A sample of metal-poor galaxies identified from the LAMOST spectral survey
Yulong Gao, Jianhui Lian, Xu Kong, Zesen Lin, Ning Hu, Haiyang Liu, Enci Wang, Zihuang Cao, Yonghui Hou, Yuefei Wang, Yong Zhang
aa r X i v : . [ a s t r o - ph . GA ] J a n Research in Astron. Astrophys. Vol. 0 (200x) No. 0, 000–000 (L A TEX: ms.tex; printed on October 10, 2018; 20:45) R esearchin A stronomyand A strophysics A sample of metal-poor galaxies identified from the LAMOSTspectral survey
Yulong Gao , , Jianhui Lian , , Xu Kong , , Zesen Lin , , Ning Hu , , Haiyang Liu , , EnciWang , , Zihuang Cao , Yonghui Hou , Yuefei Wang , and Yong Zhang CAS Key Laboratory for Research in Galaxies and Cosmology, Department of Astronomy,University of Science and Technology of China, Hefei 230026, China;
E-mail:[email protected]; [email protected] School of Astronomy and Space Science, University of Science and Technology of China, Hefei230026, China Key Laboratory of Optical Astronomy, National Astronomical Observatories, Chinese Academy ofSciences, Beijing 100012, China Nanjing Institute of Astronomical Optics & Technology, National Astronomical Observatories,Chinese Academy of Sciences, Nanjing 210042, China
Abstract
We present a sample of 48 metal-poor galaxies at z < . selected from92,510 galaxies in the LAMOST survey. These galaxies are identified for their de-tection of the auroral emission line [O III ] λ σ level, which allows a di-rect measurement of the electron temperature and the oxygen abundance. The emissionline fluxes are corrected for internal dust extinction using Balmer decrement method.With electron temperature derived from [O III ] λλ , / [O III ] λ and electrondensity from [S II ] λ / [S II ] λ , we obtain the oxygen abundances in our sam-ple which range from
12 + log(O / H) = 7 . (0.09 Z ⊙ ) to . (0.6 Z ⊙ ). We findan extremely metal-poor galaxy with
12 + log(O / H) = 7 . ± . . With multibandphotometric data from FUV to NIR and H α measurements, we also determine the stel-lar masses and star formation rates, based on the spectral energy distribution fitting andH α luminosity, respectively. We find that our galaxies have low and intermediate stellarmasses with . ≤ log(M / M ⊙ ) ≤ . , and high star formation rates (SFRs) with − . ≤ log(SFR / M ⊙ yr − ) ≤ . . We also find that the metallicities of our galaxiesare consistent with the local T e -based mass-metallicity relation, while the scatter is about0.28 dex. Additionally, assuming the coefficient of α = 0 . , we find most of our galax-ies follow the local mass-metallicity-SFR relation, while a scatter about 0.24 dex exists,suggesting the mass-metallicity relation is weakly dependent on SFR for those metal-poorgalaxies. Key words: galaxies: abundances - galaxies: evolution - galaxies: starburst - galaxies:star formation
Metal-poor galaxies are less chemically evolved galaxies and provide ideal laboratory for inves-tigating galaxy properties in extreme condition (Shi et al., 2005; Lian et al., 2016). Among them,extremely metal-poor galaxies (hereafter XMPGs), defined by their low oxygen abundance with
12 + log(O / H) ≤ . (Kniazev et al., 2003; Pustilnik & Martin, 2007; Doyle et al., 2005), are the Y.L. Gao et al. most promising young galaxy candidates in the local universe (Izotov & Thuan, 2004). These XMPGsare suspected to be primeval galaxies that are undergoing their first major mass assembly at the observedredshift (Kniazev et al., 2003). Studying these extreme objects can improve our understanding about theearly stages of galaxy assembly.The determination for abundance of elements are considered more reliable if the electron tem-perature T e can be measured directly, because the metallicity is anti-correlated with the elec-tron temperature. The electron temperature can be obtained using the auroral line ratios, such as[O III ] λ / [O III ] λλ , . This technique is often called the T e method (Aller, 1984). However,galaxies with metallicities derived from the T e method with [O III ] λ detections above 3 σ are ex-tremely rare. To date, only about 174 such objects has been found (Ly et al., 2014, 2016b).In order to enlarge the sample of metal-poor galaxies with [O III ] λ T e -based relation in Section 5. In addition, we discuss our results inthe context of other studies in Section 6. Finally, we summarize our main conclusions in Section 7.Throughout this paper, we adopt a flat cosmology with Ω Λ = 0 . , Ω M = 0 . , and H = 70 km s − Mpc − to determine distance-dependent measurements. For reference, we adopt
12 + log(O / H) ⊙ =8 . (Allende Prieto et al., 2001) for metallicity measurements quoted against the solar value, Z ⊙ . The Large Sky Area Multi–Object Fiber Spectroscopic Telescope (LAMOST, also called the Guo ShouJing Telescope) is a special reflecting Schmidt telescope with an effective aperture of 4 m and a fieldof view (FoV) of 5 ◦ (Wang et al., 1996; Su & Cui, 2004; Zhao et al., 2012; Cui et al., 2012; Luo et al.,2012). It is equipped with 4000 fibers, covering a wavelength range of 3800 − A (Luo et al., 2015)at a resolving power R ≈ r ≈ According to Song et al. (2012) and Luo et al. (2015), LAMOST 1D spectra are extracted from the CCDimages used by the LAMOST 2D pipeline. The wavelength calibration of each spectrum is accomplishedby using arc lamp spectra lines, with an average calibration error less than 0.02 ˚ A . The accuracy of therelative flux calibration of LAMOST is above 90 % .Assuming unimodal gaussian line profile, we obtain fluxes of strong emission lines such as[O II ] λ , [O III ] λ , H β , [O III ] λλ , , H α and [S II ] λλ , by fitting their lineprofile using the IDL package MPFIT (Markwardt et al., 2009). The expected location of emission linesare based on a priori redshift determined by the [O
III ] line. In addition, to estimate the signal-to-noiseratios (S/N) of emission lines, we follow the calculation method in Ly et al. (2014).
Among all the galaxies from the LAMOST ExtraGAlactic Surveys (LEGAS), we first select a subsampleof metal-poor galaxies with emission line flux ratios [N II ] λ / H α ≤ . , which consists of 665 sample of metal-poor galaxies in the LAMOST spectral survey 3 galaxies. Among them, we identify 237 objects with [O III ] λ detection at ≥ σ . We inspect these237 objects visually, and find 73 of them are false detections. We also exclude 115 objects that are H II regions in nearby large galaxies using optical images with SDSS DR12 skyserver . Finally, we checkthe right ascension and declination of the remaining sources, and note that 1 object was observed twiceby LAMOST. We keep the observation that has better spectral quality.As a consequence, our final sample consists of 48 galaxies, making up only . of all theLAMOST galaxies until DR4 Q2; this fraction is nearly the same as SDSS (Ly et al., 2014). The medianS/N of [O III ] λ is 6.1. We obtain the H β equivalent width (EW) by dividing the H β flux by con-tinuum spectral flux intensity, which is assumed as the average value of observed flux intensities within50 ˚ A wide component around the H β line. All of the galaxy spectra in our sample show strong emissionlines with a median (average) EW of H β of 42.9 (53.5) ˚ A . The EW distribution of H β is shown in Figure1. Fig. 1
The distribution of H β equivalent widths in our sample. The dashed line and dottedline represent the median and average equivalent widths of 42.9 ˚ A and 53.5 ˚ A , respectively.To exclude possible AGN contamination, we use the BPT diagram (see Baldwin et al., 1981;Veilleux & Osterbrock, 1987; Kewley et al., 2001; Kauffmann et al., 2003). Figure 2 shows the distri-bution of our sample in the BPT diagram. The grayscale 2D histogram shows the number density ofLAMOST galaxies. The blue dots represent 48 galaxies in our final sample, and the black crosses rep-resent 24 galaxies in our final sample that have also been spectrally detected by SDSS. Among these24 galaxies detected with SDSS, 19 galaxy spectra also have the [O III ] λ detections above 3 σ ; wewill compare these 19 galaxy spectra from LAMOST and SDSS in Section 6.1. The solid and dashedlines are the demarcation curves between SFGs and AGNs derived by Kauffmann et al. (2003) andKewley et al. (2001). Galaxies located between the two lines are usually classified as composite objects,which may host a mixture of star formation and AGN. It can be seen that all of the galaxies in our final http://skyserver.sdss.org/dr12/en/tools/chart/listinfo.aspx Y.L. Gao et al.
Fig. 2
BPT diagram for our metal-poor galaxy (MPG) sample and all the LAMOST galaxies.The blue dots represent our final sample galaxies. The black crosses represent these objectsin our final sample that have also been spectrally detected with SDSS. The grayscale 2Dhistogram shows the number density of LAMOST galaxies. The solid and dashed lines arethe demarcation curves between SFGs and AGNs defined by Kauffmann et al. (2003) andKewley et al. (2001), respectively.sample are located in the star-forming region; however, this is unsurprising since we initially selectedsources with [N II ] λ / H α ≤ . .Figure 3 shows example spectra for eight galaxies in our sample that have both been spectrallydetected by LAMOST and SDSS. For each object, the left panel shows the LAMOST spectrum, whilethe right panel shows the SDSS spectrum. All of these spectra show strong emission lines such as[O II ] λ (for all LAMOST spectra and part of SDSS spectra), H β , [O III ] λλ , , H α and[S II ] λλ , . The inserted panels show the zoomed spectra adjacent to [O III ] λ lines. It canbe seen that the weak [O III ] λ lines are all detected in the spectra of these galaxies. We correct the emission-line fluxes for internal dust attenuation using the Balmer decrement measure-ments, which estimate the dust extinction by inspecting the change of Balmer line ratio, such as H α/ H β ,from intrinsic value. Generally, the underlying stellar absorption in the Balmer lines should be well de-termined to obtain a reliable emission line measurement (Hu et al., 2016). In this work, we first subtractthe underlying stellar continuum and stellar absorption for each spectrum using the STARLIGHT spectralsynthesis code (Cid Fernandes et al., 2005). We assume the intrinsic flux ratio of ( H α/ H β ) = 2 . (Hummer & Storey, 1987) under Case B recombination and use the Calzetti et al. (2000) reddening for- sample of metal-poor galaxies in the LAMOST spectral survey 5 malism to derive the color excesses E ( B − V ) , and then correct the emission line fluxes. In addition,we manually set the color excesses E ( B − V ) to zero when the H α/ H β ratios are less than 2.86.The resulting reddening-corrected emission line fluxes relative to H β and color excesses are listedin Table 1. As showed in panel a of Figure 4, the measured dust extinction are very low with an average E ( B − V ) value as 0.03 mag. With significant detection of [O
III ] λ , we can determine the metallicity using the so-called T e method. In this work, we use the python package PYNEB (Luridiana et al., 2015) to calculate theelectron densities ( n e ) and electron temperatures ( T e ), which is evolved from the IRAF nebular package(Shaw & Dufour, 1995; Shaw et al., 1998). Nicholls et al. (2013) have demonstrated that the electrontemperatures would be overestimated, and thus the oxygen abundances would be underestimated whenusing older collision strength data and approximate temperature calibration methods from Izotov et al.(2006). Therefore, we need to set the atomic recombination data and atomic collision strength data be-fore the calculation. We adopt the atomic recombination data of Froese Fischer & Tachiev (2004) for O + , O ++ , and Tayal & Zatsarinny (2010) for S + . For collision strength data, we adopt those fromKisielius et al. (2009) for O + , Storey et al. (2014) for O ++ , and Tayal & Zatsarinny (2010) for S + .As encouraged in Luridiana et al. (2015), we use a cross-converging method to calcu-late the electron temperatures of O ++ regions ( T e ( [O III ] ) ) and electron densities ( n e ) withratios of [O III ] λ / [O III ] λλ , (Nicholls et al., 2013) and [S II ] λ / [S II ] λ (Tayal & Zatsarinny, 2010). Once the electron temperatures T e ( [O III ] ) and densities n e are determined,we can obtain the ionic oxygen O ++ abundances using the [O III ] λλ , / H β ratios with the re-lation derived from Izotov et al. (2006). In order to derive the electron temperatures T e ( [O II ] ) of O + regions, we follow an iterative method used in Nicholls et al. (2014), T e ( [O II ] ) = T e ( [O III ] ) × (3 . − . − . + 0 . ) , (1)where Z is the total oxygen abundance,
12 + log(O / H) . The temperature T e ( [O II ] ) and abundance Z will converge within five iterations, starting by using the O ++ abundance as the total oxygen abun-dance. Here, the O + abundance is determined from T e ( [O II ] ) and the [O II ] λ / H β ratio using theIzotov et al. (2006) relation. Using other methods from Garnett (1992) or L´opez-S´anchez et al. (2012),we will get a higher T e ( [O II ] ) about 0.02 dex and a lower metallicity about 0.05 dex given a T e ( [O III ] ) .Similarly, we will get a higher T e ( [O II ] ) about 0.03 dex and thus get a lower metallicity about 0.05 dexwhen adopt the standard two-zone temperature model from Izotov et al. (2006) and Andrews & Martini(2013). In our final sample, 31 galaxies also have the [O II ] λλ , detections above 3 σ . Usingthe [O II ] λ / [O II ] λλ , ratios to derive T e ( [O II ] ) , we will get lower T e ( [O II ] ) and highermetallicities, the differences between these average values on T e ( [O II ] ) and metallicities are about 360 K and 0.06 dex, respectively.To estimate the uncertainties of electron temperatures, electron densities and oxygen abundances,we repeat the calculation 2000 times. For every object, we produce a series of fluxes for each emissionline with a Gaussian distribution, assuming its average is the measured line flux and standard deviationis the measured error. Then, our final temperature, density and oxygen abundance are deemed to bethe median values of these 2000 calculations, and the corresponding errors are estimated as the half of − range of their distributions. We list the final electron temperatures, electron densities andoxygen abundances of our sample in Table 1.Figure 4 shows the distributions of color excesses, electron densities, electron temperatures and oxy-gen abundances. The electron densities and temperatures in our sample range from . to . − and (0 . − . × K , with median values of . − and . × K , respectively. Theiroxygen abundances range from 7.63 to 8.46, with a median of 8.16. The only XMPG found in oursample is ID26 with
12 + log(O / H) = 7 . ± . , which has already been found by Izotov et al. Y.L. Gao et al. (2012a) with
12 + log(O / H) = 7 . ± . . Interestingly, galaxy ID33, also named as RC2 A1228+12,was regarded as an XMPG in Kunth & ¨Ostlin (2000) and Brorby et al. (2014), but the metallicity
12 + log(O / H) = 7 . ± . indicates it is not an XMPG. This judgement was also supportedby Pustilnik et al. (2002) and Izotov et al. (2012b) with metallicity measurements of . ± . and . ± . , respectively. To determine the galactic stellar masses of our sample galaxies, we use the IDL code library
FAST devel-oped by Kriek et al. (2009) to perform the spectral energy distribution (SED) fitting.
FAST compares thephotometry measurements with stellar population synthesis models, based on the minimum χ template-fitting procedure, to determine mass-to-light ratios, which can be used to estimate the stellar masses ofgalaxies. We use the stellar templates of Bruzual & Charlot (2003) and a Chabrier (2003) initial massfunction (IMF) to synthesize magnitudes. These models span four metallicities (0.004, 0.008, 0.02, 0.05 Z ⊙ ) and an exponentially decreasing star formation models (SFR ∝ e − t /τ ) with a step ∆ log ( τ ) = 0 . from . ≤ log ( τ ) ≤ . . We assume the dust attenuation law from Calzetti et al. (2000) allowing E ( B − V ) to vary from 0.0 to 2.0 and stellar population ages ranging from 0 to 100 Gyr. To deter-mine the uncertainties of stellar masses, we use the Monte Carlo simulations and define the number ofsimulations as 1000. We choose the confidence interval as 68 % .Photometric measurements are collected from various survey catalogue. We adopt values of MOD - ELMAG magnitudes of u, g, r, i , and z bands from SDSS DR12 photometry catalogue (Abazajian et al.,2004; Alam et al., 2015), magnitudes of J, H , and K s bands from 2MASS All-Sky Point Source cat-alog (PSC) and 2MASS All-Sky Extended Source Catalog (XSC) (Skrutskie et al., 2006), magnitudesof W (3.4 µm ) and W (4.6 µm ) from All WISE Source Catalog (Wright et al., 2010), and magni-tudes of FUV and NUV from GALEX
GR6/7 Data Release Catalog (Bianchi et al., 2014). However,not all of our sample galaxies have these photometric measurements. For example, 45 galaxies haveFUV photometry, while 3 galaxies are not located in
GALEX surveyed areas. For these three galaxies,we just use their magnitudes from u band to W band to perform the SED fitting. We find our samplegalaxies spanning three orders with . ≤ log(M / M ⊙ ) ≤ . . We should note that we do not makethe point spread function (PSF) matching for our photometric data using same observation aperture,which may lead to some uncertainties in stellar mass measurements. The average and median valueson stellar mass measurement uncertainties are 0.14 dex and 0.12 dex, respectively. However, comparingour results with total stellar masses in MPA-JHU catalog (Kauffmann et al., 2003; Brinchmann et al.,2004) for these galaxies included in MPA-JHU catalog, we find the differences of average and medianvalues are about 0.1 dex and 0.03 dex, respectively. sample of metal-poor galaxies in the LAMOST spectral survey 7 Fig. 3
Example spectra of objects in our sample that have both been spectrally detected withLAMOST and SDSS. For every object, the left panel shows LAMOST spectrum, while theright panel shows SDSS spectrum. Inserted panels show the zoomed in spectra adjacent to the[O
III ] λ lines. Y.L. Gao et al.
Fig. 4
The distributions of color excesses, electron densities, electron temperatures and oxy-gen abundances of our sample galaxies. The dashed lines represent the median values of theseparameters, 0.00 mag, 67.0 cm − , 1.20 × K and 8.16, respectively. The dotted lines rep-resent the average values 0.03 mag, 98.9 cm − , 1.24 × K and 8.13, respectively. s a m p l e o f m e t a l - poo r g a l a x i e s i n t h e L A M O S T s p ec t r a l s u r v e y9 Table 1: The sample of metal-poor galaxies in LAMOST surveyID a RA b DEC b z b I ( λ ) /I ( H β ) c I ( H β ) c EW ( H β ) d E ( B − V ) e T e [O III ] f n f e
12 + log(O / H) Te log(M) log(SFR) SDSS g (deg) (deg) [O II ] λ [O III ] λ [O III ] λ [O III ] λ H α [S II ] λ [S II ] λ ( ˚A) (mag) ( K) ( cm − ) ( M ⊙ ) ( M ⊙ yr − )1 0.04352 4.93125 0.031 1.748 0.049 1.760 5.432 2.326 0.114 0.063 1761.90 81.46 0.00 1.16 38.78 8.22 8.250 -0.556 00.054 0.013 0.018 0.051 0.022 0.007 0.007 15.87 0.22 0.01 0.10 42.91 0.12 0.050 0.0092 0.22617 18.50614 0.055 2.205 0.030 1.267 3.924 2.135 0.178 0.133 7058.70 46.76 0.00 1.05 99.45 8.29 9.180 0.527 00.018 0.004 0.007 0.019 0.010 0.002 0.001 33.06 0.23 0.01 0.05 15.32 0.06 0.090 0.0053 17.64580 2.11408 0.016 2.258 0.032 1.333 3.876 2.685 0.193 0.159 3282.40 34.17 0.00 1.07 234.91 8.28 8.720 -0.782 00.018 0.004 0.004 0.010 0.007 0.002 0.002 8.34 0.31 0.01 0.05 22.81 0.05 0.120 0.0034 19.02766 1.03444 0.035 2.164 0.072 1.335 3.977 2.054 0.130 0.108 439.67 43.01 0.00 1.47 272.39 7.88 7.600 -1.185 10.759 0.021 0.022 0.059 0.028 0.007 0.007 5.88 0.26 0.01 0.19 142.65 0.16 0.175 0.0145 23.00047 -2.74083 0.018 1.681 0.056 1.632 5.063 2.362 0.177 0.152 797.69 62.08 0.00 1.19 303.92 8.16 7.880 -1.382 00.062 0.010 0.016 0.043 0.021 0.008 0.006 6.63 0.48 0.01 0.08 110.75 0.09 0.180 0.0096 31.87386 4.73166 0.011 3.279 0.019 1.089 3.269 2.579 0.267 0.193 5354.90 32.06 0.00 0.97 50.44 8.44 8.230 -0.863 00.026 0.003 0.003 0.007 0.006 0.001 0.001 11.49 0.52 0.01 0.05 8.88 0.07 0.110 0.0027 37.72624 1.91723 0.025 3.333 0.069 0.997 2.839 2.860 0.379 0.229 1143.89 12.62 0.20 1.82 30.48 7.67 8.050 -0.931 01.790 0.035 0.031 0.068 0.067 0.017 0.016 26.25 0.58 0.02 0.37 34.73 0.25 0.125 0.0248 39.07656 1.75273 0.023 2.130 0.053 1.828 5.206 2.860 0.200 0.144 6174.59 55.21 0.25 1.23 79.89 8.17 8.020 -0.187 00.037 0.009 0.018 0.046 0.026 0.009 0.008 54.01 0.07 0.01 0.06 65.41 0.06 0.015 0.0099 43.15004 19.68750 0.029 3.078 0.044 1.270 3.582 1.836 0.174 0.128 4755.70 27.36 0.00 1.22 73.19 8.15 8.010 -0.304 00.015 0.003 0.003 0.008 0.004 0.002 0.002 10.31 0.16 0.01 0.04 21.25 0.04 0.100 0.00210 47.66477 5.23287 0.064 2.962 0.015 1.111 3.332 2.853 0.242 0.182 3494.70 35.56 0.00 0.96 113.42 8.44 9.150 0.199 00.053 0.005 0.011 0.029 0.025 0.002 0.002 30.42 0.23 0.01 0.07 15.29 0.10 0.140 0.00911 47.91879 2.57282 0.020 1.685 0.098 2.115 6.153 2.582 0.139 0.086 6170.00 103.02 0.00 1.37 15.47 8.05 8.070 -0.389 00.021 0.003 0.005 0.013 0.005 0.001 0.001 12.86 6.06 0.01 0.00 4.26 0.00 0.035 0.00212 49.35046 3.50387 0.039 2.179 0.047 1.523 4.753 2.860 0.177 0.129 2297.92 81.85 0.09 1.17 67.05 8.19 8.350 -0.161 00.181 0.009 0.010 0.027 0.016 0.005 0.004 12.85 0.73 0.01 0.08 44.49 0.08 0.140 0.00613 51.95898 1.02635 0.109 2.973 0.030 1.111 3.332 2.816 0.443 0.300 1148.30 45.87 0.00 1.11 6.72 8.25 9.020 1.945 11.280 0.232 0.193 0.458 0.403 0.154 0.132 151.78 0.21 0.01 0.06 7.13 0.07 0.395 0.00514 123.48118 23.14714 0.015 3.171 0.052 1.245 3.601 2.339 0.226 0.161 1515.70 50.33 0.00 1.33 45.20 8.06 6.600 -1.290 00.077 0.008 0.007 0.017 0.011 0.003 0.003 6.85 0.31 0.01 0.09 29.09 0.08 0.305 0.00515 129.35571 37.51241 0.042 2.942 0.050 1.118 3.206 2.639 0.340 0.230 942.03 30.75 0.00 1.34 25.58 8.01 8.550 -0.561 10.178 0.018 0.017 0.038 0.032 0.009 0.009 10.81 0.22 0.01 0.21 26.69 0.18 0.195 0.01216 136.71599 41.36413 0.135 3.205 0.056 1.107 3.224 2.166 0.231 0.182 446.71 32.70 0.00 1.43 187.71 7.95 9.270 0.031 10.077 0.016 0.030 0.070 0.045 0.012 0.008 9.18 0.42 0.02 0.17 100.22 0.14 0.280 0.02117 139.46461 40.97089 0.024 2.608 0.031 1.403 3.912 2.207 0.262 0.189 721.63 42.77 0.00 1.06 52.69 8.32 8.110 -1.160 00.075 0.013 0.015 0.035 0.020 0.004 0.004 6.30 0.10 0.01 0.14 23.70 0.17 0.095 0.00918 139.95418 4.48470 0.012 2.452 0.057 1.376 4.164 2.338 0.227 0.167 2868.10 34.57 0.00 1.32 73.19 8.04 8.070 -1.211 00.194 0.015 0.016 0.041 0.023 0.005 0.004 27.82 0.09 0.01 0.14 38.93 0.13 0.180 0.01019 140.17699 5.73677 0.038 1.029 0.082 2.450 7.272 2.860 0.100 0.071 11516.86 96.12 0.23 1.20 38.78 8.22 8.220 0.525 00.014 0.003 0.007 0.019 0.008 0.001 0.001 30.64 0.31 0.01 0.02 23.27 0.02 0.205 0.00320 140.92796 3.36689 0.012 2.760 0.050 1.340 3.788 2.312 0.210 0.150 1090.50 34.17 0.00 1.30 52.69 8.07 7.330 -1.629 10.086 0.014 0.015 0.035 0.021 0.006 0.005 9.79 0.16 0.01 0.14 41.58 0.12 0.355 0.01021 142.63454 34.43079 0.017 1.689 0.045 1.443 4.180 2.860 0.190 0.104 1297.69 34.84 0.01 1.10 56.29 8.24 7.740 -1.114 10.037 0.010 0.012 0.029 0.020 0.006 0.006 8.78 0.21 0.01 0.10 66.58 0.11 0.300 0.00722 145.71992 35.79055 0.015 1.676 0.055 1.538 4.217 2.210 0.166 0.081 2180.30 100.80 0.00 1.21 19.54 8.10 7.339 -1.147 1Continued on next page Y . L . G a o e t a l . Table 1: – continued from previous pageID a RA b DEC b z b I ( λ ) /I ( H β ) c I ( H β ) c EW ( H β ) d E ( B − V ) e T e [O III ] f n f e
12 + log(O / H) Te log(M) log(SFR) SDSS g (deg) (deg) [O II ] λ [O III ] λ [O III ] λ [O III ] λ H α [S II ] λ [S II ] λ ( ˚A) (mag) ( K) ( cm − ) ( M ⊙ ) ( M ⊙ yr − )0.021 0.005 0.006 0.015 0.008 0.002 0.002 7.31 0.15 0.01 0.03 10.87 0.03 0.074 0.00423 146.00691 50.87919 0.038 2.683 0.028 1.048 2.923 2.471 0.320 0.235 1322.90 37.23 0.00 1.12 73.19 8.19 8.960 -0.485 10.036 0.009 0.008 0.018 0.015 0.004 0.003 7.63 0.64 0.01 0.11 22.38 0.13 0.185 0.00624 147.03999 2.52892 0.021 2.681 0.032 1.348 4.014 2.844 0.279 0.194 2646.10 35.59 0.00 1.18 30.48 8.19 8.270 -0.638 10.086 0.009 0.010 0.026 0.019 0.005 0.006 16.94 0.09 0.01 0.10 26.41 0.10 0.175 0.00725 148.70041 38.45017 0.017 3.040 0.024 0.939 2.683 2.440 0.370 0.269 1971.30 29.46 0.00 1.09 62.79 8.24 8.320 -1.003 10.036 0.008 0.008 0.016 0.014 0.004 0.004 11.02 0.17 0.01 0.13 17.34 0.15 0.070 0.00626 154.10216 37.91277 0.004 0.687 0.097 1.346 4.106 2.860 0.074 0.056 3778.46 138.04 0.01 1.64 115.85 7.63 6.550 -2.029 10.022 0.003 0.003 0.007 0.005 0.001 0.002 6.15 0.19 0.01 0.02 52.03 0.01 0.070 0.00227 157.45534 16.18091 0.011 3.230 0.040 1.187 3.468 2.007 0.278 0.203 1701.80 27.67 0.00 1.22 64.22 8.16 8.430 -1.548 10.059 0.010 0.012 0.028 0.016 0.005 0.005 13.14 0.21 0.01 0.11 33.81 0.11 0.005 0.00828 161.47824 1.06829 0.026 2.243 0.045 1.549 4.167 2.860 0.216 0.161 22426.28 108.84 0.18 1.17 91.11 8.19 7.860 0.478 10.012 0.002 0.003 0.006 0.004 0.001 0.001 33.64 0.24 0.01 0.02 7.99 0.02 0.005 0.00129 162.63853 22.31531 0.046 2.084 0.052 1.394 3.964 2.826 0.313 0.218 1269.00 39.66 0.00 1.25 16.87 8.07 8.360 -0.308 10.030 0.007 0.008 0.019 0.014 0.004 0.004 6.06 0.19 0.01 0.06 15.18 0.06 0.305 0.00530 163.09535 32.63737 0.005 2.665 0.028 1.020 2.611 2.412 0.319 0.225 1828.20 20.58 0.00 1.18 22.43 8.12 7.640 -2.135 10.022 0.004 0.004 0.007 0.006 0.002 0.001 4.74 0.06 0.01 0.06 8.89 0.06 0.110 0.00231 170.34760 6.66713 0.009 2.816 0.064 1.052 3.069 2.860 0.257 0.193 1169.43 27.91 0.01 1.30 99.45 8.03 7.410 -1.762 10.073 0.013 0.012 0.027 0.025 0.006 0.006 9.96 0.05 0.01 0.15 51.12 0.14 0.280 0.00932 186.62852 28.84698 0.027 3.039 0.046 0.929 2.844 2.860 0.423 0.315 1498.23 25.58 0.03 1.44 89.21 7.91 8.650 -0.727 00.062 0.012 0.011 0.025 0.024 0.007 0.006 12.52 0.35 0.01 0.15 32.31 0.12 0.075 0.00933 187.70251 12.04523 0.004 1.118 0.109 1.565 4.442 2.697 0.109 0.080 4437.80 85.03 0.00 1.67 73.19 7.70 6.390 -1.976 10.016 0.002 0.003 0.008 0.005 0.001 0.001 7.57 0.16 0.01 0.02 26.96 0.01 0.070 0.00234 188.68443 10.71915 0.032 2.124 0.052 1.531 4.283 2.740 0.185 0.171 1392.00 67.30 0.00 1.23 464.63 8.13 8.240 -0.583 10.017 0.004 0.005 0.012 0.008 0.002 0.002 3.93 0.43 0.01 0.04 34.42 0.04 0.020 0.00335 189.23994 10.13030 0.027 1.485 0.039 1.366 4.083 2.856 0.249 0.183 445.10 37.11 0.00 1.13 85.32 8.14 8.640 -1.205 10.064 0.016 0.019 0.049 0.034 0.008 0.008 5.16 0.31 0.01 0.17 62.00 0.19 0.225 0.01236 189.52870 10.16557 0.004 0.642 0.099 1.953 5.567 2.703 0.089 0.062 2579.20 156.19 0.00 1.43 16.87 7.90 6.590 -2.177 10.019 0.003 0.004 0.011 0.006 0.001 0.001 5.14 0.05 0.01 0.02 15.98 0.01 0.035 0.00237 189.73721 38.09029 0.007 2.820 0.032 1.046 2.960 2.729 0.367 0.257 1454.40 11.09 0.00 1.19 20.55 8.13 8.320 -1.887 10.040 0.007 0.007 0.014 0.013 0.004 0.004 6.43 0.23 0.01 0.09 15.20 0.09 0.060 0.00538 196.86958 54.44713 0.033 2.544 0.035 1.524 4.133 2.860 0.242 0.190 10397.45 73.00 0.03 1.08 162.86 8.30 8.590 0.381 10.014 0.003 0.004 0.009 0.006 0.001 0.001 22.36 0.13 0.01 0.03 9.56 0.03 0.075 0.00239 198.22404 17.20867 0.052 2.129 0.049 1.436 4.157 2.860 0.210 0.150 3750.41 70.01 0.21 1.21 57.52 8.12 8.020 0.285 10.071 0.014 0.016 0.039 0.027 0.005 0.006 34.19 0.11 0.01 0.13 44.45 0.14 0.155 0.00940 212.67758 38.71842 0.025 3.061 0.028 1.114 3.227 2.860 0.378 0.274 1216.35 23.52 0.00 1.09 52.69 8.27 8.370 -0.796 10.041 0.008 0.008 0.019 0.017 0.005 0.004 6.84 0.26 0.01 0.10 22.31 0.12 0.160 0.00641 220.85265 28.30123 0.013 2.490 0.029 1.199 3.575 2.551 0.291 0.206 2186.30 34.68 0.00 1.07 27.92 8.28 8.210 -1.161 00.038 0.007 0.007 0.018 0.013 0.003 0.003 10.51 0.67 0.01 0.09 16.76 0.11 0.120 0.00542 222.59503 11.40265 0.006 1.260 0.036 1.543 4.234 1.543 0.115 0.214 2445.60 68.75 0.00 1.10 475.04 8.39 7.500 -1.823 012.190 0.007 0.008 0.019 0.007 0.003 0.002 10.98 0.37 0.01 0.07 48.24 0.25 0.080 0.00543 321.52011 8.68640 0.010 1.850 0.061 1.617 4.840 2.809 0.200 0.144 2200.90 49.17 0.00 1.24 44.22 8.12 7.900 -1.386 00.030 0.007 0.009 0.023 0.014 0.002 0.002 10.48 0.21 0.01 0.06 22.79 0.06 0.100 0.005Continued on next page s a m p l e o f m e t a l - poo r g a l a x i e s i n t h e L A M O S T s p ec t r a l s u r v e y11 Table 1: – continued from previous pageID a RA b DEC b z b I ( λ ) /I ( H β ) c I ( H β ) c EW ( H β ) d E ( B − V ) e T e [O III ] f n f e
12 + log(O / H) Te log(M) log(SFR) SDSS g (deg) (deg) [O II ] λ [O III ] λ [O III ] λ [O III ] λ H α [S II ] λ [S II ] λ ( ˚A) (mag) ( K) ( cm − ) ( M ⊙ ) ( M ⊙ yr − )44 325.30099 8.68733 0.032 3.758 0.033 1.055 3.104 2.678 0.343 0.248 1105.70 23.31 0.00 1.19 55.05 8.20 8.600 -0.675 00.032 0.007 0.006 0.016 0.012 0.003 0.003 4.89 0.34 0.01 0.09 15.64 0.09 0.325 0.00445 327.59015 7.60371 0.027 1.501 0.094 1.929 5.831 2.420 0.109 0.082 1003.90 75.16 0.00 1.37 116.11 8.02 8.030 -0.952 00.098 0.009 0.012 0.033 0.014 0.003 0.004 5.56 0.24 0.01 0.05 72.01 0.04 0.130 0.00646 336.10513 6.25363 0.016 2.475 0.054 1.631 4.790 2.327 0.156 0.130 2647.40 63.85 0.00 1.20 260.72 8.19 8.200 -0.962 00.010 0.007 0.008 0.020 0.010 0.003 0.003 10.75 0.34 0.01 0.05 53.29 0.05 0.160 0.00447 341.07831 5.36256 0.006 2.681 0.046 1.436 4.053 2.669 0.228 0.177 6841.20 55.27 0.00 1.19 154.47 8.18 7.960 -1.344 00.022 0.003 0.003 0.007 0.005 0.001 0.001 11.57 0.46 0.01 0.02 10.14 0.02 0.190 0.00248 352.58768 5.52682 0.014 2.313 0.057 1.646 4.242 2.860 0.234 0.173 13399.27 66.73 0.10 1.26 79.89 8.10 7.800 -0.306 00.018 0.003 0.003 0.007 0.005 0.001 0.002 21.79 0.29 0.01 0.02 14.80 0.02 0.010 0.002N OTES : — Basic information, emission line fluxes, electron temperatures and oxygen abundances for our sample galaxies. For every object, the first (second) line presents the parameter (error) values. a ’ID’ is the serial number for every object and it will be referred to throughout this paper. b The right ascension (J2000) and declination (J2000) of our sample galaxies are given in units of degrees. The RA, DEC, and redshift are obtained from the header of the spectral FITS files. c Reddening corrected emission line fluxes for our sample galaxies measured from the LAMOST spectra are relative to H β . The H β fluxes are reported in units of − erg s − cm − . d The H β equivalent widths are given in units of ˚ A , assuming the mean values of observed flux intensities within 50 ˚ A wide component around the H β as the continuum spectral flux intensities. e The nebular color excesses are derived from the observed flux ratios H α/ H β , and are assumed to be zero when the observed flux ratios H α/ H β are less than 2.86. f Electron temperatures are computed from the oxygen emission line ratios [O
III ] λλ , / [O III ] λ . Electron densities are calculated from an iterative process with [O III ] λλ , / [O III ] λ and[S II ] λ / [S II ] λ ratios. g The flag numbers indicate the spectral detected states for our objects with SDSS. ”1” (”0”) represent this object has (not) been spectroscopically detected by SDSS.
In this work, we use the H α emission line luminosities to determine dust-corrected SFRs, assuming aChabrier (2003) IMF and solar metallicity. The SFR can be calculated from H α luminosity as: SFR(M ⊙ yr − ) = R × L(erg s − ) , (2)where R = 4 . × − . However, the latest work of Ly et al. (2016a) demonstrated that the aboveparameter R would overestimate the SFR at lower metallicities, and gave the metallicity-dependentparameter R as: log( R ) = log( SFRL( H α ) ) = − .
34 + 0 . y + 0 . y , (3)where y = log(O / H) + 3 . . Above all, the final stellar masses and SFRs in our sample are listed inTable 1. In panel a of Figure 5, we plot the mass-metallicity relation (MZR) with T e -based metallicities for oursample. These dot symbols of our galaxies are colour-coded by their SFRs. For comparison, we alsoshow the MZRs obtained by Andrews & Martini (2013) and Berg et al. (2012) for their galaxy samplein local universe, which are shown as solid and dotted-dashed black lines, based on T e metallicitycalculation.The MZR in Berg et al. (2012) is a simple linear fit for a small sample of low luminosity metal-poor galaxies with stellar masses log(M / M ⊙ ) ranging from 5.9 to 9.15. As is shown in panel a , themetallicities of our metal-poor galaxies are systematic higher than the MZR in Berg et al. (2012) byabout 0.25 dex. The MZR of Andrews & Martini (2013) is fitted with a asymptotic logarithmic formulafor about two hundred thousands nearby star-forming galaxies in stellar mass from log(M / M ⊙ ) =7 . − . . Most of our galaxies are in good agreement with the MZR in Andrews & Martini (2013),the average and median values of residuals between metallicities and MZR are 0.0015 dex and 0.025dex, respectively. We find that the scatter in the MZR from LAMOST data, relative to the MZR ofAndrews & Martini (2013), is 0.28 dex. The mass-metallicity-SFR relation, also referred to as the fundamental metallicity relation (FMR), isproposed by Mannucci et al. (2010) to describe the anti-correlation between metallicity and SFR at fixedstellar mass. Mannucci et al. (2010) defined a new quantity µ α = log(M) − α log(SFR) to minimize thedispersion in MZR for local galaxies. Using the semi-empirical “strong-line” metallicity calibration ofMaiolino et al. (2008), Mannucci et al. (2010) yielded α = 0 . . However, Andrews & Martini (2013)found a new value of α = 0 . based on the T e metallicity calculation method. In this work, we assumethe value of α = 0 . , since metallicities of our sample galaxy are also determined with T e method.The panel b of Figure 5 shows the FMR for our metal-poor galaxies. The solid black line repre-sents the FMR relation derived by Andrews & Martini (2013). Most of our galaxies are consistent withthe FMR, the average and median values of the residuals between metallicities and FMR are 0.002dex and 0.009 dex, respectively. The scatter in the FMR from LAMOST data, relative to the FMR ofAndrews & Martini (2013), is about 0.24 dex. Among our metal-poor galaxy sample, 24 galaxies are also spectrally detected by SDSS, and are markedwith flag ”1” in ’SDSS’ column of Table 1. We select these galaxies from SDSS DR12 by matching sample of metal-poor galaxies in the LAMOST spectral survey 13
Fig. 5
The T e method mass-metallicity relation and fundamental metallicity relation forour sample galaxies. In both panels, these dot symbols of our galaxies are colour-codedby their SFRs. Panel a : the solid and dotted-dashed black lines represent these MZRsderived from nearby star forming galaxies by Andrews & Martini (2013) and Berg et al.(2012), respectively. Panel b : the solid black line represents the FMR relation derived byAndrews & Martini (2013), which assumed the coefficient on log(SFR) is 0.66 with µ . =log(M) − . . The uncertainties of stellar masses are presented at 68 % confidenceinterval limits.the RA and DEC with our sample within one arcsec. We also obtain the emission line fluxes fromthese 24 SDSS galaxy spectra and find that there are 19 spectra with [O III ] λ detections above3 σ . Similarly, we calculate their metallicities with the T e method. Figure 6 shows the comparisons ofS/Ns for weak [O III ] λ lines, [O III ] and [S II ] line fluxes ratios ([O III ] λλ , / [O III ] λ ,[S II ] λ / [S II ] λ ), electron temperatures ( T e ( [O III ] ) ), electron densities ( n e ) and oxygen abun-dances derived from LAMOST spectra and from SDSS spectra. The quality of the SDSS spectra aregenerally better than those from LAMOST with higher S/N on the weak [O III ] λ . Panels b and d present strong correlation for [O III ] ratios and electron temperatures between the LAMOST and SDSSmeasurements. Although there are several objects that have large dispersion in the comparison for [S II ]ratios and electron densities, the differences between the final oxygen abundances from these two mea-surements are less than 0.01 dex. III ] λ galaxy samples All of galaxies in our sample are selected from the local universe ( . ≤ z ≤ . ), and havestellar masses spanning three orders with . ≤ log(M / M ⊙ ) ≤ . . The only XMPG is de-tected with
12 + log(O / H) = 7 . ± . in our sample, however, it has already been found byIzotov et al. (2012a). In the past decades, there have been many efforts to search for [O III ] λ galaxies and XMPGs in local universe. For example, Kniazev et al. (2003) discovered 12 XMPGs with . ≤
12 + log(O / H) ≤ . using SDSS spectroscopy. Izotov et al. (2006) found 6 new XMPGs in310 [O III ] λ galaxies from SDSS DR3. And Berg et al. (2012) also researched 19 [O III ] λ lowluminosity galaxies with MMT telescope. Additionally, for the intermediate and high redshift universe,Kakazu et al. (2007) mapped 12 XMPGs to z = 1 . with Keck II DEIMOS. Ly et al. (2014) identified4 XMPGs in 20 emission-line galaxies with [O III ] λ at z = 0 . − . by MMT and Keck tele-scope. Amor´ın et al. (2014) also discovered 4 XMPGs from 31 low-luminosity extreme emission line Fig. 6
The comparisons of signal noise ratios for weak [O
III ] λ lines, [O III ] line fluxesratios ( R [O III ] = [O III ] λλ , / [O III ] λ ), [S II ] line fluxes ratios ( R [S II ] = [S II ] λ / [S II ] λ ), electron temperatures, electron densities and oxygen abundancesderived from LAMOST spectra and from SDSS spectra for 19 galaxies in our sample. Thesolid line indicate equality between the LAMOST and SDSS measurements.galaxies out to z = 0 . in the VIMOS Ultra-Deep Survey. Recently, Ly et al. (2015) found 28 metal-poor galaxies with stellar mass spanning . × − . × M ⊙ in DEEP2 at redshift z ∼ . . Ly et al.(2016a) also presented a larger sample of 164 galaxies with weak [O III ] λ line at z = 0 . − . from the “Metal Abundances across Cosmic Time” survey. Compared with these samples, the galaxynumber in our sample is small, which may be caused by a limiting magnitude selection of LAMOST.However, the fraction of galaxies with [O III ] λ detections in LAMOST data is nearly same as thatin SDSS. The MZR relation, which was established originally by Lequeux et al. (1979) and developed byGarnett & Shields (1987), Skillman et al. (1989), Brodie & Huchra (1991), Zaritsky et al. (1994),Tremonti et al. (2004), indicates that the metallicities of galaxies correlate with their stellar masses.Taking SFR into consideration, Mannucci et al. (2010) found that metallicity decreases with increasingSFR at low stellar mass, while does not depend on SFR at high stellar mass ( log(M / M ⊙ ) ≥ . ).However, Yates et al. (2012) suggested that high-mass ( log(M / M ⊙ ) ≥ . ) galaxies have lowermetallicities when their SFRs are lower. These different results may be caused by different metallic-ity calculation methods. In addition, the MZR is also affected by other physical parameters, such asstellar age and gas fraction. Lian et al. (2015) found that the metallicity is strongly dependent on the D n (4000) , which interpreted galaxies with older stellar ages as having higher metallicities at a fixedstellar mass. Hughes et al. (2013) found that galaxies with higher gas fraction have lower metallicities sample of metal-poor galaxies in the LAMOST spectral survey 15 at a fixed mass. Lara-L´opez et al. (2010) and Mannucci et al. (2010) argued that the MZR is in facta projection of FMR. In the past years, many efforts (e.g., Mannucci et al. (2010); Berg et al. (2012);Andrews & Martini (2013); Salim et al. (2014); Ly et al. (2015, 2016b)) have been made to explorethe MZR and FMR feasibility from low mass to high mass, as well as the evolution with redshift. Inthe local universe, the metallicity increases with increasing stellar mass, and decreases with increasingSFR at a fixed stellar mass when log(M / M ⊙ ) ≤ . . Salim et al. (2014) found that the metallic-ity is anti-corrected with specific SFR regardless of different metallicity indicators or methods usedwhen . ≤ log(M / M ⊙ ) ≤ . , while the dependence is weak or absent for massive galaxies when log(M / M ⊙ ) > . . Salim et al. (2014) also demonstrated that the relative specific SFR is a morephysically motivated second parameter for the MZR, and found that the overall scatter in the FMR re-lation does not significantly decrease relative to the dispersion in the MZR. Recently, Bothwell et al.(2016) reported that the FMR is between stellar mass, metallicity and gas mass instead of the SFR. Inaddition, Kashino et al. (2016) measured the metallicity of star-forming galaxies based on Dopita et al.(2016) and Maiolino et al. (2008) calibrations, and found that whether the FMR exists or not dependon the metallicity measurement method. The dependence on metallicity and SFR at high stellar mass isstill in argument (Kashino et al., 2016). For the intermediate redshift universe, Ly et al. (2016b) showedclearly that the MZR evolves toward lower metallicity at fixed stellar mass with increasing redshift z ,and found a much weaker dependence of MZR on SFR than in the local universe.In panel a of Figure 5, we colour-code our galaxy points with their SFRs. Figure 5 shows thatmost of galaxies in our sample have higher metallicities than that of galaxies in Berg et al. (2012),but are consistent with the result in Andrews & Martini (2013). The difference between our work andBerg et al. (2012) may be caused by difference in sample selection and calibrations for electron tem-peratures T e ( [O III ] ) and T e ( [O II ] ) . Andrews & Martini (2013) found that the scatter in MZR for theM-SFR stacks with T e -based metallicity is 0.22 dex, while the scatter in the FMR is 0.13 dex. Thedecrease of scatter value in Andrews & Martini (2013) reflects a strong SFR-dependence on the MZR.From visual examination, we do not find strong dependence of MZR on SFR. However, the scatter inFMR is 0.24 dex, lower than the 0.28 dex scatter in MZR, suggesting MZR is weakly dependent onSFR. We note that the average and median values of metallicity measurement uncertainties are 0.09 dexand 0.08 dex, respectively. The average and median values on stellar mass measurement uncertaintiesare about 0.14 dex and 0.12 dex, respectively. The larger scatters in MZR and FMR compared withAndrews & Martini (2013) relations may be caused by the small galaxy sample size, as well as themeasurement uncertainties on stellar mass. We inspect all the 92,510 galaxies in LAMOST DR3, DR4 Q1 and Q2, and select 48 galaxies with[O
III ] λ detected at ≥ σ as our metal-poor galaxy sample. Using the T e method, we obtain themetallicities of these metal-poor galaxies with a median
12 + log(O / H) = 8 . , spanning from 7.63to 8.46. The most metal-deficient galaxy in our sample is ID26 with
12 + log(O / H) = 7 . ± . ,which is the only XMPG we found, but has already been discovered by Izotov et al. (2012a). We alsoconfirm that the galaxy ID33 (RC2 A1228+12) is not an XMPG.With multiband photometric data from FUV to NIR and H α measurements, we determine the stellarmasses and dust-corrected SFRs, based on the SED fitting and reddening corrected H α luminosities,respectively. We compare the relationship between stellar mass, T e -based metallicity and SFR of ourgalaxies with galaxies in the local universe. We find that the metallicities of our galaxies are in goodagreement with the local T e -based MZR in Andrews & Martini (2013) with average and median valuesof residuals as 0.0015 dex and 0.025 dex, respectively. However, the MZR in Berg et al. (2012) may besystematic lower than the metallicities of our metal-poor galaxies. Assuming the coefficient of α = 0 . ,we find most of our galaxies are consistent with the FMR in Andrews & Martini (2013). However, thescatter in FMR is 0.24 dex, lower than the 0.28 dex scatter in MZR, suggesting MZR has a weakdependence on SFR. Acknowledgements
We are very grateful to the referee’s insightful suggestions and comments, whogreatly improve the manuscript for this work. Guoshoujing Telescope (the Large Sky Area Multi-ObjectFiber Spectroscopic Telescope LAMOST) is a National Major Scientific Project built by the ChineseAcademy of Sciences. Funding for the project has been provided by the National Development andReform Commission. LAMOST is operated and managed by the National Astronomical Observatories,Chinese Academy of Sciences.This work is supported by the Strategic Priority Research Program ”The Emergence ofCosmological Structures” of the Chinese Academy of Sciences (No. XDB09000000), the NationalBasic Research Program of China (973 Program)(2015CB857004), and the National Natural ScienceFoundation of China (NSFC, Nos. 11225315, 1320101002, 11433005 and 11421303).
References
Abazajian, K., Adelman-McCarthy, J. K., Ag¨ueros, M. A., et al. 2004, AJ, 128, 502Alam, S., Albareti, F. D., Allende Prieto, C., et al. 2015, ApJS, 219, 12Allende Prieto, C., Lambert, D. L., & Asplund, M. 2001, ApJ, 556, L63Aller, L. H., ed. 1984, Physics of thermal gaseous nebulae,
Astrophysics and Space Science Library ,vol. 112Amor´ın, R., Sommariva, V., Castellano, M., et al. 2014, A&A, 568, L8Andrews, B. H., & Martini, P. 2013, ApJ, 765, 140Baldwin, A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 817Berg, D. A., Skillman, E. D., Marble, A. R., et al. 2012, ApJ, 754, 98Bianchi, L., Conti, A., & Shiao, B. 2014, Advances in Space Research, 53, 900Bothwell, M. S., Maiolino, R., Peng, Y., et al. 2016, MNRAS, 455, 1156Brinchmann, J., Charlot, S., White, S. D. M., et al. 2004, MNRAS, 351, 1151Brodie, J. P., & Huchra, J. P. 1991, ApJ, 379, 157Brorby, M., Kaaret, P., & Prestwich, A. 2014, MNRAS, 441, 2346Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000Calzetti, D., Armus, L., Bohlin, R. C., et al. 2000, ApJ, 533, 682Chabrier, G. 2003, PASP, 115, 763Cid Fernandes, R., Mateus, A., Sodr´e, L., Stasi´nska, G., & Gomes, J. M. 2005, MNRAS, 358, 363Cui, X.-Q., Zhao, Y.-H., Chu, Y.-Q., et al. 2012, Research in Astronomy and Astrophysics, 12, 1197Dopita, M. A., Kewley, L. J., Sutherland, R. S., & Nicholls, D. C. 2016, Ap&SS, 361, 61Doyle, M. T., Drinkwater, M. J., Rohde, D. J., et al. 2005, MNRAS, 361, 34Froese Fischer, C., & Tachiev, G. 2004, Atomic Data and Nuclear Data Tables, 87, 1Garnett, D. R. 1992, AJ, 103, 1330Garnett, D. R., & Shields, G. A. 1987, ApJ, 317, 82Hu, N., Su, S.-S., & Kong, X. 2016, Research in Astronomy and Astrophysics, 16, 006Hughes, T. M., Cortese, L., Boselli, A., Gavazzi, G., & Davies, J. I. 2013, A&A, 550, A115Hummer, D. G., & Storey, P. J. 1987, MNRAS, 224, 801Izotov, Y. I., Stasi´nska, G., Meynet, G., Guseva, N. G., & Thuan, T. X. 2006, A&A, 448, 955Izotov, Y. I., & Thuan, T. X. 2004, ApJ, 616, 768Izotov, Y. I., Thuan, T. X., & Guseva, N. G. 2012a, A&A, 546, A122Izotov, Y. I., Thuan, T. X., & Privon, G. 2012b, MNRAS, 427, 1229Kakazu, Y., Cowie, L. L., & Hu, E. M. 2007, ApJ, 668, 853Kashino, D., Renzini, A., Silverman, J. D., & Daddi, E. 2016, ApJ, 823, L24Kauffmann, G., Heckman, T. M., White, S. D. M., et al. 2003, MNRAS, 341, 33Kewley, L. J., Dopita, M. A., Sutherland, R. S., Heisler, C. A., & Trevena, J. 2001, ApJ, 556, 121Kisielius, R., Storey, P. J., Ferland, G. J., & Keenan, F. P. 2009, MNRAS, 397, 903Kniazev, A. Y., Grebel, E. K., Hao, L., et al. 2003, ApJ, 593, L73Kriek, M., van Dokkum, P. G., Labb´e, I., et al. 2009, ApJ, 700, 221Kunth, D., & ¨Ostlin, G. 2000, A&A Rev., 10, 1 sample of metal-poor galaxies in the LAMOST spectral survey 17
Lara-L´opez, M. A., Cepa, J., Bongiovanni, A., et al. 2010, A&A, 521, L53Lequeux, J., Peimbert, M., Rayo, J. F., Serrano, A., & Torres-Peimbert, S. 1979, A&A, 80, 155Lian, J., Hu, N., Fang, G., Ye, C., & Kong, X. 2016, ApJ, 819, 73Lian, J. H., Li, J. R., Yan, W., & Kong, X. 2015, MNRAS, 446, 1449L´opez-S´anchez, ´A. R., Dopita, M. A., Kewley, L. J., et al. 2012, MNRAS, 426, 2630Luo, A.-L., Zhang, H.-T., Zhao, Y.-H., et al. 2012, Research in Astronomy and Astrophysics, 12, 1243Luo, A.-L., Zhao, Y.-H., Zhao, G., et al. 2015, Research in Astronomy and Astrophysics, 15, 1095Luridiana, V., Morisset, C., & Shaw, R. A. 2015, A&A, 573, A42Ly, C., Malhotra, S., Malkan, M. A., et al. 2016a, ApJS, 226, 5Ly, C., Malkan, M. A., Nagao, T., et al. 2014, ApJ, 780, 122Ly, C., Malkan, M. A., Rigby, J. R., & Nagao, T. 2016b, ApJ, 828, 67Ly, C., Rigby, J. R., Cooper, M., & Yan, R. 2015, ApJ, 805, 45Maiolino, R., Nagao, T., Grazian, A., et al. 2008, A&A, 488, 463Mannucci, F., Cresci, G., Maiolino, R., Marconi, A., & Gnerucci, A. 2010, MNRAS, 408, 2115Markwardt, C. B., Swank, J. H., Barthelmy, S. D., et al. 2009, The Astronomer’s Telegram, 2258, 1Nicholls, D. C., Dopita, M. A., Sutherland, R. S., Kewley, L. J., & Palay, E. 2013, ApJS, 207, 21Nicholls, D. C., Dopita, M. A., Sutherland, R. S., et al. 2014, ApJ, 786, 155Pustilnik, S. A., Kniazev, A. Y., Masegosa, J., et al. 2002, A&A, 389, 779Pustilnik, S. A., & Martin, J.-M. 2007, A&A, 464, 859Salim, S., Lee, J. C., Ly, C., et al. 2014, ApJ, 797, 126Shaw, R. A., de La Pena, M. D., Katsanis, R. M., & Williams, R. E. 1998, in Astronomical Data AnalysisSoftware and Systems VII,