On the Metallicities of Kepler Stars
Subo Dong, Zheng Zheng, Zhaohuan Zhu, P. De Cat, J.N. Fu, X.H. Yang, Haotong Zhang, Ge Jin, Yong Zhang
aa r X i v : . [ a s t r o - ph . E P ] J un On the Metallicities of
Kepler
Stars
Subo Dong , Zheng Zheng , Zhaohuan Zhu , , P. De Cat , J.N. Fu , X.H. Yang , , HaotongZhang , Ge Jin , Yong Zhang ABSTRACT
We use 12000 stars from Large Sky Area Multi-Object Fiber SpectroscopicTelescope (LAMOST) spectroscopic data to show that the metallicities of
Kepler field stars as given in the Kepler Input Catalog (KIC) systematically underes-timate both the true metallicity and the dynamic range of the
Kepler sample.Specifically, to the first order approximation, we find[Fe / H] KIC = − .
20 + 0 . / H] LAMOST , with a scatter of ∼ .
25 dex, due almost entirely to errors in KIC. This relationis most secure for − . < [Fe / H] LAMOST < +0 . >
200 compar-ison stars per 0.1 dex bin and good consistency is shown between metallicitiesdetermined by LAMOST and high-resolution spectra. It remains approximatelyvalid in a slightly broader range. When the relation is inverted, the error in truemetallicity as derived from KIC is (0.25 dex)/0.43 ∼ . Subject headings: planetary systems – stars: abundances Kavli Institute for Astronomy and Astrophysics, Peking University, Yi He Yuan Road 5, Hai DianDistrict, Beijing, 100871, China Department of Physics and Astronomy, University of Utah, Salt Lake City, UT 84112, USA Department of Astrophysical Sciences, Princeton University, Princeton, NJ, 08544 Hubble Fellow Royal observatory of Belgium, Ringlaan 3, B-1180 Brussel, Belgium Department of Astronomy, Beijing Normal University, 19 Avenue Xinjiekouwai, Beijing 100875, China National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100012, China University of Science and Technology of China, Hefei 230026, China Nanjing Institute of Astronomical Optics & Technology, National Astronomical Observatories, ChineseAcademy of Sciences, Nanjing 210042, China
1. Introduction
Of the several thousand planetary candidates found by
Kepler , only a few hundred havehigh-resolution spectra of their hosts. The number of “control sample” stars (without knownplanets) with such spectra is much smaller. Hence, large-sample statistical studies generallymust rely on the Kepler Input Catalog (KIC, Brown et al. 2011). It is well known that theKIC was not designed for this purpose, and KIC metallicities are known to be particularlyproblematic. Brown et al. (2011) cautioned that “anyone with a particular interest in stellarmetallicities should not use the KIC for their estimates of log(Z).” Using stellar parametersdetermined from 34 high-resolution spectra of
Kepler target stars, they found that KICmetallicities were ∼ .
17 dex smaller and there were indications of significant systematics.But the faintness of
Kepler stars has meant that high-resolution spectra are expensive intelescope resources.An alternate approach is to obtain medium resolution spectra, which are generallyadequate for estimating basic stellar parameters, i.e., effective temperature T eff , gravity log g ,and metallicity [Fe/H]. Medium resolution spectrographs have the advantage that they canbe easily multiplexed. For example, the Sloan Digital Sky Survey (SDSS) in its variousincarnations has characterized of order 6 × stars using an R ∼ Kepler field with its optical spectrograph, although SDSS-III has begun observing brighter
Kepler stars with its high-resolution APOGEE infrared multi-object spectrograph.Because of the high science value of planetary hosts, exceptional efforts have neverthelessbeen made to obtain spectra. Buchhave et al. (2012) obtained high-resolution spectra for 152hosts, and Everett et al. (2013) obtained R ∼ ∼
115 deg Kepler field withspectroscopy. LAMOST is a Schmidt telescope with a ∼
4m effective aperture and 4000 3 –fibers that can be deployed a 5 ◦ diameter field of view. Here we use data from Data Re-lease 1 (DR1) and Data Release 2 (DR2) from LAMOST (Zhao et al. 2012; Cui et al. 2012;Luo et al. 2012) with R ∼ ∼ .
58 dex. Hence wequantitatively confirmed the warning issued by Brown et al. (2011) that KIC metallicitiesmust be used with extreme caution.LAMOST DR1 and DR2 reports stellar parameters for ∼ Kepler stars with nopreference for known planet hosts as part of the “LAMOST-
Kepler project” to observe alltarget stars in the
Kepler field (De Cat et al. 2014). The LAMOST samples should eventu-ally enable solid statistical investigations that are able to accurately characterize both the“numerators” (targets hosting planets) and the “denominators” of various subsamples. Weourselves are working on analyses regarding dependence of planet frequency on metallicitiesand various other host properties. However, our purpose here is to apply DR1 and DR2Season 1 to a much more limited question: quantifying the systematics of KIC metallicities.
2. LAMOST
Kepler
Sample
We query the LAMOST DR1 and DR2 AFGK-type stars catalog for Kepler stars,but not those that were specifically targeted because they had planets. We find 16959 starswith KIC identifications, of which 317, or about 1.9%, host planetary candidates. This isstatistically indistinguishable from the
Kepler catalog as a whole, which has 2716 candidatehosts out of ∼ ,
000 stars, or 1.8%. We eliminate those with LAMOST log g < . − . < [Fe / H] < +0 . T eff , log g and [Fe/H] in the catalog are determined by the LAMOST Stellar Param-eter pipeline. This pipeline has been built upon the algorithm in Wu et al. (2011) analysingthe commissioning LAMOST data, but it has been significantly improved since then, in par-ticular taking considerable care in handling problems associated with relative flux calibrationof the LAMOST spectra, which was found to be a main source of systematics shown in the
3. Comparison of LAMOST to KIC Metallicities
Figure 1 shows a comparison of LAMOST to KIC metallicities in 0.1 dex bins of LAM-OST metallicity. The outer error bars show the standard deviation and the inner ones showthe standard error of the mean. We makes a linear fit to all the data (the solid line) to gainan understanding of the relation between KIC and LAMOST metallicities to the first orderapproximation. The second and third highest-metallicity bin and the three lowest metallicitybins appear to differ noticeably from the trend. The reason for this is unclear. It could be arelatively large statistical fluctuation or it could be that either the KIC and/or LAMOST de-terminations actually change their trends. After all, the three highest and lowest metallicitybins only contain 6% of stars in the sample, and stars with [Fe / H] . − . / H] & +0 . / H] KIC = ( − . ± . . ± . / H] LAMOST . (1)The scatter in the individual bins is about 0.25 dex. We conclude that not only are theKIC metallicities too low, their dynamic range is substantially compressed relative to themetallicity range of the underlying stars. If the above linear relation is inverted to find truemetallicity from KIC [Fe/H], the observed scatter is ∼ . .
25 dex / .
43. We cautionthat the linear fit given here is to understand the systematics of KIC metallicities. Giventhe large scatter, this relation should not be used to “correct” the KIC metallicity.Figure 2 shows the metallicity distributions of the overlapping LAMOST/KIC sampleas determined by each catalog. Note that the mean LAMOST [Fe/H] is close to solar whilethe mean KIC [Fe/H] is about − . − .
4. Comparison of LAMOST to High-resolution Spectroscopic Metallicities
Buchhave et al. (2012) presented the largest homogeneous high-resolution spectroscopysample of
Kepler stars. They introduced a new stellar parameter classification (SPC) tech-nique that reports an average abundance [M/H] of the elements producing absorption linesbetween 5050 ˚A and 5360 ˚A. In order to compare the SPC [M/H] to LAMOST [Fe/H], wemake use of the study by Torres et al. (2012), who systematically examined SPC-determined[M/H] with [Fe/H] as measured from the widely-used Spectroscopy Made Easy (SME) pack-age (Valenti & Piskunov 1996) and the spectral synthesis code MOOG (Sneden 1973).The upper panel of Figure 3 shows the [M/H] by SPC and [Fe/H] by SME of 44 com-mon stars and by MOOG of 36 common stars observed with high-resolution spectra fromTorres et al. (2012) in filled red circles and green circles, respectively. They have mean dif-ferences of 0 . ± .
015 dex and − . ± .
019 dex, respectively, but their difference showsnoticeable trends in difference fashions as a function of metallicity. These trends have am-plitudes at about 0 . − . ± .
015 dex. The standarderror of the difference is 0 .
10 dex, at essentially the same level of systematics exhibited inthe comparison of three different methods. The middle and lower panels of Figure 3 showthe difference between LAMOST [Fe/H] and SPC [M/H] as a function of effective tempera-ture ( T eff ) and surface gravity (log( g )), and the difference show no noticeable trend over theavailable T eff and log( g ) ranges.The above comparison demonstrates that [Fe/H] measurements from the LAMOSTpipeline are in good agreement with those using high-resolution spectroscopy over a widerange of metallicity from ∼ − . ∼ +0 . K . T eff . K , which corresponds to the T eff range for the majority of the LAMOST sample.The stars used in the Torres et al. (2012) sample to cross-calibrate SPC, SME and MOOGare in the range of 4600 K < T eff < K and − . < Fe / H < +0 .
5, which covers theparameter space of overlapping LAMOST and Buchhave et al. (2012) stars. We cautionthat the reliability of the LAMOST metallicity for stars with T eff outside this range shallbe examined with other high-resolution spectroscopic data. Comprehensive calibrationsof LAMOST stellar parameters using a large, homogeneous high-resolution spectroscopicsample covering a broader range of parameters are underway. 6 –
5. Conclusion
In our view, LAMOST metallicities should be used in place of KIC metallicities wheneverthey are available (and if there are no high-resolution spectra available).And extreme caution is indicated when KIC metallicities are the only ones available. Inparticular, if Equation (1) is inverted to try to derive real metallicities from KIC metallicities,the observed scatter in the individual bins (0.25 dex) must be divided by 0 .
43 to obtain thefinal error, i.e., 0.6 dex.We thank Andrew Gould for stimulating discussions. We are grateful to Boaz Katz,Xiaowei Liu, Ali Luo, Yan Wu, and Marc Pinsonneault for helpful comments. GuoshoujingTelescope (the Large Sky Area Multi-Object Fiber Spectroscopic Telescope LAMOST) is aNational Major Scientific Project built by the Chinese Academy of Sciences. Funding forthe project has been provided by the National Development and Reform Commission. S.D.is supported by “the Strategic Priority Research Program-The Emergence of CosmologicalStructures” of the Chinese Academy of Sciences (Grant No. XDB09000000). LAMOST isoperated and managed by the National Astronomical Observatories, Chinese Academy ofSciences. JNF and XHY acknowledge the support from the Joint Fund of Astronomy ofNational Natural Science Foundation of China (NSFC) and Chinese Academy of Sciencesthrough the Grant U1231202, and the support from the National Basic Research Programof China (973 Program 2014CB845700).
REFERENCES
Ahn, C.P., Alsexandroff, R., Allende Prieto, C., et al. 2013, ApJS, submitted,arXiv:1307.7735Aihara, H., Allende Prieto, C., An, D., et al. 2011, ApJ, 193, 29Brown, T. M., Latham, D. W., Everett, M. E., & Esquerdo, G. A. 2011, AJ, 142, 112Buchhave, L.A., Latham, D.W., Johansen, A., et al. 2012, Nature, 486,375Casagrande, L., Sch¨onrich, R., Asplund, M., et al. 2011, A&A, 530, A138Cui, X., Zhao, Y., Chu, Y., Li, G., Li, Q., Zhang, L., Su, H., et al., 2012, RAA, 12, 1197Dawson, R. I., & Murray-Clay, R. A. 2013, ApJ, 767, L24 7 –De Cat, P., Fu, J.N., et al. 2014, submittedDodson-Robinson, S. E. 2012, ApJ, 752, 72Luo, A-li, Zhang, Haotong, Zhao Yongheng, et al. 2012, RAA, 12, 1243Nordstr¨om, B., Mayor, M., Andersen, J., et al. 2004, A&A, 418, 989Everett, Mark E., Howell, S.B. Silva, D.R., & Szkody, P. 2013, ApJ, 771, 107Schlaufman, K. C., & Laughlin, G. 2011, ApJ, 738, 177Sneden, C. A. 1973, PhD thesis, Univ. Texas, AustinTorres, G., Fischer, D. A., Sozzetti, A., et al. 2012, ApJ, 757, 161Valenti, J.A., & Piskunov, N. 1996, A&AS, 118, 595Wang, J., & Fischer, D. 2013, ApJ, submitted arXiv:1310.7830Wu, Y., Luo, A., Li, H., Shi, J., Prugniel, P., Liang, Y., Zhao, Y., et al., 2011, RAA, 11, 924Zhao, G., Zhao, Y., Chu, Y., Jing, Y., Deng, L., 2012, RAA, 12, 723
This preprint was prepared with the AAS L A TEX macros v5.2.
Kepler star metallicities as determined by KIC as a function of LAMOST spec-troscopic metallicity. Outer error bars show standard deviations and inner error barsshow standard errors of the mean. Dotted lines enclose the central 68 .
3% of the dis-tribution. Solid line is fit to all the data, while dashed line removes the bins < − . . [Fe / H] LAMOST . +0 . − . ± .
002 and 0 . ± . −1 −0.5 0 0.5 100.020.040.060.080.10.12 [Fe/H] KIC F r ac ti on mean =−0.04median=−0.03mode =−0.06LAMOST mean =−0.20median=−0.18mode =−0.19KIC Fig. 2.— Metallicity distribution of the sample. Solid: LAMOST [Fe/H]; Dashed: KIC[Fe/H]. The mean, median and mode of each distribution are displayed. 10 – -0.4 -0.2 0 0.2 0.4-0.4-0.200.20.4 LAMOSTSMEMOOG
Fig. 3.— Comparison of [Fe/H] as determined by LAMOST (blue), the SME technique(red) or the MOOG technique (green) with [M/H] as determined by the SPC technique(Buchhave et al. 2012). The LAMOST/SPC comparison is based on 47
Kepler stars incommon, while the SPC/SME and SPC/MOOG comparisons are based on 44 and 36 high-resolution stars, respectively from Torres et al. (2012). All three comparison show smallmean offset (LAMOST/SPC: − . ± .
015 dex, SME/SPC: 0 . ± .
015 dex, MOOG/SPC: − . ± .
019 dex). SME/SPC and MOOG/SPC comparisons show trends in differentfashions at amplitudes of ∼ . .
10 dex, suggesting that LAMOST [Fe/H] determinations are reliableat the level that present high-resolution spectroscopic methods are most secure. The middleand lower panels plot the LAMOST/SPC difference as a function of T eff and log( gg