Unique distant classical Cepheid OGLE GD-CEP-1353 with anomalously high abundances of s- and r-process elements

While looking for recently discovered distant Cepheids with an interesting chemical composition, we noticed one star (OGLE GD-CEP-1353) with extremely large equivalent widths of spectral lines of heavy elements. The aim of this work is to perform an abundance analysis, and to find a possible explanation for the found chemical anomaly. Quantitative analysis of the equivalent widths and synthetic spectrum synthesis were used to derive abundances in this star. Both local and nonlocal thermodynamic equilibrium (LTE and NLTE) approximations were used in our analysis. Abundances of 28 chemical elements from carbon to thorium were derived. While light and iron peak elements show abundances typical for distant Cepheids (located in the outer disk), the s-process elements are overabundant about one dex. r-process elements are slightly less overabundant. This makes the star a unique Cepheid of our Galaxy.


Introduction
Cepheids are yellow supergiant stars of spectral classes F to K, which are at an evolutionary stage when they are crossing the instability strip in the Hertzsprung-Russell diagram (HRD).During the crossing time, these stars pulsate in radial modes.
Observationally, their atmospheres do not exhibit remarkable chemical peculiarities.A few anomalies that are an inherent feature of Cepheids are connected to their internal structure.After hydrogen depletion in the stellar core, it contracts and heats.A shell source, where fresh hydrogen is converted into helium, develops.At this evolutionary stage, the star expands and settles in the red giant branch (RGB) region.Here the first dredge-up occurs (Iben 1967).To the surface of the star, it brings material processed in the incomplete CNO cycle.Here, carbon is partially converted into nitrogen, and if the first dredge-up happens before nitrogen is converted into oxygen, one can detect that the surface abundances of carbon and nitrogen are altered: carbon is depleted and nitrogen is enhanced (see, for example, the recent large-scale study of the Cepheids' chemical properties by Luck 2018, Fig. 19).This phenomenon is not observed in the F and G unevolved dwarfs (e.g., Reddy et al. 2003).The next stage after the RGB evolution is determined by the helium burning in the stellar core.As the helium in the core ignites in a smooth regime, the star contracts and increases its effective temperature.Thus, the star starts to perform a blue loop in the HRD.At this time interval, the star can be noticed as a Cepheid.This scenario is well known, and its description can be found in many literature sources.What is important is that the Cepheid atmospheres preserve the C-N abundance anomalies.This is a wellestablished observational fact (see, for example, Lambert 1981, Luck & Lambert 1985, Luck 2018 and references therein).Cepheid stars show a slightly increased abundance of sodium (e.g., Andrievsky et al. 2003, Luck 2018).Elements that are formed through the capture of free thermalized neutrons by seed nuclei (s-process elements) normally do not show significantly increased abundances, and this is clear from the evolutionary point of view.The internal structure of Cepheids does not provide a neutron source.Nearly normal relative abundances of the neutron capture elements ([s/Fe]) in Cepheids were reported recently by da Silva et al. (2016) and Luck (2018).As a rule, all of the studied heavy element abundances, including r-process elements such as Eu, are in the range from zero to 0.2 dex (s/Fe).For zirconium and lanthanum, abundances are 0.1-0.2dex higher but, according to Luck (2018), with a big error bar.
In summary, any anomalies of heavy element abundances that can be found in a Cepheid deserve special attention.In this paper we report our detection of remarkable overabundances of s-and r-process elements in the distant classical Cepheid OGLE GD-CEP-1353.Its characteristics are given in Table 1 (together with atmospheric parameters described in the next Section).The high-resolution spectrum of this star was obtained using the Ultraviolet and Visual Echelle Spectrograph (UVES; Dekker et al. 2000) at the Very Large Telescope of ESO at Paranal (Chile).The resolving power and signal-to-noise ratio (S/N) are 42 300 and 34, respectively.Skowron et al. (2019) give the heliocentric (d) and Galactocentric (R G ) distances that we list in Table 1.We note that from the Gaia DR3 (Data Release 3) database OGLE GD-CEP-1353 has a parallax of 0.0815 mas, which means its heliocentric distance is about 12 kpc, being close to above the mentioned value.The remainder of this paper is organized as follows.In Sect.2, we describe our abundance analysis of OGLE GD-CEP-1353.In Sect.3, we discuss our results and present our conclusions.

Abundance results
The first attempt to derive abundances in this star was recently made by Trentin et al. (2023;ASASSN-V J074354.86-323013.7).Their paper lists abundance results for 24 elements from carbon to neodymium for a large number of Cepheids (abundances of only 18 elements were determined for this star; from heavy species only for zirconium and barium).Unfortunately, this unique star was apparently "lost" in the large number of the program stars, and the authors did not pay due attention and did not discuss its individual chemical properties.

Atmosphere parameters
The effective temperature, T eff , was derived from the line-depth ratios (Kovtyukh 2007), a technique commonly employed in studies of Cepheid variables (e.g., Andrievsky et al. 2016;Luck 2018;Lemasle et al. 2013;da Silva et al. 2022;Kovtyukh et al. 2022).Once T eff is determined, the surface gravity (log g) is found by imposing the iron ionization balance (the same iron abundance derived from the lines of neutral and ionized iron).The microturbulent velocity, V t , was derived assuming that there is no dependence between the iron abundance, obtained from Fe I lines, and the equivalent widths (EWs) of the same lines (Fig. 1).The adopted value [Fe/H] is that, which is derived from the Fe I lines, since we assumed the ionization balance and because they outnumber Fe II lines.The atmospheric parameters T eff , log g, and V t are listed in Table 1.Since Trentin et al. (2023) used practically the same methods of the atmosphere parameters' determination, we have very close results on the temperature, gravity, and microturbulent velocity.

LTE results
The abundances of different elements were derived in the LTE approximation using atmosphere models interpolated for the atmosphere parameters within the grid of ATLAS9 models by Castelli & Kurucz (2003).We discarded strong lines (with EWs > 150 mÅ) due to noticeable damping effects.The list of the lines of the heavy elements measured in the spectrum of our program star is given in Table A.1.The oscillator strengths, log g f , were adopted from the Vienna Atomic Line Database (VALD, version 2023, Ryabchikova et al. 2015).The reference solar abundances were taken from Asplund et al. (2009) or determined by us in the NLTE calculations.
The abundances of some heavy elements were calculated via direct fitting observed and synthetic profiles of individual spectral lines.We used the SynthV code of Tsymbal et al. (2019) in combination with the ATLAS9 model.LTE approximation was assumed in this spectrum synthesis.Some examples of synthetic spectra fragments are shown in Figs. 2 and 3.All of the results are given in Table 2.
The general method of the NLTE calculations we used in our previous papers is the following.In order to find atomic level populations for the ions of interest, we employed the code MULTI (Carlsson 1986).For our aim, this program was modified by Korotin et al. (1999).MULTI allows one to calculate a single NLTE line profile.If the line of interest is blended, we performed the following procedure.With the help of MULTI, we first calculated the departure coefficients for those atomic levels that are responsible for the formation of the considered line.Then we included these coefficients in the LTE synthetic spectrum code SYNTHV (Tsymbal et al. 2019).This allowed us to calculate the source function and opacity for each studied line.Simultaneously, the blending lines were calculated in LTE with the help of the line list and corresponding atomic data from the VALD database (Ryabchikova et al. 2015) in the wavelength range of the line under study.For all our computations, we used 1D LTE atmosphere models computed with the ATLAS9 code by Castelli & Kurucz (2003).
Table 2 shows the averaged individual chemical abundances and their uncertainties.We note that abundance results may suffer from a combined influence of the errors in T eff , log g, V t , the position of the continuum, and the accuracy of the log g f .Abundances of those elements that are represented in the spectrum by less than three lines must be taken with caution.
The program star abundance pattern that referred to solar abundances is also presented in Fig. 4. The extremely high barium abundance might not be realistic due to the very strong lines of this element.These lines are effectively formed in the upper  layers of the Cepheid atmosphere, where the microturbulent velocity can significantly exaggerate the value found from the iron lines.We could not account for this possible phenomenon, so we simply state that the barium abundance from our analysis may be overestimated.
As mentioned, all derived abundances were normalized to the iron content in our program star.It should be noted that the iron content in Cepheids shows the expected dependence upon Galactocentric distance: the larger the distance, the lower the iron content.For instance, this is shown in Andrievsky et al. (2016), Luck (2018), Trentin et al. (2023), da Silva et al. (2023), and the references therein.From these data, it can be seen that the distant Cepheids in the outer Galactic disk have an iron content 2-3 times lower than in the Sun (see, for example, Fig. 23 in Luck 2018 based on period-luminosity distances).The distance to our program Cepheid is mentioned in Table 1.Cepheids at such a distance show absolute iron content from 7.0 to 7.4 (7.5 for the Sun), and thus our program star fits in this range.

Discussion and conclusion
First of all, it should be noted that none of the Cepheids of our Galaxy studied to date show such an abnormal chemical composition.In some sense, it resembles the chemical composition of the chemically peculiar (CP) stars (Ryabchikova et al. 1997, Yushchenko et al. 2008), or even Przybylski's star (Shulyak et al. 2010), of course, in a much lesser extent.It is believed that CP stars gain their anomalies from atomic diffusion in the dynamically stable atmospheres (Michaud 1973).This cannot be the case for yellow supergiants with atmospheric convection.As to Przybylski's star, several hypotheses exist explaining its extreme peculiarities (for more details, readers can refer to the overview in Andrievsky 2022).In addition, this author considered processes in a binary system consisting of Przybylski's star and a neutron star, which is the source of high-energy γ radiation, which may affect the atmosphere of Przybylski's star.
Studying the derived abundance distribution in our program star, we must describe a few characteristic features of it (Fig. 4), namely, the deficiency of iron-peak elements; the rather high relative-to-iron abundance of carbon; the increased sodium abundance, for example for the Cepheids with a pulsational period of 3 days, Genovali et al. (2015) give an average sodium overabundance [Na/Fe] of about 0.3 dex, while our program star shows [Na/Fe] = 0.6 dex; and a significantly increased abundance of the s-process elements (elements that formed as a result of the slow neutron capture by seed nuclei).In addition, the following remarks can be made: there is a fairly high level of overabundance of the light s-process elements (Y, Zr); and with europium, lutetium, and thorium, being r-process elements (rapid neutron capture), they have apparently high abundances, too.We note that according to da Silva et al. (2016, their Fig.6), with the relative-to-iron europium abundance [Eu/Fe] being extrapolated to a distance of about 12 kpc, it is about zero, while the corresponding value for our program star is 0.6 dex.
As we mentioned in the Introduction, Cepheid atmospheres exhibit the results of the dredge-up event.From the observational point of view, the result of the first dredge-up event was described by Lambert (1981) and Luck & Lambert (1985).After dredge-up carbon becomes deficient, while nitrogen is overabundant.The region of the spectrum of our program star that is available to us does not contain nitrogen lines, so we cannot A140, page 3 of 11 Kovtyukh, V. V., et al.: A&A, 679, A140 (2023) Luck (2018).As it was already mentioned in the Introduction, all studied s-process elements normally do not show significant excesses in the relative-to-iron abundances.If we compare our results on s-process abundances presented in Table 2 with those given in the two abovementioned papers, we would see that Cepheid OGLE GD-CEP-1353 has an overabundance of these elements by several times larger than any studied Cepheid until now (more than 500 stars).It makes this object really unique in our Galaxy.Below we consider several possible hypotheses that may help to understand its unique chemical properties.Our program star OGLE GD-CEP-1353 has a light curve typical of classical Cepheids (Udalski et al. 2015).Additionally, its Galactic latitude is small (Table 1), which means it can be a thin disk star.Since its fundamental pulsational period is about 3.15 days, according to Turner (1996; mass-pulsation period relation), its mass should be about 4 solar masses.The lifetime of a star of this mass (B-star progenitor + following a Cepheid stage), t ≈ 10 10 ( M ⊙ M * ) 2.5 yr, should be about 3.7 × 10 8 yr (the core heliumburning stage takes about 25-30% of the main-sequence time, see, e.g., Chiosi 1990).According to the age-pulsation period relation by Bono et al. (2005), the age of a Cepheid is 1.1 ×10 8 yr (for [Fe/H]= -0.5).
1. Suppose now that this Cepheid had a companion of a slightly higher mass.After a certain time, such a companion ends its evolution as an asymptotic giant branch (AGB) star.Having finished this evolutionary stage, the envelope of this rather massive AGB star (more than 4 solar masses) could contaminate the present Cepheid's atmosphere (or its B-star progenitor), enriching it in s-process elements.The more massive component after all becomes a C-O white dwarf.This scenario can explain the appearance of the s-process elements in the Cepheid's atmosphere.More massive AGB stars produce a larger amount of the weak s-process elements, such as yttrium and zirconium, and this is seen in our program Cepheid (see, Fig. 4).
According to Goswami & Goswami (2023), see their Fig.14, Panel a, the AGB model of 4 solar masses with a metallicity of about -0.5 (recall that the iron peak elements in our program star show the same abundance level) yield enough the s-process elements, and the light s-process elements (Sr, Y, and Zr) are at the same abundance level as the heavy s-process elements (Ba-Sm), [El/Fe] ≈0.8-0.9.This is exactly what we see in our program Cepheid.It should be noted that in the lower-mass AGB stars, the light s-process elements are less abundant compared to the heavy s-process elements.Finally, as is mentioned above, sodium is apparently increased in our program star.This may be connected to the fact that the highest sodium production is reached in 4 solar mass AGB stars (Di Criscienzo et al. 2016).
However, despite its attractivity, this hypothesis faces a problem with unexpectedly high abundances of the typical r-process elements, such as europium, lutetium, and thorium.As believed, the nuclei of these heavy elements cannot be formed inside the AGB star because of the insufficient neutron flux.The typical neutron density in the low-mass AGB stars is about 10 7 cm −3 (Straniero et al. 2006).
2. This Cepheid has a companion of a significantly higher mass.In this case its companion ended its evolution as a neutron star after the Supernova II (SN II) explosion.For this to happen, the B-star progenitor must have a mass larger than 8 solar masses.If it was a close explosion, the possibility of a single contamination event would be possible.However, there is one weak place in this hypothesis.Even for the lower limit of the SN II mass, the expected lifetime of its progenitor B star is about 6 × 10 7 yr.Soon after this time, the star explodes.Its companion (presently a Cepheid) is still evolving as a late B star.If the atmosphere of this B star is contaminated, then after the following evolution to the red giant stage and first dredge-up, all of the material synthesized in the SN explosion that contaminated the B star atmosphere will be mixed with the stellar interior.Thus, the surface abundance anomalies will hardly survive.
3. Finally, suppose that the OGLE GD-CEP-1353's more massive companion ended its evolution as a neutron star.If this neutron star is a source of γ radiation and if this radiation is directed toward the present-day Cepheid atmosphere, then one can expect that irradiation of the atmosphere gas will produce free neutrons.Such free neutrons (provided their flux is sufficiently high) can lead to the production of the s-and r-process elements.A similar mechanism was proposed by Andrievsky (2022) to explain the phenomenon of a chemically peculiar Przybylski's star (see details in that paper).
An additional question may arise concerning the first and third scenario.How long can the contaminated atmosphere of the Cepheid star preserve the chemical composition abundance anomalies?To estimate the characteristic time, we used the following formula (Sweet 1950): t = 8 × 10 12 M 3 LR 4 1 Ω 2 yr.All values here are in solar units.As previously mentioned, the mass of our program star with its pulsation period of about 3 days, is about 4 solar masses.According to Gieren et al. (1999), a Cepheid with such a period has a radius of about 30 solar radii.The period -luminosity relation gives the luminosity of this star of about 10 3 L ⊙ .To determine the angular rotation velocity of the Cepheid (the radii of the Cepheid and the Sun are known), we adopted a typical radius of its progenitor, the main-sequence B star, to be about 3.3 R ⊙ , and a linear equatorial velocity about 200 km s −1 (a very rough estimate according to Brott et al. 2011 gives v sin i of about 100 km s −1 for stars of about 10 solar masses, see their Fig. 1).Then, after expanding from 3 to 30 R ⊙ , the rotational velocity of the Cepheid decreased to a few km s −1 (approximately the same as the solar rotational rate).All adopted values give us a characteristic time of the large-scale meridional circulation of about 5 × 10 8 yr or more.This value exceeds our program Cepheid lifetime (see the estimate above).
To evaluate the validity of the hypothesis described above, a necessary condition has to be met.One needs to prove that OGLE GD-CEP-1353 is a component of a binary system with a secondary component being a compact object.Regarding the OGLE GD-CEP-1353 visual magnitude, it seems to be impossible now.
In summary, we have described a very unusual chemical composition of this star, in particular the high abundance of heavy elements in its atmosphere.It may be a component of a binary system originally consisting of two stars with slightly or moderately different masses.The further evolution of the stars in this system led to contamination of the OGLE GD-CEP-1353 atmosphere with heavy elements, and thus forming in this way its unique chemical properties.

Fig. 2 .
Fig. 2. Observed (open circles) and LTE synthetic profiles of Y II 5728.89Å, Eu II 6645.10Å, Lu II 6221.89Å, and Th II 5989.045Å lines (solid line).The dashed line indicates no Y, Eu, Lu, or Th.The solid lines show the abundance variation of ±0.20 dex for these elements and the best fit.

Fig. 3 .
Fig. 3. NLTE profiles of some lines of C, Na, and Ca (solid line) compared to the observed profiles (open circles).LTE profiles were calculated with the same abundances as NLTE profiles (dotted line).

Fig. 4 .
Fig. 4. Graphical representation of the content of Table 2.

Table 2 .
Silva et al. (2016) and)in OGLE GD-CEP-1353.Notes.NL is the number of lines used.(El/H) is the absolute abundance value on the scale where the hydrogen abundance is 12.00.provethisfact.Yellow supergiant stars and Cepheids, in particular, show a moderate overabundance of sodium(Andrievsky et al. 2003), which can be a sign of the neon-sodium cycle operation.OurNLTEcalculations of the sodium abundance in this star confirm this fact.However, increased sodium in our program Cepheid may have another origin.Nearly normal relative abundances of the neutron capture elements ([s/Fe]) in Cepheids were recently reported by daSilva et al. (2016) and