ACTA ASTRONOMICA

Vol. 66 (2016) pp. 0–0

OGLE-LMC-ECL-09937: The Most Massive Algol-Type Binary

System With A Mass Measurement Accurate to 2%
D. M. S k o w r o n 1 , M. K o u r n i o t i s 2,3 , J. L. P r i e t o 4,5 , N. C a s t r o 6 ,
arXiv:1801.05845v1 [astro-ph.SR] 17 Jan 2018

A. Z. B o n a n o s 3 , D. K. P i e ń k o w s k i 1
1
Warsaw University Observatory, Al. Ujazdowskie 4, 00-478 Warszawa, Poland
e-mail: [email protected]
2 Astronomický ústav AVČR, Ondřejov, 25165, Czech Republic
3 IAASARS, National Observatory of Athens, Penteli 15236, Greece
4 Núcleo de Astronomía de la Facultad de Ingeniería, Universidad Diego Portales,

Av. Ejército 441, Santiago, Chile

5 Millennium Institute of Astrophysics, Santiago, Chile
6 Department of Astronomy, University of Michigan, 1805 S.University, Ann Arbor,

MI 48109, USA

Received December 11, 2017

ABSTRACT

This paper presents a detailed analysis of the light and radial velocity curves of the semi-detached
eclipsing binary system OGLE-LMC-ECL-09937. The system is composed of a hot, massive and lu-
minous primary star of a late-O spectral type, and a more evolved, but less massive and luminous
secondary, implying an Algol-type system that underwent a mass transfer episode. We derive masses
of 21.04 ± 0.34 M ⊙ and 7.61 ± 0.09 M ⊙ and radii of 9.93 ± 0.06 R ⊙ and 9.18 ± 0.04 R ⊙ , for the
primary and the secondary component, respectively, which make it the most massive known Algol-
type system with masses and radii of the components measured with < 2% accuracy. Consequently,
the parameters of OGLE-LMC-ECL-09937 provide an important contribution to the sparsely pop-
ulated high-mass end of the stellar mass distribution, and an interesting object for stellar evolution
studies, being a possible progenitor of a binary system composed of two neutron stars.
Key words: binaries: eclipsing – Stars: early-type – Stars: fundamental parameters – Stars: mas-
sive

1. Introduction

Mass is the most important property of a star. The initial mass and metallicity
to a large extent determine the whole life and fate of a star. Moreover, the evo-
lution of the star in a binary system is significantly influenced by its companion
(Sana et al. 2012). Most stars with well-determined masses are Galactic stars with
roughly solar metallicity. Obtaining accurate parameters for stars in environments
Vol. 66 1

with different metallicities is crucial for testing stellar evolutionary theories and
for measuring distances to nearby galaxies (e.g., Bonanos et al. 2006). Moreover,
investigating stars in the lowest metallicity environments is a step toward explor-
ing the early Universe and puts us closer to the Population III regime of metal-free
stars (even if we lack true zero metallicity stars). The stellar mass distribution, par-
ticularly the initial mass function (IMF), is fundamental for constructing accurate
models of galaxies and tracing their evolution.

The high-mass end of the stellar mass distribution is very poorly constrained,
which is a serious problem for verifying theoretical models of stellar evolution. At
the same time, the evolution of massive stars is violent and their deaths lead to
many interesting phenomena. For example, massive binary stars are thought to be
the progenitors of Gamma Ray Bursts, especially in the case of Population III stars
(e.g., Toma et al. 2016). Furthermore, procuring accurate fundamental measure-
ments of the masses of high-mass stars is essential for determining the slope of the
IMF.

The review by Torres et al. (2010) lists all detached binary systems of any mass
in which the mass and radius of both components are measured with ≤ 3% accu-
racy. This is the precision necessary to test models of stellar evolution, and provide
constraints strong enough that models with incorrect underlying physics can be re-
jected. The list contains 190 individual stars, of which only two are extragalactic.
The most massive star on the list is a component of V3903 Sgr which has a mass
of only 27.27 M ⊙ , and there are only three stars with masses greater than 20 M ⊙
and 17 with masses above 10 M ⊙ . DEBCat (Southworth 2015) is a living catalog
that provides physical properties of stars in detached eclipsing binary systems that
have been determined with ≤ 2% accuracy. As of today, the list contains 390 stars
(195 systems) and the record still belongs to V3903 Sgr. The number of stars with
masses greater than 20 M ⊙ is four, while with masses greater than 10 M ⊙ is 22.
For stars with M > 30 M ⊙ the situation is even worse, there are very few measure-
ments accurate to less than 10% (e.g., Bonanos 2009).

Motivated by the deficiency of massive stars with accurate mass measurements

in low metallicity environments, we undertook a spectroscopic survey of the most
luminous stars in the Large Magellanic Cloud (LMC), with an aim to measure
their masses with at least 2% accuracy, necessary for precise astrophysical studies.
The targets are the eight most luminous binary systems in the LMC identified by
Szczygieł et al. (2010). In this paper, we present the analysis of the first candidate
massive binary system from our sample, namely BI 98 = OGLE-LMC-ECL-09937.

This paper is organized as follows. Section 2 gives an overview of the obser-

vational data and its preparation. In Section 3 we present the process of estimating
physical parameters of the system, and discuss the results in Section 4. We sum-
marize the paper in Section 5.
2 A. A.

2. Observations and Data Preparation

OGLE-LMC-ECL-09937 was identified by Szczygieł et al. (2010) in the All

Sky Automated Survey (ASAS, Pojmański 1997) data as an eclipsing binary star
ASAS 051230-6727.3. It is located in the LMC at α = 78.◦ 1266632 and δ =
−67.◦ 4539948, has a period of 3.765184 d, magnitude V = 14.03 mag and am-
plitude of 0.62 mag. It has been assigned a B1:V spectral type by Massey et al.
(1995) and was included in the infrared study of massive stars by Bonanos et al.
(2009).

2.1. Photometric data

The original ASAS (Pojmański 2002) light curve of ASAS051230-6727.3 that
led to its identification as an eclipsing binary (Szczygieł et al. 2010), is of very
low quality. However, the catalog of eclipsing binaries from the OGLE survey
(Graczyk et al. 2011) contains the high quality I- and V-band light curves of the
object, extracted from the third phase of the OGLE project (Udalski et al. 2008).
There are 476 and 40 points in the I- and V-band light curves of OGLE-LMC-
ECL-09937, respectively. The data were reduced and calibrated with the standard
OGLE image subtraction pipeline (Udalski 2003). Thanks to the high quality of the
OGLE data we were able to refine the orbital period using the TATRY code based
on the multi-harmonic periodogram of Schwarzenberg-Czerny (1996) and the new
value is P = 3.765229 d. The corrected value of the time of primary eclipse is
HJD0 = 2453637.487389 d.

2.2. Spectroscopic data

Spectroscopic observations were carried out with the Du Pont 2.5 m and the
Magellan Clay 6.5 m telescopes at the Las Campanas Observatory in Chile. In the
Du Pont we used the Echelle spectrograph with a 1′′ × 4′′ slit and 2 × 2 binning,
yielding a resolution of R = 30 000, which corresponds to 10 km/s, sufficient for
velocity separations of ≈ 100 km/s. Separate exposures were collected, in order to
facilitate the removal of cosmic rays and artifacts. For the wavelength calibration,
ThAr lamp spectra of 20 s were taken before and after each science exposure. Milky
Flats and Bias frames were collected in the afternoon preceding each night. Typical
signal to noise is about 60 for the combined spectra. In the Magellan Clay we used
the Magellan Inamori Kyocera Echelle (MIKE; Bernstein et al. 2003) spectrograph
with a 1 ′′ slit and 1× 1 binning, yielding a resolution of R = 30 000. For calibration
we obtained bias frames, quartz and milky flats, and ThAr lamps for each science
exposure. Table 1 provides the log of spectroscopic observations.
The Du Pont Echelle data were reduced using standard IRAF (Tody 1986, Tody
1993) packages. First, a master Bias frame was constructed from single Bias frames
and subtracted from each science observation. Then a normalized Milky Flat was
created from a median of single images, and corrected for bad pixels. Finally,
Vol. 66 3

Table 1
Spectroscopic observations of LMC3.

DATE HJD PHASE EXPTIME AIRMASS S/N TELESCOPE

2009-12-24 2455189.664040 0.24 1200 s 1.28 73 Clay/MIKE
2010-10-08 2455477.770863 0.75 3 × 1800 s 1.41 35 Du Pont/Echelle
2010-10-10 2455479.823712 0.30 3 × 1800 s 1.31 35 Du Pont/Echelle
2010-11-09 2455509.705694 0.24 3 × 1800 s 1.36 45 Du Pont/Echelle
2011-09-02 2455806.872237 0.16 700 s 1.38 58 Clay/MIKE
2014-01-08 2456665.570596 0.22 2400 s 1.31 61 Clay/MIKE
2017-01-10 2457763.565667 0.83 3 × 1800 s 1.31 55 Du Pont/Echelle
2017-01-11 2457764.687263 0.13 3 × 1800 s 1.34 50 Du Pont/Echelle
2017-01-12 2457765.592431 0.37 3 × 1800 s 1.28 38 Du Pont/Echelle
HJD is calculated for the mid-exposure of the total of all three frames

science spectra were divided by the master Milky Flat. An initial cosmic ray rejec-
tion on single images was performed using the DCR program (Pych 2004), which
works well with spectroscopic data. The Clay MIKE data were reduced using the
Carnegie pipeline1 , which uses similar reduction steps as described for the Du Pont
Echelle reduction.
The FITS images were further processed with the noao.imred.echelle package
from IRAF. Both science and ThAr spectra were extracted with the task apall. The
line identification in ThAr spectra was done with the ecidentify and ecreidentify
tasks and the wavelength calibration of science spectra was achieved with refspec-
tra and discpcor tasks. The spectra were normalized with the continuum task and
merged with scombine. Finally, a barycentric correction was applied with the use
of rvsao.bcvcorr.

3. Binary Modeling

3.1. Spectroscopy and radial velocities

The spectrum of OGLE-LMC-ECL-09937 is characterized by the presence of
neutral helium and hydrogen lines in absorption. Both components are distinguish-
able, and most significant spectral lines are pictured in Figs. 1 and 2 at different
orbital phases. We classify the stronger, primary component following the criteria
and the optical atlas for OB stars by Walborn and Fitzpatrick (1990). The weak,
questionable HeII 4200 Å implies that the star is of late-O/early-B spectral type.

1 http://code.obs.carnegiescience.edu/mike
 4
Hγ HeI HeII HeII Hβ HeI HeI HeI Hα HeI
1.2
1.1
.131.0
0 0.9
0.8
0.7

1.2
1.1
241.0
0. 0.9
0.8
0.7

1.2
1.1
3 1.0
0. 0.9
Relative Flux

0.8
0.7

1.2
1.1
371.0
0. 0.9
0.8
0.7

1.2
1.1
761.0
0. 0.9
0.8
0.7

1.2
1.1
.831.0
0 0.9
0.8
0.7
4341 4349 4473 4479 4543 4549 4686 4693 4860 4870 4922 4929 5016 5024 5877 5886 6563 6576 6680 6690

Wavelength (Å)
Fig. 1. Selected H and He features in Du Pont spectra of OGLE-LMC-ECL-09937 shown at different orbital phases, marked on the left. The red solid and dotted
lines correspond to the primary and secondary component, respectively, shifted at their measured velocities given in Table 2. The combined fit to the observations

A. A.
(black line) is shown with a cyan solid line. The vertical solid line marks the rest wavelength of the spectral line, while the vertical dashed line, the location of the
systemic velocity.
 Vol. 66 5

The low ratio HeII 4541/HeI 4471 indicates a spectral type later than O7, whereas
the presence of HeII 4686 in absorption is typical for high-gravity stars earlier than
B0.5. Based on these diagnostics, the primary component can be assigned a late-
O main-sequence spectral type. Absorption of HeII lines is not evident for the
companion thus favoring a B-type classification, although its small contribution to
the total flux prevents identification of intrinsically weak lines.

Hγ HeI Hβ HeI HeI HeI Hα HeI

1.2
1.1
1 1.0
6
0. 0.9
0.8
0.7

1.2
Relative Flux

1.1
221.0
0. 0.9
0.8
0.7

1.2
1.1
241.0
0. 0.9
0.8
0.7

4341 4349 4473 4479 4860 4870 4922 4929 5019 5024 5877 5886 6563 6576 6680 6690
Wavelength (Å)

Fig. 2. Same as Fig. 1, but for the MIKE spectra. The four first features are from the BLUE arm and
the four last from the RED arm of MIKE.

We proceeded to measure radial velocities (RVs) and estimate spectroscopic

temperatures in two steps. First, we pursued a preliminary fit to the kinematics
of the system and inferred stellar properties that would physically constrain the
final selection of spectroscopic models. In a second step, we derived spectroscopic
temperatures and fine-tuned the RVs throughout the phases.
We employed a set of plane-parallel, NLTE, TLUSTY high-gravity models
(Hubeny and Lanz 1995) at Z = 0.5 Z ⊙ , with T eff = 30 000 K consistent with
the late-O/early-B spectral type of the system. To fit the observed spectra, we run a
least-square minimization algorithm under a Levenberg-Marquardt scheme, setting
free the RVs of the components and the flux ratio. The algorithm was run over
spectral features visible for both components, namely H lines (4340 Å, 4861 Å,
6563 Å) and HeI lines (4471 Å, 4922 Å, 5016 Å, 5876 Å, 6678 Å), which were
selected due to their strength and S/N ratio. The resulting RV curves along with
the OGLE light curves were then processed with PHOEBE v1.0 (Prša and Zwitter
2005), which wraps the renowned Wilson-Devinney code for modeling the physics
of eclipsing binaries (Wilson ans Devinney 1971).
The system was initially modeled in the “detached” mode. The fit solution
however, converged to the semi-detached configuration with the secondary com-
ponent filling the Roche lobe, hence we switched to that configuration. A by-eye
reasonable fit of the light curves was achieved from the manual adjustment of the
surface potential of the primary component Ω1 , effective temperature of the sec-
6 A. A.

ondary Teff2 , and inclination i. The mass ratio q, semi-major axis a, and systemic
velocity γ were adjusted to optimally fit the RV curves. Bolometric and passband
coefficients for the logarithmic limb-darkening law were taken from Van Hamme
(1993). After choosing the initial values, all six parameters were set free, to allow
for the built-in fit optimization of PHOEBE based on the method of differential
corrections. The period and HJD 0 were fixed to the values determined from the
light curve analysis.
The best-fit PHOEBE model yielded surface gravity values of 3.75 and 3.40
for the hot and cool component, respectively. The resulting V-band light ratio of
2.3 was used to fix the spectroscopic flux ratio. Having stellar models constrained
by these values, we repeated the RV velocity optimization using the Du Pont ob-
servations of 2017-01-10 with the highest S/N, now setting the temperature of the
models free over a grid of TLUSTY models with Teff = 19 000−40 000 K. Addi-
tionally, the models were broadened over a wide range of rotational velocity values
from 50 km/s to 200 km/s, with a step of 20 km/s.
We found that minimization of the residuals of the selected spectral features
was achieved with a primary component temperature Teff1 = 32 500 K. The tem-
perature of the secondary was found to be similar, but poorly constrained. Broad-
ening of the primary was optimized at v sin i = 150 km/s. Because of the 2500 K
step of the TLUSTY model grid, we repeated the same process using models gen-
erated with FASTWIND stellar atmosphere code (Santolaya-Rey et al. 1997, Puls
et al. 2005), which enables non-LTE calculations assuming a spherical symmetry
geometry. We generated a grid of FASTWIND templates with a step of 1000 K
at the inferred broadening with TLUSTY and using the gravity values from the
preliminary study with PHOEBE. The best fit was achieved again with a primary
temperature of 32 000 ± 1600 K and a secondary of temperature 34 000 ± 4000 K.
Interestingly, by running the fit process for every spectrum, we derived a consis-
tent Teff1 throughout the orbital phases, whereas Teff2 was largely deviating within
errors from 21 000 K to even 40 000 K due to its low strength in the composite
spectrum. We decided to infer Teff2 from the PHOEBE model and measure RVs
using two templates of 32 000 K. The resulting values are listed in Table 2 with the
best-fit model displayed over the observations in Figs. 1 and 2.

3.2. Fundamental parameters of OGLE-LMC-ECL-09937

Having measured the radial velocities, we used them together with the high
quality OGLE-III V- and I-band light curves to calculate the final fit parameters
with PHOEBE. Since the uncertainties of the parameters provided by the GUI in-
terface of PHOEBE are underestimated, we used the scripter to estimate their re-
alistic values. First, we generated a 1000 bootstrap samples of both the light and
radial velocity curves. The light curves were created by k -multicombination of the
original light curve consisting of k points (where k is 476 and 40 for the I- and V-
band, respectively). Due to the small number of points, the RV curves were created
Vol. 66 7

Table2
Radial velocity measurements for OGLE-LMC-ECL-09937

HJD RV1 (km/s) RV2 (km/s) PHASE TELESCOPE

194 (4) 599 (8) Clay/MIKE BLUE
2455189.664040 0.24
198 (4) 618 (8) Clay/MIKE RED
2455477.770863 419 (4) -2 (10) 0.76 Du Pont/Echelle
2455479.823712 196 (4) 589 (10) 0.30 Du Pont/Echelle
2455509.705694 195 (4) 599 (8) 0.24 Du Pont/Echelle
206 (4) 546 (8) Clay/MIKE BLUE
2455806.872237 0.16
207 (6) 557 (12) Clay/MIKE RED
200 (4) 603 (4) Clay/MIKE BLUE
2456665.570596 0.22
207 (4) 628 (6) Clay/MIKE RED
2457763.565667 403 (4) 29 (8) 0.83 Du Pont/Echelle
2457764.687263 215 (4) 537 (8) 0.13 Du Pont/Echelle
2457765.592431 218 (4) 550 (6) 0.37 Du Pont/Echelle

Uncertainties (2σ) are given in parentheses.

by randomly selecting each point within its 2σ uncertainty around the measured
value (listed in Table 2), assuming a Gaussian distribution. The temperature of the
primary was also randomized over its spectroscopic value 32 000 ± 1600 K. We
then ran PHOEBE on each of the 1000 bootstrap samples and derived 1000 sets of
parameters of the system. The mean values of the parameters and their 1σ uncer-
tainties are listed in Table 3 and the resulting PHOEBE model is shown in Fig. 3.
The fundamental parameters of the components are given in Table 4. We derived
masses of 21.03 ± 0.34 M ⊙ and 7.64 ± 0.09 M ⊙ and radii of 9.65 ± 0.08 R ⊙ and
9.20 ± 0.03 R ⊙ , for the primary and the secondary component, respectively. The
surface temperature of the secondary is measured to be Teff2 = 21 800 ± 1200 K,
which is marginally consistent with the poorly determined spectroscopic one.

3.3. Spot model

The resulting best-fit light curve model from PHOEBE shows room for further
improvement (see Fig. 3 and the left panel of Fig. 4). The suppression of the pho-
tometric baseline in both the I- and V-band light curves prior to the occultation of
the primary implies the presence of an additional cool component. Since spots are
expected in stars of later types than the one studied here, we attribute the observed
fluctuation in the light curve to the reflection effect, in which the hotter primary
heats the facing surface of the secondary, increasing its effective temperature. The
temperature of the unreflected region on the secondary (i.e., the projected surface
of the secondary at phase zero) would then indicate the actual Teff2 value.
8 A. A.

Fig. 3. Light curves and radial velocity curves of the best fit model with PHOEBE. The top and
middle panels show the I- and V-band OGLE-III light curves, respectively, together with residuals.
The bottom panel shows the RV curve of the primary (magenta) and the secondary (cyan) with the
best fit model for the primary (solid line) and secondary (dotted line), and the residuals.

Fig. 4. Comparison of two light curve models from PHOEBE – without a cool spot (left panel) and
with the spot (right panel). Blue dots in the top panels show the OGLE-III I-band light curve, while
the black line denotes the model. Bottom panels show the residuals between the data and the model.
Vol. 66 9

Table3
Parameters of the OGLE-LMC-ECL-09937 system measured with PHOEBE

Parameter Unit Non-Spot Model Spot Model(a)

Period P [d] 3.765229 ± 0.000003
HJD0 [d] 2453637.48739 ± 0.0002
Semimajor axis, a [R⊙ ] 31.2 ± 0.1 31.2 ± 0.1
Systemic velocity, γ [km/s] 304 ± 1 304 ± 1
Inclination, i [deg] 90.0 ± 0.4 89.6 ± 0.4
Mass ratio, q 0.364 ± 0.004 0.362 ± 0.004
Surface potential of the primary, Ω1 3.64 ± 0.02 3.55 ± 0.02
Flux ratio in the V −band 2.25 ± 0.05 1.90 ± 0.02
Flux ratio in the I−band 2.08 ± 0.04 1.80 ± 0.02

(a) A spot with longitude 190◦, a radius of 70◦, and a temperature 0.86 of the stellar surface.

Fig. 5. Location of the large cool spot on the secondary star is shown on a mesh plot imported from
PHOEBE at three different orbital phases.

Once again we used PHOEBE to model the system, this time with the addition
of a large “cool spot” on the secondary star, that would imitate the reflection ef-
fect. We assumed an equatorial location of the spot, i.e., the colatitude of 90 ◦. The
remaining three parameters of the spot: longitude, radius, and temperature were
determined via multiple fit iterations; each time a different set of the spot param-
eters from a wide range of values was chosen, the observations were fit, and the
10 A. A.

χ2 value was returned. The longitude of the spot of 190 ◦ as measured from the L1
point was found to yield the lowest χ2 values. The fact that these two points are
not anti-diametrical, is believed to be due to rotation. For spot radii above 60 ◦, the
optimization of the fit models was achieved for a spot temperature of 0.86 of that
of the stellar surface. Assuming a spot radius of 70 ◦, spanning the entire projected
surface of the secondary at phase zero (see Fig. 5), we fixed the three parameters to
the above values and repeated the iterative fit process described in Section 3.2.
The updated parameters of the model are also listed in Table 3 with the new
model shown in the right panel of Fig. 4. The residuals of the fit are greatly
improved when compared to the non-spot model, indicating a good model selec-
tion. The fundamental parameters of the updated model with the spot are given
in the right column of Table 4. The updated masses are 21.04 ± 0.34 M ⊙ and
7.61 ± 0.09 M ⊙ for the primary and the secondary component, respectively, which
are well consistent, within their 1–2% precision, with those derived from the pre-
viously discussed non-spot model.
Table 4
Fundamental parameters of the components of OGLE-LMC-ECL-09937 measured with PHOEBE

Parameter Unit Non-Spot Model Spot Model(a)

Mass of the primary, M1 [M⊙ ] 21.03 ± 0.34 21.04 ± 0.34
Mass of the secondary, M2 [M⊙ ] 7.64 ± 0.09 7.61 ± 0.09
Radius of the primary, R1 [R⊙ ] 9.65 ± 0.08 9.93 ± 0.06
Radius of the secondary, R2 [R⊙ ] 9.20 ± 0.03 9.18 ± 0.04
Teff1 [K] 32000 ± 1600 32000 ± 1600
Teff2 [K] 21800 ± 1200 21100 ± 1400(b)
log(g1 ) 3.791 ± 0.008 3.767 ± 0.006
log(g2 ) 3.394 ± 0.002 3.393 ± 0.002
log(L1 /L⊙ ) 4.94 ± 0.09 4.96 ± 0.09
log(L2 /L⊙ ) 4.23 ± 0.10 4.17 ± 0.10

(a) A spot with longitude 190◦, a radius of 70◦, and a temperature 0.86 of the stellar surface.
(b) The temperature of the cool spot, i.e., unreflected region.

4. Discussion

By comparing dynamical masses to the predicted evolutionary ones, eclips-

ing binaries with spectroscopically measured parameters serve as a unique tool
for evaluating the current theoretical models. We employed stellar evolutionary
models and isochrones from MESA Isochrones and Stellar Tracks (MIST) v.1.1.
(Dotter 2016, Choi et al. 2016, Paxton et al. 2011), assuming rotation at the 40%
of the critical velocity. The metallicity of the models was set to the average value
Vol. 66 11

of [Fe/H] = −0.39 dex for the LMC, which is measured from the calibration of
the spectroscopic data of Red Giant Branch stars (Choudhury et al. 2016). Stel-
lar tracks for masses of 8 M ⊙ to 24 M ⊙ and isochrones of 1 Myr, 3 Myr, 5 Myr,
10 Myr, and 30 Myr, are shown on the Hertzsprung-Russell diagram of Fig. 6. The
black squares show the location of the components on the diagram, as inferred from
our modeling with PHOEBE with the addition of the cool, unreflected region.

5.0

24 M ⊙
22 M ⊙
20 M ⊙
4.5
18 M ⊙
logL/L ⊙

16 M ⊙

14 M ⊙

4.0
12 M ⊙

10 M ⊙

3.5
8 M⊙
4.60 4.55 4.50 4.45 4.40 4.35 4.30
logT eff (K)
Fig. 6. Hertzsprung–Russell diagram for the components of OGLE-LMC-ECL-09937 (black squares)
studied with PHOEBE. We show MESA evolutionary stellar tracks for initial masses 8 M ⊙ to
24 M ⊙ , assuming rotation at 40% of the critical velocity. Isochrones of 1 Myr, 3 Myr, 5 Myr,
10 Myr, and 30 Myr are shown with dotted lines.

We find a good agreement between the evolutionary and dynamical mass of

21 M ⊙ for the primary component, which appears to reside well within the main-
sequence strip. The secondary component is found to be ≈ 30% overluminous for
its mass and further indicates an evolved state not far beyond the main sequence.
As the cooler, fainter, and less massive secondary is found to fill the Roche lobe, it
is reasonable to consider the system as an Algol-type binary (Giuricin et al. 1983).
Accordingly, the secondary should have been initially a massive early-type star that
underwent a rapid mass transfer to its companion, which is now the more massive
and luminous primary (Peters 2001). The agreement between the primary mass and
its theoretical single-evolved counterpart could imply that the gainer had enough
time to stabilize to its present state and that the mass transfer is currently absent.
12 A. A.

In an attempt to detect possible changes in the orbital period, that would suggest
an ongoing mass transfer, we calculated the orbital period of the system separately
for each OGLE observing season. In order to extend the baseline of the observa-
tions, we supplemented the OGLE-III I-band light curve with the OGLE-IV (Udal-
ski et al. 2015) data from the catalog of eclipsing binaries by Pawlak et al. (2016).
The combined dataset spans over 14 years of observations (14 observing seasons).
Furthermore, we downloaded the V-band light curve of OGLE-LMC-ECL-09937
from the All-Sky Automated Survey for Supernovae (ASAS-SN, Shappee et al.
2014) using the ASAS-SN Light Curve Server (Kochanek et al. 2017), spanning
over 3.5 years of observations in four observing seasons. This further extended the
baseline to almost 17 years. We calculated an accurate period value for each of the
observing seasons within OGLE-III and ASAS-SN data and for each two seasons
in OGLE-IV data (due to fewer points) with the TATRY code and found no evident
variability in the period within uncertainties.
OGLE-LMC-ECL-09937 is one of the few massive Algol-type systems with an
accurate mass measurement of 2%. In Fig. 7, we show the mass–radius plot for the
components of the Algol-type systems from the catalog of Budding et al. (2004)
updated with the most recent mass values for the components of the massive system
ET Tau (Williamon et al. 2016). Light gray dots represent stars for which there is
no error estimate for the mass, dark gray dots stand for stars for which the error on
the mass is more than 2%, and large black circles mark the stars for which masses
were determined with up to 2% accuracy. The most massive binary in the catalog
is RY Sct, which has had a number of mass measurements over the past decades,
ranging from 8 M⊙ + 26 M⊙ (Skulskii 1992) to 36 M⊙ + 46 M⊙ (Cowley and
Hutchings 1976). The more recent estimates oscillate around ≈ 10 M⊙ + 30 M⊙
but rarely provide any radii measurement. Hence we adopt the newest result by
Grundstrom et al. (2007) of 7.1 ± 1.2 M ⊙ and 30.0 ± 2.1 M ⊙ . Since the radius is
not well defined, we follow Grundstrom et al. (2007) and adopt the radii of 18 R ⊙
and 9.6 R ⊙ , respectively.
Fig. 7 shows that the primary component of OGLE-LMC-ECL-09937 (black
square) is not only the second most massive known star in an Algol-type system,
but it is also the most massive Algol component with ≤ 2% accurate mass mea-
surements. It is also one of the few massive stars with accurate physical parameters,
making it an important ingredient of the stellar mass distribution and an excellent
target for verifying stellar evolution scenarios of semi-detached systems.

5. Summary

This paper presents a detailed analysis of the photometric and spectroscopic

data on the semi-detached eclipsing binary system OGLE-LMC-ECL-09937. The
spectra of the system show lines of both components, allowing to determine both ra-
dial velocity curves, which together with the excellent quality OGLE I- and V-band
Vol. 66 13

Fig. 7. Mass–radius plot for the components of known Algol-type systems. Light gray dots represent
stars, for which there is no error estimate for the mass, dark gray dots represent stars, for which
the error on the mass is more than 2%, and large black circles mark the stars for which masses were
determined with up to 2% accuracy. The black square represents masses of the components of OGLE-
LMC-ECL-09937. The solid line denotes the zero-age main sequence at the Galactic metallicity and
the dashed at the LMC metallicity, taken from MIST (Dotter 2016, Choi et al. 2016, Paxton et al.
2011). The vertical dotted line drawn at M = 7.5 M ⊙ represents the limit above which there are no
Algol-type component masses known with ≤ 2% uncertainty.

light curves, allowed us to obtain an accurate model of the system using PHOEBE.
The system is composed of a hot, massive and luminous primary star of an early
B type, and a more evolved, but less massive and luminous secondary, implying an
Algol-type system that underwent a mass transfer episode.
Further modeling of the binary shows that the light curve fit is greatly im-
proved when adding a large cool region on the secondary, which is attributed
to the reflection effect. The physical parameters of the system derived from the
non-spotted and cool spot models are consistent within their errors. We derive
masses of 21.04 ± 0.34 M ⊙ and 7.61 ± 0.09 M ⊙ and radii of 9.93 ± 0.06 R ⊙ and
9.18 ± 0.04 R ⊙ , for the primary and the secondary component, respectively.
The < 2% accuracy of the mass and radius measurements make OGLE-LMC-
ECL-09937 an important contribution to the sparsely populated high-mass end of
the stellar mass distribution, and an interesting object for stellar evolution studies,
as a potential progenitor of a binary system composed of two neutron stars.
14 A. A.

Acknowledgements. We thank Kris Stanek for initiating and supporting this

project. D.M.S and M.K. are supported by the Polish Ministry of Science and
Higher Education under the grant “Iuventus Plus” No. 0420/IP3/2015/73. D.M.S.
is supported by the Polish National Science Center (NCN) under the grant no.
2013/11/D/ST9/03445 and the NCN grant MAESTRO 2014/14/A/ST9/00121 to
Andrzej Udalski. Support for J.L.P. is provided in part by FONDECYT through
the grant 1151445 and by the Ministry of Economy, Development, and Tourism’s
Millennium Science Initiative through grant IC120009, awarded to The Millennium
Institute of Astrophysics, MAS.

REFERENCES

Bernstein, R., Shectman, S. A., Gunnels, S. M., Mochnacki, S., Athey, A. E. 2003, SPIE, 4841, 1694.
Bonanos, A. Z., et al. 2006, ApJ, 652, 313.
Bonanos, A. Z. 2009, ApJ, 691, 407.
Bonanos, A. Z., et al. 2009, AJ, 138, 1003.
Budding, E., Erdem, A., Çiçek, C., Bulut, I., Soydugan, F., Soydugan, E., Bakiş, V., Demircan, O.
2004, A&A, 417, 263.
Choi, J., Dotter, A., Conroy, C., Cantiello, M., Paxton, B., Johnson, B. 2016, ApJ, 823, 102.
Choudhury, S., Subramaniam, A., Cole, A. 2016, MNRAS, 455, 1855.
Conti, P. 1971, ApJ, 170, 325.
Cowley, A. P., Hutchings, J. B. 1976, PASP, 88, 456.
Dotter, A. 2016, ApJ, 222, 8.
Giuricin, G., Mardirossian, F., Mezzetti, M. 1983, ApJS, 52, 35.
Graczyk, D., et al. 2011, Acta Astron., 61, 103.
Grundstrom, E. D., Gies, D. R., Hillwig, T. C., McSwain, M. V., Smith, N., Gehrz, R. D., Stahl, O.,
Kaufer, A. 2007, ApJ, 667, 505.
Hubeny, I., Lanz, T. 1995, ApJ, 439, 875.
Kochanek, C. S. et al. 2017, PASP, 129, 4502.
Massey, P., Lang, C. C., Degioia-Eastwood, K., Garmany, C. D. 1995, ApJ, 438, 188.
Pawlak, M., et al. 2016, Acta Astron., 66, 421.
Paxton, B., Bildsten, L., Dotter, A., Herwig, F., Lesaffre, P., Timmes, F. 2011, ApJ, 192, 3.
Peters, G. J. 2001, ASSL, 264, 79.
Pojmański, G. 1997, Acta Astron., 47, 467.
Pojmański, G. 2002, Acta Astron., 52, 397.
Prša, A., and Zwitter, T. 2005, ApJ, 628, 426.
Puls, J., Urbaneja, M. A., Venero, R., Repolust, T., Springmann, U., Jokuthy, A., Mokiem, M. R.
2005, A&A, 435, 669.
Pych, W. 2004, PASP, 116, 148.
Sana, H. et al. 2012, Sci, 337, 444.
Santolaya-Rey, A. E., Puls, J., Herrero, A. 1997, A&A, 323, 488.
Schwarzenberg-Czerny, A. 1996, ApJ, 460, L107.
Shappee, B. J. et al. 2014, ApJ, 788, 48.
Skulskii, M. Y. 1992, SvA, 36, 411.
Southworth, J. 2015, ASPC, 496, 164.
Szczygieł, D. M., Stanek, K. Z., Bonanos, A. Z., Pojmański, G., Pilecki, B., Prieto, J. L. 2010, AJ,
140, 14.
Tody, D. 1986, SPIE, 627, 733.
Tody, D. 1993, ASPC, 52, 173.
Toma, K., Yoon, S-C., Bromm, V. 2016, SSRv, 202, 159.
Vol. 66 15

Torres, G., Andersen, J., Giménez, A. 2010, AARv, 18, 67.

Udalski, A. 2003, Acta Astron., 53, 291.
Udalski, A., Szymański, M. K., Soszyński, I., Poleski, R. 2008, Acta Astron., 58, 69.
Udalski, A., Szymański, M. K., and Szymański, G. 2015, Acta Astron., 65, 1.
van Hamme, W. 1993, AJ, 106, 2096.
Walborn, N. R., Fitzpatrick, E. L. 1990, PASP, 102, 379.
Williamon, R. et al. 2016, PASP, 128, 4202.
Wilson, R. E., Devinney, E. J. 1971, ApJ, 166, 605.
van Hamme, W. 1993, AJ, 106, 2096.