Lyman break and ultraviolet-selected galaxies at z ~ 1 - II. PACS 100μm/160μm FIR detections
Abstract
In this work, we report the Photodetector Array Camera and Spectrometer (PACS) 100 μm/160 μm detections of a sample of 42 GALEX-selected and far-infrared (FIR)-detected Lyman break galaxies (LBGs) at z ~ 1 located in the Cosmic Evolution Survey (COSMOS) field and analyse their ultraviolet (UV) to FIR properties. The detection of these LBGs in the FIR indicates that they have a dust content high enough so that its emission can be directly detected. According to a spectral energy distribution (SED) fitting with stellar population templates to their UV-to-near-IR observed photometry, PACS-detected LBGs tend to be bigger (Reff ~ 4.1 kpc), more massive [log (M*/M⊙) ~ 10.7], dustier [Es(B - V) ~ 0.40], redder in the UV continuum (β ~ -0.60) and UV-brighter [log (LUV/L⊙) ~ 10.1] than PACSundetected LBGs. PACS-detected LBGs at z ~ 1 are mostly disc-like galaxies and are located over the green valley and red sequence of the colour-magnitude diagram of galaxies at their redshift. By using their UV and IR emission, we find that PACS-detected LBGs tend to be less dusty and have slightly higher total star formation rates (SFRs) than other PACS-detected UV-selected galaxies within the same redshift range. As a consequence of the selection effect due to the depth of the FIR observations employed, all our PACS-detected LBGs have total IR luminosities, LIR, higher than 1011 L⊙ and thus are luminous IR galaxies. However, none of the PACS-detected LBGs are in the ultra-luminous IR galaxy (ULIRG) regime, LIR =1012 L⊙, where the FIR observations are complete. The finding of ULIRGs-LBGs at higher redshifts (z ~ 3) suggests an evolution of the FIR emission of LBGs with cosmic time. In an IRX-β diagram, PACS-detected LBGs at z~1 tend to be located around the relation for local starburst similarly to other UV-selected PACS-detected galaxies at the same redshift. Consequently, the dust-correction factors obtained with theirUVcontinuum slope allowus to determine their total SFR, unlike at higher redshifts. However, the dust attenuation derived from UV to NIR SED fitting overestimates the total SFR for most of our PACS-detected LBGs in an age-dependent way: the overestimation factor is higher in younger galaxies. This is likely due to the typical degeneracy between dust attenuation and age in the SED fitting with synthetic templates and highlights the importance of the FIR measurements in the analysis of star-forming galaxies at intermediate redshifts.
Full text
MNRAS 435, 158–186 (2013) doi:10.1093/mnras/stt1244 Advance Access publication 2013 August 16 Lyman break and ultraviolet-selected galaxies at z∼1 – II. PACS 100 µm/160 µm FIR detections I. Oteo,1,2†G. Magdis,3´ A. Bongiovanni,1,2,4 A. M. P´ erez-Garc´ ıa,1,2,4 J. Cepa,1,2 B. Cedr´ es,1,2 A. Ederoclite,5M. S´ anchez-Portal,4,6 J. A. L. Aguerri,1,2 E. J. Alfaro,7 B. Altieri,8P. Andreani,9,10 T. Aparicio-Villegas,7,11 H. Aussel,12 N. Ben´ ıtez,7 S. Berta,13 T. Broadhurst,14 J. Cabrera-Ca˜ no,15 F. J. Castander,16 M. Cervi˜ no,1,7 A. Cimatti,17 D. Cristobal-Hornillos,5,7 E. Daddi,18 D. Elbaz,18 A. Fernandez-Soto,19,20 N. F¨ orster Schreiber,13 R. Genzel,13 R. M. Gonzalez-Delgado,7C. Husillos,7L. Infante,21 E. Le Floc’h,18 D. Lutz,13 B. Magnelli,13 R. Maiolino,22 I. M´ arquez,7V. J. Mart´ ınez,23,24 J. Masegosa,7 I. Matute,7M. Moles,5A. Molino,7A. del Olmo,7J. Perea,7R. P´ erez-Mart´ ınez,25 I. Pintos-Castro,1,2,26 A. Poglitsch,13 J. Polednikova,1,2 P. Popesso,13 M. Povi´ c,7 F. Pozzi,17 F. Prada,7J. M. Quintana,7L. Riguccini,18 E. Sturm,13 L. Tacconi,13 I. Valtchanov6and K. Viironen5 1Instituto de Astrof´ ısica de Canarias (IAC), E-38200 La Laguna, Tenerife, Spain 2Departamento de Astrof´ ısica, Universidad de La Laguna (ULL), E-38205 La Laguna, Tenerife, Spain 3Department of Physics, University of Oxford, Keble Road, Oxford OX1 3RH, UK 4Asociaci´ on ASPID, Apartado de Correos 412, La Laguna, Tenerife, Spain 5Centro de Estudios de F´ ısica del Cosmos de Arag´ on, Plaza San Juan 1, Planta 2, E-44001 Teruel, Spain 6Herschel Science Centre (ESAC), Villafranca del Castillo, Spain 7Instituto de Astrof´ ısica de Andaluc´ ıa (CSIC), Glorieta de la Astronom´ ıa s/n, E-18008 Granada, Spain 8Herschel Science Centre, European Space Astronomy Centre, Villanueva de la Canada, E-28691 Madrid, Spain 9ESO, Karl-Schwarzschild-Str 2, D-85748 Garching, Germany 10INAF Osservatorio Astronomico di Trieste, Via Tiepolo 11, I-34143 Trieste, Italy 11Observatrio Nacional-MCT, Rua Jos Cristino 77, CEP 20921-400, Rio de Janeiro-RJ, Brazil 12Laboratoire AIM-Paris-Saclay, CEA/DSM/Irfu – CNRS – Universite Paris Diderot, CE-Saclay, F-91191 Gif-sur-Yvette, France 13Max-Planck-Institut f¨ ur extraterrestrische Physik, Postfach 1312, Giessenbachstraße 1, D-85741 Garching, Germany 14School of Physics and Astronomy, Tel Aviv University, Israel 15Departamento de F´ ısica At´ omica, Molecular y Nuclear, Facultad de F´ ısica, Universidad de Sevilla, Sevilla, Spain 16Institut de Ciencies de l’Espai, IEEC-CSIC, Barcelona, Spain 17Dipartimento di Astronomia, Universit di Bologna, Via Ranzani 1, I-40127 Bologna, Italy 18Laboratoire AIM, CEA/DSM-CNRS-Universit Paris Diderot, IRFU/Service d Astrophysique, Bt. 709, CEA-Saclay, 91191 Gif-sur-Yvette Cedex, France 19Instituto de F´ ısica de Cantabria (CSIC-UC), E-39005 Santander, Spain 20Unidad Asociada Observatorio Astronmico (Universitat de Val` encia/IFCA-CSIC), Parc Cientfic UV, E-46980 Paterna, Spain 21Departamento de Astronom´ ıa, Ponticia Universidad Catolica, Santiago, Chile 22Cavendish Laboratory, University of Cambridge, 19 J J Thomson Avenue, Cambridge CB3 0HE, UK 23Observatori Astronmic de la Universitat de Val` encia, Parc Cientfic UV, E-46980 Paterna, Spain 24Departament d’Astronomia i Astrofsica, Universitat de Val` encia, E-46100 Burjassot, Spain 25XMM/Newton Science Operations Centre (ESAC), Villafranca del Castillo, Spain 26Centro de Astrobiolog´ ıa, INTA-CSIC, PO Box – Apdo. de correos 78, Villanueva de la Canada Madrid E-28691, Spain Accepted 2013 July 5. Received 2013 July 3; in original form 2013 March 11 ABSTRACT In this work, we report the Photodetector Array Camera and Spectrometer (PACS) 100 µm/160 µm detections of a sample of 42 GALEX-selected and far-infrared (FIR)-detected Lyman break galaxies (LBGs) at z∼1 located in the Cosmic Evolution Survey (COSMOS) Herschel is an ESA space observatory with science instruments provided by European-led Principal Investigator consortia and with important participation from NASA. †E-mail: [email protected] C 2013 The Authors Published by Oxford University Press on behalf of the Royal Astronomical Society Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
Lyman break and UV-selected galaxies at z∼1159 field and analyse their ultraviolet (UV) to FIR properties. The detection of these LBGs in the FIR indicates that they have a dust content high enough so that its emission can be directly detected. According to a spectral energy distribution (SED) fitting with stellar population templates to their UV-to-near-IR observed photometry, PACS-detected LBGs tend to be bigger (Reff ∼4.1 kpc), more massive [log(M∗/M)∼10.7], dustier [Es(B−V)∼0.40], redder in the UV continuum (β∼−0.60) and UV-brighter [log (LUV/L)∼10.1] than PACS- undetected LBGs. PACS-detected LBGs at z∼1 are mostly disc-like galaxies and are located over the green valley and red sequence of the colour–magnitude diagram of galaxies at their redshift. By using their UV and IR emission, we find that PACS-detected LBGs tend to be less dusty and have slightly higher total star formation rates (SFRs) than other PACS-detected UV-selected galaxies within the same redshift range. As a consequence of the selection effect due to the depth of the FIR observations employed, all our PACS-detected LBGs have total IR luminosities, LIR, higher than 1011 Land thus are luminous IR galaxies. However, none of the PACS-detected LBGs are in the ultra-luminous IR galaxy (ULIRG) regime, LIR ≥1012 L, where the FIR observations are complete. The finding of ULIRGs-LBGs at higher redshifts (z∼3) suggests an evolution of the FIR emission of LBGs with cosmic time. In an IRX–β diagram, PACS-detected LBGs at z∼1 tend to be located around the relation for local starburst similarly to other UV-selected PACS-detected galaxies at the same redshift. Consequently, the dust-correctionfactorsobtainedwiththeir UV continuumslopeallowusto determine theirtotal SFR, unlike at higher redshifts. However, the dust attenuation derived from UV to NIR SED fitting overestimates the total SFR for most of our PACS-detected LBGs in an age-dependent way: the overestimation factor is higher in younger galaxies. This is likely due to the typical degeneracy between dust attenuation and age in the SED fitting with synthetic templates and highlights the importance of the FIR measurements in the analysis of star-forming galaxies at intermediate redshifts. Key words: galaxies: evolution– galaxies: high-redshift –galaxies: star formation– infrared: galaxies – radio continuum: galaxies – ultraviolet: galaxies. 1 INTRODUCTION There are some methods which have been traditionally used to look for star-forming (SF) high-redshift galaxies. Among them, one of the most employed and successful is the Lyman break or dropout technique, which segregates the so-called Lyman break galaxies (LBGs). The technique is based on sampling the Lyman break of galaxies with two broad-band filters close in wavelength, each one located on one side of the break. Since the location in wavelength of the Lyman break depends upon redshift, different combinations of filters pick up galaxies at different redshifts. Many samples of LBGs have been found following that procedure, mainly at z3by employing optical filters in ground-based telescopes (Madau et al. 1996; Steidel et al. 1996, 1999, 2003; Iwata et al. 2007; Verma et al. 2007; Yabe et al. 2009), but also at z∼2 (Ly et al. 2009, 2011; Hathi et al. 2010; Basu-Zych et al. 2011; Haberzettl et al. 2012), and at lower redshifts (Burgarella et al. 2006, 2007; Barger, Cowie & Wang 2008; Chen et al. 2013; Oteo et al. 2013a). In order to accurately study the evolution and the bolometric emission of LBGs, it is essential to have information of the light that is absorbed by dust in the UV and re-emitted in turn in the far-infrared (FIR). With a careful analysis of the UV slope, coupled with appropriate dust attenuation models, Vijh, Witt & Gordon (2003) identify some of the most heavily attenuated specimens in LBGs at z∼2−4. Furthermore, Finkelstein et al. (2009) find that LBGs at z∼4 can be detectable with Atacama Large Millimeter/submillimeter Array (ALMA) after a few hours of integration. However, until recently, a few high-redshift LBGs had been individually detected in FIR (Chapman et al. 2000; Chapman & Casey 2009; Siana et al. 2009; Rigopoulou et al. 2010) and, therefore, a complete and accurate study at different redshifts has not been possible yet. With the advent of the ESA’s Herschel space observatory (Pilbratt et al. 2010) and its Photodetector Array Camera and Spectrometer (PACS; Poglitsch et al. 2010) and Spectral and Photometric Imaging REceiver (SPIRE; Griffin et al. 2010) instruments, we are in possession of deep FIR data that enable us to study the FIR spectral energy distribution (SED) around the dust emission peak of LBGs as never before. Magdis et al. (2010c) studied for the first time using PACS data the FIR SED of infrared luminous LBGs at z∼3. Although none of their galaxies are individually detected with Herschel, a stacking analysis suggests a median IR luminosity of LIR =1.6 ×1012 L. Additionally, complementing their study with multiwavelength data they obtained that for their IR luminosity, MIPS-LBGs are warmer than submillimetre luminous galaxies while they fall in the locus of the LIR−Tdrelation of the local ultra-luminous IR galaxies (ULIRGs). Also with a stacking analysis, Lee et al. (2012) study the FIR emission of LBGs at 3.3 z4.3 obtaining that their IR-to-UV luminosity ratio (LIR/LUV) is low compared to that observed for z∼2 LBGs. Furthermore, they obtain that the correlation between the UV continuum slope and the LIR/LUV ratio suggests dust properties similar to those of local starbursts (SB) except for the galaxies with the highest luminosities. Davies et al. (2013) employed stacked detections in SPIRE bands of LBGs at z∼3 to find that a significant fraction of the star formation in these systems is obscured and that the extinction-corrected star formation rate (SFR) does not exhibit Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
160 I. Oteo et al. the large values that would be predicted by their UV continuum slope. The latter indicates that the method of assuming an intrinsic UV slope and correcting for dust attenuation may be invalid for this sample and that these are not in fact the most actively SF systems. Burgarella et al. (2011) report the first SPIRE 250µmand SPIRE 350µm FIR detections of LBGs by employing data from the Herschel Multi-tiered Extragalactic Survey (HerMES) project (Oliver et al. 2010). They find SPIRE detections for 12 LBGs at 0.7 z1.6 and one at z∼2.0. All these galaxies are high-mass, luminous IR galaxies (LIRGs), and have redder NUV−Uand U−R colours than other SPIRE-undetected galaxies. Oteo et al. (2013b) find 16 LBGs at z∼3 which are individually detected in PACS under the depth of the observations carried out in the frame of the GOODS-Herschel survey (Elbaz et al. 2011). They found that PACS-detected LBGs at z∼3 are ULIRGs or hyper-luminous IR galaxies and that the dust attenuation factor obtained from the SED- derived dust attenuation with Bruzual & Charlot (2003) templates or from the UV continuum slope cannot reproduce the total SFR of these galaxies. In this study, we continue the work started in Oteo et al. (2013a) about the physical properties of LBGs and other UV-selected galaxies at 0.8 ≤z≤1.2 in the Cosmic Evolution Survey (COSMOS; Scoville et al. 2007) field. In that work, we analyse the stellar populations of a sample of GALEX-selected LBGs by combining UV GALEX measurements and the optical-to-near-IR data coming from the Advanced Large, Homogeneous Area Medium Band Redshift Astronomical (ALHAMBRA) survey (Moles et al. 2008). This combination provides an excellent photometric coverage of the UV-to-near-IR SED of those galaxies. As result, it was found that LBGs at z∼1 are mostly young galaxies, with a median age of 341Myr, and have intermediate dust attenuation, Es(B−V)∼0.2. Due to their selection criteria they are UV-bright galaxies and have high SFRs. The median value of the stellar mass of the LBGs in the sample is log M∗/M=9.74. Now, we enlarge the photometric coverage of the SEDs of those galaxies by employing IRAC 5.8µm, IRAC 8.0 µm, MIPS 24 µm and PACS 100 µm/PACS 160 µm data. Actually, one of the advantages of dealing with galaxies at z∼1 is that their typical observed FIR fluxes make a significant number of objects detectable under the depth of PACS observations used in this work (see Section 2). This allows us to carry out the study of their physical properties avoiding the uncertainties of the opticalbased methods that do not take into account the dust emission in the FIR, like the SED fitting with Bruzual & Charlot (2003) templates. The combination of GALEX, ALHAMBRA, IRAC, MIPS and PACS data provides a photometric coverage of the SEDs of galaxies which has not been reached with previous data sets. The paper is structured as follows: in Section 2 we define the selection criteria employed to look for our SF galaxies at z∼1, explain how their SEDs were fitted with synthetic templates and report their FIR counterparts. In Section 3, we study the dust attenuation of our PACS-detected galaxies as a function of the rest-frame UV, total IR, and bolometric luminosity and UV continuum slope. Section 4 shows the study of the SFR and the dust-correction factors. We analyse the relation between the stellar mass and the total and specific SFR (sSFR) in Section 5. The location of our PACS-detected galaxies in a colour–magnitude diagram (CMD) is presented in Section 6, and their morphology is examined in Section 7. The radio emission of our SF galaxies at z∼1 is shown in Section 8. Finally, we summarize the main conclusions of the work in Section 9. Throughout this paper, we assume a flat universe with (m, ,h 0)=(0.3,0.7,0.7), and all magnitudes are listed in the AB system (Oke & Gunn 1983). 2 UV-SELECTED GALAXIES AND THEIR FIR DETECTIONS In Oteo et al. (2013a), we selected those LBGs whose Lyman break is located between the GALEX FUV and NUV filters, which are centred at the effective wavelengths of 1528 and 2271Å and have bandwidths of 1344–1786 and 1771–2831Å, respectively. The LBGs were selected by imposing the colour criterion FUV −NUV >1.5.(1) Additionally, and in order to rule out as much interlopers as possible, we limit the photometric redshift range of the sources to 0.8 ≤z≤1.2. It should be noted that the combination of GALEX and ALHAMBRA data provides accurate photometric redshifts at z∼1 (Oteo et al. 2013a). Those galaxies that are undetected in the FUV channel due to the presence of a strong Lyman break were also included in the sample. FUV observations in the COSMOS field have a limiting magnitude of FUV ∼27 mag (Oteo et al. 2013a). Then, galaxies brighter than that value in the wavelength range covered by the FUV filters should be detected. Since we select galaxies with an FUV −NUV colour cut of 1.5, that limiting magnitude would imply a limit of 25.5mag in the NUV channel. To be conservative, we choose a limit in the NUV channel of 25 mag andinclude in the previous sample of LBGs those galaxies that are brighter than 25mag in the NUV channel, are at 0.8 ≤z≤1.2 and are undetected in the FUV channel. The non-detection in the FUV channel under those restrictions is likely due to a strong Lyman break. AGN contamination was ruled out by discarding galaxies with X-ray detections (Elvis et al. 2009; Salvato et al. 2011). In total, 1225 GALEX-selected LBGs are segregated, with a median redshift of 0.93. The photometric redshift and other properties such as age, dust attenuation, stellar mass or UV continuum slope were derived by fitting(Bruzual& Charlot2003,hereafter BC03)templates totheirobserved GALEX+ALHAMBRA fluxes. These templates were built byassuming aSalpeter (1955)initialmassfunctiondistributingstars from 0.1 to 100M, a fixed value of the metallicity (Z=0.2 Z) and a time-independent star formation history (SFH). We considered a fixed value of metallicity (see also Gonz´ alez et al. 2011; Stark et al. 2013) because it is a parameter difficult to constrain with an SED-fitting procedure (see for example de Barros, Schaerer & Stark 2012; Oteo et al. 2012a). Actually, by running the Zurich Extragalactic Bayesian Redshift Analyzer (ZEBRA) with BC03 templates associated with different metallicities, we obtained in Oteo et al. (2013a) very similar values of the χ2 rfor individual galaxies, preventing us from constraining the metallicity of our galaxies. The changes in the SED-derived parameters associated with fits with different metallicities are not strongly significant, but they are within their typical uncertainties. The choice of the constant SFR is justified for the comparison with the results obtained with the Kennicutt (1998) calibrations. Therefore, the SED-derived parameters with BC03 templates presented in this work should be understood as those obtained with the previous assumptions for metallicity and SFH. We considered values of age from 1Myr to 7 Gyr, in steps of 10 Myr from 1Myr to 1 Gyr and in steps of 100Myr from 1 Gyr to 7 Gyr. Dust attenuation is included in the templates via the Calzetti et al. (2000) law and parametrized through the colour excess in the stellar continuum, Es(B−V), for which we select values ranging from 0 to 0.7 in steps of 0.05. We also include intergalactic medium absorption adopting the prescription of Madau (1995). The fits are carried out by using the ZEBRA (Feldmann et al. 2006) code which, in its maximum-likelihood mode, employs a χ2minimization Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
Lyman break and UV-selected galaxies at z∼1161 algorithm over the templates to find the one which fits the observed SED of each input object best. We have not included the contribution of emission lines in the SED fits (Zackrisson, Bergvall & Leitet 2008; Schaerer & de Barros 2009, 2010; Atek et al. 2011; Schaerer, de Barros & Stark 2011; de Barros et al. 2012; Schaerer, de Barros & Sklias 2013) since we will deal with dusty FIR-detected galaxies and, according to de Barros et al. (2012), the contribution of emission lines in their SED-derived parameters is not expected to be significant. Each fitted template has a characteristic value of the χ2. We define the reduced χ2,χ2 r, as the ratio between the χ2asso- ciated with each fitted template and the numbers of filters minus one employed in the SED-fitting procedure, χ2 r=χ2/(N−1) (see for example de Barros et al. 2012). Only galaxies with χ2 r<10 were includedinthe sample.Theuncertainties in theSED-derivedparam- eterswerederivedasthe differencebetween the best-fittedvalue and its corresponding weighted average, defined as WA =N iPifi/N, where Piis the probability that a given template, i, can represent the observed SED of a given galaxy, fiis the value of one of the physical properties associated with the ith template and Nis the number of templates (see Oteo et al. 2013a for more details). In the case of the UV continuum slope, the uncertainties considered here are those rising from the fit of the UV continuum to the exponential law. In Oteo et al. (2013a), we obtained that the photometric redshifts derived with the combination of GALEX and ALHAMBRA data are very accurate, σz=(|zphot −zspec|)/(1 +zspec)<0.05, for most of our galaxies. As in Oteo et al. (2013a), we adopt the spectroscopic redshifts for those galaxies with available optical spectroscopy in the zCOSMOS survey (Lilly et al. 2007) to redo the SED fits. For those galaxies without zCOSMOS counterpart, we employ the photometric redshift and their associated SED-fitting results. Once each SED is fitted with a BC03 template, the UV continuum slope is obtained for each galaxy directly from its best-fitted template by fitting a power law of the form fλ∼λβ(Calzetti, Kinney & Storchi- Bergmann 1994) to their UV continuum. In this process, we employ the photometric points corresponding to the rest-frame wavelength range 1300 λ3000 Å. This approach has the advantage of using all the available fluxes of each source, resulting in higher SNR determinations. The rest-frame UV luminosity, defined in a νLνway, for each source is obtained from its best-fitted template shifted to the rest frame and by convolving it with a top-hat filter, 300 Å width, centred in 1500 Å. With the aim of placing LBGs in a general scenario of FIR- detected SF galaxies at 0.8 <z<1.2, we select photometrically a sample of UV-detected galaxies in the same redshift range as the GALEX-selected LBGs: we consider all the galaxies detected in GALEX (at least in the NUV band) that have an ALHAMBRA counterpart, are detected in PACS 100 µm or PACS 160 µm, and their photometric redshifts are within 0.8 ≤z≤1.2 It should be pointed out that LBGs are contained in this general sample of UV-selected galaxies. The subsample of UV-selected and PACS-detected galaxies which are not LBGs will be called UV-faint galaxies. We emphasize that we consider LBGs and UV-faint galaxies independently because the main goal of this work is to analyse the FIR SED of the intermediate-redshift analogues of the colour-selected high-redshift LBGs. Due to the selection criteria, the main difference between LBGs and UV-selected galaxies is their rest-frame UV luminosity. The selection of LBGs in the FUV channel or the condition NUV <25 for those FUV-undetected implies that GALEX-selected LBGs are intrinsically brighter in the rest-frame UV. The general sample of UV-selected galaxies contains galaxies fainter in the rest-frame UV Table 1. Summary of the stacking results for the different bins of rest-frame UV luminosity, stellar mass and age (see the text for more details). Galaxy PACS 100 µm PACS 160 µm Both PACS bands LBGs 13 11 18 UV-faint 23 14 28 than the LBGs because their only restriction is the photometric redshift and the detection in the NUV and FIR (regardless of the FUV and NUV magnitude). The PACS FIR data employed in this work are taken within the frameworkof the PACS EvolutionaryProbe (PEP) Herschel guaranteed time key project, optimized for FIR studies of galaxy evolution (Lutz et al. 2011). As part of the PEP project, the COSMOS field was targeted in PACS 100 µm and PACS 160 µm. The PACS fluxes used in this study were extracted by using MIPS 24µm priors (Le Floc’h et al. 2009; Magnelli et al. 2009). The usage of MIPS 24 µm priors is justified because the 24 µm emission is strongly correlated to the FIR emission and, therefore, the catalogues created will not contain a large excess of sources without FIR counterparts. This substantially avoids deblending FIR sources into several unrealistic counterparts,as itmighthaveoccurred if,for example,optical-based catalogue with very high source density had been employed. The catalogues were extracted by using a point spread function (PSF) fitting method. The PSFs were directly derived from the science maps when performing a blind source extraction with the code STARFINDER (Diolaiti et al. 2000a,b) and were cut to a 3 pixel radius to extract the catalogues in order to minimize filtering effects on the wings of bright sources. The aperture corrections were determined by independent observations of very bright objects, such as the asteroid Vesta, by simply accounting for the limited radius in the observed PSFs. More details of the reduction and source extraction can be found in Lutz et al. (2011) and references therein. The limiting 3σ fluxes are 5.0 and 11.0 mJy in the PACS 100 µm and PACS 160 µm bands, respectively. We look for the possible PACS detections of our UV-selected galaxies by using their optical ALHAMBRA positions as reference and considering a matching radius of 2arcsec. It should be noted that since we are using PACS catalogues built with MIPS 24µm priors, the coordinates of the sources detected by PACS are those corresponding to the MIPS 24 µm observations. We find that 42 LBGs and 65 UV-faint galaxies are individually detected either in PACS 100 µm or PACS 160µm. We call those galaxies which are detected in, at least, one PACS band as PACS-detected galaxies. Table 1 summarizes the detections in each PACS band for each kind of PACS-detected galaxy. By inspecting the ACS I-band (F814W) images of the PACS-detected galaxies (see Section 7), we do not find evidence that the FIR fluxes of our sources are contaminated by close neighbours. In Fig. 1, we show the optical ACS I-band cutouts along with the IRAC 3.6 µm and MIPS 24 µm contours of two PACS-detected LBGs. With the aim of studying the possible effect of source confusion and analysing how good are the MIPS-based coordinates (employed as priors) and the optical ones, we represent one of the best favourable (big galaxy with no close companions) and the least favourable cases (where it can be thought that there is source confusion with close neighbours). It can be seen that in the two cases, the MIPS 24µm emission (and therefore the PACS emission) truly comes from the PACS-detected LBG and, thus, there is no evidence of source confusion. In Oteo et al. (2013a), we found that the SED fitting with ALHAMBRA data in the optical and Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
162 I. Oteo et al. Figure 1. ACS I-band images along with the IRAC 3.6 µm (blue) and MIPS 24 µm (red) contours for two PACS-detected LBGs. The PACS-detected LBGs are the galaxies located in the centre of each image. To show how good are MIPS coordinates with respect to optical data, we have selected for the plot one of the best cases and one of the worst cases of FIR identifications. Left: favourable case where the PACS-detected LBG is a big galaxy without close neighbours. Right: unfavourable case where the PACS-detected galaxy is small and it is surrounded by some close neighbours. In the two cases, it is clear that the MIPS 24µm emission (and therefore the PACS emission) truly comes from the PACS-detected LBG. Figure 2. Distribution of the ALH-613 band magnitude for our PACS- detected LBGs (red filled histogram) and PACS-detected UV-selected galaxies (green dashed histogram). The vertical dashed line represents a magnitude of 24.5mag and indicates the approximate minimum brightness limit down to which the SED fitting in the UV, optical and near-IR with ALHAMBRA data can be considered reliable. near-IR could be considered reliable only for galaxies brighter than 24.5mag in the ALH-613 band. Fainter galaxies in the optical have low values of χ2 rlikely due to high photometric uncertainties rather thantotrulygoodfitting.InFig.2,werepresenttheopticalALH-613 magnitude of our PACS-detected galaxies. It can be seen that most of the galaxies are brighter than the previous value and, therefore, the values of χ2 rare likely due to a good SED fitting rather than high photometric uncertainties. Additionally, Fig. 2 indicates that PACS-detected LBGs are typically brighter (mALH-613 ∼22.4) than UV-faint galaxies (mALH-613 ∼23.2). This is a direct consequence of their selection criteria. Furthermore, PACS-detected LBGs are among the optical brightest galaxies in the whole sample of GALEX- selected LBGs presented in Oteo et al. (2013a). Summarizing, we define the samples of PACS-detected LBGs, UV-selected and UV-faint galaxies as {LBGs :}={GALEX +ALHAMBRA : (i) FUV −NUV >1.5or NUV <25 and FUV −undetected (ii) 0.8<z<1.2 (iii) non −AGNs (iv) PACS −detected} {UV −selected :}={GALEX +ALHAMBRA : (i) NUV-detected with no restriction on FUV (ii) 0.8<z<1.2 (iii) non −AGNs (iv) PACS −detected} {UV-faint}={UV-selected}−{LBGs}. Fig. 4 shows the UV-to-FIR SED of nine PACS-detected LBGs that represent the typical SED-fitting results for our galaxies. It can be seen that the combination of GALEX, ALHAMBRA, IRAC, MIPS and PACS data provides an excellent photometric coverage of the SEDs of the studied galaxies. Mid- and far-IR detection rates of UV-selected galaxies vary with basic properties of the galaxies. Because of the larger detection rate in the mid-IR compared to the far-IR we discuss mid-IR detection rates, but trends in the FIR are consistent. Fig. 3 gives the detection rate in MIPS 24 µm of the whole sample of UV-selected galaxies at 0.8 <z<1.2 built in Oteo et al. (2013a), as a function of their SED- derived rest-frame UV luminosity, UV continuum slope, age, dust attenuation and stellar mass. MIPS 24 µm detections were searched by using a matching radius of 2arcsec and using the data coming from the S-COSMOS survey (Sanders et al. 2007). First, it can be seen that the detection rate in MIPS 24 µmincreases with the rest-frame UV luminosity, i.e. UV-bright galaxies are more likely detected in the FIR. This trend has been also reported in other works which analyse the IR properties of Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
Lyman break and UV-selected galaxies at z∼1163 Figure 3. Undetected fraction in MIPS 24µm of the whole population of PACS-detected LBGs and UV-faint galaxies at z∼1 as a function of their rest-frame UV luminosity, UV continuum slope, age, dust attenuation and stellar mass. The error bars have been obtained assuming Poisson statistic. The wider error bars in the points associated with the UV-brightest, oldest, dustiest and most massive galaxies are due to a low number of galaxies in those specific bins. UV-selected galaxies at lower (see for example Oteo et al. 2012b) and higher redshifts (see for example Reddy et al. 2010). Furthermore, the SPIRE-detected LBGs reported in Burgarella et al. (2011) are UV-bright galaxies, with an average value of logLUV/L=10.7, most of them being UV-luminous galaxies. Other clear tendencies between the detection rate in the FIR and the SED-derived physical properties of galaxies are those related to the UV continuum slope, dust attenuation and stellar mass. Fig. 3 indicates that galaxies which have redder UV continuum and are dustier and more massive are more likely detected in the FIR. We do not find any significant relation between the MIPS 24µm detection rate and the age of the galaxies. 2.1 Rest-frame UV, IR and bolometric luminosities As indicated before, the rest-frame UV luminosities, LUV, for the galaxies studied in this work are obtained from the BC03 templates which fit the observed UV-to-near-IR SED of each galaxy best by shifting them to the rest-frame and convolving them with a top-hat filter centred in 1500 Å with 100Å width. Throughout this study, we consider that the rest-frame UV luminosities indicated as LUV are expressed in νLνunits, whereas the rest-frame UV luminosities indicated as L1500 are expressed in Lνunits. With the aim of obtaining the total IR luminosities, LIR[8−1000µm], of our PACS-detected galaxies, we carry out SED fits with ZEBRA to their observed IRAC 8.0µm, MIPS 24µm and PACS 100 µm/PACS 160 µm fluxes with (Chary & Elbaz 2001, hereafter CE01) templates. Once the best templates are found, they are shifted to the rest-frame and then integrated between 8 and 1000µm to obtain LIR. In this way we avoid applying the relation between LIR and SED shape, known to fail at high redshift (Elbaz et al. 2010, 2011). We have computed the uncertainties in the total IR luminosities by modifying the IRAC/MIPS/PACS fluxes randomly within their uncertainties and obtaining the new LIR with the same procedure above. We repeat this 1000 times and find that the typical deviations in log (LIR/L) are within 0.1 dex. For comparison, we also fit the FIR SED of our PACS-detected galaxies (in the same bands) with (Dale & Helou 2002, hereafter DH02) templates. The typical uncertainties for the total IR luminosity obtained with DH02 templates are obtained in the same way as for those obtained with CE01 templates and have also a value of log(LIR/L)=0.1dex. The upper panel of Fig. 5 shows the relation between the total IR luminosities derived with CE01 and DH02 templates. It can be seen that DH02 templates tend to give higher values for our PACS-detected galaxies, although many of the points are within the uncertainties. In Fig. 4, nine examples of IR SED fittings can be found (red curves). In the bottom panels of Fig. 5, we compare the total IR luminosities derived with PACS (with both CE01 and DH02 templates) data with those obtained from single-band extrapolations with MIPS 24 µm data only (see Elbaz et al. 2010). It can be seen that MIPS 24 µm alone tends to slightly overestimate the total IR luminosities of our PACS-detected sources in only 0.1 dex typically when employing CE01 templates. For DH02 templates, the points are nearer the one-to-one relation. For both kinds of FIR templates and for most of the galaxies, the deviations are lower than 0.2dex. Therefore, there is good agreement between those quantities within the uncertainties of the total IR luminosities. This is only true for FIR- bright galaxies with PACS detections, but it does not have to be true for FIR-fainter sources detected in MIPS 24µmandPACS- undetected. There is no reason to choose between one of the two Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
164 I. Oteo et al. Figure 4. Rest-frame UV-to-FIR SED of nine randomly selected PACS-detected LBGs. These examples are representative of the SED-fitting results for our whole sample of our PACS-detected galaxies. The grey points denote the observed UV-to-FIR fluxes of the galaxies. The green curves represent the best-fitted BC03 templates to the UV-to-IRAC 4.5µm fluxes of each galaxy. The BC03 templates considered in the representations and fits are associated with time-independent SFH and a fixed metallicity of Z=0.2 Z(see Oteo et al. 2013a for more details). SED-derived redshift, age and dust attenuation associated with the represented BC03 templates are also indicated. The orange and red curves denote the best-fitted CE01 and DH02 templates, respectively, to the IRAC 5.8µm to PACS fluxes of each galaxy. sets of FIR templates. It should be remarked that the fact that the points are located closer to the one-to-one relation when using DH02 templates does not mean that these templates provide better determinations of the total IR luminosity since the MIPS 24µm single-band extrapolations do not have to apply for our galaxies. From now on, we decide to report the results obtained with CE01 templates. Finally, we define the bolometric luminosity, Lbol, of our galaxies as the sum of the UV and IR luminosities: Lbol =LUV +LIR (see for example Reddy et al. 2006). Fig. 6 shows the distribution of LUV,LIR and Lbol for our PACS- detected galaxies. As expected from their selection criteria and supported by the results of a Kolmog´ orov-Smirnov (K-S) test, LBGs tend to be UV-brighter than UV-faint galaxies. However, despite their difference in the UV selection criteria, the LIR span within a similar range. Their total IR luminosities place PACS-detected LBGs and UV-faint galaxies in the LIRG regime, i.e. they have 1011 <LIR/L<1012.Themedian valuesofLIR for PACS-detected LBGs and UV-faint galaxies are log(LIR/L)=11.7 and 11.6, respectively. It should be noted that the fact that all our PACS-detected galaxies have LIR >1011 Lis a consequence of the depths of the PACS observations in the COSMOS field, since they are not deep enough to recover galaxies with LIR <1011 L.However,wedonot find any PACS-detected LBG at 0.8 z1.2 in the ULIRG regime (1012 <LIR/L<1013) or with higher total IR luminosities, where the observations are complete. A population of ULIRG-LBG do appear at higher redshifts, suggesting evidence for an evolution of the FIR emission of LBGs (Oteo et al. 2013b). A similar behaviour has Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
Lyman break and UV-selected galaxies at z∼1165 Figure 5. Upper panel: comparison between the total IR luminosity for our PACS-detected LBGs (red dots) and UV-faint galaxies (green diamonds) with CE01 and DH02 templates. The blue bars indicate the typical uncertainties in the determinations of the total IR luminosity. Bottom panels: comparison between the MIPS 24 µm-derived and PACS-derived total IR luminosities for our PACS-detected LBGs (red dots) and UV-faint galaxies (green diamonds) with CE01 (left-hand panel) and DH02 (right-hand panel) templates. Only galaxies with available MIPS and PACS fluxes are included. The MIPS-based LIR are obtained by using single-band extrapolations from the observed MIPS 24 µmtoLIR employing the CE01 templates. The PACS-based total IR luminosities are derived by fitting CE01 templates to the observed IRAC 8.0 µm, MIPS 24 µm, PACS 100µm and PACS 160 µm fluxes of the sources. The straight line indicates where both determinations would agree and the dashed lines indicate deviations of ±0.2 dex around the one-to-one relation. been suggested for the other classical population of high-redshift SF galaxies: Lyαemitters (Oteo et al. 2011, 2012a,b). The distributions of the bolometric luminosities are similar to the IR ones, since the UV contribution to the bolometric luminosity is low for LIRGs. 2.2 Stacking analysis Many of our GALEX-selected LBGs are undetected in PACS 100µm/PACS 160 µm. In order to gain insight about the FIR properties of these PACS-undetected LBGs, we perform a stacking analysis in the PACS 160 µm band. We choose this band for the stacking because it is the closest one to the dust emission peak and, therefore, the determination of the FIR properties will be more accurate than with the PACS 100 µm band. We apply here the traditional stacking procedure previously adopted in some other works based on Herschel data (Magdis et al. 2010c; Rodighiero et al. 2010; Reddy et al. 2012). We extract 60 arcsec ×60arcsec images centred in the optical position of the PACS-undetected galaxies and then these cutouts are combined by using a median algorithm. We use the median image instead of the mean to ensure that the result is not very dependent on the presence of bright outliers. In the stacking we use the cutouts extracted from residual PACS 160 µm images, which are obtained from the science images by removing all the PACS 160 µm detected sources in the catalogue with SNR >3. We consider different samples to stack among the whole sample of PACS-undetected LBGs related to their different properties that will be studied in this work in the next sections. In this way, we arbitrarily split our sample of PACS-undetected LBGs into bins of Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
166 I. Oteo et al. Figure 6. Distribution of the rest-frame UV (left), IR (middle) and bolometric (right) luminosities for our PACS-detected LBGs (red shaded histograms) and the whole sample of LBGs and UV-faint galaxies (green dashed histograms). Table 2. Summary of the stacking results for the different bins of rest-frame UV luminosity, stellar mass and age considered in this work (see the text for more details). Bin Nf160µm(mJy) 9.75 ≤log(LUV/L)≤10.0 435 0.81 ±0.21 10.00 <log(LUV/L)≤10.25 338 0.59 ±0.24 10.25 <log(LUV/L)≤10.50 108 0.78 ±0.40 9≤log(M∗/M)≤10.0 612 0.48 ±0.13 10.0 ≤log(M∗/M)≤11.0 352 0.98 ±0.23 Age <100Myr 238 0.45 ±0.27 rest-frame UV luminosities, stellar mass and SED-derived age, as indicated in Table 2. We distinguish only one range of age with the aim of studying if the youngest galaxies in our sample behave differently from the rest of the population. The number of galaxies in each bin is also indicated in Table 2. In order to measure the uncertainties in the stacked images, we followa similarprocedure asthose employed,forexample,in Reddy et al. (2012) or Magdis et al. (2010c). We perform stacking in the PACS 160 µm residual map in Nrandom positions, where Nis the number of galaxies in each bin of rest-frame UV luminosity, stellar mass and age. We then measure the resultant fluxes in the stacked images and repeat this process 1000 times. It should be noted that the stacking in this case should be done with the residual images since otherwise there would be a high contribution of FIR emission of real PACS-detected sources. The width of the distribution of the stacked fluxes in each bin is considered as the uncertainty in the stacked PACS 160 µm flux in each case. The lower the number of galaxies in each bin, the higher is the uncertainty in the stacked flux. These uncertainties span from f160 µm=0.40 to 0.13mJy. The uncertainties in each bin are shown in Table 2. In addition to giving an estimation of the uncertainty of the stacking procedure, this process also provides information about the probability of obtaining significant stacked fluxes when stacking in random positions along the images and, therefore, representing fluxes not related to real sources. If the width of the distribution was as high as the stacked flux obtained in the real optical position of the LBGs, there would be a non-negligible probability of a chance measurement of stacked flux in a position which does not correspond to real sources and, therefore, the stacking procedure would not be reliable. Following this idea, we will consider that we obtain a stack detection if each stacked PACS 160 µm flux is higher than 3 ×f160 µm. As a first step, and in order to study the reliability of the stacking analysis, we perform a stacking in the PACS 160 µm band of all the GALEX-selected sources which are detected in MIPS 24 µm but undetected in PACS. At the redshift range of our galaxies, MIPS 24 µm single-band extrapolations to the total IR luminosities give reasonablygood results(Elbazetal.2010,2011).Therefore, inthis casewe can compare the MIPS-derived total IR luminosities with those obtained from the stacked PACS 160µm flux and study the agreement between the results. The PACS 160 µm stacked flux obtained for the MIPS-detected LBGs with the cutouts taken from the PACS 160 µm residual image is f160 µm=2.7±0.4mJy. In order to obtain its associated total IR luminosity, we employ a single-band extrapolation method from the PACS 160 µm flux to the total IR luminosity, obtaining a value of log (LIR/L)=11.0. The median value of the MIPS 24µm flux for the MIPS-detected/PACS-undetected LBGs is f24µm=0.2±0.1, where the indicated uncertainty corresponds to the standard deviation of the distribution of the MIPS 24µmfluxes. Applying again a single-band extrapolation, we obtain that its associated total IR luminosity is log (LIR/L)=11.3. This value is slightly higher than that obtained from stacking in PACS 160 µm but still compatible with the typical uncertainties of the single-band extrapolations for obtaining the total IR luminosity and the uncertainties of the MIPS 24 µm and PACS 160 µm fluxes of the stacked sources. For each of the bins, we applied the stacking procedure outlined above with the PACS 160 residual images. We only find stacked detections in the bin associated with the lowest rest-frame UV luminosity and the bin associated with the highest stellar mass. Fig. 7 shows the stacked PACS 160 µm images in those two ranges where we recover a stacked detection. The stacked fluxes are shown in Table 2. The total IR luminosities associated with the stacked PACS 160 µm fluxes are obtained by using single-band extrapolations (Elbaz et al. 2010). In the following analysis, the values of the rest-frame UV luminosities, stellar mass and age for each bin are represented by the median values for all the galaxies belonging to each one. 2.3 SED-derived properties of PACS-detected galaxies In Oteo et al. (2013a), we analysed the physical properties of GALEX-selected LBGs at 0.8 z1.2 by employing an SED- fitting procedure with BC03 templates. This allowed us to study their rest-frame UV luminosity, UV continuum slope, age, dust Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
Lyman break and UV-selected galaxies at z∼1173 Figure 14. Top: total SFR derived from the combination of UV and FIR measurements against the total SFR derived by correcting the rest-frame UV luminosity with the SED-derived Es(B−V) value obtained from SED fitting with BC03 templates associated with a constant SFR and constant sub-solar metallicity Z=0.4Z. PACS-detected LBGs and UV-selected galaxies are represented by red filled dots and green open triangles, respectively. The solid line shows the one-to-one relation. The median loci in the diagram of the PACS-detected LBGs and UV-selected galaxies are represented by the dark blue and light blue bars, respectively. The bars cross in the average value of each parameter for each kind of galaxy and the length of the bars is the standard deviationof the distribution of values. We denote with purple filled dots the point corresponding to the stacking in the three bins of rest-frame UV luminosity. The arrows indicate upper limits for the two ranges where no stacked detections are recovered. The blue error bars indicate the typical uncertainties (taken from Oteo et al. 2013a) of the parameters shown in the horizontal and vertical axes. Bottom left: the same as in the top panel but different points corresponding toLBGs and UV-faint galaxies with different values of the UV continuum slope, as indicated. The red and blue filled dots represent PACS-detected LBGs, while the green and orange open triangles represent the general population of UV-selected galaxies. Bottom right: the same as bottom left but different values of age, as indicated. The dashed straight line indicates a deviation of −0.6dex with respect to the one-to-one relation (solid line) and it is represented to guide the eye about the overestimation of the SED-derived total SFR for young galaxies. We plot in the bottom-right and left-hand panels the upper value corresponding to the stacked PACS 160µm flux of the galaxies younger than 100Myr with a purple filled dot. et al. (2000) law. We also represent the median loci of each kind of galaxy in the diagram. It can be seen that, although in average the SED-derived dust attenuation reproduces the UV+IR-derived total SFR with some underestimation, the dispersion around the one-to-one relation is relatively high. We also include in that panel the points associated with the stacking analysis in the three bins of rest-frame UV luminosity. It can be seen that there is good agreement between the SED-derived and UV+IR-derived total SFRs, although the PACS-stacked points associated with the highest values of the rest-frame UV luminosity are more deviated from the one- to-one relation, suggesting again that the SED-derived Es(B−V) gives overestimated results of the total SFR for UV-bright galaxies. This is opposite to what is found at z∼3, where the SED-derived Es(B−V) underestimates the total SFR for most PACS-detected Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
174 I. Oteo et al. Figure 15. Left: total SFR derived from the combination of UV and IR measurements against the total SFR obtained by correcting the rest-frame UV luminosity with the UV continuum slope and applying M99 relation. PACS-detected LBGs and UV-selected galaxies are represented with red filled dots and green open triangles, respectively. The median loci in the diagram of the PACS-detected LBGs and UV-selected galaxies are represented by the dark blueand light blue bars, respectively. The bars cross in the median value of each parameter for each kind of galaxy and the length of the bars is the standard deviation of the distribution of values. The solid line shows the one-to-one relation. PACS-stacked LBGs in the bins of rest-frame UV luminosity are represented as filled purple dots and the PACS-stacked LBGs in the age bin are indicated with a filled blue dot. The arrows indicate upper values where no stacked detection is found. The blue error bars indicate the typical uncertainties (taken from Oteo et al. 2013a) of the parameters shown in the horizontal and vertical axes. Right: the same as in the left-hand panel but considering different IRX relations: M99 in green, Overzier et al. (2011) in orange and T12 in purple. In this plot,and with the aim of clarifying the results, LBGs and UV-faint galaxies are considered as a whole population. LBGs. It should be remarked that the fact that the uncertainties in the SED-derived and UV+IR-derived total SFR are very similar does not mean, a priori, that the SED-derived is as good an estimator of the total SFR as the UV+IR-derived. The uncertainties in the SED-derived total SFR only reflect the uncertainties in the SED-derived dust attenuation and rest-frame UV luminosity but do not reflect the capacity of the SED-fitting procedure to reproduce the FIR emission of the galaxies. There are some physical reasons from that we expect that the UV+IR-derived total SFR is a good estimator of the real total SFR of a galaxy, but there is no physical reason to expect that it is possible to recover the total SFR for any kind of galaxy. This is only an expectation from the SED-fitting procedure employed, but it is something that needs to be checked. In the lower panels of Fig. 14 we further analyse the relation between SED-derived and UV+IR-derived total SFRs by dividing the sample of PACS-detected galaxies as a function of the UV continuum slope and the SED-derived age of the galaxies. It can be seen that the PACS-detected LBGs and UV-faint galaxies which are further the one-to-one relation are those with younger stellar populations. The stacked point corresponding to the galaxies with age younger than 100 Myr is also in agreement with the previous tendency. However, we do not find a significant correlation with the UV continuum slope. It should be noted that the tendency with age can be a consequence of the degeneracy between the dust attenuation and the stellar age in the SED-fitting procedures. The reddening of the SED of a galaxy can be caused either by the presence of an old stellar population or by the presence of high values of dust attenuation. A galaxy with a red SED can be young and dusty but also old and less dusty. In this way, the overestimation of the total SFR for young galaxies, produced by high values of the dust attenuation, can be due to the fact that the SED fitting cannot distinguish between old and young and dust-free and old galaxies. In the left-hand panel of Fig. 15, we represent the SED-derived dust-corrected total SFR obtained from the dust attenuation derived from the values of the UV continuum slope of each galaxy by employing the M99 law. The accuracy of this method is directly related to the location of our PACS-detected galaxies in the IRX–β diagram. When dust-correcting with this method, the SED-derived dust-corrected total SFR of PACS-detected LBGs and UV-selected galaxies is in very good agreement with the one derived from IR and UV measurements. The points associated with the PACS-stacked galaxies in the three bins of rest-frame UV luminosities are also very near to the one-to-one relation. Comparing the median loci of our PACS-detected galaxies in Figs 14 and 15, it can be seen that the dust attenuation derived from the UV continuum slope and the M99 relation gives much better estimations of the total SFR than that derived with the SED-derived Es(B−V). As will be explained in Section 8, this differential behaviour is also found when obtaining the total SFR with radio-VLA data instead of FIR ones. The right-hand panel of Fig. 15 shows evidence that the dustcorrectionfactorsobtainedwiththeM99laware alsoagedependent. Younger galaxies than 300 Myr tend to be located below the one- to-one relation, whereas older galaxies tend to be located above. This might be due to a correlation between age and the location of galaxies in the IRX–βdiagram (see also Reddy et al. 2010, 2012). We compare in Fig. 16 the dust-correction factors obtained after theapplicationof theM99and thenew relationsofT12 andOverzier et al. (2011, hereafter O11) for our PACS-detected galaxies. For clarity reasons, we consider in this plot our PACS-detected LBGs and UV-faint galaxies as a whole population of PACS-detected and UV-selected galaxies at z∼1. It can be seen that the IRX–βrelation that represents the best our PACS-detected galaxies is the M99 one. This is a direct consequence of the location of our galaxies in the IRX–βdiagram. However, our PACS-detected galaxies represent a Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
Lyman break and UV-selected galaxies at z∼1175 Figure 16. Relation between the UV+IR-derived total SFR and the dustcorrected total SFR by using the UV continuum slope and the application of three different IRX–βrelations: the green triangles correspond to the M99 law, the orange triangles to the O11 law and the purple triangles to the T12 law. The median loci of the galaxies in the diagram for each of the three laws are represented with bars of the same colour. The bars cross in the median value of the parameters presented in the x-andy-axes and the length of the bars is the standard deviation of the distribution of values. The solid line shows the one-to-one relation and the dashed lines represent deviations of ±0.5 and ±0.8dex with respect to the one-to-one relation. The blue bars represent the typical uncertainties in the derivation of the total SFR with each of the two methods. low percentage of the whole population of SF galaxies at z∼1. Additionally, for each UV slope, PACS only detects the dustiest galaxies. Therefore, an IRX–βrelation that gives lower values of the dust attenuation for each UV continuum slope (such as the O11 and T12 ones) might be more appropriate to recover the dust attenuation of SF galaxies at z∼1 from their UV continuum slope. 5 SFR VERSUS STELLAR MASS DIAGRAM In Oteo et al. (2013a), we study the location of the whole sample of GALEX-selected LBGs in an SFR versus stellar mass diagram (Daddi et al. 2007; Elbaz et al. 2007; Noeske et al. 2007; Salim et al. 2007; Pannella et al. 2009; Gonz´ alez et al. 2010; Rodighiero et al. 2010; Elbaz et al. 2011; Karim et al. 2011; Salmi et al. 2012; Sawicki 2012). It was found that, according to the definition of Elbaz et al. (2011) of the MS of galaxies at z∼1, most GALEX- selected LBGs are located in the SB region, giving clues to their SF activity. Additionally, it was found that more massive galaxies have higher total SFR and lower sSFR. In that case, we used SED- derived values of the dust-corrected total SFR, but not those derived from direct IR measurements. Now, we go back to the analysis of the location of our UV-selected galaxies at z∼1 in the SFR–mass diagram but focusing on the PACS-detected ones and, thus, utilizing the values of the total SFR derived from the combination of UV and IR measurements (equation 5). The left-hand panel of Fig. 17 represents the SFR−M∗for our PACS-detected LBGs and UV-faint galaxies. The SFRs which appear in the vertical axis of the plot are the total SFRs determined with the combination of UV and IR data (equation 5) and the stellar masses are those obtained by fitting the GALEX+ALHAMBRA photometry with BC03 templates built by assuming a constant SFR Figure 17. Left: location of our PACS-detected galaxies in an SFR–mass diagram. PACS-detected LBGs and UV-selected galaxies are represented with red filled dots and green open diamonds, respectively. The solid black line shows the MS of SF galaxies at z∼1 taken from Elbaz et al. (2011) and the dashed curve is for twice the values of that MS. The solid grey line shows the MS at z∼1 of Elbaz et al. (2007). The values of the dust-corrected total SFR adopted are those obtained from the UV continuum slope and the application of the M99 relation (see Section 4.1). The blue error bars indicate the typical uncertainties (taken from Oteo et al. 2013a) of the parameters shown in the horizontal and vertical axes. Right: specific star formation rate, sSFR =SFR/M∗as a function of the stellar mass for our PACS-detected LBGs (red filled dots) and UV-selected galaxies (green open diamonds). The purple filled dots represent the results of the stacking in PACS 160µm in two bins of stellar mass. The arrow represents the upper limit of the total IR luminosity in the bin associated with the lowest stellar masses, where no stacked PACS 160 µm detection is found. In both panels, for comparison, we also show the location in the SFR–mass and sSFR–mass diagrams of the PACS-undetected LBGs at z∼1 (grey open squares). The blue error bars indicate the typical uncertainties (taken from Oteo et al. 2013a) of the parameters shown in the horizontal and vertical axes. Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
176 I. Oteo et al. (Oteo et al. 2013a). Along with the data points, the SFR–mass relation for the MS of galaxies at z∼1 (Elbaz et al. 2011) is also represented. Elbaz et al. (2011) assume for their definition of MS a slope of the SFR–mass relation equal to 1 and, therefore, that the sSFR is mass independent at a fixed redshift. First, it can be seen that most PACS-detected galaxies at z∼1 are over the ‘MS’ of galaxies at their redshift and, therefore, theyare normal SF galaxies. Only the most massive galaxies are below the MS and the less massive are in the SB region. There is no significant relation between SFRtotal and stellar mass. In Oteo et al. (2013a), we reported that galaxies with higher masses tend to have higher SFRs. The reason why we do not see such a relation here is that in the present work we only analyse PACS-detected galaxies and, therefore a subsample of IR-bright, dusty and massive galaxies of the whole sample analysed in Oteo et al. (2013a). The range of IR luminosities of our galaxies does not allow us to constrain a general relation between SFRtotal and stellar mass. However, we do see that despite having a narrow range of values of SFRtotal, there is a large dispersion in the stellar mass values. This result is consistent with that found in Giovannoli et al. (2011) for their MIPS 24 µm selected LIRGs at ∼1. They also found that galaxies with stellar populations younger than 300 Myr exhibit a larger dispersion in the stellar mass than galaxies whose SFR is constant over a longer period. With the aim of gaining insights about the location of IR-faint galaxies in the SFR−M∗diagram, we also plot the two points associated with the stacking results of the galaxies located in the two bins of stellar mass defined in Section 2.2. It can be seen that the stacked point corresponding to the LBGs in the range of the highest stellar masses is right over the MS of galaxies. The upper limit of the total SFR for PACS-stacked LBGs in the range of the lowest values of stellar mass indicates that most of these galaxies might also be located over the MS. This indicates that most GALEX- selected LBGs, regardless of their detection in the FIR, tend to be located over the MS of galaxies at their redshift. We find (see the right-hand panel of Fig. 17) a trend between the sSFRandthestellarmass, whichis similartothat foundin Oteoetal. (2013a) with UV, optical and near-IR data and in some other works at different redshifts (Feulner et al. 2005a,b; Erb et al. 2006; Damen et al. 2009; Dunne et al. 2009; Rodighiero et al. 2010): galaxies with higher stellar masses tend to have lower sSFRs. This relation is characterized by its slope. In Oteo et al. (2013a), by fitting a linear relation of the form log(sSFR[Gyr−1]) =a+b×log(M∗[M]) to the whole population of GALEX-selected LBGs, it was found that a=5.20 ±0.25 and b=−0.49 ±0.03 when dust-correcting the rest-frame UV luminosity with the SED-derived Es(B−V) and a=3.95 ±0.17 and b=−0.38 ±0.02 when dust-correcting the rest-frame UV luminosity with the dust attenuation obtained from the UV continuum slope and the relation of M99. Carrying out the same linear fit to the PACS-detected LBGs studied in this work, we find a=10.1 ±0.8 and b=−0.95 ±0.07. For PACS- detected LBGs, the correlation between the sSFR and the stellar mass has much lower scatter and a higher slope than that for the whole population of LBGs due to the bias produced by the FIR selection. Actually, it can be seen in the right-hand panel of Fig. 17 that, for each stellar mass, PACS only sees the galaxies with the highest sSFRs. However, the sSFR for the most massive galaxies in the sample does not have values as high as the ones derived for the least massive galaxies. This agrees with the downsizing scenario where massive galaxies form their stars earlier and faster than lower massive ones. The absence of massive galaxies with sSFR values as high as those for the least massive ones is not a bias due to the FIR-selection criteria, since massive galaxies with high sSFR shouldhavebeendetected with PACS. We do notfindanysignificant difference in the trend between LBGs and UV-faint galaxies in the sSFR–mass diagram: all our PACS-detected galaxies have similar values of the sSFR. To further analyse the correlation between the sSFR and the stellar mass for IR-fainter LBGs, we also plot in the right-hand plot of Fig. 17 the two points associated with the stacked PACS 160 µm fluxes for the LBGS in the two stellar mass ranges defined in Section 2.2. In agreement with Fig. 8, the values of the sSFR for PACS-detected and PACS-stacked galaxies are within a similar range. The stacked points are deviated from the linear relation found for PACS-detected galaxies towards lower values of the sSFR. This indicates that PACS detections segregate galaxies with the highest sSFR for a given value of stellar mass and that the relation between the sSFR and the stellar mass becomes much wider when introducing IR-faint galaxies. 6 CMD OF THE PACS-DETECTED SOURCES In this section we study the location of our PACS-detected galaxies in a colour–mCMD. Such a diagram was already studied in Oteo et al. (2013a) for the whole population of LBGs at z∼1andit was found that most LBGs occupy the blue cloud of the galaxies at their redshift. Only the dustiest and/or oldest galaxies are distributed over the green valley or the red sequence. In Section 2, we observed that dustier sources are more likely detected in PACS. Therefore, it should be expected that PACS-detected galaxies are located mostly above the blue cloud. This is confirmed in Fig. 18, where LBGs and UV-faint galaxies are represented by red filled dots and green open triangles, respectively. As in Oteo et al. (2013a), the u−r are associated with the magnitudes in the uand rbroad-band filters of the Sloan Digital Sky Survey (SDSS). The apparent uand rand absolute rmagnitudes are obtained by convolving the best-fitted template of each galaxy with the transmission of the uand rSDSS filters shifted in wavelength according to the redshift of each source. ItcanbeseenthatPACS-detectedgalaxiesaremainlylocatedover the green valley or the red sequence at their redshift. Additionally, there is a difference between LBGs and UV-faint galaxies in the sense that LBGs tend to have bluer optical colours than UV-faint galaxies as a consequence of their lower dust attenuation. UV-faint galaxies are more attenuated by dust and are mostly located near the red sequence despite being actively SF galaxies according to their total SFR. We also plot in Fig. 18 the location of the PACS-undetected LBGs in order to analyse the difference between PACS-detected and PACS-undetected sources regarding their location in a CMD. Following the same idea as in Oteo et al. (2013a), we divide the sample of PACS-undetected LBGs into old LBGs, those whose ages are older than 1200 Myr, and young LBGs, those whose ages are younger than 1200 Myr. They are represented by filled and open grey dots, respectively. It can be seen that PACS-detected LBGs are located over a similar zone as old LBGs, which indicates that both old and/or dusty LBGs tend to depart from the blue cloud to the red sequence. This result is similar to that found in Oteo et al. (2013a). It should be noted that here, instead of dividing the sample depending on dust attenuation, we consider PACS-detected and PACS-undetected galaxies separately. Despite this, we still obtain the same result: PACS-detected and/or old LBGs tend to be located over the green valley or the red sequence rather than over the blue cloud. Those LBGs which are located over the blue cloud are only those young and less dusty. Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
Lyman break and UV-selected galaxies at z∼1177 Figure 18. Location of our PACS-detected galaxies in a CMD. The colour u−ris calculated in the rest-frame (see Oteo et al. 2013a). PACS-detected LBGs and UV-selected galaxies are represented with red filled dots and green open triangles, respectively. The grey filled dots represent PACS-undetected LBGs older than 1200Myr and the grey open dots represent PACS-undetected LBGs younger than 1200 Myr. The black contours represent the CMD of a general sample of galaxies z∼1 selected from the ALHAMBRA survey. Therefore, Fig. 18 shows a clear tendency in the CMD as a function of age and dust attenuation, from the less dusty and young galaxies located in the blue cloud to the UV-faint and dusty galaxies located over the red sequence. It also indicates that the properties of LBGs at z∼1 are diverse. The Lyman break technique seems to select galaxies with different properties but with specific characteristics. 7 MORPHOLOGY OF PACS-DETECTED LBGs This section is devoted to the study of the morphological properties ofourPACS-detectedgalaxiesbycarryingoutbothananalyticaland visualprocedure. Toperform aprecise study,high-spatial-resolution ACS images are mandatory. We employ here the publicly available ACS I-band images of the COSMOS field. At z∼1theACSIband samples the rest-frame Uband approximately, with little variation according to the redshift distribution of the sources. All our PACS- detected galaxies have available ACS information. On the visual side, we visually classify the PACS-detected galaxies into five groups, as was done in Oteo et al. (2013a): disclike galaxies, interacting/merging galaxies, compact galaxies, chain galaxies (CH) or clump clusters (CC), and irregular galaxies. We define disc galaxies as those having a clear disc structure, regardless of the presence of spiral arms or a central bulge. For CH and CC, we employ the definition given in Elmegreen et al. (2009a). Finally, we consider a galaxy as irregular if it cannot be classified within any of the other types. According to a visual classification of their ACS images (see Fig. 19), we obtain that most (68 per cent) PACS- detected LBGs have clear disc-like morphologies. Among them, 59 percent have flocculent structures (see for example Elmegreen et al. 2009b). For the remaining galaxies, 13per cent have signs of interaction, 7 per cent are irregular, 7 per cent are CC and 5 per cent belong to the CH class. On the analytical side, and with the aim of obtaining the physical sizes and examining the radial dependence of the light distribution of the PACS-detected galaxies, we carry out fits to their radial light curves with GALFIT (Peng et al. 2010). In this step, we consider S´ ersic profiles (Sersic 1968), which can be described as (r)=eexp −κr Reff 1/n −1,(6) where eis the pixel surface brightness at the effective radius Reff and nis the concentration parameter or the S´ ersic index. The effective radius is the radius which encloses half the light of the galaxy. To make this definition true, the dependent variable κis coupled to n(Peng et al. 2010). For each input galaxy, GALFIT provides the effective radius (in pixels) and the S´ ersic index. In order to convert the effective radius in pixels into the physical size in kpc, we employ the ACS pixel scale and the assumed cosmology for calculating the plate scale at the redshift of each galaxy. The left-hand panel of Fig. 20 represents the distributions of the effective radii of PACS-detected (red shaded histogram) and PACS-undetected LBGs (green histogram). PACS-detected LBGs are large galaxies with effective radii ranging from about 2 to 8 kpc, with a median value of 4 kpc. PACS-undetected LBGs have a median effective radius of Reff =2 kpc, indicating that FIR detections tend to segregate LBGs with large sizes. There is no significant difference in the physical sizebetweenPACS-detected LBGsandPACS-detectedUV-selected Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
178 I. Oteo et al. Figure 19. I-band optical ACS cutouts of the 42 PACS-detected LBGs. The size of each image is 5 arcsec ×5 arcsec. At z∼1, the plate scale is 8.008 kpcarcsec−1under the assumed cosmology, and the ACS Iband samples the rest-frame Uband approximately. Figure 20. Results of the analytical morphological study of our PACS-detected galaxies. Left: distribution of the physical sizes. Right: distribution of the S´ ersic indices. In both panels, PACS-detected LBGs, UV-selected galaxies and PACS-undetected LBGs are represented with red shaded, green and black dashed histograms, respectively. Histograms have been normalized to their maximum in order to clarify the representations. galaxies. Their S´ ersic indices of our PACS-detected LBGs and UV- selected galaxies, shown in the right-hand panel of Fig. 20, vary mostly within 0 n0.75, in agreement with the visual classification that indicates that most PACS-detected LBGs are disc-like galaxies. 8 RADIO MEASUREMENTS Dust absorbs the UV radiation created from massive stars and reradiates this light in the IR. When those massive stars explode as SNe, they generate cosmic ray electrons which lose their energy in radio wavelengths, mostly via synchrotron emission. Therefore, it is expected that the radio emission from galaxies is in part related to their emission in the IR. In this way, radio measurements also provide a powerful tool to obtain SFR. Following this idea, we analyse in this section the radio emission of the whole sample of LBGs and UV-faint galaxies built in Oteo et al. (2013a) in order to study a possible difference between them and to study the relation between IR- and radio-derived luminosities and SFRs. To this aim, we look for radio counterparts of the whole sample of LBGs and UV-faint galaxies selected in Oteo et al. (2013a) by using data coming from VLA observations of the COSMOS field (Schinnerer et al. 2010). Using again a matching radius of 2 arcsec, we find that 18 LBGs and 46 UV-faint galaxies are detected at 1.4GHz. These percentages of VLA detections are lower than those corresponding to PACS detections. In order to obtain the rest-frame 1.4 GHz luminosities, since our VLA-detected Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
Lyman break and UV-selected galaxies at z∼1179 Figure 21. Upper: distribution of the rest-frame 1.4GHz luminosities for our VLA-detected LBGs (red shaded histogram) and UV-selected galaxies (green histogram). Bottom: relation between the total IR and radio luminosities for our samples of PACS-detected LBGs (red filled dots) and UV-selected galaxies (green open diamonds). The straight line represents the relation of Condon (1992). The blue error bars represent the typical uncertainties of the parameters shown in the x- and y-axes. galaxies are located at z∼1, we must carry out a K-correction. To do that, we assume that the radio continuum behaves as fradio ∼να, where αis the radio spectral index, for which we consider a value of α=−0.8 (Ivison et al. 2010a,b; Magdis et al. 2010b). We plot in the upper panel of Fig. 21 the distribution of rest-frame 1.4GHz luminosities for our VLA-detected LBGs and UV-faint galaxies. The low number of detections in radio wavelength of both kinds of galaxies prevents us from distinguishing them according to their radio emission. Actually, both histograms are very similar. A K-S test yields a probability factor of 0.5, reinforcing the similarity between both distributions. Obtaining radio-derived SFR requires calculating the IR luminosities associated with the rest-frame 1.4 GHz luminosities by assuming an IR–radio correlation (for example Condon 1992). Then, we first compare in the bottom panel of Fig. 21 the total IR luminosities and the rest-frame 1.4 GHz luminosities for our VLA-detected and PACS-detected galaxies to check if the Condon (1992) relation applies for our galaxies. Nine LBGs and 27 UV-faint galaxies are detected both in PACS and VLA. We also plot there the Condon (1992) IR–radio correlation. It can be seen that PACS-detected UV- faint galaxies are around the Condon (1992) relation and, therefore, for those galaxies, the IR-derived and radio-derived SFRs with the procedure outlined above would agree. PACS-detected LBGs tend to be located slightly above the Condon (1992) relation. This could introduce some differences in the VLA-derived total IR luminosities with respect to those that would be obtained with PACS data. For those PACS-undetected and radio-detected galaxies, we can assume that the Condon (1992) relation is still valid with the aim of calculating the IR luminosity associated with their radio emission. It should be noted that, according to the IR–radio correlation, the SFRradio is a tracer of the SFRIR, but not of the SFRtotal. Therefore, in order to calculate the total SFR from radio observations, we also have to add the term associated with the UV emission (equation 5). In Fig. 22, we plot the distribution of the total SFR derived from the combination of UV and radio measurements for our VLA-detected galaxies. The median values are 67 ±65 and 76 ±80 Myr−1 for VLA-detected LBGs and UV-faint galaxies, respectively. The uncertainties indicate the standard deviations of the distributions, as a measurement of their width. The total SFRs obtained in Section 4 from the combination of UV and FIR measurements were 91 ±22 and 73 ±39 Myr−1for PACS-detected LBGs and UV- faint galaxies, respectively. The distributions of radio-derived total SFR are much wider than those obtained from PACS measurements. This is due to the existence of galaxies with very high values of the Figure 22. Distribution of the total SFR as derived for the combination of UV- and radio-derived IR luminosities for the VLA-detected LBGs (red shaded histogram) and UV-selected galaxies (green histogram). Note that given that we assume that the IR–radio correlation is valid for both IR- detected and IR-undetected sources, in this plot are included all radio detected sources, regardless of the fact that they are detected or not in PACS. Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
180 I. Oteo et al. Figure 23. Left: total SFR derived from the combination of UV and radio measurements against the total SFR derived by correcting the rest-frame UV luminosity with the Es(B−V) value obtained from SED fitting with BC03 templates associated with a constant SFR and sub-solar metallicity Z=0.4 Z. Right: total SFR derived from the combination of UV and radio measurements against the total SFR obtained by correcting the rest-frame UV luminosity with the UV continuum slope and applying the M99 relation. In both panels, VLA-detected LBGs and UV-selected galaxies are represented with red open dots and green open diamonds, respectively. The median loci in the diagrams of the PACS-detected LBGs and UV-selected galaxies are represented by the dark blueand light blue bars, respectively. The bars cross in the median value of each parameter for each kind of galaxy and the length of the bars is the standard deviation of the distribution of values. The lines represent where both estimators would agree. The grey error bars represent the typical uncertainties of the parameters showninthex- and y-axes. radio-derived total SFR compared to themedian of the distributions. Despite this, there is very good agreement between the radio and FIR determinations. Finally, we study in Fig. 23 the relation between the SED-derived dust-corrected total SFR and the total SFR derived from the combination of UV and radio measurements. These are similar plots as those presented in Figs 14 and 15. As in that case, we use the dust-corrected SED-derived total SFRs obtained from thebest-fitted Es(B−V) value and from the dust attenuation obtained with the UV continuum slope by applying the M99 relation. It can be seen that both procedures tend to give similar results, providing estimation of the total SFR within 0.8 dex for most of the galaxies. Furthermore, this spread around the one-to-one relation is similar to that found in Figs 14 and 15 when using PACS data to obtain the UV+IR-derived total SFR. 9 CONCLUSIONS In this work, they have analysed the UV-to-FIR SED of a sample of LBGs and other UV-selected galaxies at z∼1 located in the COSMOS field. To this aim, we have used data coming from GALEX, ALHAMBRA survey, IRAC, MIPS and PACS 100 µm/160µm observations. The combination of all these measurements provides an unprecedented coverage of the SED of galaxies at the studied redshift. The sample of LBGs was selected via their FUV −NUV colour. Due to their selection criteria, LBGs are UV-bright galaxies. For comparison, we also retain a sample of UV-selected and PACS-detected galaxies fainter in the UV than LBGs but located at the same redshift range. We call these sources UV-faint galaxies. The main conclusions of the work can be summarized as follows. (i) We have found that, among a sample of 1225 UV-selected LBGs at z∼1, 42 of them are individually detected in either PACS 100 µm or PACS 160 µm under depths of 5.0 and 11.0 mJy in each band, respectively. These measurements, in combination with ancillary IRAC 8.0 µm and MIPS 24 µm data, provide a good sampling of the mid-IR/FIR SED of these galaxies. This allows an accurate determination of their total IR luminosity, dust attenuation, and total and sSFR. PACS-detected LBGs tend to have higher restframe UV luminosities, dust attenuation and stellar mass, and have redder UV continuum than PACS-undetected galaxies. There is no relation in age between PACS-detected and PACS-undetected LBGs. (ii) All PACS-detected LBGs have total IR luminosities 1011 LIR/L1012 and,therefore, areallLIRGs. Thelower limit range of total IR luminosities is a direct consequence of the depth of the PACS observations employed in this work. However, there are no PACS-detected LBGs in the ULIRG regime, LIR/L≥1012, where the observations are complete. At z∼3, the PACS-detected LBGs found so far are all ULIRGs. This suggests that the FIR emission of galaxy is changing with cosmic time in the sense that at higher redshifts, there is a population of extreme IR-bright, red and dusty LBGs which is not found at intermediate redshifts. (iii) By using their UV and IR emissions and applying the Buat etal.(2005) calibration,wefind thatPACS-detectedLBGshavedust attenuation spanning 2 A1200 Å[mag] 5, with a median value of 4mag. PACS-detected LBGs, due to their brightness in the UV as a consequence of their selection criteria, are less dusty than other UV-fainter galaxies at the same redshift. PACS-detected LBGs with higher rest-frame UV luminosities are less attenuated. Furthermore, the dust attenuation of PACS-detected LBGs increases with the total IR luminosity. (iv) According to the combination of their UV and IR emission, PACS-detected LBGs have a median total SFR of 90 Myr−1. The SED-derived dust attenuation obtained with BC03 templates Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
Lyman break and UV-selected galaxies at z∼1181 associated with a constant SFR tends to overestimate the total SFR obtained from the combination of UV and IR measurements for the youngest galaxies in the sample. This is likely due to the typical degeneracy between age and dust attenuation in the SED-fitting procedures with synthetic templates and highlights the importance of the usage of direct UV and IR measurements when obtaining the total SFR and dust attenuation in galaxies. (v) The dust attenuation derived from rest-frame UV to NIR SED fitting overestimates the total SFR for most of our PACS-detected LBGs in an age-dependent way: the overestimation factor is higher for younger galaxies. This is likely due to the degeneracy in the SED-fitting procedures between dust attenuation, age and SFH, and highlights the importance of using direct FIR detections of galaxies to obtain accurate values of their IR luminosities, dust attenuation and total SFR. (vi) By analysing the location of our PACS-detected LBGs in an IRX–βdiagram, we find that they are mostly around the M99 relation for local SB, but above the corrections of T12 and O11 to the M99 law. This is also true for other UV-selected PACS-detected galaxies. Consequently, the dust-correction factors obtained with the M99 relation can be used to recover the dust attenuation of our PACS-detected galaxies. This does not have to be true for other less dusty and PACS-undetected galaxies, since PACS only detects the dustiest and IR-brightest galaxies at z∼1. (vii) In a CMD diagram, PACS-detected LBGs have red optical colour compatible with them being mostly located over the green valley or the red sequence despite being active SF galaxies. UV-faint galaxies are also located over the green valley and the red sequence, although there is a larger percentage of galaxies populating the red sequence than in the sample of PACS-detected LBGs. PACS- undetected LBGs with ages larger than 1200Myr tend to be also located over the green valley and the red sequence. This indicates that the properties of LBGs at z∼1 are diverse, from blue, young, PACS-undetected LBGs to old or PACS-detected red LBGs. (viii) Morphologically, ACS I-band (sampling the rest-frame U band at z∼1) imaging indicates that 68 per cent PACS-detected LBGs are disc-like galaxies, with 59per cent of them having flocculent structures, and 13 per cent have signs of interactions. There is a low percentage (<7 percent) of irregular, CH or CC galaxies. The median effective radius of the PACS-detected LBGs is 4 kpc. This value is higher than that for PACS-undetected LBGs, Reff =2 kpc, indicating that FIR detections tend to segregate galaxies with larger sizes. The S´ ersic indices of PACS-detected are compatible with them being disc galaxies. (ix) The radio measurements of our galaxies also provide accurate determinations of their total SFR. The IR/UV- and radio/UV- derivedtotalSFRsareinagreement dueto thevalidity ofthe Condon (1992) relation for most of our galaxies at z∼1. The dust correction obtained from the UV continuum slope and the application of the M99 relation gives much better results for the total SFR than the dust correction carried out with the SED-derived Es(B−V). ACKNOWLEDGEMENTS The authors would like to thank the referee for providing the motivating comments that have improved the presentations of our results. This research has been supported by the Spanish Ministerio de Econom´ ıa y Competitividad (MINECO) under the grant AYA2011- 29517-C03-01. AFS and VJM acknowledge the support from the Spanish Ministry project AYA2010-22111-C03-02 and the Generalitat Valenciana project PROMETEO-2008/132. Some/all of the data presented in this paper were obtained from the Multimission Archive at the Space Telescope Science Institute (MAST). STScI is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555. Support for MAST for non-HST data is provided by the NASA Office of Space Science via grant NNX09AF08G and by other grants and contracts. This work is based on observations made with the European Southern Observatory telescopes obtained from the ESO/ST-ECF Science Archive Facility. This work is also based on zCOSMOS observations carried out using the Very Large Telescope at the ESO Paranal Observatory under Programme ID: LP175.A-0839. Herschel is an ESA space observatory with science instruments provided by European-led Principal Investigator consortia and with important participation from NASA. The Herschel spacecraft was designed, built, tested and launched under a contract with ESA managed by the Herschel/Planck Project team by an industrial consortium under the overall responsibility of the prime contractor Thales Alenia Space (Cannes), including Astrium (Friedrichshafen) responsible for the payload module and for system testing at the spacecraft level, Thales Alenia Space (Turin) responsible for the service module and Astrium (Toulouse) responsible for the telescope, with excess of a hundred subcontractors. PACS has been developed by a consortium of institutes led by MPE (Germany), including UVIE (Austria); KUL, CSL, IMEC (Belgium); CEA, OAMP (France); MPIA (Germany); IFSI, OAP/AOT, OAA/CAISMI, LENS, SISSA (Italy); IAC (Spain). This development has been supported by the funding agencies BMVIT (Austria), ESAPRODEX (Belgium), CEA/CNES (France), DLR (Germany), ASI (Italy) and CICYT/MICINN (Spain). Funding for the SDSS and SDSS-II has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, the US Department of Energy, the National Aeronautics and Space Administration, the Japanese Monbukagakusho, the Max Planck Society and the Higher Education Funding Council for England. The SDSS website is http://www.sdss.org/. The SDSS is managed by the Astrophysical Research Consortium for the Participating Institutions. The Participating Institutions are the American Museum of Natural History, Astrophysical Institute Potsdam, University of Basel, University of Cambridge, Case Western Reserve University, University of Chicago, Drexel University, Fermilab, the Institute for Advanced Study, the Japan Participation Group, Johns Hopkins University, the Joint Institute for Nuclear Astrophysics, the Kavli Institute for Particle Astrophysics and Cosmology, the Korean Scientist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos National Laboratory, the Max-Planck-Institute for Astronomy (MPIA), the Max- Planck-Institute for Astrophysics (MPA), New Mexico State University, Ohio State University, University of Pittsburgh, University of Portsmouth, Princeton University, the United States Naval Observatory and the University of Washington. Financial support from the Spanish grant AYA2010-15169 and from the Junta de Andaluc´ ıa through TIC-114 and the Excellence Project P08-TIC-03531 is acknowledged. REFERENCES Atek H. et al., 2011, ApJ, 743, 121 Barger A. J., Cowie L. L., Wang W.-H., 2008, ApJ, 689, 687 Basu-Zych A. R., Hornschemeier A. E., Hoversten E. A., Lehmer B., Gronwall C., 2011, ApJ, 739, 98 Boissier S. et al., 2007, ApJS, 173, 524 (B07) Boquien M. et al., 2009, ApJ, 706, 553 Boquien M. et al., 2012, A&A, 539, A145 Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018
182 I. Oteo et al. Bouwens R. J., Illingworth G. D., Blakeslee J. P., Franx M., 2006, ApJ, 653, 53 Bouwens R. J. et al., 2009, ApJ, 705, 936 Bouwens R. J. et al., 2012, ApJ, 754, 83 Bruzual G., Charlot S., 2003, MNRAS, 344, 1000 (BC03) Buat V. et al., 2005, ApJ, 619, L51 Buat V., Takeuchi T. T., Burgarella D., Giovannoli E., Murata K. L., 2009, A&A, 507, 693 Buat V. et al., 2010, MNRAS, 409, L1 Buat V. et al., 2012, A&A, 545, A141 Burgarella D. et al., 2006, A&A, 450, 69 Burgarella D., Le Floc’h E., Takeuchi T. T., Huang J. S., Buat V., Rieke G. H., Tyler K. D., 2007, MNRAS, 380, 986 Burgarella D. et al., 2011, ApJ, 734, L12 Calzetti D., Kinney A. L., Storchi-Bergmann T., 1994, ApJ, 429, 582 Calzetti D., Armus L., Bohlin R. C., Kinney A. L., Koornneef J., Storchi- Bergmann T., 2000, ApJ, 533, 682 Capak P. et al., 2007, ApJS, 172, 99 Castellano M. et al., 2012, A&A, 540, A39 Chapman S. C., Casey C. M., 2009, MNRAS, 398, 1615 Chapman S. C. et al., 2000, MNRAS, 319, 318 Chary R., Elbaz D., 2001, ApJ, 556, 562 (CE01) Chen Z., Shu C. G., Burgarella D., Buat V., Huang J.-S., Luo Z. J., 2013, MNRAS, 431, 2080 Condon J. J., 1992, ARA&A, 30, 575 Daddi E. et al., 2007, ApJ, 670, 156 Dale D. A., Helou G., 2002, ApJ, 576, 159 (DH02) DamenM.,F¨ orsterSchreiberN.M., Franx M., Labb´ eI.,ToftS.,vanDokkum P. G., Wuyts S., 2009, ApJ, 705, 617 Davies L. J. M., Bremer M. N., Stanway E. R., Lehnert M. D., 2013, MNRAS, 433, 2588 de Barros S., Schaerer D., Stark D. P., 2012, arXiv:e-prints Diolaiti E., Bendinelli O., Bonaccini D., Close L., Currie D., Parmeggiani G., 2000a, A&AS, 147, 335 Diolaiti E., Bendinelli O., Bonaccini D., Close L. M., Currie D. G., Parmeggiani G., 2000b, Proc. SPIE, 4007, 879 DunlopJ.S., McLure R. J., RobertsonB. E., EllisR. S.,Stark D. P., Cirasuolo M., de Ravel L., 2012, MNRAS, 420, 901 Dunne L. et al., 2009, MNRAS, 394, 3 Elbaz D. et al., 2007, A&A, 468, 33 Elbaz D. et al., 2010, A&A, 518, L29 Elbaz D. et al., 2011, A&A, 533, A119 Elmegreen B. G., Elmegreen D. M., Fernandez M. X., Lemonias J. J., 2009a, ApJ, 692, 12 Elmegreen D. M., Elmegreen B. G., Marcus M. T., Shahinyan K., Yau A., Petersen M., 2009b, ApJ, 701, 306 Elvis M. et al., 2009, ApJS, 184, 158 Erb D. K., Steidel C. C., Shapley A. E., Pettini M., Reddy N. A., Adelberger K. L., 2006, ApJ, 647, 128 Feldmann R. et al., 2006, MNRAS, 372, 565 Feulner G., Goranova Y., Drory N., Hopp U., Bender R., 2005a, MNRAS, 358, L1 Feulner G., Gabasch A., Salvato M., Drory N., Hopp U., Bender R., 2005b, ApJ, 633, L9 Finkelstein S. L., Malhotra S., Rhoads J. E., Hathi N. P., Pirzkal N., 2009, MNRAS, 393, 1174 Finkelstein S. L. et al., 2012, ApJ, 756, 164 Giovannoli E., Buat V., Noll S., Burgarella D., Magnelli B., 2011, A&A, 525, A150 Goldader J. D., Meurer G., Heckman T. M., Seibert M., Sanders D. B., Calzetti D., Steidel C. C., 2002, ApJ, 568, 651 Gonz´ alez V., Labb´ e I., Bouwens R. J., Illingworth G., Franx M., Kriek M., Brammer G. B., 2010, ApJ, 713, 115 Gonz´ alez V., Labb´ e I., Bouwens R. J., Illingworth G., Franx M., Kriek M., 2011, ApJ, 735, L34 Griffin M. J. et al., 2010, A&A, 518, L3 Haberzettl L., Williger G., Lehnert M. D., Nesvadba N., Davies L., 2012, ApJ, 745, 96 Hathi N. P., Malhotra S., Rhoads J. E., 2008, ApJ, 673, 686 Hathi N. P. et al., 2010, ApJ, 720, 1708 Hathi N. P. et al., 2013, ApJ, 765, 88 Heinis S. et al., 2013, MNRAS, 429, 1113 Ivison R. J. et al., 2010a, MNRAS, 402, 245 Ivison R. J. et al., 2010b, A&A, 518, L31 Iwata I., Ohta K., Tamura N., Akiyama M., Aoki K., Ando M., Kiuchi G., Sawicki M., 2007, MNRAS, 376, 1557 Karim A. et al., 2011, ApJ, 730, 61 Kennicutt R. C., Jr, 1998, ARA&A, 36, 189 Kong X., Charlot S., Brinchmann J., Fall S. M., 2004, MNRAS, 349, 769 Le Floc’h E. et al., 2009, ApJ, 703, 222 Lee K.-S., Alberts S., Atlee D., Dey A., Pope A., Jannuzi B. T., Reddy N., Brown M. J. I., 2012, ApJ, 758, L31 Lilly S. J. et al., 2007, ApJS, 172, 70 Lutz D. et al., 2011, A&A, 532, A90 Ly C. et al., 2009, ApJ, 697, 1410 Ly C., Malkan M. A., Hayashi M., Motohara K., Kashikawa N., Shimasaku K., Nagao T., Grady C., 2011, ApJ, 735, 91 Madau P., 1995, ApJ, 441, 18 Madau P., Ferguson H. C., Dickinson M. E., Giavalisco M., Steidel C. C., Fruchter A., 1996, MNRAS, 283, 1388 Magdis G. E., Rigopoulou D., Huang J.-S., Fazio G. G., 2010a, MNRAS, 401, 1521 MagdisG. E.,ElbazD.,DaddiE., MorrisonG.E.,DickinsonM., Rigopoulou D., Gobat R., Hwang H. S., 2010b, ApJ, 714, 1740 Magdis G. E. et al., 2010c, ApJ, 720, L185 Magnelli B., Elbaz D., Chary R. R., Dickinson M., Le Borgne D., Frayer D. T., Willmer C. N. A., 2009, A&A, 496, 57 Meurer G. R., Heckman T. M., Calzetti D., 1999, ApJ, 521, 64 (M99) Moles M. et al., 2008, AJ, 136, 1325 Murphy E. J., Chary R.-R., Dickinson M., Pope A., Frayer D. T., Lin L., 2011, ApJ, 732, 126 Noeske K. G. et al., 2007, ApJ, 660, L43 Nordon R. et al., 2013, ApJ, 762, 125 Oke J. B., Gunn J. E., 1983, ApJ, 266, 713 Oliver S. J. et al., 2010, A&A, 518, L21 Oteo I. et al., 2011, ApJ, 735, L15 Oteo I. et al., 2012a, A&A, 541, A65 Oteo I. et al., 2012b, ApJ, 751, 139 Oteo I. et al., 2013a, MNRAS, submitted Oteo I. et al., 2013b, A&A, 554, L3 Overzier R. A. et al., 2011, ApJ, 726, L7 (O11) Pannella M. et al., 2009, ApJ, 698, L116 Peng C. Y., Ho L. C., Impey C. D., Rix H., 2010, AJ, 139, 2097 Pilbratt G. L. et al., 2010, A&A, 518, L1 Poglitsch A. et al., 2010, A&A, 518, L2 Reddy N. A., Steidel C. C., Fadda D., Yan L., Pettini M., Shapley A. E., Erb D. K., Adelberger K. L., 2006, ApJ, 644, 792 Reddy N. A., Erb D. K., Pettini M., Steidel C. C., Shapley A. E., 2010, ApJ, 712, 1070 Reddy N. et al., 2012, ApJ, 744, 154 Rigopoulou D. et al., 2010, MNRAS, 409, L7 Rodighiero G. et al., 2010, A&A, 518, L25 Salim S. et al., 2007, ApJS, 173, 267 Salmi F., Daddi E., Elbaz D., Sargent M. T., Dickinson M., Renzini A., Bethermin M., Le Borgne D., 2012, ApJ, 754, L14 Salpeter E. E., 1955, ApJ, 121, 161 Salvato M. et al., 2011, ApJ, 742, 61 Sanders D. B. et al., 2007, ApJS, 172, 86 Sawicki M., 2012, MNRAS, 421, 2187 Schaerer D., de Barros S., 2009, A&A, 502, 423 Schaerer D., de Barros S., 2010, A&A, 515, A73 Schaerer D., de Barros S., Stark D. P., 2011, A&A, 536, A72 Schaerer D., de Barros S., Sklias P., 2013, A&A, 549, A4 Schinnerer E. et al., 2010, ApJS, 188, 384 Scoville N. et al., 2007, ApJS, 172, 1 S´ ersic J. L., 1968, in S´ ersic J. L., ed., Atlas de galaxias australes. Observatorio Astronomico, Cordoba, Argentina Downloaded from https://academic.oup.com/mnras/article-abstract/435/1/158/1102582 by Universidad de Sevilla user on 26 March 2018