Circumstellar masers in the Magellanic Clouds
ABSTRACT We have searched for 22 GHz H2O and 43/86/129 GHz SiO masers in bright IRAS point sources in the SMC and LMC, to test whether the kinematics of the mass loss from these stars depends on metallicity. H2O masers were detected in the red supergiants IRAS04553-6825 and IRAS05280-6910, and tentatively in the luminous IR object IRAS05216-6753 and the AGB star IRAS05329-6708. SiO masers were detected in IRAS04553-6825. The outflow velocity increases between the H2O masing zone near the dust-formation region and the more distant OH masing zone from 18 to 26 km/s for IRAS04553-6825 and from 6 to 17 km/s for IRAS05280-6910. The total sample of LMC targets is analysed in comparison with circumstellar masers in the Galactic Centre. The photon fluxes of circumstellar masers in the LMC are very similar to those in the Galactic Centre. The expansion velocities in the LMC appear to be 20% lower than for similarly bright OH masers in the Galactic Centre, but the data are consistent with no difference in expansion velocity. OH/IR stars in the LMC appear to have slower accelerating envelopes than OH/IR stars in the Galactic Centre. Masers in the LMC have blue-asymmetric emission profiles. This may be due to the amplification of stellar and/or free-free radiation, rather than the amplification of dust emission, and may be more pronounced in low metallicity envelopes. SiO maser strength increases with the photometric amplitude at 2.2 micron but is independent of the photometric amplitude at 10 micron. This suggests a strong connection between shocks in the dust-free SiO masing zone and the dust formation process. Appendices describe H2O maser emission from R Dor in the Milky Way, optical echelle spectroscopy of IRAS04553-6825, and the properties of masers in the Galactic Centre (Abridged). Comment: 19 pages, 17 figures. Accepted for publication in Astronomy & Astrophysics Main Journal
arXiv:astro-ph/0101125v1 9 Jan 2001
Astronomy & Astrophysics manuscript no.
(will be inserted by hand later)
Circumstellar masers in the Magellanic Clouds
Jacco Th. van Loon1, Albert A. Zijlstra2, Valent´ ın Bujarrabal3, Lars-˚ Ake Nyman4,5
1Institute of Astronomy, Madingley Road, Cambridge CB3 0HA, United Kingdom
2UMIST, P.O.Box 88, Manchester M60 1QD, United Kingdom
3Observatorio Astron´ omico Nacional, Campus Universitario, Apartado 1143, E-28800 Alcal´ a de Henares, Spain
4European Southern Observatory, Casilla 19001, Santiago 19, Chile
5Onsala Space Observatory, S-439 92 Onsala, Sweden
Received date; accepted date
Abstract. Results are presented of a search for 22 GHz H2O 616 → 523, 43 GHz SiOv=1(J = 1 → 0), 86 GHz
SiOv=1(J = 2 → 1) and 129 GHz SiOv=1(J = 3 → 2) maser emission from bright IRAS point sources in the Small
and Large Magellanic Clouds — mostly circumstellar envelopes around obscured red supergiants and Asymptotic
Giant Branch stars (OH/IR stars). The aim of this effort was to test whether the kinematics of the mass loss
from these stars depends on metallicity.
H2O maser emission was detected in the red supergiants IRAS04553−6825 and IRAS05280−6910, and tentatively
in the luminous IR object IRAS05216−6753 and the AGB star IRAS05329−6708, all in the LMC. SiOv=1(J =
2 → 1) maser emission was detected in IRAS04553−6825.
The double-peaked H2O maser line profiles of IRAS04553−6825 and IRAS05280−6910, in combination with the
OH (and SiO) maser line profiles, yield the acceleration of the outflows from these stars. The outflow velocity
increases between the H2O masing zˆ one near the dust-formation region and the more distant OH masing zˆ one
from v ∼ 18 to 26 km for IRAS04553−6825 and from v ∼ 6 to 17 km s−1for IRAS05280−6910.
The total sample of LMC targets is analysed in comparison with circumstellar masers in the Galactic Centre. The
photon fluxes of circumstellar masers in the LMC are found to be very similar to those in the Galactic Centre. The
expansion velocities in the LMC appear to be ∼ 20% lower than for similarly bright OH masers in the Galactic
Centre, but the data are still consistent with no difference in expansion velocity. OH/IR stars in the LMC appear
to have slower accelerating envelopes than OH/IR stars in the Galactic Centre.
The masers in the LMC have blue-asymmetric emission profiles. This may be due to the amplification of stellar
and/or free-free radiation, rather than the amplification of dust emission, and may be more pronounced in low
The SiO maser strength increases with the photometric amplitude at 2.2 µm but is independent of the photometric
amplitude at 10 µm. This suggests a strong connection between shocks in the dust-free SiO masing zˆ one and the
dust formation process. The LMC masers obey the same trend as the Galactic Centre masers.
Appendices describe H2O maser emission from the moderately mass-losing AGB star R Dor in the Milky Way,
optical echelle spectroscopy of IRAS04553−6825, and the properties of circumstellar masers in the Galactic Centre.
Key words. Masers – circumstellar matter – Stars: mass loss – Stars: AGB and post-AGB – supergiants –
In the latest stages of their evolution, both massive and
intermediate-mass stars pass through a phase of intense
mass loss at rates of 10−6to 10−3M⊙yr−1(van Loon et
al. 1999b), returning a major fraction of their initial mass
to the interstellar medium (ISM). For stars of Minitial>∼8
M⊙this occurs when they are red supergiants (RSGs), and
for stars of 1<∼Minitial<∼8 M⊙when they are Asymptotic
Giant Branch (AGB) stars before becoming a Planetary
Nebula (PN). They become enshrouded by their dusty cir-
cumstellar envelope (CSE), rendering them invisible at op-
tical wavelengths. The absorbed radiation is re-emitted by
the dust at longer wavelengths, making them very bright
IR objects. This is also when, in oxygen-rich CSEs, maser
emission from OH, H2O and SiO molecules may be ob-
served. Hence these objects are known as OH/IR stars.
The locations and intensities of masers are determined
by molecular abundance, dust temperature, gas density,
and velocity (Goldreich & Scoville 1976; Lewis 1989).
2 van Loon et al.: Circumstellar masers in the Magellanic Clouds
Where H2O is not shielded from interstellar UV radiation
it is dissociated into OH. SiO is depleted into dust grains
except close to the star. Maser strengths scale with the
local kinetic energy density or the local IR radiation field,
which both depend on the CSE temperature. Masers are
quenched above a critical density. Finally, the radiation
field in the maser transition is amplified only by molecules
that have small projected velocity differences (within the
thermal width of the maser line, i.e.<∼1 km s−1).
As a result the CSE has a layered maser structure, cor-
responding to subsequent stages in the mass-loss process
from the underlying star. Strong pulsations of the stellar
photosphere of these Long Period Variables (LPVs) eject
matter in which dust forms at typically 1 to 10 stellar radii
(R⋆). Radiation pressure accelerates the matter to veloc-
ities exceeding the escape velocity. Matter that does not
reach the dust formation radius falls back to the star. At
distances of>∼100R⋆, the stellar wind flows with constant
velocity v∞into interstellar space. SiO masers probe the
inner dust-free zˆ one, H2O masers probe the acceleration
zˆ one, and OH masers probe the final stellar wind. This
has been beautifully confirmed by interferometric obser-
vations of Galactic OH/IR stars (Diamond et al. 1984;
Richards et al. 1996; Colomer et al. 2000).
The physical conditions in the CSE, and the evolution-
ary stage of the star can be determined from the presence
or absence of different species of masers, and from their
photon fluxes. The kinematic structure of the CSE can be
determined from the maser line profiles. SiO maser radia-
tion is predominantly amplified tangentially, resulting in a
single maser peak within a few km s−1of the stellar veloc-
ity v⋆. OH maser radiation is beamed radially, resulting in
a double-peaked line profile, spanning 2×v∞. H2O masers
are single-peaked in Mira variables but double-peaked in
OH/IR stars (Takaba et al. 1994). Hence, in OH/IR stars
H2O masers yield the expansion velocity at the base of
the acceleration zˆ one in the CSE.
A significant population of confirmed and suspected
OH/IR stars in the Small and Large Magellanic Clouds
(SMC & LMC) has now been identified and studied (Wood
et al. 1992; Zijlstra et al. 1996; Loup et al. 1997; van
Loon et al. 1997, 1998a, 1999a,b). The metallicities of the
intermediate-age populations in the SMC and LMC are
∼ 7 and ∼ 3× lower than solar, and hence the dependence
of the mass loss on metallicity may be investigated (van
Loon 2000). A study of Magellanic circumstellar masers
may shed light on the metallicity dependence of the en-
velope kinematics. Here the final results are presented of
surveys for H2O and SiO masers in the Magellanic Clouds
(van Loon et al. 1996, 1998b).
2. Radio observations
A summary of the observations can be found in Table 1.
No attempts were made to better determine the positions
of (tentatively) detected maser sources.
2.1. H2O maser emission at 22 GHz with Parkes
The 64 m radio telescope at Parkes, Australia, was used
with the 1.3 cm receiver plus autocorrelator to observe
22 GHz H2O masers. Using the Dual Circular feed spec-
tra were obtained simultaneously in left and right circular
polarization. No difference was found between them, and
they were averaged. The current K-band facility at Parkes
is not as powerful as the beam-switching set-up used in
the early 1980’s that yielded Tsys ∼ 90 K (Whiteoak &
Gardner 1986). Weather conditions were generally fair.
served, for six and three hours on-source integration,
respectively. The nearby sky was measured every two
minutes, resulting in very flat baselines that required
only a shallow second-order polynomial to be subtracted.
The 22 GHz discovery spectrum of IRAS04553−6825
was already presented in van Loon et al. (1998b). In the
2000 run, on-source integration times were a few hours
per target. The sky was measured every two minutes —
most of the time at 5′N of the source. Pointing, focus
and calibration were checked regularly by observing the
nearby bright maser source R Dor (see Appendix A) at
different zenith distances. The pointing accuracy was
found to be ∼ 10′′and flux losses due to pointing errors
are less than ∼ 10%.
2.2. H2O maser emission at 22 GHz with Mopra
The 22 m radio telescope at Mopra, Australia, was used
with the 1.3 cm receiver plus autocorrelator to observe
22 GHz H2O masers. The opacity correction was usually
between 1.3 and 1.6. The conversion factor agrees with the
observed noise and flux density for R Dor (48 Jy).
The background was measured every four minutes at
both sides of the target. The baseline was estimated for
each individual spectrum by a running average (within
1-σ around the median) over 55 channels. After subtract-
ing the baselines, the flux density was obtained by av-
eraging (within 3-σ around the median) over the time
series of spectra. This procedure yielded flat baselines
and effectively rejected bad spectra. Flux conservation
was tested on the well-detected H2O maser spectra of
2.3. SiO maser emission at 43 GHz with Parkes
The 64 m radio telescope at Parkes, Australia, was used
with the 0.7 cm single polarization receiver plus correla-
tor to observe 43 GHz SiO masers. The background was
measured by alternation between two targets every 150
seconds. Reasonably flat baselines were obtained by sub-
tracting a fifth order polynomial plus a sine with a period
of 200 channels. The useful heliocentric velocity range runs
from 167 to 322 km s−1.
van Loon et al.: Circumstellar masers in the Magellanic Clouds3
Table 1. Summary of observations: observatory, dates, transition, rest frequency, centre velocity, channel velocity
width, number of channels, bandwidth, effective telescope diameter, beam Full-Width-Half-Maximum, typical system
temperature and conversion factor. See text for more SEST observations.
H2O 616 → 523
H2O 616 → 523
H2O 616 → 523
SiO 1 → 0 (v = 1)
SiO 2 → 1 (v = 1)
SiO 2 → 1 (v = 1)
SiO 2 → 1 (v = 1)
SiO 3 → 2 (v = 1) 129.363368
2.4. SiO maser emission at 86 GHz with Mopra
The 22 m radio telescope at Mopra, Australia, was used
with the 3 mm SIS receiver plus correlator to observe 86
GHz SiO masers. The spectra were obtained simultane-
ously in two orthogonal polarizations that were then av-
eraged. The pointing was checked every few hours on R
Dor, and was accurate to < 15′′rms. The background
was measured by alternation between two of these targets
every few minutes. The baselines were very flat.
2.5. SiO maser emission at 86 & 129 GHz with SEST
The 15 m Swedish-ESO Sub-mm Telescope (SEST) at the
European Southern Observatory (ESO) at La Silla, Chile,
was used with the 3 mm receiver plus the acousto-optical
High Resolution Spectrograph (HRS) to observe 86 GHz
SiO masers. This configuration was used simultaneously
with the Low Resolution Spectrograph (LRS) either tuned
at the same frequency or coupled to the 2 mm receiver to
observe 129 GHz SiO masers.
The internal absolute flux calibration is accurate to ∼
20%. The observations were done in Double Beam Switch
Mode, with a beam throw of ∼ 11.5′in azimuth. The
pointing was checked every few hours on R Dor, and was
accurate to ∼ 3′′rms. The atmospheric conditions were
very good: a relative humidity of typically 15 to 30%, an
outside air temperature of ∼ 15◦C, and little or no cirrus.
The baselines were very flat, requiring only a zeroth order
polynomial to be subtracted from each spectrum.
IRAS04553−6825, after its 86 GHz SiO maser emis-
sion had been detected (van Loon et al. 1996), was re-
observed on 7 occasions during the period from July 1996
to January 1998, with the SEST and the HRS and LRS
at 86 GHz using the same observing strategy as described
above. The total on-source integration time at 86 GHz was
65 and 51 hr with the HRS and LRS, respectively. During
this campaign IRAS05329−6708 and IRAS05280−6910
were re-observed in January 1998.
Table 2. The SMC & LMC targets, together with their
IRAS flux densities (in Jy), classificationsand whether OH
1612 MHz masers are detected (yes), not detected (no), or
not tried ( ).
†IRAS00486−7308 is from the IRAS-FSC with flux densities
from Groenewegen & Blommaert (1998).
H ii ?
3. Results from the maser searches
The targets comprise the known Magellanic OH/IR stars,
plus other IR objects in the SMC and LMC: mainly dust-
enshrouded AGB stars, RSGs or H ii emission objects
(LHA numbers from Henize 1956) often related to sites
of recent/on-going star formation. Additional 22 GHz ob-
servations of the Galactic AGB star R Dor are described
in Appendix A. The IRAS names of the targets are listed
in Table 2, together with the IRAS flux densities at 12
and 25 µm, object classification, and whether OH maser
emission has been detected (yes), not detected (no), or
4van Loon et al.: Circumstellar masers in the Magellanic Clouds
Fig.1. Diagram of 25 µm flux density versus 25 over 12
µm flux density ratio for the targets of the maser search,
distinguished according to their object classes (RSGs, H ii
regions, AGB stars). The flux densities of the SMC objects
have been scaled by +0.2 dex to the distance of the LMC.
not tried ( ) by Wood et al. (1986, 1992) or van Loon et
al. (1998a). The results for the targets (tentatively) de-
tected in at least one of the H2O or SiO maser transitions
are summarised in Table 3, whilst upper limits (3-σ noise
levels) for the non detections are summarised in Table 4.
The IRAS mid-IR flux densities and flux density ratios
often carry information about the nature of the IR source
(Fig. 1). The AGB stars amongst the targets represent
the most luminous AGB stars with the highest mass-loss
rates encountered in the MCs, with typical mid-IR flux
densities of ∼ 1 Jy. Because of their higher luminosity and
sometimes also higher mass-loss rate, RSGs may become
somewhat redder and more than an order of magnitude
brighter at 25 µm than these extreme AGB stars. RSGs,
due to their short lifetimes, may still be associated with H
ii regions that usually have very red mid-IR colours. Very
young supergiants may be dust-enshrouded inside of an
ultracompact H ii region with similarly red colours (see,
for instance, Persi et al. 1994). The individual targets are
described below — with a general reference to Loup et al.
1997 — together with the results from our maser search.
3.1. RSGs in the LMC
This source may be interpreted as a luminous RSG or
AGB star, with a high mass-loss rate of
M⊙yr−1(van Loon et al. 1999b) and detected by IRAS
even at 100 µm (Trams et al. 1999). It is located in the H
˙ M ∼ 8 × 10−4
Table 4. Upper limits (3-σ noise levels, in Jy) for the
non detections: H2O 22 GHz with Parkes (P, 2000) and
Mopra (M), SiO 43 GHz with Parkes, and SiO 86 GHz
†observed with SEST at 86 GHz (HRS & LRS) and 129 GHz,
with upper limits of 0.29, 0.09 and 0.17 Jy, respectively.
Fig.2. IRAS04530−6916: Mopra and Parkes (2000) 22
GHz spectra. The velocities are heliocentric. The bold-
faced curves are the spectra smoothed by a gaußian of
σ = 2.0 km s−1. A possible H2O maser peak at vhel∼ 200
km s−1in the Mopra spectrum could not be confirmed at
Parkes despite the much lower noise levels.
ii region DEM L15. Variability in the near-IR is reported
by Wood et al. (1992), with a period of ∼ 1260 days, but
the amplitudes are only<∼0.4 mag even in the J-band. An
I-band spectrum of the near-IR object associated with the
IRAS source reveals an early-type emission-line spectrum
van Loon et al.: Circumstellar masers in the Magellanic Clouds5
Table 3. Peak flux densities (± 1-σ, in Jy) for objects detected (boldface) in a H2O and/or SiO maser transition:
H2O 22 GHz with Parkes (P; 1997 & 2000) and Mopra (M), SiO 43 GHz with Parkes, SiO 86 GHz with Mopra and
SEST (S; HRS & LRS), and SiO 129 GHz with SEST. Tentative detections are between parentheses, 3-σ upper limits
are given for non detections.
<0.17 (0.026±0.007) <0.027
(Cioni et al., in preparation) adding some confusion with
regard to the nature of this object.
A narrow 3-σ peak (∼ 150 mJy) at vhel∼ 200 km s−1
in the Mopra 22 GHz spectrum could not be confirmed at
Parkes at much lower noise levels (Fig. 2), possibly due to
temporal variability of the H2O maser emission (see, for
instance, Persi et al. 1994) or to beamsize differences if
the source is at ∼ 2′from IRAS04530−6916. The maser
velocity would be rather different from the 21 cm H i that
peaks at vhel∼ 260 km s−1(Kim et al. 1999), but this is
also the case for some of the H2O masers in the giant star
forming region 30 Doradus (van Loon & Zijlstra 2000).
This supergiant is of spectral type M2 (WOH S71:
Westerlund et al. 1981).
IRAS04553−6825 is a very luminous obscured RSG that
has been extensively discussed in the past (van Loon et
al. 1999a,b and references therein). OH maser emission at
1612 and 1665 MHz was discovered by Wood et al. (1986,
1992), SiO maser emission at 86 GHz by van Loon et al.
(1996), and H2O maser emission at 22 GHz by van Loon
et al. (1998b).
The H2O maser emission from IRAS04553−6825 was
confirmed with Mopra, and re-observed with Parkes on
two occasions in the 2000 run. The (integration time &
system temperature weighted) average of the latter is pre-
sented in Fig. 3 (top panel). The integrated flux of the H2O
maser emission between 255 and 307 km s−1equals 0.65
Jy km s−1, corresponding to a photon flux of 1.0 × 1045
s−1. The very narrow main peak is brighter in the 2000
run at least partly due to the higher spectral resolution,
and may, in fact, still be unresolved. The double peak at
blue-shifted velocities is intriguing, as the shape closely re-
sembles that of the 1612 MHz OH emission profile (Wood
et al. 1992) but at slightly smaller velocity displacements
of −13 and −18 km s−1with respect to the main H2O
maser peak at vhel = 278.5 km s−1, instead of −14 and
−26 km s−1. This strongly suggests that the blue-shifted
H2O maser peaks and the OH maser peaks are formed
Fig.3. IRAS04553−6825: Spectra of the 22 GHz H2O
(top; Parkes 2000) and 86 GHz SiO (middle; SEST) maser
emission. The boldfaced curve in the upper panel is the
spectrum smoothed by a gaußian of σ = 0.5 km s−1, whilst
in the middle panel it is the LRS spectrum. The velocities
are heliocentric. Also indicated are the velocities of the
1612 MHz OH maser peaks (arrows) and the velocity of
the H2O maser peak (vertical dotted line). The 1612 MHz
spectrum of Wood et al. (1992) is plotted in the lower
panel for comparison.
in related but displaced regions in the CSE outflow. The
blue-shifted H2O maser emission was not noticed in the
discovery spectrum (van Loon et al. 1998b) but can be
recovered in that data a posteriori. The additional red-
shifted H2O maser peak perhaps arises from some local
density enhancement or increased coherent path length in
part of the CSE at the far side of the star.
6van Loon et al.: Circumstellar masers in the Magellanic Clouds
The composite spectrum of all SEST observations at
86 GHz of IRAS04553−6825 is presented in Fig. 3 (middle
panel). The HRS spectrum is overlaid by the LRS spec-
trum. The quality of the individual spectra is insufficient
for a detailed study of spectral changes with pulsational
phase: no variability more than a factor two is seen in
the peak or integrated flux density. The SiO maser prop-
erties have been listed in van Loon et al. (1998b). SiO
maser emission can be traced between velocities of ∼ 260
and 290 km s−1, and there is evidence for the peak of the
emission to be split into two peaks ∼ 8 km s−1apart. The
shoulder of SiO maser emission between ∼ 260 and 266 km
s−1may be related to the blue-shifted H2O maser peaks.
The velocity of the main H2O maser peak falls right in the
centre of the double-peaked SiO emission.
Two lines at 298 and 305 km s−1are detected appar-
ently in absorption (Fig. 3, middle panel), and they are
also visible at several individual epochs. SiO masing lines
have never been seen in absorption in Galactic sources for
good reasons: the excitation temperature is ∼ 1800 K. The
features may correspond to SiO emission in the off-source
beam. It is hard to trace back this source as the beam-
switching was done in azimuth and hence the off-source
beam swept across the sky during each observing session.
This is a variable star (HV2255) of spectral type M4.
This object is very bright in the IR, with possibly an early-
type supergiant underlying its CSE (Zijlstra et al. 1996).
Marginal variability is detected by Wood et al. (1992),
who suggest it may be a proto-PN or post-RSG object
and who note the similarity with IRAS04530−6916.
The 22 GHz spectrum with Parkes shows a peak of
∼ 45 mJy — a 3 to 4-σ level after some smoothing — at
vhel∼ 284 km s−1(Fig. 4). The peak has a FWHM of ∼ 3
km s−1. The integrated flux of the emission is ∼ 0.5 Jy
Fig.4. IRAS05216−6753: Spectrum of the tentative de-
tection of 22 GHz H2O maser emission (indicated by
the arrow). The velocities are heliocentric. The boldfaced
curve is the spectrum smoothed by a gaußian of σ = 0.5
Fig.5. IRAS05280−6910: Spectra of the 22 GHz H2O
(top; Parkes 2000) and 86 GHz SiO (middle; SEST) maser
emission. The boldfaced curves are the spectra smoothed
by a gaußian of σ = 0.5 km s−1. The velocities are helio-
centric. Also indicated are the velocities of the 1612 MHz
OH maser peaks (arrows) and the velocity of the H2O
maser peak (vertical dotted line). The 1612 MHz spec-
trum of Wood et al. (1992) is plotted in the lower panel
km s−1, corresponding to a photon flux of ∼ 8×1044s−1.
There may be a secondary peak at 289 km s−1.
The 1665 MHz OH maser attributed to IRAS05280−6910
is located at the centre of the cluster NGC1984 (Wood
et al. 1992). The IRAS 12, 25 and 60 µm flux densities
suggest a very cool and luminous CSE, possibly a post-
RSG object. The proposed near-IR counterpart is with
K = 8.19 mag very bright, but with (J − K) = 1.01
not very red (Wood et al. 1992). Within the Parkes beam
of ∼ 13′at OH frequencies are three luminous late type
variable stars of V ∼ 13 and I ∼ 10 mag, a PN (SMP LMC
64), several IR objects, the H ii region DEM L198 at 4.2′
W, and the cluster NGC1994 (IRAS05287−6910) at ∼ 4′
E. It may well be that only the 1612 MHz emission arises
from the CSE of an evolved star in the region of NGC1984,
whilst the 1665 MHz emission arises from the ISM.
Bright and complex H2O maser emission was detected
with Parkes (Fig. 5, top panel). The very narrow and
probably unresolved main peak is at vhel= 272 km s−1,
van Loon et al.: Circumstellar masers in the Magellanic Clouds7
which also defines the centroid velocity of the OH 1612
MHz maser emission as observed by Wood et al. (1992).
This confirms the cluster object IRAS05280−6910 to be
the source of the 1612 MHz OH maser emission. A rather
broad and bright additional peak is blue-shifted by ∼ 5
km s−1, and fainter and possibly sub-structured emission
is seen at red-shifted velocities of ∼ 3 to 8 km s−1with
respect to the main peak. The integrated flux of the H2O
maser emission between 260 and 284 km s−1equals 0.42
Jy km s−1, corresponding to a photon flux of 6.5 × 1044
s−1, not much less than IRAS04553−6825. Both the main
and the blue-shifted additional peaks can be recovered a
posteriori in the Mopra 22 GHz data of IRAS05280−6910.
The 86 GHz SiO maser spectrum of IRAS05280−6910
is presented in Fig. 5 (middle panel). A constant baseline
offset of 2 mJy has been subtracted. The strongest peak,
at a heliocentric velocity of 292 km s−1, has a peak flux
density of 0.30 Jy (= 3.8σ). There is another (2.8σ) peak
at a velocity of 284 km s−1, and a hint of faint emission at
vhel∼ 265 km s−1(at which also H2O maser emission is
seen). The integrated flux of the emission around 265, 284
and 292 km s−1equals ∼ 0.8 Jy km s−1, corresponding to
a photon flux of 1.2×1045s−1. The SiO detection is rather
uncertain, though, and might still be (strong) noise.
This is thought to be a very luminous and highly obscured
RSG (Elias et al. 1986), but this is still uncertain. Its
IRAS-LRS spectrum is odd in that it shows a flat contin-
uum with peculiar features (Kwok et al. 1997).
3.3. AGB stars in the LMC
The AGB stars amongst the LMC targets are oxygen rich,
very luminous, variable on long timescales (∼ 103days)
and severely dust-enshrouded (Wood et al. 1992; van Loon
et al. 1998a, 1999a,b). For some, mid-IR spectra have been
obtained by Trams et al. (1999), showing the 10 µm sili-
cate dust feature in absorption with some emission wings
remaining (see also Groenewegen et al. 1995 and Zijlstra
et al. 1996 for IRAS05329−6708).
These stars comprise all known AGB sources of OH
maser emission in the LMC, except the bluest amongst
them, IRAS04498−6842, that has not (yet) been detected
at 1612 MHz. The only typical saturated double-peaked
OH maser emission profile is found in IRAS05298−6957.
IRAS04407−7000 is single peaked at 1612 MHz (van Loon
et al. 1998a), and consequently no expansion velocity of
the CSE can be determined, whilst the other emission pro-
files are rather complex and/or faint.
This object is located in a small double association of stars
(#200 & 201 in Bica et al. 1999), with a cluster at ∼ 2′.
This star is in a cluster containing a carbon star and has
an initial mass of Minitial∼ 4 M⊙(van Loon et al. 2000).
The brightest AGB star in the LMC at 25 µm, it was
first identified with the optically visible semi-regular vari-
able TRM060 (Reid et al. 1990) before the IRAS source
was identified in the near-IR at ∼ 26′′S of TRM060 by
Wood et al. (1992). The region around IRAS05329−6708
is very crowded, with a PN (TRM058 = LI-LMC1280)
at 2.2′SW, and an M4 Iab RSG (HV5933 = TRM063 =
IRAS05334−6706) at 4.2′NE. The latter has IRAS 12 and
25 µm flux densities of 0.30 and 0.21 Jy, respectively.
H2O maser emission was tentatively detected (Fig. 6,
top panel) at 22 GHz at Parkes (1997 run), centred at
vhel = 347 km s−1. The peak has a FWHM of ∼ 5 km
s−1. The integrated flux of the emission is 0.11 Jy km
s−1, corresponding to a photon flux of 1.7 × 1044s−1. It
is not certain whether the 22 GHz peak corresponds to
the stellar velocity, or whether the OH and H2O masers
Fig.6. IRAS05329−6708: Spectra of the tentative detec-
tions of 22 GHz H2O (top; Parkes 1997) and 86 GHz SiO
(middle; SEST) maser emission. The boldfaced curves are
the spectra smoothed by a gaußian of σ = 2 and 0.5 km
s−1, respectively. The velocities are heliocentric. Also in-
dicated are the velocities of the strongest 1612 MHz OH
maser peaks (arrows),and heliocentric velocity 347 km s−1
(vertical dotted line). The 1612 MHz spectrum of Wood
et al. (1992) is plotted in the lower panel for comparison.
8van Loon et al.: Circumstellar masers in the Magellanic Clouds
refer to the same source. The 86 GHz SiO maser spectrum
(Fig. 6, middle panel) is noisy, but there is positive signal
(integrated flux ∼ 0.11 Jy km s−1) within 1 km s−1of
the H2O maser peak, and some spikes within the velocity
range of the OH emission.
This star is ∼ 3′W of an H ii region (DEM L275 & 277).
3.4. AGB stars in the SMC
No H2O maser emission could be positively detected. To
date, no circumstellar maser emission has been detected
from any source in the SMC.
This star is a late-M type LPV of Mbol∼ −7 mag (Wood
et al. 1992; Groenewegen & Blommaert 1998). Castilho
et al. (1998) measure some Li-enhancement. Their data
suggest that the stellar photosphere is only mildly metal-
poor — quite surprising for a star in the SMC. An H i
shell (#157 in Staveley-Smith et al. 1997) is located at
∼ 1.5′and vhel∼ 156 km s−1.
A 3-σ peak (∼ 150 mJy) at vhel∼ 153 km s−1in the 22
GHz spectrum with Mopra could not be confirmed with
Parkes (Fig. 7). It would coincide with the maximum opti-
cal depth of the self-absorbed 21 cm H i (ATCA+Parkes)
at vhel∼ 157 km s−1(Stanimirovi´ c et al. 1999).
Fig.7. IRAS00483−7347: Mopra and Parkes (2000) 22
GHz spectra. The velocities are heliocentric. The bold-
faced curves are the spectra smoothed by a gaußian of
σ = 2.0 km s−1. A possible H2O maser peak at vhel∼ 153
km s−1in the Mopra spectrum could not be confirmed at
Parkes despite the much lower noise levels.
This source is in the IRAS-FSC rather than the IRAS-
PSC. The mid-IR flux densities were determined by
Groenewegen & Blommaert (1998). It coincides with the
extended source IRAS0048−731, which may be identified
with the H ii region LHA 115-N 36 and an H i shell (#156
in Staveley-Smith et al. 1997). The most plausible coun-
terpart of the IRAS point source is GM103, a luminous
late-M type AGB star (Groenewegen & Blommaert 1998)
from which 10 µm silicate dust emission was detected by
Groenewegen et al. (1995). An emission-line star (#368 in
Meyssonnier & Azzopardi 1993) and a carbon star (#323
in Rebeirot et al. 1993) are ∼ 0.5′away.
This is a luminous M5e-type Li-enhanced variable AGB
star (Wood et al. 1983; Whitelock et al. 1989; Smith
et al. 1995; Zijlstra et al. 1996; van Loon et al. 1998a;
Groenewegen & Blommaert 1998), located ∼ 0.8′from
the centre of an H i shell (#369 in Staveley-Smith et al.
3.5. H ii regions in the LMC
No 43 GHz SiO masers were found in any of these objects,
but at rather high detection thresholds. No other radio
observations were made.
This source is associated with the H ii region DEM L2
The extremely red mid-IR colour suggests that this is a
(compact) H ii region.
This source is associated with the Wolf Rayet star
HD32014 in/and the cluster NGC1748, situated in the
complex H ii region DEM L22.
This source is located in the open cluster IC2116 that
contains other (early-type) evolved stars and that is em-
bedded in the H ii region LHA 120-N 11A (Parker et al.
1992; Rosado et al. 1996).
This luminous IR object is in a region of high stellar and
nebular density (LHA 120-N 120: Laval et al. 1992). Loup
et al. (1997) identify the IRAS point source with the WC
van Loon et al.: Circumstellar masers in the Magellanic Clouds9
star HD35517 (#559 in Bohannan & Epps 1974; see also
Laval et al. 1994). Within ∼ 1′are a multiple system
(IDS 05201−6945) containing a B0 Iab supergiant, and
an emission-line star (#560 in Bohannan & Epps 1974).
This is associated with the H ii region LHA 120-N 57A.
4. Stellar and outflow velocities from circumstellar
masers in the LMC
The presently available data on the stellar and outflow
velocities of OH/IR stars in the MCs are summarised in
Table 5. Literature values are largely based on the OH
maser emission profile alone. The listed expansion veloc-
ities in that case are derived from the separation of the
strongest peaks in the blue- and red-shifted emission com-
ponents, respectively. Some authors use instead the max-
imum extension of the emission (rather than the peaks)
as a measure for the expansion velocity (e.g. Zijlstra et
al. 1996). Groenewegen et al. (1998) compare expansion
velocities derived from the OH peaks with those derived
from the width of the thermal CO emission and find
that the former are on average smaller by a factor 1.12
(see also Lewis 1991). In some cases for which additional
SiO and/or H2O maser emission was discovered, the OH
masers appear to represent the part of the CSE in front
of the star. Hence the outflow velocities may have been
severely underestimated. Revised values are listed, us-
ing the OH peak velocities in relation to other maser
peaks. The interpretation of the maser emission from
IRAS05329−6708, and hence its stellar and expansion ve-
locities, is rather uncertain.
Table 5. Kinematics for the presently known circumstel-
lar masers in the LMC. Outflow and stellar velocities (v∞
and v⋆, in km s−1) are derived both from OH alone (half
the separation of the main peaks in the blue- and red-
shifted emission components) and from a combination of
OH and SiO and/or H2O (the separation between the
main peak of the blue-shifted OH emission and the central
peak of the SiO and/or H2O emission). Uncertain values
are within parentheses. Also given are the 21 cm H i he-
liocentric velocities from the maps in LR-1992 = Luks &
OH + SiO/H2O
4.1. Stellar velocities and local H i velocities
The velocities of the OH/IR stars may be compared with
the velocity of the local ISM. Luks & Rohlfs (1992) present
a low spatial resolution map of 21 cm H i velocities for the
LMC disk, from which approximate heliocentric velocities
are listed in Table 5. Kim et al. (1999) present a high
spatial resolution map of 21 cm H i velocities, from which
nine individual spectra centred on and around the position
of each OH/IR star are extracted and averaged (Fig. 8).
There is good agreement between the radio H i emis-
sion and the Hα emission around IRAS04553−6825 (Fig.
B2). Both agree with its SiO and H2O maser velocities,
as is to be expected for such a massive RSG. The H2O
maser emission from IRAS05216−6753 provides for the
first time a radial stellar velocity for this heavily obscured
object, consistent with the local H i velocity. The H2O
maser emission from IRAS05280−6910 peaks at the OH
1612 MHz centroid. Its stellar radial velocity deviates sig-
nificantly from the bulk of the H i emission, with which
the OH 1665 MHz emission is associated. The H2O maser
in IRAS05329−6708 suggests an exceptionally high stellar
velocity. It is seen projected on LMC4 (Meaburn 1980), a
supergiant shell aged ∼ 107yr, and peculiar velocities are
not surprising: high-velocity H2O masers are also found
in 30 Doradus (van Loon & Zijlstra 2000). Although most
supergiants in the same (projected) region have stellar ve-
locities similar to the centre velocity of the OH profile
(Pr´ evot et al. 1989), a few stars are found with radial ve-
locities ∼ 350 km s−1(Fehrenbach & Duflot 1982).
Most OH/IR stars in the LMC follow the kinematics of
the gas — in particular the AGB stars IRAS04545−7000,
IRAS05298−6957 and IRAS05402−6956. As the referee
pointed out, this is remarkable because Galactic AGB
stars show a large velocity spread. The observation that
the velocities of the OH/IR stars in the LMC do not gen-
erally deviate (much) from the motion of the gas may be
due to the fact that the OH/IR stars are relatively mas-
sive. Even the AGB stars amongst them probably have
initial masses Minitial>∼4 M⊙, otherwise they would have
become carbon stars (Wood 1998; van Loon et al. 2000),
and their ages are t<∼200 Myr. If they have formed in the
disk of the LMC, they may still trace its rotation.
4.2. Outflow kinematics in IRAS04553−6825
The H2O maser emission from IRAS04553−6825 (Fig. 2),
with its double blue-shifted emission peaks, reveals im-
portant information about the acceleration of the out-
flow. In first instance only the central main peak of H2O
maser emission was discovered (van Loon et al. 1998b).
Although most (circum-)stellar and maser properties of
IRAS04553−6825 are virtually identical to those of NML
Cyg, a Galactic RSG (Morris & Jura 1983; van Loon et al.
1998b), the H2O maser emission from NML Cyg (Richards
et al. 1996) nearly entirely arises from a bright double
peaked structure at blue-shifted velocities indicating out-
flow velocities in the H2O masing region of ∼ 15 km s−1
10van Loon et al.: Circumstellar masers in the Magellanic Clouds
Fig.8. 21 cm H i spectra, constructed by averaging nine
spectra from Kim et al. (1999) centred on and around
the position of each OH/IR star. The stellar velocities are
indicated as a vertical line (solid: SiO and/or H2O; dotted:
1612 MHz OH).
— a main peak of H2O maser emission centred at the stel-
lar velocity has only occasionally been reported for NML
Cyg. Comparison with its OH maser emission indicative
of outflow velocities of ∼ 27 km s−1led to the interpreta-
tion of the H2O masing region in NML Cyg to be located
in the accelerating part of the outflow where the matter
has not yet reached the local escape velocity (Richards et
al. 1996). The detection of similar blue-shifted H2O maser
emission from IRAS04553−6825now for the first time also
allows us to measure the acceleration of the outflow for
this metal-poor RSG in the LMC.
The main H2O maser peak in IRAS04553−6825 was
interpreted by van Loon et al. (1998b) either (i) to be
radially beamed and hence indicate very low outflow ve-
locities of ∼ 1 km s−1in a dust-free inner CSE, or (ii) to
be tangentially beamed and centred at the stellar veloc-
ity. The detection of the blue-shifted H2O maser emission
now strongly favours the latter. The blue-shifted maser
peaks suggest the emission is radially beamed and thus
measures the radial outflow velocity of the material ex-
pelled from the star, in the region where the dust forma-
tion is thought to take place. Material is accelerated from
18 km s−1in the H2O masing zˆ one to 26 km s−1in the
OH masing zˆ one. The duplicity of both the H2O and OH
blue-shifted emission suggests a second kinematic compo-
nent in the CSE, accelerating from 13 (H2O) to 14 (OH)
km s−1. This slower component may be closer to the star
as suggested by the relatively stronger mainline OH 1665
MHz maser emission from that component (Wood et al.
1992). It is remarkable that the OH 1665 MHz emission
profile is broader than that of the OH 1612 emission, a
phenomenon that is attributed to either (or a combination
of) Zeeman broadening, clumpiness, velocity fluctuations
or axi-symmetric winds (Sivagnanam & David 1999).
The kinematic data for the CSE of IRAS04553−6825
are very similar to the kinematics in the CSE of NML
Cyg, suggesting that the outflow kinematics including the
acceleration mechanism does not depend on metallicity.
Other data, however, seem to support theoretical expec-
tations of lower velocities in CSEs of lower metallicity (van
Loon 2000). One way to reconcile both views is if the dis-
tance to NML Cyg were ∼ 1.3 kpc rather than 2 kpc: the
outflow velocity scales with metallicity Z and luminosity
L as v∞ ∝
NML Cyg (solar) and Z ∼ 0.008 for IRAS04553−6825
(van Loon et al. 1998b and Appendix B).
4√L (van Loon 2000) with Z ∼ 0.02 for
The velocity range over which SiO maser emission
is seen from IRAS04553−6825 covers the entire velocity
range of the H2O maser emission. If the strong double
peak of SiO maser emission is tangentially beamed then
the velocity separation might imply a rotational velocity
component of the inner CSE. If the double SiO peak is
radially beamed, however, then it suggests moderate out-
flow velocities in the dust-free inner CSE of ∼ 4 km s−1.
In any case, the simultaneous presence of SiO emission
over a large velocity extent and strong discrete peaks of
SiO emission indicate that the velocity field of the inner
CSE is highly complex.
Results from echelle spectroscopic observations of
IRAS04553−6825 in the 0.6 to 0.9 µm region that gen-
erally support the kinematic picture for the CSE of this
star are described in Appendix B.
van Loon et al.: Circumstellar masers in the Magellanic Clouds11
4.3. Outflow kinematics in IRAS05280−6910
IRAS04553−6825, and thus carries the same poten-
tial for analysing the kinematics of its CSE: the narrow
main peak is interpreted as tangentially beamed radiation
centred at the stellar velocity, and the blue-shifted
emission (in the case of IRAS05280−6910 probably its
red-shifted counterpart is seen as well) can be explained
as radially beamed radiation indicating the radial outflow
velocity in the H2O masing region of the CSE close to
the region of dust formation. The outflow in the CSE of
IRAS05280−6910 is being accelerated from ∼ 6 km s−1
in the H2O masing region to ∼ 17 km s−1in the outer
CSE from where the OH 1612 MHz maser emission arises.
Double-peaked red-shifted SiO emission may have
been detected, with equal velocity separation but slightly
larger receding velocities than the red-shifted OH emis-
sion. This suggests that material in the dust-free inner
CSE may exhibit a wide range of velocities, possibly ex-
ceeding the final wind velocity (Cernicharo et al. 1997).
4.4. Outflow kinematics of Magellanic and Galactic
Expansion velocities from the separation of the two OH
peaks versus maximum peak flux density are plotted in
Fig. 9 for the OH/IR stars in the Galactic Centre from
Lindqvist et al. (1992a) and Sjouwerman et al. (1998),
after scaling their flux densities from the distance of the
Galactic Centre to the distance of the LMC. Also plot-
ted are the expansion velocities for the OH/IR stars in
the LMC, as derived from the combination of all detected
maser peaks. Expansion velocities tend to be below av-
erage for the brightest OH sources. For the LMC sources
the statistics are very poor, with only one well-determined
expansion velocity for an AGB star (IRAS05298−6957)
suggestive of smaller expansion velocities at lower metal-
licity (see also Wood et al. 1992; Zijlstra et al. 1996;
van Loon 2000). The new maser data presented here for
IRAS05329−6708, however, suggest that this source may
exhibit an exceptionally large expansion velocity even
when compared to the expansion velocities of Galactic
post-AGB objects (Zijlstra et al. 2000). Disregarding
IRAS04553−6825 and IRAS05329−6708, the average ex-
pansion velocity of the remaining four LMC objects is
v ∼ 12±3 km s−1, which may be compared to v ∼ 14±2
km s−1of the six brightest OH masers in the Galactic
Centre. Hence the expansion velocity of bright OH sources
in the LMC is ∼ 20% lower than that of similarly bright
OH masers in the Galactic Centre, but the data is also
consistent with no difference in expansion velocity.
In the Galactic Centre, unlike the 13 SiO masers
(Lindqvist et al. 1991) that peak within a few km s−1
of the mid-velocity of the OH peaks — i.e. centred at
the stellar velocity — the H2O detections (Lindqvist et
al. 1990) peak at or very near the OH peak velocities.
Fig.9. Expansion velocities of the circumstellar shells de-
rived from the separation of the OH maser peaks, ver-
sus the flux density of the brightest OH maser peak (at
the LMC distance). Expansion velocities derived from the
combination of all detected masers are given too for the
LMC sources (squares).
This suggests that the wind has already (nearly) reached
its terminal velocity before leaving the H2O masing zˆ one.
In three out of four cases the blue-shifted H2O peak is
(much) brighter than the red-shifted peak. H2O masers in
the LMC, however, peak at the stellar velocity indicated
by the centroid of the SiO maser emission, rather than
at (one of) the OH maser peak(s). Secondary H2O peaks
suggest that the wind must still experience substantial ac-
celeration after leaving the H2O masing zˆ one. This may
be understood by a less efficient acceleration in the H2O
masing zˆ one of low-metallicity CSEs, which results in a
stellar wind that is still being accelerated upon entering
the OH masing zˆ one. This could cause multiple, thin OH
masing shells to give rise to the rather irregular OH emis-
sion profiles as observed in the LMC.
4.5. Asymmetric emission profiles:
non-spherical outflow or radiation transfer effects?
All masers observed in the LMC are blue-asymmetric in
the sense that the masing material approaching Earth (af-
ter correctingfor the stellar velocity with respect to Earth)
appears brighter than the receding matter (Fig. 10): the
OH, H2O and SiO maser emission from IRAS04553−6825,
the OH and H2O maser emission from IRAS05280−6910,
the OH maser emission from all four AGB stars detected
by Wood et al. (1992) — if interpreting IRAS05329−6708
in combination with the tentative H2O maser detection
12van Loon et al.: Circumstellar masers in the Magellanic Clouds
Fig.10. Ratio of the flux densities of the blue- and red-
shifted OH maser peaks, versus the flux density of the
brightest OH maser peak (at the LMC distance).
— and possibly also the single-peaked OH maser emis-
sion from IRAS04407−7000.Does this reflect outflow com-
plexity (Zijlstra et al. 2000), or is it merely due to the
origin and propagation of the amplified radiation field?
Blue-asymmetric emission profiles (Norris et al. 1984;
Sivagnanam et al. 1990) arise if the (radially-beamed)
masers amplify stellar light or free-free emission from the
inner part of the CSE, rather than radiation from the
dusty CSE, and/or the receding maser spots are occulted
by the star or an optically thick free-electron reservoir.
The blue asymmetry is expected to be more pro-
nounced in RSGs such as IRAS04553−6825 than in AGB
stars such as IRAS05298−6957, because (i) at mm wave-
lengths, the ratio of stellar light to CSE radiation is larger
for RSGs than for more compact AGB stars that have
optically thicker CSEs (van Loon 2000), and (ii) RSGs
are generally warmer than AGB stars and hence the free-
electron abundance is higher around RSGs. The blue
asymmetry may be more pronounced at lower metallicity,
because (i) at mm wavelengths, the ratio of stellar light to
CSE radiation is smaller because of the lower dust-to-gas
ratio, and (ii) the free-electron abundance is higher due to
the warmer photospheres of low-metallicity stars.
5. Photon fluxes of Magellanic and Galactic
5.1. OH masers
The OH masers in the LMC have large photon fluxes com-
pared to what is typical for the Galactic Centre (see Fig.
9), where the OH luminosity function peaks at 2.5×1043
s−1(Sjouwerman et al. 1998). The OH maser emission
from IRAS04553−6825 is ∼ 100 times brighter (and so is
NML Cyg, a very similar RSG in the Milky Way). The
sixth brightest maser in the Galactic Centre is equally
bright as the sixth brightest maser in the LMC, which
suggests that a similar number of intermediate-age stars
exist in the Galactic Centre and in the LMC. Saturated
OH masers have flux densities SOH∼1
the detected OH masers in the Magellanic AGB stars meet
the saturation criterion within a factor ∼ 2, the detected
OH masers in the two Magellanic RSGs are an order of
magnitude below the saturation level.
The deepest search for OH sources in the LMC has
been performed with a detection limit of ∼ 50 mJy,
and several single-peaked sources were found, including
IRAS04407−7000 (van Loon et al. 1998a). A flux density
of 50 mJy from the LMC corresponds to 2 Jy from the
Galactic Centre. Of the 134 double and 16 single-peaked
OH sources in Lindqvist et al. (1992a), only 12 have max-
imum flux densities > 2 Jy. Out of these 12 sources, 8
would appear as a single peak. Hence, when searching for
OH masers close to the detection limit, one is likely to find
more single than double-peaked sources.
Many of the 1612 MHz peaks of van Loon et al. (1998a)
could not be identified with IR-bright point sources associ-
ated with CSEs around evolved stars. Interestingly, none
of the Galactic Centre OH sources with maximum flux
densities > 2 Jy were recovered in the near-IR by Wood
et al. (1998), who did identify a number of the fainter
Galactic Centre OH sources with near-IR objects. Only
two of these bright OH sources are identified with IRAS
point sources (Appendix C).
Amongst the LPVs with known pulsation periods and
bolometric magnitudes from Wood et al. (1998) the frac-
tion of stars with detected OH emission increases with red-
der (K−L)0: 0.22, 0.77, 0.90, 0.95 and 1.00 for colour bins
(←,1], [1,2], [2,3], [3,4] and [4,→) mag, respectively. No
OH masers are detected from LPVs with P<∼500 d. The
OH/IR stars in the LMC all have periods 900<∼P<∼1300
d, but their colours are with (K − L)0∼ 2 mag not ex-
tremely red (Trams et al. 1999) due to the low dust content
as a result of a low metallicity.
5.2. H2O masers
Lindqvist et al. (1990) detected 22 GHz H2O maser emis-
sion from 4 out of 33 OH sources in the Galactic Centre.
Their detections typically have peak flux densities of
0.5 to 1 Jy, corresponding to ∼ 20 mJy if the sources
were at the distance of the LMC. The 3-σ upper lim-
its for their non-detections correspond to ∼ 10 mJy at
the distance of the LMC, which is comparable to the
sensitivity of our search for H2O maser emission from
Magellanic sources. Of the 13 Magellanic IRAS sources
that were observed at 22 GHz, four were detected. Three
of the IRAS-PSC identifications with OH masers in the
Galactic Centre (Table C1) were observed at 22 GHz:
van Loon et al.: Circumstellar masers in the Magellanic Clouds13
OH359.675+0.069 and OH359.946−0.047 were detected
(both identified with near-IR LPVs), but the bright OH
maser OH359.762+0.120 (for which no near-IR counter-
part has been identified) was not detected. Lewis (1998)
already remarked that 22 GHz detection rates are al-
ways < 100% (namely, ∼ 80%), and even some bright
OH sources are undetected at 22 GHz. He suggests that
this may be due to the absence of density enhancements
(“blobs”) or an unfavourable orientation of a bipolar ge-
ometry. Hence, not all Magellanic OH/IR stars should
be expected to exhibit (strong) H2O masers. Significant
mass loss is required for the presence of circumstellar H2O
masers, though: AGB stars with modest mass loss are far
below the sensitivity of both the Magellanic and Galactic
Centre 22 GHz surveys (see Appendix A).
The ratio of H2O and OH photon fluxes is ∼ 0.1 to 1,
and the ratio of H2O and OH peak flux densities is ∼ 1,
both for the Galactic Centre as well as the LMC masers.
At lower metallicities the CSEs are optically thinner and
provide less self-shielding from the interstellar UV radia-
tion field, and hence H2O may be expected to be dissoci-
ated over a larger extent of CSEs in the LMC compared to
those in the Galactic Centre (Huggins & Glassgold 1982),
thus yielding relatively faint H2O maser emission com-
pared to the OH maser emission. The fact that no dif-
ference is seen between these relative intensities in the
LMC and the Galactic Centre suggests that the effect of
stronger dissociation might be cancelled by the effect of
longer coherent paths throughout the more slowly accel-
erating wind in lower metallicity CSEs.
5.3. SiO masers
Lindqvist et al. (1991) detected 43 GHz SiO maser emis-
sion from 13 out of 31 OH sources in the Galactic Centre,
peaking at S43 ∼ 1 Jy (∼ 30 mJy at the LMC). Upper
limits for the photon rates of Galactic non-detections
at 43 GHz were a few 1044s−1. The upper limits for
LMC sources using the Parkes dish are a few dozen times
brighter and thus not very useful. None of the Galactic
Centre 43 GHz masers could be detected at 86 GHz down
to typical photon rates ∼ 1044s−1. This is similar to the
SEST limits for LMC sources and thus explains the diffi-
culty in detecting 86 GHz masers in the LMC.
Two of the Galactic Centre OH masers with IRAS-
PSC identifications (Table C1) were observed at 43
GHz, and both were detected (OH359.675+0.069 and
OH359.762+0.120). Three Galactic Centre H2O masers
were observed at 43 GHz, of which two were detected (in-
cluding OH359.675+0.069). In the LMC too, the bright-
est (circumstellar) H2O masers are also the brightest
SiO masers. However, some quite bright mid-IR objects
could not be detected in any maser transition — e.g.
SiO maser emission may be detected from IR objects
with pulsation periods P>∼400 d, and it is always present
when P>∼800 d (Fig. 11). The OH/IR stars in the LMC
Fig.11. Detection fraction for SiO J = 1 → 0 masers
versus pulsation period, for the Galactic Bulge (solid:
Izumiura et al. 1994) and the Galactic Centre (dotted:
Lindqvist et al. 1991).
all have P>∼900 d and are thus expected to exhibit SiO
maser emission, probably not much below the sensitivity
limits of our searches. With (K − L)0∼ 2 mag their 86
GHz SiO maser emission is expected to be similarly bright
as their 43 GHz masers (Nyman et al. 1993). Lindqvist et
al. (1991) find that for Miras — probably the progenitors
of OH/IR stars — the SiO masers can often be 102times
as intense as the OH masers. OH luminosities scale with 35
µm luminosities, and OH outflow velocities reach a maxi-
mum for OH/IR stars (Sivagnanam et al. 1989), indicating
that OH/IR stars experience stronger mass outflows than
Miras. The ratio of SiO and OH photon fluxes (or peak
flux densities) is about unity both for the Galactic Centre
and LMC sources, which suggests that also in the LMC
the brightest OH masers are already experiencing heavy
mass loss for some time.
Alcolea et al. (1990) found a clear correlation between
the visual amplitude of pulsation and the pumping effi-
ciency of the SiO masers. For our obscured LMC sources
no visual photometric monitoring data is available. It is
known, however, that the SiO maser emission correlates
with the IR lightcurve (Nyman & Olofsson 1986; Alcolea
et al. 1999), and hence we investigate here whether the
SiO maser intensity correlates with the IR amplitude. In
Table 6 IR photometric variability data is summarised for
relatively nearby stars in the Milky Way that were moni-
tored by Harvey et al. (1974) and Le Bertre (1993). The
latter source, if available, is preferred because it is more
modern by nearly two decades. Where only narrow-band
N1,2,3was obtained the broad-band N is approximated by
an unweighted average. SiOv=1(J = 2 → 1) peak flux den-
sities were compiled from recent literature and averaged if
more data was available. For the LMC sources similar data
is compiled, but at 10 µm too few epochs of measurements
exist to allow an estimate of the amplitude.
The flux density ratio of the peak of the SiOv=1(J =
2 → 1) and the stellar flux at 2.2 µm increases for larger
2.2 µm amplitudes (Fig. 12). This means that larger am-
plitudes cause relatively stronger maser pumping, prob-
ably because of the relatively stronger shocks travelling
through the inner dust-free part of the CSE. A similar cor-
relation is found for other near-IR amplitudes and for in-
14van Loon et al.: Circumstellar masers in the Magellanic Clouds
Table 6. IR magnitudes and their variability for LMC
masers (Elias et al. 1986; Wood et al. 1992; Zijlstra et al.
1996; van Loon et al. 1998a) and Galactic masers (Harvey
et al. 1974; Le Bertre 1993), and SiOv=1(J = 2 → 1)
peak flux densities (in Jy) for LMC masers (this work)
and Galactic masers (Nyman & Olofsson 1985; Alcolea
et al. 1990; Haikala 1990; Le Bertre & Nyman 1990;
Nyman et al. 1993; Haikala et al. 1994; Bujarrabal et al.
1996; Cernicharo et al. 1997; Gonz´ alez-Alfonso et al. 1998;
Herpin et al. 1998; Nyman et al. 1998).
Milky Way masers
tegrated fluxes of the SiO maser emission, but their lower
accuracy leads to more scatter. The one secure and two
tentative detections of SiOv=1(J = 2 → 1) maser emission
in the LMC, as well as the three useful upper limits are
all consistent with SiOv=1(J = 2 → 1) masers that are
equally strong in the LMC and the Milky Way. Fig. 12
Fig.12. Flux density ratio of SiOv=1(J = 2 → 1) peak
and K-band, versus K-band amplitude for Galactic (solid
dots) and LMC masers from Table 6.
Fig.13. Same as Fig. 12, but for the N-band.
suggests ∆K ∼ 1 to 2 mag for IRAS05280−6910, but the
DENIS K-band magnitude is with 8.195 (Cioni et al. 2000)
very similar to the photometry of Wood et al. (1992).
Interestingly, the flux density ratio of the peak of the
SiOv=1(J = 2 → 1) and the stellar flux at 10 µm (N-band)
is virtually independent of any near-IR amplitude (Fig. 13;
plotted against the amplitude at 10 µm), and shows rather
little scatter. This indicates a close connection between the
shocks that pump the SiO maser, and the dust formation.
It also suggests that both the SiO luminosity and the dust
heating have a common origin, namely the stellar flux.
Again, the LMC data (no 10 µm amplitudes available) is
roughly compatible with the same trend. Thus there is no
strong evidence for a difference in maser strength at LMC
metallicities compared to ∼solar metallicity.
A five year effort to search for SiO and H2O maser emis-
sion from circumstellar envelopes in the LMC resulted in:
— The secure detection of 86 GHz SiOv=1(J = 2 → 1)
and 22 GHz H2O 616→ 523maser emission from the lu-
minous red supergiant IRAS04553−6825.The average SiO
spectrum of 65 hours worth of monitoring data shows un-
precedented kinematic detail, probably indicating outflow
and/or turbulent motions with vSiO ∼ a few km s−1up
to ∼ 18 km s−1with respect to the stellar velocity. A
new, improved signal-to-noise H2O spectrum also shows
emission components at vH2O<∼18 km s−1, that likely
correspond to matter that is being further accelerated be-
fore exhibiting OH masers at vOH∼ 26 km s−1.
van Loon et al.: Circumstellar masers in the Magellanic Clouds 15
— The first detection of 22 GHz H2O maser emission from
the cluster supergiant IRAS05280−6910. It consists of a
central emission peak and additional emission components
indicating outflows with vH2O∼ 6 km s−1, that are accel-
erated further before reaching the OH masing regions that
expand with vOH∼ 17 km s−1.
— The tentative detection of 22 GHz H2O maser emission
from IRAS05216−6753, probably a massive star inside an
H ii region. No other masers have been detected.
— The tentative detection of 22 GHz H2O maser emis-
sion from the AGB star IRAS05329−6708 is puzzling as
it suggests an extremely fast outflow of v ∼ 44 km s−1.
— Masers in the LMC are generally blue-asymmetric
and/or single-peaked. We propose that this may be due
to the amplification of stellar and/or free-free radiation,
rather than dust emission. This may be more pronounced
in low metallicity envelopes due to the low dust content.
— There is weak evidence for the expansion velocities in
LMC objects to be lower than of similar Galactic objects.
The data is also consistent with no difference in expansion
velocities, due to the limited sample of LMC objects with
reliable estimates of the expansion velocity. The acceler-
ation through the CSE also seems to be slower in LMC
objects, with the outflow velocity increasing by a factor of
two between the H2O and OH masing zˆ ones.
— SiO, H2O and OH maser emission from circumstellar
envelopes in the LMC is found to be equally strong as from
similar envelopes in the Milky Way. A larger IR amplitude
of variability leads to an increase in the flux density ratio
of the SiOv=1(J = 2 → 1) peak and the 2.2 µm contin-
uum, but the flux density ratio of the SiOv=1(J = 2 → 1)
peak and the 10 µm spectral region is virtually a constant.
This suggests a close connection between the shocked SiO
masing region and the dusty outflow, which seems to be
similar in the Milky Way and in the LMC.
Present-day facilities for observing SiO, H2O and OH
masers offer an angular resolution and sensitivity capa-
ble of detecting only the very brightest masers in the
Magellanic Clouds. In the future, ALMA may provide
considerably larger samples of Magellanic SiO and H2O
masers, but for 1612 MHz OH masers no major improve-
ment is envisaged. Detection of H2O masers in OH/IR
stars in the Magellanic Clouds would, in principle, allow to
derive the total (gas+dust) mass-loss rates. Comparison
of H2O and OH deduced expansion velocities yields the
acceleration of the radiation-driven wind, if the location
of the H2O and OH masers is known with sufficient accu-
racy. As the distances and hence the luminosities for these
stars are known, the gas-to-dust ratios and total mass-loss
rates are derived simultaneously (Netzer & Elitzur 1993).
Acknowledgements. We would like to thank the staff at the
SEST, Parkes and Mopra observatories for their kind and help-
ful support, and in particular Drs. Peter te Lintel Hekkert,
Marcus Price and Ian Stewart for help with the 22 GHz
observations at Parkes in August 1997. We also thank Dr.
Roland Gredel for help with the NTT observations at La Silla
in January 1996. We acknowledge the granting of Director’s
Discretionary Time for obtaining the NTT data. We made
Fig.A.1. R Doradus: Circumstellar 22 GHz (Parkes 2000)
H2O maser emission. The velocities are heliocentric.
use of the SIMBAD database, operated at CDS, Strasbourg,
France. We thank the referee Dr. Anders Winnberg for critical
comments that helped improve the presentation. Jacco thanks
Joana Oliveira for help in reading funny formats of data, im-
proving this manuscript, and much more.
Appendix A: H2O maser emission from the
Galactic AGB star R Doradus
R Doradus (IRAS04361−6210: F12 = 5157, F25 = 1594
Jy) is a famous 6thmagnitude (V-band) SRb variable
AGB star with a period of 338 days and a spectral type
M8 IIIe. It is at a distance of only 61 pc, making it the
biggest star on the night sky (Bedding et al. 1997 and
The 22 GHz spectrum with Parkes is presented in Fig.
A1, with an on-source integration time of 228 seconds re-
sulting in an rms noise of 74 mJy. The (heliocentric) stel-
lar restframe velocity is at v⋆ = +25 km s−1. The blue-
asymmetric CO(1 → 0) line profile indicates an expansion
velocity of vexp= 6.2 km s−1(Lindqvist et al. 1992b) and
a mass-loss rate of˙ M ∼ 10−7M⊙yr−1(Loup et al. 1993).
The H2O maser emission appears red-asymmetric, but as
the peak very nearly coincides with the stellar velocity
there is actually more emission at the blue-shifted rather
than at the red-shifted side. Possibly the red-shifted part
of the emission is being occulted by the star. The blue-
shifted emission extends to (blue-shifted) velocities of al-
most ∼ 6 km s−1with respect to the stellar restframe.
The H2O masers thus seem to trace material that includes
matter that has already nearly reached the final outflow
R Dor is a good example of an AGB star with mod-
erate mass loss. If placed at the distance of the LMC, it