Astron. Astrophys. 337, 141-144 (1998)

3. Discussion

3.1. Circumstellar H₂O masers in the LMC

The detection of IRAS04553-6825 shows that H₂O masers in the LMC with photon fluxes of  s^-1 are detected at a 5  level within 6 hours of on-source integration, with the current receiver at Parkes. Typical observed H₂O maser photon fluxes from galactic sources are between  s^-1 (Lindqvist et al. 1990) and  s^-1 (Nyman et al. 1986) for OH/IR stars,  s^-1 for irregular and semiregular red variables with mass-loss rates yr^-1 (Szymczak & Engels 1997),  s^-1 for Mira variables (Benson & Little-Marenin 1996; Yates et al. 1995), and  s^-1 for supergiants (Yates et al. 1995). This implies that many of the mass-losing RSGs and AGB stars (i.e. IRAS point sources) in the LMC are expected to have detectable H₂O maser emission.

3.2. Introduction to IRAS04553-6825

With a bolometric luminosity mag and a spectral type M7, IRAS04553-6825 was recognised to be the most luminous (very) red supergiant in the LMC by Elias et al. (1986). The progenitor mass was estimated by Zijlstra et al. (1996). The current mass-loss rate of few yr^-1 was estimated from the self-absorbed 10 µm silicate dust feature by Roche et al. (1993), who argue that the lower than expected extinction in the optical indicates a disky CSE.

IRAS04553-6825 exhibits strong OH 1612 MHz and weaker 1665 MHz mainline emission (Wood et al. 1986, 1992). Comparison of the SiO maser peak velocity and the OH maser emission profile revealed that OH is only observed from the blue-shifted part of the shell. The outflow velocity is 27 km s^-1, similar to Milky Way RSGs. (Wood et al. derived 10 km s^-1 as is suggested by the OH maser emission profile alone.) It is possible that the red-shifted emission is much weaker than the blue-shifted and therefore below the detection limit. Stronger blue-shifted emission is expected if the maser amplifies emission from within the shell. Alternatively, an asymmetric and structured OH profile can also arise from bipolar outflow (Chapman 1988; te Lintel Hekkert et al. 1988).

3.3. Similarities between IRAS04553-6825 and NML Cyg

IRAS04553-6825 is remarkably similar to NML Cyg, a well known RSG in the Milky Way (Johnson 1967; Diamond et al. 1984; Richards et al. 1996; Monnier et al. 1997), in terms of luminosity, progenitor mass, spectral type, pulsation period, mass-loss rate, 10 µm feature profile, outflow velocity, CSE geometry, OH maser emission profile and peak flux density (scaled to a common distance of 1 kpc), and SiO maser photon flux. The properties of the two stars are compared in Table 1. We compiled the infared photometry from the data provided in Gezari et al. (1993), and use W mm Jy. The H₂O properties (NML Cyg: Yates et al. 1995) are also included in Table 1.

Table 1. Comparison between the properties of the red supergiants NML Cyg in the Milky Way and IRAS04553-6825 in the LMC.

The pump efficiency of SiO masers is known to be related to the pulsation amplitude (Alcolea et al. 1990): Mira variables and semi-regular variables with visual amplitudes exceeding 2.5 mag always reach the maximum efficiency, while variables with smaller amplitudes are usually less efficient. We recalculated the pump efficiency of the SiO masers . The pumping in IRAS04553-6825 is twice as efficient as in NML Cyg, but a factor less than in Mira variables. The SiO maser emission of IRAS04553-6825 has a larger ratio than NML Cyg (at a common resolution of 0.7 km s^-1), and a few times as large as Mira variables (Alcolea et al. 1990; van Loon et al. 1996).

The near-infrared (JHKL) amplitudes are 0.4 mag for NML Cyg (Gezari et al. 1993) and 0.3 mag for IRAS04553-6825 (Wood et al. 1992). For NML Cyg, Kholopov et al. (1985) list a magnitude range of 11.19 to 12.54 mag in R, and 17.0 to 18.0 mag in V. We have sparse optical photometry of IRAS04553-6825 in the period 1994-1997, suggesting an amplitude in V of mag (smaller in R). We conclude that the pulsation properties of NML Cyg and IRAS04553-6825 are similar, with possibly IRAS04553-6825 closer resembling Mira variables.

From the similar properties of both objects we predicted to detect 22 GHz H₂O maser emission from IRAS04553-6825 at a level of mJy (van Loon 1998b). The measured peak flux density is in excellent agreement with this prediction.

3.4. Location of the H₂O masers

A striking difference between the H₂O masers of IRAS04553-6825 and NML Cyg is that in IRAS04553-6825 the emission peaks at the stellar restframe velocity, whereas in NML Cyg the line profile resembles the OH maser line profile.

Interferometric observations of the masers around NML Cyg indicate that its H₂O masers are amplified radially and that they are located in the dusty part of the CSE, where radiation pressure on the grains accelerates the outflowing matter (Richards et al. 1996). Indeed, in NML Cyg the strongest emission occurs at a blueshift of km s^-1.

The peak of H₂O maser emission from IRAS04553-6825 falls within km s^-1 of the stellar velocity (as inferred from the SiO velocity), and all emission is detected within a range of 8 km s^-1. This is similar to the situation in Mira variables, indicating low outflow velocities in the H₂O masing region of the CSE (Yates et al. 1995). In contrast, NML Cyg is similar to the OH/IR stars.

Cooke & Elitzur (1985) derive a scaling relation for estimating the inner radius of the H₂O masing region, below which the density is higher than  cm^-3 and the maser is quenched. This was confirmed by observations (Yates & Cohen 1994). The collision rate is dominated by collisions with H₂ and does therefore not depend on metallicity: we therefore expect that the same relation will hold in the LMC. With an outflow velocity at of 0.8 km s^-1 (half the FWHM of the H₂O maser peak), and an average particle mass of  g, we estimate  cm, or times the stellar radius. This is outside the region where the SiO maser occurs (Humphreys et al. 1996).

We speculate that in IRAS04553-6825 the H₂O maser originates near the dust formation radius, where the logarithmic velocity gradient is very high () and tangential amplification dominates over radial amplification. Part of the emission may also originate from inside the dust formation radius where outflow velocities are low. Both cases could explain the narrow H₂O peak coinciding with the tangentially amplified SiO maser peak (see also Engels et al. 1997). The H₂O masers in NML Cyg are located outside of the dust formation region, where the outflow velocities are already km s^-1 (Richards et al. 1996).

In this model, either the maser occurs closer to the star in IRAS04553-6825, or dust formation takes place further out (as in galactic Miras as compared to galactic OH/IR stars). The first possibility appears less likely if the inner radius is set by collisional quenching of the population inversion. It would be interesting to test whether at lower metallicity the dust formation and the acceleration of the outflow occurs at larger radii because of the relative lack of refractory elements: in this case the difference between NML Cyg and IRAS04553-6825 could be a generic difference between similar stars in the Milky Way and the LMC.

© European Southern Observatory (ESO) 1998

Online publication: August 6, 1998
helpdesk.link@springer.de