The Antares Emission Nebula and Mass Loss of Α Sco A

Astronomical Science

The Antares Emission Nebula and Mass Loss of A Sco A

Dieter Reimers, Hans-Jürgen Hagen, Antares sible a 2-D mapping of the nebula with Robert Baade, Kilian Braun A B long-slit spectroscopy. We used UVES (all Hamburger Sternwarte, Universität with a 0.4 slit and a resolution of 80 000 Hamburg, Germany) 2.7 and covered the nebula with 23 long-slit positions (Figure 1). Exposure times were between 50 and 100 s (the Antares neb- The Antares nebula has been known as ula is bright!); the typical seeing was 0.6 a peculiar [Fe II] emission nebula, ap- which means that the angular resolution parently without normal H II region lines. corresponds roughly to the slit steps as Long-slit VLT/UVES mapping shows shown in Figure 1. that it is an H II region 3 in size around the B type star A Sco B, with a Balmer Quite unexpected was that all spectra – line recombination spectrum and [N II] even 10 from the M supergiant – are lines, but no [O II] and [O III]. The reason completely corrupted by scattered light for the non-visibility of [O II] is the low from the M supergiant. This was a sur- N electron temperature of 4 900 K, while prise, also to the UVES team at ESO. ESO [NII] is seen because the nitrogen abun- finally made a test placing the UVES slit dance is enhanced by a factor of three 10 from a bright star – with the same W due to the CNO cycle. We derive a result as in our spectra. Apparently, light –6 mass-loss rate of 1.05 ± 0.3 × 10 M/yr from the bright star is scattered by the slit for the M supergiant A Sco A. The [Fe II] Figure 1. Geometry of the Antares system together jaws back to the slit viewing camera and lines seem to come mainly from the with the UVES spectrograph slit sizes and positions from there again into the spectrograph shown to scale. edges of the H II region. slit. This should be a warning for anybody using UVES on faint targets close to bright UVES/VLT spectra, the Antares nebula sources, e.g. a disc, around a bright star. An iron rich nebula? puzzle could be solved. Did this mean that our spectra were use- The story of the Antares nebula began in The most extensive previous study was less? It meant that, due to the ubiquitous 1937 with the finding of O. C. Wilson and that by Swings & Preston (1978) based on scattered light, no general background R. F. Sanford at the Mount Wilson Ob- high-resolution long-slit photographic reduction was possible and no absolute servatory that the spectrum of the B-type spectra taken with the Mt. Wilson 100 line fluxes could be deduced. Data reduc- companion of Antares, itself an M super- and Palomar 200 telescopes. They found tion required an enormous load of extra giant, showed forbidden emission lines that the ‘[Fe II] rich nebula’ is strong roughly work. In brief, the scaled M star spectrum of Fe II. Later Otto Struve showed that the 3.5 around the B star and is surrounded was used as a background template. [Fe II] lines extend ~ 2 beyond the see- by a zone of weaker lines which may ex- The template was fitted to the spectrum – ing disc of the B star. Struve & Zebergs tend in a NW-SE direction up to 15. How- allowing for an offset and a scaling – (1962) describe extensively the Antares ever, they neither observed Balmer re- outside a region of ± 30 km/s of each de- nebula and conclude: “It is strange that combination lines nor any classical H II tected emission line and was finally the nebulosity around the B type com- region lines from ions like O II, O III, S II, and subtracted from the data. ponent shows only emission lines of iron N II. Therefore the nature of the Antares and silicon, but not those of hydrogen. nebula could not be understood. In a first The final results after this elaborate data The enormous abundance of hydrogen in quantitative study of the Antares system reduction – each of the roughly 90 de- all known gaseous nebulae and the con- it could be shown that the B star creates tected emission lines had to be treated ditions of excitation and ionization result- an H II region within the wind of A Sco A individually in each of the spectra – are ing from the radiation of a B type star and that the [Fe II] lines must be formed in 2-D brightness distributions (location on would render it almost inexplicable for a this H II region (Kudritzki & Reimers, 1978). the slit versus velocity) as a function of gas of normal composition to show only Later the H II region was detected as an the location relative to the M supergiant, the iron and silicon lines in emission. optically thin radio emitter with a diameter as shown for the HA line in Figure 2 and This gave rise to the hypothesis that the of ~ 3(Hjellming & Newell, 1983). for [Fe II] 4 814.55 Å in Figure 3. material in the vicinity of Antares is metal rich, in turn leading to the supposition that the envelope may be composed of UVES 2-D spectroscopy A peculiar H II region? solid substances such as meteors that have become vaporized and are excited With the construction and operation of The main puzzle of the Antares Nebula by the radiation of the hot B star. This UVES at the VLT it was obvious that pro- has been resolved by our UVES data. The hypothesis is admittedly improbable and gress in our understanding of the kine- Balmer recombination lines expected may have to be abandoned in the light matics of the Antares nebula should be for an H II region: HA, HB ... up to H10 are of future work.” We may add here that the possible with its high spectral resolution present, and their geometrical extent mystery has lasted 45 years until, with and high pointing accuracy, making pos- on the sky is virtually identical to that of

The Messenger 132 – June 2008 33 Astronomical Science Reimers D. et al., The Antares Emission Nebula and Mass Loss of A Sco A

0.9 1.4 1.9 2.4 2.9 0.9 1.4 1.9 2.4 2.9

3.4 3.9 4.4 4.9 5.4 3.4 3.9 4.4 4.9 5.4

8 8 ) ) ( ( 6

. 6 . s s o o

P 4 P 4

2 2

0 30 0 0 30 –20 0 –40 –40 –20 Velocity (km/s) Velocity (km/s)

5.9 6.4 6.9 7.4 7.9 5.9 6.4 6.9 7.4 7.9

Figure 2. HA 2-D brightness distributions (location Figure 3. [Fe II] 4814.55 Å 2-D brightness distribu- on the slit versus velocity) as a function of spectro- tions as a function of spectrograph slit position (see graph slit position relative to A Sco A. Offsets from caption of Figure 2). Notice the double structure the position of A Sco A are shown in arcsec. Notice is probably formed by the front (blueshifted) and rear that due to the M star wind expansion, the velocity (redshifted) edges of the H II region carried by the coordinate corresponds to the spatial depth (per- wind to large distances from the system. pendicular to the plane of the sky). The density in the plots has been rescaled in order to provide maxi- mum visibility. The true line ﬂuxes vary by a factor of 150 between 1.9 and 7.9. the radio emission seen with the VLA normally prominent in H II region spectra regions, are not? There are two reasons: III II (Figure 4). So why have previous observa- like [O ] 4959/5 007 Å, [O ] 3726/3729 a low electron temperature Te and an en- tions not shown the hydrogen lines? Å, [N II] 6 583/6548 Å and [S II] 6716/6731 hanced N/O ratio. At ﬁrst, the electron

The reason is probably that the strongest Å had not been detected. Our UVES temperature Te must be so low that the lines HAand HBsuffered, in the Mt. Wil- spectra allow a more critical assessment [O II] lines are not excited and below the son and Mt. Palomar photographic spec- than earlier photographic spectra. The an- detection limit. We have modelled the tra, from seeing-dominated contamination swer is that we do see faint [N II] 6 583 Å H II region created by A Sco B in the wind by the M supergiant and the signal-to- and 6 548 Å lines, and that none of the of the M supergiant using the Cloudy noise of the photographic spectra was other lines is present. [N II] 6 583 Å follows code (version 07.02.00, last described by never sufficient to enable an efficient closely the HA distribution on the sky Ferland et al., 1998). Since the B star, with background subtraction. shown in Figure 2. Since it is close to HA, an effective temperature of 18 200 K, is we can measure the HA/[NII] 6 583.4 Å relatively cool, the resulting electron tem- The strongest lines are of [Fe II] such as ratio as a function of the location in the perature is below 5 000 K which does not 4287.4 Å, 4 359.34 Å and 4814.55 Å. nebula. Close to the B star – in the mid- allow [O III] to exist and predicts [O II]/HA Altogether we have identified some 90 dle of the H II region – HA/[NII] ≈ 3.5 while below our detection limit. But the same emission lines, besides [Fe II] a few [Ni II] 3 west of B, at the outer edge of the H II model would also predict [N II] below our and several strong allowed Fe II lines like region, it is ~ 10. detection limit! The solution to this prob- 3177.53 Å and 4 923.9/5 018.4/5169.0 Å lem is that Antares must have CNO-proc- (see below). Beside the Si II lines 3 856 For [O II] we can give an upper limit: we essed material in its atmosphere and and 3862 Å we have detected further S III have detected the Balmer line series up wind. Indeed, it has been shown by Harris lines around 5050 Å and at 6 347/6371 Å to H10, but H11 (close to [O II] 3726/3729) & Lambert (1984) that in the atmosphere (for details and a line list cf. Reimers et was not visible. Thus, the H10 intensity of Antares 13 C and 17O are enhanced, al., 2008). as predicted by recombination theory, a clear indication of CNO cycle material. can be regarded as an upper limit to [O II] Unfortunately there is no direct measure- In all previous investigations of the which yields [O II]/HA < 0.017 for 3726 Å. ment of the N abundance (or N/O). Ty - Antares nebula the question remained So why is [N II] present, while the lines pical values for the N enhancement in open why the ‘classical’ emission lines of [O II] and [O III], usually stronger in H II M supergiants of comparable luminos -

34 The Messenger 132 – June 2008 Figure 4. Comparison of the distribu- shows that at present the B star moves –26˚19 19 tion of HA emission with radio free-free nearly perpendicularly into the plane emission (Hjellming & Newell, 1983) of the sky (Reimers et al., 2008) which, 20 (blue). Dashed red lines represent half, fourth, and tenth of the central HA combined with wind expansion, pro- ) emission ﬂux. 0 duces a cometary structure (cf. Fig- . 21 0

5 ure 5). 9 1 ( 22

n – The UVES observations with 45˚ slit o i t a 23 position (Figure 1) conﬁrm what Swings n i l & Preston (1978) had seen, namely c

e 24 that faint [Fe II] emission is seen all along D the slit, i.e. in the neutral M star wind. 25 The explanation comes from the observation that one of the allowed lines, 26 Fe II 5169 Å, appears both in the neutral 16 26m20.2 20.0 19.8 wind (position 0.9 from the M star) Right Ascension (1950.0) and in the H II region. Apparently one channel for the excitation of the forbid- ity are [N/H] = 0.6 (logarithmic abun- has also been observed close to the den [Fe II] lines is line scattering of dance ratio relative to solar), while oxygen ionisation front (Bautista et al., 1994). B star photons in the strong UV reso- is slightly reduced, [O/H] = – 0.1. Our best nance lines, which was first seen with value for the nitrogen abundance de- – In addition to emission from the ionisa- IUE (Hagen et al., 1987). One of the termined independently from a fit of the tion fronts there is apparently a further strong UV resonance (scattering) lines Cloudy models to observed [N II]/HA component (cf. Figure 3, 4 to 6offset) in the UV spectrum of A Sco B is the values is [N/H] = 0.52 ± 0.1 with [O/H] ≤ in the centre of the H II region. This is Fe II UV mult. 3 line at 2 344 Å, which – 0.05 which is fully consistent with the again consistent with our Cloudy simu- shows a P Cyg type profile. In the case expectations. lations which show such a further of the 2 344 Å line, a second downward max imum. We notice, however, that the channel exists via the Fe II lines 4 923, Last but not least: the same Cloudy ‘central’ FeII emission varies strong- 5 018, and 5169 Å so that part of the models with the above mentioned abun- ly with slit position which may have its dances show that metal line cooling origin in a time-variable (episodic) II of the H II region is dominated by [N II] and supergiant wind. There is independent Figure 5. ‘Look back’ contours (solid lines) of H region borders in a plane perpendicular to the plane collisionally excited [FeII]. This is at least evidence, from both IR inter ferometric of the sky caused by the combination of wind ex- part of the explanation for the strong [Fe II] imaging of the immediate surrounding pansion with orbital motion of A Sco B relative to spectrum of the Antares nebula. of Antares and from the multiple com- A Sco A. Contours are shown for present and previ- ponent structure of CS absorption lines ous phases (–150 yrs to –450 yrs; 150 years being the wind travel time for the projected distance be- seen in HST spectra of A Sco B (Baade tween A and B). This is a qualitative model for the [Fe II] emission & Reimers, 2007), for a time-variable cometary-like structure of the [Fe II] emission regions. wind. The comparison of the distribution of the [Fe II] emission in position and velocity – Emission is clearly seen outside of the space with HA (Figures 2 and 3) shows H II region: beyond 5.9 west of A, the distinct differences: [Fe II] emission becomes more extended 0y and more structured. In velocity space −150y – The [Fe II] emission is not smoothly dis- the [Fe II] emission moves in the mean −300y –1 tributed like HA and [N II] but is appar- from 4 km s 1.9 west of A through −450y ently concentrated on the edges – the 3 km s–1 at B to 0 km s–1 5.9 west of A ionisation shock fronts – of the H II re- and ~ – 5.5 km s–1 at 9.9 west of A. gion. The [Fe II] emission thus gives rise to a ring-like structure (extending up Apparently we start to see here time- to ~ 6from Antares outside the H II re- dependent effects: the M supergiant gion itself), which ends roughly at 6, wind carries the structure imprinted by thus making a double structure. This the H II region (‘shock fronts’) to large ring-like and double-peak shape are distances from the system outside the apparent from the front and rear side of H II region, where hydrogen is again the H II region respectively, where the neutral. In addition, the systematic shift red component (rear side) is always in velocity can be explained by a com- stronger. We notice in passing that the bination of B star orbital motion with occurrence of [Fe II] in the Orion Nebula wind expansion. An analysis of the orbit

The Messenger 132 – June 2008 35 Astronomical Science Reimers D. et al., The Antares Emission Nebula and Mass Loss of A Sco A

downward cascade is via these lines H II region shock fronts – which is cer- fers from the episodic nature of mass and this is what we observe – Fe II tainly an oversimpliﬁcation. The geometry, loss (multicomponent absorption) and 5169 Å in emission. This channel popu- i.e. the location of the B star and its from the observation that dust depletion lates the upper level of the strongest H II region relative to the plane of the sky onto grains is important (Baade & Reim- [Fe II] lines 4 287 Å and 4 359 Å. In con- at the M supergiant, has been deter- ers, 2007). clusion, we have identiﬁed in detail mined using the HA and [Fe II] velocities from IUE and UVES observations the as 23˚ behind the plane of the sky. [Fe II] excitation mechanism in the cold, References neutral M star wind far outside the H II The best match of the Cloudy models Baade, R. & Reimers, D. 2007, A&A, 474, 229 region. with the observed HA intensity distribu- Bautista, M. A., Pradhan, A. K. & Osterbrock, D. E. tion (Figures 2 and 4) yields a mass-loss 1994, ApJ, 432, L135 –6 rate of 1.05 × 10 M/yr with an uncer- Ferland, G. J., Korista, K. T., Verner, D. A., et al. Mass loss of A Sco A tainty of ± 30 %. This is a mean value over 1998, PASP, 110, 761 Hagen, H.-J., Hempe, K. & Reimers, D. 1987, the crossing time of the wind through the A&A, 184, 256 The number of Lyman continuum pho- H II region of roughly 150 yrs. We believe Harris, M. J. & Lambert, D. L. 1984, ApJ, 281, 739 tons and the wind density (mass-loss that the new value for the mass-loss rate Hjellming, R. M. & Newell, R. T. 1983, ApJ, 275, 704 rate) determine the shape (size) of the is more reliable than (our own) previous Kudritzki, R. & Reimers, D. 1978, A&A, 70, 227 Reimers, D., Hagen, H.-J., Baade, R., et al. 2008, H II region. At this time we have per- measurements, mainly because the use A&A, submitted formed only static H II region model calcu- of circumstellar absorption lines in the B Struve, O. & Zebergs, V. 1962, Astronomy of the 20th lations assuming that the density struc- star spectrum, formed in the vast circum- Century, (New York: MacMillan Comp.), 303 ture is not grossly disturbed by the stellar envelope of the M supergiant, suf- Swings, J. P. & Preston, G. W. 1978, ApJ, 220, 883

Another early VLT image showing the high ionisation bipolar planetary nebula NGC 6302. This colour pic- ture is a composite of three VLT UT1 Test Camera exposures through B, V and R ﬁlters, taken in June 1998. The extremely hot (~ 200 000 K) central star is hidden within the slender wisp of dust at the core of the nebula.

36 The Messenger 132 – June 2008