Influence of Departures from LTE on Determinations of the Sulfur

Abundances in A–K Type Stars

S. A. Korotin1 and K. O. Kiselev1

Crimean Astrophysical Observatory, Nauchny, Crimea, Russia

e-mail: [email protected]
Received date; accepted: date
arXiv:2501.00774v1 [astro-ph.SR] 1 Jan 2025

ABSTRACT

The influence of departures from local thermodynamic equilibrium (LTE) on neutral sulfur lines is considered. A grid of corrections is
proposed to take into account the influence of departures from LTE for neutral sulfur lines in the visible and infrared spectral regions,
including the H-band. The grid is calculated using the atomic model of sulfur incorporating the most up-to-date collision rates with
electrons and hydrogen. The inclusion of levels and transitions of ionized sulfur in the atomic model made it possible to expand the
range of effective temperatures of stellar photospheres in the grid up to 10000 K. The atomic model was tested in determining the
sulfur abundance of 13 stars and showed its adequacy in a wide range of fundamental stellar parameters. In the spectra of all test stars,
the sulfur lines are fitted with similar abundances of the element, regardless of the degree of influence of the effects of deviation from
LTE on a particular spectral line. For lines of several multiplets, the wavelengths and oscillator strengths were refined. A list of S I
lines recommended for determining sulfur abundance has been created.
Key words. line formation, line profiles of stars, elemental abundances on the Sun

1. INTRODUCTION (2001) suggested that the [S/Fe] ratio continues to increase at

[Fe/H] < 2.0. Other studies ((Spite et al. (2011), Kacharov et al.
To understand the processes of chemical evolution in the Galaxy (2015), and others) have shown evidence of a plateau, although
and the history of star formation, researchers examine the re- this plateau has significant dispersion due to the complicated
lationships between various chemical elements in stellar atmo- analysis of weak lines at such low metallicities. In stars with
spheres. Iron-peak elements and α-elements play a fundamental near-solar heavy element abundance, sulfur behaves as a typical
role in these studies, as their nucleosynthesis occurs over differ- α-element. Within the range of -1 < [Fe/H] < 0, the [S/Fe] ratio
ent timescales. α-Elements can be used as cosmic clocks to mon- decreases to zero, which confirms the general characteristics of
itor both stellar nucleosynthesis and the history of galaxy for- α-element nucleosynthesis.
mation. Unlike other α-elements, sulfur is moderately volatile.
However, sulfur has received little attention from re-
For this reason, its abundance measured in stars within the Local
searchers. Analysis of sulfur is often omitted in favor of other
Group galaxies can be directly compared to its abundance mea-
α-elements, whose measurements are less complex. The issues
sured in extragalactic H II regions. While α-elements are pro-
at hand include the weakness of sulfur’s optical lines. In the op-
duced by Type II supernovae, iron-peak elements are produced
tical range, there are 6, 8, and 10 multiplets with relatively weak
by Type Ia supernovae over longer timescales. Among the var-
lines that nearly vanish in metal-deficient stars. Two triplets in
ious α-elements, the nucleosynthesis of sulfur is not fully un-
derstood. Sulfur is formed during the final stages of evolution the near-infrared region, λ 9212–9237 and λ 10455–10459 Å are
of massive stars (M > 20 M⊙ ). Hydrostatic burning of neon in heavily affected by departures from local thermodynamic equi-
the core at temperature of 1.2 × 109 K leads to the formation of librium (LTE) (Takeda et al. (2005), Korotin (2009)). These lines
a convective oxygen core and the production of α-elements, in- are located in a region of the spectrum that is significantly dis-
cluding sulfur. Further, various nuclear transformation processes torted by lines from Earth’s atmosphere. For these reasons, stud-
occur, resulting in the near-total destruction of the produced sul- ies of sulfur are often bypassed in favor of simpler analyses of
fur during the Si-burning phase. However, sulfur production con- other α-elements. As a result, our understanding of sulfur’s be-
tinues with oxygen burning in the outer layers around the core, havior is still insufficient compared to other α-elements.
as well as through explosive oxygen burning during Type II su- The influence of non-LTE effects on sulfur lines requires
pernova explosions (Limongi & Chieffi 2003; Kobayashi et al. detailed study. As shown in Takeda et al. (2005) and Korotin
2020). This enrichment is assumed to occur over a relatively (2009), non-LTE corrections for S I lines have complex depen-
short timescale on the order of tens of millions of years. Accord- dences on the fundamental parameters of stars and the sulfur
ingly, it can be expected that the [S/Fe] ratio will be relatively abundance itself. Model calculations Takeda et al. (2005) and
constant for low-metallicity stars. These are stars that formed Korotin (2009) show similar patterns in the behavior of non-LTE
before the ignition of the first Type Ia supernovae. This ratio corrections depending on stellar parameters, though the correc-
then decreases as iron levels increase. There is still no consen- tions differ slightly in magnitude. It should be noted that since
sus on whether this “plateau” exists in the relative sulfur abun- these model calculations were conducted, new atomic data have
dance of metal-deficient stars. The authors of Israelian & Rebolo become available for neutral and ionized sulfur. The use of new
Article number, page 1 of 14
 manuscript Korotin_Kiselev

detailed quantum- mechanical calculations of inelastic collision plied the formula from van Regemorter (1962), and for forbid-
rates with electrons and hydrogen should help avoid potential er- den transitions, the formula from Allen (1973), with an effective
rors introduced by approximate formulas. We have worked on collision strength set to 1. Collisional ionization was accounted
updating the sulfur atomic model Korotin (2009), incorporat- for using the formula from Seaton (1962) with the photoioniza-
ing the latest atomic data and verifying it with spectra of well- tion threshold cross-section from TOPBase (Cunto et al. 1993).
studied stars. Collisional interactions with hydrogen atoms become very
important in the atmospheres of stars with low effective temper-
atures and metal-poor stars, where electron density decreases.
2. SULFUR ATOM MODEL Although the actual collisional rates with hydrogen are quite
Our previously developed sulfur atom model (Korotin 2009) in- small, the high concentration of hydrogen provides a significant
cluded 64 levels of neutral sulfur and the ground level of S II. influence on the redistribution of electronic level populations.
Approximate formulas were used to account for collisional in- Recently, detailed quantum-mechanical calculations for inelas-
teractions with electrons and hydrogen, though they were often tic collisions between various atoms and hydrogen have made
of limited accuracy. Over time, our knowledge of the atomic pa- it possible to account for this effect more realistically than with
rameters of sulfur has significantly improved. Detailed quantum- approximate approaches. For the sulfur atom, such calculations
mechanical calculations have occurred for collisional rates of in- were performed in (Belyaev & Voronov 2020). We incorporated
elastic interactions between 56 levels of neutral sulfur and hydro- collisional interactions with hydrogen for forty levels of neutral
gen (Belyaev & Voronov 2020). Previously, the collisions with sulfur within a temperature range of 1000 to 10000 K. Beyond
hydrogen in the atomic model were accounted for using the so- this range, extrapolation was applied; however, the effect of col-
called “Drawin formula” (Drawin 1968; Steenbock & Holweger lisional interactions with hydrogen becomes negligible at such
1984). It is well known that this formula gives highly inaccu- high temperatures.
rate results, which researchers attempted to correct by introduc- To obtain the sulfur level populations in stellar atmospheres,
ing empirical correction factors, obtained by adjusting non-LTE we used the MULTI software package (Carlsson 1986), version
calculations to observations. Comparisons between this approx- 2.3, which we modified slightly (Korotin et al. 1999). The main
imate formula and the detailed calculations (Belyaev & Voronov changes involve using the complete opacity calculation package
2020) show discrepancies that can reach several orders of mag- from the ATLAS9 software (Castelli & Kurucz 2003), including
nitude. A similar situation exists for collisional rates with elec- the so-called opacity distribution function (ODF) arrays. After
trons. The ADAS database (Summers & O’Mullane 2011) now solving the statistical equilibrium and radiative transfer equa-
includes detailed calculations for collisional interactions of the tions together, we obtained coefficients of deviation from LTE
17 lowest levels of sulfur with free electrons, allowing us to populations, known as b-factors: the ratio of non-LTE to LTE
abandon the use of inaccurate approximate formulas. Moreover, populations (NNLTE /NLTE ). These were passed to the synthetic
applying the old atomic model to study sulfur lines in A-type spectrum calculation software SynthV (Tsymbal et al. 2019),
stars with sufficiently high effective temperatures raised the need where sulfur line profiles were calculated with non-LTE effects,
to include additional levels of S II and S III in the model. All this while lines of other elements were calculated under the LTE ap-
prompted us to upgrade the atomic model and perform the ap- proach. Line parameters necessary for synthetic spectrum calcu-
propriate testing. lations were taken from the VALD database (Ryabchikova et al.
As in the previous model, the populations of 64 energy lev- 2015).
els of neutral sulfur were considered in detail. The highest S I
level in the model is separated from the continuum by 0.21 eV,
which ensures reliable interaction at temperatures above 2500 K. 3. APPLICATION OF THE SULFUR ATOM MODEL
Additionally, 81 levels of S II and the ground level of S III were TO SOLAR SPECTRUM ANALYSIS. REFINEMENT
included. The highest S II level is separated from the continuum OF NEUTRAL SULFUR LINE PARAMETERS
by 0.45 eV, corresponding to a temperature of 5150 K. To more
fully account for particle numbers, we included one level of S I, After updating the sulfur atom model, it is necessary to validate
six levels of S III, and the ground level of S VI, with populations its accuracy. This requires comparing the model with observed
calculated in LTE. Fine structure levels were not considered. The spectra from several stars with well-known atmospheric param-
values of electronic energy levels were taken from Martin et al. eters. These stars’ atmospheric parameters should span a broad
(1990). The diagram of the sulfur atom levels is shown in Fig. 1. range of temperatures and pressures. Validation of the model is
Detailed consideration included 775 bound-bound and 146 performed by comparing observed and synthetic profiles of spec-
bound-free transitions between levels. Line profiles were cal- tral lines that belong to different multiplets and exhibit different
culated using the Voigt profile, taking into account radiative degrees of deviation from values predicted in the LTE approxi-
broadening as well as Stark and Van der Waals broadening. mation. For instance, some spectral lines show no considerable
The number of points along the profile varied from 30 to 110, non-LTE effects, while others display significant deviations from
depending on the line’s intensity. Photoionization cross sec- LTE. Such a combination of lines cannot be accurately modeled
tions and oscillator strengths were taken from the TOPBase in an LTE approach by simply adjusting the element’s abun-
database (Cunto et al. 1993). For forbidden transitions, oscilla- dance. A well-selected atomic model in non-LTE calculations
tor strengths were obtained from the catalog Hirata & Horaguchi should allow an acceptable description of this spectrum using a
(1994). consistent element abundance for all analyzed lines.
Detailed quantum-mechanical calculations of collisional ex- Only a limited number of S I lines are suitable for determin-
citation rates by electrons for transitions among 17 levels ing sulfur abundance in the visible spectrum. These include the
of neutral sulfur were obtained from the ADAS database first (λ 9212–9237 Å), sixth (λ 8694 Å), eighth (λ 6743-6757 Å),
(Summers & O’Mullane 2011). For transitions among the 32 and tenth (λ 6046-6052 Å) quintet system multiplets. In the near-
lower levels of ionized sulfur, we used calculations from infrared range, multiplet lines at λ 10455-10459 Å, λ 15400-
Tayal & Zatsarinny (2010). For other allowed transitions, we ap- 15422 Å, λ 15469-15478 Å and λ 22507-22707 Å. The remain-
Article number, page 2 of 14
 Korotin and Kiselev: NLTE-effects of the Sulfur Abundances in A–K Type Stars

200000.
3p3 4S* S II 3p2 3P S III
80000.
175000.
70000.
150000.
60000.
125000.
Energy (cm-1 )

50000.
100000.
40000.

Energy (cm-1 )
75000.
30000.

50000.
20000.

10000. 25000.

0. 0.
1S 1P 1P* 1D 1D* 1F 3S* 3P 3P* 3D 3D* 3F 3F* 5S* 5P 5D* 5F 2S 2S* 2P 2P* 2D 2D* 2F 2F* 2G 4S* 4P 4P* 4D 4D* 4F 4F* 4G

Fig. 1. Grotrian diagrams for S I (left) and S II (right). Only transitions used to determine sulfur abundances are shown.

ing lines are either too weak or blended. Notably, the eighth and (2008) (hereafter DH) cover only part of the S I lines examined
tenth multiplets each consist of three lines, and each line itself here.
is a superposition of three components with slight wavelength Oscillator strengths for the lines of the first multiplet are
shifts. Such line profiles deviate significantly from a Gaussian nearly identical in ZB, FTI, and DH, with differences not ex-
shape, necessitating synthetic spectrum calculations for compar- ceeding 0.03 dex and an estimated accuracy of 25%. For further
ison with observations, as illustrated in the figures below. Using calculations, we used data from ZB for these lines.
equivalent widths for these lines requires caution and the appli- Figure 3 compares the calculated line profiles of two IR
cation of methods for calculating equivalent widths of multicom- triplets, including non-LTE effects, with the observed solar spec-
ponent lines. trum. All lines except for λ 9212 Å, which is distorted by
For analyzing neutral sulfur lines, we used the solar spec- Earth’s atmosphere, are described well at a sulfur abundance
trum from atlases Kurucz et al. (1984) and Reiners et al. (2016), (S/H )=7.16 , with a profile fitting error of no more than ±0.02
covering the range from λ 3000 to λ 13000 Å and λ 4050 to dex. The non-LTE correction for the lines of the first IR triplet
λ 23000 Å, respectively. Calculations were based on a solar at- (λ 9212-9237 Å) is –0.23 dex, while for the λ 10456 Å line,
mosphere model Castelli & Kurucz (2003) with a microturbulent the non-LTE correction is –0.18 dex. This derived sulfur abun-
velocity of 1 km/s. The Sun’s rotational velocity was set at 1.8 dance aligns with the sulfur abundance in chondrites: (S/H)=
km/s. Element abundances in the solar atmosphere were taken 7.15 ± 0.02 (Lodders 2021). The sulfur abundance in these me-
from Asplund et al. (2021). Profiles of all lines included Van der teorites is determined with high accuracy and can be used as a
Waals broadening, calculated according to Barklem & O’Mara reference for the Solar System. Given that the Sun is a main-
(1997); Barklem et al. (1998). sequence dwarf star, its photosphere likely retains its original
Calculations of neutral sulfur level populations for the solar composition, so the sulfur abundance should match the mete-
photosphere showed results consistent with Korotin (2009). The oritic value. Agreement between the IR triplet-derived sulfur
sixth, eighth, and tenth multiplet lines form almost in LTE and abundance and meteoritic levels can serve as a confirmation of
are not influenced by non-LTE effects. The lines in the IR H- the adequacy of our sulfur atom model.
range behave similarly. Only the lines of two IR triplets, λ 9212- The analysis of other sulfur lines in the optical range
9237 and λ 10455-10459 Å are significantly enhanced due to showed that using wavelengths and oscillator strengths from
overpopulation of their lower levels, 4s 5S* and 4s 3S*, and un- the VALD (Ryabchikova et al. 2015) or NIST (Kramida et al.
derpopulation of their upper levels, 4p 5P and 4p 3P, respec- 2012) databases to calculate the synthetic spectrum did not yield
tively, at the line formation depths. The b-factor distribution of ideal line profiles. This is understandable, since the oscillator
some sulfur levels by photospheric depth is shown on the left strength calculation accuracy varies from 25 to 40% depend-
panel of Fig. 2. The population levels between which the sixth, ing on the source. For example, for the sixth multiplet lines,
eighth, and tenth multiplet transitions form are in LTE at these the log gf values from ZB and BQZ differ by 0.31 dex. Wave-
line formation depths. The right panel of Fig. 2 shows the ratio lengths are also calculated based on electronic level energies,
of the source function Sl to the Planck function Bν for both IR which are typically determined with an accuracy of ±25-30 cm−1
triplets as a function of optical depth. It is evident that at the line (Martin et al. 1990). Given that optical-range S I lines are un-
formation depths, Sl /Bν < 1. Thus, non-LTE effects strengthen affected by non-LTE effects, we refined their wavelengths and
the lines, leading to negative corrections to sulfur abundance es- oscillator strengths. This procedure of refinement of line param-
timates when using the LTE approach for these multiplet lines. eters serves for finding a balance between observational data and
Experimentally, oscillator strengths and wavelengths have theoretical values of wavelengths and oscillator strengths.
only been determined for the triplet lines at λ 10455- The sixth multiplet is the triplet at λ 8693-8694 Å. The log gf
10459 Å (Zerne et al. 1997). For other sulfur lines, only the- values of this triplet from BQZ are systematically lower than
oretical calculations are available, notably in Biemont et al. ZB by 0.31 dex and lower than DH by 0.45 dex. Profiles calcu-
(1993) (hereafter BQZ). More recent results are presented lated using BQZ oscillator strengths were much weaker than ob-
in Zatsarinny & Bartschat (2006) (hereafter ZB). Studies served. ZB values produced a more realistic synthetic spectrum,
Froese Fischer et al. (2006) (hereafter FTI) and Deb & Hibbert though profiles were still slightly weaker than observed. At the
Article number, page 3 of 14
 manuscript Korotin_Kiselev

1.0

1.0

8694
0.8

/B
0.8
3p4 3P
10455

l
bi

10455

S
4s 5S*

4s 3S* 9212
9212 0.6 S/B 9212
4p 5P
0.6
4p 3P S/B 10455

4d 5D*

5d 5D*
0.4
-2 log
-1 0 -2 log
-1 0
5000 5000

Fig. 2. Distribution of b-factors in the Sun’s atmosphere (left panel) and the variation of the Sl /Bν ratio with depth for two IR triplets (right panel).
Arrows indicate the formation depths of some sulfur lines

1.0

S I S I
S I
S I
0.9 S I
Relative flux

0.8

0.7

0.6

0.5

9227.5 9228.0 9228.5 9237.0 9237.5 9238.0 10455 10456 10457 10458 10459 10460 o
(A)

Fig. 3. Comparison of observed (circles) and synthetic profiles of IR triplet lines in the Sun’s spectrum. The non-LTE profile for sulfur abundance
(S/H) = 7.16 is shown as a solid line, and the LTE profile, calculated with the same abundance, is shown as a dashed line.

same time, oscillator strengths from DH resulted in excessively The tenth multiplet is the triplet at λ 6041-6052 Å. Each
strong calculated lines. The difference between log gf from DH component in this triplet is also a triplet of closely spaced lines.
and ZB is 0.13 dex for all components of the multiplet. The best The λ 6041 Å line is blended with a strong iron line and was
agreement between synthetic and observed profiles was achieved excluded from analysis. DH and BQZ do not provide calcula-
with values of log gfZB + 0.088. We also adjusted the multiplet tions for this multiplet, and VALD values are on average lower
component wavelengths. The shift does not exceed 0.07 Å from than ZB by 0.15 dex. Profiles calculated with these oscillator
those in VALD, which is within the stated accuracy range. strengths were weaker than observed. A satisfactory description
for the λ 6052 Å line was achieved by increasing log gfZB by
The eighth multiplet is the triplet at λ 6743-6757 Å. Each 0.113 dex for all components in this triplet. We also adjusted the
component in this triplet is itself a triplet of closely spaced lines. wavelengths of two line components, with a maximum shift of
The log gf values of these nine lines from BQZ are on average 0.03 Å. However, this oscillator strength adjustment did not pro-
lower than ZB by 0.11 dex. DH calculations for this multiplet vide a good fit for the λ 6046 Å line in the observed spectrum.
are unavailable. When using BQZ oscillator strengths with me- This line is clearly broadened by an unknown component in the
teoritic sulfur abundance, the theoretical lines are weaker than solar spectrum. Thus, we excluded this line from further analy-
observed, while ZB data produced slightly stronger lines than sis.
observed. The best agreement is achieved at log log gfZB +0.075, Figure 4 presents examples of synthetic profiles with refined
which is close to the BQZ values, but slightly lower. It should be parameters, describing the observed solar spectrum. The refined
noted that the wavelengths of these multiplet components vary line parameters are listed in Table 1.
by up to 0.1 Åacross different sources. The multiplet itself is
a combination of transitions between three lower and five up-
per levels. Lower level energies are relatively well-determined, 4. COMPARISON OF NON-LTE CALCULATIONS
particularly for the upper levels of the first multiplet lines at WITH OBSERVED STELLAR SPECTRA
λ 9212-9237 Å. By varying upper level energies, we attempted
4.1. Late-Type Stars
to achieve the best description of the observed profiles. Energy
corrections did not exceed 25 cm−1 . The maximum wavelength After refining the S I line parameters based on the solar spec-
shift relative to VALD recommendations was 0.06 Å, which is trum, we tested the sulfur atom model against spectra of late-
consistent with various discrepancies from the literature. type stars. We selected well-studied stars with reliably deter-
Article number, page 4 of 14
 Korotin and Kiselev: NLTE-effects of the Sulfur Abundances in A–K Type Stars

1.0
S I S I S I
S I
Relative flux

0.9

6052.4 6052.8 6743.2 6743.6 6744.0 6748.5 6748.9 6749.3 6756.8 6757.2 6757.6
o
(A)

Fig. 4. Comparison of observed (circles) and synthetic profiles of the eighth and tenth multiplet lines in the Sun’s spectrum. Profiles with refined
parameters are shown as solid lines, and those calculated with VALD parameters are shown as dotted lines.

Table 1. Parameters of S I lines tions. The abundances obtained from the lines within the mul-
tiplet are nearly identical. The given error in the average abun-
λ, Å E,eV log gf Γvw λ,Å E,eV log gf Γvw dance, reflecting the line-to-line scatter, shows a very small vari-
6052.53 7.87 -1.99 -6.55 10455.45 6.86 0.25 -7.32 ation. The real error in determining the element’s abundance will
6052.59 7.87 -1.15 -6.55 10456.76 6.86 -0.45 -7.32 be slightly higher, since it should include the effect of the inac-
6052.69 7.87 -0.56 -6.55 10459.41 6.86 0.03 -7.32
6743.47 7.87 -1.24 -6.65 15400.08 8.70 0.43 -7.00 curacy of the fundamental parameters of the star.
6743.54 7.87 -0.88 -6.65 15403.72 8.70 -0.30 -7.00 The IR triplet at λ 10455-10459 Å, unfortunately, could only
6743.65 7.87 -0.94 -6.65 15403.79 8.70 0.61 -7.00 be used for Procyon, Pollux, and HD 195295, since other spectra
6748.61 7.87 -1.36 -6.65 15422.20 8.70 -1.84 -7.00
6748.71 7.87 -0.77 -6.65 15422.26 8.70 -0.30 -7.00 either had excessive noise in this range or did not cover it. For
6748.85 7.87 -0.56 -6.65 15422.28 8.70 0.77 -7.00 Pollux and τ Ceti, the λ 6052 Å line was heavily blended with
6756.86 7.87 -1.71 -6.65 15469.82 8.05 -0.17 -6.90 CN molecular lines, and it was very weak for HD 84937, so it
6757.03 7.87 -0.86 -6.65 15478.48 8.05 0.06 -6.90 was excluded from analysis. Lines of the eighth multiplet were
6757.18 7.87 -0.28 -6.65 22507.56 7.87 -0.48 -7.44
8693.16 7.87 -1.29 -6.98 22519.07 7.87 -0.25 -7.44 also nearly invisible in HD 84937, with equivalent widths below
8693.95 7.87 -0.45 -6.98 22552.57 7.87 -0.04 -7.44 1 mÅ. However, there is a clear depression at the λ 6757 Å line,
8694.64 7.87 0.14 -6.98 22563.83 7.87 -0.26 -7.44 which aligns well with the sulfur abundance determined from the
9212.87 6.53 0.39 -7.37 22575.39 7.87 -0.73 -7.44 first and sixth multiplet lines. This serves as an additional sup-
9228.09 6.53 0.25 -7.37 22644.06 7.87 -0.34 -7.44
9237.54 6.53 0.02 -7.37 22707.74 7.87 0.44 -7.44 port for the derived sulfur abundance. Figure 5 shows examples
comparing synthetic and observed profiles. It can be seen that,
despite the varying sensitivity of lines from different multiplets
mined fundamental parameters. For an accurate atomic model, to non-LTE effects depending on stellar atmospheric parameters,
lines from various multiplets should be described by close sulfur they yield closely aligned sulfur abundances. This can be consid-
abundances across a broad range of stellar atmospheric param- ered as a good confirmation of the adequacy of the atomic model
eters. The study included the following stars. Three stars with used in our analysis.
solar-like composition: hot dwarf Procyon, hot giant HD 195295,
and cool giant Pollux. Two metal-poor stars: Ceti and HD 22879, 4.2. Sulfur Lines in A-Type Stars
as well as the metal-deficient star HD 84937. Thus, the stars used
in the analysis differ greatly from each other. The stars were se- The next step in validating our model involved applying it to
lected with low rotation velocities to minimize profile distortion. hotter stars whose spectra feature sulfur in both S I and S II
Fundamental parameters were taken from Jofré et al. (2015) lines. These are A-type stars with temperatures above 9500 K,
for all stars except HD 195295, for which data were taken from where ionized sulfur lines become visible while neutral sulfur
Lyubimkov et al. (2010). The spectrum for HD 195295 was ob- lines in the visible spectrum also remain relatively intense. For
tained using the 2.7-m telescope spectrograph at McDonald Ob- this analysis, we selected two stars, o Peg and θ Vir, studied
servatory (Tull et al. 1995) with a resolution R = 60 000. The in Romanovskaya et al. (2023). These stars have temperatures
spectra of the remaining objects, obtained with the UVES spec- of 9600 K and log g = 3.8 and 3.6, respectively. The spectra
trograph at the Paranal Observatory (Tull et al. 1995), were taken were acquired using the ESPaDOnS spectrograph (Petit et al.
from the ESO archive. They all have a spectral resolution of 2014) at a resolution R = 68 000. For o Peg, a procedure of
at least 75 000 and a signal-to-noise ratio of at least 200. The averaging 12 spectrograms was performed, which allowed the
spectra were processed and the continuum was drawn using the signal-to-noise ratio to be increased to 700. Details are given in
Dech package (Galazutdinov 2022). The atmosphere models of Romanovskaya et al. (2023).
the stars were calculated using the ATLAS9 software with the The parameters of the ionized sulfur lines used for the anal-
ODF from Mészáros et al. (2012). The parameters of the stars ysis are given in Table 3. The oscillator strengths are taken from
are given in Table 2. It also contains the results of determining Irimia & Froese Fischer (2005). All lines are weak, from 2 to 7
the sulfur abundance for the multiplets used in the analysis (one mÅ, but are well detected due to the high quality of the spec-
line for each multiplet) and the corresponding non-LTE correc- tra. These lines form deep in the photosphere and are therefore
Article number, page 5 of 14
 manuscript Korotin_Kiselev

Table 2. Sulfur abundance in the stars

Stellar parameters Sulfur abundance (S/H) Nl

Star Te f f log g [Fe/H] Vt V sin i 6052 6757 8694 9212 10455
Procyon 6554 3.99 0.01 1.66 2.8 7.10 7.11 7.15 7.13 7.13 12
HD195295 6570 2.32 0.00 3.60 8.0 7.02 7.07 7.06 7.08 7.08 10
Pollux 4858 2.90 0.13 1.28 2.0 7.20 7.25 7.15 7.10 6
τ Ceti 5414 4.49 -0.49 0.89 0.4 6.74 6.67 6.67 6
HD 22879 5868 4.27 -0.86 1.05 4.4 6.56 6.56 6.53 6.50 6
HD 84937 6356 4.15 -2.03 1.39 5.2 5.40 5.36 5.34 5
o Peg 9597 3.80 0.25 1.98 5.4 7.48 7.53 7.56 7.62 7.52 11
θ Vir 9600 3.60 0.15 1.42 0.5 7.40 7.46 7.44 7.44 7
HD 24040 5809 4.12 0.09 1.00 1.5 7.23 7.21 7.19 7.17 7.20 12
HD 28005 5802 4.18 0.21 1.00 3.5 7.34 7.35 7.34 7.36 7.36 10
HD 34445 5803 4.06 -0.03 1.00 4.5 7.23 7.25 7.22 7.18 7.18 12
HD 82943 5917 4.23 0.13 1.00 4.0 7.32 7.34 7.32 7.32 7.32 9
HD 87359 5645 4.40 -0.07 1.00 4.0 7.14 7.18 7.14 7.14 7.16 10
Average (S/H) Non-LTE corrections, dex
Opt. range H-range 6052 6757 8694 9212 10455
Procyon 7.12 ± 0.02 -0.02 -0.02 -0.06 -0.48 -0.51
HD195295 7.08 ± 0.04 -0.06 -0.08 -0.21 -0.81 -0.83
Pollux 7.17 ± 0.05 0.00 -0.02 -0.30 -0.18
τ Ceti 6.71 ± 0.04 0.00 0.00 -0.13
HD 22879 6.54 ± 0.02 0.00 0.01 0.00 -0.19
HD 84937 5.35 ± 0.03 0.00 0.00 -0.14
o Peg 7.55 ± 0.05 -0.03 -0.04 -0.07 -0.37 -0.34
θ Vir 7.43 ± 0.03 -0.02 -0.03 -0.06 -0.35
HD 24040 7.19 ± 0.02 7.26±0.04 0.00 0.00 0.00 -0.20 -0.15
HD 28005 7.35 ± 0.01 7.32±0.03 0.00 0.00 0.00 -0.21 -0.15
HD 34445 7.21 ± 0.02 7.23±0.03 0.00 0.00 0.00 -0.22 -0.16
HD 82943 7.32 ± 0.01 7.30±0.03 0.00 0.00 -0.01 -0.22 -0.15
HD 87359 7.16 ± 0.02 7.15±0.01 0.00 0.00 0.00 -0.17 -0.11

1.0
1.0

S I S I S I S I S I S I
1.00 1.00

0.9
Relative flux

HD 84937

0.9
0.8

HD 84937 HD 84937 Cet

Cet
0.7 Cet

0.95 0.8 0.95

6757.1 6757.8 9212.5 9213.0 9228.0 9228.5 6757.0 6757.5 8694.5 8695.0 9212.5 9213.0 9213.5
o

(A)

Fig. 5. Comparison of observed (circles) and synthetic profiles of neutral sulfur lines in the spectra of HD 84937 and t Cet. The non-LTE profile is
shown as a solid line, while the LTE profile, calculated with the same abundance, is shown as a dashed line.

nearly free from the influence of non-LTE effects, which lead to rameters. Thus, we conclude that the sulfur atom model is appli-
a very small (by several percent) enhancement. Figure 6 shows cable up to effective temperatures of 10000 K.
examples of comparison of theoretical and observed sulfur line
profiles in two degrees of ionization.
Non-LTE results for neutral sulfur lines in the analyzed A- 5. NEUTRAL SULFUR LINES IN THE INFRARED
type stars are summarized in Table 2. For the S II lines, non-LTE H-RANGE
sulfur abundances (S/H) = 7.51 ± 0.04 fnd 7.38 ± 0.04 were ob-
tained for o Peg and θ Vir, respectively. These values show a Recently, spectral studies in the H-range have expanded. The
difference of only 0.04 dex from the average sulfur abundance emergence of high-resolution IR spectrographs, such as GI-
derived from S I lines. This is smaller than the uncertainties that ANO (Origlia et al. 2014), has significantly enhanced the capa-
may arise from oscillator strengths and fundamental stellar pa- bility to determine the chemical composition of stellar atmo-
Article number, page 6 of 14
 Korotin and Kiselev: NLTE-effects of the Sulfur Abundances in A–K Type Stars

1.0
S I S I S I S II S II
S I
1.00
Relative flux

0.9

0.9 o Peg
0.8
0.95

6748.5 6749.0 6749.5 6757.0 6757.5 8693.5 8694.0 8694.5 8695.0 8695.5 4716.0 4716.5 5453.5 5454.0 o
(A)

Fig. 6. Comparison of observed (circles) and synthetic profiles of neutral and ionized sulfur lines in the spectrum of o Peg. The non-LTE profile is
shown as a solid line, and the LTE profile, calculated with the same abundance, is shown as a dashed line.

Table 3. Parameters of S II lines Takeda et al. (2005) and Korotin (2009), this grid includes lines
in the H-range and extends to higher effective temperatures. The
λ (Å) Elow log gf λ (Å) Elow log gf range of stellar atmosphere parameters spans effective tempera-
4162.665 15.944 0.780 5014.042 14.067 0.046 tures from 4000 to 10000 K, acceleration of gravity log g from 0
4294.402 16.135 0.560 5032.434 13.671 0.188 to 5 and heavy element abundance [Fe/H] from 0 to –2. The mi-
4716.271 13.617 -0.365 5428.655 13.584 -0.177 croturbulent velocity was chosen to be 2 km/s. Calculations for
4815.552 13.671 0.068 5453.855 13.671 0.442 solar metallicity atmospheres were performed with sulfur abun-
4925.343 13.584 -0.206 5606.151 13.733 0.124 dances [S/Fe] = -0.4; 0.0; +0.4 dex, while for metal-poor mod-
5009.567 13.617 -0.234 5639.977 14.067 0.258 els, the values were [S/Fe] = 0.0; +0.4; +0.8 dex, since such stars
usually exhibit excesses of α-elements. Calculations were per-
formed by adjusting the sulfur abundance so that the equivalent
spheres. This includes determining sulfur abundance based on width with non-LTE effects, matched the LTE value at the cor-
lines from three infrared multiplets: λ 15400-15422, λ 15469- responding grid point. The difference between these abundances
15478 and λ 22507-22707 Å. In Korotin et al. (2020), some anal- rep- resents the correction due to non-LTE effects. Corrections
ysis of the non-LTE effects on these lines was performed using were not calculated for lines with an equivalent width less than
the 2009 sulfur model. The lines of two multiplets in the re- 5 mÅ. It should be noted that the magnitude of non-LTE cor-
gion of λ 15400 Å show minimal sensitivity to deviations from rections can be significantly affected by microturbulent velocity,
LTE, while the remaining lines form almost in LTE in the atmo- since it can considerably alter the line formation depth for strong
spheres of solar-type dwarfs. These conclusions remained un- lines. The calculation results are presented in the form of graphs
changed with the use of a new atomic model. (Figs. 7-11) since they visually illustrate and allow better assess-
For testing, we used five stars from Korotin et al. (2020), ment of the degree of non-LTE effects for each star, compared
which have IR spectra obtained with the GIANO spectrograph, to the tables. Some general trends can be observed. As expected,
as well as optical spectra from the ESO archive obtained with the the absolute value of corrections increases from dwarfs to gi-
UVES spectrograph. This allows a comparison of sulfur abun- ants, as the less dense atmosphere of giants reduces the impact of
dances determined from sulfur lines in the optical and H ranges. collisional processes that promote thermodynamic equilibrium.
The parameters of the stars are given in Table 2. The parameters Lines of the eighth (λ 6743-6757 Å) and tenth (λ 6046-6052 Å)
of the S I lines used in the H-range are shown in Table 1. The multiplets in dwarfs have the smallest corrections and can be
oscillator strengths for lines of the two multiplets in the region used in LTE analysis. Non-LTE effects on these lines are signif-
of λ 15400 Å from BQZ and ZB have a systematic difference of icant only for giants. The same conclusion applies to all lines in
0.16 to 0.19 dex. As discussed in detail in Korotin et al. (2020), the H-range. Lines of the sixth (λ 8694 Å) multiplet are slightly
observed profiles for the Sun and the studied stars are better de- more sensitive to LTE deviations, while lines from the two IR
scribed when using the value log gf = BQZ − 0.12 dex. For triplets can only be used if non-LTE corrections are applied. The
the λ 22507-22707 Å lines, the BQZ and ZB oscillator strengths non-LTE correction grid is available in electronic form. An ex-
differ randomly, yielding similar average abundances. The cal- ample of these data is provided in Table 4, where the first column
culations used BQZ data. The results in Table 2 show that the lists the wavelength, followed by [Fe/H], [S/Fe], and T eff , with
abundances obtained from lines in the optical range agree rea- non-LTE corrections for five log g values in subsequent columns.
sonably well with those from the H-range. In general, the absolute value of corrections initially in-
creases with effective temperature, since the UV radiation field
6. NON-LTE CORRECTION GRID FOR NEUTRAL actively affects the redistribution of sulfur level populations.
SULFUR LINES However, upon reaching a maximum around 7000–8000 K, the
corrections start to decrease as more sulfur atoms become ion-
After verifying the atomic model and refining the parameters ized, sulfur lines weaken, and form deeper in the star’s atmo-
of S I lines suitable for sulfur abundance determination, we sphere where collisional processes become more influential. For
calculated a grid of non-LTE corrections. Unlike the studies stars with effective temperatures above 9500 K, non-LTE cor-
Article number, page 7 of 14
 manuscript Korotin_Kiselev

0.1 0.1 0.1

6052 A [Fe/H] = 0.0 6052 A [Fe/H] = 0.0 6052 A [Fe/H] = 0.0

[S/Fe]= -0.4 [S/Fe]= 0.0 [S/Fe]= +0.4

NLTE 0.0 0.0 0.0

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

6757 A [Fe/H] = 0.0 6757 A [Fe/H] = 0.0 6757 A [Fe/H] = 0.0

[S/Fe]= -0.4 [S/Fe]= 0.0 [S/Fe]= +0.4

0.0 0.0 0.0
NLTE

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

8694 A [Fe/H] = 0.0 8694 A [Fe/H] = 0.0 8694 A [Fe/H] = 0.0
-0.4 -0.4 -0.4
[S/Fe]= -0.4 [S/Fe]= 0.0 [S/Fe]= +0.4

-0.5 -0.5 -0.5

9212 A [Fe/H] = 0.0 9212 A [Fe/H] = 0.0 9212 A [Fe/H] = 0.0

0.0 0.0 0.0
[S/Fe]= -0.4 [S/Fe]= 0.0 [S/Fe]= +0.4

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 log g = 0
-0.6 -0.6
log g = 1

-0.8 log g = 2
-0.8 -0.8
log g = 3

-1.0 log g = 4
-1.0 -1.0
log g = 5

-1.2 -1.2 -1.2

0.0 0.0 0.0

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 -0.6 -0.6

-0.8 -0.8 -0.8

-1.0 10455 A [Fe/H] = 0.0 -1.0 -1.0

10455 A [Fe/H] = 0.0 10455 A [Fe/H] = 0.0
[S/Fe]= -0.4
[S/Fe]= 0.0 [S/Fe]= +0.4

-1.2 -1.2 -1.2

15403 A [Fe/H] = 0.0 15403 A [Fe/H] = 0.0

0.0 0.0 0.0
[S/Fe]= 0.0 [S/Fe]= +0.4

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15403 A [Fe/H] = 0.0

[S/Fe]= -0.4
-0.6 -0.6 -0.6

15469 A [Fe/H] = 0.0 15469 A [Fe/H] = 0.0

0.0 0.0 0.0
[S/Fe]= 0.0 [S/Fe]= +0.4

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15469 A [Fe/H] = 0.0

[S/Fe]= -0.4
-0.6 -0.6 -0.6

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

-0.4 -0.4 -0.4

22563 A [Fe/H] = 0.0 22563 A [Fe/H] = 0.0 22563 A [Fe/H] = 0.0

[S/Fe]= -0.4 [S/Fe]= 0.0 [S/Fe]= +0.4

-0.5 -0.5 -0.5

4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000
T (K) T (K) T (K)
eff eff eff

Fig. 7. Non-LTE corrections for metallicity [Fe/H] = 0

Article number, page 8 of 14

 Korotin and Kiselev: NLTE-effects of the Sulfur Abundances in A–K Type Stars

0.1 0.1 0.1

6052 A [Fe/H] = -0.5 6052 A [Fe/H] = -0.5 6052 A [Fe/H] = -0.5

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] =+0.8

NLTE 0.0 0.0 0.0

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

6757 A [Fe/H] = -0.5 6757 A [Fe/H] = -0.5 6757 A [Fe/H] = -0.5

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe]= +0.8

0.0 0.0 0.0
NLTE

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

8694 A [Fe/H] = -0.5 8694 A [Fe/H] = -0.5 8694 A [Fe/H] = -0.5
-0.4 -0.4 -0.4
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] = +0.8

-0.5 -0.5 -0.5

9212 A [Fe/H] = -0.5 9212 A [Fe/H] = -0.5 9212 A [Fe/H] = -0.5

0.0 0.0 0.0
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] = +0.8

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 -0.6 -0.6

-0.8 log g=0 log g=3 -0.8 -0.8

log g=1 log g=4

log g=2 log g=5

-1.0 -1.0 -1.0

-1.2 -1.2 -1.2

0.0 0.0 0.0

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 -0.6 -0.6

-0.8 -0.8 -0.8

10455 A [Fe/H] = -0.5

-1.0 -1.0 10455 A [Fe/H] = -0.5 -1.0
10455 A [Fe/H] = -0.5
[S/Fe] = 0.0
[S/Fe] =+0.4
[S/Fe]= +0.8
-1.2 -1.2 -1.2

0.0 0.0 0.0

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15403 A [Fe/H] = -0.5 15403 A [Fe/H] = -0.5

15403 A [Fe/H] = -0.5
-0.6 [S/Fe] = 0.0 -0.6 [S/Fe]=+0.4
-0.6
[S/Fe]= +0.8

0.0 0.0 0.0

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15469 A [Fe/H] = -0.5

15469 A [Fe/H] = -0.5 15469 A [Fe/H] = -0.5
-0.6 [S/Fe] = 0.0 -0.6 -0.6
[S/Fe] =+0.4 [S/Fe] =+0.8

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

-0.4 -0.4 -0.4

22563 A [Fe/H] = -0.5 22563 A [Fe/H] = -0.5 22563 A [Fe/H] = -0.5

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] =+0.8

-0.5 -0.5 -0.5

4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000
T (K) T (K) T (K)
eff eff eff

Fig. 8. Same as Fig.7 for [Fe/H]=-0.5

Article number, page 9 of 14

 manuscript Korotin_Kiselev

0.1 0.1 0.1

6052 A [Fe/H] = -1.0 6052 A [Fe/H] = -1.0 6052 A [Fe/H] = -1.0

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] =+0.8

NLTE 0.0 0.0 0.0

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

6757 A [Fe/H] = -1.0 6757 A [Fe/H] = -1.0 6757 A [Fe/H] = -1.0

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe]= +0.8

0.0 0.0 0.0
NLTE

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

8694 A [Fe/H] = -1.0 8694 A [Fe/H] = -1.0 8694 A [Fe/H] = -1.0
-0.4 -0.4 -0.4
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] = +0.8

-0.5 -0.5 -0.5

9212 A [Fe/H] = -1.0 9212 A [Fe/H] = -1.0 9212 A [Fe/H] = -1.0

0.0 0.0 0.0
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] = +0.8

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 -0.6 -0.6

-0.8 log g=0 log g=3 -0.8 -0.8

log g=1 log g=4

-1.0 log g=2 log g=5 -1.0 -1.0

-1.2 -1.2 -1.2

0.0 0.0 0.0

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 -0.6 -0.6

-0.8 -0.8 -0.8

10455 A [Fe/H] = -1.0

-1.0 -1.0 10455 A [Fe/H] = -1.0 -1.0
10455 A [Fe/H] = -1.0
[S/Fe] = 0.0
[S/Fe] =+0.4
[S/Fe]= +0.8
-1.2 -1.2 -1.2

0.0 0.0 0.0

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15403 A [Fe/H] = -1.0 15403 A [Fe/H] = -1.0

15403 A [Fe/H] = -1.0
-0.6 [S/Fe] = 0.0 -0.6 [S/Fe]=+0.4
-0.6
[S/Fe]= +0.8

0.0 0.0 0.0

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15469 A [Fe/H] = -1.0

15469 A [Fe/H] = -1.0 15469 A [Fe/H] = -1.0
-0.6 [S/Fe] = 0.0 -0.6 -0.6
[S/Fe] =+0.4 [S/Fe] =+0.8

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

-0.4 -0.4 -0.4

22563 A [Fe/H] = -1.0 22563 A [Fe/H] = -1.0 22563 A [Fe/H] = -1.0

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] =+0.8

-0.5 -0.5 -0.5

4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000
T (K) T (K) T (K)
eff eff eff

Fig. 9. Same as Fig.7 for [Fe/H]=-1.0

Article number, page 10 of 14

 Korotin and Kiselev: NLTE-effects of the Sulfur Abundances in A–K Type Stars

0.1 0.1 0.1

6052 A [Fe/H] = -1.5 6052 A [Fe/H] = -1.5 6052 A [Fe/H] = -1.5

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] =+0.8

NLTE 0.0 0.0 0.0

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

6757 A [Fe/H] = -1.5 6757 A [Fe/H] = -1.5 6757 A [Fe/H] = -1.5

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe]= +0.8

0.0 0.0 0.0
NLTE

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

8694 A [Fe/H] = -1.5 8694 A [Fe/H] = -1.5 8694 A [Fe/H] = -1.5
-0.4 -0.4 -0.4
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] = +0.8

-0.5 -0.5 -0.5

9212 A [Fe/H] = -1.5 9212 A [Fe/H] = -1.5 9212 A [Fe/H] = -1.5

0.0 0.0 0.0
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] = +0.8

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 -0.6 -0.6

-0.8 log g=0 log g=3 -0.8 -0.8

log g=1 log g=4

-1.0 log g=2 log g=5 -1.0 -1.0

-1.2 -1.2 -1.2

0.0 0.0 0.0

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 -0.6 -0.6

-0.8 -0.8 -0.8

-1.0 10455 A [Fe/H] = -1.5 -1.0 10455 A [Fe/H] = -1.5 -1.0 10455 A [Fe/H] = -1.5
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe]= +0.8
-1.2 -1.2 -1.2

0.0 0.0 0.0

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15403 A [Fe/H] = -1.5 15403 A [Fe/H] = -1.5

15403 A [Fe/H] = -1.5
-0.6 [S/Fe] = 0.0 -0.6 [S/Fe]=+0.4
-0.6
[S/Fe]= +0.8

0.0 0.0 0.0

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15469 A [Fe/H] = -1.5

15469 A [Fe/H] = -1.5 15469 A [Fe/H] = -1.5
-0.6 [S/Fe] = 0.0 -0.6 -0.6
[S/Fe] =+0.4 [S/Fe] =+0.8

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

-0.4 -0.4 -0.4

22563 A [Fe/H] = -1.5 22563 A [Fe/H] = -1.5 22563 A [Fe/H] = -1.5

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] =+0.8

-0.5 -0.5 -0.5

4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000
T (K) T (K) T (K)
eff eff eff

Fig. 10. Same as Fig.7 for [Fe/H]=-1.5

Article number, page 11 of 14

 manuscript Korotin_Kiselev

0.1 0.1 0.1

6052 A [Fe/H] = -2.0 6052 A [Fe/H] = -2.0 6052 A [Fe/H] = -2.0

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] =+0.8

NLTE 0.0 0.0 0.0

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

6757 A [Fe/H] = -2.0 6757 A [Fe/H] = -2.0 6757 A [Fe/H] = -2.0

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe]= +0.8

0.0 0.0 0.0
NLTE

-0.1 -0.1 -0.1

-0.2 -0.2 -0.2

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

8694 A [Fe/H] = -2.0 8694 A [Fe/H] = -2.0 8694 A [Fe/H] = -2.0
-0.4 -0.4 -0.4
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] = +0.8

-0.5 -0.5 -0.5

9212 A [Fe/H] = -2.0 9212 A [Fe/H] = -2.0 9212 A [Fe/H] = -2.0

0.0 0.0 0.0
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] = +0.8

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 -0.6 -0.6

-0.8 log g=0 log g=3 -0.8 -0.8

log g=1 log g=4

-1.0 log g=2 log g=5 -1.0 -1.0

-1.2 -1.2 -1.2

0.0 0.0 0.0

-0.2 -0.2 -0.2

-0.4 -0.4 -0.4

NLTE

-0.6 -0.6 -0.6

-0.8 -0.8 -0.8

-1.0 10455 A [Fe/H] = -2.0 -1.0 10455 A [Fe/H] = -2.0 -1.0 10455 A [Fe/H] = -2.0
[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe]= +0.8
-1.2 -1.2 -1.2

0.0 0.0 0.0

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15403 A [Fe/H] = -2.0 15403 A [Fe/H] = -2.0

15403 A [Fe/H] = -2.0
-0.6 [S/Fe] = 0.0 -0.6 [S/Fe]=+0.4
-0.6
[S/Fe]= +0.8

0.0 0.0 0.0

-0.2 -0.2 -0.2

NLTE

-0.4 -0.4 -0.4

15469 A [Fe/H] = -2.0

15469 A [Fe/H] = -2.0 15469 A [Fe/H] = -2.0
-0.6 [S/Fe] = 0.0 -0.6 -0.6
[S/Fe] =+0.4 [S/Fe] =+0.8

0.0 0.0 0.0

-0.1 -0.1 -0.1

NLTE

-0.2 -0.2 -0.2

-0.3 -0.3 -0.3

-0.4 -0.4 -0.4

22563 A [Fe/H] = -2.0 22563 A [Fe/H] = -2.0 22563 A [Fe/H] = -2.0

[S/Fe] = 0.0 [S/Fe] =+0.4 [S/Fe] =+0.8

-0.5 -0.5 -0.5

4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000 4000 5000 6000 7000 8000 9000
T (K) T (K) T (K)
eff eff eff

Fig. 11. Same as Fig.7 for [Fe/H]=-2.0

Article number, page 12 of 14

 Korotin and Kiselev: NLTE-effects of the Sulfur Abundances in A–K Type Stars

Table 4. Grid of non-LTE corrections for S I lines

λ, Å [Fe/H] [S/Fe] Te f f log g=0 log g=1 log g=2 log g=3 log g=4 log g=5
6052.7 0.0 -0.4 4000
6052.7 0.0 -0.4 4250 -0.06
6052.7 0.0 -0.4 4500 -0.06 -0.04
6052.7 0.0 -0.4 4750 -0.06 -0.04 -0.03
6052.7 0.0 -0.4 5000 -0.07 -0.05 -0.03
6052.7 0.0 -0.4 5250 -0.07 -0.05 -0.03 -0.02
6052.7 0.0 -0.4 5500 -0.07 -0.06 -0.04 -0.02
6052.7 0.0 -0.4 5750 -0.06 -0.07 -0.04 -0.02 -0.01
6052.7 0.0 -0.4 6000 -0.06 -0.07 -0.05 -0.03 -0.01
6052.7 0.0 -0.4 6250 -0.07 -0.06 -0.03 -0.02
6052.7 0.0 -0.4 6500 -0.07 -0.07 -0.04 -0.02 -0.02
6052.7 0.0 -0.4 6750 -0.07 -0.07 -0.05 -0.03 -0.01
6052.7 0.0 -0.4 7000 -0.07 -0.08 -0.06 -0.03 -0.02
6052.7 0.0 -0.4 7250 -0.07 -0.08 -0.06 -0.04 -0.02
6052.7 0.0 -0.4 7500 -0.08 -0.08 -0.07 -0.04 -0.02
6052.7 0.0 -0.4 7750 -0.08 -0.07 -0.04 -0.03
6052.7 0.0 -0.4 8000 -0.09 -0.07 -0.05 -0.03
6052.7 0.0 -0.4 8250 -0.07 -0.05 -0.03
... ... ... ... ... ... ... ... ... ...

rections for most lines in dwarf stars become minimal or may 7. CONCLUSIONS
even change sign. However, sulfur lines in such stars also be-
come very weak. The correction value is also highly dependent We have proposed a grid of correction to take into account the
on sulfur abundance, as it affects the intensity of sulfur spectral departures from LTE in neutral sulfur lines in the visible and
lines and, consequently, the depth at which they form. infrared spectral regions, including the H-range. The grid was
calculated using a sulfur atomic model incorporating the most
up-to-date atomic data on collisional rates with electrons and hy-
Applying non-LTE corrections to determine abundances is drogen. Inclusion of ionized sulfur levels and transitions in the
generally reliable when the absolute correction value is less atomic model has allowed us to extend the grid’s range of effec-
than 0.20–0.25 dex. However, it should be understood that the tive stellar photospheric temperatures up to 10000 K. The atomic
greater the correction, the larger its potential error due to com- model has proven its adequacy across a wide range of stellar pa-
plex dependences on microturbulent velocity, sulfur abundance, rameters: sulfur lines in all test stars showed consistent element
and other factors. Overall, the computed correction grid can be abundances regardless of non-LTE effects. The wavelengths and
used to assess the degree of non-LTE effects on sulfur lines. For oscillator strengths for several multiplets were refined. A rec-
lines with larger non-LTE corrections, individual calculations of ommended list of S I lines for determining sulfur abundance has
both b-factors and synthetic spectra are advisable. been compiled. The recommended sulfur line parameters repre-
sent a compromise between observed stellar spectra and theoreti-
If we compare our non-LTE calculations with those based cal calculations of wavelengths and oscillator strengths. The best
on the 2009 atomic model (Korotin et al. 2017), the corrections option for obtaining precise line parameters would be laboratory
for stars with near-solar metallicity are quite similar between measurements for all sulfur lines used in the analysis.
the two models. For metal-poor stars, the new model generally
shows slightly smaller departures from LTE. For instance, us- 8. SUPPLEMENTARY INFORMATION
ing the old atomic model, for the λ 9212 Å line for a star with
parameters [Fe/H] = -1.0, log g = 2, and T eff = 6000 K the cor- The online version contains supplementary material available at
rection was approximately –0.85 dex, whereas the new model https://doi.org/10.1134/S1063772924700987.
yields a smaller value: –0.75 dex. For dwarfs with log g = 4, the
difference is smaller: ˘0.39 and -0.34, respectively. This is due
References
to the use of detailed quantum-mechanical calculations of colli-
sional rates in the new model instead of approximate formulas Allen, C. W. 1973, Astrophysical quantities (University of London, Athlone
from model Korotin (2009). It is worth noting that there can be Press)
Asplund, M., Amarsi, A. M., & Grevesse, N. 2021, A&A, 653, A141
substantial differences between our data and calculations from Barklem, P. S. & O’Mara, B. J. 1997, MNRAS, 290, 102
Takeda et al. (2005). Specifically, for the λ 9212 Å line in a star Barklem, P. S., O’Mara, B. J., & Ross, J. E. 1998, MNRAS, 296, 1057
with parameters [Fe/H] = -1.0 and T eff = 6000 K, the calcu- Belyaev, A. K. & Voronov, Y. V. 2020, ApJ, 893, 59
lations in Takeda et al. (2005) yield corrections of –1.02 for a Biemont, E., Quinet, P., & Zeippen, C. J. 1993, A&AS, 102, 435
Carlsson, M. 1986, Uppsala Astronomical Observatory Reports, 33
giant (log g = 2) and –0.26 dex for a dwarf (log g = 4).The study Castelli, F. & Kurucz, R. L. 2003, in Modelling of Stellar Atmospheres, ed.
Takada-Hidai & Takeda (1996) provides non-LTE correction es- N. Piskunov, W. W. Weiss, & D. F. Gray, Vol. 210, A20
timates for lines of the sixth multiplet (λ 8694 Å) in A-stars with Cunto, W., Mendoza, C., Ochsenbein, F., & Zeippen, C. J. 1993, A&A, 275, L5
Deb, N. C. & Hibbert, A. 2008, Atomic Data and Nuclear Data Tables, 94, 561
log g = 4 and T eff from 8000 to 12000 K. The correction values Drawin, H.-W. 1968, Zeitschrift fur Physik, 211, 404
for temperatures below 9500 K are similar to ours, but at higher Froese Fischer, C., Tachiev, G., & Irimia, A. 2006, Atomic Data and Nuclear
temperatures, they remain negative, albeit small in magnitude. Data Tables, 92, 607

Article number, page 13 of 14

 manuscript Korotin_Kiselev

Galazutdinov, G. A. 2022, Astrophysical Bulletin, 77, 519

Hirata, R. & Horaguchi, T. 1994, VizieR Online Data Catalog: Atomic Spec-
tral Line List (Hirata+ 1995), VizieR On-line Data Catalog: VI/69. Origi-
nally published in: Department of Astronomy, Faculty of Science, Kyoto Uni-
versity, and, National Science Museum, 3-23-1 Hyakunin-cho, Shinjuku-ku,
Tokyo
Irimia, A. & Froese Fischer, C. 2005, Phys. Scr, 71, 172
Israelian, G. & Rebolo, R. 2001, ApJ, 557, L43
Jofré, P., Heiter, U., Soubiran, C., et al. 2015, A&A, 582, A81
Kacharov, N., Koch, A., Caffau, E., & Sbordone, L. 2015, A&A, 577, A18
Kobayashi, C., Karakas, A. I., & Lugaro, M. 2020, ApJ, 900, 179
Korotin, S., Andrievsky, S., Caffau, E., & Bonifacio, P. 2017, in Astronomi-
cal Society of the Pacific Conference Series, Vol. 510, Stars: From Collapse
to Collapse, ed. Y. Y. Balega, D. O. Kudryavtsev, I. I. Romanyuk, & I. A.
Yakunin, 141
Korotin, S. A. 2009, Astronomy Reports, 53, 651
Korotin, S. A., Andrievsky, S. M., Caffau, E., Bonifacio, P., & Oliva, E. 2020,
MNRAS, 496, 2462
Korotin, S. A., Andrievsky, S. M., & Luck, R. E. 1999, A&A, 351, 168
Kramida, A., Ralchenko, Y., & Reader, J. 2012, in APS Meeting Abstracts,
Vol. 43, APS Division of Atomic, Molecular and Optical Physics Meeting
Abstracts, D1.004
Kurucz, R. L., Furenlid, I., Brault, J., & Testerman, L. 1984, Solar flux atlas from
296 to 1300 nm (National Solar Observatory, Sunspot, New Mexico)
Limongi, M. & Chieffi, A. 2003, ApJ, 592, 404
Lodders, K. 2021, Space Sci. Rev., 217, 44
Lyubimkov, L. S., Lambert, D. L., Rostopchin, S. I., Rachkovskaya, T. M., &
Poklad, D. B. 2010, MNRAS, 402, 1369
Martin, W. C., Zalubas, R., & Musgrove, A. 1990, Journal of Physical and Chem-
ical Reference Data, 19, 821
Mészáros, S., Allende Prieto, C., Edvardsson, B., et al. 2012, AJ, 144, 120
Origlia, L., Oliva, E., Baffa, C., et al. 2014, in Society of Photo-Optical Instru-
mentation Engineers (SPIE) Conference Series, Vol. 9147, Ground-based and
Airborne Instrumentation for Astronomy V, ed. S. K. Ramsay, I. S. McLean,
& H. Takami, 91471E
Petit, P., Louge, T., Théado, S., et al. 2014, PASP, 126, 469
Reiners, A., Mrotzek, N., Lemke, U., Hinrichs, J., & Reinsch, K. 2016, A&A,
587, A65
Romanovskaya, A. M., Ryabchikova, T. A., Pakhomov, Y. V., Korotin, S. A., &
Sitnova, T. M. 2023, MNRAS, 526, 3386
Ryabchikova, T., Piskunov, N., Kurucz, R. L., et al. 2015, Phys. Scr, 90, 054005
Seaton, M. J. 1962, in Atomic and Molecular Processes, ed. D. R. Bates (New
York: Academic Press)
Spite, M., Caffau, E., Andrievsky, S. M., et al. 2011, A&A, 528, A9
Steenbock, W. & Holweger, H. 1984, A&A, 130, 319
Summers, H. P. & O’Mullane, M. G. 2011, in American Institute of Physics
Conference Series, Vol. 1344, 7th International Conference on Atomic and
Molecular Data and Their Applications - ICAMDATA-2010, ed. A. Bernotas,
R. Karazija, & Z. Rudzikas (AIP), 179–187
Takada-Hidai, M. & Takeda, Y. 1996, PASJ, 48, 739
Takeda, Y., Hashimoto, O., Taguchi, H., et al. 2005, PASJ, 57, 751
Tayal, S. S. & Zatsarinny, O. 2010, ApJS, 188, 32
Tsymbal, V., Ryabchikova, T., & Sitnova, T. 2019, in Astronomical Society of
the Pacific Conference Series, Vol. 518, Physics of Magnetic Stars, ed. D. O.
Kudryavtsev, I. I. Romanyuk, & I. A. Yakunin, 247–252
Tull, R. G., MacQueen, P. J., Sneden, C., & Lambert, D. L. 1995, PASP, 107, 251
van Regemorter, H. 1962, ApJ, 136, 906
Zatsarinny, O. & Bartschat, K. 2006, Journal of Physics B Atomic Molecular
Physics, 39, 2861
Zerne, R., Caiyan, L., Berzinsh, U., & Svanberg, S. 1997, Phys. Scr, 56, 459

Article number, page 14 of 14