A Binary Star in the Superbubble N160 in the Large Magellanic Cloud

The Spitzer Space Telescope observations of the LMC, especially those from the program Surveying the Agents of Galaxy Evolution (Meixner et al. 2006), have been used in conjunction with ground-based optical and near-IR photometric surveys to identify young stellar objects (YSOs) in the LMC (Gruendl & Chu 2009). During the course of that work, it is noticed that over 100 stars are blue, with $U-B\lt 0$, and have 8 μm or 24 μm excesses. These blue stars with IR excesses could be YSOs or evolved stars. To probe the true nature of these stars, we have obtained long-slit medium-dispersion spectra, which allow us to examine properties of not only the stars but also the ambient ISM. One of the blue stars with IR excesses, 053949.2–693747.4 (Gruendl & Chu 2009), exhibits stellar velocity that is offset from the ambient ISM velocity by almost 150 km s⁻¹, suggesting a possible runaway star. It is nevertheless also possible that this peculiar velocity originates from orbital motion in a binary system; thus, additional spectroscopic observations have been obtained to distinguish between the two possibilities.

The multiwavelength photometric measurements of 053949.2–693747.4 were taken from various sources. Its UBVI photometry is from the Magellanic Cloud Photometric Survey (MCPS; Zaritsky et al. 2004), JHK photometry from the Two Micron All-Sky Survey (2MASS; Skrutskie et al. 2006), and Spitzer IRAC 3.6, 4.5, 5.8, and 8.0 μm and MIPS 24 μm photometry from Gruendl & Chu (2009). These photometric measurements are listed in Table 1 and the spectral energy distribution (SED) is plotted in Figure 1.

Zoom In Zoom Out Reset image size

The first long-slit medium-dispersion spectra of 053949.2−693747.4 were obtained with the Goodman Spectrograph (Clemens et al. 2004) on the SOAR 4.1 m telescope at Cerro Pachon on 2010 February 3. We performed additional observations at two epochs, on 2020 January 9 and 2020 February 12, with the same spectrograph but increasing spectral resolutions. Narrower slits were used, and the spectral resolutions estimated from the FWHM of comparison lines in lamp spectra are ∼500, ∼1000, and ∼4000 at the wavelength of 5000 Å. The log of these spectroscopic observations is presented in Table 2.

Table 2.  Log of Spectroscopic Observations with Goodman Spectrograph on SOAR 4.1 m Telescope

Date	UT	Exp. Time	Spectral Range	Grating/Filter	Slit Width	Spectral Scale	Spatial Scale	Airmass
	hh:mm:ss	(s)	(Å)			(Å pixel−1)	(arcsec pixel−1)
2010 Feb 3	03:05:42	300	∼3700–8500	RALC 300/GG-385	103	∼2.6	∼0.30	1.34
	03:00:20	300	∼3700–8500	RALC 300/GG-385	103	∼2.6	∼0.30	1.35
2020 Jan 9	02:47:56	600	∼3700–7100	SYZY 400/NoFilter	10	∼2.0	∼0.32	1.30
	02:58:06	600	∼3700–7100	SYZY 400/NoFilter	10	∼2.0	∼0.32	1.30
	03:08:17	600	∼3700–7100	SYZY 400/NoFilter	10	∼2.0	∼0.32	1.30
2020 Feb 12	02:21:12	1200	∼3855–5155	RALC 1200 RED/NoFilter	06	∼0.32	∼0.14	1.34
	02:41:41	1200	∼3855–5155	RALC 1200 RED/NoFilter	06	∼0.32	∼0.14	1.36
	03:02:16	1200	∼3855–5155	RALC 1200 RED/NoFilter	06	∼0.32	∼0.14	1.38

Download table as:  ASCIITypeset image

The long-slit spectrogram from 2010 is shown in the top panel of Figure 2. These spectroscopic data were reduced with standard procedures in the software package IRAF. The images were bias-subtracted and flat-fielded first. Then the wavelength was calibrated by comparisons with HgAr lamp spectra taken before and after the target observation. The flux calibration was done using a spectrum of the spectrophotometric standard star EG 21 taken on the same night. Finally, spectra were extracted from apertures centered on the star and background on both sides of the star along the slit, as marked on the spectrogram in Figure 2. Because the surface brightness of the background nebular emission varies steeply along the slit, simply subtracting an aperture-scaled background spectrum from the stellar spectrum would leave prominent nebular line residuals. On the other hand, the nebular excitation does not vary as steeply as the surface brightness, thus we scaled the background spectrum according to the [O iii] λλ4959, 5007 lines until these forbidden lines were mostly subtracted from the stellar spectrum. The background-subtracted spectrum of 053949.2−693747.4 is plotted in the bottom panel of Figure 2. The difficulty and uncertainty in the background subtraction is illustrated by the [O iii] doublet, of which the λ5007 line is over-subtracted and λ4959 is under-subtracted. The nebular Hα is also over-subtracted, so the stellar Hα will be omitted in our analysis.

Zoom In Zoom Out Reset image size

We have searched the HST archive and found Hα and [O iii] λ5007 images of 053949.2−693747.4 taken with the Wide Field Planetary Camera 2 (WFPC2) from program 8247 (PI: Heydari-Malayeri). The program target LMC-N160A1 was in the WF2 field and 053949.2−693747.4 fell in the PC1 field. The Hα observations were made with the F656N filter in 4 × 1040 s exposures, and the [O iii] observations were made with the F502N filter in 4 × 1200 s exposures. The PC1 camera has a 36''  × 36'' field of view and a scale of 0045 pixel⁻¹ (Baggett et al. 2002).

The SOAR Goodman long-slit spectrogram (Figure 2) samples not only the spectrum of 053949.2−693747.4, but also those of the ionized gas in the background. Spectra of the background ISM are extracted at positions with 3''  ∼ 5'' offset from the star on either side, as marked in the top panel of Figure 2. The normalized spectra of this star at three epochs are plotted along with that of an O7 V star in the top left panel of Figure 5, and the background nebula spectrum in February 2010 (normalized to the continuum) is presented in the panel beneath. The right panels in Figure 5 show the mean profiles of stellar absorption lines (upper) and the background nebular emission lines (lower) for observations at three epochs.

Zoom In Zoom Out Reset image size

Radial velocity of the background ISM is measured by Gaussian-fitting of profiles of the bright Hγ, Hβ, and [O iii] λλ4959, 5007 lines separately for the two spectra from either side of the star and for the average spectrum of the two. Since the radial velocities from either side of the star are consistent with each other, we only list the radial velocities measured from the average spectra in Table 3. The individual radial velocity measurements, the median values, and the radial velocity measurements for mean profiles of stellar lines are listed in Table 3. The mean radial velocities of the background ionized gas are ∼300 km s⁻¹ for three epochs, and are close to the mean radial velocity of the LMC.

The H Balmer absorption lines of 053949.2−693747.4 are contaminated by the background nebular emission, and the background subtraction is rendered difficult because of the large variations in the nebular emission over small spatial scales, except for the last observation with a higher resolution. We therefore measure the stellar radial velocity using the much less contaminated He i λ4026 and λ4471, and uncontaminated He ii λ4200, λ4686, λ4541, and λ5411 lines. The radial velocities determined by Gaussian-fitting of the He i and He ii line profiles and mean profiles of the stellar lines are also listed in Table 3. We have also cross-correlated the observed spectra of 053949.2−693747.4 with a template spectrum of an O7 V star to determine its radial velocity.

While the radial velocities of nebular emission lines are consistent with one another at three epochs as shown in the lower panel, the radial velocities of stellar absorption lines at the latest two epochs are lower than the value measured from the 2010 spectrum and closer to the background nebula velocity of ∼300 km s⁻¹. From the 2010 spectrum, the radial velocity of the mean profile of stellar absorption lines is measured to be 462 ± 23 km s⁻¹. The median of the radial velocity measurements for individual lines is 449 ± 48 km s⁻¹. A radial velocity of 441 ± 11 km s⁻¹ is derived from the cross-correlation with a template spectrum. These values are consistent with one another within the error limits, and are significantly higher than the background nebula velocity of ∼300 km s⁻¹. On the other hand, from the 2020 January and February spectra, the radial velocities of stellar absorption lines are consistent with the background nebular velocity within the error limits, as shown in the right panels in Figure 5.

The two spectra taken in 2020 show similar stellar radial velocities that are consistent with those of nebular emission lines. To exclude the possibility that the discrepancy of radial velocities between nebular lines and stellar lines measured from the 2010 spectrum is due to some systematic cause such as the misalignment of the target, we examined spectra of other stars obtained with the same setting on the same night. First, we checked slit images for each target, and confirmed that the seeing on the night was worse than the slit width. Then, we checked the peak position on the slit for each target, and confirmed that the miscentering of each target on the slit is less than 0.4 pixel, or < 1/8 of the slit width, which is too small to explain 150 km s⁻¹. In addition, other stars with comparable miscentering do not show the peculiar velocity. We, thus, concluded the observed large variation of stellar radial velocities of 150 ± 30 km s⁻¹ is real and indicates that 053949.2−693747.4 is a binary system.

In this section, we investigate the binary parameters of the system from the variable information. Because spectral lines of the O7 V star's companion are not clearly detected in the observed spectra, the companion must be fainter than the O7 V star. This is also supported by the absolute V magnitude of the star consistent with that of a single O-type main-sequence star at the distance of the LMC. No X-ray point source is detected there, thus, the companion is not a compact object. It is thus reasonable to consider that the O7 V star is the primary and the companion is the secondary in this binary system.

The minimum value of the semiamplitude of the radial velocity curve, K₁, is 1/2 of the 150 km s⁻¹ velocity difference between the 2010 and 2020 spectra, 75 km s⁻¹, for a systemic velocity that is the average of the 2010 and 2020 stellar velocities, or 75 km s⁻¹ from the background nebular velocity. If the systemic velocity is the same as the background nebular velocity, the minimum value of the K₁ velocity would be ∼150 km s⁻¹. The hidden companion star should be massive enough to cause the large radial velocity variation of the O7 V star in its orbital motion. We, thus, surmise that the companion star is an early B V star.

Assuming a circular orbit, the distance of the O7 V star to the center of mass, ${a}_{1}$, can be expressed as

$$\begin{eqnarray*}&&{a}_{1}\,=\,{P}* {K}_{1}/(2\pi \ast \sin i),\end{eqnarray*}$$

where P is the orbital period and i is the angle between the orbital plane and the sky plane. The distance between the two stars, a, is

$$\begin{eqnarray*}&&a={a}_{1}\ast ({M}_{1}+{M}_{2})/{M}_{2},\end{eqnarray*}$$

where M₁ and M₂ are the masses of the O7 V star and its early B V companion, respectively. With the above two expressions and the assumption of i = 0°, Kepler's third law can be rewritten as

$$\begin{eqnarray*}&&P=G\ast (2\pi )\ast {M}_{2}^{3}/{\left({M}_{1}+{M}_{2}\right)}^{2}/{K}_{1}^{3},\end{eqnarray*}$$

where G is the gravitational constant.

Using this Kepler's law and reasonable parameter values of ${M}_{1}={M}_{{\rm{O}}7{\rm{V}}}$, $0.3\lt {M}_{2}/{M}_{1}\lt 1$, and ${K}_{1}\gt 75$ km s⁻¹, we can constrain the period P for a given binary separation a. The separation between the two stars has to be greater than the radius of the primary O7 V star, i.e., a  >  11 R_⊙; thus, the period P should be longer than 1 d. On the other hand, if K₁ is ∼150 km s⁻¹, the period will be shorter than 20 d. If we assume the companion is a B0 V star, then M₂/M₁ ∼0.5, the orbital period of the system will be ∼35 days for ${K}_{1}\sim 75$ km s⁻¹, and ∼5 days for K₁ ∼ 150 km s⁻¹.

The mean profiles of stellar absorption lines in the 2010 spectrum appear asymmetric. If the asymmetry in the profile is due to a fainter and cooler companion, the asymmetry must be seen to vary for temperature- and luminosity-dependent lines in the spectrum. Because the resolution and data quality of the 2010 spectrum is inadequate to examine each line's profile, we compared mean smoothed profiles of He i and He ii lines with a model spectrum of a O7 V+B0 V binary system. For the model spectrum, we constructed synthetic spectra of O7 V and B0 V stars taken from a synthetic spectral library (Munari et al. 2005) with respective radial velocity shifts, then convolving with Gaussian function to reduce the resolution of the synthetic spectra to the observed one of ∼500. Figure 6 shows the mean smoothed profiles of He i and He ii absorption lines (solid and dashed black lines, respectively) in the 2010 spectrum and the model spectrum (red lines). The mean profile of observed He i absorption lines is more asymmetric than that of He ii absorption lines, which is not inconsistent with the profiles of the model spectrum; however, the rather low quality of the data prevents us from further examination of details. Light blue lines are best-fit Gaussian profiles by minimizing the ${\chi }^{2}$ value against the observed data points. We note that the model spectrum is best fitted to the mean smoothed profile with a systemic velocity of ∼340 km s⁻¹, which is slightly larger than the radial velocity of the background nebular lines; though, the uncertainty in the observed spectrum is too large to make a concrete conclusion on the systemic velocity.

Zoom In Zoom Out Reset image size

The mean profiles of stellar absorption lines in spectra at the two epochs in 2020 appear more symmetric than the mean profiles of the 2010 spectrum. This suggests radial velocities of two stars due to the orbital motion are close to zero at the two epochs in 2020, thus the systemic velocity of the star is close to the nebular velocity.

Before we obtained the 2020 spectra of 053949.2−693747.4, we thought the large radial velocity observed in 2010 was caused by the runaway nature of the star. An O-type star moving supersonically through the ISM, its fast wind will interact with the ISM to form a bow shock. We have found archival HST Hα and [O iii] images of 053949.2−693747.4. As shown in Figure 7, the images show a small ionized nebula around the star and the south rim of the nebula appears to resemble a bow shock. However, examined closely, the ISM around the star has a very complex structure in juxtaposition to dust lanes whose bright rims may have given the impression of a bow-shock-like morphology, rendering the bow-shock nature uncertain. We note in passing that this small ionized gas region is within the apertures used in Spitzer photometry (Gruendl & Chu 2009), and its heated dust may well be responsible for 053949.2−693747.4's excess 8 and 24 μm emission in its SED.

Zoom In Zoom Out Reset image size

We have checked the Gaia DR2 catalog (Gaia Collaboration et al. 2018) for proper motions of 053949.2−693747.4 and 126 stars projected within 3' distance that are brighter than 18 mag and have measurements with standard errors smaller than 0.3 mas yr⁻¹ in proper motions and 0.1 mas in parallax. The proper motion of 053949.2−693747.4 is 1.771(66) mas yr⁻¹ in R.A. direction and 0.615(65) mas yr⁻¹ in the decl. direction, which is within the errors identical to the median of the proper motions of the nearby stars, 1.825(20) mas yr⁻¹ in R.A. direction and 0.617(23) mas yr⁻¹ in decl. direction. A transverse velocity of the star against the median motion of its neighboring stars is less than ∼35 km s⁻¹.