CHEMICAL ABUNDANCES IN FIELD RED GIANTS FROM HIGH-RESOLUTION H-BAND SPECTRA USING THE APOGEE SPECTRAL LINELIST

There are five stars analyzed here: the well-studied α Boo, a mildly metal-poor K1.5 giant, μ Leo, the prototype metal-rich star (Spinrad & Taylor 1969), which is a K1 giant, two near-solar-metallicity M giants, β And (M0III) and δ Oph (M0III), and HD199799, an asymptotic giant branch (AGB) star of spectral type MS with a mild enhancement of ¹²C and a slightly elevated ratio of C/O (relative to solar) due to third dredge-up (Smith & Lambert 1990). Figure 1 illustrates a small portion of the APOGEE wavelength coverage for four of the red giants here. This region from 15560 to 15590 Å includes representative species and how they vary with spectral type (or effective temperature) and metallicity. Lines from the main CNO-containing molecules, CO, OH, and CN fall within this region, as noted in the figure. Of particular interest is the metal-rich K-giant μ Leo, which exhibits quite strong CN lines; much of the variation in the spectrum of μ Leo compared to the other giants result from the high line density of CN. This figure also points to the desirability of analyzing large numbers of red giants via spectrum synthesis techniques using a detailed linelist.

Zoom In Zoom Out Reset image size

With a significant fraction of its red giant targets in the thin and thick disks, plus the inner bulge and bar, the five types of red giants studied here span a range of temperature and metallicity that provides a good test for ASPCAP. The APOGEE survey targets red giants with effective temperatures from T_eff = 3400 K–5000 K, having surface gravities from log g = 3.0 to −0.5, and metallicities from [Fe/H] = −5.0 to +1.0. These nearby giants have well-measured photometry and parallaxes, so temperatures, luminosities, and masses are all rather well constrained. In addition, four of the five stars have previous abundance analyses in the literature and thus provide comparison tests for the abundances derived from the APOGEE linelist.

Deriving stellar chemical abundances from high-resolution spectra requires as basic input the effective temperature (T_eff), surface gravity (characterized as log g with g as cm s⁻²), and overall metallicity as the fundamental parameters of the model atmosphere. In addition, high-resolution abundance analyses of cool stars that use static 1-D stellar atmospheres also require the derivation of a non-thermal Doppler-like broadening term called microturbulence (ξ), which is determined as part of the spectral analysis. The first step is the determination of T_eff and, for these relatively nearby red giants, is set by the near-infrared magnitudes of J and K. This particular choice is used because the APOGEE targets are selected from the Two Micron All Sky Survey (2MASS) catalog (Skrutskie et al. 2006) and therefore all have (J − K) available. The aim is to focus on utilizing near-IR data and then compare to published literature results.

All of the red giants here have parallaxes measured by Hipparcos and thus have distances known to various levels of accuracy. The parallax is used to set the distance and thus M_K. Bolometric corrections from Bessell et al. (1998) are used in conjunction with (J − K) to determine M_bol and consequently stellar luminosity. Basic data for the red giants are presented in Table 1.

The most important derived quantity affecting the overall shape of the spectrum is T_eff, which here is based on an average of two T_eff–(J − K) calibrations; one from Gonzalez Hernandez & Bonifacio (2009) and the other from Bessell et al. (1998). The Gonzalez Hernandez & Bonifacio calibration is defined in the 2MASS photometric system of (J − K_S), so the original Johnson (1966) (J − K) colors and K magnitudes for these bright red giants (which have saturated magnitudes in the 2MASS catalog) were transformed to 2MASS values using the prescription from Carpenter (2001). The differences between the calibrations of Bessell et al. (1998) and Gonzalez Hernandez & Bonifacio (2009), for a given metallicity, are less than 30–50 K for T_eff ⩽4500 K, but above this temperature the Bessell et al. (1998) becomes cooler by about 100 K for a given (J − K_S) (where we have transformed the Bessell et al. scale onto the 2MASS system). There is also a small metallicity effect on the (J − K_S) calibration which, for a given (J − K_S), can be ∼100 K over the broad metallicity range of [m/H] = −2.0 to +0.0 (where m is taken to represent the generic abundance of metals, i.e., those elements other than H and He). Given the differences between the two independent T_eff calibrations of less than 30–50 K over the temperature range of the target stars, as well as their modest range in metallicity, and well-determined magnitudes with ∼0.02 uncertainty in (J − K_S) (which translates to ΔT_eff = 30 K), the derived values of T_eff have uncertainties of ±50 K or less.

Given the stellar luminosity and effective temperature, these fundamental properties are then compared to stellar evolution models presented in Bertelli et al. (2008, 2009; with data from these papers taken from the Web site http://stev.oapd.inaf.it/YZVAR) to estimate stellar masses and, ultimately, surface gravities through

$$\begin{eqnarray*} &&{g/g_{\odot } = (M/M_{\odot }) \times (L_{\odot }/L) \times (T_{\rm eff}/T_{\rm eff \odot })^{4}.} \end{eqnarray*}$$

Determining the surface gravity from evolutionary tracks is an iterative process because it requires knowledge of the stellar metallicity. This is illustrated in Figure 2 where the stellar luminosity and T_eff are plotted along with evolutionary models from a range of masses, with each panel having a different metallicity: Z = 0.017 (top panel), 0.008 (middle panel), and 0.004 (bottom panel), corresponding to overall metallicities of [m/H] = +0.07, −0.26, and −0.56, respectively, when taking Z = 0.0145 for the Sun, as suggested by Lodders (2010). The illustrated evolutionary tracks follow stellar evolution up to the tip of the first-ascent red giant branch (which we label RGB). Initial model metallicities for the stars studied here are based on previously published results, with β And, δ Oph, and HD199799 started with solar metallicity, α Boo with [Fe/H] = −0.5, and μ Leo with [Fe/H] = +0.3. As there are numerous Fe i lines in this spectral region, a sample of Fe i lines are used to determine the iron abundance. Given the limited range of metallicity spanned by the nearby red giants studied here, the Fe abundance is a good proxy for the overall stellar metallicity. If the derived Fe abundance differed by more than 0.1 dex from the assumed value, a new surface gravity was derived and a new model generated in order to re-analyze the Fe i lines until convergence between model abundance and derived abundance. For these relatively well-studied red giants, no more than two iterations were required until convergence. Table 2 presents, for these red giants, the derived parameters T_eff, log (L/L_☉), mass, log g, microturbulent velocity (ξ; see Section 4.2), and overall metallicity, which is represented by [Fe/H] as discussed in Section 4.2.

Zoom In Zoom Out Reset image size

Given the position of the stars in the log L versus T_eff plane shown in Figure 2, it can be ascertained that four stars (α Boo, μ Leo, δ Oph, and β And) are most likely first-ascent red giants of low to moderate mass. Based on its warmer temperature and somewhat lower luminosity, it may be more likely that μ Leo is in or very near the core-He burning phase (clump giant) and has thus already ascended the RGB and experienced the core-He flash. It is also possible that the other three stars already noted above may be post-He core-burning stars ascending the AGB; however, lifetimes on the two separate evolutionary sequences would favor them as being on the RGB. All fall above the "luminosity bump" (Fusi Pecci et al. 1990) and would have thus experienced any extra mixing that may occur during that phase of stellar evolution.

The coolest and most luminous red giant in this sample, HD199799, is an AGB star that is experiencing the third dredge-up, as evidenced by the observation of Tc I in its spectrum and its elevated s-process abundances, such as [Y/Fe] ∼ +0.5 or [Nd/Fe] ∼ +0.4 (Smith & Lambert 1988, 1990).

Due to both its relatively large cosmic abundance and large number of energy levels, iron has historically been used as a diagnostic species for certain stellar parameters, as well as an overall metallicity indicator, so the results for Fe i lines selected from the APOGEE linelist (M. Shetrone et al., in preparation) are discussed here. In particular, Fe i lines are well suited for setting the microturbulent velocity, ξ, which is required for 1-D model atmosphere abundance analyses.

A large number of Fe i lines fall within the APOGEE spectral window (15100–16900 Å), although there are no Fe ii lines strong enough to be detected in these red giants. Since all abundance determinations are done via spectrum synthesis, the Fe i lines were culled to include those lines that contain only contributions from Fe i (this was determined by preliminary synthesis of the red giant spectra, then setting the Fe abundance to zero and ensuring that the spectral feature in question vanished); since the Fe i lines are so numerous, this stringent selection still results in an adequate list of lines. Table 3 presents the Fe i lines used for the iron abundance determinations. As is true of most of the atomic lines in the H band, the excitation energies are dominated by rather high-excitation lines, although for Fe i there are two lower-excitation lines.

The Fe i lines span a large enough range in line strength that they can be used to set the microturbulent velocity. This parameter is defined by the value of ξ that produces no trend in the Fe abundance as a function of line strength. For a range of microturbulent velocities Fe abundances were determined from each Fe i line and the adopted value of ξ was one in which a linear regression between A(Fe) as a function of log(W_λ/λ) yielded a slope of zero. This procedure is illustrated graphically in Figure 3, where the abundances derived for each individual line are plotted versus ξ (the continuous lines); the circle with error bars shows where the slope of A(Fe) with line strength goes to zero.

Zoom In Zoom Out Reset image size

The microturbulent velocities set by the Fe i lines, as well as the Fe abundances are listed in Table 2 as part of the derived set of stellar parameters. The individual line-by-line abundances as determined for each Fe i line are also listed in Table 3. The Fe abundances were used to set the final overall metallicities of the ATLAS9 model atmospheres for the analysis of additional elements via atomic or molecular lines.

Abundances for ¹²C, ¹⁴N, ¹⁶O, and the minor carbon isotope ¹³C are derived from combinations of vibration–rotation (V-R) lines of CO (X¹Σ⁺) and OH (X²Π), along with electronic transitions of CN (A²Π–X²Σ). Although the details of the ingredients that went into these molecular lines in the APOGEE linelist will be found in M. Shetrone et al. (in preparation), a few highlights are noted here. The adopted dissociation energies (D₀) are 11.092 eV for CO, 4.395 eV for OH, and 7.76 eV for CN. The gf-values are from Goorvitch (1994) for CO, Goldman et al. (1998) for OH, and the CN gf-values are taken from the baseline linelist of Kurucz (1993), with updates to these values from the prescription given in Melendez & Barbuy (1999).

The procedure used for the CNO analysis, since all red giants studied here have C/O ⩽ 1, is to use CO to set the carbon abundance. With that carbon abundance, OH provides an O abundance, which, if different from the initial value (taken to be the solar value scaled by the stellar value of [Fe/H]), is used to re-analyze the CO lines. This process is repeated until the C and O abundances yield consistent abundances from CO and OH. With these values of C and O, the CN lines are used to derive the abundance of nitrogen. In general, the abundance of N has little to no effect on the CO and OH lines, however, the respective final C, N, and O abundances all provide self-consistent results from CO, OH, and CN.

All spectral features were synthesized and the molecular regions that were used to extract the abundances are listed in Table 4. In this case, if a range of wavelengths is given it indicates that a spectral window containing numerous molecular lines was used (for CO and OH), while for CN individual transitions were identified and synthesized. As with Table 3, the abundances derived for each feature are also listed in Table 4. In the case of ¹³C, as has been customary for minor isotopes, the abundance is listed as the ratio of ¹²C/¹³C, with A(¹²C) being set by the mean abundance of carbon-12.

With the Fe and C, N, and O abundances in hand, additional elements are derived via atomic spectral lines in the wavelength interval 15100–16900 Å. The APOGEE linelist was searched for suitable species, all neutral over the T_eff range of the red giants here, and some 12 other elements were deemed detectable and able to be analyzed quantitatively: Mg i, Al i, Si i, K i, Ca i, Ti i, V i, Cr i, Mn i, Co i, Ni i, and Cu i. The additional atomic lines are listed in Table 5. As with the molecular lines, if a precise wavelength is shown, the transition in question is represented by a single line, however, if the wavelength is approximated by an integer value it indicates multiple transitions due to hyperfine (hfs) splitting or isotopic splitting. In the case of isotopic components, solar-system isotopic fractions are assumed and, in general, this assumption has no significant effect on the derived total atomic abundances. Figure 4 illustrates the fitting procedure for an atomic line from Mn I in δ Oph, with the underlying hfs components illustrated, as well as nearby blending features, with several synthetic spectra plotted which have differing Mn abundances. As with the previous Tables 3 and 4, individual abundances are provided for each line or feature. As noted previously, the details of the generation of the APOGEE linelist will be found in M. Shetrone et al. (in preparation).

Zoom In Zoom Out Reset image size

Table 6 provides a summary table of all abundances listed as their mean values and standard deviations from the lines or spectral intervals shown in Tables 3– 5. In cases where only one line or region was observed, no standard deviation is given. The abundance of ¹³C is given as the isotopic ratio of ¹²C/¹³C.

Most of the abundances presented here are derived from more than one atomic or molecular line or features of lines, thus internal consistency in the average abundances is defined by the scatter from the individual features. This scatter is characterized by the standard deviations of Table 6. Because multiple features arise from differing line strengths (weak versus strong), excitation potentials, or ionization fractions, these respective standard deviations provide some estimate as to how well the 1-D LTE analysis can recover the details of the underlying stellar spectrum from the basic model atmosphere parameters T_eff, log g, metallicity, and derived value of ξ. The typical standard deviations from Table 6 are ∼ ±0.05 dex. Of note is the metal-rich K-giant μ Leo, which exhibits exceptionally strong CN lines, that are numerous throughout this H-band window. The background CN absorption creates blending which lowers the overall signature of other abundances and increases the uncertainty in these other abundances; this is reflected in the generally larger standard deviations found in the μ Leo abundances.

Besides internal consistency from the various spectral lines of different species, it is useful to explore their sensitivities to stellar parameters in order to characterize the magnitudes of possible systematic effects due to uncertainties in the four primary parameters of effective temperature, surface gravity, metallicity, and 1-D derived microturbulent velocity. For this discussion T_eff is written as T, log g as G, model atmosphere metallicity ([m/H]) as m, and microturbulent velocity as ξ. Because the abundance for a given element, A = log [N(A)/N(H)] + 12. = A(T, G, m, ξ), the incremental change in this abundance due to small variations in the parameters is given by

$$\begin{eqnarray*} &&dA=(\partial {A}/\partial {T})dT + (\partial {A}/\partial {G})dG + (\partial {A}/\partial {\xi })d\xi + (\partial {A}/\partial {m})dm. \end{eqnarray*}$$

The scatter in the abundance caused by small changes in the stellar parameters is then defined as

$$\begin{eqnarray*} &&\sigma ^{2}_{\rm A} = \langle dA^{2}\rangle \end{eqnarray*}$$

and dA² can be written as

$$\begin{eqnarray*} (dA)^{2} & = & [(\partial {A}/\partial {T})dT]^{2} + [(\partial {A}/\partial {G})dG]^{2}\\ &&+ [(\partial {A}/\partial {\xi })d\xi ]^{2} + [(\partial {A}/\partial {m})dm]^{2}\\ &&+(\partial {A}/\partial {T})(\partial {A}/\partial {G})[dTdG + dGdT]\\ &&+(\partial {A}/\partial {T})(\partial {A}/\partial {\xi })[dTd\xi + d\xi dT]\\ &&+(\partial {A}/\partial {T})(\partial {A}/\partial {m})[dTdm + dmdT]\\ &&+(\partial {A}/\partial {G})(\partial {A}/\partial {\xi })[dGd\xi + d\xi dG]\\ &&+(\partial {A}/\partial {G})(\partial {A}/\partial {m})[{dGdm} + dmdG]\\ &&+(\partial {A}/\partial {\xi })(\partial {A}/\partial {m})[d\xi dm + dmd\xi ]. \end{eqnarray*}$$

The partial derivative terms are calculated by changing each model parameter and determining the resulting change in logarithmic abundance, A. Since the perturbation in stellar parameters is relatively small, the changes are approximated by linear trends. As this paper is a limited study of a few stars as a test of the APOGEE linelist, a test of abundance sensitivity to stellar parameters is carried out for a representative model taken with T_eff = 4000 K, log g = 1.3, [m/H] = +0.0, with ξ = 2.0 km s⁻¹. The predicted equivalent widths were generated for the lines used in the abundance analysis and the stellar parameters were then perturbed by dT = +50 K, dG = +0.2 dex, dm = +0.1 dex, and dξ = +0.2 km s⁻¹, with new abundances derived for each separate perturbation, giving the change in abundance for each parameter change. These coefficients, which reflect the change in abundance, are listed in Table 7 and provide an overall view of how sensitive the derived abundances are to changes in stellar parameters that are representative of realistic errors for these particular red giants.

Columns 6 and 7 in Table 7 provide the values of dA two cases: one using only the first four terms which contains the change in abundance due to changes in each primary parameter (dA' in Column 6, which is valid if all parameters are independent of each other), while Column 7 includes the covariant terms. Within the analysis technique applied here, it was found that the microturbulent velocity did not measurably depend on the derived effective temperature or gravity, and the dT−dξ and dG−dξ cross terms are not included. As can be seen by the comparison of dA' and dA, the covariant terms contribute significantly, in many species, to the uncertainty. The largest covariant terms are typically those between T_eff and log g, caused by the slope of the RGB in the T_eff–log(L/L_☉) plane (since L ∼T⁴_eff/g).

The abundances derived here rely on the APOGEE H-band spectral window from 15100 to 16900 Å, along with parallaxes and photometry. Historically, most detailed abundance patterns obtained from high-resolution spectroscopy have tended to be based on wavelengths below 10000 Å, so it is useful to compare the abundances based on the APOGEE linelist with those from other studies for stars in common. Within the sample of red giants discussed here, previous high-resolution abundance analyses have been conducted on α Boo, μ Leo, β And, and HD 199799.

4.5.1. α Boo

4.5.2. μ Leo

4.5.3. β And

4.5.4. HD 199799

Based on their luminosities and effective temperatures (Figure 2), four of the red giants studied here are either first-ascent giants or, perhaps for μ Leo, a clump red giant. HD 199799 is in a more advanced stage of stellar evolution where it is undergoing thermal pulses and third dredge-up on the AGB. The combination of nucleosynthesis and convective mixing in evolved red giants should result in surface abundances of ¹²C, ¹³C, and ¹⁴N that have been altered from their initial (main sequence) values due to CN-cycle H-burning and first dredge-up.

The result of CN-cycle mixing is seen mostly easily in the ¹²C/¹³C ratios which are much higher in main-sequence stars, with the Sun having a value of 89. All four of the RGB or clump giants have quite low values, ranging from ¹²C/¹³ C ∼ 6–20. Gilroy (1989) and many subsequent studies of both open and globular clusters (e.g., Mikolaitis et al. 2012 or Briley et al. 1995) have noted that the carbon isotopic ratios correlate with stellar mass in stars above the luminosity bump on the RGB or in the clump; red giants more massive than M ∼ 2.5 M_☉ have values of ¹²C/¹³C ∼ 22–30, while lower-mass red giants exhibit decreasing isotopic ratios with declining mass. The mass estimates derived here for field stars from distance–luminosity and stellar evolutionary tracks are modestly well defined and the top panel of Figure 5 shows the ¹²C/¹³C values versus the estimated stellar mass. Also plotted are representative results for open clusters from Gilroy (1989) and Mikolaitis et al. (2012), and the globular cluster M71 from Briley et al. (1995). The cluster red giants have recently evolved from the cluster turn-offs, which have a well-determined mass. For the lower-mass stars included in these cluster studies, evolution along the RGB is rapid (∼10⁷ yr) when compared to main-sequence lifetimes (∼10⁸–10⁹ yr) so the red giant mass is similar to the turn-off mass. This can be demonstrated using the evolutionary tracks from Bertelli et al. (2009), where an isochrone with a turn-off mass of 4.0 M_☉ has a red giant tip mass of 4.2 M_☉, or a ratio of RGB to turn-off (TO) masses of 1.05. This ratio of masses becomes even closer to 1.0 for lower masses, which include all of the points in Figure 5. The points from Gilroy represent the mean values of ¹²C/¹³C for 19 open clusters with a total of 55 red giants in her sample. The mean carbon isotopic ratios for two other open clusters (with a total of 10 red giants from Cr261 or NGC 6253) from Mikolaitis et al. (2012) are also shown (open blue squares). A representative globular cluster, M71, which has a lower main-sequence mass (M ∼ 0.9 M_☉) from Briley et al. (1995) is plotted as the open blue triangle and represents the mean of two of the CN-weak stars in this cluster. The CN-strong giants that Briley et al. also studied are not plotted, as they exhibit slightly lower isotopic ratios but almost certainly represent second-generation stars in M71 that were formed from large fractions of the ejecta of first-generation intermediate-mass red giants and thus are not representative of single-star red giant evolution.

Zoom In Zoom Out Reset image size

The curves in the top panel of Figure 5 represent stellar evolutionary model results from Charbonnel & Lagarde (2010); the solid black line illustrates their predictions for standard RGB dredge-up (no "extra mixing" or rotation included). Clearly, standard RGB first dredge-up does not fit the observed trend of ¹²C/¹³C for red giants with M ⩽ 2.2 M_☉. This poor comparison between theory and observation has led to work in trying to identify other types of "extra mixing" mechanisms that would lead to lower carbon isotopic ratios in the lower-mass red giants. Two such hypothesized mechanisms are thermohaline mixing or thermohaline mixing coupled with rotational mixing, as discussed recently by Charbonnel & Lagarde (2010). The dashed curves present results from Charbonnel & Lagarde also, with the short dashes for thermohaline mixing only and the long dashes for thermohaline mixing plus rotational mixing for models with initial equatorial rotational velocities of 110 km s⁻¹. The stellar models that incorporate these particular types of extra mixing provide fair agreement with the observations, although these are not the only types of processes that have been studied and are used here to merely illustrate the effect of extra mixing. The trends found in the clusters also overlap well with the field stars analyzed here, except for HD 199799, which is in a different evolutionary phase than the other red giants. This star has enriched its surface in ¹²C that was synthesized by He-burning as a TP-AGB star. In the simplest case of pure ¹²C dredge-up, the surface ¹²C/¹³C ratio would increase, which is where HD 199799 lies in the top panel of Figure 5 relative to the other red giants. Such an effect has been noted by both Lambert et al. (1986) and Smith & Lambert (1990), where these studies find increasing values of ¹²C/¹³C as TP-AGB stars evolve from M to MS to S to C stars (with increasing values of ¹²C/¹⁶O).

The bottom panel of Figure 5 plots the abundance by number ratio of N(C)/N(N) versus stellar mass, where the carbon abundance is now the total of ¹²C + ¹³C: for reference, the solar ratio is 3.80. Nitrogen is also the total abundance, however, in the case of the observations, ¹⁵N is ignored as its abundance is very low (and not detected in these red giants) when compared to ¹⁴N. There is an observed decrease in C/N with increasing mass as predicted by both standard RGB dredge-up and dredge-up plus thermohaline mixing (Charbonnel & Lagarde 2010), which are shown as the solid and dashed curves, respectively, and are from the same models as in the top panel. In the case of the C/N ratios, the total carbon and nitrogen abundances are not as strongly influenced by non-standard mixing as the carbon isotopic ratios.

Figure 6 combines the C/N ratios versus ¹²C/¹³C ratios for both the observed red giants and the models from Charbonnel & Lagarde (2010). When displayed this way, the different behavior between C/N versus ¹²C/¹³C is striking and these two values are purely spectroscopically derived quantities. The trend of C/N with ¹²C/¹³ seems to provide a sensitive indicator of red giant mixing processes, as well as yields a mass estimate for stars on the RGB. Again, HD 199799 stands apart from the other red giants as it has internally enhanced its ¹²C abundance by about a factor of two since it was an RGB star, based on its ¹²C/¹³C and mass, relative to other RGB stars in the top panel of Figure 5. The dash-dotted line connects to the region in this diagram where HD 199799 would fall if its surface ¹²C abundance were reduced by a factor of two, which is a simplified sketch of how TP-AGB stars might evolve across such a diagram from the RGB.

Zoom In Zoom Out Reset image size

The assumption of only CN-cycle mixing taking place in the red giants is tested in Figure 7, where the total carbon abundance (¹²C + ¹³C) is plotted versus the nitrogen abundance (here ¹⁴N). The solar values of C and N are also shown, along with the dashed line illustrating scaled solar abundances for both C and N. The continuous blue curves represent decreasing carbon and increasing nitrogen abundances from the scaled solar lines, with the constraint that the number of C + N nuclei is conserved, as in the CN-cycle (with no leakage into the ON-cycle). Each curve is scaled in its initial C and N abundances by the stellar value of [Fe/H] for α Boo, β And, δ Oph, and μ Leo. CN-cycle mixing alone is sufficient to represent the derived abundances of C and N with initial abundances having [C + N/Fe] ∼ 0.0 for each RGB star. Due to having undergone a third dredge-up and ¹²C enrichment, HD 199799 will be offset from its CN curve, so no curve is plotted for it. The dash-dotted line simply shows the shift in A(C) if HD 199799 has doubled its RGB ¹²C abundance. Having nearly the same metallicity as β And, HD 199799 might be expected to have been near the CN curve for β And before it began its evolution as a TP-AGB star.

Zoom In Zoom Out Reset image size

The APOGEE H-band wavelength region presents good opportunities for probing many nuclear and mixing processes that occur over the various phases of red giant evolution.

Other than the C and N isotopes, the vast majority of APOGEE red giant targets will not have altered their surface abundances of the other studied elements, so the remaining elements are dominated by chemical evolution within the various stellar populations of the Milky Way (or its accreted systems). The derived abundances of each of the other elements are discussed in the light of observed trends of these elements with A(Fe) from other studies in the literature. The number of published abundances is now enormous; however, in this first exploratory study using the APOGEE linelist, a comparison with a small number of selected literature sources is sufficient. For comparison purposes, in this paper a stellar sample representative of the Milky Way stellar populations was constructed using Reddy et al. (2003; thin disk with some thick disk and having 58 stars), Reddy et al. (2006; thick disk with 176 stars), Johnson (2002; halo with 23 stars), and Fulbright (2002; halo with 178 stars).

The investigation of elements derived from the APOGEE linelist is via a number of figures, beginning with Figure 8, which presents values of O/Fe, Mg/Fe, and Al/Fe versus A(Fe); the shorthand notation here of x/Fe denotes the logarithmic number ratio of element x to iron, log[N(x)/N(Fe)], or A(x)–A(Fe). This nomenclature is chosen over the [x/Fe] in the plots as it presents a direct logarithmic number ratio with the solar abundance ratio being noted in each panel, thus no reference to the underlying solar abundance is needed, but the relative scales are equivalent to [x/Fe]. The top panel of Figure 8 shows the behavior of O/Fe and, as has been well known for decades, values of O/Fe are elevated in the thick disk and halo stars. Among the field red giants here, the slightly metal-poor objects α Boo and β And both have O/Fe enhanced by ∼ +0.2–+0.3. The near-solar metallicity M-giant δ Oph has a near-solar ratio of O/Fe, while the metal-rich giant μ Leo also exhibits a near-solar value O/Fe.

Zoom In Zoom Out Reset image size

The behavior of Mg/Fe is shown in the middle panel of Figure 8 and, like oxygen, Mg/Fe ratios are typically elevated in the thick disk and halo populations. The four red giants closest to solar in A(Fe) all exhibit basically solar Mg/Fe ratios, with α Boo having a slight enhancement of ∼ +0.1 dex, which overlaps the thick disk populations.

Aluminum abundances are plotted in the bottom panel of Figure 8 and the behavior of Al/Fe versus A(Fe) is not so well defined, with relatively large scatter. Mu Leo is found to be modestly enhanced in Al/Fe by about +0.10–+0.15 dex, as is α Boo, with an enhancement of ∼+0.2.

The abundances for the rest of the elements studied here are illustrated in Figure 9 (for Si, K, and Ca), Figure 10 (for Ti, V, and Cr), Figure 11 (for Mn, Co, and Ni), and Figure 12 (for Cu). In all elements, the general trends found in the general Galactic populations are recovered in the small sample of field red giants included here. The one exception is Ti in δ Oph and μ Leo, both of which are found to exhibit enhanced values of Ti/Fe of +0.2 dex. There is no obvious explanation for this, however, the continuing analysis of APOGEE spectra with this linelist will reveal if this possible effect appears in significant numbers of other red giants. The results presented and discussed here indicate that H-band spectra analyzed in LTE with the APOGEE linelist and 1-D model atmospheres can provide accurate results for chemical abundances.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size