Free Access
Issue
A&A
Volume 560, December 2013
Article Number A46
Number of page(s) 8
Section Stellar atmospheres
DOI https://doi.org/10.1051/0004-6361/201322586
Published online 03 December 2013

© ESO, 2013

1. Introduction

High-resolution spectroscopy in the near-infrared (NIR) is a powerful tool for measuring chemical abundances of cool giant and supergiant stars. It is especially critical in very reddened environments such as the inner Galaxy, where extinction is so severe as to prevent any reliable measurement at shorter wavelengths.

Recently, our team made the first technical commissioning of GIANO, a cross-dispersed NIR spectrograph for the Telescopio Nazionale Galileo (TNG) at the Roque de Los Muchachos Observatory in La Palma (Spain). GIANO delivers a spectrum that covers in a single exposure the wavelength range from 0.95 μm to 2.4 μm at a resolving power R ≃ 50 000. The main disperser is a commercial R2 echelle grating with 23.2 lines/mm working in quasi-Littrow configuration on a d = 100 mm collimated beam. Cross dispersion is achieved via a network of fused silica and ZnSe prisms that work in double pass, that is, they cross-disperse the light both before and after it is dispersed by the echelle gratings, thus producing curved spectral orders. The echellogram on the 2k × 2k detector spans 49 orders, from #32 to #80. The spectral coverage is complete up to 1.7 μm. At longer wavelengths the orders become larger than the detector. The effective spectral coverage in the K-band is about 75%. Light from the telescope feeds a bundle of two IR-transmitting ZBLAN fibers, with a core of 85 μm, corresponding to a sky-projected angle of 1 arcsec. The two fibers are aligned and mounted inside a custom connector. The cores are at a distance of 0.25 mm, equivalent to a sky-projected angle of about 3 arcsec.

Owing to the constraints set by the visitor focus, the fiber entrance was coupled to the TNG using a provisional, simplified focal adapter that consisted of a commercial CaF2 singlet lens positioned 26 mm before the fibers. The focal adapter was mechanically mounted at a fiducial position, no further adjustment of the optical axis was possible. This unfortunately resulted in a very reduced efficiency of the system, and only bright targets were observable at that time. More technical details on the instrument can be found in Oliva et al. (2012a,b, 2013).

As a first science verification of the GIANO performances, we observed three bright red supergiants (RSGs) in the star cluster RSGC2. RSGC2 is one of the few young, massive (4 × 104  M) clusters in the Galactic plane, located at the base of the Scutum-Crux arm and at the tip of the Galactic bar (Davies et al. 2007) at a distance of ≃3.5 kpc from the Galactic center. Such rare clusters are rich in RSG stars and represent ideal laboratories for studying the evolution of massive stars as well as for physically and chemically characterizing the most recent and violent star formation events in the inner Galaxy. However, one can only study them in the IR or at radio wavelengths, because of high visual extinction (AV ≈ 10 − 30 mag) that affects the Galactic plane. Chemical abundances of Fe, C, and alpha-elements have been measured in 12 RSG stars, members of RSGC2, by using medium-resolution (R ≃ 17 000) spectra obtained with NIRSPEC at the KeckII Telescope (Davies et al. 2009b, hereafter D09). In some of these RSGs Verheyen et al. (2012) also found SiO maser emission.

We present and discuss the main atomic and molecular lines identified in the GIANO spectra of the observed three RSGs, which are members of the RSGC2 star cluster and are in common with the D09 sample. We also report the inferred chemical abundances of CNO and F from molecular lines, iron-peak, alpha, and other light elements and a few s-process elements from atomic lines.

Table 1

RSG stars observed with GIANO.

thumbnail Fig. 1

GIANO 2D (sky-subtracted) spectra of one of the observed RSG stars. Sky subtraction has been performed by nodding on fiber, resulting in one positive and one negative spectrum.

Open with DEXTER

2. Observations and data reduction

For the observed targets, the 2MASS reference name, coordinates, spectral type and magnitudes, reddening and radial velocities from Davies et al. (2007) are reported in Table 1.

The reddening estimates do not affect the line equivalent width measurements since at NIR wavelengths the continuum is still dominated by the stellar photosphere and reddening dilutes both the line and the adjacent continuum flux by the same amount.

The data were collected during the technical night on July 30, 2012. A first exposure of 5 min was acquired by centering the star+sky on fiber #1 and the sky alone on fiber #2. Then we acquired a second exposure with the same integration time but with star+sky on fiber #2 and the sky alone on fiber #1 by nodding the telescope. By subtracting the 2D spectrum of the second exposure from the 2D spectrum of the first exposure we obtained the pure star spectrum (see Fig. 1). Nodding on fiber is very effective to properly subtract sky emission and instrumental background. The geometry of the orders was determined using flat exposures with a tungsten calibration lamp. The 2D spectrum was thus rectified and the spectra from each order were extracted by summing six pixels around each fiber, in the direction perpendicular to dispersion. Wavelength calibration was determined by feeding the fibers with the light from a U-Ne lamp. The wavelengths of the uranium lines were taken from Redman et al. (2011), while for neon we used the table available on the NIST (Kramida et al. 2012). The λ versus pixel relationship was obtained starting from a physical model of the instrument. This procedure is part of the pipeline that we are developing for the instrument. The resulting wavelength accuracy was about λ/300 000 rms, that is, 0.05 Å for lines in the H-band, while the overall signal-to-noise ratio is about 50. For the three observed stars we found radial velocities consistent with the values reported by Davies et al. (2007) and listed in Table 1.

Figure 2 shows a few examples of the observed spectra for iron-peak and s-process elements, Fig. 3 for alpha elements and Fig. 4 for other light elements.

thumbnail Fig. 2

GIANO spectra around some iron-peak and s-process element lines for the three observed RSG stars (dotted lines). Our best-fit models are overplotted as solid lines. Rest frame wavelengths are defined in air and thickmarks are every Å in each panel.

Open with DEXTER

thumbnail Fig. 3

GIANO spectra around some alpha element lines for the three observed RSG stars (dotted lines). Our best-fit models are overplotted as solid lines. Rest frame wavelengths are defined in air and thickmarks are every Å in each panel.

Open with DEXTER

thumbnail Fig. 4

GIANO spectra around some light element lines for the three observed RSG stars (dotted lines). Our best-fit models are overplotted as solid lines. Rest frame wavelengths are defined in air and thickmarks are every Å in each panel.

Open with DEXTER

3. Chemical abundance analysis

Deriving chemical abundances of RSG stars is an intrinsically difficult task for several reasons, namely the high level of molecular blending and blanketing in their spectra, the degeneracy in the determination of surface parameters, and the various broadening effects due to micro and macro turbulence. The use of high-resolution spectroscopy in the NIR significantly mitigates the problem of the molecular blending and blanketing, which remains critical at optical wavelengths.

To measure chemical abundances from the GIANO spectra, we used full spectral synthesis techniques and equivalent width measurements of selected lines, sufficiently isolated, free from significant blending and/or contamination by telluric absorption and without strong wings. The presence of possible telluric absorption was carefully checked on an almost featureless O-star (Hip89584) spectrum.

We computed a large grid of suitable synthetic spectra to model RSG stars by varying the stellar parameters and the element abundances. We used the same code as in D09 to facilitate the comparison. This is an updated version (Origlia et al. 2002) of the program first described in Origlia et al. (1993).

The code was also successfully used to obtain abundances of bulge field (Rich et al. 2012, and references therein) and globular cluster (Origlia et al. 2008, and references therein) giants, young clusters dominated by RSGs (Larsen et al. 2006, 2008), and to study the chemical abundances of RSGs in the Galactic center (Davies et al. 2009a). The code uses the LTE approximation, is based on the molecular blanketed model atmospheres of Johnson et al. (1980) at temperatures ≤4000 K and on the ATLAS9 models for temperatures above 4000 K, and it includes thousands of NIR atomic transitions from the Kurucz database1, Biemont & Grevesse (1973), and Melendez & Barbuy (1999), while molecular data are taken from our (Origlia et al. 1993, and subsequent updates) and Plez (priv. comm.) compilations. The reference solar abundances are taken from Grevesse & Sauval (1998).

Equivalent width measurements of the OH and CN molecular lines in the H-band, and of one HF line in the K-band were used to determine the oxygen, nitrogen and fluorine abundances.

12C and 13C carbon abundances were mostly determined from the CO bandheads in the H and K-bands, respectively, by means of full spectral synthesis, because of the high level of crowding and blending of the CO lines in these stars. Figure 5 shows the GIANO spectra of the three RSGs centered on some of the CO bandheads used for determining the carbon abundances and our best-fit models.

Equivalent width measurements of neutral atomic lines over the full spectral range covered by the GIANO spectra were used to derive abundances of Fe and other iron-peak elements such as V, Cr, Ni, of alpha (Mg, Si, Ca and Ti) and other light elements such as Na, Al, K, Sc, and of the s-process element Y, while Sr abundances were obtained from ionized lines. Table 3 (available online) reports the list of lines used in the present analysis and their measured equivalent widths.

Some other atomic lines of S (Y-band), Mn (J-band), Co, and Cu (H-band) are present in the spectra of the observed RSGs, but they are either too saturated or blended and/or contaminated by telluric absorption, which makes them ineffective for deriving reliable abundances.

This compilation is a first census of the usable lines for abundance analysis and it does not pretend to be complete. Work is in progress to identify other potential lines of interest that, however, will be checked on forthcoming observed spectra of less extreme stars for a proper calibration. We note that our compilation contains several lines in common with the Smith et al. (2013) APOGEE spectral line list.

thumbnail Fig. 5

GIANO H-band spectra of the 12CO (3–0), (5–2), (6–3) and (8–5) bandheads and K-band spectra of the 13CO (3–0) bandhead for the three observed RSG stars (dotted lines). Our best-fit models are overplotted as solid lines. Rest frame wavelengths are defined in air.

Open with DEXTER

We compared the observed spectra with models in which temperature, gravity, and microturbulence velocity are as carefully determined in D09, that is Teff = 3600 K, log g = 0.0 and ξ = 2 km s-1 for all the three stars. These parameters were fine-tuned to enable simultaneous spectral fitting of the CO and OH molecular bands and the few atomic lines in the NIRSPEC spectra. The many more CO, OH, and CN molecualr lines as well as neutral atomic lines available in the GIANO spectra allowed us to carefully verify the reliability of the parameters adopted in D09. Models with these atmospheric parameters satisfactorily reproduce both the D09 and the current GIANO spectra. However, while D09 obtained good fits to the data without including an additional macroturbulence velocity broadening, since this was not resolved at their spectral resolution of R ≃ 17 000, this is not the case for the GIANO spectra at R ≃ 50 000. We obtained a good fit to the observed spectra by modeling macroturbulence velocity with a Gaussian σ broadening of about 6.4, 7.1, and 5.7 km s-1, or equivalently Doppler-broadening of 9, 10, and 8 km s-1, for stars #1, #2 and #3, respectively. We did not find other appreciable line broadening by stellar rotation.

The final average abundances are quoted in Table 2 and plotted in Fig. 6. The impact of using slightly different assumptions for the stellar parameters on the derived abundances is discussed in Sect. 3.1.

Recently, Davies et al. (2013) discussed the problem of the most appropriate temperature scale for RSGs (see also Levesque et al. 2005), given that temperatures differing by several hundreds degrees can be inferred, using different scales. For the sake of consistency we therefore also explore models with significantly warmer (Teff > 4000 K) and cooler (Teff < 3200 K) temperatures than the one adopted in this analysis as best-fit value. We found that for temperatures above 3800K we were still able to fit the observed spectra by using very peculiar enhanced N and O abundances, providing unlikely [O/Fe]  > 0.9 dex and [C+N/Fe] > 1.0 dex. For temperatures below 3200 K, we were barely (at >1.5 sigma level) able to fit the spectra with very peculiar depleted N and O abundances, providing [O/Fe] < −0.3 dex and [C+N/Fe] ≈ −0.5 dex. The impact on the overall iron and iron-peak elemental abundances is much weaker (within 0.1–0.2 dex).

Table 2

Chemical abundances of the RSG stars observed with GIANO.

3.1. Error budget

We can quantify random and systematic errors in the measurements of the equivalent widths and in the derived chemical abundances as follows: the typical random error of the measured line equivalent widths is between 10 and 20 mÅ, mostly arising from a ±2% uncertainty in the placement of the pseudo-continuum, as estimated by overlapping the synthetic and the observed spectra. These random uncertainties in the line equivalent width measurements correspond to abundance variations ranging from a few hundredths to 1 tenth of a dex. This ≤0.1 dex error is lower than the typical 1σ scatter in the derived abundances from different lines, which normally ranges between 0.1 and 0.2 dex. The errors quoted in Table 2 for the final abundances were obtained by dividing these 1σ errors by the squared root of the number of used lines, ranging from a few to about 20. A 10–20 mÅ variation of line equivalent width value was also obtained by varying the Gaussian line broadening (due to macro turbulence) by ≃1 km s-1, but this is mostly a systematic effect.

The other systematics arise from varying the adopted stellar parameters. To properly quantify them, we generated a grid of test models with Teff between 3400 K and 3800 K, log g between 0.0 and 1.0 dex, and ξ between 2 and 4 km s-1.

thumbnail Fig. 6

Open circles: [X/H] chemical abundances as a function of the atomic number for the three RSGs observed with GIANO. Solid dots: corresponding [X/H] abundances from D09. The horizontal lines mark the [Fe/H] abundance (solid line) and the ±0.3 dex values (dashed lines), as inferred in the present analysis.

Open with DEXTER

We found that variations of ±100 K with respect to the adopted temperature of 3600 K have a weak effect on the measured equivalent widths of atomic lines (on average the variation is <10 mÅ  corresponding to a variation of few hundredths dex of the element abundance). Molecular OH and CN lines are more sensitive to temperature variations: their equivalent widths can vary by 15–20 mÅ  by varying temperature of ±100 K, corresponding to ≤0.1 dex variation of the element abundance. A variation of log g by ±0.5 dex has a negligible impact on the equivalent width of the OH lines, while it produces a variation of ≃20 mÅ (≃0.15 dex in abundance) and ≃30 mÅ (≃0.20 dex in abundance) in the equivalent widths of the atomic and CN lines, respectively. A variation of ξ by ±0.5 km s-1 has a negligible impact on the equivalent widths of CN lines, while it produces a variation of ≃15–20 mÅ in the equivalent widths of atomic and OH lines, corresponding to a variation of ≤0.1 dex of the element abundance. A somewhat conservative estimate of the overall systematic uncertainty in the abundance (A) determination, caused by variations of the atmospheric parameters, can be computed as follows: (ΔA)2 = (∂A/∂T)2T)2 + (∂A/log   g)2(Δlog   g)2 + (∂A/∂ξ)2ξ)2 and it amounts to 0.15–0.20 dex.

We also determined the statistical significance of our best-fit solution for the spectral synthesis of the CO features and the derived carbon abundances. As a figure of merit of the statistical test we adopted the difference between the model and the observed spectrum. To quantify systematic discrepancies, this parameter is more powerful than the classical χ2 test, which is instead equally sensitive to random and systematic scatters (see Origlia & Rich 2004, for more discussion and references).

Our best-fit solutions always show >90% probability to be representative of the observed spectra, while those with ±0.1 dex carbon abundance are significant at ≥1.5σ level only.

We also computed other test models with ΔTeff =  ±200 K, Δlog   g =  ±0.5 dex and Δξ =  ±0.5 km s-1, and also with corresponding simultaneous variations of the C abundance (0.1–0.2 dex) to reproduce the depth of the molecular features. CO molecular bands are very sensitive thermometers in the range of temperature between 4500 K and 3800 K. Indeed, temperature sets the fraction of molecular versus atomic carbon. At temperatures below 3800 K most of the carbon is in molecular form, which drastically reduces the dependence of the CO band strengths on the temperature itself. At temperatures ≥4500 K molecules barely survive, most of the carbon is in atomic form, and the CO spectral features become very weak. Solutions with Δlog   g =  ±0.5 dex or with Δξ =  ±0.5 km s-1 and corresponding 0.1–0.2 dex variation of the carbon abundance are significant at ≥1.5σ level only.

4. Results and discussion

The abundances of all chemical elements with measurable lines in the GIANO spectra are very similar in all the three RSGs, as expected because they are members of a star cluster.

We found iron abundances between half and one third solar, in good agreement with the previous estimates by D09. The other iron-peak elements (V, Cr, Ni) show abundances that are consistent with the iron values to within ±0.15 dex, thus fully confirming the sub-solar metallicity of the cluster.

The alpha (O, Mg, Si, Ca, Ti) as well as most of the other light (F, Na, Al,K, Sc) elements behave in a similar fashion, with [X/Fe] abundance ratios about solar (±0.15 dex); the only exceptions are [F/Fe] and [Sc/Fe] in star #1, [Na/Fe] in star #2 and [Na/Fe] and [Sc/Fe] in star #3, which are slightly enhanced (but always within a factor of two).

Interestingly, the measured [F/O] abundance ratios (+0.18 in star #1, +0.07 in star #2 and +0.21 in star #3), are fully consistent with the values measured in K-M dwarfs in the Orion Nebula Cluster by Cunha & Smith (2005) and are representative of the disk composition.

The Sr and Y s-process elements behave somewhat differently from each other: [Sr/Fe] is slightly enhanced, while [Y/Fe] is slightly depleted (by a factor of ≤2) with respect to solar.

[C/Fe] is depleted by a factor between two and three, similarly to what was found by D09, and the 12C/13C ratio is also rather low (between 9 and 11), indicating that some extra-mixing processes in the stellar interiors are at work during the post-main-sequence evolution. Evolutionary tracks of massive stars with rotation (e.g. Meynet & Maeder 2000) can account for both the overall depletion of carbon and the low 12C/13C isotopic ratio, and indeed, deep, rotationally enhanced mixing was previously suggested by D09.

[N/Fe] is enhanced in such a way that [C+N/Fe] is about 0.0 (+0.00 in star #1, +0.12 in star #2, and +0.16 in star #3), consistent with standard CN nucleosynthesis.

The mostly solar-scaled [X/Fe] abundance patterns of iron-peak, alpha, and other light elements as measured in the observed three RSGs of RSCG2 are fully consistent with the chemistry of the thin disk (see e.g. Reddy et al. 2003), that underwent chemical enrichment over long timescales with the contribution of both type II and type I supernovae. At variance, the sub-solar metallicity of RSGC2 and of its companion cluster RSGC1 (see D09) is intriguing. Indeed, abundance measurements of Cepheids in the inner disk (see e.g. Genovali et al. 2013; Andrievsky et al. 2013, and references therein) indicate the existence of a positive metallicity gradient toward the inner regions and metal abundances well in excess of solar. However, these measurements still sample a region at a minimum Galactocentric distance of about 4 kpc because of the huge extinction closer to the center. Only a few IR measurements of RSGs exist so far in the innermost disk region and in the Galactic center (e.g. Ramirez et al. 2000; Martins et al. 2008; Najarro et al. 2009; Davies et al. 2009a,b), consistent with a metal abundance of about solar and some level of alpha-element enhancement (see e.g. Cunha et al. 2007).

It is therefore clear that neither RSGs in the Scutum star clusters (D09 and the present work) nor RSGs in the center of the Galaxy follow the radial Galactic trend traced by Cepheids and other young stellar populations at Galactocentric distances RGC > 3 − 4 kpc (see D09 and Genovali et al. 2013, for more discussion and references). However, this is not surprising, because the innermost region of the Galaxy shows several different substructures, such as the bulge/bar and the ends of the spiral arms, with strong fluctuations in stellar/gas density and star formation rates. In such a complex physical and kinematic environment, the chemical enrichment process is also expected to have been quite inhomogeneous.

5. Conclusions

As a preliminary test of the scientific performances of the GIANO spectrograph during its first commissioning in July 2012, we observed three bright RSGs in the Scutum star cluster RSGC2.

A NIR spectrograph such as GIANO, which provides high spectral resolution and full spectral coverage in a single exposure, is a unique instrument to perform detailed chemical studies of cool stars in any Galactic environment, including highly reddened star clusters such as RSGC2 in the inner Galactic disk.

The high spectral resolution has allowed us to resolve and measure the macro turbulence broadening in the observed RSGs and to better identify unblended lines for accurate equivalent width measurements and chemical abundance determinations.

The simultaneous access to the YJH and K-bands offers the possibility of sampling most of the elements of interest for a

complete check of the chemical clock and of measuring from a few to several tens of lines per element, thus enabling an accurate and statistically significant abundance analysis. We used the Y-band to measure Cr and Sr, the J-band to measure K and some Fe, Cr, Ca, Mg, Si, and Ti lines, the H-band to measure 12C, N, O, Al, V, and Ni, most of the Fe lines and some Al Ca, Mg, Si and Ti lines, and finally the K-band to measure 13C, F, Na, Sc, Y and a few Al, Ti and Fe lines.

For the three observed RSGs we found overall abundance patterns consistent with the thin-disk chemistry, and we confirmed the sub-solar metallicity of the RSGC2 star cluster, as previously suggested by D09, which indicates that the inner disk in particular, but more generally the inner Galaxy, has quite an inhomogeneous chemical composition.


Acknowledgments

Part of this work was supported by the grant TECNO-INAF-2011. This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. We thank Ben Davies for his carefull referee report.

References

Online material

Table 3

Lines identified in the RSG stars observed with GIANO.

All Tables

Table 1

RSG stars observed with GIANO.

Table 2

Chemical abundances of the RSG stars observed with GIANO.

Table 3

Lines identified in the RSG stars observed with GIANO.

All Figures

thumbnail Fig. 1

GIANO 2D (sky-subtracted) spectra of one of the observed RSG stars. Sky subtraction has been performed by nodding on fiber, resulting in one positive and one negative spectrum.

Open with DEXTER
In the text
thumbnail Fig. 2

GIANO spectra around some iron-peak and s-process element lines for the three observed RSG stars (dotted lines). Our best-fit models are overplotted as solid lines. Rest frame wavelengths are defined in air and thickmarks are every Å in each panel.

Open with DEXTER
In the text
thumbnail Fig. 3

GIANO spectra around some alpha element lines for the three observed RSG stars (dotted lines). Our best-fit models are overplotted as solid lines. Rest frame wavelengths are defined in air and thickmarks are every Å in each panel.

Open with DEXTER
In the text
thumbnail Fig. 4

GIANO spectra around some light element lines for the three observed RSG stars (dotted lines). Our best-fit models are overplotted as solid lines. Rest frame wavelengths are defined in air and thickmarks are every Å in each panel.

Open with DEXTER
In the text
thumbnail Fig. 5

GIANO H-band spectra of the 12CO (3–0), (5–2), (6–3) and (8–5) bandheads and K-band spectra of the 13CO (3–0) bandhead for the three observed RSG stars (dotted lines). Our best-fit models are overplotted as solid lines. Rest frame wavelengths are defined in air.

Open with DEXTER
In the text
thumbnail Fig. 6

Open circles: [X/H] chemical abundances as a function of the atomic number for the three RSGs observed with GIANO. Solid dots: corresponding [X/H] abundances from D09. The horizontal lines mark the [Fe/H] abundance (solid line) and the ±0.3 dex values (dashed lines), as inferred in the present analysis.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.