The dis-
tribution in the position-velocity space of these knots and
filaments suggests that, unlike BL2, material in BL3 is not
completely confined to the walls of the conical lobes. This
is illustrated by the feature seen in the echellograms at PA
98◦and offset 14′′ South and 19′′ South at relative position
∼ −15′′ and (v−vsys)∼ −100 km s−1(Figs. 2 and 7). This
feature looks like a small velocity ellipse whose spatial size
and difference in velocity increase from the echellogram at
offset 14′′ South to that at 19′′ South, suggesting that this
filament is opening into a conical structure.
The kinematics of the knots and filaments of BL3 de-
rived from these echellograms show that their expansion
velocity follows a Hubble law. We have modeled the kine-
matics of the BL3 lobes assuming that they have a conical
shape with opening angle ∼50◦, and that the expansion
velocity is directed along the walls of the cone and in-
creases linearly outwards from the central star of Mz 3.
Following this model, we have fit the observed kinematics
(Fig. 7) and derived an inclination angle of the symme-
try axis with the plane of the sky of 12◦±3◦. The de-
projected expansion velocity at 60′′ from the central star
would be 180±30 km s−1and the kinematical age of BL3
is (1,800±200)×(D
kpc ) yr. The inclination angle and kine-
matical age derived from this fit have greater uncertainty
than these fitting BL1 and BL2, because the discrete na-
ture of BL3 makes difficult to judge the goodness of the
fit and to determine the best-fit parameters.
3.4. EE: the Equatorial Ellipse
The HST images of Mz 3 displays an additional feature
unnoticed in previous images, a closed ellipse with size
82′′×32′′ oriented along PA∼85◦, i.e., almost along the
nebular equator (Fig. 1). This structure, referred to as
the Equatorial Ellipse (EE), is delineated by filamentary
arcs especially prominent at the northeast and southwest
of Mz 3.
EE is revealed as dramatic high-velocity arcs in the
echelle observations along the slits oriented at PA 98◦and
offsets 3′′ North, and 4′′, 14′′, and 19′′ South of the cen-
tral star, as well as in the slits at PAs 8◦, 52◦, 43◦, and
−28◦(Fig. 3). The measured expansion velocity is close
to 200 km s−1with respect to the systemic velocity. It is
interesting to note that the arcs in the echellograms of the
slits passing through the central star are disrupted by ra-
dial filaments of BL3. It is also interesting to note that the
arcs detected in the echellograms at PA 98◦and offsets 3′′
North and 4′′ South show marked point-symmetry.
It is clear from these results that Mz 3 shows an equato-
rial outflow moving at high velocity. Its three-dimensional
geometry, however, is difficult to envision because the frag-
mented information revealed by the observations and the
likely interaction of EE with BL3. In the following, we will
consider four different geometrical models for this outflow:
(a) an extended equatorial disk, (b) a ring collimating a
bipolar ejection, (c) a pair of wide-opened bipolar lobes,
and (d) an oblate ellipsoid-like shell.
Although equatorial disks have been proposed to play
an important role in the collimation of bipolar PNe, there
is no detection of high velocity equatorial disks in PNe.
An example of equatorial disk can be found in the bipolar
nebula around ηCarinae which shows an equatorial struc-
ture that seems to be an extended equatorial disk (Smith
2002). If EE in Mz 3 were a circular equatorial disk, then
the expansion law with radius on the disk can be inferred
from any echellogram of a slit passing through its center,
simply by applying a scaling factor that depends on the
inclination of the disk with respect to the line of sight,
because all velocities along such a line share the same in-
clination angle with the line of sight. For a circular disk, its
inclination angle can be derived from the observed minor-
to-major axes ratio of the pro jected ellipse. The size of
EE of 82′′×32′′ corresponds to an inclination against the
plane of the sky of the rotation axis of the circular disk of
∼23◦. Using this value for the inclination of the disk and
the information on the expansion velocity law on the slit
at PA 52◦passing through the center of Mz3, we have de-
termined the expansion law with radius on the disk which
is plotted in Figure 8. The velocity in the disk decreases
smoothly with radius up to a given radius, when the ve-
locity decreases sharply. Using this velocity law, we have
produced the position-velocity plots expected for significa-
tive slit positions (Fig. 9). The model deviates significantly
from the position-velocity arcs observed along PA 98◦with
different offsets from the central star (Fig. 9). We conclude
that EE cannot be interpreted as an expanding disk.
A detailed study of the spatio-kinematical properties of
an expanding ring collimating a pair of bipolar lobes is
presented by Solf & Ulrich (1985) for the bipolar nebula
around the symbiotic Mira variable R Aqr. In an expand-
ing ring, the ring itself projects an ellipse onto the sky,
long-slit echellograms along the ellipse major axis show
two arcs in the position-velocity space, one shifted to the
blue and the other to the red, and long-slit echellograms
along the ellipse minor axis reveal a characteristic hour-
glass shaped line. The morphology and kinematics of Mz 3
observed in the echellograms of the slits along PA 98◦are
compatible with this model expectations; however, the slit
at PA 98◦and offset 26′′ South of the central stars does not
detect emission outside the observed ellipse, nor the slits
at PAs 52◦, 43◦, 8◦, and −28◦show the expected hourglass
shape. We conclude that an expanding ring that collimates
bipolar lobes is not appropriate for the three-dimensional
geometry of EE.
We have also considered the possibility that EE is com-
posed by a pair of wide-opened, champagne-glass-shaped
bipolar lobes tilted with the line of sight so that the flow
vector points almost directly to us at the location of the
equatorial waist. This model would explain the observed
kinematics: at locations near the central star, the observed
velocity is large because the line of sight is close to the di-
rection of the flow vector, while at increasing distances
from the central star, the lobes bend and close so that
the direction of the flow vector diverges from the line of
sight and the observed velocity decreases. The pro jection
of these lobes onto the plane of the sky, however, would
The Multipolar Nebula Mz 3 5
not produce an elliptical shape, but two interwined arcs
pointing at opposite directions, as observed, e.g., in the
central regions of MyCn 18 (Sahai et al. 199 9). We thus
disregard this model as the three-dimensional geometry of
EE.
Finally we consider an oblate shell which expands much
faster along the equator than along the poles and whose
symmetry axis is close to the plane of the sky. Assum-
ing a homologous expansion for this shell, we have pro-
duced synthetic position-velocity plots that can be com-
pared with the observed ones to determine the best fit
parameters (Figure 10). The best-fit shell model is a flat
ellipsoid-like whose symmetry axis is tilted against the line
of sight by 70◦±5◦, and the ellipsoid-like has an equa-
torial expansion velocity ≃200 km s−1, a polar velocity
≃70±20 km s−1, and a kinematical age (1,000±50)×(D
kpc )
yr. This model explains satisfactorily the high velocity
arcs observed in the slits at PA 98◦. It also accounts for
the disruption of EE by BL3, which has bored a hole near
the polar regions of EE. Finally, this model also explains
the point-symmetric distribution of arcs observed in the
slits at PA 98◦and offsets 3′′ North and 4′′ South; at
these locations, the detectability of the shell is optimized
because the shell is seen tangentially and the optical path
is thus larger than at other locations.
4. the multipolar structure of mz 3
Previous spatio-kinematical studies of Mz 3 have re-
vealed an increasing level of complexity in this nebula.
L´opez & Meaburn (1983) studied the inner bipolar lobes
(BL1) and concluded that these lobes are hourglass in
shape with the symmetry axis close to the line of sight.
In a later paper, Meaburn & Walsh (1985) determined
with greater accuracy a spatio-kinematical model of the
inner bipolar lobes. Moreover, they extended the spatio-
kinematical study of Mz 3 to the outer regions, reporting
the presence of different sets of bipolar lobes. The low
spatial resolution of the narrow-band images available by
then, however, hampered Meaburn & Walsh’s study: the
bipolar lobes BL2 and BL3 were not distinguished from
each other; the equatorial ellipse was interpreted as an ad-
ditional bipolar lobe; and the detection of a high-velocity
component in the Na iline, correctly interpreted as related
to a high-velocity outflow from Mz 3, was not associated
to the equatorial ellipse EE. More recently, Redman et al.
(2000) obtained high-dispersion spectroscopic observations
along the ma jor axis of Mz 3 that allowed them to describ e
the kinematics of the bipolar lobes BL2 and to find high-
velocity, ∼200 km s−1, components at the location of the
blowout at the tips of the inner bipolar lobes BL1. Be-
cause of the limited spatial coverage of their study, the
association between these high-velocity kinematical com-
ponents and the equatorial ellipse EE was not as clearly
seen as evidenced in our echelle observations obtained at
different slit positions (Fig. 3). Finally, in a simultaneous
study of Mz 3, Santander-Garc´ıa et al. (2004) have de-
rived spatio-kinematical models and kinematical ages for
the three pairs of bipolar lobes that are in complete agree-
ment with these derived here.
The present study reconciles many of the previously re-
ported kinematical features of Mz 3 into a more compre-
hensive view of its physical structure. Mz 3 consists of
four distinct structures, an oblate ellipsoid-like shell and
three pair of bipolar lobes with almost coincident symme-
try axes. The properties of these structures are especially
singular among similar structures observed in bipolar PNe.
Unlike the slowly expanding rings or tori observed in some
bipolar PNe, the oblate ellipsoidal-like shell expands at
high velocity along the equator of the bipolar lobes. Simi-
larly, very few multipolar PNe have pairs of lobes exhibit-
ing the notable differences in opening angle, morphologies
and detailed small-scale structures as the three pairs of
bipolar lobes of Mz 3; BL1 has hourglass-shaped expand-
ing bubbles filled with X-ray-emitting hot gas (Kastner
et al. 2003), while BL2 and BL3 are composed of knots
and filaments following a Hubble flow with cylindrical and
conical shapes, respectively.
Multipolarity has become a common feature among
bipolar nebulae. A growing number of bipolar nebulae
have been noted to have multiple systems of bipolar lobes
either sharing the same symmetry axis or having differ-
ent symmetry axes, e.g., M 2-9, M 2-46, NGC 2440, and
Hen 2-104 (Hora & Latter 1994; Manchado, Stanghellini,
& Guerrero 1996; L´opez, Meaburn, Bryce, & Holloway
1998; Solf 2000; Corradi et al. 2001). Among these mul-
tipolar nebulae, the case of Mz 3 is of especial interest
because the kinematical ages of the different systems of
bipolar lobes in Mz 3 are small and of the order of the
difference in kinematical ages among them. The inner
bipolar lobes BL1 have a kinematical age 500–600×(D
kpc )
yr, the cylindrical lobes BL2 and the equatorial ellipsoid
EE are ∼1,000×(D
kpc ) yr old, and the outer bipolar lobes
BL3 have a somewhat more uncertain kinematical ages of
∼1,800×(D
kpc ) yr. Mz 3 is thus a multipolar nebula in the
making, where BL1 corresponds to the most recent ejec-
tion from Mz 3 central star, EE and BL2 are older and
probably coeval, and BL3 is finally the oldest structure,
although its kinematical age is the most uncertain and we
cannot rule out a formation closer in time to that of BL2.
The different kinematical properties of the three pairs
of bipolar lobes suggest distinct formation scenarios. The
ballistic motion of the two outermost bipolar lobes of Mz 3,
BL2 and BL3, indicates that the gas within these lobes
expands freely under its own inertia. Most likely, these
lobes are the result of two episodes of explosive mass
ejection or outbursts that occurred ∼1,800×(D
kpc ) yr and
∼1,000×(D
kpc ) yr ago. The last episode of mass ejection
responsible of BL2 also resulted in high velocity ejecta
along the equatorial plane that formed EE, the equatorial
ellipse. On the other hand, the morphology and hot gas
content of the innermost pair of lobes, BL1, indicate that
they resulted from the interaction of highly pressurized
hot gas with the surrounding material. This hot gas may
be produced by the onset of a fast stellar wind. An alter-
native origin has been proposed by Kastner et al. (2003)
who attribute the X-ray emission to the action of an X-
ray jet along the symmetry axis of Mz 3. Our observations
indeed reveal bipolar collimated outflows along the sym-
metry axis of Mz 3 (the knots further away the leading
edges of the BL1 lobes as seen in Fig. 4), but not with
the high velocities required to produce the observed X-ray
emission. Note, however, that the outflow detected in our
observations may trace high density material accelerated
6 Guerrero et al.
by a much higher velocity jet that, being responsible of
the X-ray emission, would elude optical detection because
its low density.
The oblate shell forming the equatorial ellipse EE of
Mz 3 is a very singular structural component. Many bipo-
lar PNe show equatorial disks or tori, but all of them have
modest expansion velocities, ∼30 km s−1. Bipolar nebu-
lae around symbiotic stars also show equatorial disks or
tori, but expansion velocities are modest, too. The only
exception among symbiotic stars is the remarkable ellip-
tical shell or ring around Hen 2-147 with an expansion
velocity ∼100 km s−1(Corradi et al. 1999). Thus, the
&200 km s−1equatorial outflow in Mz 3 is the most ex-
traordinary among bipolar PNe and nebulae around sym-
biotic stars.
The equatorial outflow of Mz3 rivals that of the mas-
sive star ηCar. The equatorial outflow around ηCar
shares many similarities with this found in Mz 3: the neb-
ula around ηCar has several systems of bipolar lobes
(Ishibashi et al. 2003) and the formation of the equatorial
outflow has been timed during or around the moment when
the main bipolar lobes in ηCar, the Homunculus Nebula,
were formed. Despite these similarities, the equatorial out-
flows in both nebulae are notably different. The equatorial
outflow in ηCar has been described as an extended equato-
rial disk expanding with velocity proportional to the angu-
lar distance to center (Davidson et al. 2001; Smith 2002),
while the physical structure of the equatorial outflow in
Mz 3 is best described by an oblate shell. Furthermore,
their detailed morphologies are different and very likely
indicate different origins: in ηCar, the equatorial outflow
seems to be composed of multiple jet-like features located
along the equatorial plane, while in Mz 3 the equatorial
outflow shows the limb-brightened morphology character-
istic of a thin shell.
The formation of multipolar nebulae can be explained
as the result of recurrent outbursts as those observed in
massive stars in binary systems during the Luminous Blue
Variable (LBV) phase, e.g. ηCar. In low mass stars,
recurrent outbursts can be related to nova-like eruptions
on the accreting hot component of a symbiotic star or to
structural instabilities in the late evolution of the central
star of a PN (e.g., thermal pulses). In symbiotic novae, the
timescales of successive outbursts are determined by the
mass of the accreting white dwarf, the mass loss rate of the
red giant, and the accretion efficiency of the wind capture
which is related to the binary interaction (e.g. Prialnik &
Kovetz 1995). Recurrence periods of a few hundred years
are typical of symbiotic novae (Prialnik & Kovetz 1995;
Corradi et al. 1999). The formation of multipolar PNe is
far more difficult to explain, as it requires the alternation
between a dense, slow wind and a fast, tenuous wind. The
evolution of the central star of the PN in a binary system
provides a natural scenario for recurrent outbursts during
the evolution through a common envelope phase or as the
result of accretion and nova-like outbursts on the white
dwarf component of a symbiotic star. This raises the simi-
larities between Mz 3 and other symbiotic stars like R Aqr
and Hen 104, or other suspected symbiotic stars yet clas-
sified as PNe, e.g., M 2-9, and casts doubts on the true
nature of Mz 3 as a PN.
Even if we accept that Mz 3 has formed as the result of
recurrent nova-like outbursts in a symbiotic star, the phys-
ical structure of this bipolar nebula is rather unique. The
successive collimated ejections in Mz 3 are rather regular in
time, but they have very different morphological and kine-
matical properties, which suggest very distinct conditions
and formation mechanisms. In Mz 3, we are thus witness-
ing the formation of a multipolar nebula which evolves
dramatically between periodic outburst episodes.
M.A.G. and L.F.M. acknowledge support from the grant
AYA 2002-00376 of the Spanish MCyT (cofunded by
FEDER funds). We thanks Miguel Santander Garc´ıa for
providing us with the results on their spatio-kinematical
modeling of Mz 3 before publication. We also thank the
referee, Dr. Matt Redman, for his valuable comments.
REFERENCES
Corradi, R. L. M., Ferrer, O. E., Schwarz, H. E., Brandi, E., & Garc´ıa,
L. 1999, A&A, 348, 978
Corradi, R. L. M., Livio, M., Balick, B., Munari, U., & Schwarz, H.
E. 2001, ApJ, 553, 211
Davidson, K., Smith, N., Gull, T. R., Ishibashi, K., & Hillier, D. J.
2001, AJ, 121, 1569
Hora, J. L., & Latter, W. B. 1994, ApJ, 437, 281
Kastner, J. H., Balick, B., Blackman, E. G., Frank, A., Soker, N.,
Vrtilek, S. D., & Li, J. 2003, ApJ, 591, L37
Kastner, J. H., Li, J., Vrtilek, S. D., Gatley, I., Merrill, K. M., &
Soker, N. 2002, ApJ, 581, 1225
Ishibashi, K. et al. 2003, AJ, 125, 3222
L´opez, J. A. & Meaburn, J. 1983, MNRAS, 204, 203
L´opez, J. A., Meaburn, J., Bryce, M., & Holloway, A. J. 1998, ApJ,
493, 803
Manchado, A., Stanghellini, L., & Guerrero, M. A. 1996, ApJ, 466,
L95
Meaburn, J. & Walsh, J. R. 1985, MNRAS, 215, 761
Prialnik, D. & Kovetz, A. 1995, ApJ, 445, 789
Quinn, D. E., Moore, T. J. T., Smith, R. G., Smith, C. H., &
Fujiyoshi, T. 1996, MNRAS, 283, 1379
Redman, M. P., O’Connor, J. A., Holloway, A. J., Bryce, M., &
Meaburn, J. 2000, MNRAS, 312, L23
Sahai, R. et al. 1999, AJ, 118, 468
Santander-Garc´ıa, M., Corradi, R. L. M., Balick, B., & Mampaso,
A. 2004, submitted to A&A
Schmeja, S. & Kimeswenger, S. 2001, A&A, 377, L18
Smith, N. 2002, MNRAS, 337, 1252
Smith, N. 2003, MNRAS, 342, 383
Solf, J. 2000, A&A, 354, 674
Solf, J. & Ulrich, H. 1985, A&A, 148, 274
Zhang, Y. & Liu, X.-W. 2002, MNRAS, 337, 499
The Multipolar Nebula Mz 3 7
Fig. 1.— HST WFPC2 images of Mz 3 in the Hα(top) and [N ii]λ6583 (bottom) emission lines. The different morphological components
of this nebula are marked as described in the text: BL1 are the inner bipolar lobes, BL2 the cylindrical lobes, BL3 the conical lobes, and EE
is the equatorial ellipse. Both the Hαand [N ii] images are displayed at two intensity contrasts and spatial scales to highlight these different
morphological components.
Fig. 2.— Left: Hαto Hβintensity ratio map of Mz 3. Bright regions correspond to higher Hα/Hβratio and thus to higher extinction.
Right: HST WFPC2 Hαimage of Mz 3 overlaid with X-ray contours in the 0.5–1.8 keV energy range extracted from a Chandra ACIS-S
observation.
Fig. 3.— [N ii]λ6583 image (top-left) and echellograms of Mz 3 along 9 different slit positions. The slits positions of the echelle observations
are plotted over the [N ii] image. The different morphological components of this nebula are marked on the echellograms. Note that the
spatial scale of the image and echellograms are not coincident. Note also that velocities have been referred to the systemic velocity of M z 3.
Fig. 4.— [N ii]λ6583 image (left) and echellogram along PA 8◦(right) of Mz 3. Both the image and echellogram have the same orientations
and spatial scales to make easy a fair comparison. The arrows indicate the locations in the image and the echellogram of the knots at the
tips of the innermost bipolar lobes described in the text. Contrast in the image has been chosen to highlight these features.
Fig. 5.— [N ii]λ6583 image (left) and echellogram along PA 8◦(right) of Mz 3 overlaid by the best model fit for the hourglass bipolar lobes
described in §3.1. Image and echellogram are shown at the same spatial scale.
Fig. 6.— [N ii]λ6583 echellograms of Mz 3 along selected s lit p ositions marked on the figure. The echellograms are overlaid by the
position-velocity plots derived from the best model fit for the cylindrical bipolar lobes BL2 described in §3.2.
Fig. 7.— [N ii]λ6583 echellograms of Mz 3 along selected s lit p ositions marked on the figure. The echellograms are overlaid by the
position-velocity plots derived from the best model fit for the conical bipolar lobes BL3 described in §3.3.
Fig. 8.— Radial dependence of the expansion velocity of an expanding equatorial disk as inferred from the information of the EE component
in the echellogram along PA 52◦. The circular disk has an inclination 23◦consistent with the minor-to-major axes ratio of EE.
Fig. 9.— [N ii]λ6583 echellograms of Mz 3 along selected s lit p ositions marked on the figure. The echellograms are overlaid by the
position-velocity plots derived from the best model fit for an expanding equatorial disk with the expansion law given in Figure 8. This best
model fit reproduces the kinematics of EE seen in the echellograms for slit positions passing through the nebular center, but the fit is very
poor for the slit positions offset from the center and at directions orthogonal to the projected symmetry axis of the nebula.
Fig. 10.— [N ii]λ6583 echellograms of Mz 3 along selected slit positions marked on the figure. The echellograms are overlaid by the
position-velocity plots derived from the best model fit for an oblate shell as described in §3.4. This model produces an acceptable fit for all
slit positions.
8 Guerrero et al.
Table 1
Archival HST WFPC2 Observations of Mz 3
Emission Line Exposure Time Location Program ID
(sec) PC1/WF3
Hα350 PC1 9050
Hα900 WF3 6856
Hβ1300 WF3 6502,6856
[N ii] 1300 PC1 9050
[N ii] 900 WF3 6856
Table 2
Echelle Observations
Offset Position Angle Exposure Time
(′′) (◦) (sec)
0 8 600
2 W 43 900
0 52 900
3 N 98 1800
4 S 98 1800
8 S 98 1800
14 S 98 1800
19 S 98 1800
26 S 98 1800
0−28 900
Table 3
Fits of the Physical Parameters of BL1
Bipolar Lobe vpveiPA Kin. Age
(km s−1) (km s−1) (◦) (◦) (yr)
Southern Lobe 100 15 15 8 600
Northern Lobe 80 15 20 12 600
Northern Lobe and Blister 140 15 20 10 520
