Systematic non-LTE study of the −2.6≤ [Fe/H] ≤0.2 F and G dwarfs in the solar neighbourhood. II. Abundance patterns from Li to Eu
G. Zhao, L. Mashonkina, H. L. Yan, S. Alexeeva, C. Kobayashi, Yu. Pakhomov, J. R. Shi, T. Sitnova, K. F. Tan, H. W. Zhang, J. B. Zhang, Z. M. Zhou, M. Bolte, Y. Q. Chen, X. Li, F. Liu, M. Zhai
aa r X i v : . [ a s t r o - ph . S R ] O c t Accepted by ApJ for publication
Preprint typeset using L A TEX style AASTeX6 v. 1.0
SYSTEMATIC NON-LTE STUDY OF THE − . ≤ [FE/H] ≤ . G. Zhao , L. Mashonkina , H. L. Yan , S. Alexeeva , C. Kobayashi , Yu. Pakhomov , J. R. Shi , T. Sitnova , K.F. Tan , H. W. Zhang , J. B. Zhang , Z. M. Zhou , M. Bolte , Y. Q. Chen , X. Li , F. Liu , M. Zhai , Key Laboratory of Optical Astronomy, National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, China Institute of Astronomy, Russian Academy of Sciences, RU-119017 Moscow, Russia School of Astronomy and Space Science, University of Chinese Academy of Sciences, Beijing 100049, China School of Physics, Astronomy and Mathematics, Centre for Astrophysics Research, University of Hertfordshire,College Lane, Hatfield AL10 9AB, UK Department of Astronomy, School of Physics, Peking University, Beijing 100871, China Kavli Institute for Astronomy and Astrophysics, Peking University, Beijing 100871, China UCO/Lick Observatory, University of California, 1156 High Street, Santa Cruz, CA 95064, USAand Research School of Astronomy and Astrophysics, Australian National University, Canberra, ACT 2611, Australia
ABSTRACTFor the first time, we present an extensive study of stars with individual non-local thermodynamicequilibrium (NLTE) abundances for 17 chemical elements from Li to Eu in a sample of stars uniformlydistributed over the − . ≤ [Fe/H] ≤ +0 .
24 metallicity range that is suitable for the Galactic chemicalevolution research. The star sample has been kinematically selected to trace the Galactic thin andthick disks and halo. We find new and improve earlier results as follows. (i) The element-to-ironratios for Mg, Si, Ca, and Ti form a MP plateau at a similar height of 0.3 dex, and the knee occursat common [Fe/H] ≃ − .
8. The knee at the same metallicity is observed for [O/Fe], and the MPplateau is formed at [O/Fe] = 0.61. (ii) The upward trend of [C/O] with decreasing metallicity existsat [Fe/H] < − .
2, supporting the earlier finding of Akerman et al. (iii) An underabundance of Narelative to Mg in the [Fe/H] < − ≃ − .
5. (iv)The K/Sc, Ca/Sc, and Ti/Sc ratios form well-defined trends, suggesting a common site of the K-Tiproduction. (v) Sr follows the Fe abundance down to [Fe/H] ≃ − .
5, while Zr is enhanced in MPstars. (vi) The comparisons of our results with some widely used Galactic evolution models are given.The use of the NLTE element abundances raises credit to the interpretation of the data in the contextof the chemical evolution of the Galaxy.
Keywords:
Line: formation – Stars: abundances – Stars: atmospheres – Stars: late-type – Galaxy:
Zhao et al. evolution INTRODUCTIONThe understanding of the formation and evolution ofthe Galaxy mainly relies on the stellar spectroscopicanalysis, which gives the chemical compositions of un-evolved cool stars. Pioneering studies in this field (e.g.Wallerstein 1962; Zhao & Magain 1990a,b) have foundthat [ α /Fe] versus [Fe/H] shows a plateau below [Fe/H]= −
1, and then there is a steady decline to [ α /Fe] ∼ [email protected]@inasan.ru In the classical notation, where [X/H] = log( N X /N H ) star − log( N X /N H ) Sun . solar neighborhood, they have distinct abundance dis-tributions (see, e.g. Nissen & Schuster 1997, 2010), in-dicating the complex chemical evolution of the Galaxyat different locations. To decipher the spectral finger-prints in terms of abundances requires realistic mod-els for the stellar atmospheres and the line-formationprocesses. Still today, the vast majority of abundanceanalysis of late-type stars relies on the assumption oflocal thermodynamic equilibrium (LTE). It is expectedthat this approach quite often gives misleading results,and for many elements such systematic errors may bevery severe (see Fig.7 of Gehren et al. 2006, as an ex-ample). The principles of NLTE line formation andcodes capable of such calculations have been aroundfor a long time but have only been explored in a moresystematic fashion for a wide range of stellar parame-ters over the past decade or so (e.g. Zhao et al. 1998;Zhao & Gehren 2000; Mashonkina & Gehren 2001;Takeda et al. 2002; Gehren et al. 2004; Fabbian et al.2006; Mashonkina & Zhao 2006; Mashonkina et al.2007a, 2008; Zhang et al. 2006b; Bergemann & Gehren2008; Zhang et al. 2008; Lind et al. 2009; Shi et al.2009; Andrievsky et al. 2010; Bergemann et al. 2010;Bergemann & Cescutti 2010; Spite et al. 2011, 2012;Yan et al. 2015).Before the observations are used to give constraintto the GCE models, we need to take into account theanalysis errors of the stellar abundances, such as the de-partures from LTE and uncertainties in the atmosphericparameters. In particular, most high-resolution spectro-scopic studies are based on the LTE assumption, whichmay not be valid for spectral lines. It is suspected thatboth the unexpected behaviour of the scatter of theabundance ratios and the different behaviour betweendwarfs and giants found in the LTE studies could bedue at least partly to the neglect of the departures fromLTE (Andrievsky et al. 2010). Actually, NLTE analysisis important in the sense that it can improve the ac-curacy of stellar abundances, while LTE can achieve avery high precision with a large systematic deviation. Inorder to link the observations of abundance ratios withGCE models, we need more accuracy than the precision.In particular, a modelling technique allowing for depar-tures from LTE can be used to accurately predict ironabundances and spectroscopic stellar parameters for aset of benchmark late-type stars.The [ α /Fe] ratios can be more accurately derived byperforming the NLTE analysis of both α -elements (Mg,Si, Ca, Ti) and iron abundances. It is well known that[O/Fe] and [ α /Fe] ratios are the most important indica-tors for distinguishing of different chemical enrichment on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu I , C I , O I , Na I , Mg I , Al I , Si I -Si II , K I , Ca I ,Sc II , Ti II , Fe I -Fe II , Cu I , Sr II , Zr II , Ba II andEu II . For each listed species, the original model atomwas treated and tested by our previous studies (see Ta-ble 1 for details). The wavelength range is selected sothat the lines of the NLTE elements are presented inthe spectral coverage.In this work, we aim to define a large sample, the F &G benchmark stars, which include 51 F & G dwarfs andsubgiants in a limited range of temperatures, gravitiesand metallicities. These stars should be representativeof the different stellar populations of the Galaxy. Mostof these stars were studied under the LTE assumption inthe past years. Their accurate stellar parameters havebeen determined carefully by Sitnova et al. (2015, here-after, Paper I). It is important to have new abundancesderived from the NLTE analysis, which will better con-strain the models of the Galactic chemical evolution andthe yields of the Supernovae (e.g. Matteucci & Francois1989; Woosley & Weaver 1995; Thielemann et al. 1996;Nomoto et al. 2006).In this paper, Sect. 2 describes the stellar sample, ob-servations, and atmospheric parameters. Details of theNLTE calculations, including the atomic models andmechanisms of the departures from LTE are given inSect. 3. Section 4 presents the abundance results for thesample stars. In Sect. 5 we discuss the implications forthe GCE model and nucleosynthesis, followed by a shortsection of conclusions. STELLAR SAMPLE, OBSERVATIONS, ANDATMOSPHERIC PARAMETERSOur stellar sample, the observed stellar spectra, andthe determination of atmospheric parameters were pre-sented in Paper I. Here, we remind the main points anddescribe briefly a reduction of the infrared (IR) spec-tra undertaken to remove fringes in the echelle orders,where the O I The sample includes 51 nearby stars uniformly dis-tributed in the − . ≤ [Fe/H] ≤ +0 .
24 metallicityrange. We selected unevolved stars, i.e. mostly dwarfs,with a few subgiants added. The Galactic thin disk stel-lar population is well represented in our sample by 27stars, with [Fe/H] down to − .
78. We have eight thickdisk stars in the − . ≤ [Fe/H] ≤ − .
70 range overlap-ping only a little with that of the thin disk and 16 halostars. A membership of individual stars to the galacticstellar populations was identified based mainly on thestar’s kinematics.
Observations.
Spectra of 48 stars were obtained forour project using the Hamilton Echelle Spectrographmounted on the Shane 3 m telescope of the Lick ob-servatory. Their resolving power is R = λ/ ∆ λ ≃
60 000, the spectral coverage is 3700-9300 ˚A, andthe signal-to-noise ratio (S/N) at 5500 ˚A is higherthan 100 for most stars. For two stars their spec-tra were obtained with CFHT/ESPaDOnS, as de-scribed in Paper I. High-quality observed spectrum ofHD 140283 was taken from the
ESO UVESPOP sur-vey (Bagnulo et al. 2003). For some stars our ob-servational material was complemented with the datafrom our earlier projects, namely, VLT2/UVES, 67.D-0086A (HD 74000, BD − ◦ forBD − ◦ forHD 59374 (074.C-0364(A), R ≃
115 000), HD 59984(076.B-0133(A), R ≃
57 000), HD 100563 ( R ≃
115 000),and HD 108177 ( R ≃
115 000), 1.93 m/SOPHIE ( R = 76 500) and 1.93 m/ELODIE ( R = 42 000) forHD 64090 and BD +66 ◦ Fringes reduction.
The O I http://archive.eso.org/wdb/wdb/adp/phase3-main/query http://atlas.obs-hp.fr/sophie/ http://atlas.obs-hp.fr/elodie/ Zhao et al. standard flat fielding could be used to remove fringes,this procedure is limited by the bright scattered light.Here we apply a statistical procedure based on a set ofstellar spectra with similar exposures taken at one night.We show below that due to a comparable level of thescattered light in stellar spectra the fringes can be re-covered with a reasonable precision. Moreover, differentstellar radial velocities provide us with the possibilityto recover the fringes even in the continuum around ab-sorption lines. Obviously, an accuracy of this statisticalprocedure depends on the spectra sample volume.The statistical approach was applied at each night tofit the fringes in the vicinity of oxygen lines. For theobservational set in March, 2011 we used seven workingstars, while 14 to 19 working stars for the 2012 observa-tions. The processing starts from raw CCD images withsubtracted bias. For each star we find positions of echelleorders and the light distribution along the slit. We thenextracted 11 spectra of one pixel height along the slitand processed them independently. The 6th spectrumcorresponding to the slit center is shown in Fig. 1a and b.All stellar and telluric lines were removed in order to useonly the continuum spectrum. We select the star withthe spectrum I ( λ ) of the highest signal-to-noise ratioand reduce spectra I ∗ ( λ ) of other working stars to theselected spectrum using the relation I ′ ∗ ( λ ) = S ( λ ) I ∗ ( λ ),where S ( λ ) is smooth spline approximation of the func-tion I ( λ ) /I ∗ ( λ ).The median averaging of the I ′ ∗ ( λ ) spectra of all work-ing stars gives us 11 spectra of the fringes along the slit,which are then used to normalize the stellar spectra.Eleven normalized spectra for each star were averagedwith weights depending on the light distribution alongthe slit. This procedure was applied to the second over-lapping order as well. The reduced spectra are shown inFig. 1c. Both spectra were then averaged with weightsdepending on their CCD signal level. A precise wave-length calibration is required to correctly perform thespectra averaging. We used a wavelength solution ofthe Ta-Th-Ar hollow cathode lamp (Pakhomov 2015)for the observations taken in 2011, March and the Ti-Arlamp (Pakhomov & Zhao 2013) for the observations of2012.In Fig. 1d and e we compare spectra of HD 49933 andHD 142091 observed with the Hamilton spectrograph ( R = 60 000) and reduced in this study with the correspond-ing ESPaDOnS spectra, which are free of fringes. It isworth noting, the latter R = 80 000 and S/N >
200 spec-tra were degraded to R = 60 000. In case of HD 49933the root mean square (rms) of the difference betweenthe Hamilton and ESPaDOnS spectra amounts to 0.0114that corresponds to S/N = 90, slightly lower than S/N= 110 of the original infrared spectrum of this star. Incase of HD 142091 the rms value is 0.0059 (S/N = 170 CC D c oun t s CC D c oun t s R e l a ti v e f l ux R e l a ti v e f l ux Wavelength, A o R e l a ti v e f l ux a)b)c)d)e) Figure 1 . Fringes reduction in the HD 49933 and HD 142091spectra.
Panel a : fragment of the HD 49933 spectrum fromthe slit center of the 97th echelle order (black curve) andthe recovered fringes for the same position on the slit (redcurve).
Panel b : the same as in panel a for overlapping partof 98-th echelle order.
Panel c : reduced and normalizedspectra of 97-th (black curve) and 98-th (blue curve) echelleorders.
Panel d : the reduced Hamilton (black curve) andthe ESPaDOnS (green curve) spectrum of HD 49933. Thedifference between the two spectra is shown in the upper partof the panel.
Panel e : the same as in panel d for HD 142091. versus the original S/N = 250). Thus, the statisticalapproach is efficient in removing the fringes.
Stellar atmosphere parameters.
A combination of thephotometric and spectroscopic methods was applied toderive a homogeneous set of the stellar atmosphere pa-rameters: effective temperature T eff , surface gravitylog g, [Fe/H], and microturbulence velocity ξ t . Ourspectroscopic analyses took advantage of employing theNLTE line formation for Fe I - Fe II . Paper I estimatedthe systematic and statistical errors of T eff to be 50 Kand 70 K, respectively, the uncertainty in log g / ξ t tobe 0.04 dex / 0.14 km s − , and statistical error of [Fe/H]was defined by the dispersion, σ , for lines of Fe II in a on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu σ = p Σ( x − x i ) / ( N − NLTE CALCULATIONS3.1.
NLTE methods
Our present investigation is based on the NLTE meth-ods treated in our earlier studies and documented ina number of papers, where atomic data and the prob-lems of line formation have been considered in de-tail. Table 1 lists the investigated chemical speciesand cites the related papers. Compared with the pub-lished model atoms, collisional data were updated forseveral chemical species. For Ca II and Sr II weapply here electron-impact excitation rate coefficientsfrom ab initio calculations of Mel´endez et al. (2007) andBautista et al. (2002), respectively. For Li I , Mg I , Al I ,and Si I inelastic collisions with neutral hydrogen par-ticles are treated using accurate rate coefficients fromquantum-mechanical calculations of Belyaev & Barklem(2003); Barklem et al. (2012); Belyaev (2013), andBelyaev et al. (2014), respectively. For the remainingspecies hydrogen collisions are computed using the for-mula of Steenbock & Holweger (1984) with a scaling fac-tor S H estimated empirically in the literature from theirdifferent influence on the different lines of a given atomin solar and stellar spectra. The references and recom-mended S H values are indicated in Table 1.In order to solve the coupled radiative trans-fer and statistical equilibrium equations for metals,we use a revised version of the DETAIL program(Butler & Giddings 1985) based on the acceleratedlambda iteration, which follows the efficient methoddescribed by Rybicki & Hummer (1991, 1992). Theupdate was presented by Mashonkina et al. (2011).The obtained departure coefficients were then usedby the codes SIU (Reetz 1991) and synthV-NLTE (Ryabchikova et al. 2016) to calculate the synthetic lineprofiles.As in Paper I, in this study we used the MARCS modelstructures (Gustafsson et al. 2008). Table 1 . Atomic models used in this study and treat-ment of A + H I inelastic collisions. Species Model atom A + H I Li I Shi et al. (2007) BB03C I Alexeeva & Mashonkina (2015) S H ∗ = 0.3O I Sitnova et al. (2013) S H = 1.0Na I Gehren et al. (2004) S H = 0.05Mg I Mashonkina (2013) BB12Al I Baumueller & Gehren (1996) B13Si I -Si II Shi et al. (2008) BYB14K I Zhang et al. (2006a) S H = 0.05Ca I Mashonkina et al. (2007a) S H = 0.1Sc II Zhang et al. (2008) S H = 0.1Ti II Sitnova et al. (2016) S H = 1.0Fe I -Fe II Mashonkina et al. (2011) S H = 0.5Cu I Shi et al. (2014) S H = 0.1Sr II Mashonkina & Gehren (2001) S H = 0.01Zr II Velichko et al. (2010) S H = 1.0Ba II Mashonkina et al. (1999) S H = 0.01Eu II Mashonkina & Gehren (2000) S H = 0.1 ∗ Scaling factor to the Drawinian rates
Note —BB03 = Belyaev & Barklem (2003); BB12 =Barklem et al. (2012); B13 = Belyaev (2013); BYB14 =Belyaev et al. (2014).
Line list and solar abundances for a differentialanalysis
The lines used in the abundance analysis were selectedfrom the lists of our NLTE papers (see Table 1 for refer-ences). They are listed in Table 3 along with the adoptedatomic parameters.The van der Waals damping was computed follow-ing the perturbation theory, where the data were avail-able, using the van der Waals damping constants Γ /N H at 10 000 K as provided by Barklem et al. (2000). Anexception was the selected lines of some elements, forwhich we used the C -values derived from solar line-profile fitting. If no other data were available, the Γ /N H values from Kurucz’s calculations were employed.Some elements considered here are represented by ei-ther a single isotope with an odd number of nucleons(Sc), or multiple isotopes with measured wavelength dif-ferences (∆ λ ≥ .
01 ˚A for Li I , Cu I , Sc II , Ba II , andEu II ). Nucleon-electron spin interactions in odd- A iso-topes lead to hyper-fine splitting (HFS) of the energylevels, resulting in absorption lines divided into multi-ple components. Without accounting properly for HFSand/or isotopic splitting (IS) structure, abundances de- http://kurucz.harvard.edu/atoms.html Zhao et al. termined from the lines sensitive to these effects can beseverely overestimated.Hyperfine structure (HFS) and/or isotope struc-ture (IS) is taken into account when necessarywith the data from Sansonetti et al. (1995, Li I ),Zhang et al. (2008, Sc II ), Shi et al. (2014, Cu I ),Borghs et al. (1983, Sr II ), Robert Kurucz’s website( http://kurucz.harvard.edu/atoms.html , Ba II ),and Lawler et al. (2001, Eu II ). For Li, Cu, Sr, Ba,and Eu, we use the fractional isotope abundances cor-responding to the solar system matter (Lodders et al.2009).The Sun is used as a reference star for a subse-quent stellar abundance analysis. The solar flux ob-servations were taken from the Kitt Peak Solar Atlas(Kurucz et al. 1984). The calculations were performedwith the MARCS model atmosphere 5777/4.44/0(Gustafsson et al. 2008). A depth-independent micro-turbulence of 0.9 km s − was adopted. Our syntheticflux profiles were convolved with a profile that combinesa rotational broadening of 1.8 km s − and broadening bymacroturbulence with a radial-tangential profile. The V mac values varied mainly between 2.6 and 3.3 km s − for the strong lines and between 3.4 and 4.0 km s − forthe weak lines. For comparison, Gray (1977) found so-lar macroturbulence velocities varying between 2.9 and3.8 km s − for a small sample of the solar Fe I lines. So-lar LTE and NLTE abundances from the individual linesare presented in Table 3.3.3. Departures from LTE for individual spectral lines
Our calculations show that the departures from LTEare different for lines of different chemical species anddifferent for lines of any given species. For each indi-vidual line the NLTE effects depend on stellar parame-ters. Figure 2 displays the NLTE abundance corrections,∆
NLTE = log ε NLTE − log ε LTE , for the representativelines of different species in our stellar sample, and Fig. 3illustrates the departures from LTE in the line profiles.All the investigated NLTE species can be separated infive groups depending on a dominant NLTE mechanism.1.
The photoionization dominated minority speciesMg I , Al I , Si I , Ca I , Ti I , Fe I , and Cu I . De-partures from LTE for these species are mainly causedby superthermal radiation of non-local origin below thethresholds of the ground state and/or low excitation lev-els, resulting in the overionization, i.e. depleted levelpopulations compared with their TE values. Most in-vestigated lines are weakened in NLTE compared withtheir LTE strengths, resulting in positive NLTE abun-dance corrections. They are overall small in the close-to-solar metallicity stars and grow toward lower metallic-ity due to decreasing ultraviolet (UV) opacity resultingin increasing photoionization rates. For given chemi- cal species different lines reveal similar NLTE effectsin a given model atmosphere. An exception is linesof Ca I , see next paragraph. Pronounced NLTE ef-fects for the resonance in contrast with the subordi-nate lines were found for Al I . For example, ∆ NLTE (Al I NLTE ≤ − .
78. Thiscan be easy understood. The excited levels of Al I though are subject to overionization, but are closely cou-pled to the ground state of the majority species Al II via the charge-transfer reactions Al I (nl) + H I (1s) ↔ Al II (3s ) + H − (Belyaev 2013), resulting in small∆ NLTE for Al I I is separated by 3.14 eV inenergy from the excited levels, and its population ismainly decided by overionization. It is worth not-ing, the NLTE correction for Al I NLTE =0.20 dex and 0.36 dex for the two least luminous and[Fe/H] ≃ − ◦ − .
06) andBD+29 ◦ − . NLTE rangesbetween 0.48 dex and 0.61 dex for the remaining [Fe/H] < − . I and Mg I can havenegative ∆ NLTE in the close-to-solar metallicity mod-els and positive correction in the low-metallicity models(Ca I I NLTE forsome lines of Al I , Si I , and Cu I .2. The collision dominated minority species Li I , Na I ,and K I . In the stellar parameter range, with which weconcern, these species are subject to the overrecombi-nation resulting in strengthened lines of Li I , Na I , andK I and negative NLTE abundance corrections. The ori-gin of the overpopulation of the ground and first excitedstate is the photon suction process described in detail byBruls et al. (1992). The departures from LTE are largerfor K I than Na I and for Na I than Li I because ofsmaller photoionization cross sections for K I than Na I and for Na I than Li I . A magnitude of ∆ NLTE is smallfor the Na I resonance lines in the [Fe/H] > − . on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Figure 2 . NLTE abundance corrections for the selected lines in the investigated stars. The Sun is shown by the symbol insidethe larger-size open circle.
Zhao et al. because the lines are strong and their total absorption ismostly contributed from the line wings formed in deepatmospheric layers, where the departures from LTE aresmall. It is worth noting, the LTE abundances fromNa I ≤ − . The majority species C I , O I , Si II , Ti II , and Fe II ,with negative NLTE abundance corrections for the inves-tigated lines. For each of these species its total numberdensity and population of the ground state keep theirTE values throughout the atmosphere. Populations ofthe excited levels are decided by a competition of theUV radiative pumping transitions, which produce an en-hanced excitation in the line-formation layers, and pho-ton losses in the lines, when the line optical depth dropsbelow unity, resulting in an underpopulation of the up-per levels of the corresponding transitions. For C I , O I ,Si II , Ti II , and Fe II the lower levels of the investigatedtransitions are overpopulated in line-formation layers,resulting in strengthened lines and negative NLTE abun-dance corrections. For different lines of Ti II and Fe II ∆ NLTE is overall small. In line with the previous NLTEstudies of C I (Alexeeva & Mashonkina 2015, and ref-erences therein) and O I (Sitnova et al. 2013, and refer-ences therein), pronounced NLTE effects were computedfor the infrared (IR) lines in the [Fe/H] > − . NLTE up to − . I and O I ∆ NLTE s areoverall small because the lines are weak and form indeep atmospheric layers.4.
The majority species Sc II , Zr II , and Eu II ,with positive NLTE abundance corrections for the in-vestigated lines. Here, for each line, NLTE leads to itsweakening relative to the LTE strength, owing to thelarger overpopulation of the upper than the lower levelrelative to the corresponding TE populations that re-sults in the increase in the line source function abovethe Planck function in the line-formation layers.5.
The majority species Sr II and Ba II , with asign of the NLTE correction depending on the lineand stellar parameters. As found theoretically byMashonkina et al. (1999), NLTE may lead either tostrengthening or to weakening the Ba II lines dependingon stellar parameters and element abundance. In ourstellar sample, ∆ NLTE is negative for Ba II > − > − . II , NLTE leads to a strengthening of the resonancelines and, in contrast, to a weakening of the subordi- nate line at 4161 ˚A. This can be understood as follows.In each model, the ground state keeps the TE popu-lation throughout the atmosphere and the upper level,5p, of the resonance transition is underpopulated in theuppermost atmospheric layers due to photon losses inthe resonance lines themselves resulting in an enhancedabsorption of the 4077 ˚A and 4215 ˚A lines. The Sr II DETERMINATION OF STELLARABUNDANCESTo minimize the effect of the uncertainty in gf -valueson the final results, we applied a line-by-line differentialNLTE and LTE approach, in the sense that stellar lineabundances were compared with individual abundancesof their solar counterparts. Throughout this study, theelement abundance is determined from line profile fit-ting. The synthetic line profiles were computed witheither the code SIU (Reetz 1991) or the codes synthV-NLTE (Ryabchikova et al. 2016) + binmag3 . Themetal line list has been extracted from the ViennaAtomic Line Database (VALD3 Ryabchikova et al.2015). Our test calculations of the C I and Zr II linesin a broad wavelength range from 4209 ˚A to 9111 ˚A inthe solar model atmosphere prove that using SIU and synthV-NLTE + binmag3 does not produce system-atic shifts in derived abundances, namely the abundancedifference nowhere exceeds 0.03 dex.In order to compare the theoretical profiles with ob-servations, they were convolved with a profile that com-bines instrumental broadening with a Gaussian profile,rotational broadening, and broadening by macroturbu-lence with a radial-tangential profile. Rotational broad-ening and broadening by macroturbulence were treatedseparately for the six stars with v sin i ≥ − ,namely HD 58855 ( v sin i = 10 km s − ), HD 89744(9 km s − ), HD 92855 (10 km s − ), HD 99984 (6 km s − ),HD 100563 (10 km s − ), and HD 106516 (7 km s − ). Wetreated the overall effects of rotation and macroturbu-lence for the remaining stars as radial-tangential macro-turbulence. The v sin i values and most probable macro-turbulence velocities V mac were determined in this studyfrom the analysis of an extended list of lines of variouschemical species. For a given star, V mac was allowedto vary by ± − (1 σ ). We selected a mildly MPstar HD 134169 (5890/4.02/ − .
78) to illustrate in Fig. 3a quality of the line fits in a broad spectral range from ∼ oleg/download.html http://vald.astro.univie.ac.at/ vald3/php/vald.php on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu σ ), and the number oflines used to determine the mean abundances. For mostspecies their abundances are based on analysis of thetwo to 20 lines. An exception is Li I , K I , and Eu II ,with a single line measured.For every species with more than one line mea-sured the differences in differential NLTE abundance be-tween different lines were found to be consistent within0.05 dex, on average, for the entire stellar sample. Fig-ure 4 displays the abundance differences for the selectedpairs of lines. We comment below on individual chemicalspecies.4.1. Notes on individual chemical speciesLithium.
The lithium abundances are derived fromthe Li I I asymmetric profile shape (see Fig.3 for HD 134169) is mainly caused by the two doubletstructure components, Li I I > − . Li in all the halo starsare also evaluated. Assuming a meteoric isotopic ratioof 12.3 ( Li/ Li) gives 0.034 dex smaller abundance forLi in average.
Carbon.
We used three carbon abundance indicators,namely the atomic C I and the molecular CH and C lines. Suitable lines of C I are located in the visible andnear-IR spectral range (Table 3). They all have closetogether an excitation energy of the lower level, E exc ,but different oscillator strengths, with smaller values forthe visible than the near-IR lines. The C I visible lineswere used in the close-to-solar down to [Fe/H] = − . ◦ ◦ I lines we use their list together with the atomic parameters fromAlexeeva & Mashonkina (2015). The C lines are ratherweak and cannot be measured in the [Fe/H] < − . − ◦ − ◦ ◦ ± I (NLTE) lines (Fig. 5). Applying the 3Dcorrections from Gallagher et al. (2016) to the CH G-band decreases a scatter of abundance differences onlya little, resulting in CH(3D) - C I = 0 . ± . I and C amountsto − . ± . I -basedabundances for final carbon abundances. The CH-basedabundances were employed for the stars with [Fe/H] < −
1, with no C I line measured. Oxygen.
Determination of the element abundancesfrom the O I IR lines takes advantage of using the spec-tra with removed fringes, as described in Sect. 2. Em-ploying only the visible O I − . I lines giveconsistent within the error bars abundances, as shownin Fig. 4 for O I Sodium abundances were determined using six Na I lines including the strong Na I D lines, because for somevery metal-poor stars, only the Na I D lines could beused for abundance determination. However, for starswith [Fe/H] > − .
5, the Na I D lines (5889 ˚A and5895 ˚A) were not used in calculating the final averageabundances. Magnesium abundances were determined using fiveneutral Mg lines as shown in Table 1. The strong Mg I b lines were not employed for abundance determination. Aluminum.
Determining Al abundances is very chal-lenging for our sample stars. We used seven lines of Al I .However, for most stars their Al abundance is based oneither the resonance line, Al I II I . For each star the LTE analysis ob-tained a 0.25-0.45 dex lower element abundance from theresonance line compared with that from the subordinatelines. Thanks to implementing quantum-mechanical0 Zhao et al.
LiI CI OI NaI
MgI
AlI
SiI KI CaI R e l a ti v e f l ux ScII
TiII
CuI
SrII
ZrII
Wavelength, A o BaII
EuII
Figure 3 . The best NLTE fits (continuous curves) of the observed spectrum of HD 134169 (bold dots). For comparison, theLTE profiles computed with the corresponding NLTE abundances are shown by dashed curves. on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Figure 4 . Differences in differential NLTE abundance between individual lines. The mean difference together with the standarddeviation is qouted in each panel. Zhao et al. -3 -2 -1 0[Fe/H]-0.2-0.100.10.2 C I - C H -3 -2 -1 0[Fe/H]-0.2-0.100.10.2 C I - C Figure 5 . Differences in abundance derived from lines ofC I in NLTE and molecular lines of CH (top panel) and C (bottom panel) in the investigated stellar sample. data on Al I +H I collisions by Belyaev (2013) in theSE calculations (Mashonkina et al. 2016), an abundancediscrepancy between different lines was largely removedin NLTE. Silicon.
We applied 11 Si I and two Si II lines toderive the Si abundances. Two strong ultraviolet Si I lines, 3905 and 4102 ˚A, were not used for the stars with[Fe/H] > − I and Si II lead to consis-tent NLTE abundances, with the mean difference (Si I - Si II ) = 0.00 ± I and Si II . Potassium.
Abundances of K were obtained by us-ing the K I lines for most sample stars.The potassium lines are also affected by very strongfringing effects. So, we could not determine potas-sium abundances for most very metal-poor stars with[Fe/H] < − . Calcium.
Among investigated elements in this paper,calcium covers many visible lines, and 21 Ca I lines wereemployed in our abundance determination. The Ca I resonance line at 4226 ˚A is very strong in all our programstars, and it was not used in the abundance analysis. Forstars with [Fe/H] > − I Scandium.
Because the Sc I lines in metal-poor starsare extremely week, the nine lines of Sc II are employedin the abundance determinations, although the number of lines are decreased to 1-3 for very metal-poor stars.Again, the NLTE corrections for all Sc II lines are small. Titanium.
For final Ti abundances we prefer to em-ploy lines of Ti II because of small NLTE effects. Indeed,∆ NLTE ≤ II lines in our calculations. Copper.
We applied three Cu I lines for determina-tions of copper abundances, namely 5105 ˚A, 5218 ˚A, and5782 ˚A, which are the same as that in our very recentstudy (Yan et al. 2016). Among them, 5105 ˚A is thestrongest and least blended (the only weak Fe II line atthe very blue wing) line, and thus is a good indicator ofthe copper abundance. The 5218 ˚A line is weak, and itsblue wing is blended by a Fe I line, which usually has acomparable equivalent width with the Cu I line. We thustook the two lines together into the consideration duringthe line profile fitting. The 5782 ˚A line is also blendedby several weak lines (i.e. Cr I , Cr II , Fe I , Fe II ). Tak-ing all the lines into account gives a consistent copperabundance with the other two lines. The Cu abundancedifference between including and ignoring the blendedlines near 5782 ˚A is ∼ .
02 dex, on average. The Cu I lines are weakened towards lower metallicity, and no cop-per abundance can be derived from these three lines forstars with [Fe/H] < − . Strontium.
Three lines of Sr II were employed inthe abundance determinations. The subordinate line at4161 ˚A was measured in the [Fe/H] ≥ − .
98 stars, andit gives the Sr abundance in line with that from Sr II ± II I > − II II − . ± .
08 for 17 commonstars.
Zirconium.
Only three lines of Zr II are suitable forstellar abundance determinations. The Zr II − ◦ − . II line was detected. To account for the blending Cr I4208.95 ˚A line ( E exc = 3.85 eV, log gf = − .
528 ac-cording to VALD) correctly, we controled the chromiumLTE abundance using a nearby line of Cr I 4209.365 ˚A,with E exc = 3.85 eV and log gf = − .
263 (VALD).In contrast, Zr II > − .
88 stars.Another line, Zr II I II < − .
19. We obtained consistentabundances from all the lines, with a mean abundance on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu − . ± .
04 dex between Zr II ± II Barium.
For the majority of stars their barium abun-dance was determined from the Ba II subordinate lines,which are almost free of HFS effects. According to ourestimate for Ba II II > − ◦ − ◦ II cannot beextracted from noise, and the barium abundance givenin Table 4 was determined from the resonance lines. It isworth noting, Ba II Ba :
Ba :
Ba :
Ba :
Ba = 2.4 : 6.6 : 7.9 : 11.2 : 71.7(Lodders et al. 2009) to the r-process one
Ba :
Ba:
Ba = 24 : 22 : 54 (Travaglio et al. 1999). For exam-ple, Ba II EW ) of 20 m˚A, and the abundance shift between usingthe solar and the r-process Ba isotope mixture amountsto 0.02 dex. Europium.
Three lines of Eu II were employed inthe abundance determinations. The subordinate lineat 6645 ˚A was measured in the [Fe/H] ≥ − .
78 stars,and it appears to give systematically higher abundancecompared with that from Eu II ± II at 4204.878-4205.117 ˚A isblended by numerous metal lines, which cannot be takeninto account correctly even using the synthetic spec-trum approach. As a result, the abundance differencebetween Eu II − . ± .
06 dex for 37 common stars. We avoided usingEu II − .
73 and − .
20, where Eu II ◦ II can be extracted from noise in ourmore MP stars.4.2. Uncertainties in derived abundances
We choose a mildly metal-deficient star HD 134169([Fe/H] = − .
78) to perform a detailed error analysisand to estimate the uncertainties in the abundance mea- surements for all the investigated species. Stochasticerrors ( σ obs ) caused by random uncertainties in the con-tinuum placement, line profile fitting, and gf -values, arerepresented by a dispersion in the measurements of mul-tiple lines around the mean, as given in Table 4 when N ≥ −
70 K in T eff , +0.07 dex in log g,and − . − in ξ t . The quantity ∆( T, g, ξ ) listed inCol. 7 is the total impact of varying each of the threeparameters, computed as the quadratic sum of Cols. 4,5, and 6.
Table 2 . Error budget for elements in HD 134169.
Atom σ obs λ ∆ T ∆ log g ∆ ξ ∆(˚A) −
80 K 0.07 − . − ( T, g, ξ )(1) (2) (3) (4) (5) (6) (7)Li I − . − .
01 0.01 0.08C I I I − .
02 0.02 0.00 0.03Mg I − . − .
04 0.01 0.08Al I − . − .
02 0.00 0.05Si I I I − .
02 0.01 0.03 0.04Sc II − . − .
07 0.02 0.13Ti II − .
01 0.04 0.04 0.06Cu I II − .
04 0.01 0.01 0.04Zr II − .
01 0.04 0.01 0.04Ba II II − .
02 0.04 0.00 0.04
Notes on individual starsPlanet-host stars.
In our sample, five stars havebeen reported to harbor one or more planets accord-ing the catalog listing of The Extrasolar Planets En-cyclopaedia . Three of them, HD 30562, HD 82943,and HD 89744, are metal rich stars with [Fe/H] ≃ http://exoplanet.eu/catalog/ Zhao et al.
Figure 6 . Differences in element abundance ratios, [X/Fe],between HD 74000 and HD 24289. All the results are fromthe NLTE calculations.
Stars with the thin-disc kinematics, but the thick-diskchemistry.
Our two most MP stars with a thin-disc kine-matics, HD 105755 ([Fe/H] = − .
73) and HD 134169([Fe/H] = − . α - and r-process enhancementstypical of the thick disk stars, with [Mg/Fe] = 0.29 and0.34 and [Eu/Ba] = 0.50 and 0.51. Halo star HD 74000 ([Fe/H] = − . ) reveals typicalabundances of the α -process elements, but overabun-dance of sodium and underabundance of europium com-pared with the stars of close metallicity (Fig. 8, 10, 13,11, and 14). We selected HD 24289 ([Fe/H] = − . ≃ − Halo star G090-003 ([Fe/H] = − . ). We draw anattention to high abundances of Na and Al in this star.The Na I resonance lines in its observed spectrum, bothare affected by the emissions of, probably, the telluricorigin. Since a quality of the spectrum of G090-003 isvery good, we could measure the Na abundance fromNa I THE GALACTIC CHEMICAL EVOLUTION5.1.
Stellar abundance trends
Stellar abundances for different elements, classifiedfrom their nucleosynthesis histories, for a large sampleof stars with different metallicities play a key role inthe study of the chemical evolution of these elementsthemselves, their origins and the chemical evolution ofthe Galaxy. This study is of particular importance be-cause it presents, for the first time, abundances of manyelements in a broad metallicity range that were homo-
Figure 7 . Stellar NLTE abundances of Li as a functionof metallicity (top panel) and effective temperature (bot-tom panel). Different symbols correspond to different stellarpopulations, namely the thin disc (open circles), the thickdisc (filled circles), and the halo (asterisks). The dashedline and shaded area show the predicted primordial lithiumabundance, log ε Li (CMB+BBN) = 2.64 ± geneously derived from the NLTE analysis. Among allthe investigated species lithium holds a specific position,because it is of primordial origin and considered a keydiagnostic to test and constrain our description of theearly Galaxy, of stellar interiors and evolution, and ofspallation physics. Elements beyond carbon are of stel-lar origin. Their abundances suffer from the so-calledeven-odd effect, which gives rise to different yields fordifferent elements despite of their same nucleosynthesispath. Therefore, in the C to Ti range we group theeven-nuclear charge ( Z ) elements and the odd- Z ele-ments. Elements beyond the iron group are believed tobe produced in the neutron-capture nuclear reactions.We discuss separately Sr to Eu and copper, because forthe latter its production mechanisms are still debated. Lithium.
We found that Li abundances of thewarm ( T eff ≥ ε Li = 2.2 (Fig. 7). Thisis in line with the earlier discovery of a remarkablyflat and constant Li abundance among Galactic halodwarf stars spanning a wide range of effective temper-atures and metallicities — the so-called Spite plateau(Spite & Spite 1982). Careful re-analysis of the lit-erature data led Charbonnel & Primas (2005) to de- on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu ε Li = 2.177 ± ≤ − . T eff ≥ ε Li = 2.64 ± ± I T eff ≤ ε Li variesbetween 1.8 and 2.7, and a temperature dependence isnot evident. An outlier is the cool giant HD 142091 ( T eff = 4810 K, log g = 3.12), where log ε Li = 0.17, in linewith the star’s evolutionary status. We confirm an exis-tence of the Li-desert at T eff ≃ ε Li ≃ . Even-Z elements.
Carbon abundance increases rel-ative to the Fe one, when metallicity decreases fromsuper-solar values down to [Fe/H] ≃ − . ± − .
11 and 0.03 were obtained forHD 103095 and BD +66 ◦ I lines.Carbon NLTE abundances were calculated byFabbian et al. (2006, hereafter, F06) for a sample of − . < [Fe / H] < − . I atoms. The final carbon NLTE abun-dances were obtained by F06 assuming negligible colli-sions with hydrogen. We did include collisions with H I and used S H = 0.3. The abundance difference betweenapplying S H = 0 and 0.3 is non-negligible. Calcula-tions of F06 with S H = 0 resulted in 0.1-0.15 dex lowerabundances compared with those for S H = 1 (Fig. 9in F06). Our calculations with S H = 0, 0.3, and 1show that the abundance difference between applying S H = 0 and 1 is larger than that between S H = 0.3and 1. For example, in HD 59374 (5850/4.38/ − . ε ( S H = 1 - S H = 0) = 0.08 dex, while it amounts0.03 dex, when comparing the S H = 1 and 0.3 basedabundances. The second source concerns, probably,with a different treatment of background opacity. Asshown by Alexeeva & Mashonkina (2015), their NLTEabundance corrections agree well with those of F06 inthe [Fe/H] ≥ − S H = 1, and they are less negative at lowermetallicities, by 0.08 dex in the 6000/4/ − −
3. Our test calculations showthat a variation in background opacity, for example ex-cluding H +2 , metal lines, quasi H molecular absorption,can lead to stronger departures from LTE and to 0.2 dexmore negative NLTE corrections for lines of C I in the5777/3.70/ − . ≃ − . ≃ . − . ≤ [Fe/H] < − I lines from the literature and performing detailed 3DNLTE radiative transfer calculations.From LTE analysis of the − . ≤ [Fe/H] ≤ − . −
1. Having applied the NLTE corrections to the LTEabundances of Akerman et al. (2004), F06 recovered asimilar behaviour of C/O, with the upturn at [Fe/H] ≃ − .
2, where [C/O] ≃ − . − . S H Zhao et al.
Figure 8 . Stellar element-to-iron NLTE abundance ratios: even-Z elements C, O, Mg, Si, Ca, and Ti. The same symbols areused as in Fig. 7. Asterisks inside the circles show the halo stars with only the molecular CH lines available. value and grows at lower metallicities. Our data arequalitatively similar, namely C/O is, on average, solarin the thin disk stars with [Fe/H] > − . − . − .
26 (Fig. 9). We confirm the upturn in[C/O] at [Fe/H] ≃ − .
2. The eleven more metal-poorstars form a linear regression of [C/O] = − . − . σ = 0.06. The observed C/O trend is im-portant for better understanding nucleosynthesis in theearly Galaxy.Multiple abundance determinations can be found inthe literature for Mg, Si, Ca, and Ti. However, homoge-neous NLTE abundances of all the four species and alsooxygen in the stellar sample covering a broad metal-licity range were obtained in this study for the firsttime. Magnesium, silicon, calcium, and titanium reveal a common behavior that is typical of the α -process el-ements. They are enhanced relative to Fe in the haloand the thick disk stars, with nearly constant [X/Fe] ra-tios at [Fe/H] < − . ± ± ± ± > − . α /Fe] ratios.We obtained that abundance ratios among Mg, Si, Ca,and Ti are close to solar value, independent of metallic-ity (see Si/Mg in Fig. 9), while each of these elementsis deficient relative to oxygen in the halo and the thick on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Figure 9 . Stellar NLTE abundance ratios between the even-Z elements. The same symbols are used as in Fig. 7. disk stars (Mg/O and Ca/O in Fig. 9). Most thin diskstars have, on average, solar [ α /O] ratios. Outliers arethe three stars, HD 59984, HD 105755, and HD 134169,with a thin-disc kinematics, but a low Fe abundance of[Fe/H] = − . − .
73, and − .
78, respectively. Their[ α /Fe] and [ α /O] ratios suggest a thick-disk origin. Astep-like increase of [ α /O] in the thick disk-to-thin disktransition is, in particular, clearly seen, when plottingthese elemental ratios as a function of [O/H].Obtained [Mg/Fe] ratios of our halo and thick diskstars are 0.1 dex lower compared with [Mg/Fe] ≃ < − α -process elements, they concluded that an α -enhancement for the thick disk and the stars with [Fe/H] < − . Odd-Z elements.
We consider Na and Al together, al-though conclusions related to Al are less firm due tolower accuracy of the derived Al abundances as dis-cussed in Sect. 4.1. Both Na and Al follow the Fe abun-dance in the thin and thick disc stars (Fig. 10). In mosthalo stars the Na/Fe and Al/Fe ratios are subsolar, witha rather large scatter of data. In contrast, a well-defineddownward trend is observed for Na/Mg, when [Fe/H] de-creases from super-solar values to −
1, and the more MPstars form a plateau at [Na/Mg] ≃ − . ≥ ≃
0. Thelatter (G090-003) has also high [Al/Fe] = 0.27, but nor-mal Na/Al. See notes on these two stars in Sect. 4.3.One more star, HD 108177, has higher [Na/Fe] = 0.04and [Na/Mg] = − .
15 compared with the halo stars ofsimilar metallicity.A metal-poor plateau for Na/Mg was reported inthe earlier NLTE studies by Gehren et al. (2006), with[Na/Mg] = − . − . ≤ [Fe/H] < − . ≃ − . − . ≤ [Fe/H] < − ≃ − .
5, and the literaturedata is, most probably, due to overestimated magne-sium NLTE abundances in Gehren et al. (2006) andAndrievsky et al. (2010). For example, the latter pa-per reported the mean [Mg/Fe] = 0.61 for their stellarsample, while, in this study, a MP plateau was obtainedat [Mg/Fe] = 0.28. The difference in Mg abundances is,in turn, probably due to different treatment of inelasticcollisions with H I atoms. Our study takes advantage ofemploying the Mg I + H I collision rates from quantum-mechanical calculations of Barklem et al. (2012), whileGehren et al. (2006) and Andrievsky et al. (2010) usedthe formula of Steenbock & Holweger (1984) with S H =0.05 and 1/3, respectively.The heavier odd- Z elements, K and Sc, behave like8 Zhao et al.
Figure 10 . The same as in Fig. 8 for the odd- Z elements Na, Al, K, and Sc. the α -elements in the thin and thick disk stars, at [Fe/H] > − ≃ . ≃ .
2. In the halo stars, potassium remainsto be enhanced relative to Fe, however, with the lowermagnitude, [K/Fe] < .
2, while Sc/Fe is close to the so-lar value. As a result, the trends are non-monotoneous,and a group of the thick disk stars at [Fe/H] around − . Copper.
The thin and thick disk stars with [Fe/H] ≥ − − .
41 and − .
50 in the halostars and a slightly higher value of − .
29 in HD 94028.Our data combined with the three − . < [Fe/H] < − − . ≃ − − . ≃ − . Neutron-capture elements: Sr, Zr, Ba, and Eu.
Here,we concentrate mostly on the − . ≤ [Fe/H] ≤ +0 . r -process poor star HD 140283 (see Siqueira-Mello et al.2015, and references therein) that is strongly underabun-dant in Sr and Ba relative to Fe and has about 0.4 dexlower Zr/Fe ratio compared with that for the remaininghalo stars. In our most MP star, BD − ◦ − . − . ≤ [Fe/H] ≤ +0 .
24 stars Sr and Ba follow the Fe abundance, al-though with a substantial scatter of ± < − ≃ .
5, and downward trend of Eu/Fe, witha rather small scatter of data, is observed at the highermetallicities. Such a behavior is typical of the r -processelements, and the knee at [Fe/H] ≃ − on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Figure 11 . Stellar NLTE abundance ratios involving the odd- Z elements. The same symbols are used as in Fig. 7. Figure 12 . The same as in Fig. 8 for copper. small scatter of data for stars of close metallicity. Bar-ium follows the Sr abundance suggesting their com-mon origin during the period when the Fe abundanceof the Galactic matter grew from [Fe/H] ≃ − . − M ⊙ during the asymptotic giant branch (AGB)phase, while 9 % of solar Sr originate from the weaks-process occurring in the helium burning core phaseof massive stars ( M > M ⊙ ). The remaining solarBa originates from the rapid (r) process. Astrophysical sites for the r-process are still debated, although they arelikely associated with explosions of massive stars, with M > M ⊙ . Analysis of Sr, Y, and Zr in the r-processenhanced stars and extremely MP ([Fe/H] < −
3) starsled Travaglio et al. (2004) to suggest the lighter ele-ment primary process (LEPP) that in the early Galaxycontributed to the light neutron-capture elements, butdid not to the heavy ones, beyond Ba. Travaglio et al.(2004) estimated empirically the LEPP contribution tosolar Sr as 8 %. Based on our data for stellar Sr/Ba, weinfer that, if it existed, the LEPP contribution to galac-tic Sr did not change during the − . < [Fe/H] ≤ +0 . ± r ≃ r ≃ = 0.87 in the large-scale parameterised dynamical network calculations ofFarouqi et al. (2010) in the context of an adiabaticallyexpanding high-entropy wind (HEW), as is expected to0 Zhao et al.
Figure 13 . The same as in Fig. 8 for the neutron-capture elements Sr, Zr, Ba, and Eu.
Figure 14 . Stellar NLTE abundance ratios between the neutron-capture elements. The same symbols are used as in Fig. 7.The dashed and dash-dotted lines indicate the SSr ratios, as predicted by GCE calculations of Bisterzo et al. (2014) andTravaglio et al. (1999), respectively. on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu r = 0.71(Travaglio et al. 1999) and 0.80 (Bisterzo et al. 2014).Our data on Eu/Ba (top right panel in Fig. 14) providean evidence for dominant contribution of the r-processto a production of Ba and Eu in the early Galaxy, whenthe halo and thick disk stellar population formed, andrapidly growing enrichment of the Galactic matter bys-nuclei, when metallicity increased from [Fe/H] ≃ − . r ≃ = 0.22. This is smaller than [Zr/Ba] ob-served in the halo and thick disk stars. As expected,the Zr/Sr ratio is close to the solar value in the thindisk stars, but it grows steeply in the thick disk andhalo stars, approaching [Zr/Sr] ≃ . − . r+LEPP = 0.22 and [Zr/Sr] LEPP = 0.35 for pro-duction of Sr and Zr in the r- and LEPP-process to-gether and in a pure LEPP-process. An origin of Zrin the thick disk stars can be attributed to these twoprocesses. However, further efforts should be investedto understand high Zr/Sr ratios observed in the [Fe/H] < − II , Zr II , and Eu II are small in the stellar parameterrange, with which we concern (see Sect. 3.3). The up-ward trend in [Zr/Fe] was reported by Mashonkina et al.(2007b) and Mishenina et al. (2013), although the lat-ter paper studied a narrow range of metallicity, down to[Fe/H] ≃ −
1. Enhancement of Eu relative to Fe andBa in the halo and/or thick disk stars was obtainedearlier by Mashonkina & Gehren (2000); Burris et al.(2000); Barklem et al. (2005); Bensby et al. (2005), andMishenina et al. (2013). With different stellar sample,Mashonkina et al. (2007b) found the Galactic trends forZr/Ba and Zr/Sr that are very similar to ours.5.2.
Influence of NLTE on the Galactic abundancetrends
As noted, this study employs a line-by-line differentialanalysis with respect to the Sun. Here, we discuss animpact of NLTE on determination of the mean elementabundances [X/H] and elemental ratios, depending onthe star’s metallicity. Figure 15 displays the differencesbetween the NLTE and LTE [X/H] ratio for the 14 in-vestigated species. We do not show the data for Li I (seeFig. 2 for the NLTE effects), Ti II due to minor differ-ences, and Fe I -Fe II , which were discussed in Paper I.A differential approach largely cancels the (NLTE -LTE) differences in [X/H] for most species in the [Fe/H] > − NLTE for lines of C I can be up to − .
4, while the[C/H] differences between the NLTE and LTE do notexceed 0.1 dex in absolute value. However, notable( > . > − − .
47) LTE leads to[O/H] = − . ± .
11, [Ca/H] = − . ± .
08, [Ba/H]= − . ± .
13, while remarkably smaller statistical er-rors are obtained in NLTE, with [O/H] = − . ± . − . ± .
04, and [Ba/H] = − . ± . −
1. The (NLTE -LTE) differences in [X/H] grow in absolute value to-wards lower metallicity and, for most species, can reach0.2 dex and even more. Exceptions are [Mg/H], [Si/H],[K/H], [Zr/H], and [Eu/H], where the departures fromLTE are small. NLTE is, in particular, important forelemental ratios involving the species with (NLTE -LTE) of different sign, like [Na/Mg], [Na/Al], [Na/Cu],[Sr/Ba]. For example, the mean for the halo stars, ex-cluding HD 74000 and G090-003, amounts to [Na/Mg]= − . ± .
10 in NLTE and − . ± .
19 in LTE.NLTE makes Al following Na over the whole metallic-ity range under investigation, with the mean [Na/Al] = − . ± ± Comparison with the Galactic chemical evolutionmodels
In this section, we compare our observational datawith a series of GCE models from literature. Wewill mainly discuss the models of K11 (Kobayashi et al.2
Zhao et al.
Figure 15 . Differences in differential abundance [X/H] between NLTE and LTE for the investigated sample. on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Figure 16 . Comparison with the Galactic chemical evolution models. The models used in the figure are C97 (violet solid line),S98 (turquoise dashed line), GP00m2 (gray dashed line), F04m2 (green solid line), R10m15 (blue solid line), K11 (orange solidline), and K15 (red dashed line), where GP00m2, F04m2 and R10m15 represents the model of ‘thick curve’ in GP00, the modelof Fig4-6 in F04, and the model 15 in R10, respectively. Zhao et al.
Figure 17 . Comparison with the Galactic chemical evolution models. The colors are as same as that in Fig. 16. on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu . − M ⊙ , andthe Type Ia Supernova (SNe Ia) model based onthe single degenerate scenario (Kobayashi & Nomoto2009) with the metallicity effect (Kobayashi et al. 1998).The metallicity-dependent nucleosynthesis yields aretaken from Kobayashi et al. (2006, hereafter K06) andKobayashi et al. (2011b) for supernovae and hypernovae(with 0.5 fraction of hypernovae at ≥ M ⊙ ), andfrom Karakas (2010) for asymptotic giant branch (AGB)stars, respectively. The yield sets are identical tothose in Nomoto et al. (2013). Beside, in the K15model, the effect of 2D jet-like explosions is applied (seeSneden et al. 2016, for the details). These GCE resultsare consistent with the metallicity distribution function,the present star formation rate, and the present gas frac-tion.Besides K11 and K15, we also compared our re-sults with other widely used GCE models which arefrom Chiappini et al. (1997, hereafter C97), Samland(1998, hereafter S98), Goswami & Prantzos (2000, here-after GP00), Fran¸cois et al. (2004, hereafter F04) andRomano et al. (2010, hereafter R10), respectively. C97presented the model which assumes two main infallepisodes that formed the halo/thick-disk and the thin-disk, respectively. S98 developed a chemodynamicalmodel of an isolated disk galaxy to be consistent with theobservations deriving empirical yields. The model takesinto account the galactic dynamical process for variouskinds of stars and ISM. GP00 described an indepen-dently evolved halo+disk model, with short timescaleoutflows for halo and slow infall of the disk. F04 alsopresented a two-infall model that is similar to C97, butwith empirical stellar yields. R10 tested 15 GCE mod-els with various sets of stellar yields from literatures.Fig. 16 and Fig. 17 show the comparison between ourobservational data and the predictions from those GCEmodels. The offsets of solar abundances from differentworks have been corrected. Carbon – The [C/Fe] ratio predicted by K11 and K15shows a waved line, which slightly decreases from [Fe/H] ∼ − − .
7, due to the smaller envelope mass thatcontains C of massive progenitor stars. The rapid in-crease from [Fe/H] ∼ − . − ∼ − M ⊙ . From[Fe/H] ∼ −
1, [C/Fe] decreases due to the delayed enrich-ment of SNe Ia. Although the lifetimes of these AGBstars (0 . − . > ∼ − .
5. At [Fe/H] < ∼ − .
5, the models predict ∼ − Oxygen – The observed [O/Fe] trend is in good agree-ment with the K11 and K15 models, where the plateauat [O/Fe] ∼ . ∼ − α elements,i.e., O, Mg, Si, S, and Ca. It is very important that thisevolutionary change appears sharply, and there is nostars with [O/Fe] < ∼ . −
1. As in Figure 6of K11, without the metallicity effect of SNe Ia, the evo-lutionary change occurs much more gradually, being in-consistent with our new observational data. Other mod-els did not show the sharply changing at [Fe/H]= − Sodium – In the K11 and K15 models, Na productionhighly depends on the metallicity of progenitors of core-collapse supernovae, which causes the increasing trendfrom [Fe/H] ∼ − ∼ − .
5. This agrees with the ob-servational data very well. From [Fe/H] ∼ − .
5, [Na/Fe]ratios increase quickly due to AGB stars. Note that withthe updated reaction rates, the Na yields of AGB starshave been reduced. However, Na is still over-producedby AGB stars. S98 may predict a better trend, but ituses empirical yields that are determined from the ob-servations.
Magnesium – The observed [Mg/Fe] ratios show thesame trend as [O/Fe], but there is a ∼ .
25 dex offsetbetween the observations and K11/K15 models. Thismeans that [O/Mg] is not zero at a wide range of metal-licity. This could be partially solved with the mass de-pendence of core-collapse supernovae, where [O/Mg] isslightly higher for more massive supernovae ( > ∼ M ⊙ ,see Fig 1-4 of K06). This could also be solved by uncer-tain reactions rates in the hydrostatic burning of pro-genitor stars, as shown in Figure 9 of K06. Note thatO and Mg are synthesized roughly the same region ofsupernova ejecta, and hence [O/Mg] should not dependon the parameters of supernova explosions very much.The Mg production of AGB stars is negligible in GCEmodels (except for the isotopic ratios, K11). Aluminum – Similar to Na, the trend predicted by K11and K15 is consistent with the observational data, butsimilar to Mg, there is a ∼ .
25 dex offset. These couldbe due to the reaction rates, the rotational/convectivemixing, or the combination of both. The trend predictedby the model 15 of R10 is in good agreement with theobservation.
Silicon – Similar to Mg, the observed trend is well re-6
Zhao et al. produced with the K11 and K15 models, but the modelis ∼ . − . − M ⊙ stars exist, the [Si/Fe] ratios become muchhigher, being inconsistent with the observations (e.g.,Cayrel et al. 2004). Potassium – The underproduction of K in K11 andK15 is, at least partially, due to the lack of the neutrinoprocess (Kobayashi et al. 2011a). This element has notbeen well studied before because of the uncertainty ofthe NLTE effect, but our observational data can givestrong constraints on supernova nucleosynthesis. Simi-lar underproduction of K is seen in the model 15 of R10.
Calcium – Similar to [O/Fe], the K11 and K15 modelsexcellently reproduces the observed [Ca/Fe] ratios, andthe sharp evolutionary change at [Fe/H] = − Scandium – In K15 the Sc abundance can be increasedby the 2D jet-like explosions (Maeda & Nomoto 2003),as shown in Fig. 17 by dashed line. This could alsobe enhanced by the neutrino process as for K. It worthnoting, all the models except S98 failed to reproduce theobserved [Sc/Fe] trend in the whole metallicity range.
Titanium – This is the long-standing Ti problem inGCE, where the predicted [Ti/Fe] ratios are much lowerthan observed. The 2D jet effect should increase Tiabundances (dashed lines in Fig. 17), but may notenough to solve this problem. Ti is produced in almostthe same region as Fe in supernova ejecta, and hence itis categorized not α element but an iron-peak element.Nucleosynthsis with multidimensional explosions is nec-essary to understand the Ti production. Copper – The trends predicted by K11 and K15 mod-els are similar with those of observed, although themodel is a bit higher at [Fe/H] > − .
0. This is causedby the star formation rates connected with IMF. [Cu/Fe]increases toward higher metallicity because Cu is anodd-Z element, the production of which depends on theprogenitor metallicity. This agreement suggests thatthe main producer of Cu is core-collapse supernovae,not the weak slow neutron-capture process suggested byPignatari et al. (2010).Without normalized with respect to Fe, it may be pos-sible to constrain uncertain processes that are importantfor some specific elements. [Na/Mg] ratio in 17 impliesthat the metallicity dependence may be smaller thanthose in K11 and K15. The [C/O] ratio may suggestthat the mixing and/or rotation may be more importantthan those in K11 and K15. Note that in the model 15 ofR10, the C yields in stellar winds are added, but a part ofwhich have already been included in supernova yields, so the high [C/O] ratio should be due to the double-count of C production. In K15, the [Ca/O] ratios areconsistent with the observed ones at [Fe/H] < ∼ − .
5, butare lower at higher metallicity, which may be due to thecontribution to observed Ca from SNe Ia. [K/Sc] and[Ti/Sc] are in particular interesting since [(K,Sc,Ti)/Fe]is underabundant in K11. The low [K/Sc] may suggestthe importance of ν process, and the [Ti/Sc] support the2D effect applied in K15 to some extent. These figuresshould be used to test the next generation of nucleosyn-thesis yields with multi-dimensional calculations. CONCLUSIONSUsing accurate atmospheric parameters determined inPaper I and high-resolution (R ≃
60 000) stellar spectraobserved for our project with the Shane/Hamilton spec-trograph (the Lick observatory) and also taken from thearchives, we calculated the NLTE abundances for 17 el-ements in a sample of stars uniformly distributed overthe − . ≤ [Fe/H] ≤ +0 .
24 metallicity range. Thestar sample has been kinematically selected to trace theGalactic thin and thick disks and halo. This is the firstextensive NLTE study of the stellar sample suitable forthe Galactic chemical evolution research.We derive differential abundances relative to the Sun,and such an approach largely cancels the difference be-tween NLTE and LTE for [C/H], [Na/H], [Ca/H], and[Ba/H] for the [Fe/H] > − . > . −
1, in par-ticular, for elemental ratios involving the species with(NLTE - LTE) of different sign, like [Na/Mg], [Na/Al],[Na/Cu], [Sr/Ba].In line with the earlier studies, we obtained that thehalo dwarf stars, which are expected to keep the pris-tine Li abundance, reveal a clear temperature depen-dence of their Li abundance. In our warm ( T eff ≥ ε Li = 2.2, which isconsistent with log ε Li = 2 . ± .
071 deduced byCharbonnel & Primas (2005) for the [Fe/H] ≤ − . T eff ≥ ε Li = 2.72 ± T eff < on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu ≃− .
8, and a downward trend for higher metallicity. Inthe halo and thick disk stars [O/Fe] = 0.61 and a 0.3 dexlower [X/Fe] ratio was obtained for Mg, Si, Ca, and Ti.An upward trend of [C/Fe] with decreasing metallicity isobserved in the thin disk stars and a very similar valueof [C/Fe] = 0.21 in the thick disk stars. In contrastto [C/Fe] that reveals a substantial scatter in the halostars, a well defined trend was obtained for the C/Oratios, with the upturn at [Fe/H] ≃ − .
2, in line with theearlier finding of Akerman et al. (2004). We obtainedno systematic shift between the NLTE abundances fromlines of C I and the CH-based abundances over the entiremetallicity range under investigation.A rather large scatter of data is observed when com-paring abundances of the odd- Z elements Na, K, andSc with iron, however, it is largely removed in the ratiosbetween these elements, like K/Sc, and in the ratios be-tween nearby odd- Z and even- Z elements, like Na/Mgand Sc/Ti, suggesting a common production site for Nato Ti. We find a nearly constant underabundance of Narelative to Mg in the [Fe/H] < − ≃ − . ≃ − .
5, but Zr is enhanced relative toFe and Sr in the MP stars. The [Zr/Sr] ratio is closeto the solar value in the thin disk stars, grows to 0.4 at[Fe/H] = -2, and approaches to 0.8 at [Fe/H] = − .
5. Inline with the earlier studies, the upward trend in [Eu/Fe]exists for [Fe/H] > −
1, and europium is enhanced rel-ative to Fe by more than 0.3 dex in the halo stars. Aplateau of [Eu/Ba] at 0.50 is formed by the halo andthick disk stars, the knee occurs at [Fe/H] ≃ − .
8, andthe downward trend in [Eu/Ba] is observed for highermetallicities.The use of the NLTE element abundances raises creditto the interpretation of the data in the context of thechemical evolution of the Galaxy. Although GCE mod-els are not calibrated with our NLTE abundances in thispaper, K15 model predictions are in good agreement forC, O, Ca, and Fe in some metallicity coverages and theoverall shapes. The underproduction of K, Sc, and Ti issomewhat known, and is due to the lack of ν processes(Sneden et al. 2016). The offsets in odd-Z elements (i.e.,Na, Al, Cu) give important constraints on the uncertainprocesses such as mixing. Despite the agreement for O,the offsets for Mg may be the most problematic sinceboth elements have formed in relatively robust stellarevolution phase. If [Mg/Fe] is as low as in our NLTEanalysis, that requires a different C/O ratio due to themixing, mass-loss, and/or reaction rates in the progeni-tor stars, which should be studied in future works. The authors thank Klaus Fuhrmann and ThomasGehren for providing the FOCES spectra at our dis-posal, Oleg Kochukhov and Vadim Tsymbal for pro-viding the codes binmag3 and synthV-NLTE , Do-natella Romano for providing the data of GCE mod-els. This study was supported by the Russian Foun-dation for Basic Research (grant 14-02-91153 and 16-32-00695), the National Natural Science Foundation ofChina (grants 11390371, 11233004, 11222326, 11103034,11473033, 11473001), the National Basic Research Pro-gram of China (grant 2014CB845701/02/03), and theSwiss National Science Foundation (SCOPES projectNo. IZ73Z0-152485). We made use the Simbad,MARCS, and VALD databases.8 Zhao et al.
Table 3 . Line data, references to their sources and the obtained solar element LTE and NLTE abundances, log ε ⊙ . λ , ˚A E exc log gf Ref. log C Ref. log ε ⊙ λ , ˚A E exc log gf Ref. log C Ref. log ε ⊙ (eV) LTE NLTE (eV) LTE NLTELi I Ca I I I II I II I I I I II II Table 3 continued on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Table 3 (continued) λ , ˚A E exc log gf Ref. log C Ref. log ε ⊙ λ , ˚A E exc log gf Ref. log C Ref. log ε ⊙ (eV) LTE NLTE (eV) LTE NLTE6243.81 5.62 -1.29 SGM -29.67 ABO 7.53 7.53 4208.98 0.71 -0.51 LNA -32.00 2.57 2.586244.47 5.62 -1.29 SGM -29.67 ABO 7.54 7.54 5112.27 1.67 -0.85 LNA -32.00 2.70 2.71Si II Ba II I I Eu II Note —ABO = Barklem et al. (2000), BG96 = Baumueller & Gehren (1996), B00 = Keith Butler (private communication), B75 = Bielski (1975),BM98 = Barklem & O’Mara (1998), GLS = Gehren et al. (2004), K07, K12 = R. Kurucz’s website http://kurucz.harvard.edu/atoms.html , LD89= Lawler & Dakin (1989), LNA = Ljung et al. (2006), LWD = Lawler et al. (2001), MZG = Mashonkina et al. (2008), NIST = Ralchenko et al.(2010), RCW = Reader et al. (1980), S81 = Smith (1981), S88 = Smith (1988), SGM = Shi et al. (2009), SN75 = Smith & O’Neill (1975), SR81 =Smith & Raggett (1981), Sun = the solar line profile fits, WLS = Wood et al. (2013), ZGZ = Zhang et al. (2008); IS, Sansonetti et al. (1995); HFS,ZGZ; HFS, Shi et al. (2014); HFS, Borghs et al. (1983); HFS, K07; HFS, LWD. Zhao et al.
Table 4 . Summary of the obtained stellar abundances.
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]HD 19373 6045/4.24/ 0.10, ξ t = 1.2 Thin disk HD 22484 6000/4.07/ 0.01, ξ t = 1.1 Thin disk3 Li I 1 2.50 2.47 3 Li I 1 2.37 2.336 C I 5 -0.00 ± ± ± ± ± · · · · · · - CH 13 0.07 ± · · · · · · ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.0 Thick disk HD 24289 5980/3.71/-1.94, ξ t = 1.1 Halo3 Li I 1 1.71 1.69 3 Li I 1 2.34 2.336 C I 1 -0.65 0.19 -0.65 0.19 6 C I 0- CH 0 - CH 12 -1.55 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.3 Thin disk HD 30743 6450/4.20/-0.44, ξ t = 1.8 Thin disk3 Li I 1 2.74 2.68 3 Li I 1 2.60 2.556 C I 4 0.06 ± ± ± ± Table 4 continued on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]- CH 10 0.06 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.2 Thin disk HD 43318 6250/3.92/-0.19, ξ t = 1.7 Thin disk3 Li I 1 2.06 2.04 3 Li I 16 C I 5 -0.05 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.5 Thin disk HD 45205 5790/4.08/-0.87, ξ t = 1.1 Thick disk3 Li I 1 2.37 2.33 3 Li I 1 2.03 2.016 C I 4 -0.12 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 4 continued Zhao et al.
Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]14 Si I 9 -0.11 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.7 Thin disk HD 52711 5900/4.33/-0.21, ξ t = 1.2 Thin disk3 Li I 1 3 Li I 1 1.84 1.826 C I 3 -0.43 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.6 Thin disk HD 59374 5850/4.38/-0.88, ξ t = 1.2 Thick disk3 Li I 1 2.42 2.39 3 Li I 1 1.80 1.806 C I 5 -0.20 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 4 continued on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]22 Ti II 10 -0.22 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.4 Thin disk HD 62301 5840/4.09/-0.70, ξ t = 1.3 Thick disk3 Li I 1 2.47 2.44 3 Li I 1 2.04 2.026 C I 3 -0.60 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 0.7 Halo HD 69897 6240/4.24/-0.25, ξ t = 1.4 Thin disk3 Li I 1 1.26 1.27 3 Li I 1 2.74 2.696 C I 0 6 C I 5 -0.23 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 4 continued Zhao et al.
Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]40 Zr II 2 -1.32 ± ± ± ± ± ± ± ± ξ t = 1.3 Halo HD 76932 5870/4.10/-0.98, ξ t = 1.3 Thick disk3 Li I 1 2.41 2.39 3 Li I 1 2.17 2.156 C I 0 6 C I 3 -0.65 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.2 Thin disk HD 84937 6350/4.09/-2.16, ξ t = 1.7 Halo3 Li I 1 2.44 2.41 3 Li I 1 2.19 2.176 C I 4 0.13 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.7 Thin disk HD 90839 6195/4.38/-0.18, ξ t = 1.4 Thin disk Table 4 continued on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]3 Li I 1 1.93 1.90 3 Li I 1 2.68 2.616 C I 3 0.05 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.3 Thin disk HD 94028 5970/4.33/-1.47, ξ t = 1.3 Thick disk3 Li I 1 2.37 2.34 3 Li I 1 2.27 2.256 C I 2 -0.20 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.8 Thin disk HD 100563 6460/4.32/ 0.06, ξ t = 1.6 Thin disk3 Li I 1 2.55 2.51 3 Li I 1 2.73 2.666 C I 2 -0.25 ± ± ± ± ± ± ± ± ± ± ± Table 4 continued Zhao et al.
Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]12 Mg I 5 -0.29 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.5 Thin disk HD 103095 5130/4.66/-1.26, ξ t = 0.9 Halo3 Li I 1 1.94 1.91 3 Li I 16 C I 4 0.12 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.2 Thin disk HD 106516 6300/4.44/-0.73, ξ t = 1.5 Thick disk3 Li I 1 1.97 1.96 3 Li I 16 C I 3 -0.48 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 4 continued on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]20 Ca I 16 -0.52 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.1 Halo HD 110897 5920/4.41/-0.57, ξ t = 1.2 Thin disk3 Li I 1 2.28 2.26 3 Li I 1 1.98 1.966 C I 0 6 C I 2 -0.54 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.1 Thin disk HD 115617 5490/4.40/-0.10, ξ t = 1.1 Thin disk3 Li I 1 2.61 2.58 3 Li I 16 C I 2 -0.07 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 4 continued Zhao et al.
Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]29 Cu I 3 -0.00 ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.1 Thick disk HD 134169 5890/4.02/-0.78, ξ t = 1.2 Thin disk3 Li I 1 1.36 1.36 3 Li I 1 2.36 2.356 C I 1 -0.63 0.17 -0.63 0.17 6 C I 4 -0.59 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.3 Thin disk HD 140283 5780/3.70/-2.46, ξ t = 1.6 Halo3 Li I 1 3 Li I 16 C I 5 0.23 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 4 continued on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]HD 142091 4810/3.12/-0.07, ξ t = 1.2 Thin disk HD 142373 5830/3.96/-0.54, ξ t = 1.4 Thin disk3 Li I 1 0.14 0.17 3 Li I 1 2.44 2.416 C I 1 0.05 0.12 0.05 0.12 6 C I 2 -0.50 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ◦ ξ t = 1.3 Halo BD +09 ◦ ξ t = 1.3 Halo3 Li I 1 2.30 2.27 3 Li I 1 2.35 2.336 C I 8 -1.41 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ◦ ξ t = 1.5 Halo BD +29 ◦ ξ t = 0.8 Halo3 Li I 1 2.15 2.14 3 Li I 1 2.26 2.246 C I 0 6 C I 0- CH 0 - CH 10 -1.66 ± Table 4 continued Zhao et al.
Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]8 O I 2 -1.81 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ◦ ξ t = 1.0 Halo BD +66 ◦ ξ t = 0.6 Halo3 Li I 1 1.51 1.53 3 Li I 16 C I 2 -1.83 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± − ◦ ξ t = 1.4 Halo BD − ◦ ξ t = 1.4 Halo3 Li I 1 3 Li I 1 2.18 2.166 C I 0 6 C I 3 -2.61 ± ± ± ± ± ± ± ± ± ± Table 4 continued on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu Table 4 (continued)
Z Species N LTE NLTE Z Species N LTE NLTE[X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe] [X/H] [X/Fe]14 Si II 0 14 Si II 019 K I 0 19 K I 020 Ca I 0 20 Ca I 2 -2.24 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ξ t = 1.3 Halo3 Li I 1 2.47 2.466 C I 0- CH 11 -1.74 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± REFERENCES
Adibekyan, V. Z., Sousa, S. G., Santos, N. C., et al. 2012, A&A,545, A32Akerman, C. J., Carigi, L., Nissen, P. E., Pettini, M., &Asplund, M. 2004, A&A, 414, 931Alexeeva, S. A., & Mashonkina, L. I. 2015, MNRAS, 453, 1619Allende Prieto, C., Barklem, P. S., Lambert, D. L., & Cunha, K.2004, A&A, 420, 183Amarsi, A. M., Asplund, M., Collet, R., & Leenaarts, J. 2015,MNRAS, 454, L11Andrievsky, S. M., Spite, M., Korotin, S. A., et al. 2010, A&A,509, A88Bagnulo, S., Jehin, E., Ledoux, C., et al. 2003, The Messenger,114, 10 Barklem, P. S., Belyaev, A. K., Spielfiedel, A., Guitou, M., &Feautrier, N. 2012, A&A, 541, A80Barklem, P. S., & O’Mara, B. J. 1998, MNRAS, 300, 863Barklem, P. S., Piskunov, N., & O’Mara, B. J. 2000, Astron. andAstrophys. Suppl. Ser., 142, 467Barklem, P. S., Christlieb, N., Beers, T. C., et al. 2005, A&A,439, 129Baumueller, D., & Gehren, T. 1996, A&A, 307, 961Bautista, M. A., Gull, T. R., Ishibashi, K., Hartman, H., &Davidson, K. 2002, MNRAS, 331, 875Belyaev, A. K. 2013, A&A, 560, A60Belyaev, A. K., & Barklem, P. S. 2003, PhRvA, 68, 062703 Zhao et al.
Belyaev, A. K., Yakovleva, S. A., & Barklem, P. S. 2014, A&A,572, A103Bensby, T., Feltzing, S., Lundstr¨om, I., & Ilyin, I. 2005, A&A,433, 185Bensby, T., Feltzing, S., & Oey, M. S. 2014, A&A, 562, A71Bergemann, M., & Cescutti, G. 2010, A&A, 522, A9Bergemann, M., & Gehren, T. 2008, A&A, 492, 823Bergemann, M., Pickering, J. C., & Gehren, T. 2010, MNRAS,401, 1334Bielski, A. 1975, JQSRT, 15, 463Bisterzo, S., Travaglio, C., Gallino, R., Wiescher, M., &K¨appeler, F. 2014, ApJ, 787, 10Borghs, G., de Bisschop, P., van Hove, M., & Silverans, R. E.1983, Hyperfine Interactions, 15, 177Bruls, J. H. M. J., Rutten, R. J., & Shchukina, N. G. 1992,A&A, 265, 237Burris, D. L., Pilachowski, C. A., Armandroff, T. E., et al. 2000,ApJ, 544, 302Butler, K., & Giddings, J. 1985, Newsletter on the analysis ofastronomical spectra, No. 9, University of LondonCarbon, D. F., Barbuy, B., Kraft, R. P., Friel, E. D., & Suntzeff,N. B. 1987, PASP, 99, 335Cayrel, R., Depagne, E., Spite, M., et al. 2004, A&A, 416, 1117Charbonnel, C., & Primas, F. 2005, A&A, 442, 961Chiappini, C., Matteucci, F., & Gratton, R. 1997, ApJ, 477, 765,(C97)Coc, A., Goriely, S., Xu, Y., Saimpert, M., & Vangioni, E. 2012,ApJ, 744, 158Edvardsson, B., Andersen, J., Gustafsson, B., et al. 1993, A&A,275, 101Fabbian, D., Asplund, M., Carlsson, M., & Kiselman, D. 2006,A&A, 458, 899, (F06)Farouqi, K., Kratz, K.-L., Pfeiffer, B., et al. 2010, ApJ, 712, 1359Fran¸cois, P., Matteucci, F., Cayrel, R., et al. 2004, A&A, 421,613, (F04)Fu, X., Bressan, A., Molaro, P., & Marigo, P. 2015, MNRAS,452, 3256Fuhrmann, K. 2008, MNRAS, 384, 173—. 2011, MNRAS, 414, 2893Gallagher, A. J., Caffau, E., Bonifacio, P., et al. 2016, ArXive-prints, arXiv:1605.07215Gehren, T., Liang, Y. C., Shi, J. R., Zhang, H. W., & Zhao, G.2004, A&A, 413, 1045Gehren, T., Shi, J. R., Zhang, H. W., Zhao, G., & Korn, A. J.2006, A&A, 451, 1065Goswami, A., & Prantzos, N. 2000, A&A, 359, 191, (GP00)Gray, D. F. 1977, ApJ, 218, 530Gustafsson, B., Edvardsson, B., Eriksson, K., et al. 2008, A&A,486, 951Ishigaki, M. N., Aoki, W., & Chiba, M. 2013, ApJ, 771, 67Karakas, A. I. 2010, MNRAS, 403, 1413Kobayashi, C., Izutani, N., Karakas, A. I., et al. 2011a, ApJL,739, L57Kobayashi, C., Karakas, A. I., & Umeda, H. 2011b, MNRAS,414, 3231, (K11)Kobayashi, C., & Nomoto, K. 2009, ApJ, 707, 1466Kobayashi, C., Tsujimoto, T., Nomoto, K., Hachisu, I., & Kato,M. 1998, ApJL, 503, L155Kobayashi, C., Umeda, H., Nomoto, K., Tominaga, N., &Ohkubo, T. 2006, ApJ, 653, 1145, (K06)Korn, A. J., Grundahl, F., Richard, O., et al. 2006, Nature, 442,657Kratz, K.-L., Farouqi, K., Pfeiffer, B., et al. 2007, ApJ, 662, 39Kurucz, R. L., Furenlid, I., Brault, J., & Testerman, L. 1984,Solar flux atlas from 296 to 1300 nm (New Mexico: NationalSolar Observatory) Lawler, J. E., & Dakin, J. T. 1989, Journal of the OpticalSociety of America B Optical Physics, 6, 1457Lawler, J. E., Wickliffe, M. E., den Hartog, E. A., & Sneden, C.2001, ApJ, 563, 1075Lind, K., Asplund, M., & Barklem, P. S. 2009, A&A, 503, 541Ljung, G., Nilsson, H., Asplund, M., & Johansson, S. 2006,A&A, 456, 1181Lodders, K., Plame, H., & Gail, H.-P. 2009, in Landolt-B¨ornstein- Group VI Astronomy and Astrophysics Numerical Data andFunctional Relationships in Science and Technology Volume4B: Solar System. Edited by J.E. Tr¨umper, 2009, 4.4., 44–54Maeda, K., & Nomoto, K. 2003, ApJ, 598, 1163Mashonkina, L. 2013, A&A, 550, A28Mashonkina, L., Belyaev, A. K., & Shi, J.-R. 2016, AstronomyLetters, 42, 366Mashonkina, L., & Gehren, T. 2000, A&A, 364, 249—. 2001, A&A, 376, 232Mashonkina, L., Gehren, T., & Bikmaev, I. 1999, A&A, 343, 519Mashonkina, L., Gehren, T., Shi, J.-R., Korn, A. J., & Grupp, F.2011, A&A, 528, A87Mashonkina, L., Gehren, T., Travaglio, C., & Borkova, T. 2003,A&A, 397, 275Mashonkina, L., Korn, A. J., & Przybilla, N. 2007a, A&A, 461,261Mashonkina, L., & Zhao, G. 2006, A&A, 456, 313Mashonkina, L., Zhao, G., Gehren, T., et al. 2008, A&A, 478, 529Mashonkina, L. I., Vinogradova, A. B., Ptitsyn, D. A.,Khokhlova, V. S., & Chernetsova, T. A. 2007b, AstronomyReports, 51, 903Matteucci, F., & Francois, P. 1989, MNRAS, 239, 885McWilliam, A., Preston, G. W., Sneden, C., & Shectman, S.1995, AJ, 109, 2736Mel´endez, M., Bautista, M. A., & Badnell, N. R. 2007, A&A,469, 1203Mishenina, T. V., & Kovtyukh, V. V. 2001, A&A, 370, 951Mishenina, T. V., Pignatari, M., Korotin, S. A., et al. 2013,A&A, 552, A128Nissen, P. E., & Schuster, W. J. 1997, A&A, 326, 751—. 2010, A&A, 511, L10Nomoto, K., Kobayashi, C., & Tominaga, N. 2013, ARA&A, 51,457Nomoto, K., Tominaga, N., Umeda, H., Kobayashi, C., &Maeda, K. 2006, Nuclear Physics A, 777, 424Pakhomov, Y. V. 2015, Astronomy Reports, 59, 952Pakhomov, Y. V., & Zhao, G. 2013, AJ, 146, 97Pignatari, M., Gallino, R., Heil, M., et al. 2010, ApJ, 710, 1557Prochaska, J. X., Naumov, S. O., Carney, B. W., McWilliam, A.,& Wolfe, A. M. 2000, AJ, 120, 2513Ralchenko, Y. A., Kramida, E., Reader, J., & Team, N. A. 2010,NIST Atomic Spectra Database (version 3.1.5) (USA)Ram´ırez, I., Fish, J. R., Lambert, D. L., & Allende Prieto, C.2012, ApJ, 756, 46Reader, J., Corliss, C. H., Wiese, W. L., & Martin, G. A. 1980,Wavelengths and transition probabilities for atoms and atomicions: Part 1. Wavelengths, part 2. Transition probabilitiesReddy, B. E., Tomkin, J., Lambert, D. L., & Allende Prieto, C.2003, MNRAS, 340, 304Reetz, J. K. 1991, Diploma Thesis (Universit¨at M¨unchen)Romano, D., Karakas, A. I., Tosi, M., & Matteucci, F. 2010,A&A, 522, A32, (R10)Roˇskar, R., Debattista, V. P., Quinn, T. R., Stinson, G. S., &Wadsley, J. 2008, ApJL, 684, L79Ryabchikova, T., Piskunov, N., Kurucz, R. L., et al. 2015, PhyS,90, 054005Ryabchikova, T., Piskunov, N., Pakhomov, Y., et al. 2016,MNRAS, 456, 1221Rybicki, G. B., & Hummer, D. G. 1991, A&A, 245, 171 on-LTE study of F and G dwarfs. II. Abundance patterns from Li to Eu43