Long-period eclipsing binaries: towards the true

mass-luminosity relation. II. Absolute parameters of the

NN Del system
Alexei Kniazev1,2,3
arXiv:2010.05551v1 [astro-ph.SR] 12 Oct 2020

Abstract We present results of our study of the long- 1 Introduction

period eclipsing binary star NN Delphini (hereafter
NN Del). The results are based on spectral data ob- The ratio between the mass of a star and its luminos-
tained with the HRS échelle spectrograph of the South- ity on the main sequence (mass-luminosity ratio, MLR
ern African Large Telescope (SALT). Our constructed hereafter) is a fundamental law used in various fields of
velocity curve is based on 19 spectra obtained between astrophysics. It is particularly important to restore the
2017 and 2019 years and covers all phases of the binary’s initial mass function (IMF) using the luminosity func-
orbit. The orbital period, P = 99.252 days, was deter- tion (LF) of stars, and it should be noted that the IMF
mined from our spectral data and coincides with the can only be obtained by this method, that is, through
period determined in previous studies, as well as the the LF and MLR.
system eccentricity of e = 0.517. Calculated velocity Independent determination of the mass of a star and
amplitudes of both components allow us to determine its luminosity is only possible for binary system com-
the masses of both system components M1 = 1.320 M ponents of certain types. One of these types are vi-
and M2 = 1.433 M with the accuracy about of one sual binary stars with known orbital parameters and
percent (0.8% and 1.1%), respectively. Luminosities of trigonometric parallaxes (Fernandes et al. 1998; Malkov
both components are presented as L1 = 4.164 L and et al. 2012; Docobo et al. 2016). Usually, these stars
L2 = 6.221 L , and the effective temperatures of both are wide pairs, whose components do not interact with
components were directly evaluated (Teff1 = 6545 K each other and, in an evolutionary sense, are similar to
and Teff2 = 6190 K) together with the metallicity of the single stars. Another main source of independent mass
system [Fe/H] = −0.19 dex and its color excess E(B- determinations are detached eclipsing binary stars with
V)=0.026 mag. Comparison with evolutionary tracks components on the main sequence, whose spectra show
shows that the system age is 2.25 ± 0.19 Gyr, and both lines of both components (double-lined eclipsing bina-
components are on the main sequence and have not yet ries, DLEB; Popper 1980; Andersen 1991; Gorda and
passed the turn point. Spectral type is F5V for the Svechnikov 1998; Kovaleva 2001; Torres et al. 2010).
hotter component and F8V for another one. These are mainly close pairs whose component’s spins
are synchronized by tidal interaction, and they evolve
Keywords stars: luminosity function, mass function slightly different compare to single stars due to the
— stars: binaries: spectroscopic — stars: individual slowdown of rotation. At the same time, for masses
(NN Del) M/M > 2.7, MLR is based exclusively on the data
obtained for DLEBs, and it is used for studies of sin-
Alexei Kniazev gle stars. Apparently, such use of MLR could be not
1 South African Astronomical Observatory, PO Box 9, 7935 Ob- very correct and leads to systematically wrong results
servatory, Cape Town,South Africa. when evaluating the characteristics of stars in the dis-
2 SouthernAfrican Large Telescope Foundation, PO Box 9, 7935 cussed mass range, as well as when restoring the ini-
Observatory, Cape Town, South Africa. tial mass function with its help. For example, Malkov
3 Sternberg Astronomical Institute, Lomonosov Moscow State (2003) compared the radii of DLEBs and single stars
University, Moscow, Russia. and found a noticeable difference between the obser-
vation parameters of the B0V-G0V components of the
DLEBs and single stars of similar spectral classes.
2

Table 1 Parameters of NN Del system collected from the literature.

Parameter GF (2003)a G(2014)b S(2018)c G(2019)d
Orbital period P (d) 99.2684±0.0005 99.244±0.019 99.26849±0.00015 99.2690±0.0009
Eccentricity e 0.51759±0.00002 0.5168±0.0029 0.51944±0.00055 0.5197±0.0004
Rad. vel. semi-amplitude K1 (km s−1 ) – 39.45±0.27 39.62±0.15 39.407±0.037
Rad. vel. semi-amplitude K2 (km s−1 ) – 36.04±0.21 36.22±0.11 36.191±0.023
Systemic heliocentric vel. γ (km s−1 ) – −8.19±0.08 −9.405±0.050 −9.485±0.015
Inclination i (degrees) 89.488±0.003 90.0 (fixed) 89.6342±0.0076 89.90±0.11
The longitude of the periastron ω (degrees) 171.710±0.005 170.0±0.4 169.75±0.45 170.00±0.07
R.m.s. residuals of Keplerian fit (km s−1 ) – 0.56 0.11 0.10 (0.18)
m1 (M ) (Component A) – 1.328±0.021 1.337±0.011 1.3266±0.0021
m2 (M ) (Component B) – 1.454±0.025 1.462±0.013 1.4445±0.0029
r1 0.01584±0.00006 – 0.01260±0.00015 –
r2 0.0153 – 0.01720±0.00015 –
Teff (K) (Component A) 6437±5 – – –
Teff (K) (Component B) 6500.0 – – –
log g (dex) (Component A) – – 4.1532±0.0051 –
log g (dex) (Component B) – – 3.9226±0.0035 –
Radius R (R ) (Component A) 2.2 – 1.604±0.014 –
Radius R (R ) (Component B) 2.2 – 2.188±0.015 –
v sin i (km s−1 ) (Component A) – 7.7±0.6 – –
v sin i (km s−1 ) (Component B) – 7.4±0.3 – –
Spectral Type F5 IV (both) F8 IV-V (both) – –
a
– Gómez-Forrellad et al. (2003); b – Griffin (2014); c – Sybilski et al. (2018); d – Gallenne et al. (2019)

Malkov (2007) collected data on fundamental param- 2 Previous studies of NN Del

eters of 19 components of long-period DLEB. These
stars presumably have not undergone synchronization The system NN Del (HD 197952, HIP 102545) was dis-
of rotation with the orbital period and therefore ro- covered as variable based on Hipparcos satellite data
tate rapidly and evolve similarly to single stars. Possi- (Makarov et al. 1994), but the only information pub-
bly, only such kind of data should be used to construct lished was the star changing its brightness by more
relations (in particular, MLR) for “isolated” stars for than half of a magnitude. The brightness curve for
masses M/M > 2.7. The masses of the components the NN Del system was obtained and thoroughly stud-
of other types of binary stars (orbital, resolved spec- ied by Gómez-Forrellad et al. (2003). They concluded
tral binaries) rarely exceed this limit. Note that of the that both components of the system are almost equal to
19 DLEB stars mentioned above, only 13 have a mass each other, the period of the system is 99.268 days and
greater than 2.7 M/M , which is clearly insufficient the eccentricity of the system is 0.5176, i.e. the stellar
to draw any definite conclusions. For this reason, we components have non-circular orbits. Gómez-Forrellad
have begun a systematic study of long-period DLEB et al. (2003) also determined the types of each star as
stars to obtain their mass and luminosity parameters F5, based on photometric data, and since the radii of
and to compare these parameters with those obtained both components were calculated to be 2.2 R , which
for short-period DLEB stars. The definition of the test meant that both stars had already left the main se-
sample as well as a brief description of the software quence, their luminosity class was defined as IV. Table 1
package fbs created to analyze spectra of binary stars is shows all parameters and errors, for NN Del that were
described in our first paper (Kniazev et al. 2020). Here, defined by Gómez-Forrellad et al. (2003) and following
we present the first results obtained for the long-period works, where this binary system was studied.
eclipsing binary star NN Del, which belongs to our test Griffin (2014) carried out spectroscopy of NN Del us-
sample and has some previously published studies to ing a photoelectric spectrometer designed to measure
help evaluate the accuracy of the parameters based on the radial velocities of stars (Griffin 1967). It should
our spectral data. Hereafter we call the brighter star in be noted here that the spectra were not obtained, but
the DLEB system NN Del component A. the velocities of both components were determined. In
total, 37 measurements were made during three years,
and for the first time a velocity curve was constructed
and masses of both components were determined. The
3

Table 2 HRS observations for NN Del and found heliocentric velocities

Date MJD Exp.time Vhel1 Vhel2
(day) (Sec) (km/sec) (km/sec)
(1) (2) (3) (4) (5)
20170504 2457878.65465 315. 43.528±0.012 -68.014±0.025
20170508 2457882.63785 315. 38.422±0.012 -61.257±0.026
20170520 2457894.64881 400. -2.161±0.045 -16.526±0.041
20170805 2457971.46249 315. 12.869±0.013 -34.007±0.034
20170814 2457980.36746 450. 42.930±0.016 -66.858±0.029
20170830 2457996.37722 315. -7.087±0.023 -12.431±0.039
20170927 2458024.28117 315. -25.935±0.017 8.459±0.021
20171003 2458030.27510 315. -26.714±0.017 9.568±0.024
20171024 2458051.23722 240. -23.874±0.019 5.794±0.035
20180428 2458237.66367 315. -28.411±0.029 8.636±0.056
20180429 2458238.65980 315. -27.308±0.014 8.903±0.031
20180926 2458388.33887 315. 3.222±0.014 -23.580±0.024
20180930 2458392.23932 315. -5.554±0.016 -15.491±0.031
20181007 2458399.29359 315. -14.738±0.016 -5.158±0.032
20181015 2458407.27283 315. -21.292±0.015 2.574±0.027
20190728 2458693.50237 315. -10.364±0.041 -8.896±0.070
20190801 2458697.46381 315. -14.558±0.016 -4.101±0.031
20190810 2458706.45261 315. -21.766±0.016 3.275±0.030
20190818 2458714.44764 315. -24.539±0.014 7.221±0.030

final accuracy of the velocity curve (scattering of resid- ×10−12

uals after model subtraction) was 0.56 km sec−1 . The 1.6 NN Del

obtained values of the orbital period P and eccentric- 1.4

ity e agreed with numbers found from photometric data

Relative intensity (counts)

1.2

(Gómez-Forrellad et al. 2003), where the inclination i of 1.0

the system was assumed to be 90 degrees, but the longi-
0.8
tude ω of the periastron w was different at the 4σ level,
0.6
although Griffin (2014) noted the unbelievable high ac-
curacy of all the parameters given by Gómez-Forrellad 0.4

et al. (2003). 0.2

Sybilski et al. (2018) carried out spectral observa- 0.0

tions of NN Del using the échelle spectrograph CHI- 4000 5000 6000 7000 8000 9000
Wavelength (Å)
RON (R ∼ 80000) installed on the 1.5m telescope of
CTIO observatory (Chile). Only seven observations
Fig. 1 An example of a fully processed spectrum of
were taken, but the spectroscopy was combined with
NN Del. The spectrum consists of 70 échelle orders from
photometry from Gómez-Forrellad et al. (2003) to con- both blue and red arms merged together and corrected for
struct both the velocity and the brightness curves si- sensitivity.
multaneously. The obtained masses of both compo-
nents of the system agreed with those obtained by
Griffin (2014) taking into account errors. The accu- components from the obtained spectra, and presented
racy of the velocity curve was 0.11 km sec−1 that is only the ratio of the luminosities of the components as
much higher compare to Griffin (2014). The use of LB /LA = 1.6564 ± 0.0076.
spectral and photometric data together made it pos- Gallenne et al. (2019) is the most recent work in
sible to determine masses of both components and which the characteristics of the NN Del system were de-
their radii, although the authors noted that defining fined. In that paper spectral data were obtained with
all these characteristics was not the main purpose of the échelle spectrograph STELLA (R ∼ 55000) and
their work. For the same reason, the authors did not were used along with interferometric data obtained with
try to estimate temperatures and luminosities of both VLTI/PIONIER. 40 échelle spectra were taken during
4

2014–2018. The simultaneous use of spectral and astro- NN Del, P=99.27 days
metric data allowed the determination of the masses of
8.4
both components and the distance to NN Del. The ac-
curacy of the velocity curve was 0.10 km sec−1 for com-
ponent A and 0.18 km sec−1 for component B, which 8.5

Magnitude (mag)
made it possible to determine the masses of the com-
ponents with an accuracy of 0.18%, and the distance 8.6

to the system NN Del with an accuracy of 0.4%. It

was noted that the resulting parallax is greater than 8.7

the parallax in Gaia DR2 (Gaia Collaboration 2018),

but agrees with the estimate obtained by Hipparcos. 8.8

Further, Gallenne et al. (2019) tried to evaluate the evo-

lutionary status of both components of NN Del. They −0.2 0.0 0.2 0.4
Phase
0.6 0.8 1.0 1.2

used their own mass estimates of both components, ob-

tained with high accuracy and adopted temperatures, Fig. 2 Photometric data from the ASAS survey converted
fractional radii and luminosities for both components of to the period P=99.27 days. There are only few points
NN Del. Finally, after analysis of all available data, it that can indicate the shape of the primary and secondary
was concluded that both components of the NN Del sys- minima.
tem are on the main sequence and have not yet passed
the turn point. Each HRS spectrum of NN Del was additionally cor-
rected for bad columns and pixels and was also cor-
rected for the spectral sensitivity curve obtained clos-
3 Spectral observations and data reduction est to the date of the observation. Spectrophotometric
standards for HRS are observed once a week as part
Spectral observations of NN Del were made during of the HRS Calibration Plan. An example of a fully
2017–2019 at the Southern African Large Telescope processed spectrum of NN Del, which was used in fur-
(SALT; Buckley et al. 2006; O’Donoghue et al. 2006) ther analysis, is shown in Figure 1. The spectrum con-
using the fiber échelle spectrograph HRS (Barnes et al. sists of 70 échelle orders from both the blue and the
2008; Bramall et al. 2010, 2012; Crause et al. 2014). red arm merged together and corrected for sensitivity.
HRS is a thermostabilised dual-beam échelle spectro- Unfortunately, SALT is a telescope where the unfilled
graph. The blue arm of the spectrograph covers the entrance pupil of the telescope moves during the ob-
spectral range 3900–5550 Å, and the red arm covers servation. For that reason, absolute flux calibration is
the spectral range 5550–8900 Å, respectively. HRS is not feasible with SALT. However, since all optical el-
equipped with four pairs of fibers (object fiber and sky ements are always the same, relative flux calibration
fiber) and can be used in low (LR), medium (MR), high can be used for SALT data. Additionally, since HRS
resolution (HR) and high stability modes. All spec- is a fiber-fed échelle spectrograph, the relative distribu-
tral observations of NN Del were taken with MR mode tion of energy could be done with high enough accuracy
(R=36 500–39 000), with fibers with a diameter of 2.23 (Kniazev 2020; in preparation).
arcseconds. Both the blue arm and the red arm CCDs
were used with binning 1×1. The dates of observations,
their Julian days, and exposure times are presented in 4 Available photometric data
Table 2.
Since HRS is the vacuum échelle spectrograph that is Unfortunately, there are no good photometric data for
installed inside a temperature-controlled enclosure, all NN Del among all existing public surveys. The best
standard calibrations are performed once a week, which available data are from the ASAS survey (Pojmanski
is enough to achieve an accuracy 300 m/sec for MR 1997) as shown in Figure 2. It is obvious, that even
mode. The primary HRS data reduction was done au- these data have too few points outlining the positions
tomatically using the standard SALT pipeline described and shapes of the narrow primary and secondary min-
by Crawford et al. (2010). The following échelle data ima and it is impossible to use these data for any mod-
reduction was performed using the standard HRS data eling. Therefore, it was decided to use our spectral data
pipeline described in detail by Kniazev et al. (2016, together with the results of the analysis of photometric
2019). data from Gómez-Forrellad et al. (2003), which were
also used by Sybilski et al. (2018).
5

5 Spectral data analysis to use the parameters derived by Sybilski et al. (2018).
(2) The presence of known temperature-log g degener-
To investigate the fully processed HRS spectra we used ation for stars colder than about 7000 K. With the full
fbs (Fitting Binary Stars) software specially devel- set of spectral and photometric data in hands, this de-
oped by our group (Katkov et al. 2020 in prepara- generation can be avoided.
tion) and partially described in Kniazev et al. (2020). Our iterative procedure was as follows:
This program uses a library of theoretically calculated
1. During the first iteration, the program fbs was used
high-resolution stellar spectra and is designed to de-
only with one limitation – it was assumed that both
termine radial velocities and stellar parameters (Teff ,
log g, v sin i, [Fe/H]) for both components of the bi- components had the same metallicity ([F e/H]1 ≡
nary system, and parameters E(B-V) and W1,2 that [F e/H]2 ; (Hawkins et al. 2020)). All 19 obtained
describe the reddening of both spectra and the con- spectra were fitted simultaneously and 19 veloci-
tribution of each component to the observed spectrum ties were obtained for both components together
at the wavelength of 5500 Å that is very close to the with the parameters of both components (Teff , log g,
effective wavelength for the V-filter. By definition, v sin i, [Fe/H])1,2 . The result of the analysis of a sin-
W1 + W2 = 1. The program simultaneously approx- gle spectrum is shown as an example in Figure 3.
imates the observed spectrum by a model, which is ob- The upper panel of the figure shows the result of the
tained by interpolating over the grid of stellar models fit in the spectral range 4000-5300 Å, and the bot-
and convolves it with a function that takes into account tom panel shows the result of the same fit in the Hβ
broadening caused by the rotation of a star (v sin i) line region. In each panel, the black and red lines
with a shift corresponding to a specific value of the correspond to the observed spectrum and its model.
radial velocity at a given epoch. In case of a binary The blue and orange lines refer to the model spectra
star, the fitting routine uses two model spectra for the of the first and second component, respectively. The
components, each with its own radial velocity and stel- bottom part of each panel shows the difference be-
lar atmosphere parameters, and as result the observed tween observed spectrum and its model with a black
spectrum is decomposed into separate spectra of two line, where blue lines indicate errors in the observed
components. If there are several spectra of the binary spectrum.
system obtained at different epochs, the program finds 2. Based on the obtained velocities and their errors,
a solution, in which the parameters (Teff , log g, v sin i, the radial velocity curves for both components were
[Fe/H], W)1,2 and E(B-V) satisfy all spectra that can constructed using the program fbs, and the orbital
be fitted simultaneously, and velocities of both com- parameters of the system and their errors were calcu-
ponents V1,2j
are determined for each particular epoch lated. Parameters such as the orbital period P , ve-
j. The present study uses theoretical stellar models locity semi-amplitudes of both components K1 and
from Coelho (2014). K2, system eccentricity e, system heliocentric ve-
locity γ and the longitude of the periastron ω were
obtained.
6 Results of our study of NN Del 3. Using previously determined P , e, K1 and K2 and
the inclination angle i (taken from research by Sybil-
The measured heliocentric velocities for both compo- ski et al. 2018), the masses of M1 and M2 of both
nents of the system NN Del and their errors are pre- components and their errors were determined by the
sented for each epoch in Table 2. To determine the equations:
velocities, as well as the parameters of the components
and orbital parameters of the binary system NN Del, we M1 = 1.036149 10−7 (K1 + K2 )2 K2 P e3/2 sin−3 (i)
used the iterative procedure described below. The ne- M2 = 1.036149 10−7 (K1 + K2 )2 K1 P e3/2 sin−3 (i)
cessity of this iterative procedure was due to two facts:
(1)
(1) We used only spectral observations, but to obtain
the complete set of parameters to calculate masses and Here M1,2 is in M , K1,2 is in km/s, P is in day.
luminosity, a simultaneous analysis of spectral and pho- 4. Next the absolute sizes of the major and minor axes
tometric data is necessary, as was done, for example, by of the NN Del system were calculated as:
Sybilski et al. (2018). Potentially, masses for both com-
ponents can be obtained by analyzing simultaneously a1 = 0.01976569 K1 P e1/2 sin−1 (i)
spectral and astrometric data, as was done by Gallenne (2)
a2 = 0.01976569 K2 P e1/2 sin−1 (i)
et al. (2019), but for the further analysis of the evolu-
tionary status of NN Del these authors were still forced
6

1.5

1.0
Relative intencity

0.5

0.0

−0.5
4000 4200 4400 4600 4800 5000 5200
Wavelength(Å)

1.50

1.25

1.00
Relative intencity

0.75

0.50

0.25

0.00

4820 4830 4840 4850 4860 4870 4880 4890 4900

Wavelength(Å)

Fig. 3 The results of the analysis of one spectrum of NN Del obtained with HRS. The upper panel shows the result of the
fit in the spectral region 4000-5300 Å, and the bottom panel shows a small spectral region near the Hβ line. In each panel,
the black and red lines correspond to the observed spectrum and its model, respectively. The orange and blue lines refer
to the model spectra of the first and second component, respectively. The bottom part of each panel shows the difference
between observed spectrum and its model with a black line, where blue lines indicate errors in the observed spectrum.
7

25
Velocity, km s−1

−25

−50
Residuals, km s−1

0.0 0.2 0.4 0.6 0.8 1.0

1
0
−1
0.0 0.2 0.4 0.6 0.8 1.0
Phase

Fig. 4 The final radial velocity curve for NN Del as the function of phase. The fitted parameters are shown in Table 4.
The model is constructed by the points listed in Table 2. The velocity curve of component A is shown in orange, and the
velocity curve of component B is shown in blue. The differences between the model and the observed points are shown in
the bottom panel. The final accuracy of the velocity curve (residuals after model subtraction) is 0.494 km s−1 , what is
close to the nominal accuracy of 0.300 km s−1 for HRS spectra obtained in MR mode (Kniazev et al. 2019).

25
Velocity, km s−1

−25

−50
Residuals, km s−1

2457800 2458000 2458200 2458400 2458600 2458800

0
−1
2457800 2458000 2458200 2458400 2458600 2458800
MJD

Fig. 5 The final radial velocity curve for NN Del as the function of Julian date. Designations are the same as in Figure 4.
From this figure one can see that the observations were made during ten periods of NN Del.
8

as well as the radii R1 and R2 for both components: where γ is the systemic heliocentric velocity of the
NN Del system.
Unstable solutions far from q = 1 were discarded
R1 = r1 ∗ (a1 + a2 )
(3) (approximately 30% of decisions) iteratively with a clip-
R2 = r2 ∗ (a1 + a2 ) ping level of 2.5σ, and all other solutions were used to
calculate the average and its error for each of the pa-
where the relative radii (r1 /r2 ) for both components
rameters (Teff , v sin i, [Fe/H])1,2 and E(B-V). The final
were taken from Sybilski et al. (2018), since they
calculated errors are given in Table 3. The calculated
used both photometric and spectral data simultane-
radial velocity curves as a function of phase and Julian
ously, and their relative radii are more correct com-
date are shown in Figures 4 and 5, respectively. The fi-
pare to the ones from Gómez-Forrellad et al. (2003).
nal accuracy of the velocity curve (scatter of the points
Here a1,2 and R1,2 is in R .
after model subtraction) was 0.406 km sec−1 , which is
5. Finally, log g1 and log g2 were recalculated using:
close to the nominal accuracy of 0.300 km s−1 for HRS
log g1 = 4.438068 + log M1 − 2 log R1 spectra obtained in MR mode (Kniazev et al. 2019).
(4) The values for the orbital parameters of the binary star
log g2 = 4.438068 + log M2 − 2 log R2
NN Del and their errors are given in Table 4.
Numerical coeeficients of Eqs. (1,2,4) are taken from With all the values of the physical parameters of the
Prša et al. (2016) and based on the IAU 2015 Res- NN Del system and their errors, as well as the orbital
olution B3 on Recommended Nominal Conversion parameters of the system, the masses and radii of both
Constants for Selected Solar and Planetary Proper- components were calculated by Eq.1–3. Using the ob-
ties (Mamajek et al. 2015). tained radii and temperatures, the luminosity of each
If the new values were different from the previ- of the components was then calculated as:
ous values by more than 0.01 dex, all step were re-  4
peated, with the new, fixed, values for log g1 and Teff
L(spec) = R2 , (6)
log g2 . When the difference between the old and new 5780
log g1 and log g2 was less of 0.01 dex, the procedure where Teff of the Sun is 5780 K. The obtained abso-
was stopped and all calculated parameters were de- lute parameters of the NN Del system are presented in
clared as the final ones. The final heliocentric veloc- Table 5.
ities for both components are presented in Table 2
with their errors.
Since the output errors of the fbs program are fit-
ting errors, which are often underestimated, errors of
the output parameters need to be estimated in more
realistic way. A statistical evaluation was used, where
all possible pairs of the observed spectra were taken, pa-
rameters log g1,2 were fixed and these pairs were fitted
by the program. With 19 observed spectra 171 differ-
ent pairs were fitted. After that, for each solution, the
parameter q (mass ratio of the components) was calcu-
lated as:

q = −(V1 − γ)/(V2 − γ), (5)

Fig. 6 Comparison of our mass measurement for both com-
ponents of NN Del with previously published results (see
Table 3 Parameters for both stellar components. Table 1 for more details).

Parameter Component A Component B

Teff (K) 6545±180 6190±85
log Teff (K) 3.82±0.02 3.79±0.01
7 Discussion
log g (dex) 4.16±0.01 3.92±0.01
v sin i (km s−1 ) 1.15±0.13 0.36±0.28
7.1 Comparison of parameters for the NN Del system
W (at 5550 Å) 0.378±0.009 0.622±0.009
[Fe/H] (dex) −0.19 ± 0.05 To evaluate the correctness of our orbital parameters
E(B-V) 0.026 ± 0.002 calculation program, we fitted the 40 radial velocities
9

Table 4 Best-fit orbital elements.

Parameter Value %
Epoch at radial velocity maximum T0 (d) 2457779.76 ± 0.06 0.00
Orbital period P (d) 99.252±0.024 0.02
Eccentricity e 0.517±0.002 0.38
Radial velocity semi-amplitude K1 (km s−1 ) 39.191±0.179 0.46
Radial velocity semi-amplitude K2 (km s−1 ) 36.101±0.099 0.27
Systemic heliocentric velocity γ (km s−1 ) −9.692±0.066 0.68
The longitude of the periastron ω (degrees) 170.307±0.532 0.31
Root-mean-square residuals of Keplerian fit (km s−1 ) 0.406 –

of the system NN Del given in Table A.2 of Gallenne and L2 (phot) = 4.28±0.15L , which agree with the lu-
et al. (2019) and obtained full agreement with the or- minosities L1,2 (spec) presented in Table 5. From these
bital elements presented in Gallenne et al. (2019) up to photometric luminosities it is possible to calculate pho-
the final accuracy of the velocity curves: 0.10 km sec−1 tometric radii using Eq.6 and temperatures of the com-
for the component A and 0.18 km sec−1 for the compo- ponents from Table 3 as R2 (phot) = 2.31 ± 0.07R and
nent B. R1 (phot) = 1.61 ± 0.07R . As can be seen from the
It was obvious from the beginning, that since we use comparison of luminosities and radii, given in Table 5,
échelle spectra with over twice lower resolution compare both spectral and photometric luminosities and radii of
of the CHIRON échelle spectrograph (R ∼ 80000) used the component A agree with a difference of only 0.3σ
by Sybilski et al. (2018) and about one and a half times and 0.5σ. The spectral and photometric luminosities
lowere than the resolution of the STELLA échelle spec- and radii of the component B are slightly less agree-
trograph (R ∼ 55000) used by Gallenne et al. (2019), able at about 2.0σ and 2.2σ, respectively, but are still
our accuracy of the obtained parameters could be worse close to each other. Such a comparison of absolute pa-
or comparable to the accuracy obtained by those au- rameters, counted in a different way and matched to
thors. Figure 6 shows a comparison of our mass esti- each other, is a good indicator of the correctness of the
mates for both components of NN Del with previously absolute parameters found for the NN Del system.
published results. It can be seen that our mass esti- We can also independently estimate the distance to
mates do not show any systematics and are in very good NN Del and compare it with the distance from Gaia
agreement with the most accurate values obtained by
DR2 (Gaia Collaboration 2018). Our estimation of
Gallenne et al. (2019), with differences not exceeding
the distance could be done with use of the known to-
1σ of the total error. That also means that the radial
tal Vtot magnitude, found extinction shown in Table 3
velocity semi-amplitudes determined by us do not show
and an absolute magnitude M(Vtot ). The M(Vtot )
any systematics and positively characterize the quality
could be found as the sum of absolute magnitudes for
of HRS data.
both components M(V1,2 ), where the absolute mag-
With found absolute parameters of the system
nitude for each component could be calculated with
NN Del, we can try to check their correctness. First,
use of luminosities based on the spectroscopic radii
using the distance to NN Del according to Gaia DR2
(the parallax 5.6393 ± 0.0636 is translated to a distance L(spec) from Table 5 and temperatures Teff from Ta-
177.33 ± 2.00 parsec; Gaia Collaboration 2018) as well ble 3 and bolometric corrections calculated with use of
as the color excess E(B − V ) resulting from our study, (Straižys 1992). The final result is also shown in Ta-
we can estimate the absolute magnitude of NN Del as ble 5. Our estimated distance is smaller than the dis-
M(V) = 2.08 ± 0.03 mag. Since we know from our spec- tance according to Gaia and this result is in the agree-
tral data the contributions (parameter W in Table 3) ment with the result obtained by Gallenne et al. (2019)
of each component at the equivalent wavelength of the as d = 167.99 ± 0.65 pc. However, we have to admit
V-band, we can separate the resulting absolute mag- here that the large error of our assessment means that
nitude into contributions of each component. Taking difference we found is not significant with a difference
into account the bolometric corrections from (Straižys of only 1.62σ.
1992) for the Teff , presented in Table 3, we are then In the Torres et al. (2010) review the data on lumi-
able to calculate their luminosities (photometric lumi- nosities and masses of 190 stars, from 95 DLEB sys-
nosities hereafter; L1,2 (phot)) and compare these lumi- tems, have been collected. A comparison of our ob-
nosities with the spectral luminosities (L1,2 (spec)) ob- tained characteristics for NN Del with data from Torres
tained by Eq.6. We obtain L2 (phot) = 7.05 ± 0.20L et al. (2010), as presented in Figure 7, shows that our
10

Table 5 The absolute parameters for NN Del

Parameter Value %
M1 (M ) 1.320±0.011 0.83
M2 (M ) 1.433±0.015 1.06
log M1 0.121±0.004 –
log M2 0.156±0.005 –
a1 (R ) 65.813±0.315 0.48
a2 (R ) 60.624±0.188 0.31
R1 (spec) (R ) 1.594±0.016 1.00
R2 (spec) (R ) 2.176±0.018 0.82
log R1 (spec) 0.203±0.004 –
log R2 (spec) 0.338±0.004 –
R1 (phot) (R ) 1.614±0.070 4.35
R2 (phot) (R ) 2.314±0.071 3.06
log R1 (phot) 0.208±0.019 –
log R2 (phot) 0.364±0.013 –
L1 (spec) (L ) 4.164±0.346 8.31
L2 (spec) (L ) 6.221±0.371 5.96
log L1 (spec) 0.620±0.036 –
log L2 (spec) 0.794±0.026 –
L1 (phot) (L ) 4.280±0.151 3.53
L2 (phot) (L ) 7.051±0.205 2.90
log L1 (phot) 0.631±0.015 –
log L2 (phot) 0.848±0.013 –
Distance (pc) 169.71±4.27 2.52
Age (Gyr) 2.25±0.19 8.44
Spectral Type (A) F5 –
Spectral Type (B) F8 –

characteristics of the NN Del components are fully con-

sistent with the masses, luminosities and radii of the
stars in this large sample.
Fig. 7 Diagram for all stars from Torres et al. (2010)
with position of both components of NN Del. Component
7.2 Evolutionary state A shown in green and component B in red. Sizes of the
symbols are significantly larger than the errors.
Since NN Del is a detached system where components
do not influence each other’s evolution, it is reasonable
to assume that evolutionary status of each component the components on these tracks, varying the metallicity
can be assessed based on evolutionary models of single [Fe/H] from -0.2 to 0.2 dex in increments of 0.05 dex.
stars. Since Gallenne et al. (2019) showed that results We then searched for the minimum of the equation:
from the models PARSEC (PAdova and TRieste Stellar  !2 
2 2 2 2
Evolution Code, Bressan et al. 2012), BaSTI (Bag of
  
 ∆L ∆Teff ∆R ∆g
X
2
χ = + + + , (7)
Stellar Tracks and Isochrones, Pietrinferni et al. 2004) i=1
σL i σTeff σR i σg i
i
and MIST (MESA Isochrones and Stellar Tracks, Choi
where the summation is done on both components
et al. 2016) agree well, we only used the MIST model in
(i = 1, 2), the symbol ∆ shows the logarithmic differ-
our work. Since the masses of the NN Del components
ence between the model and the value obtained from
are known to us with the highest degree of accuracy, we
the observations, and σ is also used in the logarithmic
retrieved from MIST1 the evolutionary tracks for stars
scale. The search was performed for each model metal-
of these masses and considered the positions of each of
licity under the assumption that both components of
the system have the same metallicity.
1 http://waps.cfa.harvard.edu/MIST/interp isos.html
11

Figure 8 shows the log L–log Teff plot for two solu-
tions: (1) for metallicity [Fe/H] = −0.19 dex, which is
derived from our star models (see Section 6) and (2) for
metallicity [Fe/H] = 0.05 dex, which showed the best
χ2 from Eq.7. For metallicity [Fe/H] = −0.19 dex, the
age of the system NN Del is estimated to be 2.06 Gyr,
and for [Fe/H] = 0.05 dex it is 2.44 Gyr. In both cases,
as can be seen in Figure 8, both components have not
yet passed the turn point and are on the main sequence,
which coincides with the conclusion by Gallenne et al.
(2019).
It is possibly to explain the resulting difference in
metallicities by the systematic shift between the metal-
licity scales of the stellar models from Coelho (2014)
and the MIST models. We therefore took stellar models
from phoenix Husser et al. (2013) and recalculated our
data according to the methodology described in Sec-
tions 5 and 6, and resulted the value [Fe/H] = −0.20 dex
as well. We would like to note that the phoenix mod-
els match the observed spectra less well and therefore
we prefer to work with the models from Coelho (2014).
As the final result, we adopt the age assessment for the
NN Del system as tavg = 2.25 ± 0.19 Gyr, which also
agrees with the output from Gallenne et al. (2019).
Since fbs software splits observational spectrum into
two separate components (see Figure 3) and we know
that both of them have luminosity class V, we can esti-
mate the spectral type of each of them. For both com-
ponents Ca K/(H+Ca H) intensity ratio ∼1, but the
CH G-band at ∼4300 Å is not clear visible yet implies
that both spectra have spectral types F0–F5 (Evans
and Howarth 2003; Evans et al. 2004). Additionally,
taking into account criteria from Jaschek and Jaschek
(1990) about EWs of different lines, I finally classify the
component A as F5 based on EW(Fe i λ4045)= 0.51 ±
0.02 Å, EW(Sr ii λ4077)= 0.31 ± 0.02 Å, EW(Hγ)=
4.95 ± 0.10 Å and EW(Hβ)= 5.15 ± 0.10 Å and compo-
nent B as F8 based on EW(Fe i λ4045)= 0.71 ± 0.01 Å,
EW(Ca i λ4226)= 0.75 ± 0.02 Å, EW(Hγ)= 4.20 ±
0.10 Å and EW(Hβ)= 4.05 ± 0.10 Å. This classifica-
Fig. 8 Evolutionary tracks of stars of mass 1.44 M (up- tion for component A coincides with the classification
per panel) for the component A and mass 1.32 M (lower
from Gómez-Forrellad et al. (2003) and for components
panel) for the component B for two metallicities: [Fe/H] =
-0.19 dex (blue) and [Fe/H] = 0.05 dex (red). The burgundy B with the classification from Griffin (2014).
dots include the NN Del system components, and the black
dotted lines show the distance to the optimal solution as
found by Eq.7. 8 Conclusions

The long-period eclipsing binary star NN Del has been

studied using spectral data obtained with the HRS
échelle spectrograph of the SALT telescope. A veloc-
ity curve has been constructed which is based on 19
spectra obtained during 2017–2019 years and covers all
phases of this binary system. The obtained orbital and
12

absolute parameters of the components, together with

their errors, agree with the results obtained in early
works, which shows the good quality of HRS spectra
and correctness of the applied processing and analysis
methodology. The luminosities of both components are
calculated and the effective temperatures are directly
evaluated altogether with the metallicity of the system.
The analysis of the calculated parameters is presented
that allows to estimate the obtained absolute param-
eters in various ways and shows the repeatability of
these assessments. The age and evolutionary status of
both components have been estimated, as well as their
spectral types.

Acknowledgments

A. K. is grateful to D. Graczyk (the referee) for use-

ful comments and suggestions on the manuscript. This
work is based on observations obtained with the South-
ern African Large Telescope (SALT) under programs
2017-1-MLT-001 and 2019-1-SCI-004 (PI: Kniazev).
This research is supported by the National Research
Foundation (NRF) of South Africa and the Russian
Foundation for Basic Research grant 19-02-00779.
13

References Evans, C.J., Howarth, I.D., Irwin, M.J., Burnley, A.W.,

Harries, T.J.: Mon. Not. R. Astron. Soc. 353(2), 601
Andersen, J.: Astron. Astrophys. Rev. 3, 91 (1991). (2004). astro-ph/0406409. doi:10.1111/j.1365-2966.2004.08096.x
doi:10.1007/BF00873538
Barnes, S.I., Cottrell, P.L., Albrow, M.D., Frost, N., Gra- Fernandes, J., Lebreton, Y., Baglin, A., Morel, P.: Astron.
ham, G., Kershaw, G., Ritchie, R., Jones, D., Sharples, Astrophys. 338, 455 (1998)
R., Bramall, D., Schmoll, J., Luke, P., Clark, P., Tyas, Gaia Collaboration: Astron. Astrophys. 616, 11 (2018).
L., Buckley, D.A.H., Brink, J.: The optical design of the 1804.09380. doi:10.1051/0004-6361/201832865
Southern African Large Telescope high resolution spec- Gallenne, A., Pietrzyński, G., Graczyk, D., Pilecki, B.,
trograph: SALT HRS. Society of Photo-Optical Instru- Storm, J., Nardetto, N., Taormina, M., Gieren, W.,
mentation Engineers (SPIE) Conference Series, vol. 7014, Tkachenko, A., Kervella, P., Mérand , A., Weber,
p. 70140 (2008). doi:10.1117/12.788219 M.: Astron. Astrophys. 632, 31 (2019). 1910.03393.
Bramall, D.G., Sharples, R., Tyas, L., Schmoll, J., Clark, doi:10.1051/0004-6361/201935837
P., Luke, P., Looker, N., Dipper, N.A., Ryan, S., Buckley, Gómez-Forrellad, J.M., Sánchez-Bajo, F., Corbera-Subirana,
D.A.H., Brink, J., Barnes, S.I.: The SALT HRS spectro- M., GarcÍa-Melendo, E., Vidal-Sainz, J.: Astrophys.
graph: final design, instrument capabilities, and opera- Space Sci. 283(2), 297 (2003). doi:10.1023/A:1021324730755
tional modes. Society of Photo-Optical Instrumentation
Engineers (SPIE) Conference Series, vol. 7735, p. 77354 Gorda, S.Y., Svechnikov, M.A.: Astronomy Reports 42, 793
(2010). doi:10.1117/12.856382 (1998)
Bramall, D.G., Schmoll, J., Tyas, L.M.G., Clark, P., Griffin, R.F.: Astrophys. J. 148, 465 (1967). doi:10.1086/149168
Younger, E., Sharples, R.M., Dipper, N.A., Ryan, S.G.,
Buckley, D.A.H., Brink, J.: The SALT HRS spectro-
Griffin, R.F.: The Observatory 134, 109 (2014)
graph: instrument integration and laboratory test re-
Hawkins, K., Lucey, M., Ting, Y.-S., Ji, A., Katzberg, D.,
sults. Society of Photo-Optical Instrumentation Engineers
Thompson, M., El-Badry, K., Teske, J., Nelson, T., Car-
(SPIE) Conference Series, vol. 8446, p. 84460 (2012).
rillo, A.: Mon. Not. R. Astron. Soc. 492(1), 1164 (2020).
doi:10.1117/12.925935
1912.08895. doi:10.1093/mnras/stz3132
Bressan, A., Marigo, P., Girardi, L., Salasnich, B., Dal
Husser, T.-O., Wende-von Berg, S., Dreizler, S., Homeier,
Cero, C., Rubele, S., Nanni, A.: Mon. Not. R. Astron.
D., Reiners, A., Barman, T., Hauschildt, P.H.: Astron.
Soc. 427(1), 127 (2012). 1208.4498. doi:10.1111/j.1365-
Astrophys. 553, 6 (2013). 1303.5632. doi:10.1051/0004-
2966.2012.21948.x
Buckley, D.A.H., Swart, G.P., Meiring, J.G.: Comple- 6361/201219058
tion and commissioning of the Southern African Large Jaschek, C., Jaschek, M.: The Classification of Stars,
Telescope. Society of Photo-Optical Instrumentation En- (1990)
gineers (SPIE) Conference Series, vol. 6267, p. 62670 Kniazev, A.Y., Gvaramadze, V.V., Berdnikov, L.N.: Mon.
(2006). doi:10.1117/12.673750 Not. R. Astron. Soc. 459(3), 3068 (2016). 1604.03942.
Choi, J., Dotter, A., Conroy, C., Cantiello, M., Paxton, doi:10.1093/mnras/stw889
B., Johnson, B.D.: Astrophys. J. 823(2), 102 (2016). Kniazev, A.Y., Usenko, I.A., Kovtyukh, V.V., Berd-
1604.08592. doi:10.3847/0004-637X/823/2/102 nikov, L.N.: Astrophysical Bulletin 74(2), 208 (2019).
Coelho, P.R.T.: Mon. Not. R. Astron. Soc. 440(2), 1027 doi:10.1134/S199034131902010X
(2014). 1404.3243. doi:10.1093/mnras/stu365 Kniazev, A.Y., Malkov, O.Y., Katkov, I.Y., Berdnikov,
Crause, L.A., Sharples, R.M., Bramall, D.G., Schmoll, J., L.N.: arXiv e-prints, 2004 (2020). 2004.04115
Clark, P., Younger, E.J., Tyas, L.M.G., Ryan, S.G., Kovaleva, D.A.: Astronomy Reports 45, 972 (2001).
Brink, J.D., Strydom, O.J., Buckley, D.A.H., Wilkinson, doi:10.1134/1.1426128
M., Crawford, S.M., Depagne, É.: Performance of the Makarov, V., Bastian, U., Hoeg, E., Grossmann, V.,
Southern African Large Telescope (SALT) High Resolu- Wicenec, A.: Information Bulletin on Variable Stars
tion Spectrograph (HRS). Society of Photo-Optical In- 4118, 1 (1994)
strumentation Engineers (SPIE) Conference Series, vol. Malkov, O.Y.: Astron. Astrophys. 402, 1055 (2003).
9147, p. 91476 (2014). doi:10.1117/12.2055635 doi:10.1051/0004-6361:20030313
Crawford, S.M., Still, M., Schellart, P., Balona, L., Buck- Malkov, O.Y.: Mon. Not. R. Astron. Soc. 382, 1073 (2007).
ley, D.A.H., Dugmore, G., Gulbis, A.A.S., Kniazev, doi:10.1111/j.1365-2966.2007.12086.x
A., Kotze, M., Loaring, N., Nordsieck, K.H., Pickering, Malkov, O.Y., Tamazian, V.S., Docobo, J.A., Chulkov,
T.E., Potter, S., Romero Colmenero, E., Vaisanen, P., D.A.: Astron. Astrophys. 546, 69 (2012). doi:10.1051/0004-
Williams, T., Zietsman, E.: PySALT: the SALT science 6361/201219774
pipeline. Society of Photo-Optical Instrumentation En- Mamajek, E.E., Prsa, A., Torres, G., Harmanec, P., As-
gineers (SPIE) Conference Series, vol. 7737, p. 773725 plund, M., Bennett, P.D., Capitaine, N., Christensen-
(2010). doi:10.1117/12.857000 Dalsgaard, J., Depagne, E., Folkner, W.M., Haberreiter,
Docobo, J.A., Tamazian, V.S., Malkov, O.Y., Campo, P.P., M., Hekker, S., Hilton, J.L., Kostov, V., Kurtz, D.W.,
Chulkov, D.A.: Mon. Not. R. Astron. Soc. 459, 1580 Laskar, J., Mason, B.D., Milone, E.F., Montgomery,
(2016). 1609.03392. doi:10.1093/mnras/stw709 M.M., Richards, M.T., Schou, J., Stewart, S.G.: arXiv
Evans, C.J., Howarth, I.D.: Mon. Not. R. Astron. Soc. e-prints, 1510 (2015). 1510.07674
345(4), 1223 (2003). astro-ph/0308125. doi:10.1046/j.1365-
2966.2003.07038.x
14

O’Donoghue, D., Buckley, D.A.H., Balona, L.A., Bester, D.,

Botha, L., Brink, J., Carter, D.B., Charles, P.A., Chris-
tians, A., Ebrahim, F., Emmerich, R., Esterhuyse, W.,
Evans, G.P., Fourie, C., Fourie, P., Gajjar, H., Gordon,
M., Gumede, C., de Kock, M., Koeslag, A., Koorts, W.P.,
Kriel, H., Marang, F., Meiring, J.G., Menzies, J.W., Men-
zies, P., Metcalfe, D., Meyer, B., Nel, L., O’Connor, J.,
Osman, F., Du Plessis, C., Rall, H., Riddick, A., Romero-
Colmenero, E., Potter, S.B., Sass, C., Schalekamp, H.,
Sessions, N., Siyengo, S., Sopela, V., Steyn, H., Stoffels,
J., Scholtz, J., Swart, G., Swat, A., Swiegers, J., Tiheli,
T., Vaisanen, P., Whittaker, W., van Wyk, F.: Mon. Not.
R. Astron. Soc. 372(1), 151 (2006). astro-ph/0607266.
doi:10.1111/j.1365-2966.2006.10834.x
Pietrinferni, A., Cassisi, S., Salaris, M., Castelli, F.:
Astrophys. J. 612(1), 168 (2004). astro-ph/0405193.
doi:10.1086/422498
Pojmanski, G.: Acta Astron. 47, 467 (1997). astro-
ph/9712146
Popper, D.M.: Annu. Rev. Astron. Astrophys. 18, 115
(1980). doi:10.1146/annurev.aa.18.090180.000555
Prša, A., Conroy, K.E., Horvat, M., Pablo, H., Kochoska,
A., Bloemen, S., Giammarco, J., Hambleton, K.M., De-
groote, P.: Astrophys. J. Suppl. Ser. 227(2), 29 (2016).
1609.08135. doi:10.3847/1538-4365/227/2/29
Straižys, V.: Multicolor Stellar Photometry, (1992)
Sybilski, P., Pawlaszek, R.K., Sybilska, A., Konacki, M.,
Helminiak, K.G., Kozlowski, S.K., Ratajczak, M.: Mon.
Not. R. Astron. Soc. 478(2), 1942 (2018). 1805.00520.
doi:10.1093/mnras/sty1135
Torres, G., Andersen, J., Giménez, A.: Astron. Astrophys.
Rev. 18, 67 (2010). 0908.2624. doi:10.1007/s00159-009-
0025-1

This manuscript was prepared with the AAS LATEX macros v5.2.