SPECTROSCOPIC STUDY OF THE N159/N160 COMPLEX IN THE LARGE MAGELLANIC CLOUD

The images were taken during two consecutive nights in 2003 November, at the Las Campanas Observatory (LCO), Chile, with the 2.5 m Irénée du Pont telescope and the Wide Field Reimaging CCD Camera, in the multi-object spectroscopic mode. The detector was a Tek\#5 CCD with 2048 × 2048 (24 μm) pixels providing images of 25 arcmin diameter at a scale of 32.3 arcsec mm⁻¹ (0.774 arcsec pixel⁻¹). The H&K grism, centered at 3700 Å, was used resulting in spectrograms with reciprocal dispersion of 1.3 Å pixel⁻¹ and ∼3.1 Å resolution. We used nine different slit masks grouping stars of similar brightness, selected in order to produce the minimum overlapping among the resulting spectra. Each mask has an average of 25 slits of variable length and 1 arcsec wide. The number of spectra obtained in each mask depends on the number of stars falling in each slit. Individual exposures were 1200 s long; two or three exposures were taken for each slit mask to combine into the final spectra.

No optical photometry was available at the time the multislit masks were made, therefore program stars were selected from colors calculated from their Two Micron All Sky Survey (2MASS) JHK photometry. We derived an IR analogue to the Q parameter; Q_IR = (J − H) − (H − K)E_J−H/E_H−K. Although we were aware that infrared photometry is even less sensitive than UBV for early-type stars, this parameter has behaved well in detecting OB stars in the area. Astrometry was derived using a set of stars common to our acquisition images and the 2MASS catalog.

The data were reduced using the IRAF⁷ software package at the La Plata Observatory. The bias level was removed but the images were not divided out by flat fields because the quartz lamp exposures were obtained with the multislit masks in and adjacent spectra slightly overlap each other. The subtraction of background nebular emission was not trivial in some of the spectra. Two situations are worth mentioning to illustrate this point.

• 1.  
  In our multi-object spectroscopy masks several stars did not fall in the center of the slit, but rather near one of the ends, making it difficult to take a background sample large enough to be properly removed. This results in background contamination that is not easy to remove due to the fact that the nebular background emission in which the program stars are immersed is bright and highly variable on small spatial scales. Many of these are stars that were not originally included in the program but showed up as bonus spectra in program star slits.
• 2.  
  Because the spectra are slightly curved along the dispersion direction, some spectra from consecutive slits are overlapped. This contamination from other slits prevented us from obtaining a proper sample of adjacent background emission.

These are the reasons that in some of the spectra presented the background nebular emission is noticeable, although in most cases it does not affect the spectral classification. The signal-to-noise ratio (S/N) covers a range from 20 for the noisier spectra to 210 for those of best quality. The histogram distribution among these values is nearly flat.

The spectra were classified mainly following the criteria of the Digital Atlas of Optical Spectral Classification of OB Stars (Walborn & Fitzpatrick 1990). Other sources were consulted for some particular spectra, among them are the Digital Atlas of Peculiar Spectra (Walborn & Fitzpatrick 2000), The Spectral Classification System for the Earliest O Stars (Walborn et al. 2002) and R. O. Gray's Digital Atlas of Spectral Classification.⁸ In Figure Set 1, the best quality spectra from our data set are presented, ordered by sequence of spectral type and luminosity class. In Table 1, all the stars for which we obtained a spectral classification are listed, along with their J2000 celestial coordinates and V magnitudes from Zaritsky et al. (2004). Correlation between our star list and the photometric catalog was made by coordinates and we exclude magnitudes for three stars that seem to be misidentified: star 1 is very close to the limit of our field image and stars 41 and 42 are very close together, immersed in a bright background. In Table 2, we have listed our program stars that had already been classified by other authors. As mentioned in the introduction, only a few of the stars in our sample have previous spectral classifications based on spectroscopic observations, although many of them are included in several studies, see Meynadier et al. (2004), Degioia-Eastwood et al. (1993), Heydari-Malayeri et al. (2002), and Jones et al. (2005) among others. As can be seen, only nine stars were included in Sanduleak (1970) and the spectral class given is no more precise than "OB" type, except for Star 7. A few stars were also found in common with other catalogs of spectral classification through careful review of finding charts, since the difference in accuracy between the 2MASS coordinates and those in old catalogs made it impossible to search objects by coordinates.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

View figure set (20 panels)

Star 33 (Figure 1.8) has a peculiar spectrum, which was classified as O9.7 Iabpe. It presents emission at Hγ and all the lines appear peculiarly weak for a supergiant star.

Star 35 (Figure 1.9) is a widely studied object because it is the optical counterpart to the X-ray source LMC X-1. Hutchings et al. (1983) found it to be an O7 star with a binary period of approximately 4 days; they also suggested that the secondary could be a black hole. It was later classified as O7-9 III by Pakull (1984) and Bianchi & Pakull (1985). We find a type of O8(f)p, where the peculiarity refers to the reversed He ii λ4686 emission. For this reason it has been associated with the Onfp class by N. R. Walborn et al. (2009, in preparation), of which the He ii profile is a defining characteristic, although in this object the absorption lines are not broadened, and the He ii profile may be related to the X-ray binary system.

Star 37 (Figure 1.4) was classified as O8 here, although Conti & Fitzpatrick (1991) classified it as O3–O6, based on the presence of He ii λλ4200, 4541, 4686 absorption lines and possible N iii λλ4634-40-42 emission lines. The stellar He i absorption lines were obliterated by the very strong nebular emission lines in their observation. For our classification we used mainly the ratios of He ii λ4541/He i λ4387 and He ii λ4200/He i λ4144. We also used the oversubtracted nebular [O iii] λ4363 line present in the final spectrum to estimate the nebular oversubtraction (see Bosch et al. 1999). We found a correction of less than 15%, which indicates that the stellar He i λ4471 absorption is indeed deeper than He ii λ4541. In addition, the C iii λ4070 absorption blend is present. All of these considerations indicate that the spectral type of Star 37 must be closer to O8. Unfortunately, our wavelength coverage for this object did not allow us to check for the presence of N iii emission, as suspected by Conti & Fitzpatrick (1991). The luminosity class could not be assessed due to the lack of the He ii λ4686 line.

Star 38 (Figure 1.17) is a B3 supergiant that at one time was also a candidate for identification with LMC X-1. Fortunately, the weak He ii λ4686 emission seen in high S/N digital data was not visible in the early observations, or this misidentification would have been considered definite! Ramsey et al. (2006) have clearly shown that this anomalous emission feature actually arises in an unusual high-ionization nebulosity in the field, which may be related to the X-ray source.

Star 39 (Figure 1.8) was classified as O9.7 Ia+pe. It is certainly an unusual spectrum, probably related to the Ofpe/WN9 or WNL class. See, for comparison, the spectrum of HD 152408 (Walborn et al. 2000). However, these characteristics have not been seen before at such a late O type.

Star 41 (Figure 1.1) is the hottest star we have found among the sample. It was classified as O3 Vz((f*)), although the z nature (He ii λ4686 absorption stronger than other He lines, possibly an indicator of subluminosity and extreme youth) is marginal. N iii λλ4634-40-42 and N iv λ4058 are clearly detected in emission and N v λλ4604-20 absorption lines are also present. Si iv λλ4089-4116 seem to be in emission but the features are not conspicuous.

Star 93 (Figure 1.10) was classified as BN0 Iabp due to enhanced N iii and deficient C and O lines, characteristic of OBN stars (Walborn 1971a, 1976). The additional peculiarity is the presence of P Cygni profiles at Hγ and He i lines.

Star 179 (Figure 1.10) was classified as B Iae P Cyg. The P Cygni features can be seen in H and Hδ, but not in Hγ which is totally in emission. This object is quite similar to HD 87643 (Walborn & Fitzpatrick 2000), which belongs to the group of Iron Stars, of which η Carinae is the most famous member.

Stars 54, 70, 76, 107, and 112 (Figure 1.20) show either anomalously broad or asymmetric profiles in their absorption lines, suggesting a possible binary nature. The wavelength resolution does not allow us to reliably analyze radial velocity information. Higher resolution spectra would be needed to assess the possibility of these being SB2 stars.

Stars 72, 82, and 151 (Figure 1.9) belong to the Onfp category (Walborn 1973). In their spectra, the broad emission in He ii λ4686 with absorption reversal can be observed (see the Appendix). Stars 72 and 82 are also denoted with "+" due to Si iv λ4116 emission; they also present C iii λ4650 in emission. A luminosity class cannot be inferred for these stars because of the peculiar He ii λ4686 profiles (N. R. Walborn et al. 2009, in preparation).

30 Doradus is the current optical "hot spot" near the northeastern edge of a much larger, oval region several kpc in diameter, itself northeast of the LMC Bar, in which star formation has been active for at least 10⁸ yr. This region contains half the WR content of the LMC, numerous young clusters of various ages, many red supergiants, and SN 1987A. The field south of 30 Dor is particularly rich in clusters, associations, and nebulae, including the N159/N160 complex studied here. This field also contains the most massive CO concentrations in the LMC, so it may be expected that in a few million years 30 Dor's successor will appear there. In order to analyze the current, global evolutionary trends within the N159/N160 complex, the observed stars have been grouped following qualitative criteria based on those defined and used in Walborn & Blades (1997); see also Bosch et al. (1999). We have defined three groups of spectral types and luminosity classes. These groups comprise distinct age ranges in massive stellar evolution. Then from the total of 189 spectra in the field of N159/N160, we have selected those that belong to one of the following three groups.

• 1.  
  Young phase: O3-5 V-I and O Vz.
• 2.  
  Middle-age phase: O6-9 V-I and B0 I.
• 3.  
  Evolved phase: B0-2 V-III and B1-8 I.

Of course, some of the main-sequence stars in the second and third groups could actually belong to the younger ones, but this grouping is a first approximation for our purposes. The three groups are plotted in a direct image of the field in Figure 2, with different symbols for each. If a global sequential star formation process were taking place (as suggested by CO studies mentioned in Section 1 and also studies of neighboring regions like N158 (Testor & Niemela 1998)), we would expect to find systematic trends in the object distribution according to their evolutionary stages. Indeed, there is evidence for such effects in Figure 2.1: there is a clear tendency for the two younger phases to be associated with the nebulae and clusters, while the majority of the third phase are in the dispersed northeast quadrant of the field, which is devoid of nebulosity. On the other hand, young phase objects are found in the northern part of the complex (N160) as well as in the southern part (N159), while some more evolved objects also appear in the latter. Of course, we are looking at a two-dimensional projection of three-dimensional structures. It is quite likely that the younger concentrations are embedded in and/or projected against a field of more evolved and dispersed objects, just as occurs in 30 Dor itself (Walborn & Blades 1997; Selman et al. 1999). Also, some isolated very young objects may be runaways from the young clusters. Our data favor a scenario in which star formation was initiated throughout the region, perhaps in the event producing the kpc-scale structure noted above, and that this strong burst led to subsequent, localized star formation events wherever sufficient remanent molecular material was available, in an ongoing process that continues at the present time. This scenario is in good agreement with other studies of subregions in the N159/N160 complex, as in Heydari-Malayeri et al. (2002) where the presence of compact high-excitation H ii blobs demonstrates that this is indeed a very young subregion likely containing different (early) evolutionary stages. Stellar spectra alone do not provide the definitive ages that are required to construct a quantitative star formation history of the region. A complete photometric study will be essential to disentangle the overlapping populations and their extinctions. Such data are available from Zaritsky et al. (2004) and will be extended by new observations we are currently undertaking. Subsequently, we shall undertake a more detailed analysis including H–R diagrams and comparison with the extensive Hα, IR, and CO observations available for this region.

Zoom In Zoom Out Reset image size

View extended figure (4 panels)