Red giants and yellow stragglers in the young open cluster NGC 2447

Abstract

In this work we analysed, using high-resolution spectroscopy, a sample of 12 single and 4 spectroscopic binary stars of the open cluster NGC 2447. For the single stars, we obtained atmospheric parameters and chemical abundances of Li, C, N, O, Na, Mg, Al, Ca, Si, Ti, Ni, Cr, Y, Zr, La, Ce, Nd, Eu. Rotational velocities were obtained for all the stars. The abundances of the light elements and Eu and the rotational velocities were derived using spectral synthesis technique. We obtained a mean metallicity of [Fe/H] = −0.17 ± 0.05. We found that the abundances of all elements are similar to field giants and/or giants of open clusters, even for the s-process elements, which are enhanced as in other young open clusters. We show that the spectroscopic binaries NGC 2447-26, 38, and 42 are yellow-straggler stars, of which the primary is a giant star and the secondary a main-sequence A-type star.

1 INTRODUCTION

1.1 Open clusters

Open clusters offer a possibility to study several aspects of stellar evolution, correlations between Galactocentric distances with abundances and also to study the importance of binary stars in their population. The luminosities of the stars are known because the distances of the open clusters are well determined. Therefore, any chemical peculiarity can be related to the evolutionary status of the stars. In addition, since their ages are also known, possible relations between age and abundances can be explored. This is particularly true for the giants in open clusters because for field giants distances and ages are less constrained. The knowledge of the open clusters will be greatly improved by the Gaia mission (Gaia Collaboration et al. 2016; Jordi et al. 2016), with distances and proper motions being accurately determined the membership probability can be better constrained. In addition, probing distant stars with old and intermediate ages can help identify new open clusters through their evolved giants.

In this work, we continue our analysis of giants of open clusters in the Southern hemisphere through high-resolution spectroscopy. We dedicate our study to the red giants in the open cluster NGC 2447, also referred to as M 93. As in our previous papers, Santrich, Pereira & Drake (2013) and Sales Silva et al. (2014), the stars for our study were selected from the radial velocity and proper motion surveys of Mermilliod, Mayor & Udry (2008) and Frinchaboy & Majewski (2008), respectively. These studies identified red giant stars in many open clusters. Following the same strategy of our previous papers, we observed all giant stars that were recognized as cluster members. Observing all known stars of same cluster, both singles and binaries, increases the possibility for new findings of chemically peculiar stars, such as barium stars (Katime Santrich, Pereira & de Castro 2013). This is specially true for red giant stars in binary systems in open clusters because there are few high-resolution spectroscopic observations of this kind of stars. Therefore, we observed all the 16 giants known in NGC 2447 including the three known spectroscopic binaries and one star (NGC 2447-38) that has been previously suspected to be a binary star by Mermilliod & Mayor (1989). Our spectroscopic observation indeed confirms its binary nature.

1.2 NGC 2447

Fig. 1 shows the colour–magnitude diagram and Table 1 shows the observational parameters of the cluster. In Fig. 2, we plot a dependence of the turn-off mass on the age for some clusters including NGC 2447. From the isochorone fit, we derived a turn-off mass of 2.76 ± 0.01 M_⊙, which is in good agreement with the value of 2.74 M_⊙ adopted by Hamdani et al. (2000). The error in the turn-off mass is determined by the discretization of the points of the isochrone.

Also based on isochorone fit, we obtained an age of 0.40 ± 0.03 Gyr. Our result, |$\log t\ = $| 8.6 ± 0.1, is also in good agreement with other values from the literature, such as |$\log t\ = $| 8.65 (Hamdani et al. 2000), 8.59 (Reddy, Giridhar & Lambert 2015), 8.61 (Netopil et al. 2016). Our error cannot be better than 0.1 (in log t) since this is the interval of isochrones given by Bertelli et al. (1994) for ages down to 10.0 Gyr.

A few giant stars of the open cluster NGC 2447 have already been spectroscopically observed using high resolution but all the studies focused on the analysis of the same three giants (stars #28, #34 and #41; Hamdani et al. 2000; Santos et al. 2009, 2012; Smiljanic et al. 2009; Reddy et al. 2015). In this work, thanks to high-resolution spectroscopy, we could do a broad abundance analysis of the giants in NGC 2447. Using the spectral synthesis technique, we determined the abundances of the light elements, CNO and lithium, one element created by the r-process, europium, and we also determined the rotation velocity for these stars. The abundances of Na, Al, α-elements (Mg, Si, Ca and Ti), iron-peak elements (Cr and Ni), and the elements created by the s-process (Y, Zr, La, Ce, and Nd) were obtained by measuring equivalent widths of the corresponding lines.

As was earlier mentioned, we also observed the all known (four) red giants in binary systems of NGC 2447. As it will be seen, three of the four binaries are indeed ‘yellow straggler’ stars (stars #26, #38 and #42) considering their position in the colour–magnitude diagram. These stars of NGC 2447 have the same spectral characteristic of the other yellow straggler stars analysed in NGC 2360, NGC 3680, and NGC 5822 by Sales Silva et al. (2014): strong veiling due to the presence of a main-sequence A-type secondary star. Finally, the other binary star, NGC 2447-25, has (B − V) colour similar to other single red giants, which indicates that the secondary is probably a faint main-sequence star (Mermilliod & Mayor 1989).

2 OBSERVATIONS

The high-resolution spectra of the 16 red giants were obtained with the Fibre fed Extended Range Optical Spectrograph (FEROS) (Kaufer et al. 1999) echelle spectrograph at the 2.2 m ESO telescope at La Silla (Chile). The FEROS spectral resolving power is R = 48 000, corresponding to 2.2 pixels of 15 |$\mu$|m, and the wavelength coverage goes from 3800 to 9200 Å. The S/N ratio was evaluated by measuring the rms flux fluctuation in selected continuum windows, and the values range from 100 to 150 around 6000 Å. The observational parameters of the sample stars are given in Table 2.

The (B − V) colours of the objects from Frinchaboy & Majewski (2008), namely CD-23°2813, CD-23°6042, and TYC 6540-4084-1, are redder because the authors’ B and V magnitudes were based on the Tycho system. These values were corrected using the calibration given in Bessell (2000). Table 2 provided the V magnitude and (B − V) for these three stars after correction.

3 ANALYSIS AND RESULTS

3.1 Line selection, measurement, and oscillator strengths

The atomic absorption lines selected in this study are basically the same as those used in the previous studies devoted to the analysis of photospheric abundances of the single stars of the open cluster NGC 3114 (Katime Santrich et al. 2013) and in the study of the binary stars in the open clusters NGC 2360, NGC 3680, and NGC 5822 (Sales Silva et al. 2014). In Table 3, we show the Fe i,ii lines employed in the analysis, the lower excitation potential (χ) of the transitions, the log gf values and the measured equivalent widths. The log gf values for the Fe i,ii lines were taken from Lambert et al. (1996). The equivalent widths were obtained by fitting Gaussian profiles to the observed ones using the task splot in iraf. In Table 4, we show the same for the other species.

3.2 Atmospheric parameters

In order to determine the chemical abundance of the stars, it is necessary to know the effective temperature (T_eff), surface gravity (log g), the microturbulence (ξ) and metallicity [Fe/H] (in the notation [X/H] = log (N_X/N_H)_⋆−log(N_X/N_H)_⊙). Following our previous analysis, the temperature was derived by forcing null correlation between the iron abundance and the excitation potential and also null correlation between the iron abundance and the equivalent widths of the lines to obtain the microturbulent velocity. The surface gravity was obtained by imposing that the Fe i abundance be equal to that of Fe ii. The atmospheric parameters were determined using the local thermodynamic equilibrium (LTE) atmosphere models of Kurucz (1993) and the spectral analysis code moog (version 2013, Sneden 1973). The adopted atmospheric parameters are given in Table 5. Table 6 provides atmospheric parameters determination from the literature.

The errors in the effective temperatures and microturbulent velocity were estimated considering the uncertainties in the correlations between the Fe i abundances and excitation potentials and between the Fe i abundances and the equivalent widths. For gravity, the uncertainty was estimated considering that the difference between the Fe i and Fe ii abundances differed by 1σ of the standard deviation of the mean value of Fe i abundances.

The cooler stars of our sample (NGC 2447-3 and 7) have larger errors in the temperature when compared with NGC 2447-93 which has a higher temperature. For NGC 2447-3 and 7, we have an error of ∼110 K while for NGC 2447-93 we have 40 K. Like the temperature, the error in the surface gravity is higher for these two cooler stars when compared with the hotter ones. For NGC 2447-3 and 7, we found a typical error of 0.4 dex while for NGC 2447-93 we found a typical error of 0.2 dex. The reason for these differences is related to the uncertainty in the continuum placement which affects the measurement of the equivalent widths, which is higher for cooler stars than for hotter ones.

For the distance, we adopted r =  1.057 kpc and A_V = 0.13. The bolometric corrections were calculated using the relation given in Alonso, Arribas & Martínez-Roger (1999). We found a mean difference between the spectroscopic and evolutionary gravities of 0.05 ± 0.13. This mean difference is in good agreement with uncertainties given by the standard deviation in log g. We also estimated the luminosities based on the observed V magnitude, interstellar absorption, and distance (photometric luminosities) and those based on spectroscopic gravities and turn-off mass (spectroscopic luminosities). They are also given in Table 5. The mean difference between them is 0.08 ± 0.05.

The stellar parameters for the three yellow stragglers, were obtained in the same way as was done for the other giant stars of NGC 2447, that is, the temperature and gravity were obtained based on the excitation and ionization equilibrium respectively. However, these stars are strongly affected by veiling (Section 4.3.7). Veiling produces an additional continuum over the intrinsic stellar spectrum and changes the equivalent widths of the spectral lines. An additional continuum added by the secondary reduces the measured equivalent values thus influencing determination of the stellar parameters. Looking for a possible solution, we found near zero microturbulent velocity. Null microturbulence has no physical meaning and, as it was above mentioned, it happens because we are forcing to find a solution based on the excitation and ionization equilibrium. For this reason, we do not provide the atmospheric parameters for the stars NGC 2447-26, 38, and 42 yet we published the equivalent widths of their Fe i and Fe ii absorption lines.

3.3 Abundance analysis

The abundances of the chemical elements were also determined with the LTE model atmospheres of Kurucz (1993) and using the code moog. The abundances of the elements were obtained either by measuring the equivalent widths or by means of the spectral synthesis technique. We normalized our abundances to the Sun using the solar abundances given in Santrich et al. (2013). For barium and europium, we obtained a solar abundance of, respectively, 2.13 and 0.54. For Stark broadening, we used the standard option available in moog, while for the collisional broadening we used the Unsöld approximation for all the lines.

We did not obtain the abundances for the stars NGC 2447-26, 38, and 42. These stars show evidence of veiling in their spectra as it will be discussed in Section 4.3.7.

The abundances of the light elements, lithium, carbon, nitrogen, and oxygen, were determined with spectral synthesis technique in the local thermodynamic equilibrium (LTE) atmosphere models of Kurucz (1993). Lithium abundance was derived from the synthetic spectra using the Li i λ6708 Å resonance doublet. The CN lines in the vicinity of the Li i doublet were included in the line list. The wavelengths and oscillator strengths for the individual hyperfine and isotopic components of the lithium line were taken from Smith, Lambert & Nissen (1998) and Hobbs, Thorburn & Rebull (1999). A solar ⁶Li/⁷Li = 0.081 isotopic ratio was adopted in the calculations of the synthetic spectrum. For the carbon abundance determination, we used the C₂ features at 5086 Å and at 5635 Å. The oscillator strengths and wavelengths of C₂ at 5086 Å lines were taken from Lambert & Ries (1981) and the oscillator strengths and wavelengths of C₂ at 5635 Å were taken from Drake & Pereira (2008). The oxygen abundance was inferred from the forbidden [O i] line at 6300.304 Å. In our calculation for this line, we used the oscillator strength log gf = −9.72 obtained by Allende Prieto, Lambert & Asplund (2001) in their analysis of oxygen abundance of the Sun. To determine the nitrogen abundance and the ¹²C/¹³C isotopic ratio, we used the ¹²CN and ¹³CN lines of the (2, 0) band of the CN red system A²Π − X²Σ between 7994 and 8020 Å.

The barium abundance was obtained using the Ba ii line at 5853.69 Å, where the hyperfine structure was taken from McWilliam (1998). The other barium lines at 4554.04 and 4934.09 Å have equivalent widths greater than 250 mÅ and are not suitable for abundance determination because they lie outside the linear part of the curve of growth.

The europium abundance was determined using the Eu ii line at λ6645.13 Å and the hyperfine splitting was taken from Mucciarelli et al. (2008). For two stars, NGC 2447-41 and 93 we could not determine a reliable europium abundance due to a very uncertain local continuum.

Figs 3 –5 show the observed and synthetic spectra for the spectral regions where the abundances of the lithium, oxygen, and europium were obtained. Table 7 gives the abundances of the light elements, the derived mean cluster abundances and their respective standard deviations. Tables 8 and 9 provide the abundances for each star in the notation of [X/Fe] ratios. For the elements of the s-process (Y, Zr, La, Ce, and Nd), we also give the mean abundance of these elements, s, in the notation [s/Fe].

3.4 Abundance uncertainties

The errors in the abundances were estimated in the same way as in our previous analyses of other open clusters (Katime Santrich et al. 2013; Sales Silva et al. 2014), that is, we changed T_eff, log g, and ξ according to their standard errors as well as the equivalent widths of the lines of each element and compared the new abundances with the obtained for those elements with more than three available lines. For the equivalent widths, the errors were estimated by the S/N ratio and the resolution of the spectra. For a resolution of 48 000 and an S/N equal to 100 and using the expression given in Cayrel de Strobel & Spite (1988), the uncertainty in the equivalent width is approximately 3 mÅ.

Tables 10 and 11 show the results for the stars NGC 2447-3 and NGC 2447-28. The seventh column provides the total uncertainty of each abundance calculated by composing quadratically the uncertainties due to T_eff, log g, microturbulent velocity, metallicity, and the equivalent widths. The last columns of Tables 10 and 11 give the observed abundance dispersion between the lines for each element with more than three available lines. As expected, the abundances derived using the lines of the neutral elements are more sensitive to the effective temperature variations, while the abundances based on lines of ionized elements are more sensitive to the variations in the surface gravity. Tables 12 and 13 show the changes in abundances of carbon, nitrogen, and oxygen related to variations in T_eff, log g, and ξ. Derived CNO abundances are not very sensitive to the variations of the microturbulent velocity since weak lines were used for their determination. Uncertainties in the carbon abundance also result in variation of nitrogen abundances, since the CN molecule lines are used for the N abundance determination and uncertainties in the oxygen abundance affect the carbon abundance and vice versa. In the last column, we present the resulting abundance uncertainties, calculated as the root square sum of the various sources of uncertainties.

4 DISCUSSION

4.1 The binary star NGC 2447-25 and the suspect binary NGC 2447-71

As seen in Table 2, NGC 2447-25 is a spectroscopic binary of which we did not find significant traces of the secondary component in its spectrum. However, we suspect that some contamination by a secondary star may have happened since NGC 2447-25 presents the lowest microturbulent velocity of the giants analysed in this work, 1.1 km s⁻¹ (Table 5). As discussed in Section 3.2, null or in some cases low microturbulent velocities, such as seen in NGC 2360 - 92 (0.8 km s⁻¹; Sales Silva et al. 2014) is a indication that veiling modified the determination of the atmospheric parameters. In addition, in Fig. 6 we compare the spectrum of NGC 2447-25 with the spectra of NGC 3680 - 11 and 34 in the region around H α. These two stars analysed in Sales Silva et al. (2014) are examples of spectroscopic binaries where in one star (#11) we did not detect any traces of the secondary star, while in another star (#34) the presence of the secondary star is evidenced by broader wings in the H α line. Comparing the spectrum of NGC 2447-25 with the spectra of NGC 3680 - 11 and 34 it seems that the H α line in NGC 2447-25 is slightly broader than the Hα in NGC 3680 - 11 but is not as broad as in the spectrum of NGC 3680 - 34. It is interesting that NGC 2447-25 has a lower Geneva (B − V) colour index than other giants (table 2 of Mermilliod & Mayor 1989).

As far as NGC 2447-71 is concerned, we suspected that it is a binary only based on its E/I ratio, seen in table 2 of Mermilliod & Mayor (1989). The E/I ratio is related to the ratio of the dispersion of the radial velocities over the sum of the sources of the instrumental errors. NGC 2447-71 presents a E/I ratio basically equal to that of NGC 2447-38, a yellow straggler star. This ratio was a criterion, apart from the Geneva (B − V) colour index and the search for radial velocity variation, considered by Mermilliod & Mayor (1989) to also conclude for the presence of possible binaries in NGC 2447. We, however, agree that our suspicion is weak and the evidence is not strong as is in the case of NGC 2447-25. Nevertheless, we also show the spectrum of NGC 2447-71 in the region of the H α. Given the binary nature of NGC 2447-25 and our suspicion for NGC 2447-71, we did not obtain the abundances for these two stars.

4.2 Rotational velocities

The determination of the rotational velocity |$v$| sin i was made by using spectral synthesis for the Fe i 6151.6 Å line. The macroturbulent velocity adopted in the synthesis was 3.0 km s⁻¹ for all stars. Fig. 7 shows the fit for the Fe i line with different rotational velocities. Table 14 shows the results for all stars. The mean rotational velocity of the giant stars in NGC 2447, excluding the stars #25 and #71 is 3.64 ± 0.55 km s⁻¹. Fig. 8 shows our results in comparison with field giant stars based on the data taken from Carlberg et al. (2011). Our mean value is slightly smaller than that of the giants analysed by Carlberg et al. (2011). In fact, excluding from the sample of Carlberg et al. (2011) those stars that are more than 1σ of their standard deviation, that is, those stars with rotational velocities higher than 9.0 km s⁻¹ which represent only 3 per cent of their sample of 1288 stars, the mean rotational velocity is 4.5 ± 1.2 km s⁻¹

4.3 The abundance pattern

4.3.1 Metallicity

The mean metallicity of NGC 2447 based on the Fe i lines is −0.17 ± 0.05 excluding the stars NGC 2447-25 and 71. This value is in good agreement with the value obtained by Reddy et al. (2015) which is −0.13 ± 0.02 based on the analysis of three stars. Hamdani et al. (2000) also used the same stars as Reddy et al. (2015) and obtained a slightly higher metallicity, +0.03 ± 0.03, and Smiljanic et al. (2009) obtained −0.01 ± 0.01. Clariá et al. (2005) obtained [Fe/H] = −0.09 ± 0.06 using Washington photometry which is also close to our values. Santos et al. (2009) and Santos et al. (2012) did two different studies for this cluster using the stars #28, #34, and #41. Santos et al. (2009) used two different line lists, one from Sousa et al. (2008) and one from Hekker & Meléndez (2007). The first one gives [Fe/H] = −0.03 ± 0.03, and the second one, [Fe/H] = −0.10 ± 0.03. In the second study, Santos et al. (2012) obtained [Fe/H] = 0.00 ± 0.04 (with the line list from Sousa et al. 2008) and [Fe/H] = −0.08 ± 0.02 (with the line list from Hekker & Meléndez 2007). Table 6 shows the results.

4.3.2 Li and the ¹²C/¹³C isotopic ratio

In giants of open clusters, lithium abundances have already been investigated by Gilroy (1989) and Pasquini, Randich & Pallavicini (2001) for NGC 3680, Pasquini et al. (2004) for IC 4651, Gonzalez & Wallerstein (2000) for M 11, Katime Santrich et al. (2013) for NGC 3114, Böcek Topcu et al. (2015) and Böcek Topcu, Afşar & Sneden (2016) for NGC 752 and NGC 6940, respectively, and Delgado Mena et al. (2016) for several open clusters. These studies showed that the lithium abundances in giants of the open clusters exhibit a large spread, from negative values to maximum values around 1.3–1.4 in the notation of log ε (Li). The higher values for the lithium abundances are in good agreement with models considering that the first dredge-up is the main mechanism of mixing as the star evolves to the red giant branch. Low values for log ε (Li) can be explained by extra-mixing mechanism that is thermohaline mixing or thermohaline plus rotation-induced mixing (Charbonnel & Lagarde 2010).

For the giants of NGC 2447, we found a mean value of 0.71 ± 0.43, for the lithium abundance. The large spread in the lithium abundance uncertainty thus reflects that different mixing mechanisms happened in the interior of these stars. Another interesting aspect of our lithium analysis is the fact the two stars, NGC 2447-3 and 7, that display very low values for the lithium abundance are the most evolved stars of our sample, with log g = 1.1 and log g = 1.5, respectively. There are few open cluster giants with low gravities analysed before. The open cluster NGC 7789 (Tautvaišiene et al. 2005) and NGC 6404 (Magrini et al. 2010) also have stars with log g < 1.5. Some cluster giants analysed by Gilroy (1989) also have low surface gravities.

The low values for log ε (Li) in NGC 2447-3 and 7 would indicate that these two stars have suffered deep mixing and as a consequence of that, it would be expected to determine a low ¹²C/¹³C isotopic ratio. However, we derived value of 22.0 which means a low content of ¹³C. The same kind of behaviour was already seen by Gilroy (1989). The most evolved stars in the sample of Gilroy (1989), those with low gravities, also have low log ε (Li) values but do not show low ¹²C/¹³C isotopic ratio. Gilroy (1989) considered that mass loss on the giant branch could modify the lithium abundance without changing the ¹²C/¹³C isotopic ratio.

As far as the other stars are concerned, those less evolved than NGC 2447-3 and 7, we could derive only lower limits for the ¹²C/¹³C isotopic ratios, which indicate that these stars did not suffer deep mixing in their interiors. Our results for the ¹²C/¹³C isotopic ratio are in a good agreement with those predicted by Charbonnel & Lagarde (2010) considering standard mixing models for an open cluster with a turn-off mass of 2.7 M_⊙.

Some works link rotational velocity and lithium abundance (Randich et al. 1999; do Nascimento et al. 2000), but there have been no consensus on the effect of rotation on the lithium content. De Medeiros et al. (2000) obtained a correlation coefficient between lithium abundance and rotational velocity for different stellar masses. They found that stars with M > 3.5 M_⊙ show a correlation coefficient of 0.297 and stars in the mass range 2.5 M_⊙ < M < 3.5 M_⊙ have a correlation coefficient of 0.558. The behaviour of the Li abundance and rotational velocity can be seen in Fig. 9 using stars from De Medeiros et al. (2000) and the stars from our sample.

From Fig. 9 there may indeed be some dependence between rotational velocity and lithium abundance, but we did not observe any Li abundance correlation with rotation in our sample, although our stars are slow rotators compared to the ones with high Li abundance.

4.3.3 Carbon, nitrogen, and oxygen

In Fig. 10, we show the [C/Fe] and [N/Fe] ratios for giant stars in NGC 2447 in comparison with the same ratios taken from Luck & Heiter (2007) and Mishenina et al. (2015). It can be seen that the stars from our sample have abundances similar to field giants. Like the field giants, the giant stars of NGC 2447 present nitrogen overabundance and carbon underabundance which indicates that the first dredge-up has taken place in the interior of these stars. In fact as seen from Table 7, the mean [C/Fe] and [N/Fe] ratios for NGC 2447 are, respectively, −0.25 ± 0.10 and +0.40 ± 0.18. The [N/C] ratio can be used as a diagnostic of the first dredge-up for giants in open clusters and this ratio can be compared with standard evolutionary models or additional models which consider that mixing can be due to rotation and/or to thermohaline effects (Charbonnel & Lagarde 2010). For NGC 2447, we obtain a [N/C] ratio of 0.65 ± 0.21. This value is in good agreement with the predictions given by Charbonnel & Lagarde (2010) for an open cluster with a turn-off mass of 2.7 M_⊙.

Previously CNO abundances for NGC 2447 were obtained by Smiljanic et al. (2009), where the authors analysed the stars NGC 2447-28, 34, and 41. They obtained a mean [C/Fe] ratio of −0.17 ± 0.01 and mean [N/Fe] ratio of 0.53 (with NGC 2447-28 and 34 only). Our results based on the same stars are, respectively, [C/Fe] = −0.22 ± 0.05 and [N/Fe] = 0.49 ± 0.04. The giants of NGC 2447 have nitrogen abundances similar to other giants of other open clusters: Smiljanic et al. (2009), Mikolaitis et al. (2010), Katime Santrich et al. (2013), Drazdauskas et al. (2016), Böcek Topcu et al. (2015, 2016), just to name a few. The oxygen abundances of our cluster giants all show low [O/Fe] ratios. Smiljanic et al. (2009) obtained −0.14 ± 0.03. Similar [O/Fe] ratios for the giants of other open clusters have already been reported by Böcek Topcu et al. (2015, 2016).

4.3.4 Other elements: Na to Ni

Figs 11 and 12 show the abundance ratios of the elements Na, Al, Mg, Si, Ca, Ti, Cr, and Ni in comparison with field giants. No significant difference was found between the [X/Fe] ratios for these elements of the giants of NGC 2447 and field giants. In addition, Table 15 shows that our results are in good agreement with the values obtained by Reddy et al. (2015). The comparison between our abundance ratios and those given by Reddy et al. (2015), Hamdani et al. (2000), and Smiljanic et al. (2009) (only for sodium) is given in Table 15. We found a mean difference between our derived ratios and those of Reddy et al. (2015) and Hamdani et al. (2000) of 0.07 ± 0.07 and 0.10 ± 0.09, respectively. The comparison with Hamdani et al. (2000) gives a larger uncertainty probably due to the different metallicities and temperatures obtained by them and the use of different lines with different gf values from ours.

4.3.5 Heavy elements: s-process

Fig. 13 shows the abundance ratios [X/Fe] versus metallicity for the heavy elements Y, La, Ce, Nd and the mean value of these chemical elements in the notation [X/Fe] for the giant stars of NGC 2447 analysed in this work (red squares). The abundance ratios of these elements are also compared with the abundance ratios obtained in two previous studies done for field stars, Mishenina et al. (2006) and Luck & Heiter (2007). Zirconium was not included in this comparison because both studies above mentioned did not determine its abundance. In addition, we also obtained the mean value [s/Fe] for the sample of Mishenina et al. (2006) and Luck & Heiter (2007). Fig. 13 shows that the giant stars of NGC 2447 are slightly enriched in the elements of the s-process with respect to the field giants. Previous heavy-element abundance determination for the open cluster NGC 2447 has already been reported by Reddy et al. (2015) for the stars #28, #34, and #41. The heavy-element abundance pattern for these stars is also shown in Fig. 14 (blue squares) and is in good agreement with our results.

The barium abundance is also given in Table 9. As we can see all the stars show an enhancement of barium. However, the large overabundance is not followed by the other s-process, which have smaller increases. The most likely explanation for this apparent contradiction, was given by Reddy & Lambert (2017). In that paper, the authors showed that the observed barium overabundance was not related with a nucleosynthesis process but is due to a underestimation of the microturbulent velocity since the strong Ba ii line used for barium determination is formed in the upper layers of the photosphere where microturbulent velocity is higher. In fact considering a microturbulent velocity of 0.3 and 0.6 km s⁻¹ greater than derived for the stars NGC 2447-3 and 4, the barium abundance can decrease between 0.3 and 0.8 dex.

In Fig. 14, we compare our mean values of the abundance ratios for the elements of the s-process (red squares) with the mean values of the heavy elements abundance ratios of other open clusters from the literature, analysed with high-resolution spectroscopy. As defined in Section 3.3, the mean value of the abundance ratios of the elements of the s-process, s, in the notation [s/Fe] is given by the abundance ratios ([X/Fe]) of the elements Y, Zr, La, Ce, and Nd. Barium was not included in the computation of the mean [s/Fe].

The mean abundance ratios from Reddy et al. (2015) are also shown in this figure as blue squares. Young open clusters, those with ages younger than 1.0 Gyr (green squares) show heavy-element abundance enhancements compared to the field giants. Maiorca et al. (2011) noticed that the youngest open clusters are the most s-process enriched when compared to older clusters, those with ages older than 1.0 Gyr. In order to explain why young open clusters present an overabundance of s-process elements compared to older ones, Maiorca et al. (2012) considered that AGB stars with masses smaller than 1.5 M_⊙ would present extra-mixing phenomena following by a very efficient production of neutrons via ¹³C. In fact this is seen in Fig. 15, where we plot the mean abundance of s-process elements, in the notation [s/Fe], versus the age of the open cluster using the results derived by Maiorca et al. (2011) and by several other recent spectroscopic analysis for some open clusters. The open cluster NGC 2447 also follows the trend noticed by Maiorca et al. (2011).

4.3.6 Heavy elements: Eu

We also determined the abundance of the europium, an element created by r-process in Type II supernovae, neutron-star mergers or in kilonova emission from compact binary mergers, that is, radioactive decays of r-process nuclei with the bulk of the emission in the optical and near-infrared wavelengths (Tanaka 2016; Tanaka et al. 2017). So, this means that it europium is not created in low-mass stars (Burris et al. 2000; Wanajo & Ishimaru 2006; Rosswog et al. 2014). The europium abundance for the giant stars of NGC 2447 determined in this work by spectral synthesis technique is shown in Fig. 5. Fig. 16a shows the [Eu/Fe] ratio of the giants in NGC 2447 (red squares) in comparison with field giants, using abundance data from Luck & Heiter (2007) and Mishenina et al. (2007). Europium abundance for the open cluster NGC 2447 was also determined by Reddy et al. (2015) for the stars #28, #34, and #41. In Figs 16(a) and (b), we also show the [Eu/Fe] ratios for these three stars (blue squares). In Fig. 16b, we show the mean [Eu/Fe] ratio of NGC 2447 in comparison with other open clusters including the mean [Eu/Fe] ratio determined by Reddy et al. (2015). NGC 2447, as well as other clusters, those younger than 1.0 Gyr and those older than 1.0 Gyr, follows the trend for field giants.

4.3.7 Yellow stragglers in NGC 2447

Besides the 12 single giants and one spectroscopic binary analysed in this work, our spectroscopic observations confirm the ‘yellow straggler’ nature of three stars in NGC 2447. The stars NGC 2447-26 and 42 have already been identified as confirmed spectroscopic binaries and NGC 2447-38 as a probable spectroscopic binary by Mermilliod & Mayor (1989). They have (B − V) colours that are ‘between that of the turn-off and the red giant branch, but brighter than the subgiant branch’ as defined by Clark, Sandquist & Bolte (2004) for stars in both globular and open clusters. In fact as seen in Fig. 1 and in Table 2, they have redder colours than the turn-off stars and bluer (or yellow) colours than the red giant stars in the CMD. A spectroscopic description of these kind of stars and a definition of the terms red and yellow stragglers was already given in Sales Silva et al. (2014), where this kind of object was also identified in the open clusters NGC 2360, NGC 3680 and NGC 5822. Here, we report that NGC 2447-26, 38, and 42 share the same spectroscopic characteristics of the stars of these open clusters : (i) strong presence of veiling in their spectra as evidenced by their very low microturbulent velocities and metallicities lower than the mean open cluster metallicity (Table 5) and (ii) the source of veiling is due to the presence of an A-type secondary star in the binary system.

In Fig. 17, we show the spectra of the stars NGC 2447-26, 38, and 42 in comparison with another yellow straggler star NGC 5822 - 4. Following Sales Silva et al. (2014), we also show the spectra of HD 65810 (red line), an A-type main-sequence star. Like the spectra of the ‘yellow-straggler’ stars in the open clusters NGC 2360, NGC 3680, and NGC 5822, the spectra of the ‘yellow-stragglers’ in NGC 2447 are strongly contaminated by the spectrum of an A-type star, since the hydrogen lines in NGC 2447-26, 38, and 42 have the same strength as in the spectrum of HD 65810.

Finally, it is worth noting that our value for the radial velocity of one of the ‘yellow-straggler’ stars, NGC 2447-38, is +10.76 ± 0.30 km s⁻¹ (Table 2). This value is very different to the one reported by Mermilliod et al. (2008), +22.91 ± 0.31 km s⁻¹. Although regarded as a suspected binary by Mermilliod & Mayor (1989), this star was not later considered as a spectroscopic binary in the radial velocity survey of the red giants by Mermilliod et al. (2008). Probably because the mean radial velocity of NGC 2447-38, based on seven measurements between 1983 February 26 and 1996 May 15, was not different from the mean cluster radial velocity of 22.14 ± 0.75 km s⁻¹, based on the nine giants shown in Table 2, excluding the spectroscopic binaries. In table 1 of Clariá et al. (2005), this object also appears as a possible spectroscopic binary. The authors derived a radial velocity of 22.86 km s⁻¹, in agreement with the value of Mermilliod et al. (2008). It is interesting that the early work of Mermilliod & Mayor (1989) indicated that NGC 2447-38 could be a long period binary but later Mermilliod et al. (2008) did not confirm this. Probably NGC 2447-38, lies in a very eccentric orbit where significant radial velocity variations would occur only in a small phase range. Our spectroscopic analysis of this star undoubtedly confirms its binary nature.

5 CONCLUSIONS

The main conclusions of our abundance analysis employing high-resolution optical spectra of all the known giants in NGC 2447 can be summarized as follows:

• Analysis of the light elements reveals a lower carbon and higher nitrogen abundance with respect to iron, which is similar to field giant stars and giants in other clusters. We found a low mean [O/Fe] ratio for NGC 2447, −0.10 ± 0.08. Low oxygen abundances have already been determined in other giants in other young clusters. The abundances of other elements, such as sodium, aluminum, α-elements, and iron-peak elements (Ni and Cr) are similar to field giants and giants of other open clusters.

• The stars NGC 2447-3 and 7 are the most evolved stars of our sample and have lowest lithium abundance, probably indicating that deep mixing has taken place in their interiors. Conversely, to the other giants of this cluster for these two stars, we were able to obtain the ¹²C/¹³C isotopic ratio.

• For the first time, we determined the rotational velocities of the giants of NGC 2447. There are few determinations of rotational velocities in giants of open clusters including single and binary stars. Our results also showed that we did not detect any difference between the rotational velocities of single and binary stars in NGC 2447.

• The abundance of the elements produced by the s-process of NGC 2447 follows the same trend seen in Maiorca et al. (2011), that is young clusters have higher heavy-element abundances compared to older clusters that do not show such enrichment. The [Eu/Fe] ratios determined for the giants of NGC 2447 are similar to the field giants and other giants of open clusters as expected by an element mainly produced by massive stars or by short time-scale process, as neutron–neutron mergers.

• Our spectroscopic observations confirm the binary nature of the stars NGC 2447-26, 38, and 42 already known by Mermilliod & Mayor (1989) as spectroscopic binaries. They are ‘yellow straggler’ stars, that is, binary stars with a red giant primary and a main sequence secondary A-type star. The combination of the light of a giant star and an A-type star changes the (B − V) colours of these binary systems to a position between the main sequence and the red giant branch in the colour–magnitude diagram. The yellow stragglers in NGC 2447 display the same spectroscopic characteristics of those already identified in the open clusters NGC 2360, NGC 3680 and NGC 5822 (Sales Silva et al. 2014).

Acknowledgements

NAD acknowledges FAPERJ, Rio de Janeiro, Brazil, for Visiting Researcher grant E-26/200.128/2015 and financial support by RFBR according to the research project 18-02-00554. This research has made use of: the SIMBAD data base, operated at CDS, Strasbourg, France. NAD acknowledges FAPERJ, Rio de Janeiro, Brazil, for Visiting Researcher grant E-26/200.128/2015 and the Saint Petersburg State University for research grant 6.38.335.2015; iraf (Image Reduction and Analysis Facility).

REFERENCES