Online material

Appendix A: Spectral classification of emission-line stars

We have spectra of 29 emission line stars. These stars were classified in previous studies as either Ofpe/WN9 type, WN, or WC type stars. Ofpe/WN9 types have narrower lines and are cooler (T_eff = 10 000−20 000 K) than WN and WC stars (T_eff> 30 000 K). We found some disagreement with previously reported spectral classifications for some of the emission-line stars.

Id 144/AF and Id 1237/IRS 7E2(ESE): we find that the two stars Id 144/AF and Id 1237/IRS 7E2(ESE), which were listed as Ofpe/WN9 by Paumard et al. (2006), have broad emission lines and are rather WN8 or WN9 stars. Paumard et al. (2006) stated that their spectra of these stars are of high quality. But we also have a high S/N, 78.2 for Id 144/AF and 35.6 for Id 1237/IRS 7E2. Furthermore, the high-resolution spectra of Tanner et al. (2006) agree with Id 144/AF being a broad emission-line star, but they have no data for Id 1237/IRS 7E2(ESE). Paumard et al. (2003) also found a broad He I line in the spectrum of Id 144/AF. The resolving power R reported by Tanner et al. (2006) was 14 000 and 23 300, but only 1500 and 4000 in the data used by Paumard et al. (2006). Our data set with R = 4300 agrees with the high-resolution results from Tanner et al. (2006). Because of their broad emission lines (FWHM ~ 700 km s^-1) and the resemblance of the spectra of Id 144/AF and Id 1237/IRS 7E2 with the WN8 and WN9 spectra in our data set, we classify Id 144/AF and Id 1237/IRS 7E2 as broad emission-line stars, probably WN8 or WN9. Id 1237/IRS 7E2 is also classified as a WN8 star by Martins et al. (2007). Martins et al. (2007) used non-LTE atmosphere models to derive the properties of Galactic centre stars. For Id 144/AF they found a degeneracy between the effective temperature T_eff and the helium abundance He/H. In addition, the wind of this star could be stronger than the wind of the Ofpe/WN9 stars, and Id 144/AF may be more evolved. Martins et al. (2007) suggested that Ofpe/WN9 stars evolve to WN8 stars.

Id 666/IRS 7SW: we reclassify the star Id 666/IRS 7SW as WN8/WC9. This star was classified as WN8 in Paumard et al. (2006). The C III and C IV lines distinguish a WN8/WC9 star from a WN8 star. These lines are weaker than the He and H lines. Therefore a low S/N can lead to a misidentification as a WN8 star, but Paumard et al. (2006) state that their spectrum of Id. 666/IRS 7SW is of high quality. Our spectrum of this star has a S/N of 59.9, and we can clearly identify the C IV doublet at 2.0796 μm, and 2.0842 μm C III at 2.325 μm (see Fig. 11). The spectrum is very similar to the spectrum of the WN8/WC9 star Id 491/IRS 15SW, therefore we conclude that Id 666/IRS 7SW is also a WN8/WC9 type star. The two stars Id 491/IRS 15SW and Id 666/IRS 7SW are also in a similar location at the

colour-magnitude diagram ((H−K_S)₀ = −0.02 and −0.11, K_S,0 = 9.40 and 9.59, Fig. 3), which confirms their similarity. This classification agrees with that of Martins et al. (2007).

Id 185/IRS 29N, Id 283/IRS 34, Id 303, and Id 638/IRS 29NE1: four of the eleven WC stars have only shallow emission lines in our data set. These stars are Id 185/IRS 29N, Id 283/IRS 34, Id 303, and Id 638/IRS 29NE1. The lines are very broad, but only weakly pronounced. Previous studies (Paumard et al. 2001; Tanner et al. 2006) did not detect any distinct He I emission for Id 185/IRS 29N, while Paumard et al. (2003) reported a broad He I emission line for the same star. Tanner et al. (2006) suggested that Id 638/IRS 29NE1 is variable and that the spectral features changed with time. Rafelski et al. (2007) studied the light curve of IRS 29N over a time line of ten years and found photometric variability. They suggested that these sources could be a wind-colliding binary system. Gamen et al. (2012) showed that stars can change their spectra within months.

Apart from the weak but broad lines, all the four sources are also very red ((H−K_S)₀> 0.55). Their continua rise steeply with wavelength. This is an indication that these sources are embedded in dust (see e.g. Geballe et al. 2006 for IRS 8). The continuum in the spectra might not be the stellar continuum, but the continuum of the circumstellar dust, which dominates the lines (Figer et al. 1999; Chiar & Tielens 2001). Therefore the emission lines appear only as weak, broad bumps in the spectrum.

Circumstellar dust is common for WC9 stars such as Id 185/IRS 29N and Id 283/IRS 34. For earlier types such as WC8, dust formation is rather uncommon and might indicate colliding winds (Sander et al. 2012). The two stars Id 303 and Id 638/IRS 29NE1 are WC8/9 stars. Id 303 is located close to the minispiral, at least in projection. So it might be a bow-shock that causes the reddening of Id 303. The bow-shock sources Id 161/IRS 5 and Id 25347/IRS 1W are probably also embedded WR stars (Tanner et al. 2005; Sanchez-Bermudez et al. 2014), but their emission lines are outshone by the bow-shock continuum. Id 638/IRS 29NE1, however, is not located inside the minispiral. But as mentioned earlier, the spectral features seem to change with time. This could be explained by circumstellar dust.

Id 243/IRS 34W: the star with a narrow emission line, Id 243/IRS 34W, has moderate reddening ((H−K_S)₀ = 0.24) and a steeply rising continuum. As Paumard et al. (2003) already pointed out, this star is fainter than the other stars with narrow emission lines (see also the colour–magnitude diagram, yellow circle with black cross, Fig. 3). The star Id 243/IRS 34W shows a long-term photometric variability (Ott et al. 1999; Paumard et al. 2003). Therefore this star may also be dust embedded. Id 243/IRS 34W is a LBV candidate, and Humphreys et al. (1999) showed for the case η Carinae that LBV eruptions are accompanied with circumstellar dust obscurations.

Appendix B: O/B star tables

© ESO, 2015