CAPOS: The bulge Cluster APOgee Survey IV. Elemental Abundances of the bulge globular cluster NGC 6558

This study presents the results concerning six red giant stars members of the globular cluster NGC 6558. Our analysis utilized high-resolution near-infrared spectra obtained through the CAPOS initiative (the APOgee Survey of Clusters in the Galactic Bulge), which focuses on surveying clusters within the Galactic Bulge, as a component of the Apache Point Observatory Galactic Evolution Experiment II survey (APOGEE-2). We employ the BACCHUS (Brussels Automatic Code for Characterizing High accUracy Spectra) code to provide line-by-line elemental-abundances for Fe-peak (Fe, Ni), $\alpha$-(O, Mg, Si, Ca, Ti), light-(C, N), odd-Z (Al), and the $s$-process element (Ce) for the 4 stars with high signal-to-noise ratios. This is the first reliable measure of the CNO abundances for NGC 6558. Our analysis yields a mean metallicity for NGC 6558 of $\langle$[Fe/H]$\rangle$ = $-$1.15 $\pm$ 0.08, with no evidence for a metallicity spread. We find a Solar Ni abundance, $\langle$[Ni/Fe]$\rangle$ $\sim$ $+$0.01, and a moderate enhancement of $\alpha$-elements, ranging between $+$0.16 to $<+$0.42, and a slight enhancement of the $s$-process element $\langle$[Ce/Fe]$\rangle$ $\sim$ $+$0.19. We also found low levels of $\langle$[Al/Fe]$\rangle \sim $+$0.09$, but with a strong enrichment of nitrogen, [N/Fe]$>+$0.99, along with a low level of carbon, [C/Fe]$<-$0.12. This behaviour of Nitrogen-Carbon is a typical chemical signature for the presence of multiple stellar populations in virtually all GCs; this is the first time that it is reported in NGC 6558. We also observed a remarkable consistency in the behaviour of all the chemical species compared to the other CAPOS bulge GCs of the same metallicity.


INTRODUCTION
The Milky Way bulge is a centrally concentrated component that has an associated system of globular clusters (Minniti 1995).Glob-★ E-mail: danilo.gonzalez@ucn.clular clusters (GCs) in the central region (≲ 3.5 kpc; Barbuy et al. 2018a,b) of the Galactic Bulge (GB) could contain the oldest stars in the Galaxy.Indeed, the rapid chemical evolution in the deep potential well of the GB means that GCs could be potentially older than their halo cousins, which were born and raised in much less massive satellites (Cescutti 2008; Barbuy et al. 2018b;Savino et al. highly reddened regions.This is indeed the goal of the CAPOS project (Geisler et al. 2021).
In this paper, we present -band spectroscopy for six members of NGC 6558, using spectra from the bulge Cluster APOgee Survey (CAPOS, Geisler et al. 2021), an internal program of the Apache Point Observatory Galactic Evolution Experimental II survey (APOGEE-2, Majewski et al. 2017) and the Sloan Digital Sky IV survey (Blanton et al. 2017;Ahumada et al. 2020).These spectra make it possible to derive detailed elemental abundances for a large number of giant stars, as well as to achieve RVs with precision better than 1 km s −1 .
This paper is organized as follows: Section 2 and 3 present an overview of the CAPOS data analysed toward the NGC 6558 field, while Section 4 describes the adopted atmospheric parameters employed to determine the elemental abundances.The resulting elemental abundances obtained with the BACCHUS code are described in Section 5. We discuss our results and compare them with previous literature in Section 6.Our conclusions are presented in Section 7.

OBSERVATIONS AND DATA
The second phase of the Apache Point Observatory Galactic Evolution Experiment (APOGEE-2; Majewski et al. 2017) is a highresolution ( ∼ 22, 500) near-infrared (NIR) spectroscopic survey containing observations of ∼ 657, 135 unique stars released as part of the SDSS-IV survey (Blanton et al. 2017), targeting these objects with selections detailed in Zasowski et al. (2013), Zasowski et al. (2017), Beaton et al. (2021), and Santana et al. (2021).These papers additionally provide an extensive explanation of the targeting approach employed in the APOGEE-2 survey.The spectra were obtained using the cryogenic, multi-fiber (300) APOGEE spectrograph (Wilson et al. 2019), mounted on the 2.5m SDSS telescope (Gunn et al. 2006) at Apache Point Observatory, to observe the Northern Hemisphere (APOGEE-2N), and expanded to include a second APOGEE spectrograph on the 2.5m Irénée du Pont telescope (Bowen & Vaughan 1973) at Las Campanas Observatory to observe the Southern Hemisphere (APOGEE-2S).Each instrument registers the H-band in wavelengths between the 15100 Å and 17000 Å, utilizing three detectors.There are small gaps in coverage between them of about 100 Å around ∼15800 Å and ∼16400 Å , respectively.Additionally, each fiber provides an on-sky field of view with an diameter of ∼2 ′′ in the Northern instrument and 1. ′′ 3 in the Southern instrument.
The final version of the APOGEE-2 catalog was published in December 2021 as part of the 17 th data release of the Sloan Digital Sky Survey (DR17; Abdurro'uf et al. 2022b) and is available publicly online through the SDSS Science Archive Server and Catalog Archive Server 1 .The spectra were reduced following the procedures extensively described in Nidever et al. (2015).For detailed information, we refer the interested reader to that reference.The stellar parameters and chemical abundances in APOGEE-2 were derived within the APOGEE Stellar Parameters and Chemical Abundances Pipeline (ASPCAP; García Pérez et al. 2016).ASPCAP derives stellar-atmospheric parameters, radial velocities, and as many as 20 individual elemental abundances for each APOGEE-2 spectrum by comparing each to a multidimensional grid of theoretical MARCS model atmospheres grid (Zamora et al. 2015), employing a 2 minimization routine with the code FERRE (Allende Prieto et al. 2006) to derive the best-fit parameters for each spectrum.
Fig. 1 shows the spatial distribution of the stars toward the NGC 6558 field in the APOGEE-2S footprint (top-left panel), together with the six identified cluster members listed in this work.The main APOGEE-2S plug-plate containing the NGC 6558 cluster (toward  J2000 = 18 h 10 m 18.38 s and  J2000 = −31 d 45 m 48.6 s ) was centered on (l,b) ∼ (0.0, −0.6) degrees, and 15 out of 550 science fibers were located inside the cluster tidal radius (r  = 10.′ 44 ; Harris 1996Harris , 2010, dashed circumference in Fig. 1).
CAPOS targets were originally selected on the basis of Gaia DR2 (Gaia Collaboration et al. 2018) to have proper motions (PMs) within an approximate radius of ∼0.5 mas yr −1 around the nominal PMs of NGC 6558;   cos ()= −1.72 and   = −4.14(see, e.g.Baumgardt & Vasiliev 2021b).We initially selected these values to estimate candidate members from the PMs to reduce the potential field star contamination in both optical and NIR color-magnitude diagrams (CMDs).We re-examined the PMs from the original sample, but we did not find important differences between Gaia DR2 (Gaia Collaboration et al. 2018) and Gaia DR3 PMs (Gaia Collaboration et al. 2022) toward NGC 6558.The left-bottom panel in Fig. 1 shows the updated version of the PMs using Gaia DR3Gaia (Gaia Collaboration et al. 2022).Besides PMs, the selected stars are distributed along both the red giant branch (RGB) and the asymptotic giant branch (AGB) of the cluster, as illustrated in the bottom-right panel of Fig. 1.
Our target stars not only lie inside the tidal radius of the cluster, but also share similar PMs, follow the same evolutionary track in the optical+NIR CMDs, and share similar radial velocity (RV) and metallicity ([Fe/H], top-right panel), which confirm that the 5 CAPOS stars (the red open circles) are very likely cluster members.Both the RV and [Fe/H] were extracted from the spectra reduced with the ASPCAP pipeline, and these values in the metallicity were used only for the member selection.In fact, Fernández-Trincado et al. (2020b) shows that there is usually an offset of ∼0.2 dex on average between the abundances measured with ASPCAP pipeline with respect to the BACCHUS code in most of the chemical species.The methodology followed to determine the final values of the abundances is described in detail in Sections 4 and 5.In this sense, the zoomed black box rectangle in the top-right panel of the Fig. 1 shows the revised Fe abundances as black dashed open circles, together with the ASPCAP values (red open circles).The APOGEE-2 designations of the stars, the coordinates, Gaia DR3 and VVV magnitudes, RV, and S/N are listed in Table 1.
Additionally, we also examined two overlapping APOGEE-2S plug-plates toward NGC 6558, which contain the APOGEE-2S programs: kollmeier_17a (238 science fibers) and apogee (332 science fibers).We refer the interested reader to Santana et al. (2021) for further details on these programs.
From kollmeier_17a, we have identified one star (2M18101623−3145423) satisfying almost all the cluster membership criteria imposed in this work.However, this star is only recorded in Table 1 together with the APOGEE-2 RV, Gaia DR3 astrometry, and photometry information because its spectral S/N is too low (< 10) to obtain reliable elemental abundances.This star is positioned along the cluster CMD's horizontal branch (HB), and is likely associated with the RR Lyrae population of NGC 6558.
Finally, we imposed a quality cut in the APOGEE-2 spectra using an S/N threshold > 70 per pixel to ensure the analysis of goodquality spectra and maximize the precision of the stellar abundances.Thus, elemental abundances are restricted to the 4 stars with high S/N listed in Table 1.

ATMOSPHERIC PARAMETERS
The atmospheric parameters (T eff and log ) were derived using the same methodology as described in Fernández-Trincado et al. (2022a), that is, the   vs.  −   CMD shown in Fig. 1 (bottomright panel) was reddening corrected by using giant stars and adopting the reddening law of Cardelli et al. (1989) andO'Donnell (1994) and a total-to-selective absorption ratio RV = 3.1 (see, Barbuy et al. 2018b;Cohen et al. 2021).We chose the RGB stars located within a 10 ′ radius from the cluster, ensuring that their proper motions closely matched those of NGC 6558 within a margin of 0.5 mas yr −1 .Subsequently, following the reddening slope, we proceeded to determine the separation between these selected stars and a previously defined reference line, which outlines the RGB.The vertical and horizontal separation yields the differential absorption A K s and reddening E(BP − K s ) values of each individual star, respectively.Our next step involved the selection of the three closest RGB stars within the field to calculate the average absorption and reddening values.We then corrected the K s magnitude and BP − K s colour index based on assumed  eff and log g to be the effective temperature and gravity at the point of the isochrones that has the same K s magnitude as the star.Finally, microturbulence velocities   were determined using the relation from Gratton et al. (1996).Table 2 lists the determined photometric atmospheric parameters, microturbulence velocity, and S/N for the four stars in our sample for which elemental abundances are provided in this study.

ELEMENTAL ABUNDANCES
Elemental abundances were derived from a local thermodynamic equilibrium (LTE) analysis for the four stars listed in Table 2 using the Brussels Automatic Code for Characterizing High accuracy Spectra (Masseron et al. 2016), which relies on the radiative code Turbospectrum (Alvarez & Plez 1998;Plez 2012) and the MARCS model atmosphere grid Gustafsson et al. (2008).With the atmospheric parameters determined in Section 4, the first step consisted in determining the metallicity ([Fe/H]) from selected Fe I lines and broadening parameters.
With the [Fe/H] and main atmospheric parameters fixed, we then computed the elemental abundance of each chemical species following the same methodology as described in Fernández-Trincado et al. (2020c), Fernández-Trincado et al. (2021b), Fernández-Trincado et al. (2021c), and Fernández-Trincado et al. (2022a), and summarized here for guidance.We performed a synthesis using the full set of atomic line lists described in Smith et al. (2021), which is used to fit the local continuum.Subsequently, the cosmic rays and the telluric lines are removed, and the local S/N is estimated.After that, we follow the description given in Hawkins et al. (2016).We used four different abundance determination methods, namely, line-profile fitting, core line-intensity comparison, global goodness-of-fit estimate, and an equivalent-width comparison.For each method, we used a set of synthetic spectra of different abundances to compute the observed abundance.Only the lines with the best fit of the individual abundances are kept.We chose the  2 diagnostic for abundance determination due to its reliability.Nevertheless, we preserved data from the other diagnostic methods, including the standard deviation obtained from all four approaches to estimate the uncertainties.
To determine the C, N, and O elemental abundances, a mix of heavily CN-cycled and −poor MARCS models were used, as well as the same molecular lines adopted by Smith et al. (2021).The oxygen abundances are first estimated from the hydroxide molecular lines ( 16 OH).With this abundance known, we derived the carbon abundances from the carbon monoxide molecular lines ( 12 C 16 O), and finally, the nitrogen abundances are obtained from the cyanogen molecular lines ( 12 C 14 N).This process is repeated multiple times to eliminate any dependence on OH, CO, and CN lines.
An illustration of the quality of the model fit obtained in this study is presented in Fig. 2, where we show the observed (black dotted line) and synthetic (red line) spectra for the APOGEE-2 region around selected atomic and molecular lines obtained for our target giant star AP18102342−3146515.
Our study provides a number of elements in common with optical studies carried out by Barbuy et al. (2007) and Barbuy et al. (2018b), as well as complementary elements not previously examined.The resulting abundance derivations are listed in Table 3.
Uncertainties in spectroscopic parameters are listed in Table 4 for our four cluster stars.The reported uncertainties were obtained by once again calculating the abundances by varying one by one the mean value of the effective temperature, surface gravity, and microturbulence velocity by ΔT eff = ±100 K, Δ log g = ±0.3dex, Δ  = ±0.05km s −1 , respectively.These values represent the usual uncertainties for the stellar parameters in GCs (see, e.g., Barbuy et al. 2018b;Fernández-Trincado et al. 2019b).The final uncertainty of each chemical abundance is the quadratic sum of the individual uncertainties: where  2 mean is the standard deviation of the abundances computed for each line of a given element.Table 4 also shows the characteristic uncertainties for each abundance due to variations in the stellar parameters.
We derived a mean spectroscopic metallicity ⟨[Fe/H]⟩ = −1.15± 0.08 for NGC 6558, only 0.02 dex above of the results from Barbuy et al. (2018b) (⟨[Fe/H]⟩ = −1.17),who used UVES spectra.These authors investigated a star in common with our data set, namely, the star 2M18101768−3145246, with [Fe/H] = −1.15computed from both Fe I and Fe II lines.This value differs by 0.05 dex from ours (see Table 3).Barbuy et al. (2018b) adopted a Solar iron abundance of (Fe) ⊙ = 7.50 from Grevesse & Sauval (1998) for computing [X/Fe] abundances, in contrast to our study, which adopted that from Asplund et al. (2005), (Fe) ⊙ = 7.45.In order to analyse the entire set of data available for NGC 6558, we re-calibrated the values of those authors to be at the same relative Solar iron abundances as this work.Re-calibrated values to Asplund's Solar abundances for the star in common with Barbuy et al. (2018b) are listed in parentheses in Table 3.In this way, the star in    carbon, ⟨[C/Fe]⟩ = +0.23,which is 0.54 dex greater than our measures when their values are re-calibrated to Asplund et al. (2005).Although our data use the -band, where molecular lines permit us to achieve what we believe to be a more accurate determination of C, N, and O together, both the C and N are affected not only by stellar evolution but also by multiple populations.We did not obtain reliable measurements for C and N for our star in common, so a good agreement between the two samples is not necessarily expected.
The star 2M18101768−3145246 exhibits clear differences, about 0.18 dex lower in our measurement for [Mg/Fe] with respect to Barbuy et al. (2018b), and 0.14 dex higher in [Ca/Fe], which fall outside our error bars, in contrast with the much smaller differences of only 0.01 and 0.03 dex for [O/Fe] and [Al/Fe] abundances, respectively, which together with Si and Ti are in agreement within the error bars.(Carretta et al. 2009;Renzini et al. 2015;Mészáros et al. 2020), and potential correlations for Ce-Al and Ce-N (see, e.g., Fernández-Trincado et al. 2021d, 2022b).We also show the re-calibrated abundances for 2 stars from Barbuy et al. (2007) and 4 stars from Barbuy et al. (2018b) (open squares and circles, respectively).Our star in common with Barbuy et al. (2018b) (filled circle) is depicted as a filled star symbol.The results appear to be consistent with each other, except for one star from Barbuy et al. (2018b) in the Mg and Si abundances relative to Fe, which shows an evident shift in the Al-Si, Al-Mg, and Si-Mg diagrams.Despite the relatively few stars in our sample, anti-correlations for NGC 6558 might be present in the N-C, but the anti-correlations in Al-Mg, as well as the correlations in the Al-Si, Si-Mg, and Ce-N diagrams, are not clear.On the other hand, a weak anti-correlation between Ce and Al could be present, but the sample is too small to be definitive.We found that the NGC 6558 stars do not extend over a large range in [Al/Fe], but instead lie within a confined zone below [Al/Fe]=+0.3.This limit has been used as an easy and graphic marker between the first-and second-generation stars in GCs (Mészáros et al. 2015(Mészáros et al. , 2020)).In spite of that, the high values in N abundance for the stars of NGC 6558 in Fig. 3, and the low values in C, are together a good indicator of multiple-population stars.Thus, using only the Al abundance to distinguish between first-and second-generation stars may not be enough.Such enhanced N abundances exceed 4 above the typical Galactic level (dotted grey line at [N/Fe]∼ +0.81) at a metallicity of [Fe/H]=−1.15),as is shown in Fernández-Trincado et al. (2022c), and represented as a dashed line in Fig. 4.This figure presents a 4 fourth-order polynomial fit above the mean behaviour of the nitrogen for the Galactic stars drawn from the APOGEE pipeline ASPCAP (Pérez et al. 2016), and corrected to BACCHUS abundances using the stars belonging to GCs in (Mészáros et al. 2020, grey dots).This polynomial fit was used by Fernández-Trincado et al. (2022c) to select the nitrogenenriched red giant stars distributed across the bulge, metal-poor disk, and halo of our Galaxy, which exhibit elemental abundances comparable to those exclusively seen in GCs.These stars are thought to have been likely formed in the cluster and subsequently lost.
All the BGCs displayed in Fig. 3 exhibit a significant gap of approximately 0.3 dex in nitrogen abundance, with the exception of NGC 6558 and NGC 6522.The latter two clusters, due to the small sample, only presented stars with [N/Fe] values over the Galactic levels within this range of metallicities.When compared to Mészáros et al. (2020, their Fig. 17), this gap is solely observed in NGC 288, and barely indicated in NGC 5904.This suggests that the observed effect is likely due to the limited sample size in our clusters.
We could only measure the abundance of the -process element Ce relative to Fe for three stars, as well as to N. Despite this, [Ce/Fe] shows values that are slightly overabundant compared to the Sun, but as expected by other BGCs at the same metallicity, as can be seen in the Ce-Al and Ce-N diagrams.
The remarkable similarities observed in the chemical behaviour between NGC 6558 and the other CAPOS BGCs with the same metallicity can be seen clearly in Fig. 5 and Table 5. Violin plots displaying the abundance distributions of NGC 6558 in comparison to NGC 6522, Djorg 2, HP 1, NGC 6540, and NGC 6642 are presented.All the BGC abundance distributions are strongly consistent with each other.The mean values of [N/Fe] for NGC 6558 and NGC 6522 are slightly greater than the others, primarily due to the small sample (5 stars for NGC 6522), which include only bona fide second-generation stars.On the other hand, NGC 6522 exhibits systematically higher Al abundances compared to other CAPOS clusters with the same metallicity.In fact, according to (Fernández-Trincado et al. 2019c), this cluster exhibits a significant scatter in [Al/Fe], with a variation of ∼1 dex when considering values from the literature.
Figs. 3 and 5 show that the [Mg/Fe] abundances for all these bulge clusters, including NGC 6558, do not exceed the value ∼ +0.4.This observation is consistent with the values observed for the APOGEE bulge field stars (see, for example, Rojas-Arriagada et al. 2020;Geisler et al. 2021).However, they differ from M 4, M 12, and NGC 288, whose average ⟨[Mg/Fe]⟩ values are around ≈ +0.5 (Mészáros et al. 2020).These latter GCs share similar metallicities to NGC 6558, but they are not associated with the GB; instead, they are associated with disk formation (M 4 and M 12) or Gaia-Enceladus (NGC 288, Massari et al. 2019).

CONCLUDING REMARKS
We have presented a new high-resolution (R∼22,000) spectral analysis in the -band for six members of the relatively old, moderately metal-poor BGC NGC 6558, obtained as part of the CAPOS survey.The memberships are very likely, due to their spatial distribution, Gaia DR3 proper motions, and our own RV and metallicity from APOGEE spectra.We found a mean heliocentric RV of   = −193.4± 0.9 km s −1 , where the error is the standard error of the mean, in good agreement with Barbuy et al. (2018b).Out of the six stars, four presented spectral S/N greater than 70 per pixel, which allowed us to determine their abundance ratios We found a mean metallicity of ⟨[Fe/H]⟩ = −1.15for NGC 6558, in good agreement with Barbuy et al. (2018b) when their value is scaled to the same Solar iron abundance as ours.We also found mean values of the -(O, Mg, Si, Ca, Ti) elements that are slightly enhanced, iron-peak (Ni) essentially Solar, and the process element (Ce) slightly overabundant compared to the Sun.The high enrichment in nitrogen, together with the depleted carbon, indicates the presence of the multiple-population phenomenon in this cluster, although the Al values are below the Mészáros et al. ( 2020) limit for second-generation stars.This is the first time that the phenomenon of multiple-populations is observed for this cluster.Such high nitrogen enrichment for NGC 6558 agrees with the nitrogen abundances for the CAPOS BGCs at the same metallicities, turning them into potential progenitors of the nitrogen-enhanced moderately metal-poor field stars identified in the inner bulge region (e.g., Fernández-Trincado et al. 2016, 2017;Schiavon et al. 2017;Fernández-Trincado et al. 2019a, 2020a, 2021a, 2022b).
Despite our small statistical sample, we have observed consistent abundance patterns in all chemical elements studied in this work for all CAPOS BGCs in the same metallicity range as NGC 6558 (−1.2<⟨[Fe/H]⟩<−1.0).However, NGC 6522 stands out due to its significant spread in [Al/Fe].These clusters present nitrogen enhancement and carbon depletion, indicating the presence of multiple populations.Furthermore, the [Mg/Fe] abundances in these clusters remain within the range observed for bulge field stars, in contrast to some Galactic GCs outside the bulge at the same metallicities.

Figure 1 .
Figure 1.Distribution of the stars toward the NGC 6558 field.In all panels, the six cluster stars are identified with red open symbols.The size of the symbols is proportional to the signal-to-noise ratio (S/N).The size of the smaller S/N (red triangle symbol star) was scaled by a factor of 3 in order to make it visible.Top-left panel: spatial distribution of stars within the APOGEE-2 survey area.The black symbols refer to the overlapping APOGEE-2S plug-plates toward NGC 6558, corresponding to three APOGEE-2 programs: geisler_19b (open circles), kollmeier_17a (open triangles), and apogee (plus symbols).The cluster tidal radius ( r t =10.′ 44 ; Harris 1996, 2010) is marked with a large open black dashed circle.North is up and east is to the right.Top-right panel: plot of the radial velocity against metallicity of our member stars with the APOGEE-2S field sources in the direction of NGC 6558.Both parameters are from the ASPCAP pipeline.The black box, limited by ±0.15 dex and ±15 km s −1 and centered on [Fe/H] = −1.15 and RV = −192.63km s −1 , encloses our potential cluster members (red open circles), while the final sample of our four members whose abundances were determined with the BACCHUS code are the open dashed circles within the zoomed box (see text).Bottom-left panel: proper motion distribution of the stars within the cluster tidal radius of NGC 6558 extracted from Gaia DR3.The colour-coded bar represents the membership probabilities.The probability of membership of our six stars is greater than 95%.A circumference with a radius of 0.5 mas yr −1 is shown in the proper motions diagram.Bottom-right panel:   vs. ( −  ) colour-magnitude diagram.An isochrone with colour-coded cross-symbol temperatures has been fit to the most likely cluster members.

Figure 2 .
Figure 2. Example of the quality of the model fit obtained with BACCHUS in this study for selected molecular lines: 12 C 16 O, 12 C 14 N, 16 OH, and atomic lines: Mg I, Al I, Si I, Ca I, Ti I, Fe I, Ni I, and Ce II for one target giant in NGC 6558: AP18102342−3146515.The best-fit synthesis and observed spectra are highlighted with red lines and black dots, respectively.Each panel is centered around the selected lines, and the dashed black lines mark the position of the air wavelength lines.

Figure 4 .
Figure 4. Distribution of [N/Fe] vs. [Fe/H] for the stars in the GCs from Mészáros et al. (2020, grey dots).Symbols and colour codes are the same as in Fig. 3.The dashed black line represents a 4  ℎ order polynomial fit at the 4 level, over the usual Galactic values as described in Fernández-Trincado et al. (2022c).

Figure 5 .
Figure 5. Violin plot for the abundances of C, N, O, Mg, Al, Si, Ca, Fe, and Ce of NGC 6558 compared with other CAPOS BGCs.

Table 2 .
Estimated photometric atmospheric parameters for the NGC 6558 stars with high S/N (see text).

Table 3 .
Elemental abundances of NGC 6558 CAPOS stars.Within parentheses are the re-calibrated abundances for the star in common withBarbuy et al.

Table 4 .
Typical abundance uncertainties determined for four stars in our sample.