2.1. Sample Selection

The stars for this analysis (HD 45674, HD 180028, HD 194951 and HD 224893) were selected because they are luminous objects of Classes I and II located in the galactic plane. The derived abundances are compared with the mean abundances obtained for a sample of G-K young supergiants of Population I taken from ^{Luck (1977}; ¹⁹⁷⁸⁾.

The stellar spectra taken for HD 45674, HD 180028, HD 194951 and HD 224893 are taken from the library ELODIE¹ originally published by ^{Moultaka et al. (2001)} and updated in its version 3.1 by ^{Prugniel et al. (2007)}. The spectra were observed on July 06, 1999; on July 26, 2003; on December 18, 1999 and on July 07, 1999 respectively by means of the echelle spectrograph ELODIE placed in the 1.93m telescope located in the Haute-Provence Observatory (OHP). They cover the spectral region between 4000 and 6800 Å, and have a resolving power of R ≈ 42,000. The signal-to-noise ratio (S/N) for HD 45674, HD 180028, HD 194951 and HD 224893 is reported to be 83, 175, 82 and 145 respectively. This S/N per pixel has been derived for all spectra in the library at 5550 Å.

HD 45674, HD 180028 and HD 194951 are objects labeled as IRAS point sources, (IRAS 06262 - 0032, IRAS 19122 + 0557 and IRAS 20252 + 3410 respectively), although none of them display realiable infrared excesses since their IRAS fluxes are of low quality (Q=1) at 25, 60 and 100 microns. A review of the spectral energy distributions (SED) from the available photometry contained in the VisieR database also indicates that flux data from other sources like WISE and AKARI for these three objects do not indicate infrared excesses. Likewise, HD 224893 does not show evidence of dust around the central star. This last object was identified by ^{Bidelman (1993)} as a possible candidate to a post-AGB or normal A-and F-type supergiant star.

Figure 1 shows a representative range of 50 Å of the spectrum for the stars HD 45674, HD 180028, HD 194951 and HD 224893. The vertical dotted lines indicate the position of some absorption lines of elements such as Y II, La II, Ti II, Fe I y Ni I. The basic parameters for the total sample are shown in Table 1. This table contains the HD number, the equatorial coordinates, the apparent V and B magnitude, the galactic coordinates, the radial velocity, the parallax, the color excess and the velocity of rotation. These values were taken from the SIMBAD astronomical database.²

Fig. 1 Representative spectra of the sample stars HD 45674, HD 180028, HD 194951 and HD 224893. The locations of lines of certain important elements have been indicated by dashed lines. This stars are arranged in decreasing order the HD number. 

3.1. Uncertainty in the Abundances

We calculated the effects on the chemical abundances of errors in the measured equivalent widths (8-10%), the defined model of atmospheric parameters (+250 K in T_eff, +0.5 in log g and +0.5 km s⁻¹ in ξ_t), and the atomic data (log gf and χ (eV)). Errors due to equivalent widths are random because they depend on several factors, such as the position of the continuum, the signal to noise ratio (S/N) and the spectral type of the star.

In contrast, errors due to the atmospheric parameters and atomic data are systematic and de-pend on the quality with which they were derived. In short, the error in the gf -values can varies from element to element. For example, experimental values of high accuracy for Fe I and Fe II between 3% to 10% are available for a large fraction of the lines. For other Fe-peak elements, errors in their gf -values may range from 10 to 25%. For neutron- capture elements the accuracy is in the 10 to 50% range. The sensitivity of the derived abundances to changes in the model atmosphere parameters is displayed in Table 3.

HD 45674 and HD 194951 have the same temperature, i.e., 7500 K. For these stars the spectra are much cleaner and the temperature is not large enough for the lines to develop strong wings making line strengths inaccurate. We can see that a variation in effective temperature of 250 K generates a greater uncertainty in all lines, particularly in neutral lines except C, N, O and ionized lines like Ba, Ce and Nd. On the contrary, we did not observe significant changes in the uncertainties due to changes in log g by 0.5, ξ_t by 0.5 km s⁻¹ and Ew by 10 per cent. However, at a lower temperature (6400 K) the line strengths shows sensitivity only to changes both in T_eff by 250 K, and in log g by 0.5.

The total error σ_tot for each element is given by the square root of the quadratic sum of the random and systematic errors.

5.1. HD 45674

The star HD 45674 has been classified as F1Ia. ^{Bidelman (1993)} includes this object as a possible candidate to a post-AGB or an A-and-F normal supergiant star. The atmospheric parameters were obtained from ionization equilibrium between the lines of Fe I and Fe II with values of T_eff = 7500 K, log g = 2.0, χ_t = 4.0 and [Fe/H] = +0.16 dex respectively. The surface gravity for HD 45674 was confirmed from the equilibrium ionization of other elements such as Ca, Ti, Cr, from which Δ = [X_II/H]- [X_I/H] is 0.08, −0.02 and −0.05 respectively. A compilation of the atmospheric parameters obtained by different authors is shown in Table 2.

With respect to the abundances of the light elements CNO, the carbon abundance was obtained from two lines (4769 Å and 5380 Å) their values being [C/H] = −0.20 and [C/Fe] = −0.36. ^{Venn (1995b)} predicts a non-LTE correction of −0.25 in the C abundance for the supergiant HD 36673 (F0Ib) with an effective temperature of 7400 K. HD 45674 has an effective temperature of 7500 K; therefore no non-LTE correction was neccesary. A value of −0.25 dex led to an abundance of [C/Fe] = −0.69.

The N abundance for HD 45674 showed a moderate enrichement ([N/Fe] = +0.68). In fact, this abundance was obtained from the λ4151.4 Å line with an Ew of 10.1 mÅ. A non-LTE correction of −0.28 was derived by ^{Venn (1995b)} in the near IR region at 7400 K for this star, whis is of A9II type. The non-LTE correction for N I by ^{Luck & Lambert (1985)} indicated that at a temperature of 7500 K, an equivalent width of ≈10 mÅ and log g = 2.0, the non-LTE correction is almost zero (see Figure 8). By contrast, for the blue region ^{Przybilla & Butler (2001)} found a non-LTE correction of −0.11 dex for N I (λ3830 Å) at a temperature of 9600 K and with a negatively increasing value toward the IR lines. Taking into consideration the above we can argue that the non-LTE correction should be small for this line.

We adopted a value of ≈ −0.08 dex for our line in the blue region. Our non-LTE values are [N/H] = +0.60 dex and [N/Fe] = +0.36 dex respectively. The observed deficiency of C and the moderate enhancement of N involves a conversion of initial carbon into N through the CN-cycle and products of the 1DUP being brought to the surface.

The O abundance is near solar ([O/Fe] = +0.03), and is obtained from the λ6156.0 Å and λ6156.8 Å lines. At this temperature a non-LTE correction of about −0.15 (taken from ^{Takeda & Takada-Hidai 1998}) would lead to abundances of [O/H] = −0.12 and [O/Fe] = −0.36 respectively. The C/O ratio of 0.32 indicates that HD 45674 is O-rich.

Iron-peak elements (Sc, V, Cr, Ni) show [X/Fe] abundances very similar to solar, and ranges from −0.22 to +0.18 dex. The [α/Fe] ratio ≈+0.10±0.16, estimated from Mg, Si, Ca, Ti, is typical for objects of the thin disk (^{Reddy et al. 2003}; ²⁰⁰⁶). Si and S abundances show moderate enhancements, i.e., [Si/Fe] = +0.31 and [S/Fe] = +0.25 respectively. The Zn abundance is derived from two lines at λ4722 Å and λ4810 Å. The value of log gf = −0.25 for the line at λ4810 Å was obtained from ^{Barbuy et at. (2015)}. [Zn/Fe] = −0.33 is lower than expected for objects of the thin and thick disk population, i.e., ≈ −0.20 a −0.30 dex (^{Reddy et al. 2003}; ²⁰⁰⁶). Neutron-capture elements show [X/Fe] abundances with a tendency towards solar values and ranges from +0.06 to +0.19 dex. Of other elements identified, like Y II, Ba II, Ce II and Nd II, only Zr II seems to be moderately enriched ([Zr/Fe] = +0.45).

5.2. HD 180028

HD 180028 was classified as a F6Ib and cataloged as an IRAS point source (IRAS 19122 + 0557). The radial velocity was reported by different authors: −6.0 km⁻¹ (^{Wilson 1953}), −6.0 km⁻¹ (^{Duflot et al. 1995}) and −5.1 km⁻¹ (^{Gontcharov 2006}) and does not present variability. Its absorption lines appear slightly wide indicating that it rotates with a velocity of 23.3 km ⁻¹ (^{De Medeiros et al. 2002}).

Their atmospheric parameters were obtained from ionization equilibrium between the lines of Fe I and Fe II, with values of T_eff = 6400 K, log g = 2.1, χ_t = 4.8 and [Fe/H] = −0.06 dex respectively. The surface gravity for HD 180028 was confirmed from the equilibrium ionization of other elements such as Si, Ti, Cr, namely Δ = [X_II/H]-[X_I/H] is 0.11, −0.10 and −0.00 respectively. The atmospheric parameters obtained by different authors and the adopted parameters are displayed in Table 2.

HD 180028 exhibits a metallicity near to solar of [Fe/H]=−0.06. The abundance of C was obtained by synthesizing the line at 5380 Å. We obtained values of [C/H] = −0.28 dex and [C/Fe] = −0.22 dex respectively. At a temperature of 6100 K the non-LTE correction to C was very small (^{Takeda 1994}) and led to a value of [C/Fe] = −0.39 dex. There are no N lines or CN bands in the spectrum, so abundances could not be estimated. The O abundance was derived by synthesizing the region at 6140-60 Å. The values are [O/H] = +0.07 dex and [O/Fe] = +0.13 dex, respectively (see Figure 3). These values are higher than those obtained by ^{Luck (2014)}. The C/O ratio is about ≈0.23, which is lower than the solar value. From a sodium analysis ^{Andrievsky et al. (2002)} found a moderate enrichment of [Na/Fe] = +0.28 dex. The value obtained in this work is slightly higher, i.e. [Na/Fe] = +0.43 dex. The Fe-peak elements show a solar trend. The ratio [α/Fe] of +0.18 ±0.05 dex seems to indicate that this object belongs to the thin disk. Individually, Si, Ca and Ti present solar values while Mg shows a moderate enhancement, [Mg/Fe] = +0.25 dex. Sulfur is found be enriched, i.e., [S/Fe] = +0.44 dex. The s-elements show a solar trend except for Ba and La which are modestlly enhanced.

Fig. 3 Observed (filled squares) and synthetic (solid line) spectra of the O I triplet in the 6140-60 Å region. The synthetic spectra correspond from top to bottom to [O/H] = +0.37, +0.07 and −0.23 dex, respectively.  

5.3. HD 194951

HD 194951 was classified as F-type by different authors: F2Iab by ^{Stock et al.(1960)}, F3II by ^{Bidelman (1957)} and F1II by ^{Morgan (1972)}. The physical parameters have been determined by various authors using photometric and spectroscopic techniques (see Table 2). The values we adopted can be seen in Table 3. The surface gravity of HD 194951 was confirmed from the equilibrium ionization of other elements, such as Ti, Cr, and Ni, from which Δ = [X_II/H]-[X_I/H] is +0.05, +0.09 and −0.09 respectively. The chemical abundances of the elements identified are presented in Table 4.

The carbon abundance was determined from five lines (λ4769.9 Å, λ4775.8 Å, λ4932.0 Å, λ5380.3 Å and λ6014.8 Å). We obtained a [C/H] value of −0.16. At a temperature of 7000 K the non-LTE effects were present and a correction of −0.2 dex for carbon led to values of [C/H] = −0.36 and [C/Fe] = −0.26 respectively. For the O abundance a non-LTE correction of −0.1 dex taken from ^{Takeda & Takada-Hidai (1998)} was estimated, and the abundances of [O/H] = −0.20 and [O/Fe] = −0.10 are near to solar. These C and O abundance values are similar to those obtained by ^{Luck (2014)}. The N line at λ4151.4 Å found for HD 45674 and HD 224893 could not be measured. However, ^{Lyubimkov et al. (2011)} were able to study the N abundance for HD 194951 using high-resolution spectra and they found that N is enriched, with values of [N/H] = +0.60 and [N/Fe] = +0.73±0.18 respectively.

Our atmospheric model differs by 240 K and 0.12 in effective temperature and surface gravity relative to the model used by Lyumbikov et al., i.e., 6760 K and 1.92. In Table 3 we note that very few changes in the N abundance appear to be due to variations of 250 K and 0.5 in T_eff and log g. This means that we can use this N abundance without being affected by the model. According to the authors the N enrichment is a general characteristic present in super-giant stars of A and F type that have experienced the first dredge-up (1DUP). The deficiency in C and the enrichment of N indicates that the CN-cycle has taken place and that processed material from the H-burning has been released on the surface of the star. The C/O = 0.46 ratio is practically solar, but it is an O-rich star.

The [α/Fe] ratio value of +0.06 is consistent for an object belonging to the thin disk population. The Si abundance exhibits a moderate enrichment, i.e., [Si/Fe] = +0.33. Other elements such as Sc, Cr, Mn and Zn show abundances [X/Fe] between −0.15 and +0.09, similar to those present in objects of the thin and thick disk. The Zn abundance was obtained from the synthesized line at λ4810.5 Å. [X/Fe] of the s-elements ranges from −0.09 to +0.01, that is, it is on average solar.

5.4. HD 224893

^{Gray et al. (2001b)} determined atmospheric parameters for this star (see Table 2) and it was classified as a bright giant, A8II. Several authors studied the radial velocity in different epochs, obtaining values (in km s⁻¹) of −22.4 (^{Adams 1915}), −26.4 (^{Harper 1937}), −23.2 (^{Wilson 1953}), −21.9 (^{Bouigue et al. 1953}), −27 (^{Fehrenbach et al. 1996}) and −25.10 (^{Gontcharov 2006}). These values show very little variation. ^{Danziger & Faber (1972)} report a rotation velocity of 40 km s⁻¹.

In this paper an analysis of abundances which includes a total of 15 elements was performed for the first time. The abundances are shown in Table 4. The metallicity of HD 224893 is moderately deficient ([Fe/H] = −0.26 dex). The C abundance was obtained from the lines 4770.0 Å and 4771.7 Å respectively. From the LTE analysis we derived a value of [C/H] of −0.11 dex. By taking a non-LTE correction of −0.26 dex (^{Venn 1995b}), both [C/H] and [C/Fe] reached values of −0.37 dex and −0.19 dex respectively. The N abundances obtained from λ4151.4 Å were [N/H] = +0.89 dex and [N/Fe] = +1.15 dex respectively. The equivalent width of this line was 15.8 mÅ. With a non-LTE correction of −0.08 dex adopted at 7500K the new values were [N/H] = +0.81 dex and [N/Fe] = +0.99 dex respectively. Like HD 45674, this star shows evidence of having undergone the CN-cycle and the 1DUP event.

The O abundance is derived from the line at 6158 Å and shows signs of overabundance ([O/Fe] = +0.45). However, a non-LTE correction of -0.15 dex at a temperature of 7500 K (from ^{Takeda & Takada-Hidai 1998}) leads to values of [O/H] = +0.35 dex and [O/Fe] = +0.53 dex respectively. These values indicate that the O abundance shows a moderate enrichment. The ratio C/O = 0.26 indicates that HD 224893 is O-rich and has a value similar to HD 45674.

The abundances [X/Fe] of the Fe-peak elements, such as V, Cr and Mn show solar values except Sc, which is moderately enriched by +0.39 dex. The alpha-elements ([α/Fe] = −0.02 dex); the Zn and the s-elements (Y and Ba) abundances resemble the solar values.

6.1. Atmospheric Parameters

The atmospheric parameters taken from the literature and those adopted in this work are present in Table 2. We take a mean value of atmospheric parameters of those objects with more than two values. We can note that our adopted values for the effective temperature, surface gravity, microturbulence velocity and metallicity are consistent (within the uncertainties) with the mean values derived by other authors.

6.2. Masses

We also estimated the masses of our sample stars. These can be obtained from their position on the HR-diagram using theoretical evolutionary tracks. We employed the theoretical evolutionary tracks without rotation for masses of 5, 7 and 9 M_⊙, taken from ^{Ekström et al. (2012)} at solar metallicity (Z = 0.020). Luminosities for HD 45674, HD 194951 and HD 224893 were calculated using M_V = V₀ + 5 − log(D), where V₀ is the intrinsic colour derived from expresion V₀ = V − 3.1E(B − V) and D is the distance estimated from parallaxes (^{van Leeuwen 2007}).

For HD 180028, however, the absolute magnitude was derived from a mean value (M_V = −3.38±0.16 mag) estimated by ^{Kovtyukh et al. (2012)}. The bolometric corrections were taken from ^{Masana et al. (2006)} and we adopted a solar bolometric magnitude, M_Bol = 4.73±0.01 (^{Gray 2005}).

The color excesses for HD 180028, HD 194951 and HD 224893 were obtained using E(b − y) de-rived from Strömgren photometry by ^{Gray et al. (2001b)}. We obtained E(B − V) from the relation E(b − y) = 0.78 E(B − V) (^{Fernie 1987}). For HD 45674 the color excess was estimated from Schlafly’s map (^{Schlafly & Finkbeiner 2011}) where the dust in the Galactic disc was modelled assuming a thin exponential disc with a scale-height of 125 pc. A correction to the color excess (B −V) was applied with the last assumption. The values are listed in the eleventh column of Table 1.

Figure 4 shows the position of our sample stars on the HR-diagram. The uncertainty of the luminosities is based on the uncertainties of the parallaxes, visual magnitudes, extinctions and bolometric corrections. Individually, these uncertainties are 0.29 (HD 45674), 0.11 (HD 180028), 0.30 (HD 194951) and 0.21 (HD 224893). According to their position we could estimate that HD 180028 has a mass of between 5 and 7 M_⊙, HD 45674 and HD 194951 be-tween 7 and 7.5 M_⊙, and HD 224893 of about 9 M_⊙.

Fig. 4 Locations of HD 45675, HD 180028, HD 194951 and HD 224893 on the HR-diagram. The theoretical evolutionary tracks without rotation for masses of 5, 7 and 9 M_⊙ are taken from ^{Ekström et al. (2012)} with Z = 0.020. 

6.3. Evolutionary Status

CNO abundances are key to deduce the evolutionary status since their composition reflects the mixing process inside the stars. A summary is shown in Table 4. With respect to the light elements CNO, we note that the C abundances for HD 45674, HD 180028, HD 194951 and HD 224893 show deficiency with a non-LTE correction, i.e., they range from −0.04 to −0.69 dex.

Applying non-LTE corrections, HD 45674, HD 180028, HD 194951 and HD 224893 show values for the O abundance of −0.36, +0.02, −0.10 and +0.53, respectively. HD 224893, by contrast, has a moderate enhancement ([O/Fe] = +0.45) probably due to the CNO bi-cycle. In general, our oxygen values show the observed tendency in objects of the disc studied by ^{Bensby et al. (2014)}; ^{da Silva et al. (2015)}

For the total sample we were able to identify only one nitrogenline (λ4145.4 Å) for HD 45674 and HD 224893. The values with NLTE corrections are [N/Fe] = +0.36±0.11 dex and [N/Fe] = +0.99±0.14 dex, respectively. Both objects show signs of CN processed material that has been brought to the surface. Since we were unable to estimate the abundance of nitrogen in HD 180028, we cannot argue about the efficiency of the CN-cycle. The N abundance in HD 194951 was previously studied by ^{Lyubimkov et al. (2011)}, who found that nitrogen is enriched, i.e., [N/Fe] = +0.73±0.18 dex.

HD 45674, HD 194951 and HD 224893 show an enhancement of N abundance. These objects have shown a systematic deficiency of C and a systematic increase of N indicating the presence of CN-cycle material in the stellar surfaces. This enhancement of nitrogen was also reported by ^{Venn (1995b)}, ^{Venn & Przybilla (2003)}, ^{Luck & Lambert (1985)}, ^{Luck & Wepfer (1995)} and ^{Smiljanic et al. (2006)} for supergiant and bright giant stars.

Three processes have been suggested to explain the surface nitrogen enhancement for massive stars. Each leads to the presence of CN-cycle material in stellar surfaces. In single stars these processes can be the severe mass loss, the rotationally induced mixing during the MS phase and the dredge-up associated with the deep convective envelope. According to ^{Lyubimkov et al. (2011)} the mass loss is unimportant for B-type stars (progenitors of A-F supergiants) with masses ranging between 4-15 M_⊙. By contrast, the other two processes might be found coupled, e.g., it is an observational fact that the mixing process during the 1DUP event might differ between rotating and non-rotating stars. In binary stars, on the contrary, this enhancement can be due to mass transfer from an evolved companion.

In order to verify whether HD 45674, HD 194951 and HD 224893 have already passed through the 1DUP event is necessary to know the [N/C] ratio. The post-1DUP prediction by ^{Schaller et al. (1992)} is approximately [N/C] = +0.60 dex for stars with masses between 2 and 15 M_⊙. On the other hand, ^{Meynet & Maeder (2000)} calculate rotating (with a initial velocity of 300 km s⁻¹) and non-rotating evolutionary models for masses between 9-120 M_⊙. These authors predict [N/C] = +0.72 dex without rotation and [N/C] = +1.15 dex with rotation after the 1DUP and during the blue loop phase for stars of 9 M_⊙. For more details see Figure 14 in ^{Smiljanic et al. (2006)}.

For HD 194951, the [N/C] ratio could be obtained from the N abundance derived by ^{Lyubimkov et al. (2011)} and the C abundance estimated in this work. Here, we must emphasize that our model differs by 240 K in T_eff and by 0.12 in log g from the model used by ^{Lyubimkov et al. (2011)}. From Table 3 we can see that there are small differences of −0.02 (in T_eff ) and −0.06 (in log g) respectively. These small differences do not affect the N abundance derived by ^{Lyubimkov et al. (2011)}. Hence, it is possible to use the nitrogen abundance de-rived by ^{Lyubimkov et al. (2011)} with our C abundance. This value is [N/C] = +0.81±0.17 dex. For HD 45674 and HD 224893 we determined a [N/C] ratio of +0.88±0.09 dex and +1.00±0.13 dex respectively. We note also that these values indicate that both stars show signs of an efficient mixing process. The mean [N/C] ratio for the sample stars is [N/C] = +0.90±0.13 dex.

In short, our results for single and mean [N/C] ratios are larger than the values obtained by ^{Schaller et al. (1992)} by a factor between 0.3-0.4 dex and are slightly above those of the non-rotating model of by ^{Meynet & Maeder (2000)}, i.e., between +0.72 dex (non-rotating model) and +1.15 dex (rotating model). Our mean value +0.90±0.10 dex would be in agreement with the results for non-rotating stars with masses of 9 M_⊙. Therefore, we may conclude that within the range of masses of our objects (5-9 M_⊙) and for their [N/C] ratios, HD 45674, HD 194951 and HD 224893 have already experienced the 1DUP event and are in post-1DUP phase. Evolutionarily, the post-1DUP objects can be located only inside the red-blue loop area. ^{Claret (2004)} suggests that extensive red-blue loops occur for stars with masses from 6 to 13 M_⊙, although this extension is still a matter of controversy (see § 6 and Figure 12 in ^{Lyubimkov et al. 2011}).

On the other hand, if we consider the effective temperature and luminosities to be right, HD 45674, HD 194951 and HD 224893 appear to be located within the extended red-blue region with M = 7 − 9M_⊙ on the HR-diagram. However, the uncertainties cause an appreciable change in the luminosity so that their evolutionary status is unreliable. With regard to HD 180028, assuming a mass of 5 M_⊙, its probable evolutionary status would place it in the red-blue loop region (post-1DUP phase). On the contrary, if we assume a mass of 7 M_⊙ it would to be located on the Hertzsprung gap region and moving for the first time towards the red giant/supergiant phase (post-MS phase).

6.4. Sodium

In our sample we found a moderate enrichment of Na abundances, i.e., [Na/Fe] of +0.34, +0.43, +0.24 and +0.55 for HD 45675, HD 180028, HD 194951 and HD 224893, respectively. It is believed that the sodium enrichment is related to the first dredge-up event (^{Denissenkov & Denissenkova 1990}) although this assumption is still debated. The predictions made by the models (see Fig. 8 in ^{Karakas & Lattanzio 2014}) indicate that sodium does not show a significant enrichment in low-mass stars (M ≤ 2 M_⊙) from the first dregde-up or by extra-mixing process to solar metallicity (^{Charbonnel & Lagarde 2010}). Even for M > 2 M_⊙ this enrichment does not exceed 0.3 dex.

A non-LTE correction of about of −0.10 dex to Na was taken from ^{Lind et al. (2011)} and results in values of [Na/Fe] of +0.16, +0.22, +0.24 and +0.37 dex respectively. We note that our observational results (within their uncertainties) are in agreement with the [Na/Fe] predicted by ^{El Eid & Champagne (1995)} for intermediate mass stars (≈0.2-0.3 dex).

6.5. Heavier Elements

The tendency shown by the alpha-elements of our objects is similar to that observed in the thin disk population, i.e., [α/Fe] ranges from −0.02 to 0.18 dex. The sulfur abundance was estimated synthesizing the spectral region at (6743-6757 Å). We obtained a [S/Fe] of +0.25 for HD 45675 and +0.12 for HD 194951, showing a tendency similar to that observed for α-elements in the disc for metallicity near to solar value (^{Caffau et al. 2005}). By contrast, S for HD 180028 is overabundant, i.e., [S/Fe] = +0.44 dex. Iron-peak and neutron-capture elements in all sample stars show [X/Fe] abundances similar to the solar value. HD 45674 and HD 180028 show a modest enrichment of Zr II and La II of +0.45 dex and +0.32 dex, respectively.

6.6. Comparison With Other Supergiants

Our [X/H] abundances can be compared with the mean abundances derived from a sample of G-K young supergiant stars studied by ^{Luck (1977}; ¹⁹⁷⁸⁾ and mean CNO abundances by ^{Luck (2014)}. Figure 5 shows the [X/H] abundances versus atomic number. The abundances in the sample stars of the present paper are shown with different symbols together with their uncertainties, namely HD 45674 (open circle), HD 180028 (open triangle), HD 194951 (open square) and HD 224893 (open pentagon). The cool supergiants are represented with a blue filled circle and CNO abundances with a red filled pentagon. We can see, in general, that abundances of our sample stars (e.g., CNO, α, Fe-peak and neutron capture elements) follow the trends expected from galactic chemical evolution for Population I stars.

Fig. 5 Comparison of our abundances [X/H] with the mean abundances for a sample of G-K young supergiants taken from ^{Luck (1977}; ¹⁹⁷⁸⁾ and mean CNO abundances by ^{Luck (2014)}. Symbols and color represent data from different sources. The color figure can be viewed online.