Comparison of star-formation rates for the deep field Extended Groth Strip and its Dusty Star Forming Galaxies population by Luisa Fernanda Cardona Torres Thesis submitted in partial fullfillment of the requirements for the degree of MASTER OF SCIENCE IN ASTROPHYSICS at the Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ September 2019 Tonantzintla, Puebla Advised by: PhD. Itziar Aretxaga Tenured Researcher - INAOE PhD. Alfredo Montana˜ Conacyt Fellow - INAOE c INAOE 2019
The author hereby grants to INAOE permission to reproduce and to distribute publicly paper and electronic copies of this thesis document in whole or in part
Abstract
This masters thesis presents the study of the star formation rates (SFR) of submil- limeter galaxies (SMGs) in the deep field Extended Groth Strip (EGS). We used the SFRs reported in the CANDELS and SCUBA-2 Cosmology Legacy Survey (SCLS2) catalogues, derived from UV/optical, NIR and submillimeter measurements. We per- formed the selection of the star-forming galaxies in EGS from the CANDELS cata- logue. With this sample we adjusted the star formation main sequence to the star- forming population of EGS for redshift bins (0.5 < z < 6). Our fit best agrees with the fit by Speagle, et al. (2014) and the high-mass end of the fit by Whitaker, et al. (2014). However, the exact location of the star formation main sequence is still uncertain. Then we compared the estimations of the SFRs from both catalogues for the population of SMGs with counterparts identified in the optical and near infrared bands. The total SFRUV+IR reported in the CANDELS catalogue is derived from measurements in the UV/optical bands and is corrected for dust extinction with ob- servations at 24 µm and the SFRIR is derived from the submillimeter observations of SCLS2 at 450 and 850 µm. When comparing both estimations we found that in most cases SFRUV+IR underestimates the total SFR of the SMGs. Nonetheless, 86% of the SMGs lie within the 3σ region of our star formation main sequence fit, where 4% of the sample can be classified as starbursts as they are located above the 3σ region of the fit. We performed follow-up spectroscopic observations of some SMGs at the GTC-MOS Osiris instrument. We were able to identify emission lines in the spectra and propose candidate spectroscopic redshift when possible. However, we were unable to measure the flux of emission lines that trace the star formation activity. Finally, we propose the future work to be developed in a doctorate program.
[i] ii Abstract
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ Resumen
Esta tesis de maestr´ıa presenta el estudio de las tasas de formaci´onestelar (SFR) de la poblaci´onde galaxias submilim´etricasen el campo profundo Extended Groth Strip. Se emplearon las SFR reportadas en los cat´alogosCANDELS y SCUBA-2 Cos- mology Legacy Survey (SCLS2), las cuales han sido estimadas a partir de medidas realizadas en el UV/´optico, cercano y mediano infrarojo y sub-milim´etrico.Se re- aliz´ola selecci´onde las galaxias con formaci´onestelar en el campo EGS del cat´alogo CANDELS, observadas en el UV/´optico.Con esta muestra se realiz´oel ajuste de la secuencia principal de formaci´onestelar a esta poblaci´onde EGS, para corrimientos al rojo entre 0.5 y 6. Nuestro ajuste concuerda mejor con el ajuste presentado por Speagle, et al. (2014) y con el ajuste a altas masas de Whitaker, et al. (2014). Sin embargo, la ubicaci´onexacta de la secuencia principal de formaci´onestelar a´unes incierta. Entonces comparamos las estimaciones de SFRs de ambos cat´alogospara la poblaci´onde galaxias submilim´etricascon contrapartes identificadas en las bandas
´opticasy del cercano infrarojo. La tasa de formaci´ontotal SFRUV+IR, reportada en el cat´alogoCANDELS, est´aderivada de mediciones en el UV/´opticoy corregida por la extinci´onde polvo con observaciones a 24 µm y la SFRIR proveniente de las observa- ciones submilim´etricasde SCLS2 a 450 y 850 µm. Al comparar las dos estimaciones encontramos que en la mayor´ıade los casos la SFRUV+IR subestima la SFR total de las galaxias submlim´etricas.A pesar de esto, 86% de las SMGs yacen dentro de la region de 3σ de nuestro ajuste de la secuencia principal de formaci´onestelar, donde 4% de la muestra de SMGs pueden ser clasificadas como galaxias starbursts ya que se encuentran ubicadas sobre la regi´onde 3σ del ajuste. Finalmente, realizamos un seguimiento espectrosc´opicode algunas SMGs con el GTC/MOS Osiris, en los cuales identificamos l´ıneasde emisi´on,proponiendo un corrimiento al rojo para los que fue
[iii] iv Resumen posible. Sin embargo, no logramos medir el flujo de l´ıneasde emisi´onque trazaran la actividad de formaci´onestelar. Finalmente, se plantea una propuesta de trabajo futuro a desarrollar en una etapa de doctorado.
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ Contents
Abstract i
Resumen iii
1 Introduction 1 1.1 Dusty Star Forming Galaxies ...... 2 1.2 K-correction ...... 3 1.3 Star Formation Rate ...... 5 1.3.1 Star-formation main sequence ...... 7 1.4 Main goals and key questions ...... 8
2 Extended Groth Strip Data 9 2.1 Data sources ...... 9 2.1.1 AEGIS ...... 9 2.1.2 CANDELS catalogue ...... 11 2.1.3 Optical spectra ...... 11 2.1.4 Submillimeter data ...... 12 2.2 Optical counterparts to submillimeter galaxies ...... 14 2.2.1 Derived properties of the SCLS2 galaxies ...... 17
3 Demographics of the EGS population 21 3.1 UVJ diagram ...... 23 3.2 Star Formation Main Sequence ...... 29 3.2.1 Location of Dusty Star-Forming Galaxies within the star for- mation main sequence ...... 33
[v] vi CONTENTS
4 GTC-MOS spectra 35 4.1 Phase II - planning ...... 35 4.2 Line identification and redshift report ...... 38 4.2.1 HST-galaxies ...... 39 4.2.2 Submillimeter galaxies ...... 46
5 Conclusions and future work 53
List of Figures 57
List of Tables 59
References 61
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ Chapter 1
Introduction
One of the most important goals of modern astrophysics is to understand the forma- tion and evolution of the universe and the structures it contains. In 1996 the tenta- tive detection of the Cosmic Infrared Background (CIB) by the Cosmic Background Explorer (COBE) was presented (Puget et al., 1996), which had been predicted be- forehand as the contribution of early galaxies (Partridge & Peebles, 1967). Later on, the confirmation of the CIB measurement was presented, whose integrated energy (in the 140 to 240 µm range) is 2.5 times the integrated optical/NIR light from the ∼ galaxies in the Hubble Deep Field (Hauser et al., 1998). This implies that there is a dust enshrouded population of galaxies at high redshift and that an important frac- tion of our understanding of the star formation history of the universe is obscured by dust (Dwek et al., 1998). Since then, the technological developments have permitted the growth of a new kind of astronomical observations at submillimeter wavelengths, where the dust enshrouded population of the universe and its contribution to the CIB has been under study (Smail et al., 1997; Hughes et al., 1998; Perera et al., 2008; Hatsukade et al., 2011; Scott et al., 2012; Geach et al., 2013; Zavala et al., 2017). For instance, Fujimoto et al. (2015) presented a statistical study of sources detected in the 1.2 mm band at the Atacama Large Millimeter Array/submillimeter (ALMA) +31 where the integrated flux of these sources accounted for 104−25% of the CIB found by COBE.
In this work we focus on the physical properties of star-forming galaxies that are dust enshrouded (DSFGs) within the deep field Extended Groth Strip (EGS).
[1] 2 1. Introduction
1.1 Dusty Star Forming Galaxies
Submillimeter galaxies (SMGs) are a high-z population of star-forming galaxies which contain large amounts of dust. This implies that a great extent of their UV radiation, emitted by young stars, is obscured by the dust and thermally re-emitted at longer 12 wavelengths in the Infrared (IR). They have large IR Luminosities (LIR = 10 L ), −1 which are related to high Star Formation Rates of the order SFR 300 3000M yr ≈ − (Smail et al., 1997; Hughes et al., 1998). The heated dust grains within the DSFGs emit radiation as a modified black body, and there is not a straight forward way to discern between the two main possible mechanisms that heat the dust: Active Galactic Nuclei (AGN) or young stars (Blain et al., 2002).
SMGs were discovered in the late 90’s thanks to the technological developments that allowed observations in the submillimiter range, specifically at 850 µm with the Submillimeter Common-Use Bolometer Array (SCUBA) at the 15 m James Clerk Maxwell Telescope (JCMT, i.e. Smail et al., 1997; Hughes et al., 1998; Barger et al., 1998). This marked the beginning of a new research area in astrophysics, which has been reviewed by Blain et al. (2002) and Casey et al. (2014). As a recently discovered population, there have been efforts to understand them in terms of local known ob- jects. For instance, low-z objects considered analogues to SMGs, similar in luminosity 12 (Lbol 10 L ), are the Ultra Luminous Infra-Red Galaxies (ULIRGs Blain et al., ∼ 2002), whose dust emission dominates the SED from 8 µm to 1 mm (and therefore
Lbol LIR). However, whether the SMGs are scaled up versions of ULIRGs at high ≈ redshift or not is still under debate (Casey et al., 2014).
A very important new generation survey in the study of DSFGs is the SCUBA- 2 Cosmology Legacy Survey (SCLS2, see section 2.1.4), which observed over 3000 submillimeter sources at 850 µm approaching the confusion limit (Geach et al., 2017) of the JCMT. The confusion limit happens when the noise is neither dominated by atmospheric noise nor by instrumentation noise, but instead by the blurring of faint sources caused by the resolution limitation of the telescope (Blain et al., 2002).
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ 1.2 K-correction 3
There is a selection bias upon the wavelength at which DSFGs are detected in these surveys. For instance, Casey et al. (2013) report a higher peak z for the 850 µm detected galaxies than the 450 µm detected ones. Likewise, galaxies selected at shorter wavelengths (i.e. 450 µm) have higher dust temperatures, implying a warmer popula- tion (Casey et al., 2013). Zavala et al. (2014) further studied these selection effects and they showed that the variations upon the peak z for different observation wavelengths are consistent with a common parent distribution. In order to gain a more complete understanding of a population, it is common to associate sources detected at a certain wavelength with counterparts found at other wavelengths. Since single-dish submillimeter observations usually have poor resolution (θ 15 30arcsec), we resort to sources detected at higher positional certainties at ≈ − other wavelengths, like radio wavelenghts, given the known correlation between FIR and radio for starburst galaxies (Condon, 1992; Carilli & Yun, 2000), or mid-infrared 24 µm, where we look for the emission of dust at shorter wavelengths (Pope et al., 2006). This is, however, a very difficult endeavour for SMGs, where counterparts cannot be found for complete samples (Casey et al., 2013; Zavala et al., 2018).
1.2 K-correction
The distribution of the flux-density over wavelength (or frequency) emitted by an object is known as its Spectral Energy Distribution (SED). The shape of the SED depends upon the mechanisms of emission of the object. As it was mentioned in the previous section, the thermal emission of dust, which dominates the SED of DSFGs, is commonly modeled with a modified black body distribution. The flux-density at frequency ν is given by:
3 ν β 2hν 1 −( /ν ) Sν = [1 e 0 ] , (1.2.1) − c2 ehν/kT 1 − where ν0 is the frequency at which the emission becomes optically thick, β is the emissivity index, and T is the temperature of the emitting source, which in the case of DSFGs is considered to be the dust-temperature.
Determination of SFR of SMGs in the EGS 4 1. Introduction
Figure 1.1: (Left panel) Modified black body for a source with luminosity LIR = 12 1 10 L , dust-temperature Td = 42 K, emissivity index β = 1.8 and ν0 = c/100µm (Casey× et al., 2014). The SEDs are redshifted to lower ν as z increases. The vertical lines correspond to observed wavelengths: 2.1 mm (blue dash-dot line),1.1 mm (red line), 850 µm (green dashed line), 450 µm (cyan dotted line) and 70 µm (red dash- dot line). (Right panel) Flux density at the observed λ (following the same colors and symbols as in the left panel) for the same modified black body (LIR,β,ν0,Td). We can see that for λ = 70 µm there is a strong positive K-correction, while at larger λ (λ 450 µm) the effect is opposite (negative K-correction). ≥
As we observe sources with the same luminosity at higher redshifts, the SED shifts to higher wavelengths and the peak brightness decreases its intensity. It is necessary to apply the K-correction in order to estimate the emitted flux-density by the source. This effect is shown in Figure 1.1, with a modified black body SED redshifted to a set of values, and the resulting observed flux density as a function of redshift. In the left panel it can be appreciated how the peak of the SED approaches submillimeter observational wavelengths as the redshift of the source increases. In the right panel,
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ 1.3 Star Formation Rate 5 which shows how the flux density of a galaxy with fixed L changes as a function of z, we can see that at λ 1 mm the flux density remains almost constant (or even ∼ increases) at z 1 10. This aids the observation of this type of galaxies up to high ≈ − redshifts at wavelengths higher than λ > 250 µm (Blain et al., 2002). The observed flux density can be estimated through the equation (Casey et al., 2014):
L (1 + z) S S = IR ν , (1.2.2) ν,obs 4πD2 R 1000µm L 8µm Sν(Td)dν where DL is the luminosity distance, Sν is the source emitted flux density at rest-frame R 1000µm frequency ν, LIR the galaxy infrared luminosity, z is the redshift, and 8µm Sν(Td)dν is the modified black body emission integrated over the 8 µm 1 mm wavelength range. − Given that in the case of DSFGs, detected at submillimeter wavelengths, the effect of the K-correction is applied to decrease the observed flux-density to estimate the rest-frame flux density, then it is known as negative K-correction.
1.3 Star Formation Rate
The Star Formation Rate (SFR) is a measurement of the rate at which stars are born in a galaxy, or a particular region within it, and how they transform gas into −1 stars. It is typically measured in units of solar masses per year (M yr ). There are several calibrated methods to estimate the SFR, according to tracers associated with young stellar populations at different wavelengths. Some of the most used methods presented in Kennicutt (1998) are the following:
SFR derived from the UV luminosity: •
−1 −28 −1 −1 SFRUV(M yr ) = 1.4 10 LUV(erg s Hz ) (1.3.1) ×
SFR derived from recombination lines like Hα: •
−1 −42 −1 SFRHα(M yr ) = 7.9 10 L(Hα)(erg s ) (1.3.2) ×
Determination of SFR of SMGs in the EGS 6 1. Introduction
SFR derived from forbidden lines like [OII]: •
−1 −41 −1 SFR[OII](M yr ) = 1.4 10 L[OII](erg s ) (1.3.3) ×
SFR derived from the IR luminosity: •
−1 −44 −1 SFRIR(M yr ) = 4.5 10 LIR(erg s ) (1.3.4) ×
These relations may vary according to the initial mass function (IMF) considered by the author. The most frequently used IMFs are those by Salpeter (1955), Kroupa (2001) and Chabrier (2003). Kennicutt (1998) used the Salpeter IMF for equations 1.3.1 - 1.3.4. If, instead of the Salpeter IMF, a Chabrier IMF (2003) is employed, then the SFRIR is lower by a 1.8 factor (Casey et al., 2014).
Figure 1.2: SED fitting of the observational flux-densities for an example galaxy (irac131161) at redshift z 1. The SCLS2 450 µm and 850 µm fluxes would lie in the rightmost region of the∼ SED, where there is an extrapolation in the fit. Image from Barro et al. (2011a).
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ 1.3 Star Formation Rate 7
UV-based determinations are affected by dust extinction, which is often difficult to determine. The estimation of IR-based SFRs is also affected by different assumptions. For example, the SED templates used to fit the fluxes of high-z galaxies are adopted local low-z analogues (Barro et al., 2011a). There could also be some radiation con- tribution from obscured AGNs, which can heat the dust, but would not be related to the star-forming mechanisms (Casey et al., 2014). Furthermore, for the IR derived luminosity, few galaxies have deep data beyond 70 µm (e.g. from Spitzer/MIPS Barro et al., 2011a). This affects the calculations of SFR as there are not enough data points to correctly constrain the full IR-SED. Hence, the LIR and SFRIR determinations can be affected by the indetermination of the SED fit and the assumption of a single T throughout the galaxy. As it was mentioned previously, SMGs are bright in the sub- millimeter range and, therefore, in principle they can provide more data points to fit the SED and achieve more accurate IR-based luminosities and star-formation rates.
1.3.1 Star-formation main sequence
The star-formation main sequence is a correlation between the stellar mass (M?) and the star-formation rate (SFR) shown by the majority of star-forming galaxies. The concept was first introduced by Noeske et al. (2007) for optically selected galaxies that had SFRHα and later on expanded by Daddi et al. (2007) for 24 µm selected galaxies. There have been several works that fitted a star-formation main sequence: 64 of those fits were studied by Speagle et al. (2014) for a comprehensive relation that holds up to z = 6. Before the work by Speagle et al. (2014) there were no quantitative comparisons and, given that each work uses different calibrations and methods (e.g. IMFs, Stellar Population Synthesis models, extinction curves), they converted the necesary parameters and derived a time and stellar mass dependent equation. There are still contradictory results. For instance, Whitaker et al. (2014) studied the low-mass slope of the star-formation main sequence for the redshift range 0.5 < z < 2.5. They considered the deep photometry and grism spectra from CANDELS and 3D-HST and three methods for estimating the SFR: SFRUV+IR, β-dust corrected
UV SFR and Hα SFR. On the other hand Koprowski et al. (2016) present the main sequence for galaxies detected in the 850 µm SCLS2, in the high-mass end of the main sequence. Koprowski et al. (2016) discuss the various main sequence fits and
Determination of SFR of SMGs in the EGS 8 1. Introduction
how the FIR/sub-millimeter selected samples yield different relations to the optically
selected fits, where some find a decline in the slope for masses above logM?/M 10.5 ≈ (Whitaker et al., 2014), while others find no differences (Schreiber et al., 2015). They emphasize on the main sequence relation depending upon the sample selection, and the need of the SED fitting to encompass optical-IR data in order to estimate the SFR. Even though the star-formation main sequence is still a subject to debate it is used as a new definition of starburst galaxies. A starburst galaxy is now defined as those with elevated sSFR= SFR , when compared to the main sequence (e.g. Casey M? et al., 2014). In their 2014 review on DSFGs Casey et al. showed that it was still uncertain where the Dusty Star-Forming Galaxy population lies in relation to the main sequence, which depends upon the method and uncertainties related to the estimation of SFR and M?. Furthermore, it is unkown which would be the impact of mergers to the mass build-up of SMGs (Casey et al., 2014).
1.4 Main goals and key questions
The main goal of this thesis is to compare the SFR measurements derived from ob- servations at UV/optical and submillimeter bands. We will work with data from the CANDELS and SCLS2 catalogues for the Extended Groth Strip field. We aim to explore the demographic properties of the galaxies reported in the optical catalogue CANDELS selecting the star-forming and quiescent galaxies using a color-color dia- gram. Besides, we would like to examine how DSFGs with identified counterparts in the CANDELS and IRAC catalogues relate to the rest of the population in this field according to the derived SFR and stellar mass (M?). Throughout this work we adopted a flat ΛCDM cosmology with the following −1 −1 parameters: H0 = 67 km s Mpc , Ωm = 0.32.
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ Chapter 2 Extended Groth Strip Data
The Extended Groth Strip (EGS) field is a region between the constellations Ursae Major and Bootes, located at α = 14h19m00s and δ = +52o4800000 and it is 70 10 × square arcminutes in size. It was first observed by the Hubble Space Telescope (HST ) with the Wide Field Planetary Camera 2 (WFPC1) between March and April 1994 and named after the phycisist Edward Groth (Rhodes, 1999). He was associated with the HST and worked there as data and operations team leader, where he was P.I. for the Wide Field Camera (WFC) instrument. It is one of the most studied fields, along with: GOODS-N, GOODS-S, UDS and COSMOS (e.g. Koekemoer et al., 2011)
2.1 Data sources
The EGS field has had a large multi-wavelength coverage from several instruments and surveys. For instance, the 3D-HST, AEGIS, CANDELS and IRAC surveys include this deep field. Their data is available in the Rainbow Navigator, presented by Barro et al. (2011b) 1.
2.1.1 AEGIS
The All-wavelength Extended Groth Strip International Survey (AEGIS) is a multi- wavelength program whose goal is to understand how galaxies and large-scale struc- tures form and evolve from early times into the universe we see around us (Davis et al., 2007). The instruments involved in this survey range from those used at radio
1Rainbow Navigator (P´erez- Gonzalez and Barro): https://rainbowx.fis.ucm.es/Rainbow˙navigator˙public/
[9] 10 2. Extended Groth Strip Data
wavelengths (VLA at 6 and 20 cm) to X-Rays (Chandra/ Advanced CCD Imaging Spectrometer (ACIS) at 4 and 1 keV). It also covers the UV, NIR and IR spectral re- gions with the following telescopes: Galaxy Evolution Explorer (GALEX), HST in the V (F606W), I (F814W), J (F110W) and H (F160W) bands, Keck Telescope, Canada- France-Hawaii Legacy Survey, Palomar/ Wide-Field Infrared Camera, Spitzer/ In- frared Array Camera (IRAC) for the Mid-Infrared range (at 3.6, 4.5, 5.8 and 8 µm) and the Spitzer/ Multiband Imaging Photometer (MIPS) for the far-IR observations (at 24, 70 and 160 µm). The resulting data is presented as images, spectra and derived data products in their web page2. However, each instrument had a different coverage of the EGS field. Therefore, there are some areas where the surveys do not overlap, as can be seen in Figure 2.1.
Figure 2.1: AEGIS coverage map of the Extended Groth Strip. In both panels the background gray-scale image shows the HST /ACS mosaic composed image. (Left panel) CFH12K (pink), WIRC from Palomar (red), Keck spectroscopic observations with DEIMOS as part of the DEEP2 collaboration (black), CFHTLS MegaCam (blue). (Right panel) VLA at 20 cm(red), VLA at 6 cm(pink), Spitzer/IRAC (green), Chandra (blue), GALEX (black). In both panels the cyan region corresponds to the Mini Test Region and the yellow line marks 30 arcmin along the length of the field. Image taken from Davis et al. (2007).
2Official web-page of the AEGIS catalogue http://morta.ucolick.org/aegis/v
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ 2.1 Data sources 11
2.1.2 CANDELS catalogue
The Cosmic Assembly Near-infrared Deep Extragalactic Legacy Survey (CANDELS) is an HST survey program that covered 800 arcmin2 over five deep fields (Koekemoer et al., 2011; Grogin et al., 2011). The program was designed to observe up to two depths: the CANDELS/Deep survey (limited at H=27.7 mag) and CANDELS/Wide survey (limited at H=27 mag). The WFC3 and ACS cameras on board the HST acquired deep images during 902 orbits between 2010 and 2013. The project was led by Sandra Faber (U. of California, Santa Cruz) and Henry Ferguson (Space Telescope Science Institute). The five deep fields observed within this project were:
CANDELS/Deep survey: CANDELS/Wide survey:
GOODS-N (Great Observatories UDS (Ultra Deep Survey) • • Origins Deep Survey - North) COSMOS (Cosmic Evolution Sur- • vey) GOODS-S (Great Observatories • Origins Deep Survey - South) EGS (Extended Groth Strip) • The CANDELS survey has an extensive database for these observed deep fields. Several sub teams within CANDELS estimated redshifts and other physical parame- ters through various methods, presented and evaluated by Mobasher et al. (2015) and Santini et al. (2015). There have also been demographic studies for the GOODS-S and the UDS fields by Fang et al. (2018). The CANDELS catalogue for the EGS field was presented by Stefanon et al. (2017).
2.1.3 Optical spectra
Within the work in this thesis we present new optical spectra obtained with the 10 m Gran Telescopio de Canarias (GTC) Multi-Object Spectrograph (MOS) for the obser- vation proposal GTC-AMEX19 (PI: Aretxaga). We observed, among other objects, 3 submilimeter galaxies selected from the SCLS2 sample, with optical counterparts presented by Zavala et al. (2017) (section 2.2). Additionally, we observed 4 galaxies selected from the 3D-HST catalogue. In chapter 4 we present the planning, reduction, extracted spectra and line identification of these observations.
Determination of SFR of SMGs in the EGS 12 2. Extended Groth Strip Data
2.1.4 Submillimeter data
2.1.4.1 SCUBA-2 Cosmology Legacy Survey
Our submillimeter sample of Dusty Star-Forming Galaxies (DSFGs) are selected from data acquired with the Submillimetre Common-User Bolometer Array 2 (SCUBA-2) at the James Clerk Maxwell Telescope in 2014, as part of the SCUBA-2 Cosmology Legacy Survey (SCLS2 Geach et al., 2013, 2017). The catalogue presents observations at 450 µm and 850 µm, with depths of σ450 µm 1.9 mJy and σ850 µm 0.46 mJy ≈ ≈ (Zavala et al., 2017). The coverage of the map is 70 arcmin2 with angular FWHM ∼ of θ450 µm 7.5 arcsec and θ850 µm 14.5 arcsec. The sources for a robust sample ≈ ≈ were selected as the peak values from the S/N map, taking into account S/N> 3.5. Considering this threshold, the catalogue has 57 sources at 450 µm and 90 at 850 µm and it is displayed in Figure 2.2 over the corresponding S/N maps at each wavelength.
Figure 2.2: The SCLS2 450 µm (left panel) and 850 µm (right panel) S/N maps of the Extended Groth Strip with detected sources marked with white squares. Image taken from Zavala et al. (2017).
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ 2.1 Data sources 13
2.1.4.2 AzTEC map
There is also an EGS 1.1 mm map acquired with the AzTEC instrument at the Large Millimeter Telescope Alfonso Serrano (LMT). It is centered at α = 214.9420396o, δ = 52.87861111o and has a 91.2 arcmin2 area. The data was acquired in May 2013 and observed over 11 hours, when the LMT was operating in its 32 m configura- tion (θ1.1mm 8.5 arcsec). Figure 2.3 shows the catalogue of SCLS2 sources over ∼ the AzTEC S/N map, where the brightest source at 850 µm (S/N=24.2) appears with S/N=6.6 at 1.1 mm. As it can be seen, the depth of the 1.1 mm map is low
(S1.1 mm 0.5 mJy). There will be a new instrument, named TOLTEC, to be in- ∼ stalled at the LMT in late 2019. It is designed to observe at 1.1, 1.4 and 2.0 mm in the 50 m configuration of the LMT. One of the main projects for this instrument is the observation of deep fields as part of its legacy programs, where the confusion limit for the 1.1 mm band will be reached (S1.1 mm 0.025 mJy). ∼
34 35 72 50 43 60 5479 27 80 68 43 4022 48 1446 1721 3470 64 36 13 93 7 2023 55 254 129 51 136 52 62 89 158 49 4226 21 81 74 2738 50 48 56 28 2631 67 44 57 53 82 63 57 23 30244 88 4469126 23 3010 159 32 4931 78 207 75 77 6518 2529 53 35 18 47 71 8533 36 37 4039 4116 421711 28 58 221 2459 47 54 2921 94 5273 1410 91 61 32 51 5616 194638 33 3911 83 37 53 92 45 58 90 66 45 55
-2.9 -2.2 -1.6 -0.88 -0.21 0.47 1.1 1.8 2.5 3.1 3.8 Figure 2.3: The 1.1 mm S/N map of the EGS field observed by AzTEC at the 32 m LMT. This region closely matches the SCLS2 coverage area described in 2.1.4. The 850 µm detected sources are marked in red and the 450 µm detected sources are marked in blue.
Determination of SFR of SMGs in the EGS 14 2. Extended Groth Strip Data
2.2 Optical counterparts to submillimeter galaxies
Once the SCLS2 sources were detected, they were associated with optical counter- parts using the panchromatic data available for this field (Zavala et al., 2018). The counterparts were found using: radio VLA at 20 cm, which traces recent star forma- tion via synchrotron emission (Carilli & Yun, 2000); observations at 24 µm, as they are sensitive to warm dust emission; and 8 µm data as a tracer of the older and mass- dominant stellar population. Counterparts were found within a search radius 2.5 times the positional uncertainty of each galaxy. They found 71 counterparts. Out of the 71 identified counterparts, 58 lie within the footprint of CANDELS. The remaining 13 galaxies that fall outside the HST /CANDELS coverage, were associated to sources from the IRAC catalogue presented by Barro et al. (2011b). In Table 2.1 we present the ID and coordinates for the optical counterparts. From now on, we will work with this sample of DSFGs with identified counterparts. They have reported Stellar-Mass and SFR values estimated by various teams in the CANDELS catalogue.
Table 2.1: Optical counterparts for the SCLS2 sources. The optical counterpart coor- dinates are those of CANDELS (Stefanon et al., 2017) or IRAC (Barro et al., 2011b) catalogues. Column 1: name in the SCLS2 catalogue at 850 µm. Column 2: identifi- cation in the SCLS2 catalogue at 450 µm. Column 3: identification in the CANDELS catalogue with HST observation. Column 4: identification in the IRAC catalogue. Column 5: right-ascension of the optical/IR counterparts. Column 6: Declination of the optical/IR counterparts.
ID850 µm ID450 µm IDHST IDIRAC RAopt DECopt 850.001 450.02 21335 - 214.910882 52.900925 850.002 450.03 16498 - 214.914447 52.875959 850.003 450.05 18656 - 214.916415 52.891337 850.004 450.25 17920 - 214.946689 52.910020 850.005 450.08 - 122794 214.978122 52.811542 850.006 450.13 13658 - 214.974134 52.905954 850.007 450.20 8156 - 214.962443 52.870324 850.008 450.15 8265 - 215.014456 52.901214 850.009 450.12 - 123412 214.843676 52.910939 Continued on next page
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ 2.2 Optical counterparts to submillimeter galaxies 15
Table 2.1 – continued from previous page
ID850 µm ID450 µm IDHST IDIRAC RAopt DECopt 850.010 450.14 8340 - 214.919736 52.839658 850.011 450.39 4919 - 214.922389 52.821939 850.012 450.06 12475 - 214.946125 52.876173 850.014 450.46 - 121876 1 214.8562017 52.9287205 850.015 450.09 12874 - 214.938204 52.874313 850.016 450.56 - 120588 215.007724 52.827103 850.017 450.11 13776 - 214.898846 52.852513 850.018 450.84 4996 - 214.974861 52.860563 850.019 450.41 24050 - 214.970982 52.957352 850.020 450.23 7994 - 215.031040 52.916459 850.021 450.17 6891 - 215.054219 52.925918 850.022 450.01 9333 - 214.927684 52.847964 850.024 450.04 16318 - 214.923237 52.882156 850.025 450.29 23205 - 214.850601 52.866404 850.028 450.69 16544 - 214.876329 52.852377 850.029 450.21 21206 - 214.835440 52.843781 850.030 450.10 21235 - 214.877974 52.876880 850.032 450.63 5729 - 214.932895 52.832951 850.037 450.64 - 117087 215.050227 52.853897 850.038 450.27 27358 - 214.864723 52.899081 850.039 450.40 - 117578 215.038863 52.854429 850.041 450.16 1368 - 214.992838 52.850766 850.042 450.26 12502 - 214.979990 52.902444 850.043 450.73 23152 - 214.950132 52.938209 850.052 450.76 - 124474 1 214.827757 52.904873 850.059 450.24 18694 - 214.855508 52.848827 850.060 450.75 - 121668 214.84949 52.938104 850.065 450.18 21212 - 214.875544 52.866459 850.069 450.44 11896 - 214.954780 52.876572 Continued on next page
Determination of SFR of SMGs in the EGS 16 2. Extended Groth Strip Data
Table 2.1 – continued from previous page
ID850 µm ID450 µm IDHST IDIRAC RAopt DECopt 850.070 450.34 18143 - 214.968421 52.925046 850.073 450.52 5375 - 214.948005 52.840633 850.078 450.82 -- 215.0377 52.8708.. 850.079 450.54 12448 - 215.029611 52.936340 850.085 450.33 - 118504 1 215.015842 52.856805 850.092 450.60 8277 - 214.885212 52.815749 850.095 450.37 - 121835 1 214.996545 52.814086 850.097 450.71 14062 - 214.920611 52.865887 850.104 450.45 - 122942 214.986563 52.80205 850.044 - 2998 - 215.034896 52.891364 850.047 - 20341 - 214.859833 52.860656 850.048 - 2838 - 215.045433 52.894592 850.050 - 17067 - 214.996769 52.94075 850.051 - 10774 - 214.995061 52.906108 850.053 - 3345 - 214.993497 52.864329 850.054 - 15371 - 214.874737 52.843773 850.056 - 4619 - 215.027303 52.894668 850.057 - 7340 - 214.996947 52.889169 850.062 - 1507 - 215.061805 52.900938 850.067 - 28663 - 214.829331 52.893962 850.072 - 14167 - 215.015852 52.939408 850.075 - 26086 - 214.834060 52.870092 850.077 - - 117273 215.029710 52.867911 850.080 - 26097 - 214.923151 52.934686 850.081 - 14924 - 214.954660 52.899053 850.082 - - 115315 215.055684 52.881997 - 450.19 6840 - 214.916970 52.827376 - 450.28 15610 - 214.945929 52.894158 - 450.35 15566 - 214.902474 52.863695 Continued on next page
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ 2.2 Optical counterparts to submillimeter galaxies 17
Table 2.1 – continued from previous page
ID850 µm ID450 µm IDHST IDIRAC RAopt DECopt - 450.36 9195 - 214.936159 52.855855 - 450.42 13677 - 214.905312 52.850914 - 450.43 11923 - 215.020341 52.929760 - 450.48 27020 - 214.908677 52.927897
2.2.1 Derived properties of the SCLS2 galaxies
Zavala et al. (2017) identified the sources in the SCLS2 catalogue, detecting them at first in the S/N map and then measuring the coordinates, flux density and noise. In order to measure these quantities, they fitted to each detected source the Point Spread Function (PSF), which was estimated in the reduction procedure. This measured flux density is called raw flux density, as it does not have any boosting correction. We measured the raw flux density at the reported coordinates for each wavelength map and compared it with the raw flux density reported by Zavala et al. (2017). In Figure 2.4, we present a comparison between both reported flux densities for the 450 µm map, where we can appreciate that there is good agreement between both measured flux densities. Therefore, we will adopt the derived physical parameters estimated by Zavala et al. (2017) for our sample from the SCLS2.
2.2.1.1 Deboosted flux
When sources are detected at a low S/N, there occurs contamination from the crowd- ing of sources fainter than the detection threshold, which increases (boosts) the mea- sured flux density of the detected sources (Coppin et al., 2006). Therefore, the raw flux densities have to be deboosted before estimating physical parameters. The deboosting factor is estimated through an statistical process using Monte Carlo simulations, as described by Zavala et al. (2017). Fainter sources are more boosted than brighter ones. Considering that we have compared the raw flux densities measured directly and that they comply with a 1:1 relation (Figure 2.4), we will adopt from now on the fDeboost from Zavala et al. (2017) and use these values to estimate the infrared luminosities and star formation rates.
Determination of SFR of SMGs in the EGS 18 2. Extended Groth Strip Data
Figure 2.4: Comparison between the SCLS2 450 µm raw flux densities and fRAW measured in this work (y-axis) and those reported by Zavala et al. (2017) at map (x-axis).
2.2.1.2 Infrared Luminosity
The IR luminosity (8 1000 µm) reported by Zavala et al. (2018) was estimated by − fitting a modified black body, fixed at the redshift provided by the optical catalogues.
The modified black body that was used had an emissivity index β = 1.6 and ν0 =
c/100 µm, from equation 1.2.1. Our LIR estimation uses the same modified black body
parameters (β, ν0), the redshift reported in the catalogues for the optical counterparts,
the deboosted flux and the dust temperature (Td) as in Zavala et al. (2017, 2018). We
must point out that in Zavala et al. (2018) they fitted both the LIR and Td. Instead, we are calculating the IR-luminosity from either the 450 µm or 850 µm, assuming that the
Td is correct. The resulting LIR comparison is presented in Figure 2.5, where we can appreciate that the LIR calculated from the 850 µm flux densities are systematically located above the 1:1 relation. As the f850 µm is placed further away from the modified black body peak emission the fit is less accurate, in this case overestimating the LIR. We could safely assume this is the reason for the overestimation, given that this effect is only appreciated in the estimated LIR from 850 µm flux densities and not at 450 µm.
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ 2.2 Optical counterparts to submillimeter galaxies 19
Figure 2.5: Comparison of LIR calculated in this work and Zavala et al. (2018), using the 450 µm (upper panel) and 850 µm (lower panel) deboosted flux densities and LIR by Zavala et al. (2017). We considered a modified black body with β = 1.6 and ν0 = c/100 µm, z of the optical counterparts and Td fitted by Zavala et al. (2018).
Determination of SFR of SMGs in the EGS 20 2. Extended Groth Strip Data
2.2.1.3 Infrared star-formation rate for dusty star forming galaxies
Considering the LIR, Zavala et al. (2018) estimate the SFRIR with a Kennicutt (1998) 12 relation and Chabrier (2003) initial mass function (IMF), where LIR 10 L implies −1 ∼ a SFR 100M yr . Equation 1.3.4 considers a different IMF, so it must be scaled ∼ to account for this difference with a factor 1.8. Zavala et al. (2018) use the star- ∼ forming main sequence derived by Speagle et al. (2014), and our sample of DSFG population lies mostly within the 3σ of this main sequence fit (Figure 2.6).
Figure 2.6: Distribution of the DSFGs from the SCLS2 catalogue with optical coun- terparts in the sSFR-redshift relation by Zavala et al. (2018). The continuous black line and its 3σ uncertainty region (gray-scale) are the star forming main sequence derived by Speagle et al. (2014). The black dots are the galaxies detected at both 450 µm and 850 µm, the blue rectangles are those only detected at 450 µm and the red triangles are the galaxies detected only in the 850 µm data. Image taken from Zavala et al. (2018).
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ Chapter 3
Demographics of the EGS population
The EGS CANDELS catalogue includes physical properties calculated by various teams, as presented in Mobasher et al. (2015) and Santini et al. (2015), who com- pare the results from the teams. They concluded that the results are in overall good agreement, despite the different methods and assumptions (i.e. initial mass functions (IMFs), Stellar Population Synthesis templates, SED fitting code) that each team considered. We adopt from the CANDELS catalogue the Star Formation Rate (SFR) presented by Barro et al. (2011a), which is the total SFR SFRIR+UV, according to ∼ equation 3.0.1. In order to obtain the SFRIR+UV they estimated the IR luminosities
(LIR = L[8 1000 µm]) and luminosity at 2800 A.˚ The LIR was calculated by fitting − the MIPS 24 µm fluxes to SED templates, and L2800 was obtained from the best fit of optical SED templates (Barro et al., 2017, see Figure 1.2).
−10 SFRUV+IR = 1.09 10 (LIR + 3.3L2800) (3.0.1) ×
There have been galaxy demographic studies for the other CANDELS fields, like those in UDS and GOODS-S conducted by Fang et al. (2018). Following their proce- dure, we select a sample of galaxies from the EGS catalogue, considering the criteria below:
[21] 22 3. Demographics of the EGS population
1.- Observed magnitude at the HST F160W band H < 24.5, following the rec- ommendation by van der Wel et al. (2014). The galaxy sizes can be estimated accurately if this limit is considered when fitted in GALFIT.
2.- SExtractor parameter CLASS STAR< 0.9, in order to avoid contamination of the sample by stars.
3.- As suggested by Santini et al. (2015) we excluded the sources with photometric defects, like star spikes and hot pixels, which are tagged as PhotFlag=0.
4.- Stellar masses within the range 9.0 < log (M?/M ) < 11.0.
5.- Well-constrained GALFIT measurements, where we accepted the objects with quality flag value of 0.
The only criterium we will not comply with from Fang et al. (2018) is the redshift selection (z < 2.5), as we wish to explore the redshift evolution of the demographics up to higher redshifts. In Table 3.1 we present the number of galaxies remaining in our sample after the selection criteria have been applied, and the percentage they represent from the entire catalogue. For comparison, we show as well the Fang et al. (2018) sample in UDS, GOODS-S and for both fields combined. Another difference between our study and that of Fang et al. (2018) is the adopted SFR estimation. They did not use the SED fitting results by Santini et al. (2015), instead they used the rest-frame UV (NUV: λ 2800A)˚ luminosity corrected by dust (AV from the ≈ SED fits, where ANUV = 1.8AV ) following equation 3.0.2 for a Kroupa IMF (2001).
We will identify this SFR as SFRUV,corr. They justified this selection by comparing their estimates with the ones presented by other teams in Santini et al. (2015), and concluded they had good agreement.
−1 −10 SFRUV,corr[M yr ] = 2.59 10 LNUV,corr[L ] (3.0.2) ×
Astrophysics Department Instituto Nacional de Astrof´ısica, Optica´ y Electronica´ 3.1 UVJ diagram 23
Table 3.1: Number of galaxies in each field using: F160W < 24.5 mag, PhotFlag=0, Class star< 0.9, 9.0 < logM?/M < 11.0. Both the number of galaxies and percent- ages relate to the whole catalogue for the GOODS-S and UDS fields (taken from Fang et al., 2018), and for the EGS field (this work). Column 1: the selection criteria. Column 2: number of galaxies for the GOODS-S field. Column 3: number of galaxies for the UDS field. Column 4: number of galaxies for the GOODS-S and UDS fields, combined. Column 5: number of galaxies for the EGS field.
Characteristic GOODS-S UDS Combined EGS Full catalogue 34930 35932 70862 41457 (100%) (100%) (100%) (100%) z spec 2658 (6.41%) Selection criteria 4683 5810 10493 7663 (13.4%) (16.2%) (14.2%) (18.5%) Star-forming galaxies 3581 4479 8060 7137 (10.3%) (12.5%) (11.4%) (17.22%) Quiescent galaxies 447 628 1075 526 (1.3%) (1.7%) (1.5%) (1.3%)
3.1 UVJ diagram
We used a color-color rest-frame diagram for our star-forming galaxies selection, with colors U-V and V-J, which differentiates between quiescent and star-forming galaxies, even those reddened by dust (Patel et al., 2012). From now on we will refer to this diagram as UVJ diagram, where star-forming galaxies are red in V-J and quiescent galaxies are blue (Williams et al., 2009). In Figure 3.1 and 3.2 we present the UVJ diagram for the selected sample of galaxies from the CANDELS/EGS catalogue. The color of the dots represents their specific star-formation rate (log sSFR, considering the total SFR estimated by Barro et al., 2011a). The optical counterparts identified by Zavala et al. (2018) in the CANDELS catalogue are marked as downwards red triangles. Dusty star-forming galaxies (DSFGs) with only IR counterparts in the IRAC catalogue are not included, as these do not have reported U, V and J magnitudes. The black dashed line is the selection criteria used to discern between quiescent and star-forming galaxies for up to z = 2, since this is the redshift limit in Williams et al. (2009).
Determination of SFR of SMGs in the EGS 24 3. Demographics of the EGS population
logM logM 0 0 logM 0 0 logM logM