EDP Sciences
Free access
Issue
A&A
Volume 531, July 2011
Article Number L4
Number of page(s) 4
Section Letters
DOI http://dx.doi.org/10.1051/0004-6361/201117203
Published online 10 June 2011

© ESO, 2011

1. Introduction

Centuries ago o Ceti confused the astronomers all over the world with its appearance and disappearance on the night sky and was therefore named Mira, the wonderful, by Hevelius (1662).

Today we know that Mira is a wind-accreting binary star system (Reimers & Cassatella 1985) at a distance of 91.7 pc (corresponding to a parallax of 10.9 ± 1.2 mas; van Leeuwen 2007) consisting of the primary star Mira A (o Ceti) and its companion Mira B (VZ Ceti). Mira A is an M-type oxygen-rich star with a pulsation period of 331 d and a brightness amplitude in the visible of 8 mag. Moreover, Mira A is the prototype of stars in their late stage of evolution residing on the asymptotic giant branch (AGB). The nature of Mira B has been unknown for long; with an effective temperature of about 10000 K, it might either be a main-sequence star or a white dwarf (Karovska et al. 1997), but recent observations show strong evidence in favour of a white dwarf (Sokoloski & Bildsten 2010). From a Hubble Space Telescope observation, Karovska et al. (1997) found the separation of Mira AB to be  ≈ 0.6″, which corresponds to 55 AU. A very uncertain orbit, based on astrometric measurements since 1923, which has a period of  ≈ 500 yr, has been published by Prieur et al. (2002).

Stars on the AGB loose a significant fraction of their mass in the form of a stellar wind. For Mira A, a moderate mass loss rate of 10-7 M yr-1 was determined by Mauron & Caux (1992), which corresponds to a dust mass-loss rate of about 5.5 × 10-10 M yr-1 (Meixner 2011, priv. comm.). From mm-CO lines, Josselin et al. (2000) found a wind expansion velocity of 4 km s-1 and a molecular envelope around the binary system with a size of about 20″. This envelope is disrupted by a slow bipolar outflow of about 8 km s-1 (Josselin et al. 2000), but a high-velocity component at 160 km s-1 has been observed by Meaburn et al. (2009). In the outer environment of Mira, a spectacular observation with the GALEX ultraviolet satellite (Martin et al. 2007) revealed a cometary-like structure with an extension of more than 2° (≈ 4 pc) from head to tail. Martin et al. (2007) attribute this tail to Mira’s exceptionally high space velocity of  ≈ 110 km s-1 and the resulting strong interaction of the stellar wind with the interstellar medium (ISM). This structure was also observed in the 21 cm H I line (Matthews et al. 2008) and is well reproduced by the hydrodynamic simulations of Wareing et al. (2007), Raga & Cantó (2008) and Esquivel et al. (2010). In the vicinity of the binary system, knot-shaped enhancements of the UV and Hα emission were found at a distance of 1′ to 2′ from the binary (Martin et al. 2007; Meaburn et al. 2009), corresponding to 5500 to 11000 AU, and in a direction consistent with that of the bipolar outflow observed by Josselin et al. (2000). In the far IR Ueta (2008) resolved with the Spitzer Space Telescope the contours of Mira’s astropause including the circumbinary envelope, termination shock, and astrosheath.

The present study offers a detailed view of Mira’s circumstellar shell(s) thanks to the high spatial resolution of the PACS instrument (which has a point-spread function of full-width at half-maximum at 70 μm) onboard the Herschel satellite. The images presented here reveal clumpy broken arcs (see Sect. 3.1) that result from the complex interactions of the wind with the ISM, the companion, and the fast outflow.

Table 1

Kinematic data for Mira.

2. Observations and data reduction

The observations of Mira took place on February 9, 2010, using the PhotoArray Camera and Spectrometer (PACS; Poglitsch et al. 2010) onboard Herschel as part of the MESS (Mass-loss of Evolved StarS) guaranteed time key program (Groenewegen et al. 2011).

The PACS camera observed at 70 μm and 160 μm simultaneously, covering a field of 30′ × 30′ on the sky with a scan-map mode. For the basic data processing to produce the images displayed in Fig. 1, the Herschel Interactive Processing Environment (HIPE) was used (Ardila & Science Ground Segment Consortium 2010) together with Scanamorphos, an IDL software for building scan maps from PACS (and SPIRE) observations (Roussel 2011). Finally, we deconvolved the images as described by Ottensamer et al. (2011). Although the PACS instrument offers a resolution of 3.2″ per pixel in the 70 μm (blue) band and 6.4″ per pixel at 160 μm (red band), the images shown in Fig. 1 are oversampled by a factor 3.2, resulting in a sampling of 1″ and 2″ per pixel (at 70 and 160 μm, respectively), to be compared with the FWHM of and at those wavelengths. We estimate the average background emission to be 1.5 × 10-4 Jy pixel-1 in the blue and 3 × 10-4 Jy pixel-1 in the red band, yielding the 1σ sensitivity.

thumbnail Fig. 1

Panel a) deconvolved PACS image at 70 μm. The arrow indicates the space motion and the position in 500 yrs; b) is the same for the 160 μm band and c) is the 70 μm deconvolved PACS image with contours and arcs labelled as A, B, C, D and E. The lowest contour is the 3σ threshold (and is depicted in white), the arrow labelled with Vspace shows the direction of the space motion, the bar labelled PA orbit gives the orientation of the major axis of the apparent orbit, orthogonal to the node line. Panel d) results from the “toy model” described in Sect. 3.3 that is based on the hydrodynamical simulations of Mohamed & Podsiadlowski (2007, 2011).

Open with DEXTER

thumbnail Fig. 2

Radial intensity profiles of Mira in the 70 μm band as a function of the distance to the star. Continuous (blue) line: integrated intensity across the arcs B and A (75° ≤ PA ≤ 180°). Dashed (red) line: across the arcs D and C (190° ≤ PA ≤ 270°). The top scale gives the distance from the star computed from Mira’s distance of 91.7 pc.

Open with DEXTER

3. Results and discussion

3.1. IR images

Mira’s environment is remarkably structured and exhibits arc-like structures in the surrounding of the binary system, labelled A to E in Fig. 1c. The best-defined and brightest arc, C, stretches over position angles (PAs)  ~ 190°−270° (south-west to west), is well detached from the star and covers a range of angular separation from 67″ (6100 AU) at the closest point to 85″ (7800 AU) at the most distant point from the star. The intensity was integrated over this PA range and the resulting radial intensity profile (Fig. 2) clearly shows the arc as a plateau in the diagram. The extension of this plateau confirms that the arc is non-circular or, if circular, is not centred on the position of the star. We found another arc, D, in the same direction as arc C but much closer to the star and stretching over much smaller PAs. In the radial intensity profile, arc D is visible as a bump located at a distance of 21−28″ (1900−2600 AU). East of Mira more arcs are visible. In the south-east direction, we see arc B close to the star. It stretches over PAs 75° to 180° and is more circular than arc C and therefore appears as a narrow peak in the radial intensity profile at a distance of 41″ (3800 AU) from the binary (Fig. 2).

Farther out, directly to the east, we find another structure, arc A, stretching towards north-east. In the integrated intensity profile, it can be recognized as a small plateau representing its fairly faint nature at a distance of 68−89″ (6200−8200 AU). This matches the distance of arc C, which leads to the suggestion that both arcs are somehow related to each other.

It is also possible that another arc, E, is present in the south-east direction, very close to the binary at a distance of less than 15″ (1400 AU), of which we only see the edge at a PA of about 65°. It is noteworthy that the arc sequence E-D-B-C-A is a sequence of both increasing distance from the star and increasing PA range, arc E being the shortest and closest to the star, and A the longest and farthest.

The overall 3σ emission above the 70 μm IR background of Mira, visible in Fig. 1c as the outermost contour, spans as far as 157″ towards the south-east. It is remarkable that the extension of the emission is towards a position angle of  ≈ 160° because one would expect the emission to be aligned with Mira’s space motion, but that motion is in the direction of PA = 185°. In Fig. 1c we also indicated the orientation of the major axis of the apparent orbit, which is oriented along a PA of  ≈ 140° (Prieur et al. 2002). As discussed in more detail in Sect. 3.3, the orbital motion and the mass transfer from the binary can also shape its environment. Nevertheless, the extension of the IR emission lies between both values and makes a distinct attribution difficult.

3.2. Comparison with GALEX UV observations

Before drawing conclusions from the structures observed in the IR, it is interesting to confront these IR observations with GALEX ultraviolet observations of Mira in the far and near UV (152.8 nm and 227.1 nm, respectively; Martin et al. 2007). In the far UV image, Martin et al. (2007) found emission knots arranged north-west and south (Fig. 3). The axis of the streams is consistent with the direction of the bipolar outflow detected by Josselin et al. (2000), and moreover, the UV knots are exactly coincident with similar knots seen in Hα (Meaburn et al. 2009). In contrast, Fig. 3 reveals that in the PACS 70 μm image the southern stream exactly corresponds to a region with an IR flux below the 3σ threshold1. It thus looks as if the southern stream is penetrating and disrupting the material responsible for the IR emission. According to Meaburn et al. (2009), the southern stream is indeed inclined at an angle of 69° with respect to the plane of sky towards us. Put together, all these arguments clearly indicate that the IR emission must come from a 3D structure and is not confined to the orbital plane; otherwise, there could be no explanation for the anti-coincidence between the GALEX far UV/Hα emission and the IR emission.

Another interesting clue is that the northern lobe of the bipolar flow remains visible even though it is located behind the IR shell. The dust present in the shell will efficiently absorb the far UV radiation, so that the intensity of the northern lobe measured by GALEX should be smaller than that of the southern lobe, which has no intervening dust. In the far UV, the peak intensity of the northern lobe is indeed about 1.5 times fainter than the peak intensity of the southern lobe. Assuming that the northern and southern jets carry the same energy, this ratio may then be seen as resulting from dust extinction: (1)where N is the dust column density, a is the grain radius, and Qext,   UV is the ratio between the extinction cross section and the geometrical section. This constraint can be inserted in Kirchhoff’s law for the IR emission jIR, leading to (2)where BIR is the Planck intensity function at the dust temperature and Qabs,   IR the ratio of the absorption cross section and the geometrical section. From Draine (1985), Qabs,   70   μm/Qext,   0.15   μm ~ 10-3 (almost independent of the grain size), which leads to (3)From the measured intensity ratio at 70 and 160 μm for the northern lobe, we find a dust temperature of 36 K; combining B70   μm(36   K) with the measured j70   μm = 4.42 × 107 Jy/sr, we obtain j70   μm/B70   μm(36   K) = 1.2 × 10-4, iwhich agrees well with the value predicted by Eq. (3).

thumbnail Fig. 3

GALEX far UV image of Mira obtained by Martin et al. (2007) overplotted with contours of the Herschel 70 μm image, the thick blue contour corresponding to an IR flux just 3σ above background. The bright white regions correspond to strong UV emission at both ends of the fast jet. Note the absence of IR emission at the end of the southern jet. The red arrow indicates the position of the star in 500 yrs. The bow shock is the faint structure in the background in the south-west.

Open with DEXTER

3.3. Role of the binary system

The arcs observed around Mira (Sect. 3.1) are (almost) unique in the MESS sample2 (Groenewegen et al. 2011; Cox et al., in prep.) which makes it tempting to relate them to the binary nature of Mira.

thumbnail Fig. 4

Various vectors entering the discussion in a 3D view, with the red and green planes corresponding to the two possible orbital orientations. Because bipolar flows are generally perpendicular to the orbital plane, the green orientation seems the most likely. The sky plane is depicted in blue. The thick arrows are projections onto the sky plane. The observer is located on the bottom plane and looking upwards.

Open with DEXTER

In binary systems involving a mass-losing component, nested half-shells are seen when the orbit is orientated close to edge-on (He 2007), and they represent the 3D structure extending above the Archimedian spiral present in the plane of a circular orbit (for eccentric orbits, the spiral becomes broken). The latter was predicted by the hydrodynamics simulations of Theuns & Jorissen (1993), Mastrodemos & Morris (1999), Mohamed & Podsiadlowski (2011), and detected around AFGL 3068 by Mauron & Huggins (2006) and Morris et al. (2006). In Mira, the observed arc structure resembles neither an Archimedian spiral, nor nested half-shells; its interpretation must therefore be more complex, involving two supplementary physical processes at play, namely the presence of a bipolar jet, and the interaction with the ISM (to fix the ideas, Fig. 4 presents the respective orientations of all relevant vectors involved in the Mira system). We already argued in Sect. 3.2 that the fast bipolar outflow (160 km s-1; Martin et al. 2007; Meaburn et al. 2009) seems to cut a hole in the IR shell (see Fig. 3). Similarly, it could easily disrupt the arcs with its high velocity, and may actually be responsible for cutting the edges of arcs B, C, and D, because all these edges are located along the southern jet, at a position angle close to 180°.

On the other hand, it is likely that the interaction with the ISM, more precisely, the termination shock, plays some role in shaping (at least) arc C, as already suggested by Ueta (2008) from lower-resolution Spitzer images, and confirmed in the Herschel images by the sharp ridge seen on the contour levels in Figs. 1c and 3. Arc C is indeed located almost upstream along Mira’s space motion. On the 160 μm image (Fig. 1b), arcs A and C extend towards the north and join with each other, exactly opposite to the space motion of Mira, in a tail-like or bullet-like structure, as seen in the hydrodynamic simulations by Wareing et al. (2007) and Esquivel et al. (2010). This is clear evidence that the outer arcs feel the space motion and interact with the ISM. Another indication thereof is provided by the curvature of the bipolar-outflow southern stream, which Martin et al. (2007) attribute to its deceleration by the post-shock flow. Therefore, the arcs seen around Mira likely result from a combination of the projected 3D structures resulting from the interaction of Mira’s wind with its companion on one hand, and with the ISM on the other hand.

As a first exploration of this combination, we used the simulations of Mohamed & Podsiadlowski (2007, 2011) of an AGB star nearly filling its Roche lobe in a system 10 AU-wide3, and losing mass through a slow wind. The 3D structure resulting from the expanding wind disturbed by the companion’s gravitational pull was then intersected with a paraboloïd representing the bow shock. All particles within a thin shell along the paraboloïd surface were kept and the resulting particle density projected onto the sky plane (the respective orientations of the paraboloïd, the orbital plane, the sky plane, and the companion were taken into account, as represented in Fig. 4). If the IR emission observed by Herschel around Mira results from the reheating of the wind by the shock, this “toy model”, presented in Fig. 1d, should then give a fair account of the situation. Fig. 1d shows that there are indeed promising resemblances between the Herschel image and the one from the “toy model”, apart from the absence of brightening of the south-west arc C (where the compression from the ISM – not included in our model – is maximum), and apart from the ISM-swept appearance of the outermost northern arc. Hydrodynamical simulations of mass transfer in the Mira system, including the ISM interaction, are thus badly needed.


1

The “hole” in the IR emission located about 1′ to the East is less significant than the one considered, with fluxes about twice smaller for the latter.

2

The prototypical carbon star IRC +10216 is surrounded by a large number of thin and faint arcs, known for quite some time in the optical (Mauron & Huggins 1999, 2000), and now identified in Herschel images as well (Decin et al., in prep.). Their appearance is, however, quite different from the few thick arcs seen around Mira, in line with the fact that there is so far no evidence that IRC +10216 belongs to a binary system (Huggins et al. 2009).

3

Although this separation is much smaller than the actual orbital separation of Mira system, this difference should not impact our conclusions, because the spiral properties are largely scale-invariant.

Acknowledgments

This work was supported in part by the Belgian Federal Science Policy Office via the PRODEX Programme of ESA (Nos. C90371 and C90372). A.M. and F.K. acknowledge funding by the Austrian Science Fund FWF under project number P23586-N16, R.O. under project number I163-N16. We thank A. Zijlstra and D. Pourbaix for discussions about Mira’s orbit. S.v.E. is FNRS Research Associate.

References

All Tables

Table 1

Kinematic data for Mira.

All Figures

thumbnail Fig. 1

Panel a) deconvolved PACS image at 70 μm. The arrow indicates the space motion and the position in 500 yrs; b) is the same for the 160 μm band and c) is the 70 μm deconvolved PACS image with contours and arcs labelled as A, B, C, D and E. The lowest contour is the 3σ threshold (and is depicted in white), the arrow labelled with Vspace shows the direction of the space motion, the bar labelled PA orbit gives the orientation of the major axis of the apparent orbit, orthogonal to the node line. Panel d) results from the “toy model” described in Sect. 3.3 that is based on the hydrodynamical simulations of Mohamed & Podsiadlowski (2007, 2011).

Open with DEXTER
In the text
thumbnail Fig. 2

Radial intensity profiles of Mira in the 70 μm band as a function of the distance to the star. Continuous (blue) line: integrated intensity across the arcs B and A (75° ≤ PA ≤ 180°). Dashed (red) line: across the arcs D and C (190° ≤ PA ≤ 270°). The top scale gives the distance from the star computed from Mira’s distance of 91.7 pc.

Open with DEXTER
In the text
thumbnail Fig. 3

GALEX far UV image of Mira obtained by Martin et al. (2007) overplotted with contours of the Herschel 70 μm image, the thick blue contour corresponding to an IR flux just 3σ above background. The bright white regions correspond to strong UV emission at both ends of the fast jet. Note the absence of IR emission at the end of the southern jet. The red arrow indicates the position of the star in 500 yrs. The bow shock is the faint structure in the background in the south-west.

Open with DEXTER
In the text
thumbnail Fig. 4

Various vectors entering the discussion in a 3D view, with the red and green planes corresponding to the two possible orbital orientations. Because bipolar flows are generally perpendicular to the orbital plane, the green orientation seems the most likely. The sky plane is depicted in blue. The thick arrows are projections onto the sky plane. The observer is located on the bottom plane and looking upwards.

Open with DEXTER
In the text