star-forming galaxies - Leiden Observatory - Universiteit Leiden [PDF]

Nearby and distant

star-forming galaxies as seen through emission lines

Maryam Shirazi

ISBN 978–9–46–191906–9 Cover by Alireza Rahmati Front: The star formation rate map of a nearby star-forming galaxy based on the spectral energy distribution fitting. This image is based on Figure 5.8. Back: Astrolabe: An ancient astronomical instrument that shows how the sky looks at a specific place at a given time based on the projection of the celestial sphere onto the plane of the equator. It has been used through the ages for finding the local time and finding the time of celestial events such as sunrise or sunset. The history of the astrolabe goes back to more than two thousand years ago and was highly developed by Muslim astronomers.

Nearby and distant

star-forming galaxies as seen through emission lines

Proefschrift ter verkrijging van de graad van Doctor aan de Universiteit Leiden, op gezag van de Rector Magnificus Prof. mr. C. J. J. M. Stolker, volgens besluit van het College voor Promoties te verdedigen op dinsdag 15 oktober 2013 klokke 10:00 uur

door Maryamosadat Sadatshirazi geboren te Tehran in 1977

Promotiecommissie Promotor: Co-promotor:

Prof. dr. M. Franx Dr. J. Brinchmann

Overige leden:

Prof. dr. H. J. A. R¨ ottgering Prof. dr. A. G. G. M. Tielens Prof. dr. P. van der Werf Dr. S. Charlot (Institute d’Astrophysique, Paris, France)

Table of Contents

1 Introduction 1.1 Nebular physics . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.1 Nebular emission lines in star-forming galaxies . . . . 1.1.2 Measuring galaxy properties using emission line ratios 1.1.3 Emission lines as indirect tracers of massive stars . . . 1.1.4 Emission lines as indirect tracers of the ISM . . . . . . 1.2 Evolution in the properties of star-forming galaxies . . . . . . 1.2.1 Stellar mass evolution . . . . . . . . . . . . . . . . . . 1.2.2 Mass-metallicity evolution . . . . . . . . . . . . . . . . 1.3 Star and galaxy formation over cosmic times . . . . . . . . . 1.3.1 From clumps to bulges . . . . . . . . . . . . . . . . . . 1.4 Towards probing the small-scale properties of distant galaxies 1.5 This thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5.1 Observations . . . . . . . . . . . . . . . . . . . . . . . 1.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

2 Strongly star-forming galaxies in the local universe with nebular He IIλ4686 emission 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Sample selection and classification . . . . . . . . . . . . . . 2.2.2 AGN contamination estimation . . . . . . . . . . . . . . . . 2.3 Physical properties of the sample . . . . . . . . . . . . . . . . . . . 2.3.1 Mass measurements . . . . . . . . . . . . . . . . . . . . . . 2.3.2 Emission line derived parameters . . . . . . . . . . . . . . . 2.4 Model predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.1 CL01 predictions for nebular He II emission . . . . . . . . . 2.4.2 Starburst99 predictions . . . . . . . . . . . . . . . . . . . . 2.4.3 The effect of binary evolution on the He II 4686 emission . 2.5 The origin of nebular He II 4686 emission . . . . . . . . . . . . . .

1 2 2 3 6 7 8 8 9 10 11 11 12 14 15 16

19 20 21 22 28 30 30 31 32 33 35 37 38 v

TABLE OF CONTENTS 2.6 Why are there galaxies with He II 4686 emission but no WR features? 2.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.8 Appendix A: Fitting models to the emission lines . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 The physical nature of the 8 o’clock arc based on near-IR IFU spectroscopy with SINFONI 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1 Near-IR spectroscopy with SINFONI . . . . . . . . . . . . . 3.2.2 Data Reduction . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3 HST Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.4 PSF Estimation . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Analysis of the SINFONI data . . . . . . . . . . . . . . . . . . . . 3.3.1 Nebular emission lines . . . . . . . . . . . . . . . . . . . . . 3.3.2 The integrated Hβ profile . . . . . . . . . . . . . . . . . . . 3.3.3 Spatially-resolved emission-line properties of the 8 o’clock arc in the image plane . . . . . . . . . . . . . . . . . . . . . 3.4 The physical properties of the 8 o’clock arc . . . . . . . . . . . . . 3.4.1 SED fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.2 Parameters derived from emission line modeling . . . . . . 3.4.3 AGN contribution . . . . . . . . . . . . . . . . . . . . . . . 3.4.4 Star formation rate and dust extinction . . . . . . . . . . . 3.4.5 Metallicity and dust-to-gas ratio . . . . . . . . . . . . . . . 3.4.6 The gas surface mass density . . . . . . . . . . . . . . . . . 3.4.7 Electron density . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Source Reconstruction . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.1 Gravitational lens modeling . . . . . . . . . . . . . . . . . . 3.5.2 Reconstructed-Hβ and [O II] emission lines maps in the source plane . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.3 Hβ profile of the reconstructed source . . . . . . . . . . . . 3.6 Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.1 Hβ Kinematics . . . . . . . . . . . . . . . . . . . . . . . . . 3.6.2 Dynamical mass . . . . . . . . . . . . . . . . . . . . . . . . 3.6.3 A massive outflow of gas? . . . . . . . . . . . . . . . . . . . 3.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8 AppendixA: Gaussian decomposition . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Stars were born in significantly denser regions in the verse 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 Metallicity dependence . . . . . . . . . . . . . . . . . . . vi

42 47 57 59 61 62 64 64 64 65 67 67 67 68 69 70 70 73 76 77 79 80 82 83 83 87 88 90 90 91 93 94 97 99

early Uni103 . . . . . . 104 . . . . . . 105 . . . . . . 106 . . . . . . 107 . . . . . . 109

4.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 4.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 5 On the spatial distribution of star formation in distant and nearby galaxies 121 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 5.2 High-z sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 5.2.1 Resolved stellar population modeling . . . . . . . . . . . . . 124 5.2.2 The Voronoi binning of the UDF data . . . . . . . . . . . . 124 5.2.3 Fitting models to the color map of the UDF data . . . . . . 124 5.2.4 Integrated properties . . . . . . . . . . . . . . . . . . . . . . 125 5.3 Low-z sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 5.3.1 Resolved stellar population modeling of low-z sample . . . . 135 5.3.2 Fitting models to the color map of the SDSS data . . . . . 135 5.3.3 Integrated properties . . . . . . . . . . . . . . . . . . . . . . 136 5.4 Structural parameters . . . . . . . . . . . . . . . . . . . . . . . . . 136 5.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 5.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 Nederlandse samenvatting

151

Publications

155

Chapter

1

Introduction Galaxies with all their varieties, have been home to billions of stars during their life. It is because of the presence of these shining stars that we are able to observe them through the cosmic time. Although we observe galaxies mostly through the light emitted by their stars, we cannot resolve these stars individually unless they are very close by. Because of this, the cumulative light from billions of stars in every galaxy is analyzed using stellar population models to extract information about the evolution of galaxies. Stellar light does not reach us without passing through the interstellar medium (ISM) which contains clouds of gas and dust particles. Gas and dust can absorb and re-emit the light from stars, or scatter it towards us and make interpreting what we observe in galaxies very complicated. Despite all these difficulties, just by analyzing the total light from galaxies, we can constrain the global physical properties of galaxies such as stellar mass, star formation rate and age, based on the stellar population models. By combining stellar population models and photoionization models we can further analyze the emission line spectrum of star-forming galaxies coming from ionized gas around young stars which provide us with a wealth of information about the small-scale properties of galaxies e.g., the ISM. This thesis is an attempt in understanding the relation between these small-scale properties and global properties of starforming galaxies over cosmic time using stellar population synthesis models and photoionization models.

Introduction

1.1

Nebular physics

Star formation in galaxies can be traced not only by stars directly, but also through studying their impacts on their surrounding gas. The radiation from stars ionizes the gas around them and produces nebular emission lines as a result of ionized gas recombination. We can measure the star formation rate, along with other galaxy properties, from emission lines coming from both nearby and distant galaxies. In this thesis I extensively use nebular emission lines for measuring the intrinsic properties of galaxies in the nearby and the distant Universe. In the following, I briefly review the physics of ionized gas and mechanisms that produce strong optical emission lines. I also discuss how we can use emission line ratios to constrain various galaxy properties. I continue by discussing that how we can use emission lines to indirectly trace massive stars and also to probe small-scale properties of the ISM.

1.1.1

Nebular emission lines in star-forming galaxies

The optical emission line spectrum contains the Balmer series of Hydrogen (e.g., Hδ, Hγ, Hβ, Hα), together with lines of Helium (e.g., He ii λ4686) and numerous other emission lines such as Oxygen (e.g., [O i] λ6300, [O ii] λ3726, [O iii] λ5007), Nitrogen (e.g., [N ii]λ6584), Sulfur (e.g., [S ii]λ6716). Figure 3.7 shows an example of the emission line spectrum of a nearby actively star-forming galaxy. The source of radiation for producing these emissions in star-forming galaxies is mostly hot, luminous stars with spectral types O and B. OB stars are very massive and shortlived stars with effective temperatures, T e , > 20000 K which enable them to emit ultraviolet radiation at λ 3 ≈

α2B 3Q(t) nH  2 ( ), 4π c3

(1.1)

The ionization parameter which is a measure of intensity of ionizing sources and also Hydrogen number density can be estimated using emission line ratios of high ionization lines to low ionization lines (e.g., [O iii] λ5007/[O ii] λ3727 ratio). Emission line ratios are also used for classifying galaxies in terms of their sources of ionization/excitation (see Figure 2.3). For instance, [O iii] λ5007/Hβ is strongly correlated with hardness of source of ionization/excitation and temperature of ionized gas. Therefore, galaxies with different main ionizing sources are distributed differently in diagnostic diagrams (e.g. the BPT diagram, Bald4

Nebular physics

[OII] 2

2

[OIII] 1

1/2 P3/2

7330

7329

7320

7319

3/2

D5/2

3729 4

S3/2

S0

4363 1

2321

D2

3726

5007

3

P

4959 2 1 0

Figure 1.2 Energy level diagram of the 2p3 ground configuration of singly ionized Oxygen ([O ii] ) and doubly ionized Oxygen ([O iii] ) are illustrated based on Osterbrock & Ferland (2006). Splitting of the ground 3 p is exaggerated for the [O iii] and energy levels in the two diagrams are not comparable. All emissions are in the optical except lines 2321 ˚ A which is in the ultraviolet. All wavelengths shown are in ˚ A.

win, Phillips & Terlevich, 1981) that are based on the line ratios that correlated with the metallicity (e.g., [N ii] λ6584/Hα) and ionization properties (e.g., [O iii] λ5007/Hβ) of galaxies. In star-forming galaxies where the main source of ionization/excitation is star formation, electrons lose their energy more efficiently through optical transitions (strong [O iii] λ5007 or high [O iii] /Hβ) at low metallicity. When the metallicity increases, metal line cooling gets stronger and the electron temperature drops. Thus, electrons evacuate their energies through low ionization lines in infrared (e.g., [O iii] λ88 µm) which makes high ionization lines weaker (weak [O iii] λ5007 or low [O iii] /Hβ). Other sources of ionization such as shock and active galactic nuclei (AGN) can be the source of strong [O iii] λ5007 or high [O iii] /Hβ, at high metallicities. Using these line ratio diagrams, galaxies are generally classified as star-forming, AGN or composite galaxies (Kewley et al., 2001; Kauffmann et al., 2003; Kewley et al., 2013), where the source of ionizing radiation for composite galaxies could be a combination of star formation and AGN or shock. 5

Introduction

Figure 1.3 This plot shows the BPT diagnostic diagram that is used for classifying galaxies in terms of their main source of ionization/excitation. The distribution of emission line galaxies in the SDSS is shown by the colored 2D distribution where the color-scale shows the logarithm of the number of galaxies in each bin. The classification line presented by Kauffmann et al. (2003) is shown as a dashed line, galaxies below this line are star-forming (SF) galaxies. Kewley et al. (2001) classification line is shown as a solid line, galaxies between this line and dashed line are composite galaxies (comp) and galaxies above this line are classified as AGN.

1.1.3

Emission lines as indirect tracers of massive stars

As was mentioned earlier, the main source of ionization for producing strong emission lines in star-forming galaxies is very massive stars. However, our knowledge about these massive stars is limited because direct observations of them often cannot be carried out as these stars are often heavily enshrouded and at low metallicities, they are only found outside the Milky Way. Despite the small fraction of these stars among billions of stars in star-forming galaxies, and their very short life times (e.g., a few million years), they have a significant impact on the galaxy evolution through their hard radiation, strong winds and their explosions as supernovae. At very high energies (e.g. λ < 228 ˚ A) normal OB stars, emit a negligible number of photons. In standard models of massive stars, stars in the Wolf-Rayet (WR) phases have sufficiently hard spectra at these wavelengths. This is, however, 6

Nebular physics a poorly tested assumption in general, and particularly at low metallicities. The ionizing spectrum of WR stars is still subject to significant uncertainty and softer spectra are being predicted in more recent models (Schaerer, 1996; Smith et al., 2002). WR stars are very short lived (e.g., a few Myr) and typically have masses of 10 − 25 M and they are descended from O-type stars (Meynet & Maeder, 2005; Crowther, 2007). In order to produce the observed number ratio of WR to O stars, rotation and binary evolution should be considered in the models of massive stars (see e.g, Brinchmann et al., 2008a). These two effects help removing the outermost atmospheres of stars, thus encouraging the formation of hot WR stars. However, despite an extensive effort in modeling these massive stars, the impact of rotation on the models is still uncertain (Meynet & Maeder, 2005; Heger et al., 2005). The role of binary evolution or single star evolution, especially at low metallicities is also still open for discussion (Han et al., 2007; Eldridge et al., 2009). Although we cannot observe the high energy part of the continuum coming from massive stars due to interstellar absorption we can observe emission lines that are produced by hard ionization from these stars. We can use these high ionization emission lines such as the nebular He ii λ4686 emission line to indirectly trace massive stars. This is a recombination line with ionization potentials of 54.4 eV, coming from ionized gas around very massive stars. Using this, we can probe the high energy part of the spectral energy distribution of very massive stars.

1.1.4

Emission lines as indirect tracers of the ISM

Recent studies have shown that star formation conditions at high redshift (high-z) are different from what we observe in the local Universe. For instance, it has been shown that high-z emission line galaxies are systematically offset from low redshift (low-z) trends in emission line ratio diagrams. This is seen particularly well in the BPT diagram, log [N ii]/Hα vs. [O iii]/Hβ diagram (Brinchmann et al., 2008b; Liu et al., 2008). In this diagram, high-z star-forming galaxies unlike low-z ones are distributed in the regions that need other sources of ionization/excitation than star formation (e.g., Shapley et al., 2005; Erb et al., 2006; Newman et al., 2013). Kewley et al. (2013) recently studied the cosmic evolution of the BPT diagram and showed that the extreme ISM conditions at high-z cause this offset between distant and nearby galaxies in the BPT diagram. Therefore, we can use the emission line intensities of distant galaxies to indirectly trace the ISM at high-z and to study the evolution of intrinsic physical properties of star-forming galaxies from distant to nearby Universe. However, studying the evolution of the physical conditions at which stars are forming has proven very challenging and is hidden in the strong evolution of global mean properties of galaxies such as stellar mass and SFR. In the next section, I briefly summarize these evolutions in the global properties of star-forming galaxies from distant to nearby Universe. 7

Introduction

1.2

Evolution in the properties of star-forming galaxies

The average integrated properties of star-forming galaxies have evolved significantly during the last ∼ 12 Gyr. When the Universe was only 2-3 Gyr old (redshift, z ∼ 3 − 2), star formation in typical galaxies was happening at a rate that today is only found in the most extreme star-forming galaxies (Brinchmann et al., 2004). The star formation rate of the Universe within a comoving volume element as a function of redshift was first presented by the Madau plot (Madau et al., 1996) and confirmed by further observations (see Figure 1.4 that shows the evolution of the SFR density of Universe with redshift, data are taken from Hopkins & Beacom, 2006). These very high star formation rates have been measured based on rest-frame UV emission from young stars in high-z galaxies (e.g., Noeske et al., 2007; Daddi et al., 2007) or infrared observations that determine the contribution of obscured light to the SFR of high-z galaxies (e.g., Elbaz et al., 2007, see also Shapley, 2011 for other methods used for estimating SFR of distant galaxies). A strong correlation between SFR and stellar mass is observed at high-z known as the star-forming main sequence (Noeske et al., 2007; Daddi et al., 2007). This tight main sequence locus evolves smoothly with redshift showing that galaxies with the same stellar mass at low-z and high-z have higher SFR at high-z (see Bouch´e et al., 2010). The high SFR of these galaxies implies they are much more gas rich than local starforming galaxies. High gas fractions (several times higher than what we observe in the local galaxies) also were observed for some of these high-z galaxies (e.g., Daddi et al., 2008; Tacconi et al., 2010; Genzel et al., 2010; Tacconi et al., 2013). A more clumpy morphology has been observed for many of high-z star-forming galaxies (e.g., Elmegreen & Elmegreen, 2006; Genzel et al., 2011; Wuyts et al., 2012). These clumpy structures can be caused from gravitational instability within these very gas rich disks at high-z.

1.2.1

Stellar mass evolution

Stellar population synthesis models can be used to estimate the global physical properties of galaxies such as stellar mass. Based on stellar population synthesis models (e.g., Bruzual & Charlot, 2003), a combination of optical broadband photometry and spectral indices (the 4000˚ A spectral break and the strength of Balmer absorption lines) can be modeled to measure stellar masses in nearby Universe (e.g., Kauffmann et al., 2003). At higher redshifts, however, due to lower signalto-noise, stellar absorption features are difficult to measure and only broadband photometry tends to be used for measuring stellar masses. The rest-frame near-IR luminosity is more closely tied to stellar mass than the rest-frame optical luminosity (Bell & de Jong, 2001). The rest-frame UV emission from distant galaxies can only probe their massive stars light. Therefore, these data should be combined with longer wavelengths observations to probe older stellar populations of high-z galaxies. With availability of these observations for many of high-z galaxies in recent years, the global evolution of the stellar content in galaxies from the 8

Evolution in the properties of star-forming galaxies

Figure 1.4 The SFR density of the Universe as a function of redshift. The data have been taken from Hopkins & Beacom (2006). distant to the nearby Universe has become possible to estimate with good accuracy. The evolution in stellar mass can be described by constructing the galaxy stellar mass function at a range of redshifts. Based on recent observations in the COSMOS/ULTRAVISTA survey which confirms previous measurments, a strong evolution in stellar mass has been observed from z = 4 to z = 0.2 (e.g., Ilbert et al., 2013; Muzzin et al., 2013). These studies show the mass density of star-forming galaxies grows by a factor of 1.59 since z = 3.5 and the typical mass of a galaxy of Log(M∗ /M ) = 10.5 at z = 0.3 would be Log(M∗ /M ) . 9.5 at z = 2.

1.2.2

Mass-metallicity evolution

As galaxies evolve, they form more stars and their gas content is converted into stars and their metal content increases. Thus, stars made out of material that has been enriched for many generations will be more metal-rich. Some of the metals that are produced will be ejected out of the galaxy through outflows into the intergalactic medium (IGM). Galaxies also accrete some gas from the IGM. Gas metallicities are derived from emission line properties and stellar metallicities are derived from Lick indices (Faber, 1973). Based on these metallicity measurements, we can study the evolution in the metallicity of galaxies from distant to nearby Universe. There is an evolution in the relation between stellar mass and metallicity known as mass-metallicity relation as we look back in time. This evolution shows that galaxies with the same stellar mass have lower metallicities at high-z compared to 9

Introduction similar galaxies in the local Universe. However, Mannucci et al. (2010); Lara-L´opez et al. (2010) found that when including the SFR, mass-metallicity-SFR relation holds up to z ∼ 3 which means galaxies with the same stellar mass have higher SFR when they show lower metallicity. There should be also a more fundamental relation between atomic gas mass, SFR and metallicity as it is observed for local galaxies (Bothwell et al., 2013). This suggests that the reason for having high SFR at low metallicity is because of having more gas. However, because of the lack of enough atomic gas data available for high-z galaxies (e.g., H i gas cannot be observed at z > 0.4 with current instrumentation, and the molecular, i.e. H2 , contents of high-z galaxies are estimated from CO observations), this has not been studied yet.

1.3

Star and galaxy formation over cosmic times

After the Big Bang, the Universe had no stars but was filled with only gas and dark matter. Dark matter perturbations in the early universe grew gravitationally and ended up as galaxy dark matter haloes. The gas which is bound to dark matter haloes radiates its energy away and cools down. Because of the conservation of angular momentum, this collapsing gas forms a rotating disk within which smallscale instabilities could grow to form molecular clouds. These molecular clouds have been the birth place for most stars. Gravitational instability with the critical density that is set by turbulence from stellar feedback (both negative and positive through heating of gas, and compressing it) is believed to determine star formation processes. Although we know the formation of stars to the first order, we do not know well the physical processes that control the interplay between gas and stars. Therefore, understanding the star formation history over cosmic time remains a major theoretical and observational challenge. Understanding how galaxies were assembled across the cosmic time also remains a challenging question. For instance, how today’s Hubble Sequence with different galaxy morphologies is shaped and which physical processes can constrain and control the galaxy evolution, are important questions to be addressed. In the standard scenario, galaxies are believed to form as disk galaxies, which can then be transformed into ellipticals mainly due to major mergers. If new gas from the merger remnants is able to cool then new disks can form and this process can make disk-bulge systems (e.g., Kauffmann et al., 1993; Baugh et al., 1996). However, galaxies at the peak of star formation in the Universe show very distinct features, such as clumpy morphologies. The nature of these clumps and their evolution determines whether host galaxies have inside-out growth and form bulges from migration of these clumps towards the center. Another important question in this regard is what fuels star formation. Hydrodynamical cosmological simulations predict that at high-z, gas accretion plays a significant role for fueling star formation (Kereˇs et al., 2005). The existence of disk-like kinematics in star-forming galaxies during the peak of star formation suggests that gas accretion is the dominant process for growth of galaxies. In the local Universe, however, mergers are believed to play a more dominant role. 10

Towards probing the small-scale properties of distant galaxies

1.3.1

From clumps to bulges

Small-scale instabilities in a rotationally supported gaseous disk are unstable against gravitational collapse and can grow if the Toomre stability parameter (Toomre 1964), Qgas < 1 where the Q parameter for stability of a disk is: Qgas =

κσ πGΣgas

(1.2)

where σ is velocity dispersion and Σgas is the gas mass density. κ = a v/R is the epicyclic frequency where a is a dimensionless factor 1 < a < 2 depending on the rotational structure of the disk, v is the circular velocity and R is the radius. A clump of gas that is large enough, i.e. larger than the Jeans length L J ' σ2 /GΣgas , can collapse under its self-gravity despite its velocity dispersion. Because of its rotation within the disk this clump experiences an outward centrifugal acceleration ' L J κ2 ; if this acceleration is larger than the gravitational acceleration, GΣgas , then the disk is stable. It is widely accepted that the majority of gas clumps form from gravitational instability with Jeans scale of ≈1 kpc. Clumpy galaxies at high-z usually show disklike kinematics with high turbulent Hα and CO velocity dispersions (e.g., Genzel et al., 2006; Epinat et al., 2012; Tacconi et al., 2013). From a theoretical point of view, violent disk instabilities and high velocity dispersions are required to regulate disks with a Toomre parameter Qgas ≈ 1 (Dekel et al., 2009). Observationally, the Qgas ≈ 1 instability limit has been estimated for gas within high-z galaxies (see Genzel et al., 2011). However, local spiral disks tend to have Qgas ∼ 2 (van der Kruit & Freeman, 1986) and they cannot form gas clumps. The formation and evolution of clumps has an important impact on the formation of the central bulges in galaxies. However, it is not known yet whether they can survive stellar feedback long enough and migrate inward to build the central bulge, or whether these clumps disrupt like the molecular clouds. Recent high-resolution optical and near-infrared images and spatially-resolved observations allow us to study the stellar populations of these kpc-size clumps within high-z galaxies.

1.4

Towards probing the small-scale properties of distant galaxies

The last decade has seen a dramatic increase in our knowledge about galaxy population at z ∼ 1–3 (e.g., Shapley , 2011). This has been achieved mostly by studying the integrated properties of galaxies. However, recently using near-IR integral field unit (IFU) observations with adaptive optics (AO), many high-z galaxies have been resolved spatially. The steadily growing effort to obtain resolved near-IR spectra of high-z galaxies in a systematic manner, such as the SINS, MASSIV and LSD/AMAZE surveys (F¨ orster Schreiber et al., 2006; Contini et al., 2012; Maiolino et al., 2010), is leading to large samples of spatially-resolved emission line maps of distant star-forming galaxies. These maps provide us with spatially-resolved 11

Introduction galaxy properties such as metallicity and SFR at high-z. Based on these studies, we know a large fraction (30% or larger) of these galaxies are dominated by rotating disk kinematics with an increased fraction towards higher stellar masses (e..g., F¨ orster Schreiber et al., 2009, see their Figure 17 for kinematics of some SINS galaxies) and they show high turbulent Hα velocity dispersions (e.g., Genzel et al., 2006). Metallicity gradients were measured within these galaxies based on spatially-resolved emission line maps and it was discovered that some of these galaxies show a lower metallicity in their centers (Cresci et al., 2010) as opposed to what we observe in the local Universe for rotating disks systems, see Glazebrook (2013) for a review on kinematic studies of star-forming galaxies across cosmic time. However, obtaining sub-kpc resolution, even with AO observations, is very difficult because of the intrinsic faintness and the small sizes of high-z galaxies. Our limitation to reach sub-kpc scale resolution for high-z galaxies using current instruments can be resolved by observing high-z galaxies which are significantly magnified due to gravitational lensing. This allows us to study the properties of those high-z galaxies at a level similar to what is achieved at lower redshifts (e.g., Yee et al., 1996; Pettini et al., 2000, 2002; Teplitz et al., 2000; Savaglio et al., 2002; Siana et al., 2008). However, even the ∼100 pc resolution achieved for some lensed galaxies at high-z (e.g., Swinbank et al., 2009; Jones et al., 2010) is not enough to study the small-scale properties of the ISM in high-z galaxies at the same level that we can study those in the local Universe (e.g., Kennicutt & Evans, 2012). The Atacama Large Millimeter/submillimeter Array (ALMA) observations will soon provide us with insightful information about sub-kpc scale kinematics and distribution of gas and star formation within distant galaxies. Afterwards, integral field spectroscopic capability with the James Webb Space Telescope (JWST) will allow us to accurately map distant galaxies in emission and absorption. This will significantly change our view about the resolved small-scale properties of high-z galaxies in the coming decade.

1.5

This thesis

In this thesis I analyze emission lines from gas ionized by very massive stars in nearby and distant star-forming galaxies. In the local Universe, based on these emission lines we can determine the source of ionization for producing them at different environments. At high-z, these emission lines provide us with information about the small-scale properties of the ISM. A brief summary of the contents of this thesis is given below. Chapter 2: Strongly star-forming galaxies in the local universe with nebular He ii λ4686 emission The evolution of massive stars is a complex and not fully understood process. While we are limited by interstellar absorption in observing the stellar continuum at λ < 228 ˚ A, using the nebular He ii λ4686 emission line gives us valuable information about this high energy part of the stellar spectral energy distributions. Only the most extreme star-forming galaxies show nebular He ii emission and it 12

This thesis is generally believed that Wolf-Rayet (WR) stars provide the required ionizing radiation for them. In Chapter 2, we study the physical properties of emission line galaxies in the SDSS showing He ii emission. Based on these data, we find that the He ii is not associated with WR features in a large number of star-forming galaxies with this emission at low metallicities. This lack of WR stars has important implications for the evolution of the most massive stars at low metallicity. Non-homogenous stellar evolution models (e.g., Yoon et al., 2006) and spatial offset between the location of WR stars and He ii regions (e.g., Kehrig et. al, 2008) might be two possible explanations for this discrepancy. We also show the current stellar population models cannot produce observed He ii /Hβ ratios in low metallicity environments. This result has implications for interpreting observations of high-z galaxies where the metallicity is expected to be typically lower. Another key result from this study is to define a new diagnostic diagram using the He ii /Hβ ratio, which can be used to constrain AGN contribution in star-forming galaxies showing He ii emission. Chapter 3: The physical nature of the 8 o’clock arc based on near-IR IFU spectroscopy with SINFONI The detailed analysis of distant galaxies is limited by their small angular sizes and faint apparent magnitudes. Both of these limitations can be overcome by observing gravitationally lensed galaxies. In Chapter 3, we analyze spatiallyresolved data of the 8 o’clock arc, a lensed Lyman break galaxy, in conjunction with HST imaging of this galaxy, from which the lens model for the galaxy was reconstructed. Based on this lens modeling, the de-lensed Hβ map, velocity and velocity dispersion maps are reconstructed. We show a simple rotating disk model is unable to fit the observed velocity field of the galaxy, and a more complex velocity field is needed. The Hβ profile of the galaxy shows a broad blueshifted wing, suggesting an outflow of 200 km/s. The estimated gas surface density and gas mass of the 8 o’clock arc shows a factor of 2.5-7 higher gas content compared to similar galaxies in the SDSS. Chapter 4: Stars were born in significantly denser regions in the early Universe Most of stars that surround us today were formed several billion years ago, around the peak of star formation activity in the Universe. The conditions under which these stars were born is of great interest but very difficult to study due to limited observational resolution for distant objects. In Chapter 4, we present a novel approach to directly compare the density in the star-forming regions of galaxies that are near the peak of star formation activity in the Universe to those of nearby galaxies. To indirectly trace the ISM at high-z, we use the emission line intensities of distant galaxies. We calibrate a new relation between the [O iii] λ5007/[O ii] λ3727 emission line ratio and ionization parameter to estimate the difference between the ionization parameters in the high and low-z samples. We analyze the ionization properties of a sample of high-z galaxies at redshift 2.6–3.4, including the 8 o’clock arc, and compare them with that of galaxies with similar physical properties in the local Universe. We show that after accounting 13

Introduction for all differences in large-scale properties, such as mass and specific star formation rate, the density in the star-forming regions was eight times higher in the past. This implies that the majority of stars in the Universe were formed in gas that obeyed very different scaling relations than what we see in the present day Universe. This is a striking result that provides strong constraints on the conditions of star formation in normal galaxies in the early Universe. Chapter 5: On the spatial distribution of star formation in distant and nearby galaxies In Chapter 5, we study the differences between the spatial distribution of star formation in the distant and the nearby Universe for galaxies with the similar global properties (e.g., stellar mass and specific star formation rate). We use the multi-band imaging data available in the HUDF and compare this quantitatively with the low-z data from the SDSS. Based on this, we study the physical processes that cause clumpy star formation distribution for galaxies with similar star formation activity at low-z and high-z. We compare the resolved stellar populations of these galaxies by measuring the structural parameters of distant galaxies and their nearby counterparts. We show galaxies at high-z have more concentrated stellar content but their star formation is more extended compared to galaxies with the same global properties at z∼0. We show high-z galaxies are more clumpy in their star formation distributions than their local analogs. This clumpy morphology suggests that distant galaxies need to have more surface density of the disk compared to their local analogs.

1.5.1

Observations

In this thesis I use emission line intensities of a sample of distant galaxies from the literature (Maiolino et al., 2008; Mannucci et al., 2009; Richard et al., 2011; Dessauges-Zavadsky et al., 2011). Analyzing near-IR IFU spectroscopy of the 8 o’clock arc, a lensed Lyman break galaxy at redshift 2.735 is also added to my high-z studies. These data were taken with SINFONI on VLT covering λ = 2900 ˚ A to 6500 ˚ A in the rest-frame. The SINFONI data are analyzed in conjunction with the HST images of the galaxy. The low-z studies in this thesis are based on the Sloan Digital Sky Survey (SDSS) (York et al., 2000) data. The MPA-JHU1 value added catalogues (Brinchmann et al., 2004; Tremonti et al., 2004) for SDSS DR7 (Abazajian et al., 2009) are used and star-forming galaxies following Brinchmann et al. (2004) are selected. Furthermore, SDSS DR8 (Aihara et al., 2011) photometry are used to estimate stellar masses in Chapter 4. In this thesis I also use the multi-band imaging data available in the Hubble Ultra Deep Field (HUDF) to study the spatial distribution of star formation in some HUDF galaxies that have confirmed spectroscopic redshifts2 (Coe et al., 2006) . 1 http://www.mpa-garching.mpg.de/SDSS/DR7 2 http://adcam.pha.jhu.edu/~coe/UDF/paper/zspec.cat

14

Summary

1.6

Summary

This thesis aims at studying the physical properties of galaxies from the distant to the nearby Universe based on their emission line observations. We study these properties using Charlot & Longhetti (2001, CL01) models which combine Bruzual & Charlot (2003) stellar population models with CLOUDY photoionization models (Ferland et al., 1998). We use high ionization emission lines to probe the high energy part of the stellar SEDs at low metallicities and constrain current stellar populations. We show that stellar population models need to consider harder spectra for O type stars in order to explain observations of high ionization emission lines such as He ii λ4686. These results can be further confirmed by studying the spatially-resolved analysis of He ii galaxies. Recent observations have shown that high-z star-forming galaxies form a different population compared to star-forming galaxies in the local Universe. These studies, however, cannot tell if the main difference between low-z and high-z starforming galaxies is related to their strongly evolving global properties (e.g., mass, SFR) or their different intrinsic properties (e.g., the ISM). Based on the CL01 models and emission line ratios, we probe the intrinsic properties of high-z galaxies that show no evolution in their global properties compared to a sample of nearby star-forming galaxies. Using this, we show that star-forming regions are denser at high-z and also there is an evolution in the relation between the surface density of gas and the surface density of SFR known as the star-formation law towards less efficient relation at high-z. Emission line observations for a larger sample of star-forming galaxies at high-z and also follow up observations of emission line ratios that are density tracers can confirm this higher density at high-z. We compare also the distribution of star formation between distant and nearby galaxies with similar physical properties based on their deep imaging. We show that stellar content of distant star-forming galaxies is more compact than their local analogs. The same result has been shown before for elliptical galaxies in the local and high-z Universe. To do a more statistical analysis our study should be done for a larger sample of star-forming galaxies at z > 1.5.

15

REFERENCES

References Abazajian, K. N., Adelman-McCarthy, J. K., Ag¨ ueros, M. A., et al. 2009, ApJS, 182, 543 Aihara, H., Allende Prieto, C., An, D., et al. 2011, ApJS, 193, 29 Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 5 Baugh, C. M., Cole, S., & Frenk, C. S. 1996, MNRAS, 283, 1361 Bell, E. F., & de Jong, R. S. 2001, ApJ, 550, 212 Bothwell, M. S., Maiolino, R., Kennicutt, R., et al. 2013, MNRAS, 433, 1425 Bouch´e, N., Dekel, A., Genzel, R., et al. 2010, ApJ, 718, 1001 Brinchmann, J., Charlot, S., White, S. D. M., et al. 2004, MNRAS, 351, 1151 Brinchmann, J. ,Kunth, D., Durret, F., 2008, A&A, 485, 657 Brinchmann, J., Pettini, M., & Charlot, S. 2008b, MNRAS, 385, 769 Bruzual, G. ,Charlot, S., 2003, MNRAS, 344, 1000 Charlot, S., & Longhetti, M. 2001, MNRAS, 323, 887 Coe, D., Ben´ıtez, N., S´ anchez, S. F., et al. 2006, AJ, 132, 926 Contini T., et al., 2012, A&A, 539, 91 Cowie, L. L., Hu, E. M., & Songaila, A. 1995, AJ, 110, 1576 Cresci, G., Mannucci, F., Maiolino, R., et al. 2010, Nature, 467, 811 Crowther, P. A., 2007, ARA&A, 45, 177 Daddi, E., Dickinson, M., Morrison, G., et al. 2007, ApJ, 670, 156 Daddi, E., Dannerbauer, H., Elbaz, D., et al. 2008, ApJL, 673, L21 Dekel, A., Birnboim, Y., Engel, G., et al. 2009, Nature, 457, 451 Dessauges-Zavadsky, M., Christensen, L., D’Odorico, S., et al. 2011, A&A, 533, A15 Elbaz, D., Daddi, E., Le Borgne, D., et al. 2007, A&A, 468, 33 Elbaz, D., Dickinson, M., Hwang, H. S., et al. 2011, A&A, 533, A119 Eldridge, J. J., Stanway, E. R., 2009, MNRAS, 400, 1019 Elmegreen, B. G., & Elmegreen, D. M. 2006, ApJ, 650, 644 Epinat, B., Tasca, L., Amram, P., et al. 2012, A&A, 539, A92 Erb, D. K., Shapley, A. E., Pettini, M., Steidel, C. C., et al. 2006, ApJ, 644, 813 Faber, 1973, ApJ, 179, 731 Ferland, G. J., Korista, K. T., et al., 1998, PASP, 110, 761 F¨ orster Schreiber N., et al., 2006, Msngr, 125, 11 F¨ orster Schreiber, N. M., Genzel, R., Bouch´e, N., et al. 2009, ApJ, 706, 1364 F¨ orster Schreiber, N. M., Shapley, A. E., Genzel, R., et al. 2011, ApJ, 739, 45 Genzel, R., Newman, S., Jones, T., et al., 2011, ApJ, 733, 101 Genzel, R., Tacconi, L. J., Gracia-Carpio, J., et al. 2010, MNRAS, 407, 2091 Genzel, R., Tacconi, L. J., Eisenhauer, F., et al. 2006, Nature, 442, 786 Jones, T. A., Swinbank, A. M., Ellis, R. S., et al. 2010, MNRAS, 404, 1247 Han, Z., Podsiadlowski, P., & Lynas-Gray, A. E. 2007, MNRAS, 380, 1098 Heger, A., Woosley, S. E., & Spruit, H. C. 2005, ApJ, 626, 350 Hopkins, A. M., & Beacom, J. F. 2006, ApJ, 651, 142 Ilbert, O., McCracken, H. J., Le Fevre, O., et al. 2013, arXiv:1301.3157 Kauffmann, G., Heckman, T.M., et al. 2003, MNRAS, 346, 1055 Kauffmann, G., White, S. D. M., & Guiderdoni, B. 1993, MNRAS, 264, 201 16

REFERENCES Kehrig, C., V´ılchez, J. M., et al., D., 2008, A&A, 477, 813 Kennicutt, Jr., R. C. 1998, ApJ, 498, 541 Kennicutt, R. C., & Evans, N. J. 2012, ARA&A, 50, 531 Kewley, L. J., Maier, C., Yabe, K., et al. 2013, arXiv:1307.0514 Kewley, L. J., & Dopita, M. A. 2002, ApJS, 142, 35 Kewley, L. J., Dopita, M. A., et al., 2001, ApJ, 556, 121 Kereˇs, D., Katz, N., Weinberg, D. H., & Dav´e, R. 2005, MNRAS, 363, 2 Glazebrook, K. 2013, arXiv:1305.2469 Lara-L´ opez, M. A., Cepa, J., Bongiovanni, A., et al. 2010, A&A, 521, L53 Liu, X., Shapley, A. E., Coil, A. L., et al. 2008, ApJ, 678, 758 Madau et al. 1996, MNRAS, 283, 1388 Maiolino R., et al., 2010, Msngr, 142, 36 Maiolino, R., Nagao, T., Grazian, A., et al. 2008, A&A, 488, 463 Mannucci, F., Cresci, G., Maiolino, R., Marconi, A., & Gnerucci, A. 2010, MNRAS, 408, 2115 Mannucci, F., Cresci, G., Maiolino, R., et al. 2009, MNRAS, 398, 1915 Meynet, G., & Maeder, A. 2005, A&A, 429, 581 Muzzin, A., Marchesini, D., Stefanon, M., et al. 2013, arXiv:1303.4409 Noeske, K. G., Weiner, B. J., Faber, S. M., et al. 2007, ApJL, 660, L43 Newman, S. F., Buschkamp, P., Genzel, R., et al. 2013, arXiv:1306.6676 Osterbrock, D. E., & Ferland, G. J. 2006, Astrophysics of gaseous nebulae and active galactic nuclei, 2nd. ed. by D.E. Osterbrock and G.J. Ferland. Sausalito, CA: University Science Books, 2006 Penston, M. V., Robinson, A., Alloin, D., et al. 1990, A&A, 236, 53 Pettini, M., Rix, S. A., Steidel, C. C., et al. 2002, ApJ, 569, 742 Pettini, M., Steidel, C. C., Adelberger, K. L., et al. 2000, ApJ, 528, 96 Richard, J., Jones, T., Ellis, R., et al. 2011, MNRAS, 413, 643 Savaglio, S., Panagia, N., & Padovani, P. 2002, ApJ, 567, 702 Schaerer, D., Vacca, W. D., 1998, ApJ, 497, 618 Schaerer, D. 1996, ApJ, 467, 17 Siana, B., Teplitz, H. I., Chary, R.-R., et al. 2008, APJ, 689, 59 Shapley A., 2011, ARA&A, 49, 525 Shapley, A. E., Steidel, C. C., Erb, D. K., et al. 2005, ApJ, 626, 698 Shirazi, M., Brinchmann, J., & Rahmati, A. 2013, arXiv:1307.4758 Smith, L. J., Norris, R. P. F., Crowther, P. A., 2002, MNRAS, 337, 1309 Swinbank, A. M., Webb, T. M., Richard, J., et al. 2009, MNRAS, 400, 1121 Tacconi, L. J., Genzel, R., Neri, R., et al. 2010, Nature, 463, 781 Tacconi, L. J., Neri, R., Genzel, R., et al. 2013, ApJ, 768, 74 Teplitz, H. I., McLean, I. S., Becklin, E. E., et al. 2000, ApJ, 533, L65 Tremonti, C. A., Heckman, T. M., Kauffmann, G., et al. 2004, ApJ, 613, 898 van der Kruit, P. C., & Freeman, K. C. 1986 Wuyts, S., F¨ orster Schreiber, N. M., Genzel, R., et al. 2012, ApJ, 753, 114 Yee, H. K. C., Ellingson, E., Bechtold, J., et al. 1996, AJ, 111, 1783 Yoon, S.-C., Langer, N., Norman, C., 2006, A&A, 460, 199 York, D. G., Adelman, J., Anderson, Jr., J. E., et al. 2000, AJ, 120, 1579

17

Chapter

2

Strongly star-forming galaxies in the local universe with nebular He IIλ4686 emission We present a sample of 2865 emission line galaxies with strong nebular He ii λ4686 emissions in Sloan Digital Sky Survey Data Release 7 and use this sample to investigate the origin of this line in star-forming galaxies. We show that star-forming galaxies and galaxies dominated by an active galactic nucleus form clearly separated branches in the He ii λ4686/Hβ versus [N ii] λ6584/Hα diagnostic diagram and derive an empirical classification scheme which separates the two classes. We also present an analysis of the physical properties of 189 star forming galaxies with strong He ii λ4686 emissions. These star-forming galaxies provide constraints on the hard ionizing continuum of massive stars. To make a quantitative comparison with observation we use photoionization models and examine how different stellar population models affect the predicted He ii λ4686 emission. We confirm previous findings that the models can predict He ii λ4686 emission only for instantaneous bursts of 20% solar metallicity or higher, and only for ages of ∼ 4 − 5 Myr, the period when the extreme-ultraviolet continuum is dominated by emission from Wolf-Rayet stars. We find however that 83 of the star-forming galaxies (40%) in our sample do not have Wolf-Rayet features in their spectra despite showing strong nebular He ii λ4686 emission. We discuss possible reasons for this and possible mechanisms for the He ii λ4686 emission in these galaxies. Maryam Shirazi, Jarle Brinchmann Monthly Notices of the Royal Astronomical Society Volume 421, Issue 3, pp. 1043-1063 (2012)

Star-forming galaxies with nebular He II 4686 emission

2.1

Introduction

The ionizing continuum of stars at λ < 912 ˚ A is of major importance for interpreting emission line observations of galaxies because many of the strong lines observed in the spectra of galaxies, such as [O iii] λ5007, [Ne iii] λ3869 and He ii λ4686, have ionization potentials in excess of 13.6 eV. Despite this importance we are severely limited by interstellar absorption in observing stellar spectra in this spectral window directly (e.g. Hoare et al., 1993). Although we can get direct information at slightly longer wavelengths with space-based UV spectroscopy (e.g. Crowther et al., 2002), most of our knowledge about the λ < 912˚ A region is based on indirect evidence, even for solar metallicity. A promising way to indirectly obtain information on the stellar ionizing continuum is to compare emission line properties (e.g. flux, equivalent width) to predictions from photoionization codes such as CLOUDY (Ferland et al., 1998) or MAPPINGS III (Allen et al., 2008). In practice these kinds of studies provide modest constraints on stellar atmosphere models (e.g. Crowther et al., 1999). However where predictions of models differ significantly, this approach can yield useful information. This is the approach we will adopt in this work, where we will make use of the He ii λ4686 nebular emission line to place constraints on stellar models and in particular on the ionization mechanism for this line. The presence of a nebular He iiλ4686 line in the integrated spectrum of a galaxy indicates the existence of sources of hard ionizing radiation as the ionization energy for He+ is 54.4 eV (λ ≈ 228 ˚ A). This hard radiation can of course be produced by an active galactic nucleus (AGN), and most sources with luminous He ii emission, in a flux limited sample, are indeed galaxies with an AGN1 . However the required hard radiation can also be provided by stellar sources and He ii λ4686 emission is frequently seen in H ii-galaxies. The line appears to be associated with young stellar populations; for instance, Bergeron et al. (1997) proposed Of stars as the sources of He iiλ4686 emission in dwarf galaxies. Subsequent discussion has mostly focused on Wolf-Rayet (WR) stars, although the distinction between these two classes is rather blurred (e.g. Gr¨ afener et al., 2011). Schaerer (1996, see also Schaerer & Vacca (1998)) showed that the hard radiation field of WR stars could provide a good explanation of the nebular He ii λ4686 seen in H ii-galaxies. Guseva et al. (2000) tried to test this in a careful study of H ii-galaxies with prominent WR features. They were however unable to find WR features in 12 out of the 30 galaxies with nebular He ii λ4686 emission. The same lack of WR features in metal poor Blue Compact Dwarf (BCD) galaxies was pointed out by Thuan & Izotov (2005). They proposed that fast radiative shocks could be responsible for this emission (see also Garnett et al. 1991). Similar results were reported by Brinchmann et al. (2008, hereafter B08), who analyzed a sample of strong emission line galaxies in the Sloan Digital Sky Survey (SDSS, York et al., 2000) with He ii λ4686 emission. They showed that at least at metallicities of 12 + log O/H > 8, there appeared to be a close correlation between WR features in galaxies and the presence of He iiλ4686 emission, but this appeared 1 We will here not distinguish between the host galaxy and its nuclear power source so will refer to these galaxies as AGNs

20

Data not to be so clear-cut at lower metallicities. This apparent lack of connection of He ii emission with the hard UV radiation from WR stars has also been seen in spatially resolved spectroscopic studies. Kehrig et. al (2008) performed an integral field spectroscopy study for the H ii galaxy II Zw 70 and found that the region associated with nebular He ii λ4686 emission was a few arcsec offset from the region with detected WR features. More recently Kehrig et. al (2011) and Neugent & Massey (2011) have presented studies of He ii λ4686 emission in M33. Both studies find some regions with nebular He ii λ4686 emission that are not associated with WR stars (see also Hadfield & Crowther 2007, L´ opez-S´ anchez & Esteban 2010 and Monreal-Ibero et al. 2010). Thus a series of studies have shown that while He iiλ4686 emission frequently is found in association with WR stars, it appears not to be so in all cases, particularly at low metallicity. As mentioned above, possible additional sources of high energy photons could be X-ray binaries (Garnett et al. 1991), strong shocks (Dopita & Sutherland, 1996), low-level AGN activity and alternative models for stellar evolution (Yoon & Langer, 2005). However the existing studies do not show clear trends that allow us to distinguish between these scenarios. Crucially the samples in most of the previous studies have not been selected specifically to study He ii emission lines. To make progress in understanding this puzzle it is important to have as large as possible sample of He ii emitting galaxies to allow one to study the relationship between He ii emission and other physical properties. To this end we present here an analysis of emission line galaxies with strong He ii λ4686 emission in SDSS Data Release 7 (DR7, Abazajian et al., 2009). In section 3.2 we discuss the sample selection and carefully account for AGN emission. The physical properties of the He ii emitting galaxies are discussed in section 2.3 and the observed He ii λ4686/Hβ ratios are compared to model predictions in section 2.4. In section 2.5 we test these model predictions and investigate whether the presence of He ii λ4686 is associated with WR features. We find that low metallicity systems frequently do not show signs of WR stars. We discuss possible explanations for this finding in section 2.6 and conclude in section 5.6.

2.2

Data

Our sample is based on galaxy spectra from SDSS DR7 which cover a wavelength range of 3800-9200 ˚ A. The spectra were analyzed using the methodology discussed in Tremonti et al. 2004 (see also Brinchmann et al. 2004) to provide accurate continuum subtraction. All emission line sources were additionally analyzed using the pipeline discussed in B08 to measure a wider gamut of emission lines. For each galaxy we measure 40 lines, These lines and the number of spectra that show these lines with S /N > 5.5 are summarized in Table 2.1. Our concern in this paper is not to analyze a volume- or magnitude-limited sample of galaxies, we therefore do not impose a redshift cut nor a magnitude limit. Since the blue wavelength cut-off of the SDSS spectrograph is ∼ 3800˚ A, the [O ii] λ3727, 3729 doublet falls outside the spectral range for z < 0.02. This is a concern, because as we will see later 55% of our final sample fall in this region 21

Star-forming galaxies with nebular He II 4686 emission

Measured line [O ii] λ3726, 3729 [Ne iii] λ3869 H8 [Ne iii] λ3967 H He i λ4026 [S ii] λ4069 Hδ Hγ [O iii] λ4363 He i λ4472 [Fe iii] λ4658 He ii λ4685 [Ar iv] λ4711 [Ar iv] λ4740 Hβ [O iii] λ4959 [O iii] λ5007 [N i] λ5197 [N i] λ5200 [Cl iii] λ5518 [Cl iii] λ5538 [N ii] λ5755 He i λ5876 [O i] λ6300 [S iii] λ6312 [O i] λ6363 [N ii] λ6548 Hα [N ii] λ6584 He i λ6678 [S ii] λ6717 [S ii] λ6731 He i λ7065 [Ar iii] λ7135 [O ii] λ7318, 19, 29, 30

Number of spectra 243977 39143 92230 18367 64120 6368 1503 163534 295528 7886 15563 2626 4034 2893 1074 458324 147734 339212 1318 758 160 190 4888 75256 119328 5552 16100 361641 613338 562659 19084 410903 341386 4232 24034 659

Fraction 16.51% 2.65% 6.24% 1.24% 4.34% 0.43% 0.10% 11.07% 20.00% 0.53% 1.05% 0.18% 0.27% 0.20% 0.07% 31.02% 10.00% 22.96% 0.09% 0.05% 0.01% 0.01% 0.33% 5.09% 8.08% 0.38% 1.09% 24.48% 41.51% 38.08% 1.29% 27.81% 23.11% 0.29% 1.63% 0.04%

Table 2.1 The table shows number of spectra that have the indicated line detected at S /N > 5.5. The total number of analyzed spectra is 1,477,411. of redshift space and their oxygen abundances are therefore somewhat uncertain. When possible we use the [O ii] λ7318 − 7330 quadruplet instead (Kniazev et al., 2004), but as this is a fairly weak line and falls in a region with significant sky emission, we cannot always make use of this line.

2.2.1

Sample selection and classification

We select our sample requiring signal to noise ratio > 5.5 in He ii λ4686, the resulting data set is given in Table 2.2 2 . When the width of the He ii line is consistent with that of the strong forbidden lines, we make the assumption that it has a nebular origin. Given that He ii lines from individual WR stars typically are considerably broader than the forbidden lines in galaxies (e.g. B08), we feel this is a reasonable assumption. In addition we require a S/N> 3 in each of Hβ, [O iii]λ5007, 2 The

full table of 3292 spectra is available in electronic form in http://www.strw.leidenuniv.

nl/~shirazi/SB011/.

22

Data Hα and [N ii] λ6584 emission lines to reliably classify our galaxies (Brinchmann et al. 2004, hereafter B04). The resulting sample contains 2865 spectra with strong nebular He ii λ4686 emission. In parallel, the spectra of the sample galaxies are examined for WR signatures using the approach discussed by B08. This resulted in a total sample of 385 spectra with likely and secure WR features (Class 1, 2 and 3 from B08). While we do not discuss the sample of all WR galaxies in DR7 with Class 1–3 in detail here, we note that it intersects that of the He ii sample but is not a strict subset of it (see Table 2.3). Figure 3.7 shows the redshift distribution for the fraction of the He ii sample in the SDSS as a shaded grey histogram. The cut-off at z ∼ 0.4 is due to Hα falling outside the spectrograph range. The red histogram shows the redshift distribution for the fraction of just the star-forming (SF) galaxies in the SDSS showing He ii λ4686 emission (see below for a discussion of the classification). For each class we have divided the number of He ii emitting galaxies in that redshift bin by the number of similarly classified galaxies in the parent sample (SDSS galaxies that have S/N> 3 in each of Hβ, [O iii] λ5007, Hα and [N ii] λ6584) in that redshift bin. A constant value would therefore indicate a similar redshift distribution of the He ii sample and the parent sample. It is clear from this that full He ii sample closely follows the overall distribution of the SDSS, but that the star-forming galaxies with He ii λ4686 emission are predominantly found at low redshift. We can also see that less than 2% of all galaxies, and less than 1% of the star-forming galaxies in the SDSS DR7 show He ii λ4686 emission in their spectra. To classify the dominant ionization source in each galaxy we follow previous studies in using the Baldwin, Phillips & Terlevich (1981, BPT) line ratio diagnostic diagram of [O iii] λ5007/Hβ versus [N ii] λ6584/Hα as our starting point (Figure 2.2). As has been discussed extensively (e.g. Terlevich et al., 1991; Kewley et al., 2001, hereafter Ke01; Kauffmann et al., 2003, hereafter Ka03; Kewley et al., 2006; Stasi´ nska et al., 2006) this diagram allows a separation of AGN and starforming galaxies because of their significantly different ionizing spectra, typically leading to high [O iii] λ5007/Hβ and [N ii] λ6584/Hα when an AGN is dominating the output of ionizing photons. Ke01 used a combination of stellar population synthesis and photoionization models to compute a theoretical maximum starburst line that isolates objects whose emission line ratios can be accounted for by photoionization by massive stars (below and to the left of the curve) from those where some other source of ionization is required. Ka03 defined an empirical upper limit to the H ii region sequence of SDSS galaxies in the BPT diagram. The region lying between these two lines represents objects more naturally explained as having a composite spectrum combining H ii region emission with a harder ionizing source. As we will see later, this interpretation is further corroborated by the He ii λ4686/Hβ ratios we find for our sample. In the present study, we adopt a two-stage classification methodology. We start out by classifying all galaxies using the BPT diagram, and we will then refine our classifications for a subset of the galaxies using detailed inspection of the spectra and the He ii λ4686 line properties. For the initial classification we 23

Star-forming galaxies with nebular He II 4686 emission

Figure 2.1 The redshift distribution of the full He ii sample, relative to that of the parent sample is shown as a shaded grey histogram. The red histogram shows the redshift distribution of the He ii SF sample relative to that of all star-forming galaxies in the parent sample; the parent sample consisting of those galaxies in the SDSS DR7 having Hβ, [O iii] λ5007, Hα and [N ii] λ6584 detected at S/N> 3. In comparison with the SDSS, it is clear that the galaxies in our SF sample have a redshift distribution strongly shifted toward low redshift.

adopt a similar methodology to that of B04 and use the separation criteria defined by Ke01 and Ka03 to divide galaxies into different classes. Galaxies which are distributed above the Ke01 dividing line are considered AGNs, galaxies between the Ke01 and Ka03 limits have composite classification, which means their source of ionizing radiation could be a combination of star formation and AGN activity. Finally, galaxies below the Ka03 line have a SF classification, which as we will see might be modified subsequently. Figure 2.2 shows the BPT diagram for our sample, the Ke01 and Ka03 classification lines are shown as solid and dotted lines, respectively. The distribution of all emission line galaxies in the SDSS with S /N > 3 in Hβ, [O iii] λ5007, Hα and [N ii] λ6584 is shown as a grey-scale 2D distribution, where the grey-scale shows the logarithm of the number of galaxies in each bin. Blue circles show star-forming galaxies, triangles show composite galaxies and red squares mark AGNs. The grey dashed-dotted line shows the N2 (Pettini & Pagel 2004, hereafter PP04) metallicity calibration for 12 + log O/H = 8.2 ([N ii] λ6584/Hα = −1.2). These classifications are the final ones and incorporate further information as described in the following. While the BPT diagram is a useful classification diagram, it is not particularly sensitive to low levels of AGN contamination and some progress can be made by including lines originating in the mostly neutral ISM (e.g. Kewley et al., 2006). For our purposes we however need to be very confident in the lack of AGNs in our 24

Data

Figure 2.2 This plot shows the BPT diagnostic diagram for the sample. The Ka03 classification line is shown as a dotted line and Ke01 classification line as a solid line. The grey dashed-dotted line shows the N2 PP04 metallicity calibration for 12 + log O/H = 8.2. The distribution of emission line galaxies in the SDSS is shown by the gray-scale 2D distribution where the grey-scale shows the logarithm of the number of galaxies in each bin. Blue circles show star-forming galaxies, triangles show composite galaxies and red squares mark AGNs. As discussed in the text, for some galaxies the classification has been adjusted which is why some objects in the star-forming region are classified as AGN. We mark these galaxies with a blue plus over the red square. sample and we will use the He ii λ4686/Hβ ratio for this purpose. As remarked earlier, only photons with energy in excess of 54.4 eV can ionize He+ and thence produce the He ii λ4686 recombination line. If we consider single stellar population models from Starburst99 (Leitherer et al., 1999) at an age of 2 Myr (so that all stars are on the main sequence), we find that the integrated spectrum of this population typically contain 2-3 orders of magnitude fewer photons at this energy than at the energy required to ionize O+ (35.5 eV), which is needed to produce the [O iii] λ5007 line from collisionally excited O+2 . This line is therefore a very sensitive probe of AGN activity, particularly if used in conjunction with other line ratios. We therefore make use of the He ii λ4686/Hβ vs [N ii] λ6584/Hα diagram to further refine the classification of our sources. We show this diagram in Figure 2.3 where the symbols and colors are the same as in Figure 2.2. Note that there is generally a very significant offset between the star-forming galaxies and AGNs in this diagram, in contrast to the gradual transition in much of the BPT diagram. To make a quantitative separation, we draw a random set of star-forming galaxies from the SDSS and gradually add their emission line fluxes to that of a random set of AGNs. We quantify this by the variable f which is defined to be the fraction of the total Hβ flux comes from the AGN. The total flux 25

Star-forming galaxies with nebular He II 4686 emission

Figure 2.3 This plot shows our sample in the He ii λ4686/Hβ versus [N ii] λ6584/Hα diagnostic diagram. The dotted line shows an empirical line separating AGN and composite objects from star-forming galaxies. The grey dashed-dotted line shows the N2 PP04 metallicity calibration for 12 + log O/H = 8.2. Symbols are the same as the BPT diagram. As He ii λ4686 have a higher ionization potential in comparison to [O iii] λ5007 and is much less sensitive to the electron temperature, we can clearly see separation between the classes in this diagram. The lines with color gradients in the figure are simulated fluxes drawn from adding a random set of star-forming galaxy from the SDSS with emission line fluxes of a random set of AGN spectrum. The coloring of the lines corresponds to the fraction of the spectrum contributed by the star-forming galaxy, 1 − f , as indicated by the color bar on the side. The solid line shows the theoretical upper limit for He ii λ4686/Hβ ratio. is therefore f × AGNFlux + (1 − f ) × S FFlux . Changing f will trace out a path in the diagnostic diagram in Figure 2.3 as illustrated by the lines with color gradients in the figure. The coloring of the lines corresponds to 1 − f , as indicated by the color bar on the side. We repeat this process for a thousand AGN and SF objects located in different bins in the He ii /Hβ diagram. Based on this analysis we find that there is a well defined locus where 10% percent of the He ii λ4686 flux comes from an AGN (in this case ≈ 1% of the Hβ flux comes from the AGN, see section 2.2.2 for further details). A good fit to this relation is given by: log(

FHe ii λ4686 1 ) = −1.22 + , F[N ii] λ6584 FHβ 8.92 log( FHα ) + 1.32

(2.1)

which is shown as a dotted line in Figure 2.3. The solid line shows a theoretical maximum starburst line, similar to the Ke01 line in the BPT diagram — we discuss this further below. 26

Data Sample Total AGN Star forming Composite

He ii 2865 2474 199 179

He ii + WR 385 234 116 35

Table 2.3 An overall numerical summary of the He ii sample. See section 3.2 for a summary of the selection and classification details. See section 2.5 for details on the WR classification. Finally, we look at the spectra of galaxies that we would classify as star-forming on the basis of their location in the BPT diagram, but that are offset from the rest of the star-forming galaxies in the He ii /Hβ diagram and check whether they show AGN features such as broad Balmer lines, strong Nevλ3426, Fe ii emission or if they show a similar [O iii] λ4363/Hγ ratio to that of AGNs. If some of these features are present, we change the classification from star-forming to AGN (these objects are plotted as red squares containing a blue cross in Figures 2.2 and 2.3). This happens for 127 (39%) galaxies. This is a conservative approach as we would exclude star-forming galaxies with strong outflows and hence a broad base to the Balmer lines for instance. To estimate the maximum starburst line in Figure 2.3, we adopted the Charlot & Longhetti (2001, hereafter CL01) models. These combine evolving stellar populations models from Bruzual & Charlot (unpublished BC00 models using Padova 1994 tracks, see Bruzual & Charlot 2003 (BC03) for the current models) with the photoionization code Cloudy (Ferland et al., 1998) and adopt the simple dust attenuation prescription of Charlot & Fall (2000). The main model parameters for our calculations are the metallicity Z, the ionization parameter, U, the dust attenuation τV and the dust-to-metal ratio, ξ, the model parameters used are given in Table 2.4, see B04 for a more detailed discussion. We use the CL01 Single Stellar Population (SSP) models since these achieve the highest possible He ii λ4686/Hβ values. We then identify the maximum ratio reached by the different models and use this upper envelope to define the maximum starburst line shown as a solid line in Figure 2.3. This combination of classification methods means that there is not a one-to-one mapping between the location of an object in a diagnostic diagram and its final classification, as is clear from Figure 2.2 and 2.3. We also mention that we do not change classes for galaxies classified as AGNs or composites using the BPT diagram, but which fall within the star-forming region in the He ii /Hβ diagram and this is the reason why there are 7 AGN below our star-formation–AGN dividing line. By contrasting Figure 2.2 and 2.3 we can make a couple of interesting observations. The first is that while in the BPT diagram we see a steady increase in [O iii] λ5007/Hβ with decreasing [N ii] λ6584/Hα, in Figure 2.3 we see no major change in He ii λ4686/Hβ with [N ii] λ6584/Hα. This indicates that He ii λ4686/Hβ and consequently the ionizing spectrum of stars at λ < 228 ˚ A vary only weakly with metallicity. The second difference between the two plots is that the He ii /Hβ 27

Star-forming galaxies with nebular He II 4686 emission diagram can separate star-forming galaxies from composite galaxies better than BPT diagram since the He ii λ4686/Hβ ratio is more sensitive to the hardness of the ionizing source than [O iii] λ5007/Hβ. We can use this to further support our supposition that the gas in the galaxies falling between the Ka03 and Ke01 lines in the BPT diagrams is ionized by a combination of stars and an AGN — while they are adjacent to the star-forming sequence in the BPT diagram they nearly all clearly separate from the star-forming sequence in the He ii /Hβ diagram, corresponding to an AGN contribution to He ii λ4686 > 50%. Thus referring to these as composite objects appear to be justified.

2.2.2

AGN contamination estimation

In view of the clear separation of the AGN and star-formation branches in Figure 2.3, and our sensitivity to low-level AGN contamination, it is beneficial to study the impact of an AGN on the line ratios in more detail. This has been discussed in previous studies (e.g. B04, Stasi´ nska et al. (2006)) but here we extend those efforts to include He ii λ4686. We focus our attention on the BPT diagram as it is most widely used for emission line classification. We follow the same approach as in the previous section of adding gradually more of an AGN emission line spectrum with He ii λ4686 in emission at S/N> 3, to a star-forming one, and find where it intersects the Ka03 line (see bottom panel of Figure 4). At this point the galaxy would cease to be classified as a star-forming galaxy. We repeat this for a total of 10,000 random combinations of spectra. The median AGN contribution to Hβ at the point where a galaxy ceases to be classified as star-forming is ≈ 10%. Figure 4 shows the region where a median galaxy would be classified as starforming as the gray shaded region. On top of this we show the median trend for the fraction of flux in the indicated line as a function of the fraction of the Hβ flux coming from an AGN (shown for reference as the dashed diagonal line). We note that the exact shape and location of the [N ii] λ6584 and [O iii] λ5007 lines does depend somewhat on the AGN sample chosen but the qualitative trend remains the same. What this figure shows, is firstly the well-known result that some lines are more sensitive to the presence of an AGN than others. As mentioned before, the Balmer lines are expected to have less than 10% contribution from an AGN, while the [N ii] λ6584 line can have more than 30% of its flux coming from an AGN, putting in question its use as an abundance indicator on its own (see also Stasi´ nska et al. (2006)). But for our purposes, it is more important to note that by adopting a classification based on the BPT diagram, we would classify a galaxy as star-forming even when ∼ 65% of its He ii λ4686 flux would come from an AGN. We can now combine this with our previous result in Figure 2.3, where we found that only 10% of the He ii λ4686 emission comes from an AGN in our refined star-forming sample. Applying this to Figure 2.4, we conclude that less than 1% of the Hβ flux comes from an AGN and the other lines will also only have very small contributions from an AGN implying we have a quite pure star-forming sample. Our final sample of He ii star-forming galaxies consists of 189 star-forming 28

Data

Figure 2.4 The upper panel shows the sensitivity of different lines to the presence of an AGN as a function of the fraction of the Hβ flux originating from an AGN. The grey shading indicates the region where a typical galaxy would be classified as star-forming in the BPT diagram (see text for details). We see that we might classify galaxies as star-forming even if 10% of the Hβ flux comes from AGN. The bottom panel illustrates the method used and shows the path traced by a galaxy in the BPT diagram as an increasing amount of AGN light is added to a star-forming galaxy spectrum. The coloring of the line corresponds to the fraction of the Hβ line flux contributed by the star-forming galaxy, 1 − f , as indicated by the color bar on the side. The dashed line is the Ka03 classification line, the intersection of this line with the trajectory of the simulated flux is at f ≈ 0.1.

29

Star-forming galaxies with nebular He II 4686 emission Parameter Z, The metallicity U, The ionization parameter τV , The total dust attenuation ξ, The dust-to-metal ratio

Range −1 < logZ/Z < 0.6, 24 steps −4.0 < logU < −2.0, 33 steps 0.01 < τV < 4.0, 24 steps 0.1 < ξ < 0.5, 9 steps

Table 2.4 The model grid used for the present work. We calculate this both for a constant star formation history at t = 108 yrs as well as for an SSP. galaxies (199 spectra which have been summarized in Table 2.5). As mentioned above we have also checked these spectra for the presence of WR signatures. We will return to a detailed discussion of this in section 2.5 but will use the result of this classification in the following plots.

2.3

Physical properties of the sample

While the majority of the galaxies in our sample have some physical parameters in the MPA-JHU value added catalogues3 , our galaxies are sufficiently extreme that we need to rederive some properties and add some physical parameters to what is in the MPA-JHU catalogue. For the calculation of physical parameters we will adopt the Bayesian methodology outlined by Ka03 and B04. For each model we calculate the probability of that model given the data assuming Gaussian noise and obtain the Probability Distribution Function (PDF) of every parameter of interest by marginalisation over all other parameters (see Appendix A). We take the median value of each PDF to be the best estimate of a given parameter.

2.3.1

Mass measurements

The standard SDSS pipeline often segment nearby actively star-forming galaxies incorrectly, thus we need to redo the photometry of our galaxies. We do this using the Graphical Astronomy and Image Analysis Tools (GAIA4 ). Most of our galaxies have strong emission lines within some of the broad-band filters. Prior to fitting we therefore correct the magnitudes for emission line contributions by assuming that the relative contribution of the lines found in the SDSS fiber spectra is applicable to the galaxy as a whole. As most galaxies in our sample appear to have a uniformly blue color, presumably due to active star formation, we expect this to be a reasonable assumption. Stellar masses are calculated as outlined above, by fitting a large grid of stochastic models to the SDSS u, g, r, i, z band photometry. The grid contains pre-calculated spectra for a set of 100,000 different star formation histories using the BC03 population synthesis models, following the precepts of (Gallazzi et al., 2005, 2008). 3 http://www.mpa-garching.mpg.de/SDSS/DR7 4 http://astro.dur.ac.uk/%7Epdraper/gaia/gaia.html

30

Physical properties of the sample

Figure 2.5 This plot shows He ii λ4686/Hβ as a function of oxygen abundance (see text for details on the calculations). Oxygen abundances derived using the direct method are shown by black circles, we can not calculate oxygen abundances for 13 objects with this method as no [O ii] lines are available for them. Oxygen abundances derived from fits to the CL01 models are shown by red circles while the blue squares show the O3N2 oxygen abundance estimates. Note that the qualitative trends are similar, but the O3N2 estimator does not reach as high O/H values as the other two models. At low metallicity the three methods are in good agreement.

2.3.2

Emission line derived parameters

We use the CL01 model to analyze the emission lines in our sample. We adopt a constant star formation history (SFH) and use the same grid used by B04 (see Appendix A and B04 for further details). In total the model grid used for the fits have 2 × 105 different models. The main quantity of interest for the present discussion is the oxygen abundance, quantified as 12 + log O/H. As there are significant differences between methods for estimating oxygen abundance (Kewley & Ellison, 2008), we have complemented the estimate from the CL01 method with two independent methods: Firstly, we estimate gas-phase oxygen abundances with the empirically calibrated estimators proposed by PP04. They used the line ratios of [O iii] λ5007/Hβ/[N ii] λ6584/Hα, the O3N2 method, and [N ii] λ6584/Hα, the N2 method, as abundance indicators. For all objects with detected [O iii] λ4363 with S /N > 3, we also use the T e method, or direct method, using the fitting formulae provided by Izotov et al. (2006) to estimate the oxygen abundances. For those objects without [O iii] λ4363, we adopt the electron temperature estimates from the CL01 models fits for the direct method calculation. Whenever we do not have [O ii] λ3727, 3729, we use the [O ii] λ7320, 7330 lines to calculate abundances. However, for 13 objects we are unable to use the direct method for estimating oxygen 31

Star-forming galaxies with nebular He II 4686 emission

Figure 2.6 The contours show the mass-metallicity relation for SDSS galaxies (Tremonti et al. 2004). The present sample of star-forming galaxies with WR features is shown by blue circles, while the red triangles show the locations of those that do not show WR features. At low mass we see that our sample is offset from the rest bulk of the of SDSS but overall they sample much the same region. abundance as none of the [O ii] lines are available. Figure 2.5 compares different abundance indicators in the He ii λ4686/Hβ flux ratio versus oxygen abundance plane. The oxygen abundances derived using the direct method are shown by black circles while those derived from the fit to the CL01 model are shown by red circles, the blue squares show O3N2 oxygen abundances. The main conclusion we can draw from this comparison is that all methods agree well at low metallicity while at high metallicity the trends are similar but the O3N2 estimator reaches a lower maximum O/H. For concreteness we will adopt the CL01 estimates for the remainder of the paper but as our main focus will be on the low metallicity region, our results are robust to the estimator chosen. Figure 2.6 shows the mass-metallicity relation for the sample compared to the mass-metallicity relation for all star-forming SDSS galaxies (Tremonti et al. 2004) shown as a contour. The sample galaxies with WR features in the spectra are shown as blue circles, while those that do not show WR features are plotted as red triangles, we will use the same symbols in the following. Overall there is a reasonable agreement with the main SDSS sample except for an offset towards slightly lower metallicity at a fixed mass at low masses.

2.4

Model predictions

In the previous section we carried out an empirical analysis of the properties of star-forming galaxies in the SDSS which show strong nebular He ii λ4686 emission in their integrated spectra. Now we will build on the preceding to explore whether 32

Model predictions

Figure 2.7 This figure shows the logarithm of the ratio of the model prediction for He ii λ4686/Hβ to the observed ratio as a function of oxygen abundance. Red triangles show the ratio for objects without WR features. It is clear that there is good agreement in the range 8.4 < 12 + log O/H < 8.8, but at lower metallicity, the discrepancy between model and observations can be up to and order of magnitude. At higher than solar metallicity where an AGN contribution to the He ii λ4686 flux is more likely we see that model also fail to predict the same ratio as the observed value. current stellar models can be used to explain the He ii λ4686 emission seen in the spectra of these galaxies. We start by predicting nebular He ii λ4686 emission for galaxies in our sample with the CL01 model. Then we change the stellar population model and explore the effect of changing the model on the predicted He ii λ4686 line flux.

2.4.1

CL01 predictions for nebular He II emission

We follow the same procedure as in the calculation of PDFs for the galaxy parameters in the previous section and calculate the likelihood of the model for each object in our sample by fitting the CL01 grid of models to the five important [O ii] λ3727, 3729, Hβ, [O iii] λ5007, Hα, [N ii] λ6584 emission lines. We now want to see whether the models that reproduce the main strong lines in the optical spectrum also reproduce the He ii λ4686 emission line strength. We build the likelihood distribution of He ii λ4686 flux for each galaxy in the same way as before by weighting the He ii λ4686 flux in each model by the probability of that model. We take the median of the likelihood distribution as a prediction for nebular He ii λ4686 emission and the associated confidence interval to be the 16th84th percentile range. We follow the same approach to estimate the He iiλ4686/Hβ ratio. 33

Star-forming galaxies with nebular He II 4686 emission

Figure 2.8 This figure shows the spectral energy distribution (SED) of an instantaneous burst calculated with Starburst99. Each panel corresponds to one metallicity and shows the SED for a range of ages as indicated. Note that the appearance of WR stars 4 Myr after the burst results in a much harder UV continuum. After 5-6 Myr the WR stars disappear and the UV continuum rapidly fades. Also note that in these models, the lowest metallicity SED does not show a significant WR phase. In Figure 2.7 we show the logarithm of the ratio of this model prediction for He ii λ4686/Hβ to the observed He ii λ4686/Hβ ratio as a function of oxygen abundance. It is clear that there is acceptable agreement between the model predictions and the observations in the range 8.4 < 12 + log O/H < 8.8, but a model that can reproduce most the strong lines in the spectrum well, predicts up to one order of magnitude lower He ii λ4686/Hβ ratio than the observed value for some objects at lower metallicities. We also see a deviation at high metallicity — in this regime an AGN contribution to the He ii line flux is more likely, both because the galaxies are more massive, and also because the star formation-AGN separation is more gradual in this regime. We will not discuss this mismatch further here. The population synthesis model used in the CL01 models approximate the WR emission as black bodies at their effective temperature, note in passing that this is not the case in the current BC03 models. This will overestimate the hardness of the ionizing spectra compared to models that consider more sophisticated WR 34

Model predictions atmosphere models such as e.g. Starburst99 (Leitherer et al., 1999) and BPASS (Eldridge et al., 2008). Given the relatively simple treatment of the WR phases in the CL01 models, one might be concerned that the failure to match the data is due to an inherent weakness of the models. In the following we therefore look at the effect of different stellar evolution and atmosphere models on the prediction of He ii λ4686 emission line strengths.

Figure 2.9 Each panel compare the calculated SED of two SFHs with starburst99 for a different metallicity, solid lines show SEDs of instantaneous burst and dotted lines show SEDs with constant star formation rate and different colors show different burst ages. Fluxes have been normalized to the flux at 912 ˚ A. We can see although models with constant star formation form WR stars continuously after 4 Myr, the overall shape of the UV continuum is softer than in the instantaneous burst models because the continuous formation of luminous O stars softens the extreme UV spectrum for a fixed rate of hydrogen ionizing photons.

2.4.2

Starburst99 predictions

To better understand the origin of the He ii λ4686 emission and its dependence on the stellar models adopted, we use the latest version (6.0) of the spectral synthesis code Starburst99 (Stb99, Leitherer et al., 1999, 2010) to calculate spectral energy distributions (SEDs) predictions for a range of ages and metal abundances. We 35

Star-forming galaxies with nebular He II 4686 emission calculate models with an instantaneous burst and a constant star formation history with a Kroupa IMF. We have explored a range of stellar evolution models but the differences are small so we only show the results of one model. For this we adopt the Padova AGB evolutionary tracks combined with Pauldrach/Hillier atmospheres (Smith et al. 2002), Stb99 uses O star model spectra from Pauldrach et al. (2001) and WR model spectra from the code of Hillier & Miller (1998). The models include stellar and nebular continuum. We create models with different metallicities (0.0004, 0.004, 0.008, 0.02, 0.05), where the reference solar metallicity is Z = 0.02 (Anders & Grevesse 1989). We do not consider dust and run the models up to 100 Myr with time-steps of 105 years. Figure 2.8 shows the resulting SEDs for an instantaneous burst with a range of metallicities. Each panel corresponds to one metallicity as indicated and shows the time of the SED for six burst ages. The plots show clearly that the appearance of WR stars, 4 Myr after the burst, results in a harder UV continuum shortwards of the He+ ionizing edge at 228 ˚ A. After ∼ 5 Myr the WR stars disappear and the UV continuum becomes softer. This is in good agreement with the discussion in Schaerer & Vacca (1998, see their Figure 9), which is natural as those models lie at core of the WR modeling in Stb99. Figure 2.9 compares the SEDs of an instantaneous burst (solid line) with that of a continuous star formation model (dotted line). The SEDs have been normalized at 912˚ A, so have the same amount of hydrogen ionizing photons. The continuous star formation models form WR stars continuously after 3 Myr, but the overall shape of the UV continuum is softer than for instantaneous burst models because of the continuous formation of luminous O stars which dilute the SED for a given total mass. To calculate the emission lines, we use the UV continuum generated by the Stb99 models as an input to the photoionization code Cloudy (version c08, Ferland et al., 1998). For each time step, ionization bounded models are calculated by varying the ionization parameter log U = −2, −3, −4, where for consistency with the CL01 models we calculate the ionization parameter at the edge of the Str¨omgren sphere, and a constant hydrogen density of log nH /cm−3 = 2.5. This range in ionization parameter spans the range found by Stasi´ nska & Leitherer (1996) in their analysis of intensely star-forming galaxies (approximately −3.5 < log U < −2.5). Figure 2.10 shows the calculated He ii λ4686/Hβ ratio versus [N ii] λ6584/Hα for different metallicities and different ionization parameters in the left panel and different burst ages in the right panel. The triangles along each model line correspond to the metallicities (0.02, 0.2, 0.4, 1) Z in the left panel and to ages of 3, 4, 5, and 6 Myr in the right panel. In the left panel we fix the the age to 4 Myr, and in the right we set log U = −2. The galaxies in our sample are shown as filled purple circles. From these plots we can see how the He ii λ4686/Hβ ratio depends on age, metallicity and ionization parameter. The lowest metallicity considered in this work is Z = 0.02 Z and for this metallicity there is a strong discrepancy between the model predictions and the observed data but predictions for other metallicities at 4 Myr agree well with the observed He ii λ4686/Hβ ratios. The models can predict the He ii λ4686 emission line ratio, but only for instantaneous bursts with metallicity of 20% solar and above, and only for ages of ∼ 4 − 5 Myr, the period when the extreme-ultraviolet continuum is dominated by emission from WR stars. 36

Model predictions

Figure 2.10 The left panel shows the starburst99 instantaneous burst model prediction for the He ii λ4686/Hβ ratio for different ionization parameters at a fixed age of 4 Myr. The triangles along each line correspond to the metallicities (0.02, 0.2, 0.4, 1) Z and our sample galaxies are shown as purple circles. The right hand panel shows the same for a range of different metallicities and ages with log U set to −2. The triangles along each model line correspond to the burst ages (3, 4, 5, 6) Myr. Note that the lowest metallicity model is unable to cover the observational data. The grey dashed-dotted lines in each panel show the N2 PP04 metallicity calibration for 12 + log O/H = 8.2 For burst ages younger than 4 Myr and older than 6 Myr, and for models with a continuous star formation (not shown here), the softer ionizing continuum results in an emission spectrum that has too weak He ii lines to be consistent with the observational data.

2.4.3

The effect of binary evolution on the He II 4686 emission

The Stb99 models consider single-star evolution only, but it is well-known that massive stars are frequently found in binaries and higher order systems which can have a major effect on the evolution of massive stars. To explore this possibility we compare the observed He ii λ4686/Hβ ratio to the prediction of the Binary Population and Spectral Synthesis (BPASS, Eldridge et al 2008, 2009, 2011) model. The BPASS code includes a careful treatment of the effect of binary evolution on massive short lived stars, and Eldridge et al found that including massive binary evolution in the stellar population leads to WR stars forming over a wider range of ages up to 10 Myr which increases the UV flux at later times. In Figure 2.11 and Figure 2.12 we show the observed He ii λ4686/Hβ ratio in comparison with their instantaneous burst binary and single-star population models, respectively. Each panel shows four different metallicities (0.05, 0.2, 0.4,1) Z . The grey dasheddotted line shows the N2 PP04 metallicity calibration for 12 + log O/H = 8.2. We should note that the lowest metallicity in these plots is a factor of 2.5 higher than that of Stb99 (0.02 Z ). 37

Star-forming galaxies with nebular He II 4686 emission

Figure 2.11 The plots show the prediction of instantaneous burst binary model (BPASS code) for He ii λ4686/Hβ ratio for different metallicities (0.05, 0.2, 0.4,1) z and burst ages. The grey dashed-dotted line shows the N2 PP04 metallicity calibration for 12 + log O/H = 8.2. We get the highest value for the He ii λ4686/Hβ ratio at 3 Myr and the period with elevated He ii λ4686/Hβ lasts longer. Comparing Figure 2.11 and Figure 2.12, there is not a striking difference in the predicted peak ratio for He ii λ4686/Hβ between single-star and binary population models but clearly the binary model predict an elevated ratio for a longer period of time than the single-star models.

2.5

The origin of nebular He II 4686 emission

The preceding discussion makes it clear that in standard models for stellar evolution, only during the phases where the ionizing spectrum is dominated by WR stars will we see strong He ii λ4686 emission. This was already pointed out by Schaerer (1996) and our results are in good agreement with that work as well as a number of other previous works (Schaerer & Vacca, 1998; Guseva et al., 2000; Thuan & Izotov, 2005). What has been less studied is the direct test of this prediction, namely to ask whether WR features are seen whenever He ii λ4686 is observed. Kehrig et. al 38

The origin of nebular He II 4686 emission

Figure 2.12 The plots show the prediction of instantaneous burst single-star model using the BPASS code for the He ii λ4686/Hβ ratio for different metallicities, (0.05, 0.2, 0.4, 1) Z , and burst ages. The grey dashed-dotted line shows the N2 PP04 metallicity calibration for 12 + log O/H = 8.2. (2011) and Neugent & Massey (2011) have recently studied individual star forming regions with He ii λ4686 emission in the Local Group and have shown that while most He ii -emitting regions also show evidence of WR stars, not all do. In more distant galaxies, previous efforts have primarily looked at He ii emission in samples selected for other purposes. As an example, B08 looked at He ii galaxies in SDSS DR6 and studied whether they showed WR features, but their sample selection was not optimised for finding galaxies with He ii λ4686 emission. The excellent analysis of Thuan & Izotov (2005) is another example, where the focus was on BCD galaxies. Since the present sample is selected purely on the presence of He ii λ4686, we can carry out this study with more reliability but we need to check for WR features in the spectra of our galaxies. The presence of WR stars can be recognised via the WR bumps around λ4650 ˚ A(blue bump) and λ5808 ˚ A(red bump). The former is a blend of He ii λ4686 and several metal lines while the latter is caused by Civλ5801 − 12 (see Crowther ARA&A review (2007), B08 and references therein). We have therefore inspected the spectra in our star-forming sample following the methodology described in B08 (see Table 2.3). Each spectrum is assigned a classification of 0 (no WR), 1 (possible 39

Star-forming galaxies with nebular He II 4686 emission

Figure 2.13 This plot shows a comparison between two stacked spectra of galaxies with He ii emission. The black spectrum shows the result of stacking the spectra of all galaxies in our sample that show WR features. In this case we can clearly see the blue WR bump (gray shaded region). The red spectrum shows the result of stacking the spectra of all galaxies that show no WR features. Despite the increase in S/N we see no sign of WR features, strengthening our claim that this class of galaxies show no signs of WR features.

WR), 2 (very probable WR), or 3 (certain WR). We will consider any spectrum with class 1, 2 or 3 to show WR features. To check the reliability of these classifications we can check whether duplicate observations of the same object are given the same classification. There are four sets of duplicate observations (0417-51821-513 Class 2, 0418-51817-302 Class 3, 0418-51884-319 Class 2), (0455-51909-073 Class 0, 0456-51910-306 Class 0), (0266-51602-089 Class 1, 0266-51630-100 Class 0) and (0308-51662-081 Class 3, 0920-52411-575 Class 3). Only for 0266-51602-089 and 0266-51630-100 is there uncertainty whether spectrum show evidence of WR features or not — we choose to keep the original classifications. This is in agreement with, but somewhat better than what we find for the full sample of WR galaxies from the B08 sample where the RMS classification uncertainty from duplicates is 0.4 classes without any apparent dependence on the median S/N of the spectra down to S/N∼ 10; at lower S/N we do not have sufficient numbers of duplicate observations to make a statement. In total we find that 116 of objects show WR features. 40

The origin of nebular He II 4686 emission To check whether the non-detection of WR features is due to a low S/N, we have created stacked spectra of galaxies with and without WR features. The result of this exercise is shown in in Figure 2.13. The black line is the stack of spectra that show WR features while in red we show the stack of non-WR galaxies. We see no sign of WR features in the stack, further supporting the notion that this class of objects show no signs of WR features. We now turn to explore whether there are physical differences in the galaxies showing WR features or not. Figure 2.14 shows the ratio of He ii λ4686/Hβ versus oxygen abundance for galaxies with (blue circles) and without WR features (triangles). The symbols for the non-WR galaxies, here and in the following, are colored red for 12 + log O/H < 8.2, and orange otherwise. At oxygen abundances lower than 8.2 we see there is a trend of increasing He ii λ4686/Hβ ratio towards lower metallicities. We interpret this as being due to a harder ionizing continuum at lower metallicities (e.g. Thuan & Izotov, 2005, and references therein). We also see a trend of increasing He ii λ4686/Hβ towards higher metallicity for 12 + log O/H > 8.7. It is less clear what causes this, but since these are more massive systems with higher star formation rates and with stronger stellar winds due to their higher metallicity, it is likely that what we are seeing is due to an increased contribution of shocks and/or a low-level AGN contamination. Another striking result is that for 12 + log O/H > 8.2, essentially all He ii emitting galaxies show WR features, in agreement with what one would expect from the models discussed in the previous section. At high metallicity there appear to be some systems that show a high He iiλ4686/Hβ ratios but no sign of WR stars. Since these are more massive systems, and fall intermediate between the SF and AGN groups in the He ii λ4686/Hβ versus [N ii] λ6584/Hα diagram, we interpret this as a likely sign of a low-level AGN contribution. In Figure 2.15 we show the fraction of galaxies that show WR features as a function of metallicity. To calculate this we draw a number of random realisations of the data using the uncertainty estimates on the oxygen abundance to draw a random realisation. We also draw a random realisation of the WR classification assuming a random uncertainty in the classification of 0.4 as determined from the analysis of duplicate spectra as discussed above. We carry out 101 random realisations for each of the 101 bootstrap repetitions and calculate the median fraction of galaxies showing WR features in each metallicity bin as well as the 16%– 84% scatter around the median which is shown by the error bars in Figure 2.15. What is clear is that there is a transition at around an oxygen abundance of 12 + log O/H ≈ 8.2 ± 0.1. The uncertainty of 0.1 dex encapsulates the fact that the exact value of this transition abundance depends somewhat on the metallicity calibration adopted and 0.1 dex corresponds to the scatter found when using the different calibrations discussed earlier. As we showed in earlier, e.g. Figure 2.10, the He ii λ4686/Hβ ratio depends strongly on the age of the starbursts so it is reasonable to ask whether the systems without WR features are systematically younger or older than the systems that show WR features. In Figure 2.16 we test this by plotting the He ii λ4686/Hβ ratio versus EW(Hβ) for the sample, where we take the EW(Hβ) as a proxy for starburst age. We show galaxies with WR features by black and blue circles at high and low 41

Star-forming galaxies with nebular He II 4686 emission

Figure 2.14 The He iiλ4686/Hβ ratio versus oxygen abundance for the sample. The blue solid circles show the location of galaxies with WR features. Galaxies which do not show WR features are indicated by red and orange triangles at lower and higher oxygen abundances than 12 + log O/H = 8.2, respectively. metallicities, respectively. Galaxies without WR features are as before shown by orange and red triangles at high and low metallicities. We see the He ii λ4686/Hβ ratios decrease with EW(Hβ) for both WR and non-WR objects. However, galaxies without WR features show higher ratios than galaxies with WR feature for the same EW(Hβ), especially at lower EW(Hβ). Alternatively one might say that at a fixed He ii λ4686/Hβ the systems without WR features have a higher EW(Hβ), or with our assumption, a younger starburst. It is not possible with our data to disentangle these two possibilities.

2.6

Why are there galaxies with He II 4686 emission but no WR features?

The models that we discussed previously agree that WR stars are the source of the hard ionizing photons necessary to produce the He iiλ4686 emission. It is therefore a puzzle why some of the galaxies with nebular He ii λ4686 emission do not show stellar WR features. This lack of WR features has been pointed out before (e.g Thuan & Izotov, 2005), but this is the first time a clear trend with metallicity appears. In this section we therefore turn to discuss some possible reasons for this lack of WR features. 1. Differences in the S/N of the spectra To detect the WR features we need fairly high S/N in the continuum as they are broad and weak features. In the top panel in Figure 2.17 we show the relationship between the equivalent width of the WR blue bump and the median S/N in the 42

Why are there galaxies with He II 4686 emission but no WR features?

Figure 2.15 The fraction of objects with detected WR features in the He ii sample as a function of gas-phase oxygen abundance. The points show the median fraction in each abundance bin and the error bars the 16%-84% scatter around the median (see text for details). While essentially all high-metallicity star-forming galaxies with He ii λ4686 nebular emission show WR features, this fraction drops rapidly at metallicities below 12 + log O/H ≈ 8.2. continuum for WR galaxies in SDSS DR7. We see that a S/N of 10 is sufficient to detect even very weak features. The distribution of the S/N for WR and non-WR objects in the sample versus their metallicities in the bottom panel shows that 20 out of 70 of the non-WR galaxies at 12 + log O/H < 8.2 have S/N less than 10. Since the total number of galaxies at 12+log O/H < 8.2 is 115, if all the 20 low S/N galaxies are assumed to have WR features this brings the fraction of galaxies with WR features in this metallicity range from 30% to 43%. Thus low S/N could cause us to underestimate the WR fraction by at most ∼ 15%. But the fact that the co-added spectrum in Figure 2.13 which has a higher S/N shows no WR features is suggestive that S/N might not be the problem for non-detection of WR features. Furthermore even adding 15% would still mean that less than half He ii emitters at low metallicity would show WR features. We also studied the difference in WR classification of duplicate observations as a function S/N and as remarked earlier, we saw no trend with S/N for more uncertain classification down to a S/N of 10. The dataset is insufficient to test this at lower S/N. Thus our conclusion is that it is unlikely that the systematic absence of WR features in low metallicity objects is due to low S/N in the spectra. 2. Weak lined WN stars One possible reason for the lack of WR features in the most metal poor galaxies, is that the WR lines are too weak to be seen. It is well known (e.g. Conti et al., 1989; Crowther & Hadfield, 2006) that WR stars in the SMC have narrower and less luminous lines than equivalent stars in the Milky Way. Crowther & Hadfield 43

Star-forming galaxies with nebular He II 4686 emission

Figure 2.16 This plot shows the dependence of the He ii λ4686/Hβ ratio on the EW(Hβ) and metallicity. We show galaxies with WR features by black and blue circles at high and low metallicities, respectively. Galaxies without WR feature are shown by orange and red triangles at high and low metallicities. We see the He ii λ4686/Hβ ratios decrease with EW(Hβ) for both WR and non-WR objects. However, galaxies without WR features show higher ratios than galaxies with WR feature for the same EW(Hβ), especially at lower EW(Hβ)s, alternatively one could read this to say that they have a high EW(Hβ) (young age) for a given He ii λ4686/Hβ ratio.

(2006), for instance, find that He II lines in WR stars in the SMC are typically a factor of 4-5 weaker than the Milky Way and hence one would expect that in galaxies at the same distance, it would be harder to see WR features in lower metallicity systems. This is however countered by the fact that low metallicity systems on average are closer, and hence the SDSS fibre subtends a smaller physical size. Since WR emitting regions typically are small, they do not fill the 3” aperture in more distant galaxies, which means that the contrast of the WR features is being enhanced in low redshift systems. This is reflected in the fact that in the WR survey by B08, the equivalent width of the WR features in low metallicity systems (12 + Log O/H < 8.25) is slightly higher than that in metal rich objects (12 + Log O/H > 8.5), a mean of 5.1 ˚ A vs 4.1 ˚ A. Thus the increased contrast appears to approximately cancel out the decrease in line luminosity leading to a fairly constant detection potential with redshift. Thus we do not believe that this is the cause of the dearth of WR features in the low metallicity He ii emitting galaxies, at least down to 12 + log O/H ∼ 8. At very low metallicities, 12 + log O/H ∼ 7.5, Eldridge & Stanway (2009) found that in their models the WR features become very weak and if those results are correct it should be extremely hard to detect WR features in those galaxies. We do however caution that I Zw 18 has very prominent WR features (e.g. Legrand et al., 1997), so at least some extremely metal poor 44

Why are there galaxies with He II 4686 emission but no WR features? galaxies do show clear WR features. Furthermore, even if we are unable to detect WR features at the very lowest metallicities, the same models predict strong WR features at 12 + log O/H > 8, thus this is not a sufficient explanation for the result in Figure 2.15. 3. Shocks Thuan & Izotov (2005) studied the hard ionizing radiation in very metal poor BCD galaxies in the local Universe and concluded that fast radiative shocks could be responsible for the nebular He ii λ4686 emission. Therefore, another possibility is that there is a contribution to the He ii λ4686 from shocks in the ISM (Dopita & Sutherland, 1996). That He ii λ4686 can have some contribution from shocks is plausible, but whether it can explain the systematic lack of WR features at low metallicity is less clear. One question is whether shock models can reproduce the line luminosity in He ii at low metallicities. Using the predictions of the Dopita et al. (2005) and Allen et al. (2008) shock models for the He ii λ4686/Hβ ratio versus [N ii] λ6584/Hα, we found that we can only obtain a ratio comparable to the observed one for objects at high metallicity. A second issue centers on the observation that shocks would most likely come from supernovae and stellar winds, but the latter are though to be weaker at low metallicities. Shocks can also be induced by outflows from starbursts (e.g. galactic winds) and mergers but only two galaxies in our sample are interacting and show a perturbed morphology and we see no clear difference between the low- and high-metallicity subsamples. 4. X-ray binaries Another candidate source for He ii ionization that has been discussed in previous studies of He ii λ4686 emitting nebulae are massive X-ray binaries (Garnett et al. 1991). While these are likely present in many active star forming regions, the question here is why massive X-ray binaries should be more common at low metallicity. If the He+ ionizing photons at low metallicities come from X-ray binaries, we would expect an increasing binary fraction with decreasing metallicity. Without a theoretical justification for this, we consider an increased abundance of X-ray binaries at low metallicity to be an unlikely explanation for the trend seen in Figure 2.15. 5. post-AGB stars Binette et al. (1994) demonstrated that photoionization by post-AGB stars can produce nebular He iiλ4686 emission. After about a few 107 years (i.e, after massive stars disappear), the ionizing radiation comes from post-AGB stars. So, the ionization from post-AGB stars become more important in more evolved systems and this is not the case for our objects especially not for objects at low metallicities (c.f. Figure 2.16). 6. Spatial offset One possible explanation for non detection of WR features is that there could be a significant spatial separation between the WR stars and the region emitting He ii . Kehrig et. al (2008) saw indeed such a spatial separation based on integral field spectra of II Zw 70. The found that the location of the WR stars and the He ii λ4686 emission appear to be separated by ∼ 80pc. Similarly, Izotov et al. (2006b) studied two-dimensional spectra of an extremely 45

Star-forming galaxies with nebular He II 4686 emission

Figure 2.17 Top panel: The relationship between the equivalent width of the WR blue bump and the median S/N in the continuum for WR galaxies in SDSS DR7. We see that a S/N of 10 is sufficient to detect even very weak features. Bottom panel: The distribution of the median S/N in the continuum for WR and non-WR objects in the sample versus their oxygen abundance, we see that 20 out of 70 of the non-WR galaxies at 12 + log O/H < 8.2 have S/N less than 10. See text for more discussion.

46

Conclusion metal-deficient BCD galaxy SBS 0335-052E and showed that the He ii λ4686 emission line was also offset from the near evolved star clusters but in their case, by studying the kinematical properties of the ionized gas from the different emission lines they suggested that the hard ionizing radiation responsible for the He ii λ4686 emission was not related to the most massive youngest stars, but rather was related to fast radiative shocks. If this offset between the WR stars and the region emitting He ii might be an explanation for our non-detections of WR features, it would mean that such a spatial separation is much more common at low metallicity which is rather surprising, since stellar winds are thought to be considerably weaker there, however low metallicity galaxies are also on average closer so the SDSS fibre subtends a smaller physical scale and hence a smaller volume would need to be blown out by the wind (see Figure 2.18). To get a more quantitative estimate we calculate the gravitational binding energy of a cloud with radius 1.5” (SDSS aperture radius) at the redshift of each non-WR object. We assume a hydrogen density of ∼ 50 cm−3 . The median energy require to excavate a hole of this size, is of the order of 1055 erg for our full sample. For the lowest redshift objects the energy requirement is a much more manageable 1049 erg and thus in these cases the absence of WR features in the spectrum could be due to a spatial offset from the He ii emitting region. At higher redshift the energetics makes this a much less likely explanation. The possibility does however warrant further examination and to test this we are undertaking a spectroscopic follow-up of a subsample of these sources. 7. Chemically homogeneous stellar evolution A final explanation could be that the stellar populations at very low metallicities can have much higher temperatures than is currently expected in models. This would be the case if some stars rotated fast enough to evolve homogeneously (Maeder, 1987; Meynet & Maeder, 2007; Yoon & Langer, 2005; Yoon et al., 2006; Cantiello et al., 2007). In that case we can get a higher continuum at 228 ˚ A and correspondingly a higher He ii λ4686/Hβ ratio in comparison with non-homogeneous stellar evolution models. An appealing aspect of this, speculative, explanation is that homogeneous evolution is predicted to be more common at low metallicity. There are however currently no studies of the nebular He ii λ4686 line in the literature. Eldridge & Stanway (2011b) looked at the effect of quasi-homogeneous evolution in their binary models and showed that including it led to strengthened WR features at low redshift, in contrast to what we need. However that still leaves the possibility open that there is a period where strong nebular He ii λ4686 is seen but no WR features as that has not yet been tested.

2.7

Conclusion

We have presented a sample of rare star-forming galaxies with strong nebular He ii λ4686 emission spanning a wide range in metallicity. We have derived physical parameters for these galaxies and showed that emission line models that can reproduce the strong lines in the galaxy spectra are not able to predict the observed ratio of He ii λ4686/Hβ at low metallicities. In agreement with previous studies 47

Star-forming galaxies with nebular He II 4686 emission we found that current models for single massive stars are able to reproduce the He ii λ4686/Hβ ratio in galaxies in our sample, but only for instantaneous bursts of 20% solar metallicity or higher, and only for ages of ∼ 4 − 5 Myr, the period when the extreme-ultraviolet continuum is dominated by emission from WR stars. For stars younger than 4 Myr or older than 5 Myr, and for models with a constant star-formation rate, the softer ionizing continuum results in He ii λ4686/Hβ ratios typically too low to explain our data. Including massive binary evolution in the stellar population analysis leads to WR stars occurring over a wider range in age which leads to acceptable agreement with the data at all metallicities sampled as long as WR stars are present. However, the most notable result of our studies is that a large fraction of the galaxies in our sample do not show WR features and this fraction increases systematically with decreasing metallicity. We find that 70% of galaxies at oxygen abundances lower than 8.2 do not show WR features in their spectra. We discussed a range of different mechanism responsible for producing He ii λ4686 line apart from WR stars in these galaxies and conclude that spatial separation between WR stars and the region emitting He ii emission can be a possible explanation for non-detection of WR features in these galaxies. Moreover, if the stellar population models at very low metallicities can have much higher temperatures than is currently expected in models, as would for instance be the case if some stars rotate fast enough to evolve homogeneously, then such models might explain the origin of the He ii λ4686 line and also the metallicity trend of the He ii sample better. We will explore these possibilities in a future paper.

Acknowledgements First of all we would like to thank the anonymous referee for insightful comments and valuable suggestions which improved the paper significantly. We also would like to thank Marijn Franx, Norbert Langer, S.-C. Yoon and Brent Groves for useful discussion. Finally, we would like to express our appreciation to Daniel Schaerer, Johan Eldridge, Carolina Kehrig and Alireza Rahmati for their kind comments on this paper. Funding for the Sloan Digital Sky Survey (SDSS) and SDSS-II has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, the U.S. Department of Energy, the National Aeronautics and Space Administration, the Japanese Monbukagakusho, and the Max Planck Society, and the Higher Education Funding Council for England. The SDSS Web site is http://www.sdss.org/. The SDSS is managed by the Astrophysical Research Consortium (ARC) 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, The University of Chicago, Drexel University, Fermilab, the Institute for Advanced Study, the Japan Participation Group, The Johns Hopkins University, the Joint Institute for Nuclear Astrophysics, the Kavli Institute for Particle Astrophysics and Cosmology, the 48

Conclusion Korean Scientist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos National Laboratory, the Max-Planck-Institute for Astronomy (MPIA), the MaxPlanck-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. Many thanks go to Allan Brighton, Thomas Herlin, Miguel Albrecht, Daniel Durand and Peter Biereichel, who are responsible for the SkyCat developments at ESO, in particular for making their software free for general use. GAIA and SkyCat are both based on the scripting language Tcl/Tk developed by John Ousterhout and the [incr Tcl] object oriented extensions developed by Michael McLennan. They also make use of many other extensions and scripts developed by the Tcl community. Thanks are also due to the many people who helped test out GAIA and iron out minor and major problems (in particular Tim Jenness, Tim Gledhill and Nigel Metcalfe) and all the users who have reported bugs and sent support since the early releases and continue to do so. The 3D facilities of GAIA make extensive use of the VTK library. Also a free library. GAIA was created by the now closed Starlink UK project, funded by the Particle Physics and Astronomy Research Council (PPARC) and has been more recently supported by the Joint Astronomy Centre Hawaii funded again by PPARC and more recently by its successor organisation the Science and Technology Facilities Council (STFC). This research has made use of the Perl Data Language (PDL, http://pdl. perl.org) and the Interactive Data Language (IDL).

49

Star-forming galaxies with nebular He II 4686 emission

616-52442-364 (# 40) z= 0.0022

0667-52163-634 (# 45) z= 0.0049

1 kpc

0947-52411-569 (# 62) z= 0.0075

0752-52251-340 (# 51) z= 0.0029

1 kpc

1 kpc

1446-53080-134 (#103) z= 0.0034

1 kpc

2016-53799-185 (#151) z= 0.0162

1 kpc

1646-53498-616 (#115) z= 0.0027

1 kpc

2211-53786-486 (#163) z= 0.0021

1 kpc

2329-53725-536 (#169) z= 0.0097

1 kpc

1 kpc

Figure 2.18 Images of nine galaxies which do not show WR features in their spectra while having strong He ii λ4686 emission. The sizes of boxes are 50 ” × 50 ”. In these images north is up, east to the left. Red circles show the size of SDSS fibre aperture (3”) and yellow boxes show 100 pc around the center of the SDSS fibre.

50

+00 +00 +00 +00 +00

δ 00 00 01 01 01

06.78 10.81 24.49 25.06 33.00

Plate–MJD–FiberID 0285-51930-485 0363-51989-400 0470-51929-309 0294-51986-438 0694-52209-113

log He ii /Hβ -0.93 -0.79 -0.84 -1.12 -0.88

log[N ii] /Hα -0.14 -0.31 -0.01 -0.07 -0.36 Class AGN AGN AGN AGN AGN

Class(BPT) AGN AGN AGN AGN AGN

SN OKSN OKSN OKSN OKSN OKSN

WR Features Non-WR Non-WR WR Non-WR Non-WR

Class(WR) — — 1 0 —

Plate–MJD–FiberID 0752-52251-340 0390-51900-291 0390-51900-445 0753-52233-094 0418-51884-319 0691-52199-389 1905-53706-628 0394-51913-075 0695-52202-137 2329-53725-536 0429-51820-495 0666-52149-331 1073-52649-409 0667-52163-634 0456-51910-306 1070-52591-072 0456-51910-076 1666-52991-310 0413-51821-480 1733-53047-528 1922-53315-588 0761-52266-361 1267-52932-384 0548-51986-503

Z 0.00291 0.09437 0.09840 0.01424 0.01791 0.19554 0.15928 0.16742 0.00556 0.00975 0.05662 0.15155 0.03994 0.00492 0.08222 0.00421 0.00456 0.03992 0.06868 0.07881 0.06978 0.04592 0.04722 0.00706

α(J2000) +00 09 53.09 +00 17 28.29 +00 21 01.03 +00 24 25.95 +00 32 18.59 +00 42 52.31 +00 49 13.88 +00 55 27.46 +01 15 33.82 +01 25 34.19 +01 47 07.04 +01 59 53.07 +02 13 06.62 +02 15 13.98 +02 40 52.20 +02 42 39.86 +02 48 15.94 +03 14 31.71 +03 17 43.12 +07 29 30.29 +08 06 19.50 +08 22 27.44 +08 23 54.97 +08 26 04.80

δ +15 44 04.80 -00 56 24.98 +00 52 48.08 +14 04 10.65 +15 00 14.17 +00 27 30.09 +00 24 01.99 -00 21 48.77 -00 51 31.17 +07 59 24.40 +13 56 29.29 -08 13 48.99 +00 56 12.44 -08 46 24.39 -08 28 27.43 -00 00 58.64 -08 17 16.51 +41 05 25.95 +00 19 36.84 +39 49 41.62 +19 49 27.31 +42 23 31.14 +28 06 21.75 +45 58 07.36 ID 51 13 14 52 19 48 138 15 49 169 20 44 73 45 24 72 23 117 16 121 139 53 78 30

log He ii /Hβ -1.35 -1.57 -1.96 -1.98 -1.97 -0.75 -1.64 -1.76 -2.32 -2.02 -1.95 -1.73 -1.97 -1.12 -1.60 -1.41 -2.11 -1.95 -0.77 -1.97 -2.05 -1.09 -2.16 -2.07

log[N ii] /Hα -1.67 -0.56 -1.10 -1.36 -1.52 -0.82 -2.05 -0.89 -1.27 -2.22 -1.52 -1.43 -1.58 -1.81 -1.63 -0.33 -1.90 -1.40 -0.54 -0.93 -1.19 -0.37 -0.95 -0.76

12 + log O/H 7.92±0.09 9.27±0.05 8.83±0.04 8.10±0.12 8.01±0.07 8.88±0.05 7.85±0.06 8.90±0.04 8.31±0.35 7.80±0.04 8.02±0.05 8.04±0.06 8.09±0.05 7.74±0.09 8.02±0.04 8.93±0.25 7.87±0.08 8.08±0.19 9.07±0.11 8.90±0.04 8.83±0.04 9.25±0.07 8.88±0.04 8.64±0.26

Features Non-WR WR WR WR WR WR Non-WR WR WR Non-WR Non-WR Non-WR Non-WR Non-WR Non-WR Non-WR Non-WR WR WR Non-WR WR WR WR WR

Class 0 2 2 3 2 2 0 2 3 0 0 0 0 0 0 0 0 1 2 0 2 1 1 1

Other Names N/A N/A UM 228 N/A SHOC 022 N/A N/A N/A NGC 0450 N/A N/A SHOC 099 UM 411 SHOC 111 SHOC 133 M077 N/A N/A N/A N/A N/A N/A N/A UGC 04393

Table 2.5: The positions and identifications of the sample of star-forming galaxies with nebular He ii λ4686 emission.

α(J2000) +11 59 09.71 +15 30 26.32 +08 58 28.60 +13 06 00.68 +01 10 06.09

Table 2.2: The positions and the identifications of galaxies with nebular He ii λ4686 emission. OKSN is for galaxies having S N > 3 in their [O iii] λ5007, Hβ, [N ii] λ6584 and Hα lines. See B04 for the BPT classification. The full table is available in electronic form in http://www.strw.leidenuniv.nl/∼shirazi/SB011/.

Conclusion

51

Plate–MJD–FiberID 2425-54139-372 0445-51873-404 0828-52317-148 2278-53711-411 0564-52224-216 1875-54453-549 1785-54439-201 2430-53815-117 0551-51993-279 2086-53401-458 0566-52238-497 1194-52703-397 0899-52620-594 0554-52000-190 0553-51999-602 0485-51909-550 2580-54092-470 1594-52992-563 1305-52757-269 0266-51630-100 1947-53431-448 0769-54530-086 1306-52996-005 2583-54095-062 2364-53737-618 1745-53061-475 2588-54174-369 1745-53061-196 1955-53442-354 1427-52996-221 2366-53741-124 0575-52319-521 2591-54140-222 0999-52636-517 0947-52411-569 0875-52354-226 1998-53433-304 0274-51913-187 2478-54097-370 0578-52339-060

Z 0.01974 0.00245 0.14746 0.07219 0.09109 0.07498 0.09186 0.07592 0.00961 0.00981 0.03909 0.00513 0.02727 0.00767 0.00772 0.01366 0.10060 0.01486 0.01085 0.00478 0.01730 0.04626 0.00486 0.01522 0.00423 0.06131 0.05563 0.00957 0.03762 0.00388 0.01897 0.03319 0.04432 0.04453 0.00747 0.02850 0.00489 0.01856 0.00398 0.01287

α(J2000) +08 29 32.66 +08 37 43.48 +08 38 43.64 +08 40 00.37 +08 44 14.24 +08 51 03.67 +08 51 15.65 +08 52 21.72 +08 52 58.21 +09 05 26.34 +09 05 31.08 +09 10 28.78 +09 14 34.95 +09 20 55.92 +09 20 56.07 +09 30 06.43 +09 38 01.64 +09 42 52.78 +09 42 56.74 +09 44 01.87 +09 50 00.77 +09 51 31.77 +09 54 49.56 +09 56 42.49 +10 10 32.81 +10 10 42.54 +10 10 59.30 +10 12 27.02 +10 14 10.58 +10 16 24.52 +10 24 02.75 +10 24 29.25 +10 26 23.65 +10 33 28.53 +10 34 10.15 +10 35 08.88 +10 36 13.22 +10 39 24.38 +10 41 09.60 +10 44 57.79

δ +14 27 06.92 +51 38 30.26 +38 53 50.50 +18 05 31.01 +02 26 21.10 +62 13 26.93 +58 40 55.02 +12 16 51.76 +49 27 33.91 +25 33 02.57 +03 35 30.38 +07 11 17.97 +47 02 07.24 +52 34 07.34 +52 34 04.32 +60 26 53.40 +13 53 17.07 +35 47 25.98 +09 28 16.26 -00 38 32.18 +30 03 41.04 +52 59 36.05 +09 16 15.94 +15 38 11.34 +22 00 39.63 +12 55 16.81 +15 42 23.53 +12 20 37.50 +34 20 34.79 +37 54 45.97 +21 04 50.04 +05 24 51.02 +17 10 14.36 +07 08 01.76 +58 03 49.06 +49 21 42.47 +37 19 27.57 -00 23 21.44 +21 21 42.80 +03 53 13.15

52 ID 173 21 56 168 34 136 130 174 31 154 35 76 60 33 32 25 186 110 80 2 140 54 81 187 171 123 188 122 141 101 172 36 189 69 62 58 146 3 176 37

log He ii /Hβ -2.00 -2.38 -1.88 -2.02 -2.13 -2.06 -2.01 -2.06 -1.83 -1.93 -1.93 -2.02 -2.13 -2.10 -1.71 -2.17 -1.18 -1.92 -2.38 -1.88 -1.97 -2.05 -1.90 -2.02 -2.12 -1.92 -1.92 -2.24 -1.66 -1.70 -1.88 -2.07 -1.12 -1.99 -1.60 -1.79 -1.93 -2.16 -2.13 -1.80

log[N ii] /Hα -1.35 -1.71 -1.32 -1.53 -1.20 -1.39 -1.83 -1.45 -0.62 -1.12 -1.78 -1.77 -1.55 -1.83 -1.63 -1.59 -0.43 -1.44 -1.51 -1.98 -1.51 -1.64 -0.42 -0.91 -1.91 -1.40 -1.70 -1.77 -0.54 -2.21 -1.56 -1.72 -0.64 -0.77 -1.97 -0.76 -0.53 -0.85 -1.81 -2.55

Table 2.5: continued. 12 + log O/H 8.13±0.20 7.95±0.08 8.69±0.05 7.98±0.07 8.75±0.13 8.59±0.10 7.84±0.08 8.23±0.08 8.80±0.24 8.40±0.45 7.79±0.08 7.91±0.09 7.97±0.07 7.85±0.08 7.88±0.07 7.98±0.06 9.28±0.06 8.19±0.18 8.14±0.16 7.77±0.06 8.04±0.10 7.93±0.06 9.03±0.21 8.56±0.28 7.82±0.09 8.40±0.04 7.91±0.04 8.04±0.10 9.13±0.05 7.80±0.05 8.08±0.09 7.84±0.07 9.12±0.05 8.91±0.04 7.78±0.06 8.92±0.04 9.09±0.26 8.67±0.26 7.83±0.09 7.80±0.00 Features Non-WR Non-WR WR Non-WR WR Non-WR Non-WR Non-WR WR WR Non-WR WR Non-WR Non-WR WR Non-WR WR Non-WR WR Non-WR Non-WR WR WR WR WR WR Non-WR WR WR Non-WR WR Non-WR Non-WR WR Non-WR WR WR WR WR Non-WR

Class 0 0 1 0 1 0 0 0 3 3 0 1 0 0 1 0 2 0 3 0 0 2 3 3 1 3 0 2 1 0 2 0 0 2 0 2 3 1 3 0

Other Names N/A MRK 0094 N/A N/A N/A N/A N/A N/A SBS 0849+496 UGC 04764 N/A N/A SBS 0911+472 N/A N/A SBS 0926+606A N/A N/A UGC 05189 CGCG 007-025 N/A SBS 0948+532 NGC 3049 UGC 05342 WAS 05 N/A N/A N/A KUG 1011+345 N/A LSBC D568-03 N/A N/A CGCG 037-076 MRK 1434 SBS 1032+496 NGC 3294 IC 0633 MRK 0724 N/A

Star-forming galaxies with nebular He II 4686 emission

Plate–MJD–FiberID 2147-53491-514 0275-51910-445 1749-53357-499 1981-53463-438 2483-53852-254 0275-51910-622 0876-52669-175 2359-53826-205 1362-53050-617 2213-53792-359 2211-53786-486 1363-53053-510 2494-54174-361 1223-52781-128 1014-52707-254 2500-54178-084 1754-53385-151 0967-52636-302 0967-52636-339 1442-53050-599 2012-53493-407 2506-54179-357 2008-53473-467 0967-52636-540 2513-54141-309 2510-53877-560 2508-53875-615 1761-53376-636 0330-52370-471 1446-53080-134 1313-52790-423 0516-52017-315 1991-53446-584 2226-53819-157 1763-53463-094 2227-53820-389 0517-52024-504 2004-53737-439 2644-54210-188 2610-54476-421

Z 0.05487 0.02620 0.01061 0.02945 0.08443 0.05061 0.00435 0.00452 0.03745 0.00214 0.00214 0.02154 0.00492 0.00345 0.00988 0.00470 0.01764 0.00084 0.02601 0.01018 0.00483 0.02082 0.00601 0.00558 0.02676 0.04512 0.07911 0.00245 0.00353 0.00337 0.01726 0.05811 0.01097 0.08188 0.06675 0.05588 0.00429 0.04894 0.02435 0.00277

α(J2000) +10 45 20.42 +10 45 54.78 +10 46 53.99 +10 47 23.61 +10 50 32.51 +10 50 46.59 +10 53 10.82 +10 54 21.87 +11 00 24.90 +11 04 58.30 +11 04 58.54 +11 05 08.12 +11 17 46.30 +11 27 10.93 +11 27 32.67 +11 29 14.15 +11 32 35.35 +11 33 28.95 +11 34 45.72 +11 36 23.82 +11 36 39.57 +11 36 54.01 +11 41 07.49 +11 45 06.26 +11 48 05.45 +11 48 27.34 +11 48 40.87 +11 50 02.73 +11 52 37.68 +11 54 41.22 +11 55 28.34 +11 57 12.45 +11 57 31.73 +12 00 16.49 +12 00 33.42 +12 01 49.90 +12 08 11.11 +12 09 24.64 +12 09 27.95 +12 15 18.60

δ +09 23 49.10 +01 04 05.84 +13 46 45.77 +30 21 44.29 +15 38 06.31 +00 36 40.11 +50 16 53.21 +27 14 22.16 +43 01 11.93 +29 08 16.55 +29 08 15.72 +44 44 47.24 +17 44 24.69 +08 43 51.70 +53 54 54.47 +20 34 52.01 +14 11 29.83 +49 14 13.01 +50 06 03.33 +47 09 29.08 +36 23 42.89 +19 55 34.80 +32 25 37.22 +50 18 02.44 +21 49 45.35 +25 46 11.77 +17 56 33.02 +15 01 23.48 -02 28 06.39 +46 36 36.35 +57 39 51.97 +02 28 27.88 +32 20 30.17 +27 19 59.01 +13 43 07.99 +28 06 10.67 +02 52 41.82 +32 44 02.05 +22 06 16.69 +20 38 26.72 ID 160 4 124 143 177 5 59 170 91 164 163 92 178 77 70 179 125 64 65 102 150 180 149 66 183 182 181 126 8 103 82 27 145 165 127 166 28 148 193 191

log He ii /Hβ -2.18 -2.28 -2.23 -2.18 -1.79 -1.44 -1.70 -1.93 -2.11 -1.73 -2.18 -2.05 -2.04 -1.71 -1.98 -2.32 -2.19 -1.22 -2.25 -2.07 -1.82 -1.77 -1.82 -1.55 -1.95 -2.10 -2.00 -2.23 -2.13 -1.81 -2.20 -1.99 -2.14 -1.93 -2.05 -1.92 -1.84 -1.44 -1.91 -2.59

log[N ii] /Hα -1.30 -1.34 -1.80 -1.15 -1.76 -0.46 -1.75 -0.74 -1.29 -1.84 -1.83 -1.28 -1.69 -1.27 -1.62 -1.44 -1.48 -1.76 -1.41 -1.92 -1.52 -0.77 -1.82 -2.04 -1.27 -1.70 -1.78 -1.59 -1.75 -1.68 -1.77 -1.15 -1.13 -1.86 -1.08 -1.71 -0.27 -0.43 -0.64 -1.80

Table 2.5: continued. 12 + log O/H 8.70±0.06 8.74±0.04 7.92±0.09 8.85±0.04 7.92±0.04 9.16±0.26 7.94±0.08 8.67±0.41 8.77±0.08 7.79±0.05 7.81±0.06 8.78±0.04 7.90±0.09 8.09±0.11 8.04±0.11 8.03±0.10 8.20±0.09 7.89±0.08 8.21±0.54 7.83±0.09 8.07±0.10 8.61±0.40 7.79±0.05 7.77±0.09 8.35±0.11 7.95±0.07 7.91±0.04 7.99±0.05 7.84±0.09 7.80±0.05 8.03±0.10 8.85±0.06 8.36±0.49 7.94±0.07 8.72±0.04 7.92±0.05 9.04±0.17 9.27±0.04 9.13±0.08 7.88±0.08 Features WR Non-WR Non-WR WR Non-WR WR Non-WR WR Non-WR Non-WR Non-WR WR Non-WR WR Non-WR WR WR WR WR Non-WR WR WR Non-WR Non-WR WR Non-WR Non-WR WR Non-WR Non-WR Non-WR WR WR Non-WR WR Non-WR WR WR WR WR

Class 2 0 0 3 0 2 0 3 0 0 0 2 0 3 0 2 2 3 2 0 1 3 0 0 2 0 0 3 0 0 0 1 3 0 2 0 2 1 2 2

Other Names SCHG 1042+097 SHOC 308 N/A TON 0542 N/A N/A N/A NGC 3451 N/A N/A N/A MRK 0162 N/A IC 2828 MRK 1446 IC 0700 N/A Mrk 0178 MRK 1448 N/A NGC 3755 MRK 0182 KUG 1138+327 N/A MRK 1459 N/A N/A MRK 0750 N/A N/A MRK 0193 UM 469 NGC 3991N N/A N/A N/A NGC 4123 N/A UGC 07137 MRK 1315

Conclusion

53

Plate–MJD–FiberID 1625-53140-386 2001-53493-146 2880-54509-277 0955-52409-608 2880-54509-095 1453-53084-322 1371-52821-053 1371-52821-059 1452-53112-011 1452-53112-016 1615-53166-120 1768-53442-476 2613-54481-507 0494-51915-007 1372-53062-072 1975-53734-498 2236-53729-038 1455-53089-556 1989-53772-089 2461-54570-089 0602-52072-369 2018-53800-096 0339-51692-083 0782-52320-022 2016-53799-185 0602-52072-019 2023-53851-263 0526-52312-097 0341-51690-606 1282-52759-057 2112-53534-557 1376-53089-637 2606-54154-474 2110-53467-499 1464-53091-370 1801-54156-583 2094-53851-487 1043-52465-308 1803-54152-448 0854-52373-514

Z 0.00864 0.00061 0.00428 0.00234 0.09423 0.00100 0.00071 0.00072 0.00157 0.00192 0.00418 0.00697 0.04855 0.08778 0.04190 0.00198 0.00350 0.02377 0.02782 0.00797 0.02766 0.03595 0.00450 0.11183 0.01620 0.03678 0.00288 0.18400 0.02246 0.01637 0.01462 0.02797 0.09425 0.01613 0.01170 0.14730 0.00320 0.00594 0.05450 0.03041

α(J2000) +12 16 47.89 +12 17 49.31 +12 22 25.79 +12 25 05.41 +12 26 11.90 +12 26 15.69 +12 28 09.26 +12 28 13.86 +12 30 28.33 +12 30 38.45 +12 30 48.60 +12 31 54.67 +12 38 29.93 +12 40 49.89 +12 41 34.25 +12 43 56.70 +12 45 16.87 +12 48 46.36 +12 53 06.56 +12 54 23.74 +13 02 49.20 +13 03 54.44 +13 04 32.27 +13 04 45.63 +13 06 24.19 +13 14 26.56 +13 14 47.37 +13 22 11.96 +13 23 47.46 +13 25 19.89 +13 25 49.42 +13 28 44.05 +13 29 16.56 +13 30 17.38 +13 31 26.91 +13 39 24.24 +13 41 56.48 +13 42 51.85 +13 44 24.06 +13 45 31.50

δ +08 02 56.28 +37 51 55.50 +04 34 04.77 +61 09 11.29 +04 15 36.06 +48 29 38.43 +44 05 08.02 +44 07 10.43 +41 41 22.07 +41 39 11.31 +12 02 42.82 +15 07 36.48 +19 59 21.36 +66 24 20.17 +44 26 39.24 +32 10 14.67 +27 07 30.78 +47 42 53.45 +36 49 11.41 +58 53 40.67 +65 34 49.27 +37 14 01.88 -03 33 22.12 +62 24 20.88 +35 13 43.04 +63 33 11.37 +34 52 59.81 +01 30 34.39 -01 32 51.95 +48 02 26.15 +33 03 54.38 +43 55 50.51 +17 00 21.00 +31 19 58.02 +41 51 48.29 +07 39 27.61 +30 31 09.62 +52 42 30.57 +07 45 00.08 +04 42 32.71

54 ID 112 147 198 63 197 106 93 94 104 105 111 128 192 26 95 142 167 107 144 175 39 152 9 55 151 38 153 29 10 79 157 96 190 156 108 131 155 71 132 57

log He ii /Hβ -1.81 -2.28 -2.28 -1.97 -2.00 -2.27 -1.67 -2.50 -2.33 -2.14 -1.44 -1.82 -1.90 -1.88 -0.19 -1.77 -1.83 -2.14 -1.85 -2.03 -1.92 -2.11 -2.26 -1.84 -2.08 -1.62 -2.37 -2.04 -1.88 -1.93 -2.05 -2.16 -1.92 -2.14 -2.13 -1.80 -2.19 -1.79 -1.78 -1.97

log[N ii] /Hα -1.44 -1.72 -1.33 -1.96 -1.66 -2.11 -1.09 -1.95 -1.23 -1.31 -2.04 -0.68 -0.99 -1.49 -0.93 -1.88 -1.15 -1.48 -0.52 -1.15 -1.57 -1.39 -1.36 -0.95 -1.71 -1.15 -1.45 -1.00 -2.48 -1.64 -1.57 -1.28 -1.13 -1.41 -2.12 -0.82 -1.73 -1.53 -1.00 -1.55

Table 2.5: continued. 12 + log O/H 7.99±0.05 7.92±0.09 8.10±0.11 7.96±0.07 7.96±0.06 7.80±0.04 8.22±0.75 7.63±0.06 8.29±0.43 8.07±0.08 7.73±0.07 9.18±0.10 8.93±0.04 7.91±0.07 8.80±0.08 7.93±0.08 8.15±0.19 8.13±0.06 9.27±0.07 8.23±0.62 8.04±0.06 8.54±0.11 8.10±0.09 8.94±0.04 7.87±0.08 8.87±0.08 8.22±0.20 8.89±0.05 7.80±0.03 7.94±0.08 8.15±0.16 8.84±0.06 8.80±0.09 8.11±0.14 7.79±0.04 8.91±0.04 7.98±0.07 7.99±0.06 8.93±0.04 7.92±0.04 Features Non-WR Non-WR Non-WR WR Non-WR Non-WR WR WR WR WR WR WR WR Non-WR Non-WR WR WR WR WR WR WR Non-WR WR WR Non-WR WR WR WR Non-WR Non-WR WR WR WR Non-WR Non-WR WR WR WR WR WR

Class 0 0 0 3 0 0 1 3 3 3 1 3 2 0 0 3 3 2 3 3 1 0 3 2 0 3 3 2 0 0 2 2 3 0 0 1 2 1 2 1

Other Names VCC 0207 N/A N/A SBS 1222+614 N/A UGCA 281 NGC 4449 NGC 4449 NGC 4485 NGC 4490 N/A IC 0797 N/A SHOC 379 N/A NGC 4656 NGC 4670 N/A NGC 4774 N/A N/A N/A UGCA 322 N/A N/A N/A UGC 08323 F13196+0146 UM 570 SBS 1323+483 WAS 69 MRK 0259 N/A UGC 08496 N/A N/A Mrk 0067c3 MRK 1480 N/A TOLOLO 1343+049

Star-forming galaxies with nebular He II 4686 emission

Plate–MJD–FiberID 1776-53858-632 1158-52668-062 2770-54510-583 1324-53088-271 1378-53061-023 1323-52797-002 1323-52797-014 1325-52762-356 1325-52762-350 1642-53115-155 1324-53088-234 1325-52762-412 2786-54540-084 2746-54232-104 1381-53089-470 1644-53144-564 1827-53531-503 0305-51613-604 2137-54206-310 1709-53533-215 0920-52411-575 1646-53498-616 1383-53116-110 2145-54212-388 1843-53816-087 1844-54138-311 1399-53172-299 2911-54631-344 1651-53442-255 1679-53149-384 2163-53823-546 0616-52442-364 1725-54266-068 2167-53889-071 2524-54568-146 2527-54569-147 0624-52377-361 0364-52000-187 0624-52377-092 1570-53149-021

Z 0.02156 0.03383 0.02775 0.00058 0.00442 0.00056 0.00087 0.00085 0.00093 0.01194 0.00097 0.07731 0.00855 0.00771 0.02230 0.08624 0.17350 0.01341 0.01502 0.00466 0.02740 0.00267 0.00396 0.08403 0.00942 0.00610 0.03255 0.04684 0.06816 0.00826 0.03395 0.00225 0.03772 0.01098 0.10704 0.01217 0.00237 0.03133 0.02993 0.00910

α(J2000) +13 46 49.45 +13 59 50.92 +14 01 07.12 +14 02 28.23 +14 02 36.07 +14 03 01.17 +14 03 34.06 +14 03 39.84 +14 04 11.24 +14 04 14.87 +14 04 28.63 +14 09 56.76 +14 18 51.13 +14 23 48.53 +14 26 28.17 +14 28 05.51 +14 29 47.01 +14 30 53.51 +14 31 08.88 +14 32 48.36 +14 48 05.38 +14 48 52.02 +14 50 56.56 +14 51 33.55 +14 54 12.15 +14 56 36.63 +15 09 34.18 +15 19 47.15 +15 23 32.19 +15 26 30.31 +15 34 56.40 +15 37 04.18 +15 45 43.55 +15 46 58.88 +16 06 27.54 +16 15 17.02 +16 16 23.54 +16 24 10.11 +16 26 04.26 +16 47 10.66

δ +14 24 01.68 +57 26 22.98 +21 14 34.60 +54 16 33.08 +39 13 13.28 +54 14 29.40 +54 18 36.91 +54 18 56.87 +54 25 18.67 +36 43 32.67 +54 23 52.80 +54 56 48.89 +21 02 39.74 +14 38 16.54 +38 22 58.67 +36 27 10.40 +06 43 34.97 +00 27 46.35 +27 14 12.29 +09 52 57.15 -01 10 57.72 +34 42 42.99 +35 34 19.59 +26 46 03.56 +30 12 36.25 +30 13 52.36 +37 31 46.11 +39 45 37.85 +29 31 12.08 +41 17 22.34 +24 51 39.24 +55 15 50.62 +08 58 01.35 +17 53 03.07 +13 55 47.88 +13 01 33.08 +47 02 02.32 -00 22 02.58 +46 22 05.79 +21 05 14.51 ID 129 75 195 86 97 83 84 88 87 113 85 89 196 194 98 114 133 6 158 119 61 115 99 159 134 135 100 199 116 118 161 40 120 162 184 185 42 12 41 109

log He ii /Hβ -1.75 -1.89 -2.14 -2.12 -1.92 -2.59 -2.25 -2.29 -2.40 -2.14 -2.41 -2.00 -1.97 -2.04 -1.87 -1.93 -1.80 -2.28 -2.15 -2.31 -2.02 -1.48 -2.24 -1.51 -2.38 -2.31 -2.01 -1.98 -1.96 -1.83 -1.68 -2.46 -1.69 -1.72 -1.99 -1.96 -1.56 -2.11 -1.36 -1.83

log[N ii] /Hα -0.42 -1.42 -1.20 -0.89 -1.46 -1.75 -0.86 -0.67 -1.48 -0.88 -1.30 -1.32 -2.24 -0.48 -1.96 -1.60 -1.17 -1.67 -1.20 -1.42 -1.73 -1.66 -1.41 -0.67 -1.52 -1.50 -2.09 -0.63 -1.79 -1.04 -0.64 -1.33 -2.01 -0.40 -1.28 -1.57 -1.91 -1.58 -0.39 -2.06

Table 2.5: continued. 12 + log O/H 8.92±0.32 8.03±0.08 8.86±0.04 8.58±0.24 8.17±0.16 7.78±0.07 8.57±0.30 8.75±0.23 7.99±0.06 8.54±0.29 8.33±0.32 8.34±0.05 7.76±0.04 9.04±0.33 7.77±0.07 7.88±0.04 8.79±0.04 8.02±0.11 8.44±0.31 8.06±0.10 7.98±0.07 7.91±0.09 8.05±0.10 9.00±0.04 8.14±0.16 8.01±0.08 7.81±0.04 9.08±0.12 7.88±0.04 8.42±0.38 9.05±0.10 8.54±0.04 7.72±0.05 9.08±0.14 8.76±0.06 7.96±0.07 7.87±0.08 7.99±0.05 9.25±0.03 7.74±0.07 Features WR Non-WR WR WR WR WR WR WR Non-WR WR Non-WR WR Non-WR WR Non-WR Non-WR Non-WR WR Non-WR WR WR Non-WR WR WR Non-WR WR Non-WR WR Non-WR WR WR Non-WR Non-WR WR WR Non-WR WR WR WR Non-WR

Class 2 0 1 3 2 3 3 3 0 3 0 3 0 3 0 0 0 1 0 2 3 0 1 1 0 2 0 2 0 3 2 0 0 3 1 0 3 3 2 0

Other Names MRK 0796 MRK 1486 UGC 08929 NGC 5447 N/A N/A NGC 5461 NGC 5461 N/A MRK 1369 N/A SBS 1408+551A N/A N/A N/A N/A N/A N/A MRK 0685 NGC 5669 SHOC 486 UGC 09540 N/A N/A N/A N/A N/A N/A N/A UGC 09856 N/A N/A N/A NGC 5996 N/A N/A Arp 2 SHOC 536 N/A N/A

Conclusion

55

Plate–MJD–FiberID 1342-52793-112 0976-52413-600 0978-52441-118 0358-51818-504 1115-52914-309 0673-52162-312 1893-53239-476 0742-52263-179 0677-52606-533 0650-52143-330

Z 0.03205 0.01195 0.01483 0.04723 0.01381 0.06669 0.02061 0.03029 0.03312 0.03592

α(J2000) +16 49 05.27 +17 12 36.63 +17 18 53.45 +17 35 01.25 +20 47 59.21 +22 25 10.13 +22 38 31.12 +23 01 23.59 +23 02 10.00 +23 56 21.96

δ +29 45 31.61 +32 16 33.42 +30 11 36.20 +57 03 08.55 -00 10 53.98 -00 11 52.84 +14 00 29.78 +13 33 14.79 +00 49 38.84 -09 04 07.42

56 ID 90 67 68 11 74 46 137 50 47 43

log He ii /Hβ -1.52 -1.99 -1.92 -2.17 -2.18 -1.85 -2.16 -2.06 -1.83 -1.51

log[N ii] /Hα -0.39 -1.96 -0.84 -1.31 -1.17 -1.81 -2.22 -1.68 -2.21 -1.43

Table 2.5: continued. 12 + log O/H 9.30±0.03 7.79±0.06 8.60±0.27 8.31±0.04 8.44±0.34 7.90±0.04 7.79±0.03 7.90±0.04 7.75±0.04 8.00±0.07 Features WR WR WR WR WR Non-WR Non-WR Non-WR Non-WR WR

Class 1 1 2 2 1 0 0 0 0 1

Other Names KUG 1647+298 N/A IRAS 17169+3014 SHOC 579 N/A N/A N/A N/A N/A N/A

Star-forming galaxies with nebular He II 4686 emission

Appendix A: Fitting models to the emission lines

2.8

Appendix A: Fitting models to the emission lines

The CL01 model grid is calculated by varying the model parameters, U, ionization parameter at the edge of the Str¨ omgren sphere, τV , total V–band optical depth, ξ, the dust-to-metal ratio of ionized gas, µ, the fraction of the total optical depth in the neutral ISM contributed by the ambient ISM, and Z, the metallicity, over a certain range for 221 unequally spaced time steps from t = 0 to t = 20Gyr (see Table 2.4). In this paper we use the interpolated model grid of various luminosities for 50 time-steps from B04. This includes a total of 2 × 105 different models. To fit to the data we adopt the Bayesian methodology described by Ka03. We obtain the PDF of every parameter of interest by marginalisation over all other parameters. The resulting PDF is used to estimate confidence intervals for each estimated physical parameter. We need to fit to at least five strong emission lines, [O ii] λ3727, 3729, Hβ, [O iii] λ5007, Hα, [N ii] λ6584 to get a good constraint on the parameters. We take the median value of each parameter to be the best estimate of a given parameter. In Figure 2.19 we illustrate our technique by showing the effect of adding lines on the PDFs of parameters when we fit a model to the data. We start with [O ii] λ3727, 3729 and show how we get more well defined PDFs for the indicated parameters as we add the emission lines indicated on the left. We show the PDFs for dust attenuation parameter in V -band, gas phase oxygen abundance, ionization parameter, dust-to-metal ratio of ionized gas and the conversion factor from Hα and [O ii] luminosity to star formation rate (see CL01 for further details), for one object in our sample. The dust-to-metal ratio, ξ, is hard to constrain, except at high metallicity.

57

Star-forming galaxies with nebular He II 4686 emission

Figure 2.19 We show how the PDFs for dust attenuation parameter in V-band, gas phase oxygen abundance, ionization parameter of ionized gas, dust-to-metal ratio and Hα and [O ii] efficiency factors change when we add more emission lines to the fit. ξ is hard to constrain, except at high metallicity.

58

REFERENCES

References Abazajian K. N., et al., 2009, ApJS, 182, 543 Allen, M. G., Groves, B. A., et al., 2008, ApJS, 178, 20 Anders, E., Grevesse, N., 1989, GeCoA, 53, 197 Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 5 Bergeron, P., Ruiz, M. T., Leggett, S. K., 1997, ApJS, 108, 339 Binette, L., Magris, C., et al., 1994, A&A, 292, 13 Brinchmann, J., Charlot, S., et al. 2004, MNRAS, 351, 1151 Brinchmann, J. ,Kunth, D., Durret, F., 2008, A&A, 485, 657 Bruzual, G. ,Charlot, S., 2003, MNRAS, 344, 1000 Cantiello, M., Yoon, S.-C., et al., 2007, A&A, 465, 29 Charlot, S., Fall, S. M., 2000, ApJ, 539, 718 Charlot, S., Longhetti, M., 2001, MNRAS, 323, 887 Conti, P. S., Garmany, C. D., Massey, P., 1989, Apj, 341, 113 Crowther, P. A., Pasquali, A., et al., 1999, A&A, 350, 1007 Crowther, P. A., Dessart, L., Hillier, D. J., et al. 2002, A&A, 392, 653 Crowther, P. A., Hadfield, L. J.,2006, A&A, 449, 711 Crowther, P. A., 2007, ARA&A, 45, 177 Dopita, M. A. & Sutherland, R. S., 1996, ApJS, 102, 161 Dopita, M. A., Groves, B. A., et al., 2005, ApJ, 619, 755 Eldridge, J. J., Izzard, R. G., Tout, C. A., 2008, MNRAS, 384, 1109 Eldridge, J. J., Stanway, E. R., 2009, MNRAS, 400, 1019 Eldridge, J. J., Langer N., Tout, C. A., 2011, MNRAS, 414, 3501 Eldridge, J. J., & Stanway, E. R. 2011, MNRAS, 1595 , in press Ferland, G. J., Korista, K. T., et al., 1998, PASP, 110, 761 Gallazzi, A., Charlot, S., et al., 2005, MNRAS, 362, 41 Gallazzi, A., Brinchmann, J., et al., 2008, MNRAS, 383, 1439 Garnett, D. R., Kennicutt, R. C., et al., 1991, ApJ, 373, 458 Gr¨ afener G., Vink J. S., de Koter A., Langer N., 2011, A&A, 535, A56 Guseva, N. G., Izotov, Y. I., Thuan, T. X., 2000, ApJ, 531, 776 Hadfield, L. J., Crowther, P. A., 2007, MNRAS, 381, 418 Hillier, D. J. & Miller, D. L., 1998, ApJ, 496, 407 Hoare, M. G., Drew, J. E., Denby, M., 1993, MNRAS, 262, 19 ˇ Izotov, Y. I., Stasi˚ ADska, G., et al., 2006, A&A, 448, 955 Izotov, Y. I., Schaerer, D., Blecha, A., et al. 2006, A&A, 459, 71 Kauffmann, G., Heckman, T.M., et al. 2003, MNRAS, 346, 1055 Kehrig, C., V´ılchez, J. M., et al., D., 2008, A&A, 477, 813 Kehrig, C., Oey, M. S., et al., 2011, A&A, 526, 128 Kewley, L. J., Dopita, M. A., et al., 2001, ApJ, 556, 121 Kewley, L. J., Groves, B., et al., 2006, MNRAS, 372, 961 Kewley, L. J. & Ellison, S. L. 2008, ApJ, 681, 1183 Kniazev, A. Y., Pustilnik, S. A., et al., 2004, ApJS, 153, 429 Legrand, F., Kunth, D., Roy, et al. 1997, A&A, 326, L17 Leitherer, C., Schaerer, D., et al. 1999, ApJS, 123, 3 Leitherer, C., Ortiz Ot´ alvaro, P. A., et al. 2010, APJS, 189, 309 59

REFERENCES ´ R., & Esteban, C. 2010, A&A, 516, A104 L´ opez-S´ anchez, A. Maeder, A., 1987, A&A, 178, 159 Meynet, G., Maeder, A., 2007, A&A, 464, 11 Monreal-Ibero, A., V´ılchez, J. M., et al., 2010, A&A, 517, A27 Neugent, K. F., Massey, P., 2011, ApJ, 733, 123 Pauldrach, A. W. A., Hoffmann, T. L., Lennon, M., 2001, A&A, 375, 161 Pettini, M. & Pagel, B. E. J. 2004, MNRAS, 348, 59 Schaerer, D. 1996, ApJ, 467, 17 Schaerer, D., Vacca, W. D., 1998, ApJ, 497, 618 Smith, L. J., Norris, R. P. F., Crowther, P. A., 2002, MNRAS, 337, 1309 Stasi´ nska, G., & Leitherer, C., 1996, ApJ, 107, 661 Stasi´ nska, G., Cid Fernandes, R., et al., 2006, MNRAS, 371, 972 Terlevich, R., Melnick, J., et al., 1991, A&AS, 91, 285 Thuan, T. X., Izotov, Y. I., 2005, ApJS, 161, 240 Tremonti, C. A., Heckman, T. M., et al. 2004, ApJ, 613, 898 Yoon, S.-C., Langer, N., 2005, A&A, 443, 643 Yoon, S.-C., Langer, N., Norman, C., 2006, A&A, 460, 199 York, D. G. et al. 2000, AJ, 120, 1579

60

Chapter

3

The physical nature of the 8 o’clock arc based on near-IR IFU spectroscopy with SINFONI We present an analysis of near-infrared integral field unit spectroscopy for the gravitationally lensed Lyman break galaxy, the 8 o’clock arc, taken with SINFONI on the Very Large Telescope. We explore the shape of the spatially-resolved Hβ profile and demonstrate that we can decompose it into three main components that partially overlap (spatially) but are distinguishable when we include the dynamical information. To study the de-lensed morphology of the galaxy we make use of existing B & H imaging from the Hubble Space Telescope and construct a rigorous lens model using a Bayesian grid based lens modeling technique. We apply this lens model to the SINFONI data cube to construct the de-lensed Hβ line map and the velocity and velocity dispersion maps of the galaxy. We explore the dynamical state of the galaxy and find that the 8 o’clock arc has a complex velocity field that is not simply explained by a single rotating disk. The Hβ profile of the galaxy shows a blue-shifted wing suggesting gas outflows of ∼ 200 km s−1 . We confirm previous findings that the 8 o’clock arc lies on the stellar mass–oxygen abundance–star formation rate plane found locally, but it has nevertheless significantly different interstellar medium properties. We show that the gas surface density of the 8 o’clock arc is a factor of 2–4 higher than similar low-redshift galaxies selected from the Sloan Digital Sky Survey. We also find that the electron density in the ionized gas is approximately five times higher than in the comparison sample, which implies a higher H ii-region pressure as well as likely a higher density interstellar medium than in similar nearby galaxies. Maryam Shirazi, Simona Vegetti, Nicole Nesvadba, Sahar Allam, Jarle Brinchmann, Douglas Tucker Monthly Notices of the Royal Astronomical Society 2013, submitted

SINFONI observations of the 8 o’clock arc

3.1

Introduction

The last decade has seen a dramatic increase in our knowledge of the galaxy population at redshift z > 2. In particular, the large samples of high redshift (high-z) galaxies identified by the Lyman-break dropout technique (Steidel et al., 2003, and references therein) have allowed detailed statistical analysis of the physical properties of these galaxies (Shapley, 2011). While early studies made use of long-slit near-IR spectroscopy (Erb et al., 2006a,b,c) to study the physical properties of these galaxies, more recent studies have focused on near-IR integral field units (IFUs) (F¨ orster Schreiber et al., 2006, 2009; Genzel et al., 2008, 2010). The steadily growing effort to obtain resolved near-IR spectra of high-z galaxies in a systematic manner as in the MASSIV, SINS, SINS/zC-SINF and LSD/ AMAZE surveys is leading to samples of spatially-resolved emission line maps of distant (z ∼ 1 − 3.8) star-forming galaxies. Studying these maps has provided us with spatially-resolved physical properties, metallicity gradients and kinematics of high-z star-forming galaxies (e.g., Contini et al., 2012; Epinat et al., 2012; F¨orster Schreiber et al., 2009; Cresci et al., 2009; Genzel et al., 2011; F¨orster Schreiber et al., 2011a,b; Newman et al., 2012b; Maiolino et al., 2008; Mannucci et al., 2009; Gnerucci et al., 2011). A particularly important question for these studies is whether the observed dynamics are due to, or significantly influenced by major mergers. While this is generally difficult to establish, Genzel et al. (2006) have shown that with sufficiently high resolution integral field unit (IFU) spectroscopy, it is possible to distinguish between rotation and merging. However, variations in spatial resolution still cause inconclusive interpretations. As an example, using SINFONI observations of 14 LBGs F¨ orster Schreiber et al. (2006) argued for rotationally supported dynamics in many LBGs (7 out of 9 resolved velocity fields). In contrast, by studying spatially resolved spectra of 3 galaxies at redshift z ∼ 2 − 3, using the OSIRIS in combination with adaptive optics (AO), Law et al. (2007) showed that the ionized gas kinematics of those galaxies were inconsistent with simple rotational support. Analysis of the SINS Hα sample studied by F¨orster Schreiber et al. (2009) showed that about one-third of 62 galaxies in their sample show rotation dominated kinematics, another one-third are dispersion-dominated objects, and the remaining galaxies are interacting or merging systems. However, more recent AO data have shown that many of these dispersion-dominated sources are in fact rotating and follow the same scaling relations as more massive galaxies (Newman et al., 2013). They also show that the ratio of rotation to random motions (V/σ) increases with stellar mass. This result shows the importance of spatial resolution for studying high-z galaxies. While we are essentially limited by the intrinsic faintness of these objects, gravitational lensing can significantly magnify these galaxies and allow us to study their properties at a level similar to what is achieved at lower redshifts (e.g., MS 1512-cB58; see Yee et al., 1996; Pettini et al., 2000, 2002; Teplitz et al., 2000; Savaglio et al., 2002; Siana et al., 2008). Although near-IR IFU observations with AO have been able to spatially resolve high-z galaxies (F¨orster Schreiber et al., 2006, 2009), obtaining a resolution better than 0.2” even with AO is very difficult 62

Introduction and lensing is the only way to obtain sub-kpc scale resolution for high-z galaxies using current instruments. Studies of this nature will truly come into their own in the future with 30m-class telescopes. Given a sufficiently strongly lensed Lyman break galaxy (LBG), we might be able to study its dynamical state, the influence of any potential non-thermal ionizing source, such as a faint active galactic nucleus (AGN), and the physical properties of the interstellar medium. Spatially-resolved studies of six strongly lensed star-forming galaxies at z ∼ 1.7 − 3.1 using the Keck laser guide star AO system and the OSIRIS IFU spectrograph enabled Jones et al. (2010) to resolve the kinematics of these galaxies on sub-kpc scales. Four of these six galaxies display coherent velocity fields consistent with a simple rotating disk model. Using the same instrument, Jones et al. (2010) also studied spatially-resolved spectroscopy of the Clone arc in detail. Deriving a steep metallicity gradient for this lensed galaxy at z = 2, they suggested an insideout assembly history with radial mixing and enrichment from star formation. A detailed study of the spatially-resolved kinematics for a highly amplified galaxy at z = 4.92 by Swinbank et al. (2009) suggests that this young galaxy is undergoing its first major epoch of mass assembly. Furthermore, analyzing near-infrared spectroscopy for a sample of 28 gravitationally-lensed star-forming galaxies in the redshift range 1.5 < z < 5, observed mostly with the Keck II telescope, Richard et al. (2011) provided us with properties of a representative sample of low luminosity galaxies at high-z. The small number of bright z ∼ 2 lensed galaxies has recently been increased by a spectroscopic campaign following-up galaxy-galaxy lens candidates within the Sloan Digital Sky Survey (SDSS) (Stark et al., 2013). These high spatial and spectral resolution data, will provide us with constraints on the outflow, metallicity gradients, and stellar populations in high-z galaxies. Given its interesting configuration and brightness, the 8 o’clock arc (Allam et al., 2007) is of major interest for the detailed investigation of the physical and kinematical properties of LBGs. Indeed, there has been a vigorous campaign to obtain a significant collection of data for this object. In particular, the following observations have been made: 5-band HST imaging covering F450W to F160W, a Keck LRIS spectrum of the rest-frame UV, near-IR H- and K-band long-slit spectroscopy with the Near InfraRed Imager and Spectrometer (NIRI) on the Gemini North 8m telescope (Finkelstein et al., 2009) and X-shooter observations with the UV-B, VIS-R and NIR spectrograph arms (Dessauges-Zavadsky et al., 2010; Dessauges-Zavadsky et al., 2011, hereafter DZ10 and DZ11). Measuring the differences between the redshift of stellar photospheric lines and ISM absorption lines, Finkelstein et al. (2009) suggested gas outflows on the order of 160 km s−1 for this galaxy. DZ10 also showed that the ISM lines are extended over a large velocity range up to ∼ 800 km s−1 relative to the systematic redshift. They showed that the peak optical depth of the ISM lines is blue-shifted relative to the stellar photospheric lines, implying gas outflows of 120 km s−1 . Studying the rest frame UV, DZ10 showed that the Lyα line is dominated by a damped absorption profile with a weak emission profile redshifted relative to the ISM lines by about +690 km s−1 on top of the absorption profile. They 63

SINFONI observations of the 8 o’clock arc suggested that this results from backscattered Lyα photons emitted in the HII region surrounded by the cold, expanding ISM shell. DZ11 argued that the 8 o’clock arc is formed of two major parts, the main galaxy component and a smaller clump which is rotating around the main core of the galaxy and separated by 1.2 kpc in projected distance. They found that the properties of the clump resembles those of the high-z clumps studied by Swinbank et al. (2009), Jones et al. (2010), and Genzel et al. (2011). They also suggested that the fundamental relation between mass, SFR, and metallicity (Mannucci et al., 2010; Lara-L´ opez et al., 2010) may hold up to and even beyond z = 2.5, as also supported by two other lensed LBGs at 2.5 < z < 3.5 studied by Richard et al. (2011). In this work, we use near-IR IFU spectroscopy of the 8 o’clock arc with SINFONI to spatially resolve the emission line maps and the kinematics of this galaxy. In Section 3.2, we introduce the observed data. In this section, we also discuss the data reduction procedure and the PSF estimation as well as the SED fitting procedure. The analysis of the IFU data is covered in Section 3.3. The physical properties of the 8 o’clock arc are discussed in Section 3.4. In Section 3.5, we introduce our lens modeling technique and also our source reconstruction procedure. In this section, we also present the emission line maps and the Hβ profile in the source plane. The kinematics of the galaxy are discussed in Section 3.6. We present our conclusions in Section 5.6.

3.2 3.2.1

Data Near-IR spectroscopy with SINFONI

We obtained J, H and K band spectroscopy of the 8 o’clock arc (α(J2000) : 00 22 40.91 δ(J2000) : 14 31 10.40) using the integral-field spectrograph SINFONI (Eisenhauer et al., 2003; Bonnet et al., 2004) on the VLT in September 2009 (Program ID: 83.A-0879 A). The observation was done in seeing-limited mode with the 0.125” pixel−1 scale, for which the total field of view (FOV) is 8” × 8”. The total observing time was 4h for J, 5h for H and 3.5h for K with individual exposure times of 600s.

3.2.2

Data Reduction

The SINFONI data were not reduced with the standard ESO pipeline, but with a custom set of routines written by N. Nesvadba, which are optimized to observe faint emission lines from high-z galaxies. These routines are very well tested on SINFONI data cubes for more than 100 high-redshift galaxies, and have been used to reduce the data presented, e.g., in Lehnert et al. (2009) and Nesvadba et al. (2006a,b, 2007a,b). The reduction package uses IRAF (Tody, 1993) standard tools for the reduction of long-slit spectra, modified to meet the special requirements of integral-field spectroscopy, and is complemented by a dedicated set of IDL routines. Data are dark frame subtracted and flat-fielded. The position of each slitlet is measured 64

Data

Filter F450W F814W F160W

Band B I H

A1 AB magnitude 21.89 ± 1.61 20.98 ± 1.03 19.35 ± 0.53

A2 AB magnitude 21.76 ± 1.55 21.06 ± 1.07 19.23 ± 0.5

A3 AB magnitude 21.02 ± 1.07 20.17 ± 0.7 18.46 ± 0.34

A4 AB magnitude 22.65 ± 1.27 21.71 ± 1.44 20.35 ± 0.79

Table 3.1 HST photometry of the 8 o’clock arc images A1-A4. The AB magnitudes correspond to the total photometry of all components in each image. from a set of standard SINFONI calibration data which measure the position of an artificial point source. Rectification along the spectral dimension and wavelength calibration are done before night sky subtraction to account for some spectral flexure between the frames. Curvature is measured and removed using an arc lamp, before shifting the spectra to an absolute (vacuum) wavelength scale with reference to the OH lines in the data. To account for the variation of sky emission, we masked the source in all frames and normalized the sky frames to the average of empty regions in the object frame separately for each wavelength before sky subtraction. We corrected for residuals of the background subtraction and uncertainties in the flux calibration by subsequently subtracting the (empty sky) background separately from each wavelength plane. The three-dimensional data are then reconstructed and spatially aligned using the telescope offsets as recorded in the header within the same sequence of 6 dithered exposures (about one hour of exposure), and by cross-correlating the line images from the combined data in each sequence, to eliminate relative offsets between different sequences. A correction for telluric absorption is applied to each individual cube before the cube combination. Flux calibration is carried out using standard star observations taken every hour at a position and air mass similar to those of the source.

3.2.3

HST Imaging

Optical and NIR imaging data of the 8 o’clock arc were taken with the Wide Field Planetary Camera 2 (WFPC2) and the Near Infrared Camera and Multi-Object Spectrometer (NICMOS) instruments on the Hubble Space Telescope (Proposal No. 11167, PI: Sahar Allam). The 8 o’clock arc is clearly resolved, and was observed in the five filters WFPC2/F450W, WFPC2/F606W, WFPC2/F814W, NIC2/ F110W, and NIC2/F160W, which we will refer to as B, V, I, J and H in the following. Total exposure times of 4 × 1100 s per BV I band, 5120 s in the J band, and 4 × 1280 s in the H band were obtained. The BV I frames, with a pixel scale of 0.1”, were arranged in a four-point dither pattern, with random dithered offsets between individual exposures of 1” in right ascension and declination. The JH frames, with a pixel scale of 0.075”, were also arranged in a four-point dither pattern, but with offsets between individual exposures of 2.5”. In order to resolve the 8 o’clock arc better, the HS T images were drizzled to obtain a pixel scale of 0.05”. Figure 3.1 shows the B band HS T image of the 8 o’clock arc and defines the images A1 through A4 as indicated. 65

SINFONI observations of the 8 o’clock arc

Figure 3.1 The B band HS T image (in counts per second) of the 8 o’clock arc is shown. Three images A1-A3 form an arc and A4 is the counter image. The foreground galaxy (lens) has been removed from this image. The scale-bar in this and all following images are at the redshift of the source.

Figure 3.2 The I − H (rest frame NUV − B) color image of the arc is shown. We can see that the substructures of the arc are better resolved in this image; for instance, we can resolve better two lensed images of the same clump lying between the A3 and A2 images (see Figure 3.13, which marks the de-lensed image of this clump with a purple ellipse). We can also resolve two images of another clump which are between the A2 and A1 images (see Figure 3.13, which notes the de-lensed image of this clump by a red ellipse).

We performed photometry using the Graphical Astronomy and Image Analysis 66

Analysis of the SINFONI data Tools (GAIA1 ). Table 3.1 summarises the HS T photometry of the images A1-A4. As an illustration of the power of the multi-wavelength HS T data set, we show the I-H (rest frame NUV −B) color image of the arc in Figure 3.2. To construct this we convolved the WFPC2/F814W image to the same PSF as the NICMOS/F160W band before creating the color image. We can see that the substructures of the arc are better resolved in this image; for instance, we can resolve two individual images of the same clump that lie between the A3 and A2 images (see the de-lensed image of the clump shown by a purple dashed ellipse in Figure 3.13).

3.2.4

PSF Estimation

We created model PSFs for the HS T images using the TinyTim package 2 (Krist et al., 2011). A measure of the PSF was also obtained using a star in the field. This estimate is consistent with TinyTim PSFs; however, because the star is significantly offset from the arc, in the rest of the paper we use only the the TinyTim PSFs when analyzing the HS T data. For the SINFONI data we use the standard star observations to estimate the PSF. The standard star was observed at the end of each Observing Block (OB) at a air mass similar to that of the data, in a fairly similar direction, and with the same setup. We integrate the standard star cubes in each band along the spectral axis to obtain the 2-dimensional images of the star. We measure the FWHM size of the star along the x and y axes of the SINFONI field-of-view by fitting a 2-dimensional Gaussian to the resulting image. We then average the individual measurements of the standard star images in each direction to determine the PSF for the corresponding band. The spatial resolutions in right ascension and declination are always somewhat different for SINFONI data due to the different projected size of a slitlet (0.25”) and a pixel (0.125”). The PSFs in the J, H and K bands are [0.99”,0.7”], [0.8”,0.66”], [0.69”,0.51”], respectively.

3.3 3.3.1

Analysis of the SINFONI data Nebular emission lines

The spectrum is first analysed using the platefit pipeline, initially developed for the analysis of SDSS spectra (Tremonti et al 2004, Brinchmann et al., 2004, 2008) and subsequently modified for high-z galaxies (e.g. Lamareille et al., 2006). The nebular emission lines identified in the 8 o’clock arc images A2-A3 are summarized in Table 3.3 . Specifically, the emission lines that we can detect in the spectra of the galaxy are [O ii] λ3727, 29, Hδ, Hγ, He ii λ4686 and Hβ. Hβ is the strongest detected emission line. In the following, we therefore concentrate on this line to further study the dynamical properties of the galaxy. Due to the redshift of the 8 o’clock arc, we can not study the [O iii]λ4959, 5007, Hα and [N ii] λ6548, 6584 emission lines because they fall outside of the spectral 1 http://astro.dur.ac.uk/%7Epdraper/gaia/gaia.html 2 http://tinytim.stsci.edu/cgi-bin/tinytimweb.cgi

67

SINFONI observations of the 8 o’clock arc range of the SINFONI bands. This means that we can not place strong constraints on the ionisation parameter or the metallicity of the galaxy using the IFU data.

3.3.2

The integrated Hβ profile

The Balmer lines show asymmetric profiles, especially Hβ, as was also observed by DZ11. We therefore start with a look at the integrated Hβ profile. This offers us, among other things, the possibility of testing our reduction techniques because in the absence of significant small-scale structure, profiles are expected to be similar in shape in the different sub-images. We focus here on the spectra of the highest magnification images, A2 and A3 (see Figure 3.1), and we only integrate over the main galaxy structure, excluding the clump identified by DZ11. The counter image (A4) is complete but is not resolved; the A1 image is only partially resolved and is located near the edge of the data cube. Figure 3.3 shows the integrated Hβ profiles for the images A2 and A3, respectively. We can see that the two images show the same profile, which is what one expects as they are two images of the same galaxy. This result differs from that of DZ11, who found different profiles using their long-slit data. They suggested that this might be either due to the slit orientation not optimally covering the lensed image A3, or alternatively, due to the presence of substructure perturbing the surface brightness of the A2 image. Since the IFU data show the same profile for both images, we can rule out the possibility that substructure might have caused the differences. We can see that the integrated Hβ profiles of both images show one main component with a broad blue wing; thus, the full profile requires a second Gaussian to be well fitted. The width of the Gaussian components for both images is 1.7±0.7 ˚ A, which gives a velocity dispersion of 104 ± 42 km s−1 . The velocity offset between the two fitted Gaussian components which are shown by the red dashed curves is 278 ± 63.5 km s−1 for the A3 image and 191 ± 63 km s−1 for the A2 image which are consistent within the errors. We can clearly see this blue-shifted component in both images in Figure 3.5 and Figure 3.6. DZ11 fitted two individual Gaussians to the main component of the galaxy and concluded that these fits are related to the two components (main and clump) with velocity offset of ∼ 61 ± 8 km s−1 . Since we have resolved the clump using our IFU data and did not include it when integrating the Hβ profile in Figure 3.3, the spectra of the A2 and A3 images plotted in Figure 3.3 do not contain any contribution from the clump. To illustrate the Hβ profile of the clump, we add the spectra of the two images of the clump and show the total profile with an orange line in both panels in Figure 3.3. We measure a velocity offset of 126 ± 42 km s−1 between the clump and the main component of the galaxy. The second component seen for both images (the left Gaussian fits in Figure 3.3) is coming from the part of the arc that was not covered by the slit used by DZ11. From the lens modeling described in Section 3.5, we know that the spatially separated blue-shifted component in the A2 image is coming from the north-east part of the galaxy (see Figure 3.13). However, we see from Figure 3.6 that this blue-shifted component of the A3 image is not separated spatially from the main component of the galaxy. The difference between the two images might be due to the fact that the data have insufficient spatial resolution to resolve the 68

Analysis of the SINFONI data

Figure 3.3 Left: integrated Hβ profile of A2, the highest magnification image, showing two components that we can also resolve in individual spaxels. Middle: integrated Hβ profile of the A3 image. Both images show the same Hβ profile (see right panel). The widths of the Gaussian components for both images (red dashed lines) are 1.7 ˚ A, which gives velocity dispersions of 104 km s−1 . The velocity offset between the two fitted Gaussian components is 278 km s−1 for the A3 image and 191 km s−1 for the A2 image. To illustrate the Hβ profile of the clump, we add the spectra of two images of the clump and show this profile in orange in both panels (note that the A2 and A3 profiles shown in this figure do not contain the clump profile). The velocity offset between the clump Hβ profile and the main galaxy Hβ profile is 126 km s−1 . The bottom panels show the residuals if we fit the Hβ profile of each image with a single Gaussian. Right: Hβ profiles of the A2 and A3 images and the clump are shown; the profiles are normalised to have the same peak. components in the A3 image.

3.3.3

Spatially-resolved emission-line properties of the 8 o’clock arc in the image plane

As we saw above, the integrated Hβ profile is not well fitted by a single Gaussian, and this is also true for Hγ and can also be seen in individual spatial pixels (spaxels) for Hβ. We therefore fit these lines with two or three Gaussian components when necessary. We carry out these fits to the Balmer lines using the MPFIT package in IDL3 . During the fitting, we require the lines to have the same velocity widths. This could be an incorrect approximation in detail but it leads to good fits to the line profiles; the S/N and spectral resolution of the data are not sufficient to leave the widths freely variable. The spatially-resolved Hβ profiles generally show a main component, which we place at a systemic redshift of 2.7363 ± 0.0004 (rest-frame wavelength, λair = 4861.325) and an additional component that is blue3 http://cow.physics.wisc.edu/~craigm/idl/mpfittut.html

69

SINFONI observations of the 8 o’clock arc shifted relative to the main component by 120-300 km s−1 . The best-fit Gaussian intensity map of these blue-shifted and main components of the galaxy are shown in Appendix A for the A2 and A3 images. There is also a redshifted component that is detectable close to the clumps between the A3 and A2 images (see Figure 3.1). This component is spatially separated from the main component by 1” (mentioned also by DZ11). The velocity difference between this component and the main component is ∼ 120 km s−1 . The central map in Figure 3.4 shows the spatial distribution of Hβ line flux across the main components of the arc, where we have integrated the line flux between λ = 4855˚ A and 4867˚ A. The small panels around the Hβ line map show the Hβ profiles of different spatial pixels as indicated. These individual panels clearly show that the Hβ line shows different profiles in different regions across the lensed images. We can show these components in an alternative way, using the positionvelocity diagrams in Figure 3.5 and Figure 3.6 for the A2 and A3 images, respectively. Figure 3.5 clearly shows two spatially separated components corresponding to the A2 image. The peak of one component is blue-shifted by ∼130 km s−1 and spatially separated by ∼ 1” relative to the peak of the other. DZ11 identified these two components with the main galaxy and the clump because they could not separate the clump from the rest of the galaxy using long-slit observations. Here, using IFU data, we have excluded the clump from these position-velocity diagrams. The two retained components are associated with the galaxy and the red (in the spectral direction) component that DZ11 identified as the clump is part of the main galaxy. From the lens modeling that we describe in Section 3.5, we will see that the blue-shifted component comes from the eastern part of the galaxy (see Figure 3.13). The A3 image in Figure 3.6 also shows this blue-shifted component but not as spatially separated. We will argue below (see Section 3.6.3) that a reasonable interpretation of this component is that it corresponds to an outflow from the galaxy.

3.4 3.4.1

The physical properties of the 8 o’clock arc SED fitting

To determine the physical parameters of the 8 o’clock arc, we fit a large grid of stochastic models to the HS T BV I JH photometry to constrain the spectral energy distribution (SED). The grid contains pre-calculated spectra for a set of 100,000 different star formation histories using the Bruzual & Charlot (2003, BC03) population synthesis models, following the precepts of Gallazzi et al. (2005, 2008). Figure 3.7 shows the best-fit SED. We corrected the observed magnitudes for galactic reddening. We corrected the photometry for Galactic foreground dust extinction using E(B − V)Gal = 0.056 (Schlegel et al., 1998). We follow the Bayesian approach presented by Kauffmann et al. (2003) to calculate the likelihood of the physical parameters. We take the median values of the Probability Distribution Functions (PDFs) as our best estimated values. In particular, the parameters we extract are the stellar mass, M? , the current 70

The physical properties of the 8 o’clock arc

Figure 3.4 Middle panel shows Hβ line map of the arc. Small panels around the Hβ line map show the Hβ profiles of different spatial pixels as indicated. The Hβ line map was integrated over 4855˚ A< λrest 2 thus this is not an unreasonable result and it does not seem to be an uncommon result for high-z galaxies (Shirazi et al., 2013, submitted).

3.5

Source Reconstruction

In order to study emission line maps and the kinematics of the galaxy in the source plane, we need to reconstruct the morphology of the 8 o’clock arc using gravitational lens modeling. The lens modeling also allows us to derive the magnification factors of the multiple-lensed images which were used to estimate the corrected SFR and the stellar mass of the galaxy in the previous section. In the following, we describe our lens modeling procedure.

3.5.1

Gravitational lens modeling

To reconstruct the lens model for this system, we make use of the Bayesian grid based lens modeling technique presented by Vegetti & Koopmans (2009), which is optimized for pixelized source surface brightness reconstructions. In order to obtain a robust lens model, we first consider the high resolution and high signal-to-noise ratio B band HS T image. We assume the lens mass distribution to follow a power-law elliptical profile with surface mass density, in units of the critical density, defined as follows k(x, y) =

1−γ γ (y − y0 )2 k0 + rc 2 )( 2 ) √ (2 − ) ((x − x0 )2 + 2 2 q 2 q

(3.2)

We also include a contribution from external shear. In particular, the free parameters of the model are the mass density normalization k0 , the position angle θ, the mass density slope γ (γ = 2.0 for an isothermal mass distribution), the axis ratio q, the centre coordinates x0 and y0 , the external shear strength Γ, the external shear position angle Γθ , and the source regularization level (i.e., a measure of the level of smoothness of the source surface brightness distribution), while the core radius is kept fixed to the negligible value of rc ≡ 0. The most probable parameters 83

SINFONI observations of the 8 o’clock arc

Figure 3.13 Top-left panel: the arc and the counter image in the B band HS T image. The foreground galaxy (lens) has been removed from this image. Topright panel: the best-fit model. Lower-left panel: the residuals after subtracting this model from the data. The reconstructed B band HS T image is shown in the lower-right panel. From this image we see that the source in the rest frame UV consists of at least three components; the main galaxy component, a clump separated by 0.15”, which is shown by the purple dashed ellipse and another clump separated by 0.15”, which is shown by the red dashed ellipse.

of the model are k0 = 3.367, θ = 14.54, q = 0.618, γ = 2.009, Γ = 0.062, Γθ = 10.597. Using the same mass model for the lens galaxy, we also model the NICMOS data. While the B band HS T data probe the rest frame UV and have a higher resolution in comparison to the NICMOS data, the latter have the advantage of providing us with information about the continuum in the J and H bands, where Hβ and [O ii] emission lines are located in the spectra. The most probable mass model for the NICMOS data has the following parameter values: k0 = 3.328, θ = 14.30, q = 0.672, γ = 2.020, Γ = 0.077 Γθ = 13.37. Both results are consistent with DZ11 best-fit parameters, within the error bars. However, unlike DZ11, we do not optimize for the core radius because Einstein rings only make is possible to constrain the mass distribution at the Einstein radius. 84

Source Reconstruction

Figure 3.14 The reconstructed H band HS T image is shown in the plot. From this image we can see that the source in the rest frame optical is formed of multiple components, two main galaxy components. The clump in the reconstructed UV image is partially resolved in this image and it shows a slightly different morphology. The contour shows the HS T B-band reconstructed image. The yellow ellipse in the corner of the image shows the spatial resolution in the source plane.

Figure 3.15 The reconstructed Hβ image is shown in the plot. From this image we can see that the source in the rest frame optical is formed of multiple components, two main galaxy components. The clump in the reconstructed UV image is not resolved in this image. The contour shows the reconstructed HS T B-band image from Figure 3.13. The yellow ellipse in the corner of the image shows the spatial resolution in the source plane. 85

SINFONI observations of the 8 o’clock arc

Figure 3.16 From left to right this figure shows reconstructed Hβ maps from the blue, green and red components of the spectral line. The contour line shows the reconstructed HS T B-band image from Figure 3.13. It can be seen that the blue map predominantly contributes to the eastern part — from a detailed inspection of the lens model we find that in the image plane this is predominantly seen in the A1 and A2 images. Note also that the red component of the spectral line predominantly originates in the west.

Figure 3.17 The reconstructed [O ii] image. Here we are unable to separate two components; this might be due to a higher [O ii] /Hβ ratio in the left component. The red contour shows the reconstructed HS T B-band image from Figure 3.13. These two most probable models are used to map the image plane into the source plane and reconstruct the original morphology of the 8 o’clock arc in the B and H bands, respectively. Thanks to the Bayesian modeling technique, the most 86

Source Reconstruction probable source surface brightness distribution for a given set of lens parameters are automatically provided. The reconstructed B band HS T image is shown in the lower-right panel of Figure 3.13. From this image we can see that the source in the rest frame UV consists of multiple components, including the main galaxy component and two clumps separated by 0.15 arcsec (i.e., 1.2 kpc in projected distance) indicated by the purple and red ellipses. The reconstructed H band HS T image is shown in Figure 3.14. This image shows that the source in the rest-frame optical also consists of multiple components.

3.5.2

Reconstructed-Hβ and [O II] emission lines maps in the source plane

Here we make use of the B band HS T data modeling to reconstruct the Hβ and [O ii] emission line maps of the galaxy. These lines have the highest signal-to-noise that we obtain from the SINFONI data. For each spectral pixel image (frame) of the SINFONI data cubes, we derive the most probable source surface brightness distribution by keeping the lens parameters fixed at the best values recovered from the B band HS T data modeling (after taking into account the rotation of the image), while re-optimizing for the source regularization level. Because of the relatively low signal-to-noise SINFONI data and non-homogenous sky background, we can not use all the lensed images. We focus instead on the highest magnification image, which is the A2 image. Before reconstructing the Hβ map in the source plane, we first bin in the spectral direction by a factor of 4. This corresponds to the spectral resolution (FWHM= 7.9 ˚ A) that we measure from the line widths of the night sky lines around the Hβ line. This provides us with higher signal-to-noise image plane frames. We finally make a Hβ source cube from these reconstructed source frames and use that to derive the kinematics of the galaxy. A reconstructed Hβ map is shown in Figure 3.15. This image also shows two galaxy components. In order to better understand the morphology of the Hβ image, we divide the Hβ spectral range into three equal spectral bins defined as blue (λ(4855 − 4859)˚ A), green (λ(4859 − 4863)˚ A) and red (λ(4863 − 4867)˚ A) intervals, corresponding to three SFR maps shown in Figure 3.11. We then reconstruct the source surface brightness distribution for each of these images, using the same method that was used for the whole Hβ image. The three panels in Figure 3.16 show the reconstructed sources for these images. We can see that the west part of the Hβ line map is very weak and only dominates in the red image (right panel); on the other hand, the eastern part is dominant in the blue image (left panel). Figure 3.17 shows a reconstructed [O ii] image of the galaxy. We see that the eastern component is dominant in this image. Here we are unable to separate the two components; this might be due to a higher [O ii] /Hβ ratios in the eastern component, but could also be caused by the lower spatial resolution at these wavelengths. However, we rule out the later explanation by convolving the Hβ map with a Gaussian PSF matching the slightly different [O ii] map (J-band) PSF. 87

SINFONI observations of the 8 o’clock arc

Figure 3.18 Hβ profile derived from the reconstructed Hβ source. The widths of both Gaussian components are 1.59 ˚ A, which gives a velocity dispersion of 98 km s−1 . The velocity offset between the two components is 246 km s−1 . In the source plane, we have fewer bins than Figure 3.3, since we bin in spectral resolution by a factor 4 before reconstructing the Hβ map in the source plane. To make this profile, we interpolate between those bins to have the same binning as Figure 3.3.

3.5.3

Hβ profile of the reconstructed source

We use the same fitting method that we used in Section 3.3 to fit a two component Gaussian to the Hβ profile for every pixel. We also integrate over the total flux of the galaxy and fit a two component Gaussian to it to compare it to our study in the observed plane (see Section 3.3.2). Figure 3.18 shows the Hβ profile derived from the reconstructed Hβ source. We see that this profile also retains two components. The width of the Gaussian for both components is 1.59 ˚ A, which gives a velocity dispersion of 98±44 km s−1 . This is consistent, within the errors, with our estimated velocity dispersion for the A2 and A3 images in the observed plane (i.e., 104 ± 42 km s−1 ). The velocity offset between the two components is 246 ± 46 km s−1 and matches the offset that we derive for the A2 and A3 images in the image plane. Figure 3.19 shows the velocity and velocity dispersion maps derived from the reconstructed Hβ source. From these maps, we see that the east galaxy component has a lower velocity and velocity dispersion relative to the western component. The western component also shows a smoother velocity gradient. The Hβ line flux divided by the H-band continuum flux is shown as a proxy for EW(Hβ) in Figure 3.20. Here, the H-band continuum map is convolved to the same PSF as the Hβ map. From this we see that the outskirts of the galaxy show a clumpy and higher EW(Hβ). The eastern component of the galaxy also shows a higher EW(Hβ), which might be interpreted as a younger age relative to that of the main component. 88

Dynamics

Figure 3.19 The top and middle panels show the velocity and velocity dispersion maps, respectively. The velocity map is derived using single Gaussian fits but velocity dispersion map represents both components. These are derived from the reconstructed Hβ source. The bottom panel shows the velocity dispersion map derived using single Gaussian fits. The contours show the reconstructed HS T B-band image from Figure 3.13.

89

SINFONI observations of the 8 o’clock arc

Figure 3.20 Hβ line flux divided by the H-band continuum flux shown as a proxy for EW(Hβ). We see that the outskirts of the galaxy show a clumpy and higher EW(Hβ). The eastern component of the galaxy shows a higher EW(Hβ), i.e., a younger age, relative to that of the main component. The red contour shows the reconstructed HS T B-band image.

3.6 3.6.1

Dynamics Hβ Kinematics

To test whether the kinematics of the galaxy are consistent with those of a rotating disk, we compare the velocity and velocity dispersion maps derived from the reconstructed Hβ map with an exponential disk model. Given the low resolution and the low S/N of our data, we simulate a very simple system. The disk models is created using the DYSMAL IDL code (Davies et al., 2011, see also Cresci et al. 2009 for description of the code). The code was used extensively to derive intrinsic properties of disk galaxies (e.g., for estimating the dynamical mass of high-z galaxies; see Cresci et al., 2009). The code uses a set of input parameters which constrain the radial mass profile as well as the position angle and systemic velocity offset, in order to derive a 3D data cube with one spectral (i.e., velocity) and two spatial axes. This can be further used to extract kinematics. The best-fit disk parameters are derived using an optimized χ2 minimization routine and the observed velocity and velocity dispersion. The mass extracted from DYSMAL is that of a thin disk model assuming supported only by orbits in ordered circular rotation. We do not have any constraints on the inclination of our system. Therefore, we use a nominal inclination of 20 degrees. We account for spatial beam smearing from the PSF and velocity broadening due to the finite spectral resolution of the instrument and also rebin by a factor of 4 in the spectral direction in our modeling. We then compare this spatially and spectrally convolved disk model to 90

Dynamics

Figure 3.21 The left panel shows the observed velocity derived from the reconstructed Hβ map. The right panel shows the best velocity fit. the observations. We focus on the western component in the velocity map shown in Figure 3.19 because from the lens modeling we know that this part contains the main component of the galaxy and also shows a smoother observed gradient. The best-fitting exponential disk model for this component is shown in Figure 3.21. While the disk model can reproduce some large-scale features of the velocity field, the residuals are substantial. We can therefore rule out a single rotating disk as a reasonable description of this system. We conclude that the 8 o’clock arc has a complex velocity field that cannot be explained by a simple rotating disk. Furthermore, there appears to be a second component from a clump (see red ellipse in Figure 3.13) that partially overlaps with this component. Whether this is a sign of an on-going merger is difficult to ascertain with the present data. Indeed, the S/N in Hβ does not warrant a much more complex model to be fitted.

3.6.2

Dynamical mass

DZ11 estimated the dynamical mass of the 8 o’clock arc from the line widths via the relation presented by Erb et al. (2006b). We use the same method to estimate the dynamical mass using our estimated velocity dispersion (σ) and half-light radius. For rotation-dominated disks, DZ11 assumed that the enclosed dynamical 91

SINFONI observations of the 8 o’clock arc

Figure 3.22 The observed velocity derived from the reconstructed Hβ map along the slit (shown by dashed lines in Figure 3.21) is shown by the solid curve. The dashed line here shows the best-fit velocity model.

mass within the half-light radius, r1/2 , is Mdyn,rot (r < r1/2 ) = (2.25σ2 r1/2 )/G and multiply this resulting mass by two to obtain the total dynamical mass, where G = 4.3×10−6 kpc (km s−1 )2 M−1 is the gravitational constant. For dispersion-dominated objects, they applied the isotropic virial estimator with Mdyn,disp = (6.7σ2 r1/2 )/G, appropriate for a variety of galactic mass distributions (Binney & Tremaine, 2008). In this case, Mdyn,disp represents the total dynamical mass. For estimating the half-light radius we run galfit on the reconstructed B and H band images. This gives us r1/2 = 2.8±0.2 kpc. We measure σ = 104±42 km s−1 and a rotation-dominated dynamical mass log (Mdyn /M ) = 10.2 ± 0.3 and a dispersiondominated dynamical mass log (Mdyn /M ) = 10.7 ± 0.27 (these values are corrected for instrumental broadening). Using the de-lensed spectra, we also estimate a σ = 98 ± 44 km s−1 , which give us 10.1 ± 0.6, and 10.6 ± 0.6 for the rotationdominated and dispersion-dominated dynamical masses, respectively. The disk model fit can also provide a dynamical mass estimate, log (Mdyn /M ) = 9.5, but we do not use this here because it only accounts for the west component of the velocity map. We note that the idea of using a single line width to estimate dynamical mass is not convincing (this should be done for blue, green and red components individually). In that case the velocity field is clearly more like a merger, so neither of these dynamical mass indicators are reliable. Therefore, obtaining a robust dynamical mass estimate would require considerably more sophisticated models. The simple models are not physically constraining. 92

Dynamics

3.6.3

A massive outflow of gas?

It has been shown that many of high-z star-forming galaxies show evidence for powerful galactic outflows, indicated by studying UV absorption spectroscopy (Pettini et al., 2000; Shapley et al., 2003; Steidel et al., 2010; Weiner et al., 2009; Kornei et al., 2012) and broad Hα emission-line profiles (Shapiro et al., 2009; Genzel et al., 2011; Newman et al., 2012a). Recently Newman et al. (2012b) showed how galaxy parameters (e.g., mass, size, SFR) determine the strength of these outflows. They decomposed the emission line profiles into broad and narrow components and found that the broad emission is spatially extended over ∼ a few kpc. Newman et al. (2012b) showed that star formation surface density enforces a threshold for strong outflows occurring at 1 M yr−1 kpc−2 . The threshold necessary for driving an outflow in local starbursts is 0.1 M yr−1 kpc−2 (Heckman, 2002). The 8 o’clock arc with integrated star formation surface density of 9.2 M yr−1 kpc−2 is certainly in the regime of strong outflow. If we consider the ratio of the Gaussian flux in the blue-shifted component to that of the main component (∼ 0.5) as the Fbroad /Fnarrow , then our result is consistent with what Newman et al. (2012b) show in their Figure 2. However, we note that this definition is not exactly what Newman et al. (2012b) introduced as Fbroad /Fnarrow as we do not fit broad and narrow components but two component with the same width. This provides us with a lower limit on the line width of the blue shifted component. In our data we find a blue-shifted component to Hβ as discussed in Section 3.3.2. As we mentioned earlier we used the same Gaussian width for both main and blue-shifted components. Given the low SN, a unique broad fit with a physical meaning can not be found considering the fact that residual from sky lines might create broad line widths. The velocity offset between this component and the main component of the Hβ line is ≈ 190 km s−1 for the A2 image and ≈ 280 km s−1 for the A3 image. This blue-shifted component could be due to an outflow of gas or a minor merger. In support of the outflow picture, Finkelstein et al. (2009) and DZ10 both observed that ISM lines in the rest-UV spectrum of the 8 o’clock arc are blueshifted relative to the stellar photospheric lines. They also argued that this was a sign of an outflow of gas from the galaxy and taken together with the SINFONI results this strengthens the outflow picture. A further argument for an outflow is the fact that as we saw in section 3.4.7, the H ii regions have an elevated internal pressure, at least compared to similar galaxies locally, and it is reasonable to assume that this aids in driving an outflow (Heckman et al., 1990). In support of the merger picture, Figure 3.16 indicates that much of the blue component flux is extended and arguably galaxy shaped. Taking the evidence from the UV ISM lines, the Hβ profile, the lens model and the high ISM pressure together, it is likely that there is a significant outflow component to the blue-shifted wing, but the question here is whether these observations correspond to a reasonable amount of gas in the outflow. To answer this we also estimate the mass of ionized hydrogen from the luminosity of Hβ, LHβ , of 93

SINFONI observations of the 8 o’clock arc the blue-shifted component and main component of the 8 o’clock arc using (e.g., Dopita & Sutherland, 2003): Mionised =

mH × LHβ 1.235 × 10−25 T 4−0.86 ne

,

(3.3)

where mH is the mass of the hydrogen atom, T 4 the electron temperature in units of 10,000 K and ne the electron density. The mass of ionized hydrogen for the blue-shifted component with LHβ of 45 × 1041 erg s−1 is 107.7±1.3 M assuming ne = 600 cm−3 , which can be contrasted with the mass of ionized hydrogen in the main component which is 108.2±1.3 M (with LHβ = 127×1041 erg s−1 ). The estimate for the blue-shifted component is clearly an upper limit because from the argument above it seems likely that the blue component is not purely an outflow. Thus, taking this at face value, we would find that with an outflow rate of 10% of the star formation rate, a conservative value given local observations (e.g., Martin, 1999, 2006), we need the current star formation activity to have lasted < 3 × 108 yr, which is not unreasonable (to estimate the star formation time scale, the mass is divided by 10% of the SFR). In order to use the outflow mass quantitatively, we need to have an estimate of outflow mass in neutral hydrogen. However, we note that we can not estimate this for the blue-shifted component since NHI estimated in DZ10 is given for the whole galaxy. To estimate the total neutral hydrogen mass in the outflow and the mass outflow rate (Pettini et al., 2000), we use the following formulae given in Verhamme et al. (2008) MHI ≈ 107 ˙ HI = 6. M





r 2  NHI  M , 1 kpc 1020 cm−2

 Vexp r  NHI  M yr−1 , 20 −2 −1 1 kpc 10 cm 200 km s

(3.4) (3.5)

where the first equation relate H I mass in the shell to its column density NHI and the second equation assumes that the mechanical energy deposited by the starburst has produced a shell of swept-up interstellar matter that is expanding with a velocity of Vexp . Assuming our estimated half-light radius from the rest-frame UV reconstructed image and our assumed outflow velocity (r = 2.8 kpc, Vexp = 200 km s−1 ) and taking NHI = 1020.57 cm−2 from DZ10, we find neutral gas masses of MHI = 2.9 × 108 M ˙ HI = 62.4 M yr−1 . This gives us a mass-loading factor of and an outflow rate of M ˙ η = MHI /SFR = 0.27.

3.7

Conclusions

We present a spatially-resolved analysis of the 8 o’clock arc using near-IR IFU data. From this we recovere the Hβ map and the spatially-resolved Hβ profile. We showed that Hβ has different profiles at different spatial pixels and is composed of multiple components. We carefully modeled the strong emission lines in the 94

Conclusions galaxy and compared the results to a local comparison sample. This allowed us to conclude that • The 8 o’clock arc lies on the same M∗ -O/H-SFR manifold as similar starforming galaxies locally (Mannucci et al. (2010); Lara-L´opez et al. (2010), DZ11). • The gas surface density in the 8 o’clock arc log(Σgas / M pc−2 ) = 1.87 is more than twice (×2.164.01 1.61.46 1.55 ) that of similar galaxies locally log(Σgas(analogs) / M pc−2 ) = 1.271.48 . Comparing this with other high-z results 1.02 (e.g., Mannucci et al., 2009, who measure gas surface densities in the range of 2.5-3.3 M pc−2 ), the gas surface density for the 8 o’clock arc is lower. Note that as mentioned by Mannucci et al. (2009), they are sampling the central, most active parts of the galaxies, so those values should be considered as the maximum gas surface densities. • The electron density, and thus the H ii region pressure, in the 8 o’clock arc is ∼ 5 times that of similar galaxies locally. As (Wuyts et al., 2012) pointed out, the electron density measurements for high-z galaxies range from the low density limit to ne > 104 cm−3 . Although these differences depend on the method for measuring the electron density, these also imply a huge difference in the physical properties of star-forming regions in star-forming galaxies at z ∼ 2. The difference between electron density at low-z and high-z have been studied recently by Shirazi et al. (2013, submitted) who compared a sample of 14 high-z galaxies with their low-z counterparts in the SDSS and showed that high-z star-forming galaxies that have the same mass and sSFR as low-z galaxies have a median of 8 times higher electron densities. Taken together these results imply that although the 8 o’clock arc seems superficially similar to local galaxies with similar mass and star formation activity, the properties of the ISM in the galaxy are nonetheless noticeably different. We showed that the two images A2 and A3 have the same Hβ profiles, which of course is to be expected because they are two images of the same galaxy. But this contrasts with the results from long-slit observations of the object by DZ11 who found different profiles. The similarity of the profiles from the IFU data has allowed us to rule out a significant contribution of substructures to the surface brightness of the A2 image. The integrated Hβ profile of both images show a main component with a blue wing which can be fitted by another Gaussian profile with the same width. The width of the Gaussian components for both images are 1.7±0.7 ˚ A, which gives velocity dispersion ∼ 104 ± 42 km s−1 . The velocity offset between the two components is 278 ± 63.5 km s−1 for the A3 image and 191 ± 63 km s−1 for the A2 image which are consistent within the errors. Since both DZ11 and Finkelstein et al. (2009) showed ISM lines are blue-shifted relative to the stellar photospheric lines, suggesting gas outflows with 120-160 km s−1 , and find a comparatively high pressure in the H ii regions of the 8 o’clock arc, we interpret this blue-shifted component as an outflow. However, we cannot rule out that the blue-shifted component might represent a minor merger. 95

SINFONI observations of the 8 o’clock arc To study the de-lensed morphology of the galaxy, we used existing B and H band HS T images. Based on this, we constructed a rigorous lens model for the system using the Bayesian grid based lens modeling technique. In order to obtain a robust lens model, we used the lens modeling of the B band HS T image to reconstruct the Hβ line map of the galaxy. We then presented the de-lensed Hβ line map, velocity and velocity dispersion maps of the galaxy. As an example application we derived the Hβ profile of the reconstructed source and showed that this also requires two Gaussian components with a width of 98 ± 44 km s−1 and velocity separation of 246 ± 46 km s−1 . By fitting an exponential disk model to the observed velocity field, we showed that a simple rotating disk cannot fit the velocity field on its own. Thus, a more complex velocity field is needed, but the S/N of the present data does not allow a good constraint to be had. This also implies that obtaining an accurate dynamical mass for the 8 o’clock system is not possible at present. Similar to some of clumpy galaxies studied by Genzel et al. (2011), the 8 o’clock arc shows a blue-shifted wing but with a less broad profile. We note that as can be seen for example from Figure 3.13, the galaxy has a very clumpy nature in the source plane, but because of the lack of spatial resolution we are not able to study these clumps in more detail.

Acknowledgements We are very thankful for useful comments and suggestions of the anonymous referee. We would like to thank also Ali Rahmati for his useful comments on this paper, Raymond Oonk and Benoit Epinat for useful discussion about SINFONI data reduction, Richard Davies for providing us with DYSMAL code, Johan Richard for his help on the lens modeling and also Max Pettini, Alicia Berciano Alba,Thomas Martinsson and Joanna Holt for useful discussions. We would like also to express our appreciation to Huan Lin, Michael Strauss, Chris Kochanek, Alice Shapley, Dieter Lutz, Chuck Steidel, and Christy Tremonti for their help on the HS T proposal along with our spacial thanks to Andrew Baker. M. Sh., S. A. and D. T. acknowledge the support of Mel Ulmer at North Western University for providing them a meeting room and working place in MayJune 2011. S.V. is grateful to John McKean for useful comments and discussions on the lens modeling During part of this work S.V. was supported by a Pappalardo Fellowship at MIT. This research has made use of the Interactive Data Language (IDL) and QFitsView5 .

5 www.mpe.mpg.de/~ott/QFitsView

96

AppendixA: Gaussian decomposition

3.8

AppendixA: Gaussian decomposition

As we have shown in Section 3.3.3, the resolved Hβ profiles are not well fitted by a single Gaussian. Here we show the best-fit Gaussian intensity maps of the main and blue-shifted components of the galaxy for the image A2 and A3 in Figure 3.23 and Figure 3.24, respectively. As we mentioned earlier, during the fitting we require the lines to have the same velocity widths. In both figures, the upper panel shows the Gaussian intensity map of the main component and the lower panel shows the Gaussian intensity map of the blue-shifted component. Intensities are in unit of 10−17 erg cm−2 s−1 . We see that the main and the blueshifted components of the galaxy are offset spatially (∼ 1”) in the A2 image as we see also in Figure 3.11 showing three calculated SFR maps of the A2 image in blue, green and red channels. However, for the A3 image it is not possible to decompose these components spatially.

97

SINFONI observations of the 8 o’clock arc

Figure 3.23 Lensed image A2: upper panel shows the Gaussian intensity map of the main component; lower panel shows the Gaussian intensity map of the blue-shifted component. Intensities are in unit of 10−17 erg cm−2 s−1 .

Figure 3.24 Lensed image A3: upper panel shows the Gaussian intensity map of the main component; lower panel shows the Gaussian intensity map of the blue-shifted component. Intensities are in unit of 10−17 erg cm−2 s−1 . 98

REFERENCES

References Allam, S. S., Tucker, D. L., Lin, H., Diehl, H. T., Annis, J., Buckley-Geer, E. J., & Friemam, J. A. 2007, ApJ, 662, L51 Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 5 Binney, J., & Tremaine, S. 2008, Galactic Dynamics, 2nd Edition, Princeton University Press, Princeton, NJ USA Bonnet, H., Conzelmann, R., Delabre, B., et al. 2004, Proc. SPIE, 5490, 130 Bouch´e, N., Cresci, G., Davies, R., et al. 2007, ApJ, 671, 303 Brinchmann, J., Charlot, S., et al. 2004, MNRAS, 351, 1151 Brinchmann, J., Charlot, S., Kauffmann, G., et al. 2013, MNRAS, 1276 Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000 Chabrier, G. 2003, PASP, 115, 763 Charlot, S., Longhetti, M., 2001, MNRAS, 323, 887 Contini, T., Garilli, B., Le F`evre, O., et al. 2012, A & A, 539, A91 Cresci, G., Hicks, E. K. S., Genzel, R., et al. 2009, APJ, 697, 115 Davies, R., F¨ orster Schreiber, N. M., Cresci, G., et al. 2011, APJ, 741, 69 de Barros, S., Schaerer, D., & Stark, D., 2012, arXiv:1207.3663 Dessauges-Zavadsky, M., D’Odorico, S., Schaerer, D., Modigliani, A., Tapken, C., & Vernet, J. 2010, A&A, 510, 26 Dessauges-Zavadsky, M., Christensen, L., D’Odorico, S., Schaerer, D., & Richard, J. 2011, A & A, 533, A15 Dopita, M. A. & Sutherland, R. S. 2003, Astrophysics of the diffuse universe (Astrophysics of the diffuse universe, Berlin, New York: Springer, 2003. Astronomy and astrophysics library, ISBN 3540433627) Dopita, M. A., Fischera, J., Sutherland, R. S., et al. 2006, APJS, 167, 177 Dopita, M. A., Fischera, J., Sutherland, R. S., et al. 2006, APJ, 647, 244 Eisenhauer, F., Abuter, R., Bickert, K., et al. 2003, Proc. SPIE, 4841, 1548 Eldridge, J. J., & Stanway, E. R. 2012, MNRAS, 419, 479 Erb, D. K., Shapley, A. E., Pettini, M., Steidel, C. C., Reddy, N. A., & Adelberger, K. L. 2006a, ApJ, 644, 813 Erb, D. K., Steidel, C. C., Shapley, A. E., Pettini, M., Reddy, N. A., & Adelberger, K. L. 2006b, ApJ, 646, 107 Erb, D. K., Steidel, C. C., Shapley, A. E., Pettini, M., Reddy, N. A., & Adelberger, K. L. 2006c, ApJ, 647, 128 Epinat, B., Tasca, L., Amram, P., et al. 2012, A & A, 539, A92 Finkelstein, S. L., Papovich, C., Rudnick, G., Egami, E., Le Floc’h, E., Rieke, M. J., Rigby, J. R, & Willmer, C. N. A. 2009, ApJ, 700, 376 F¨rster Schreiber, N. M., Genzel, R., Lehnert, M. D., et al. 2006, ApJ, 645, 1062 F¨ orster Schreiber, N. M., Genzel, R., Bouch´e, N., et al. 2009, ApJ, 706, 1364 F¨ orster Schreiber, N. M., Shapley, A. E., Erb, D. K., et al. 2011, APJ, 731, 65 F¨ orster Schreiber, N. M., Shapley, A. E., Genzel, R., et al. 2011, APJ, 739, 45 Gallazzi, A., Charlot, S., et al., 2005, MNRAS, 362, 41 Gallazzi, A., Brinchmann, J., et al., 2008, MNRAS, 383, 1439 Genzel, R., Tacconi, L. J., Eisenhauer, F., et al. 2006, Nature, 442, 786 Genzel, R., Burkert, A., Bouch´e, N., et al. 2008, ApJ, 687, 59 99

REFERENCES Genzel, R., Tacconi, L. J., Gracia-Carpio, J., et al. 2010, MNRAS, 407, 2091 Genzel, R., Newman, S., Jones, T., et al., 2011, ApJ, 733, 101 Gnerucci, A., Marconi, A., Cresci, G., et al. 2011, A & A, 528, A88 Heckman, T. M. 2002, Extragalactic Gas at Low Redshift, ed. J. S. Mulchaey & J. Stocke (San Francisco, AC: ASP), 254, 292 Jones, T. A., Swinbank, A. M., Ellis, R. S., Richard, J., & Stark, D. P. 2010, MNRAS, 404, 1247 Jones, T., Ellis, R., Jullo, E., & Richard, J. 2010, APJL, 725, L176 Heckman, T. M., Armus, L., & Miley, G. K. 1990, APJS, 74, 833 Kauffmann, G., Heckman, T.M., et al. 2003, MNRAS, 346, 1055 Kennicutt, R. C., Jr. 1992, APJ, 388, 310 Kennicutt, R. C., Jr. 1998, ARA&A, 36, 189 Kewley, L. J. & Ellison, S. L. 2008, ApJ, 681, 1183 Kornei, K. A., Shapley, A. E., Martin, C. L., et al. 2012, APJ, 758, 135 Krist, J. E., Hook, R. N., & Stoehr, F. 2011, Proc. SPIE, 8127 Lara-L´ opez, M. A., Cepa, J., Bongiovanni, A., et al. 2010, A & A, 521, L53 Lamareille, F., Contini, T., Le Borgne, J.-F., et al. 2006, A & A, 448, 893 Law, D. R., Steidel, C. C., Erb, D. K., Larkin, J. E., Pettini, M., Shapley, A. E., & Wright, S. A. 2007, ApJ, 669, 929 Lehnert, M. D., Nesvadba, N. P. H., Le Tiran, L., et al. 2009, APJ, 699, 1660 Lehnert, M. D., Le Tiran, L., Nesvadba, N. P. H., et al. 2013, arXiv:1304.7734 Le Tiran, L., Lehnert, M. D., van Driel, W., Nesvadba, N. P. H., & Di Matteo, P. 2011, A & A, 534, L4 Maiolino, R., Nagao, T., Grazian, A., et al. 2008, A&A, 488, 463 Mannucci, F., Cresci, G., Maiolino, R., et al. 2009, MNRAS, 398, 1915 Mannucci, F., Cresci, G., Maiolino, R., Marconi, A., & Gnerucci, A. 2010, MNRAS, 408, 2115 Martin, C. L. 1999, APJ, 513, 156 Martin, C. L. 2005, APJ, 621, 227 Martin, C. L. 2006, APJ, 647, 222 Martin, C. L., Shapley, A. E., Coil, A. L., et al. 2012, APJ, 760, 127 Moustakas, J., Kennicutt, R. C., Jr., & Tremonti, C. A. 2006, APJ, 642, 775 Nesvadba, N. P. H., Lehnert, M. D., Eisenhauer, F., et al. 2006, APJ, 650, 661 Nesvadba, N. P. H., Lehnert, M. D., Eisenhauer, F., et al. 2006, APJ, 650, 693 Nesvadba, N. P. H., Lehnert, M. D., Genzel, R., et al. 2007, APJ, 657, 725 Nesvadba, N. P. H., Lehnert, M. D., De Breuck, C., Gilbert, A., & van Breugel, W. 2007, A & A, 475, 145 Newman, S. F., Shapiro Griffin, K., Genzel, R., et al. 2012, APJ, 752, 111 Newman, S. F., Genzel, R., F¨ orster-Schreiber, N. M., et al. 2012, APJ, 761, 43 Newman, S. F., Genzel, R., F¨ orster Schreiber, N. M., et al. 2013, APJ, 767, 104 Osterbrock, D. E. 1989, Astrophysics of gaseous nebulae and active galactic nuclei, University Science Books, Mill Valley, CA USA Pettini, M., & Pagel, B. E. J. 2004, MNRAS, 348, 59 Pettini, M., Rix, S. A., Steidel, C. C., Adelberger, K. L., Hunt, M. P., & Shapley, A. E. 2002, ApJ, 569, 742 Pettini, M., Steidel, C. C., Adelberger, K. L., Dickinson, M., & Giavalisco, M. 100

REFERENCES 2000, ApJ, 528, 96 Richard, J., Jones, T., Ellis, R., Stark, D. P., Livermore, R., & Swinbank, M. 2011, MNRAS, 413, 643 Schaerer, D., & de Barros, S. 2009, A&A, 502, 423 Schaerer, D., & de Barros, S. 2010, A&A, 515, 73 Salpeter, E. E. 1955, ApJ, 121, 161 Savaglio, S., Glazebrook, K., Le Borgne, D., et al. 2005, ApJ, 635, 260 Savaglio, S., Panagia, N., & Padovani, P. 2002, ApJ, 567, 702 Shapiro, K. L., Genzel, R., Quataert, E., et al. 2009, APJ, 701, 955 Shapley, A. E., Steidel, C. C., Pettini, M., & Adelberger, K. L. 2003, ApJ, 588, 65 Shapley, A. E., Steidel, C. C., Pettini, M., Adelberger, K. L., & Erb, D. K. 2006, APJ, 651, 688 Shapley A., 2011, ARA&A, 49, 525 Shapley, A. E., Steidel, C. C., Erb, D. K., et al. 2005, APJ, 626, 698 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525 Shirazi, M., & Brinchmann, J. 2012, MNRAS, 421, 1043 Shirazi, M., & Brinchmann, J., Rahmati, A., 2013, arXiv:1307.4758 Schmidt M., 1959, APJ, 129, 243 Siana, B., Teplitz, H. I., Chary, R.-R., Colbert, J., & Frayer, D. T. 2008, APJ, 689, 59 Stark, D. P., Auger, M., Belokurov, V., et al. 2013, arXiv:1302.2663 Steidel, C. C., Adelberger, K. L., Shapley, A. E., et al. 2003, APJ, 592, 728 Steidel, C. C., Erb, D. K., Shapley, A. E., et al. 2010, APJ, 717, 289 Swinbank, A. M., Webb, T. M., Richard, J., et al. 2009, MNRAS, 400, 1121 Teplitz, H. I., McLean, I. S., Becklin, E. E., et al. 2000, ApJ, 533, L65 Tody, D. 1993, Astronomical Data Analysis Software and Systems II, 52, 173 Tremonti, C. A., Heckman, T. M., Kauffmann, G., et al. 2004, APJ, 613, 898 Vegetti, S., & Koopmans, L. V. E. 2009, MNRAS, 392, 945 Verhamme, A., Schaerer, D., Atek, H., & Tapken, C. 2008, A & A, 491, 89 Volino, F., Wucknitz, O., McKean, J. P., & Garrett, M. A. 2010, A & A, 524, A79 Weiner, B. J., Coil, A. L., Prochaska, J. X., et al. 2009, APJ, 692, 187 Wuyts, E., Rigby, J. R., Sharon, K., & Gladders, M. D. 2012, APJ, 755, 73 Yee, H. K. C., Ellingson, E., Bechtold, J., Carlberg, R. G., & Cuillandre, J.-C. 1996, AJ, 111, 1783

101

Chapter

4

Stars were born in significantly denser regions in the early Universe The density of the warm ionized gas in high-redshift galaxies is known to be higher than what is typical in local galaxies on similar scales. At the same time, the mean global properties of the high- and low-redshift galaxies are quite different. Here, we present a detailed differential analysis of the ionization parameters of 14 star forming galaxies at redshift 2.6–3.4, compiled from the literature. For each of those high-redshift galaxies, we construct a comparison sample of low-redshift galaxies closely matched in specific star formation rate and stellar mass, thus ensuring that their global physical conditions are similar to the high-redshift galaxy. We find that the median log[O iii] 5007/[O ii] 3727 line ratio of the high-redshift galaxies is 0.5 dex higher than their local counterparts. We construct a new calibration between the [O iii] 5007/[O ii] 3727 emission line ratio and ionization parameter to estimate the difference between the ionization parameters in the high and lowredshift samples. Using this, we show that the typical density of the warm ionized gas in star-forming regions decreases by a median factor of 8 from z ∼ 3.3 to z ∼ 0 at fixed mass and specific star formation rate. We show that metallicity differences can not explain the observed density differences. Because the highand low-redshift samples are comparable in size, we infer that the relationship between star formation rate density and gas density must have been significantly less efficient at z ∼ 2 − 3 than what is observed in nearby galaxies with similar levels of star formation activity. Maryam Shirazi, Jarle Brinchmann and Alireza Rahmati Astrophysical Journal 2013, submitted

Denser star-forming regions in the early Universe

4.1

Introduction

The cosmic star-formation rate, averaged over all observed galaxies in the Universe, has dropped by a factor of > 10 during the last ∼ 10 Gyr (e.g., Hopkins & Beacom, 2006). In addition to the increasing fraction of actively star-forming galaxies with increasing look-back time, the star-formation rates of typical galaxies increases rapidly towards the earlier stages of galaxy formation (Noeske et al., 2007; Daddi et al., 2007; Elbaz et al., 2007, 2011). Several studies also provide hints that star formation conditions in distant galaxies (i.e., z ∼ 2 − 3) are significantly different from the nearby Universe: emission lines from ionized gas in and around star-forming regions show different characteristics in distant and nearby galaxies (Brinchmann et al., 2008b; Liu et al., 2008; Newman et al., 2013), actively star-forming galaxies show higher gas fractions at higher redshifts (Tacconi et al., 2010; Genzel et al., 2010) and clumpy star-forming disks become increasingly more prevalent at higher redshifts (Cowie et al., 1995; Elmegreen & Elmegreen, 2006; Genzel et al., 2011). The average density of the warm ionized gas in typical high-redshift (high-z) galaxies is also known to be significantly higher than in typical low-redshift (low-z) galaxies on similar scales (Elmegreen et al., 2009; Lehnert et al., 2009; Le Tiran et al., 2011; Newman et al., 2012; Tacconi et al., 2013; Lehnert et al., 2013). These studies have revealed that distant star-forming galaxies form a population of objects that are distinct from their nearby analogs. However, it is unclear whether the main difference between low-z and high-z star-forming galaxies is related to their strongly evolving global properties, such as stellar mass (e.g., Ilbert et al., 2013; Muzzin et al., 2013), star formation rate (e.g., Noeske et al., 2007; Daddi et al., 2007; Elbaz et al., 2007, 2011) or metallicity (e.g., Mannucci et al., 2010; Lara-L´ opez et al., 2010), or that the interstellar medium (ISM) conditions were significantly different in similar galaxies at high-z. Comparing representative samples of high-z and low-z star-forming galaxies (e.g. Rigby et al., 2011) cannot disentangle the evolution in global characteristics from the possibly evolving starformation conditions. We address this issue by selecting a comparison ensemble of low-z galaxies for each high-z star-forming galaxy in our sample, ensuring that the stellar mass and star-formation activities are similar in our high-z galaxies and their low-z comparison samples. This allows us to evaluate the differences in star-formation conditions between the high-z star-forming galaxies and their local analogs. Although observations of some lensed galaxies at high-z reach spatial resolutions of ∼100 pc (e.g., Swinbank et al., 2009; Jones et al., 2010), even this spatial resolution is insufficient to directly compare the small-scale properties of the ISM in high-z and low-z star-forming galaxies. However, these properties can be constrained through their impact on the emission line spectra of galaxies (e.g., Yeh & Matzner, 2012). Here we use emission line ratios to derive the average ionization parameter of star-forming regions. Since the ionization parameter is a measure of ionizing radiation intensity per unit density, we can use it to constrain the density of star-forming regions in distant galaxies and compare it with that of their nearby counterparts. 104

Data The structure of the paper is as follows. In Section 4.2 we introduce our highz sample and explain how we select their low-z counterparts. In Section 4.3 we introduce our new calibration for calculating the ionization parameter using the emission line ratios. We present our main results in Section 4.4 and compare the density of ionized gas in high-z and nearby galaxies. In Section 4.5 we investigate the impact of metallicity variations between the high-z and local galaxy samples on our results. We discuss the implications of our finding in Section 5.5 and end the paper with concluding remarks in Section 4.7.

4.2

Data

We have assembled a sample of 14 high-z star-forming galaxies from the literature for which published [O ii] λ3727, [O iii] λ5007 and Hβ emission line fluxes are available (they have [O iii] λ5007/Hβ > 0). This sample consists of 2 galaxies (RXJ1053, Cl0949) from Richard et al. (2011, R11); 7 galaxies from the AMAZE sample (Maiolino et al., 2008, M08); 4 galaxies from the LSD sample (Mannucci et al., 2009, M09), and the 8 o’clock arc (Dessauges-Zavadsky et al., 2011; Shirazi et al., 2013, arc). These galaxies span redshifts between z = 2.39 and z = 3.69 with a median redshift of z = 3.39. All these galaxies also have gas metallicity, stellar mass and star-formation rate estimates. To test our results further, we also use a sample of 3 galaxies in the SINS survey (F¨orster Schreiber et al., 2009, 2011) that have directly measured electron densities using [S ii] doublet (Lehnert et al., 2009). The physical properties of our high-z sample are summarized in Table 4.1. We compare these galaxies to matched samples of low-z galaxies from the Sloan Digital Sky Survey (SDSS) (York et al., 2000). We used the MPA-JHU1 value added catalogues (Brinchmann et al., 2004; Tremonti et al., 2004) for SDSS DR7 (Abazajian et al., 2009) as our parent sample and selected star-forming galaxies following Brinchmann et al. (2004), with the adjustments of the line flux uncertainties given in Brinchmann et al. (2013). Furthermore, we used SDSS DR8 (Aihara et al., 2011) photometry to estimate stellar masses. The median and 1-σ scatter of the physical properties of the low-z sample of each high-z galaxy are summarized in Table 4.2. As argued above, it is essential to take out correlations with global properties of galaxies when comparing their ISM conditions. To achieve this we select, for each high-z galaxy, all star-forming galaxies in the SDSS DR7 that have log M∗ and log SFR/M∗ within 0.3 dex of that of the high-z galaxy. We require that the SDSS galaxies to have z > 0.02 so that [O ii] λ3727, 29 are measured, they also have [O iii] λ5007/Hβ > 0. We note that for two galaxies in our high-z sample that have very high sSFR (∼ 0.3 Gyr−1 ), we had to include low-z galaxies whose log SFR/M∗ differ by up to 1 dex to find a local analog sample. By default, we do not explicitly constrain the low-z samples to match the metallicity and/or size of their high-z counterparts as this would reduce the size of our sample and in the case of metallicity is subject to systematic uncertainties (e.g. Kewley & Ellison, 2008). However, as we show below, matching metallicities and/or sizes does not 1 http://www.mpa-garching.mpg.de/SDSS/DR7

105

Denser star-forming regions in the early Universe affect our results significantly. Any significant contribution of ionizing radiation from an Active Galactic Nucleus (AGN) could bias our estimates of the ionization parameter. For the low-z sample we can exclude strong AGN using the BPT diagram (Baldwin, Phillips & Terlevich, 1981). At high-z, the galaxies from M08 and R11 do not show any evidence indicating the presence of AGNs in their rest frame UV spectra (i.e., [N v] ,[C iv] ,He ii or broad Lyα), X-ray and 24 µm Spitzer-MIPS observations (Maiolino et al., 2008; Richard et al., 2011; Shirazi et al., 2013). The LSD galaxies also show no evidence of AGN activity in X-ray observations (Mannucci et al., 2009). While the aforementioned arguments do not rule out the presence of some AGN activity that is optically thick for X-rays, this is unlikely to significantly influence the optical emission lines which originate in only moderately obscured regions. One galaxy from the SINS sample (Q2343-BX610) that we use in this study has an indication of possible AGN from mid-IR observations (F¨orster Schreiber et al., 2011; Hainline et al., 2012) and from an analysis of resolved spectroscopy presented by Newman et al. (2013). However, we note that we do not use our calibration to infer ionization parameter for the SINS galaxies. Thus, we conclude that AGN activity is unlikely to bias our results at high-z.

4.3

Methodology

The high-z galaxies all have measured [O iii]λ5007 and [O ii]λ3727 line fluxes. This allows us to use the strong sensitivity of the [O iii] λ5007/[O ii] λ3727 (hereafter O32) ratio to the ionization parameter (Penston et al., 1990) to estimate this. Kewley & Dopita (2002) derived an estimator for the ionization parameter using the dereddened O32 ratio. Since this can not easily be applied to our high-z sample in the absence of reliable reddening estimates, we here calibrate a new relation between the ionization parameter and the observed O32 ratio using the Charlot & Longhetti (2001, hereafter CL01) models that account for variations in dust properties and metallicities (see Table 4 in Shirazi & Brinchmann (2012) for the CL01 model grid used in our study). The effective ionization parameter in these models is taken to be the volume average over the Str¨omgren sphere (see equation 9 and 10 in CL01). We wish to construct a calibration between the ionization parameter and O32 ratio that treats the metallicity as a free parameter. Based on this approach, as long as our high-z and low-z samples do not differ greatly in metallicity we do not need to know this exactly. We discuss this assumption further below, but given that we still have several possible ways to construct the calibration from the CL01 models: a- Leaving all parameters in the CL01 models as free parameters in the fitting procedure (including all dust attenuation parameters, 0.01 < τV < 4 ). This is likely to give a large amount of scatter in the relationship. b- Using only models with τV ∼ 0.2 and leaving all other model parameters free. This fit is appropriate if line ratios are corrected for dust attenuation but there is no constrain on the dust-to-metal ratio (ξ). 106

Results c- Using only models with τV ∼ 0.2 and ξ ∼ 0.3 (i.e., the Galactic dust-to-metal ratio) and leaving all other model parameters free. Since ξ is expected to evolve weakly with time (Calura et al., 2008), it is reasonable to fix its value. d- Using ξ ∼ 0.3 and leaving all other model parameters free. Since ξ is likely not to differ strongly from this value, this is the best choice for a calibration when the amount of dust attenuation is unknown. These fits are plotted in Figure 4.1 from the top-left to the bottom-right, respectively. The best fits for the relation between ionization parameter and Log [O iii] λ5007/[O ii] λ3727 (Log O32) are summarised as equation 1 to 4, respectively. We use option d, Equation 4.4, as our reference in this study because in general we do not have enough information to accurately constrain the dust attenuation for the high-z galaxy sample. To derive our reference relation we fix ξ = 0.3, which is the Galactic value (see Brinchmann et al., 2013, for a discussion), and allow all other parameters to vary. We use the same fit for estimating the ionization parameter for low-z counterparts of high-z galaxies. Log U = −3.300 ± 0.017 + (0.481 ± 0.019) Log O32

(4.1)

Log U = −3.109 ± 0.039 + (0.586 ± 0.039) Log O32

(4.2)

Log U = −3.119 ± 0.027 + (0.804 ± 0.035) Log O32

(4.3)

Log U = −3.363 ± 0.011 + (0.593 ± 0.012) Log O32

(4.4)

We are primarily focused on relative statements in this work so the most important aspect of these calibrations is how they convert relative statements in O32 to relative statements about log U. Since the slope in equation 1, 2 & 4 is similar they will result in similar relative statements about log U, while that in equation 3 is even steeper and would lead to an even stronger result than that outlined below.

4.4

Results

The left and middle panels of Figure 4.2 compare the [O iii] λ5007/[O ii] λ3727 ratios and corresponding ionization parameters (from Equation 4.4) for our highz sample (colored symbols), and the median values of their low-z analogs (black circles). Error bars shown on the black circles indicate 1-σ scatter in the low-z sample of each high-z galaxy. It is evident that the high-z star-forming galaxies show significantly higher [O iii] λ5007/[O ii] λ3727 ratios (up to ≈ 0.8 dex higher) compared to their low-z analogs. This translates into significantly higher ionization parameters (up to ∼ 0.5 dex) in the high-z galaxies relative to low-z even though their star formation rates and masses are constrained to be the same. For a given production rate of hydrogen ionizing photons, Q, and after assuming that most of ionizing photons are absorbed locally, the ionization parameter in a typical ionized region can be related to the hydrogen number density, nH : U3 ∝ Q(t) nH  2 ,

(4.5)

where  is the volume filling factor of the ionized gas, which is defined as the ratio between the volume-weighted and mass-weighted average hydrogen densities 107

Denser star-forming regions in the early Universe -2.0

a

-2.0 -2.5 Log U

Log U

-2.5 -3.0

-3.5

-4.0

-4.0

-2 -1 0 1 2 Log [O III]5007/[O II]3727

-2 -1 0 1 2 Log [O III]5007/[O II]3727

c

-2.0

-3.0

-3.0

-3.5

-3.5

-4.0

-4.0

-2 -1 0 1 2 Log [O III]5007/[O II]3727

d

-2.5 Log U

-2.5 Log U

-3.0

-3.5

-2.0

b

-2 -1 0 1 2 Log [O III]5007/[O II]3727

Figure 4.1 Best-fit relations between ionization parameters and the [O iii] λ5007/[O ii] λ3727 (O32) ratios are shown by blue dashed lines. The top-left panel shows the best-fit using all CL01 models (0.01 < τV < 4), on the top-right we show the best-fit using only models with τV ∼ 0.2, the bottom-left panel shows the best-fit using only models with τV ∼ 0.2 and ξ ∼ 0.3 (i.e., Galactic dust-to-metal ratio), and in the bottom-right panel we show the best-fit to all models with ξ ∼ 0.3. The results in the paper are presented for the fit shown in the lower-right panel. (Charlot & Longhetti, 2001). This allows us to constrain the densities of starforming regions, by measuring their ionization parameters. Assuming that the production rate of hydrogen ionizing photons and volume filling factors of the ionized gas are similar in typical star-forming regions in high-z galaxies and their low-z analogs, one can translate the ratio between the ionization parameters of the high-z galaxies and their low-z counterparts into the ratio of their ionized gas densities. The difference between the density of the ionized gas in star-forming regions in our high-z galaxies and their low-z analogs is shown in the right panel of Figure 4.2. This shows up to ≈ 25 times higher densities in high-z star-forming galaxies. To derive physical densities for our high-z galaxies from the relative density differences shown in Figure 4.2, we exploit the fact that for the nearby galaxies we can estimate the electron density from the [S ii] λ6716, 6731 ratio and thus get an estimate of the electron density in the high-z galaxies. The resulting absolute densities for the ionized gas in our high-z star-forming galaxies are shown in Figure 4.3. The median values of the electron densities of the low-z samples, inferred 108

Metallicity dependence from the [S ii] 6716, 6731 doublet, are shown by the black filled circles in the figure where error bars show 1-σ scatter. The median values of the redshifts of the low-z samples and the number of low-z analogs in each sample are indicated with n in the figure. Colored symbols show our high-z sample with their redshifts indicated. The high-z values are inferred from the low-z values multiplied by ne (z)/ne (z = 0) ratios shown in Figure 4.2, and their error bars show propagation of uncertainties based on Equation 4.4. The grey small dashed and long dashed lines show the median values for the electron density at low-z and high-z, respectively. Besides the sensitivity of the ionization parameter to the density of the emitting gas, it also depends on the production rate of ionizing photons and the volume filling factor of the ionized gas (Charlot & Longhetti, 2001). Therefore, our density estimates might also be sensitive to the possible differences in the ionizing photons production rate and the volume filling factor of the ionized gas between high-z and nearby galaxies. To address this concern, in Figure 4.3 we show electron densities for a sample of five high-z star-forming galaxies in the SINS survey (F¨orster Schreiber et al., 2009; Lehnert et al., 2009) as purple diamonds. The electron density for these galaxies has been measured directly using the [S ii] λ6716, 6731 doublet and is in a good agreement with our inferred evolution in density estimated from the ionization parameter. For three of these five objects that have available stellar masses and specific star formation rates (F¨orster Schreiber et al., 2011), we constructed low-z analog samples. The comparison between the electron density of these three objects and their low-z analogs also shows good agreement (evolution in density with a median factor of 8.4) with the density ratios we obtained for our high-z star-forming galaxies using their ionization parameters (an evolution in density with a median factor of 7.9). This further strengthens our argument that an elevated density of star-forming regions in high-z galaxies is the main reason for their higher ionization parameter.

4.5

Metallicity dependence

A key result in this work is that high-z galaxies have a typically 0.5 dex higher Log O32 than low-z galaxies with the same mass and sSFR. We interpret this as primarily being due to a difference in ionization parameter but O32 is also sensitive to metallicity. Ideally we would select our high-z and low-z samples to have the same metallicity but to do this we require a metallicity estimator that can be applied equally at low-z and high-z allowing for a variation in ionization parameter. With the current data available for high-z galaxies this is not possible, thus we need to assess whether metallicity differences between the samples could be the reason for the observed offset. Mannucci et al. (2010) and Lara-L´ opez et al. (2010) showed that there is a relationship between stellar mass, metallicity and star formation rate (SFR) that appear to hold to high-z (z < 2.5 for Mannucci et al. and z < 3.5 for Lara-Lopez et al.). Therefore, if this holds for our galaxies, a selection on stellar mass and SFR should ensure that the metallicity difference between the high- and low-z sample is small. Given our small sample and considering that Mannucci et al. (2010) argued 109

Denser star-forming regions in the early Universe

Figure 4.2 A comparison between [O iii] λ5007/[O ii] λ3727 ratio, ionization parameter and electron density at low-z and high-z. The x-axis on the left panel shows the [O iii]λ5007/[O ii]λ3727 ratio, the middle panel shows the ionization parameter and the right panel shows the electron density at high-z relative to that of lowz. Colored symbols show high-z galaxies with their redshift indicated and black circles show the median values for the low-z sample of each high-z galaxy. Error bars span from the 16% to the 84% confidence level. We see that high-z galaxies show higher [O iii] λ5007/[O ii] λ3727 ratios than their low-z analogs (up to ≈ 0.8 dex higher), even though their masses and sSFR are the same. The middle panel shows the ionization parameters derived using our new calibration between the [O iii] λ5007/[O ii] λ3727 ratio and the ionization parameter. We see that high-z galaxies show up to ∼ 0.5 dex higher (median ∼ 0.3 dex) ionization parameters than their low-z analogs. This translates to up to 25 times higher electron density for high-z galaxies relative to their low-z analogs. that the multi-parameter relationship was not well established at z > 2.5, where most of our high-z galaxies lie, it is necessary to examine this assumption more carefully. It is useful to start this by asking what metallicity difference would give a O32 difference similar to what is observed. From Brinchmann et al. (2008b, their Figure 8), or directly using the CL01 models, we find that a change in metallicity from 1 Z to 0.1 Z leads to a change in Log O32 of 0.40 ± 0.07 dex. Thus we need a major difference in metallicity to explain the results. We can test for a large offset in metal content by calculating the metallicities of the high- and low-z samples in a consistent way. To do this we adopt the methodology used for AMAZE and LSD described in (Maiolino et al., 2008) for both high- and low-z galaxies. Note that, by construction, this method assumes that all variation in O32 is due to metallicity. Therefore, by using it, we will maximize the contribution of metallicity to the change in O32 and hence derive a minimum difference in ionization parameter between the low- and high-z objects. 110

Metallicity dependence

Figure 4.3 The median value of the electron density for the low-z samples inferred from the [S ii] 6716, 6731 doublet is shown by the black filled circles. The highz values are inferred from the low-z values multiplying by ne (z)/ne (z = 0) ratios shown in Figure 4.2. Colored symbols show our high-z sample sorted based on their redshifts from down to top as indicated on the figure. Five galaxies from the SINS survey that have directly measured electron densities are shown by purple diamonds. The median values of the redshifts of low-z samples are shown in black and the number of low-z analogs in each sample are indicated with n. Error bars span from the 16% to the 84% confidence level (low-z data: they show scatter in the sample, high-z data: they show propagation of uncertainties through Equation 4.4). The grey small dashed and long dashed lines show the median value for the electron density at low-z and high-z, respectively.

Based on the derived metallicities, we can calculate the maximum difference in O32 between high- and low-z galaxies due to metallicity differences, using the CL01 models and by averaging over U. This gives us the expected change in O32 due to metallicity only, and we subtract this off the actually observed difference for each galaxy. The resulting difference can be seen in the top panel of Figure 4.4. We emphasize that since we have used an abundance calibration that assumes that changes in O32 are due to metallicity, this correction should be the maximum possible correction. This gives a lower limit to the difference in O32 between high- and low-z galaxies and it is still quite sizeable. Converting this to a density difference as done in the main text we get the lower panel in that figure. This shows that the mean (median) electron density of the high-z galaxies is 5.5 (3.5) times higher than the low-z galaxies with the same sSFR and mass. 111

Denser star-forming regions in the early Universe

Min O32(z)-O32(z=0)

1.0 0.8 0.6 0.4 0.2

Min n(z)/n(z=0)

0.0 8 6 4 2 0 2.0

2.5

3.0 Redshift

3.5

4.0

Figure 4.4 Top panel: The minimum difference in O32 between the high- and lowz samples when corrected for metallicity as described in the text. Bottom panel: The resulting minimum density difference between the high- and low-z samples when corrected for metallicity.

To test further the sensitivity of our results to metallicity differences between our high-z galaxies and their low-z analogs, we made a low-z comparison sample for all high-z galaxies ensuring that their metallicities were equal to within 0.3 dex, in addition to matching their stellar masses and sSFRs2 . In this case we found that high-z galaxies show a median of ≈ 6.1 higher density compared to their low-z analogs with similar sSFRs, masses and metallicities; a result which is not significantly different from what we found without matching metallicities. We also note that the densities that are measured directly from the [S ii] doublet for the 5 high-z galaxies we selected from the SINS, are not derived using our calibration and hence are insensitive to variations in metallicity. Yet they have densities which are on average 8.4 times higher than their local analogs. It also worth noting that not all SINS galaxies have detected [S ii] which is consistent with these conclusions because [S ii]/Hα decreases with increasing U at fixed metallicity (e.g., Brinchmann et al., 2008a, their Figure 11). In conclusion, regardless of how we correct for possible differences in metallicity between the high- and low-z samples, the effect is minor and the main result of the paper is robust to these corrections. Thus we conclude that differences in metallicity can not explain the observed major offset in O32 and that systematic differences in the ionization parameter is the main cause. 2 Note

112

that this additional metallicity constraint decreases the low-z sample sizes.

Discussion

4.6

Discussion

The observed strong evolution in the global properties such as star formation intensity, stellar mass and size indicates that mean star formation conditions are different in distant galaxies compared to typical galaxies today (Cowie et al., 1995; Elmegreen & Elmegreen, 2006; Noeske et al., 2007; Daddi et al., 2007; Elbaz et al., 2007, 2011; Tacconi et al., 2010; Genzel et al., 2010, 2011). In this work we have however shown that even when the star formation intensity and mass are the same, the density in the ionized gas in high- and low-z galaxies differ dramatically. This difference would naturally imply a higher pressure in the colder ISM surrounding the ionized gas (Dopita et al., 2006), and thence its higher density. This could naturally occur if star formation at high-z was more concentrated to the central regions, so to check this we compared the u-band half-light radius for the SDSS galaxies with the half-light radius of the high-z galaxies when they are available (for 7 galaxies). Among the high-z galaxies, only one has a smaller size than the median size of its low-z counterparts. This is in agreement with the findings of Lehnert et al. (2009) and can not explain the density differences seen for any reasonable mass profile in the galaxies. We double-checked this by 2 constructing a matched low-z sample that have log SFR/πr1/2 within 0.3 dex of their high-z counterparts where r1/2 is the half-light radius. This results in a median density difference greater than 19 between low-z and high-z galaxies compared to a median difference of ∼ 8 before matching SFR densities. This shows that size differences are unlikely to be the explanation of the systematic differences. We have not required a match in SFR density in the bulk of the paper. However, because the size definitions are somewhat arbitrary and we do not have sizes for all galaxies at high-z. Assuming now that the distribution of star formation is comparable at low- and high-z, we next assume that the H ii regions are in pressure equilibrium with their surrounding ISM (Oey & Clarke, 1997; Dopita et al., 2006). Under this assumption the increased pressure in the ionized regions implies a higher pressure in the cold ISM. There are considerable uncertainties in how ionized regions expand in detail. However, in our case it is not unreasonable to assume that those complexities should be similar at high and low redshift. This is because the evolution of the H ii regions is driven by the energy injection from massive stars which should be similar at high-z and low-z, given how we selected our samples. The same applies to cosmic ray production rates which contribute to the heating (and the pressure support) of the ambient ISM. Note that this also means that the contribution of radiation pressure to the equilibrium for the H ii regions (e.g., Yeh & Matzner, 2012) should be similar at low and high redshift. It is hard to test whether the H ii regions in the high-z galaxies have reached pressure equilibrium with their surrounding ISM. However, since the mechanical input energy is the same at high and low redshift, and the life-times of the relevant stellar population is also the same, it seems unlikely that the evolutionary age of the H ii regions differs significantly between the high-z and low-z samples. This is also supported by Verdolini et al. (2013) who used a population study to show that the line emission of a galaxy will typically be dominated by the youngest 113

Denser star-forming regions in the early Universe H ii regions. Verdolini et al. (2013) also show explicitly the effect of an elevated ambient pressure on emission line ratios (their Figure 8), which is a qualitatively similar trend to what we infer here. Thus, the simplest explanation for the elevated density in the high-z H ii regions is an elevated pressure in the cold ISM relative to similar galaxies nearby. This increased pressure could arise from various sources, but in general, one would expect a pressure-density relation, P ∝ ργ with γ > 1. Thus, the increased pressure would correspond to an increased ISM density by an amount that depends on the model adopted for the ISM and we do not attempt to discuss this in detail here. The simplest model, where the ISM temperature is the same at high and low redshift, would predict that the density difference between the ISM at high- and low-z would be the same as that of the the ionized regions, i.e., ρhigh−z ∼ 8 ρlow−z . This conclusion has important implications for empirical star-formation law as well. The most popular scaling relation observed between star formation activity and gas surface density in the local Universe is the Kennicutt-Schmidt relation (Kennicutt, 1998), ΣSFR ∝ Σ1.4 (4.6) gas , where Σ denote surface densities. In our case ΣSFR is approximately the same in the high- and low-z galaxies (see above), but the gas density is much higher. If the scale-height of the gas is not significantly smaller in high-z galaxies, one can conclude that the scaling relation in the high-z galaxies is significantly different from what is observed in their low-z counterparts, being a factor ∼ 5–8 less efficient. We note however that we can not distinguish between molecular and atomic gas. Our results therefore are for the total gas and we cannot directly compare them to molecular studies (e.g., Daddi et al., 2010; Tacconi et al., 2013) at high-z and leave a discussion of this for future work.

4.7

Conclusion

In this work we compare the physical conditions of the ISM in high-z galaxies and their low-z counterparts which are selected to have similar global properties as that of high-z galaxies. This selection criteria minimize the differences between distant and nearby galaxies due to the evolution of the global properties such as mass and sSFR from high-z to low-z and can therefore be used to study the evolution of intrinsic properties of the ISM. Previous studies have already pointed out that the physical densities/properties of the star-forming regions at high-z are very different from those in the local Universe and we confirm this here. Using a novel approach, we have been able to go one step further, and show that this difference can not fully be explained by an increased star formation activity in the high redshift galaxies. Since we compare high and low-z galaxies that are matched in sSFR, their different densities must reflect an intrinsic difference in ISM conditions between high and low-z. We argue that this difference is primarily due to a difference in the density of the warm ionized gas. We have also shown that the differences between the high- and lowz galaxies can not be explained by differences in metallicity. By showing that 114

Conclusion the high-z and low-z samples are also comparable in size, we conclude that the relationship between star formation rate density and gas density must have been significantly less efficient at z ∼ 2 − 3 than what is observed locally. This, in turn, implies that most of the stars in the local Universe were formed following a different star formation scaling relation than what is observed in normal galaxies today.

Acknowledgments We thank Marijn Franx for helpful discussion. We also thank Leslie Sage and the two anonymous referees of Nature for very useful comments which improved this paper.

115

116

RXJ1053 Cl0949 SSA22a-C30 Q0302-C131 Q0302-M80 Q0302-C171 CDFa-C9 CDFS-4414 CDFS-4417 CDFS-16767 CDFS-2528 SSA22a-M38 SSA22a-aug16M16 8oclock Q2343-BX389 Q2343-BX610 Q2346-BX482

Name

Table 4.1 High-z sample.

R11 R11 M09 M09 M09 M09 M08 M08 M08 M08 M08 M08 M08 arc SINSa SINSa SINSa

ID

2.576 2.394 3.103 3.235 3.414 3.328 3.212 3.471 3.473 3.624 3.688 3.294 3.292 2.735 2.172 2.210 2.256

z

Log Mass M 9.620.75 −0.72 10.190.22 −0.18 10.330.31 −0.38 10.090.10 −0.33 10.070.23 −0.19 10.060.10 −0.28 10.180.40 −0.08 10.570.19 −0.22 10.290.37 −0.11 10.050.10 −0.16 9.760.09 −0.07 11.010.18 −0.41 10.290.20 −0.21 10.24−1.80 0.45 10.610.77 −2.16 11.002.70 −0.60 10.260.79 −0.46

Log sSFR yr−1 -8.66 -9.31 -8.87 -9.09 -8.96 -9.36 -7.76 -8.52 -7.65 -8.13 -7.76 -8.95 -8.67 -7.88 −9.22 −9.22 −8.36

SFR M yr−1 9.12.3 −2.3 7.51.5 −1.5 29.081.0 −21.0 10.06.0 −4.0 13.017.0 −8.0 2.0 5.0−2.0 265.00.0 0.0 113.00.0 0.0 438.00.0 0.0 84.00.0 0.0 101.00.0 0.0 115.00.0 0.0 0.0 42.00.0 10.0 228.0−10.0 ··· ··· ···

ΣS FR M yr−1 kpc−2 0.22 0.19 4.21 1.97 7.36 1.02 ··· ··· ··· ··· ··· ··· ··· 9.26 ··· ··· ··· r1/2 kpc 3.62± 0.45 3.50± 0.88 1.48± 0.44 1.27± 0.37 0.75± 0.24 1.25± 0.39 ··· ··· ··· ··· ··· ··· ··· 2.80± 0.20 ··· ··· ···

12 + LogO/H 8.680.11 −0.12 8.100.06 −0.05 8.160.20 −0.60 8.000.25 −0.40 8.360.15 −0.15 8.140.25 −0.45 8.100.18 −0.21 8.540.15 −0.14 8.550.09 −0.10 8.310.11 −0.17 8.070.39 −0.28 8.340.15 −0.12 7.990.26 −0.34 8.350.19 −0.19 ··· ··· ···

Log O32 0.5890.0 −0.0 0.4070.0 −0.0 0.6300.2 −0.0 0.5150.0 −0.4 0.3720.1 −0.1 0.2930.0 −0.2 0.5000.0 −0.0 0.0380.0 −0.0 0.2330.1 −0.0 0.5800.1 −0.1 0.4460.2 −0.0 0.1880.1 −0.0 0.5640.3 −0.2 0.661 ··· ··· ···

−3.010.02 −0.02 −3.120.02 −0.02 −2.990.02 −0.02 −3.060.02 −0.02 −3.140.02 −0.02 −3.190.01 −0.01 −3.070.02 −0.02 −3.340.01 −0.01 −3.220.01 −0.01 −3.020.02 −0.02 −3.100.02 −0.02 −3.250.01 −0.01 −3.030.02 −0.02 −2.970.02 −0.02 ··· ··· ···

Log U

ne cm−3 897.31145.2 −549.2 1229.41713.1 −789.8 2766.62286.8 −1623.8 2325.3 1682.9−1044.2 796.2946.5 −483.8 765.11192.4 −495.7 912.61278.7 −489.9 158.8179.9 −109.5 342.0448.9 −268.9 840.71746.3 −425.0 406.5237.1 −319.3 978.51862.6 −588.7 1698.71698.0 −934.8 391.5310.0 −310.0 1200.0700 −400 400.0700 −300 1200.0700 −400

Denser star-forming regions in the early Universe

0.150.07 −0.07 0.150.07 −0.05 0.210.05 −0.08 0.160.07 −0.05 0.170.06 −0.06 0.130.07 −0.05 0.200.07 −0.08 0.260.03 −0.10 0.190.09 −0.10 0.230.04 −0.05 0.240.04 −0.09 0.270.02 −0.09 0.220.05 −0.09 0.270.00 −0.22 0.200.05 −0.08 0.250.02 −0.09 0.220.06 −0.10

h z ia

Table 4.2 Low-z sample.

RXJ1053 Cl0949 SSA22a-C30 Q0302-C131 Q0302-M80 Q0302-C171 CDFa-C9 CDFS-4414 CDFS-4417 CDFS-16767 CDFS-2528 SSA22a-M38 SSA22a-aug16M16 8oclock Q2343-BX389 Q2343-BX610 Q2346-BX482

High-z ID

hLog Massi M 9.590.20 −0.19 10.040.19 −0.11 10.180.19 −0.11 0.21 9.96−0.12 0.21 9.94−0.13 9.930.20 −0.12 10.050.15 −0.13 10.420.21 −0.10 10.100.20 −0.08 0.17 9.90−0.11 0.16 9.48−0.02 10.860.23 −0.12 10.120.19 −0.10 10.030.00 −0.07 10.440.17 −0.10 10.840.27 −0.11 10.070.13 −0.06

hLog sSFRi yr−1 -8.81 -9.32 -9.03 -9.19 -9.08 -9.37 -8.4 -8.70 -8.0 -8.38 -7.93 -9.02 -8.84 -8.04 -9.27 -9.27 -8.55

hLog SFRi M yr−1 0.90.2 −0.3 0.80.3 −0.3 1.20.2 −0.2 0.80.2 −0.3 0.90.2 −0.2 0.60.3 −0.3 1.742.2 −0.2 1.60.4 −0.2 1.902.0 −0.3 1.60.2 −0.1 0.1 1.7−0.3 1.80.3 −0.3 1.40.2 −0.2 1.80.0 −2.0 1.20.2 −0.2 1.50.2 −0.1 1.60.1 −0.1 hΣS FR i M yr−1 kpc−2 1.91 0.93 3.70 1.35 1.86 0.53 17.63 12.18 25.71 14.91 22.25 3.72 6.25 17.63 2.49 5.23 12.15

hr50u−band i kpc 1.110.41 −0.24 1.460.65 −0.42 1.220.32 −0.27 1.320.54 −0.33 1.240.41 −0.29 1.530.84 −0.46 0.950.33 −0.95 1.090.43 −0.17 0.910.39 −0.91 1.060.21 −0.29 0.830.34 −0.14 1.893.59 −0.72 1.170.28 −0.24 3.010.00 −1.95 1.400.51 −0.36 1.550.65 −0.52 1.080.27 −0.20 0.0120.163 −0.161 −0.2600.118 −0.088 −0.1570.138 −0.101 −0.2230.137 −0.100 −0.1740.143 −0.115 −0.2810.125 −0.090 0.0480.146 −0.364 −0.0450.199 −0.128 0.0190.096 −0.316 0.092 0.129−0.068 0.3090.047 −0.130 −0.2310.277 −0.029 −0.0820.149 −0.119 0.000 0.236−0.747 −0.2050.113 −0.085 −0.1970.140 −0.061 0.0840.110 −0.087

hLog O32i

h12 + LogO/Hi M08 calib b 8.490.12 −0.05 8.700.05 −0.07 8.650.07 −0.10 8.670.07 −0.09 8.640.08 −0.10 8.710.05 −0.08 8.440.08 −8.44 8.570.13 −0.08 8.440.13 −8.44 8.450.10 −0.01 8.410.05 −0.09 8.710.04 −0.14 8.600.08 −0.12 8.780.00 −0.40 8.700.04 −0.08 8.710.04 −0.09 8.470.09 −0.03

−3.360.10 −0.10 −3.520.07 −0.05 −3.460.08 −0.06 −3.500.08 −0.06 −3.470.09 −0.07 −3.530.07 −0.05 −3.330.09 −0.22 −3.390.12 −0.08 −3.380.09 −0.16 −3.290.05 −0.04 −3.180.03 −0.08 −3.500.16 −0.02 −3.410.09 −0.07 −3.440.22 −0.22 −3.480.07 −0.05 −3.480.08 −0.04 −3.310.07 −0.05

hLog Ui

hne i cm−3 84.6107.9 −51.8 80.0111.5 −51.4 110.291.1 −64.7 81.8113.1 −50.8 100.9 84.9−51.6 73.0113.8 −47.3 143.1200.6 −76.8 112.9128.0 −77.9 142.3186.9 −111.9 132.6275.4 −67.0 232.1135.4 −182.3 335.1 176.0−105.9 120.5120.4 −66.3 68.754.4 −54.4 112.5127 −68 153.5228 −94 143.1100 −65

Conclusion

117

REFERENCES

References Abazajian, K. N., Adelman-McCarthy, J. K., Ag¨ ueros, M. A., et al. 2009, ApJS, 182, 543 Aihara, H., Allende Prieto, C., An, D., et al. 2011, ApJS, 193, 29 Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 5 Brinchmann, J., Charlot, S., Kauffmann, G., et al. 2013, MNRAS, doi:10.1093/mnras/stt551 Brinchmann, J., Charlot, S., White, S. D. M., et al. 2004, MNRAS, 351, 1151 Brinchmann, J., Kunth, D., & Durret, F. 2008a, A&A, 485, 657 Brinchmann, J., Pettini, M., & Charlot, S. 2008b, MNRAS, 385, 769 Calura, F. and Pipino, A. and Matteucci, F., 2008, A&A, 479, 669 Charlot, S., & Longhetti, M. 2001, MNRAS, 323, 887 Cowie, L. L., Hu, E. M., & Songaila, A. 1995, AJ, 110, 1576 Daddi, E., Dickinson, M., Morrison, G., et al. 2007, ApJ, 670, 156 Daddi, E., Elbaz, D., Walter, F., et al. 2010, ApJL, 714, L118 Dessauges-Zavadsky, M., Christensen, L., D’Odorico, S., Schaerer, D., & Richard, J. 2011, A&A, 533, A15 Dopita, M. A., Fischera, J., Sutherland, R. S., et al. 2006, ApJ, 647, 244 Elbaz, D., Daddi, E., Le Borgne, D., et al. 2007, A&A, 468, 33 Elbaz, D., Dickinson, M., Hwang, H. S., et al. 2011, A&A, 533, A119 Elmegreen, B. G., & Elmegreen, D. M. 2006, ApJ, 650, 644 Elmegreen, D. M., Elmegreen, B. G., Marcus, M. T., et al. 2009, ApJ, 701, 306 F¨ orster Schreiber, N. M., Genzel, R., Bouch´e, N., et al. 2009, ApJ, 706, 1364 F¨ orster Schreiber, N. M., Shapley, A. E., Genzel, R., et al. 2011, ApJ, 739, 45 Genzel, R., Tacconi, L. J., Gracia-Carpio, J., et al. 2010, MNRAS, 407, 2091 Genzel, R., Newman, S., Jones, T., et al. 2011, ApJ, 733, 101 Hainline, K. N., Shapley, A. E., Greene, J. E., Steidel, C. C., Reddy, N. A., & Erb, D. K. 2012, ApJ, 760, 74 Hopkins, A. M., & Beacom, J. F. 2006, ApJ, 651, 142 Ilbert, O., McCracken, H. J., Le Fevre, O., et al. 2013, arXiv:1301.3157 Jones, T. A., Swinbank, A. M., Ellis, R. S., Richard, J., & Stark, D. P. 2010, MNRAS, 404, 1247 Kennicutt, Jr., R. C. 1998, ApJ, 498, 541 Kewley, L. J., & Dopita, M. A. 2002, ApJS, 142, 35 Kewley, L. J., & Ellison, S. L. 2008, ApJ, 681, 1183 Lara-L´ opez, M. A., Cepa, J., Bongiovanni, A., et al. 2010, A&A, 521, L53 Lehnert, M. D., Nesvadba, N. P. H., Le Tiran, L., et al. 2009, ApJ, 699, 1660 Lehnert, M. D., Le Tiran, L., Nesvadba, N. P. H., et al. 2013, A&A, 555, A72 Le Tiran, L., Lehnert, M. D., van Driel, W., Nesvadba, N. P. H., & Di Matteo, P. 2011, A & A, 534, L4 Liu, X., Shapley, A. E., Coil, A. L., Brinchmann, J., & Ma, C.-P. 2008, ApJ, 678, 758 Maiolino, R., Nagao, T., Grazian, A., et al. 2008, A&A, 488, 463 Mannucci, F., Cresci, G., Maiolino, R., Marconi, A., & Gnerucci, A. 2010, MNRAS, 408, 2115 118

REFERENCES Mannucci, F., Cresci, G., Maiolino, R., et al. 2009, MNRAS, 398, 1915 Muzzin, A., Marchesini, D., Stefanon, M., et al. 2013, arXiv:1303.4409 Newman, S. F., Shapiro Griffin, K., Genzel, R., et al. 2012, ApJ, 752, 111 Newman, S. F., Buschkamp, P., Genzel, R., et al. 2013, arXiv:1306.6676 Noeske, K. G., Weiner, B. J., Faber, S. M., et al. 2007, ApJL, 660, L43 Oey, M. S., & Clarke, C. J. 1997, MNRAS, 289, 570 Penston, M. V., Robinson, A., Alloin, D., et al. 1990, A&A, 236, 53 Richard, J., Jones, T., Ellis, R., et al. 2011, MNRAS, 413, 643 Rigby, J. R., Wuyts, E., Gladders, M. D., Sharon, K., & Becker, G. D. 2011, ApJ, 732, 59 Shirazi, M., & Brinchmann, J. 2012, MNRAS, 421, 1043 Shirazi, M., Vegetti, S., Nesvadba, N., et al. 2013, submitted to MNRAS, arXiv:1306.6282 Swinbank, A. M., Webb, T. M., Richard, J., et al. 2009, MNRAS, 400, 1121 Tacconi, L. J., Genzel, R., Neri, R., et al. 2010, Nature, 463, 781 Tacconi, L. J., Neri, R., Genzel, R., et al. 2013, ApJ, 768, 74 Tremonti, C. A., Heckman, T. M., Kauffmann, G., et al. 2004, ApJ, 613, 898 Verdolini, S. and Yeh, S. C. C. and Krumholz, M. R. and Matzner, C. D. and Tielens, A. G. G. M., 2013, ApJ, 769, 12 Wisnioski, E., Glazebrook, K., Blake, C., et al. 2012, MNRAS, 422, 3339 Wuyts, E., Rigby, J. R., Sharon, K., & Gladders, M. D. 2012, APJ, 755, 73 Yeh, S. C. C., & Matzner, C. D. 2012, ApJ, 757, 108 York, D. G., Adelman, J., Anderson, Jr., J. E., et al. 2000, AJ, 120, 1579

119

Chapter

5

On the spatial distribution of star formation in distant and nearby galaxies Distant star-forming galaxies show more clumpy structures in the color maps that are star formation indicators compared to nearby star-forming galaxies. At the same time, the mean global properties of galaxies can be significantly different between distant and nearby Universe. In this work, we study the differences between the spatial distribution of star formation between a sample of distant galaxies from the Hubble Ultra Deep Field and nearby star-forming galaxies in the Sloan Digital Sky Survey with similar stellar mass and specific star formation rate. We construct spatial maps of physical properties using multi-band imaging data and compare the maps derived for high-redshift galaxies to those derived for low-redshift data. In general, we find that the stellar mass is more centrally concentrated in distant galaxies that we study here, but star formation is more extended compared to local galaxies with the same global properties. More massive galaxies at high-redshift are more concentrated than their low-redshift counterparts while less massive galaxies show more similar concentration derived from the specific star formation maps at low and high redshifts. On the basis of the maps of specific star formation rate, we find that high- and low-redshift galaxies with the same global properties also show more clumpy distributions of specific star formation rate. Maryam Shirazi and Jarle Brinchmann Monthly Notices of the Royal Astronomical Society 2013, to be submitted

Spatial distribution of star-formation

5.1

Introduction

During the past decade, high resolution observations of high-redshift (high-z) star-forming galaxies have revealed an increased fraction of clumpy actively starforming galaxies (Griffiths et al., 1994; Windhorst et al., 1995; Cowie et al., 1995; Abraham et al., 1996; Elmegreen & Elmegreen, 2006; Elmegreen et al., 2007; Genzel et al., 2011; Wuyts et al., 2012). This finding has motivated many studies to investigate the nature of these clumps and their implications for galaxy evolution, from both observational and theoretical point of view. To study the nature of these clumps observationally, the resolved color information and stellar population of their host galaxies and/or IFU observations are needed. These have been provided by analyzing the rest-frame UV and also the rest-frame optical emission of z ∼ 2 galaxies using Near-IR observations. Studies of the spatial variations of the stellar populations properties on ∼ kpc scales were started by Abraham et al. (1999) at intermediate redshifts using the Hubble Deep Field data and continued by Thompson et al. (1999), Dickinson et al. (2000), Elmegreen et al. (2009) and F¨ orster Schreiber et al. (2011b) for a sample of high-z objects using NICMOS observations. Detailed analyses of the resolved colors were also studied for a sample of clumpy galaxies in CANDELS by Guo et al. (2011a,b). Recently, Wuyts et al. (2012) studied both the resolved colors and stellar populations of a complete sample of 649 star-forming galaxies at 0.5 < z < 2.5 with stellar masses of > 1010 M and specific star formation rate (sSFR) of log(sS FR) > −9.76 at z ∼ 1, and log(sS FR) > −9.51 at z ∼ 2 using deep 7-band ACS+WFC3 imaging in the GOODS-South field. Using this high resolution data, they showed that the inferred stellar distributions of these galaxies are less clumpy than their star formation distributions. They argued that the clumpy rest-frame UV morphologies and the smooth stellar mass distribution of these galaxies might imply that the star formation history at a given galactocentric radius is uniform, but varies spatially on timescales similar to the lifetime of OB stars. IFU observations of clumpy star-forming galaxies at high-z have also provided insightful information about the physical nature of these clumps and their formation and evolution. The spatially-resolved observations have revealed large-scale outflows from individual star-forming clumps in some of these high-z galaxies (e.g., Genzel et al., 2011; Newman et al., 2012a,b). These powerful outflows play an important factor in setting the lifetimes of these clumps (e.g., Genzel et al., 2011). The spatially-resolved studies of the ionized gas and molecular gas find disk-like kinematics for many of these clumpy galaxies (Genzel et al., 2006; Shapiro et al., 2008; Bournaud et al., 2008; Contini et al., 2012; Daddi et al., 2010; Epinat et al., 2012; Tacconi et al., 2013). In the low-redshift (low-z) Universe however, mergers are known to be responsible for making clumpy morphologies (Conselice et al., 2003; Lotz et al., 2004). As only a minority of the observed high-z galaxies are in the process of merging (see e.g., Bundy et al., 2009; L´ opez-Sanjuan et al., 2009; Williams et al., 2011; Lotz et al., 2008, 2011a; Conselice et al., 2009), mergers can not be the dominant physical process causing clumpy morphologies at high-z. The formation mechanism for making the majority of these clumps is known 122

High-z sample to be gravitational instability (e.g., Genzel et al., 2011) where these instabilities can be caused by high fraction of dense molecular gas (Daddi et al., 2010; Tacconi et al., 2010). From a theoretical point of view, cosmological simulations predict that at high-z, gas accretion are playing a significant role for fueling star formation (Kereˇs et al., 2005; Dekel et al., 2009). Based on this scenario, gas accretion into the halo can continuously fuel the disk and actively star-forming clumps seen in the Hα and rest-frame UV maps of high-z galaxies can then be caused by gravitational instabilities within these gas-rich disks in this picture (Genzel et al., 2011). While the morphologies of high and low-z galaxies are well studied, they can be challenging to compare because the mean properties of the galaxies can be significantly different. Indeed, the energy input into the galaxies can be a factor of several higher at high-z because the typical star formation rates are much higher. This difference in energy balance could naturally lead to differences in the overall structure of star formation between high- and low-z galaxies. To compare the physical processes that make clumpy morphologies at high-z and low-z, one should select galaxies that have the same global properties such as mass and sSFR. Studying the resolved color and stellar populations of galaxies with similar properties at low-z and high-z makes it possible, at least in principle, to take out the effect of the star formation rate (SFR) on the morphological properties of the galaxies. For instance, whether an internal process such as SFR or external effects such as gas accretion which is not likely at low-z are responsible. In this work, we study the distribution of star-formation in high-z star-forming galaxies and their low-z counterparts. We use the multi-band imaging data available in the Hubble Ultra Deep Field (HUDF) and compare this quantitatively with low-z data from the Sloan Digital Sky Survey (SDSS). In Section 5.2, we introduce our high-z data and infer resolved physical parameter maps for the sample. In this section, we also contrast the properties inferred from the resolved analysis to those inferred from modeling of the integrated photometry. In Section 5.3, we present the sample selection of the low-z counterparts and explain the resolved stellar population modeling of our low-z sample. We compare the clumpiness (M20), concentration (Gini) of the low-z and high-z samples in Section 5.4. Finally, we discuss our results in 5.5 and conclude in 5.6.

5.2

High-z sample

In this study, we make use of the objects in the HUDF catalogue by Coe et al. (2006) that have confirmed spectroscopic redshifts 1 . That amounts to a total of 57 objects with 0.127 < z < 5.819 and median redshift of 0.667, among which 36 objects have 0.5 < z < 1.5. In our analysis, we exclude 7 objects from the high-z sample for which we infer an integrated sSFR < −11. The total HUDF catalogue consists of 18,700 objects and comes with a segmentation map. We use the same segmentation map as the basis for identification of our objects. 1 http://adcam.pha.jhu.edu/~coe/UDF/paper/zspec.cat

123

Spatial distribution of star-formation

5.2.1

Resolved stellar population modeling

Spatially resolved spectral energy distribution (SED) modeling of broad-band multi-band imaging of galaxies can be have spatial biases because of the significant contribution of low signal-to-noise (S/N) pixels in the outskirts of the galaxies (Welikala et al., 2009). Therefore, in order to have a sufficiently high S/N ratio for resolved SED modeling, we first bin our data following the Voronoi 2D binning technique by Cappellari & Copin (2003) to have approximately constant S/N across the images.

5.2.2

The Voronoi binning of the UDF data

As we mentioned above, to improve the color measurements in the outskirts of the UDF galaxies where the noise is significant, we use the Voronoi tesselation technique used for IFU data by Cappellari & Copin (2003) and later extended to X-ray data by Diehl & Statler (2006). In this work, we use the implementation presented by Diehl & Statler (2006) and carry out our procedure for both target S/N of 10 and 20. Based on the Voronoi binning technique, we aggregate pixels in approximately circular bins until the desired S/N is reached in that bin. The SED fitting is sensitive to the colors so to avoid biases we require that the color maps have approximately uniform S/N. To do this, we create our bins using one color image and apply this binning to the flux images subsequently. Ideally we should create bins using all images at the same time, but here we focus on a single color. When the integrated S/N in the B-band image is twice the target S/N, we use the B − V color image, otherwise we use the V − I image. This choice is a balance between creating a reasonable number of bins per object and good S/N per effective pixel. For the higher redshift objects the S/N in the B-band image can be rather poor and using this for binning would lead to very few bins. Examples of the Voronoi maps are shown in Figure 5.1 for six different objects. Four panels that belong to one object shows the unbinned V-band image and the Voronoi binned B − V, V − I and i0 − z0 images, as indicated in the panels. From these maps we can see how the Voronoi binning procedure isolates regions of different colors — which often translates into isolating regions with distinct physical properties (e.g., star formation rates). Note for instance 3088 for which the high-S/N color maps highlights a blue region along the major axis of the galaxy while also showing a ring-like red feature around the galaxy. The target S/N for this illustration is 10.

5.2.3

Fitting models to the color map of the UDF data

We fit Bruzual & Charlot (2003, , hereafter BC03) models to the B, V, i0 and z0 fluxes for each effective pixel of the UDF data that is given by the Voronoi binning procedure. We adopt the approach introduced by Kauffmann et al. (2003). This relies on a set of models with variable star formation histories, as described in detail by Gallazzi et al. (2005, 2008). For each effective pixel we calculate the χ2 between our observed fluxes and the stochastic model grid, which gives us a likelihood for each model, P(model|B, V, i0 , z0 ) under the assumption that the 124

High-z sample

Figure 5.1 The procedure of Voronoi binning galaxy images of the UDF data is illustrated for six different objects. For each object the B − V, V − i0 , i0 − z0 (binned) and V-band images are shown. The spectroscopic redshift and also the ID of each object as presented in the Coe et al. (2006) catalogue are also indicated above each panel. The color range in the color images is the same in each panel. uncertainties involved are Gaussian. Then we construct a likelihood distribution of various quantities for each effective pixel. We calculate the stellar mass (M∗ ), the star formation rate (SFR), sSFR (sSFR= SFR/M∗ ) and the r-band weighted age, and store the 100 most likely models and their likelihoods. As the area belongs to each effective pixels derived from Voronoi binning might be large, we draw random values from these likelihoods using Monte-Carlo resampling to populate each pixel of the derived maps. The result of this SED fitting procedure can be seen in Figure 5.2 for all 47 UDF objects studied in this work. The color images are shown in the left-most column, where the redshift and the ID of the UDF objects are also indicated. Other columns from left to right show the Voronoi binned B − V color map, the stellar mass map, the SFR map, the sSFR map and the age map, respectively. The pixel scale in these maps is 0.03 arcsec/pixel.

5.2.4

Integrated properties

In order to measure the integrated physical properties of the UDF sample, we fit BC03 models to the integrated B, V, i0 , z0 photometry presented in the Coe et al. (2006) catalogue. This gives us M∗ , SFR, sSFR, and the r-band weighted age for each UDF object. We also derive the total values of the physical parameters from the spatially resolved stellar population modeling. A comparison of global stellar 125

Spatial distribution of star-formation

Stamp Z= 0.13

B-V bins 1"

Log Mass 1.2 1.0

Log SFR 5.0 4.5

0.8 0.6 0.4 0.2

3088

Z= 0.66

0.0

1"

4.0 3.5 3.0

Log sSFR -8.5

-5.0

-9.0

-5.5

-9.5

-6.0

-10.0

-6.5

1.0 6.0

Log age

-4.5

9.4 9.2 9.0 8.8 8.6 8.4

-10.5

8.2

-8.5

9.0

-3.0

0.8

8.8

0.6

-3.5

-9.0

-4.0

-9.5

8.6

5.5 0.4

8.4

0.2 5.0

1375

Z= 0.74

0.0

1"

1.0 0.8

7.0 6.5

0.6

8.2

-4.5

-2.0

-8.5

-2.5

-9.0

-3.0

-9.5

-3.5

-10.0

-4.0

-10.5

6.0 0.4 0.2

8810

Z= 0.74

0.0

1"

4142

Z= 0.95

1"

6206

Z= 1.09

1"

8261

Z= 1.10

8749

5.5

1"

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

7.0 6.5 6.0 5.5 5.0

7.5

-1.0 -1.5 -2.0 -2.5 -3.0 -3.5 -4.0

9.6 9.4 9.2 9.0 8.8 8.6 8.4 8.2

-10.0

9.4 9.2 9.0 8.8 8.6 8.4 8.2

-10.5

9.4

-8.5 -9.0 -9.5

-3.5 -4.0

7.0

-11.0 -4.5 -5.0

-11.5

-5.5

-12.0

9.2

6.5

9.4 7.5 7.0 6.5

8.0 7.5 7.0 6.5 6.0

-1.5 -2.0 -2.5 -3.0 -3.5 -4.0 -4.5

-0.5 -1.0 -1.5 -2.0 -2.5 -3.0 -3.5

-9.0 -9.5 -10.0 -10.5 -11.0 -11.5 -12.0

-8.5

9.2 9.0 8.8 8.6 8.4

8.8 8.6

-9.0 8.4 -9.5 8.2 -10.0

8.0

Figure 5.2 The results of the SED fitting procedure for 47 UDF galaxies studied here. The left column shows the color JPEG image of the object with the redshift and UDF ID indicated (note that the region and orientation shown here is different from the images in the following columns). The next columns from left to right show the Voronoi binned B − V color map for S/N of 10, the stellar mass maps (log M∗ /M ), the SFR maps (log SFR/M yr−1 ), the sSFR maps (log sSFR/yr−1 ) and r-band luminosity-weighted age maps (log age/yr) all on a spatial scale of 0.03 arcsec/pixel. 126

High-z sample

Z= 1.22

-1.5

1"

0.6

7.0 -2.0

0.4

6.5

-8.5

9.2

-9.0

9.0 8.8

-2.5 -9.5 -3.0

0.2 6.0

Z= 1.22

8.6 8.4 8.2

0.0

1"

8.0 0.8 7.5 0.6 7.0 0.4 6.5 0.2 6.0

4396

Z= 1.24

-10.0 -3.5

4816

-0.5 -1.0 -1.5 -2.0 -2.5 -3.0 -3.5

-9.0

9.4 9.2

-9.5 9.0 -10.0

8.8

-10.5

8.6

0.0

1"

-1.5

0.6

-8.5

7.0

9.0

-2.0 0.4

6.5

-9.0

8.8

-2.5 0.2

5.5

1829

Z= 1.24

6.0

8.6 -3.0 -3.5 -10.0

0.0

1"

0.8 0.6

-9.5

-1.0

-8.5

-1.5

-9.0

7.0

9.0

6.5

Z= 1.29

-3.5

0.0

1"

-9.5

8.8

-10.0

8.6

-10.5

8.4

-2.5 -3.0

0.2

1266

8.2

9.2 7.5

-2.0 0.4

8.4

0.8

9.2

-2.0 7.0

-9.0

0.6

9.0

-2.5 0.4 0.2

7995

Z= 1.31

-3.0

-9.5

-3.5

-10.0

8.6

6.0 8.4

0.0

1"

6188

Z= 1.32

8.8

6.5

1"

1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

0.6

-3.0 7.5

-10.5

-3.5 -11.0

9.4

-4.0 7.0

-11.5 -4.5 -5.0

7.0

-2.0

0.4

6.5

-2.5

0.2

6.0

-3.0

9.2 -12.0

9.0 -9.0 8.8 8.6 8.4

7725

Z= 1.43

-9.5

0.0

1"

0.8

7.5

-1.5

0.6

7.0

-2.0

0.4

6.5

-2.5

0.2

8461

6.0

-8.5 9.0 -9.0

-3.0

8.8 8.6

-9.5

8.4

0.0

Figure 5.2 Continued.

127

Spatial distribution of star-formation

Z= 0.13

1"

5670

Z= 0.21

1"

-4.5 1.2 1.0 0.8 0.6 0.4 0.2 0.0

5.0

1.0

5.0

-5.0

-9.0

-5.5

-9.5

4.0 -6.0 3.5

-10.0

-6.5

0.4

8.6 8.4

-3.5 -4.0

8.8 -8.5 8.6

4.5 0.6

8.8

-10.5

3.0

0.8

9.2 9.0

4.5

-4.5

8.4 -9.0

4.0

-5.0

3.5

-5.5

0.0

1.0

5.0

-3.5

8.2

0.2

5620

Z= 0.21

1"

0.6 0.4

8.6 -4.5

4.0

5606

Z= 0.32

-9.0

-5.0 -5.5

0.0

1"

8.8

-4.0 4.5

0.2

Z= 0.23

8.0

9.0 -8.5

0.8

1000

-9.5

8.4 8.2

-9.5

8.0

-9.5

9.2

-10.0

9.0

-3.5 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

1"

6.0 -4.0 5.5 5.0

-4.5 -5.0

4.5

-5.5

4.0

-6.0

8.8 -10.5

6.0

9.4

-3.5 0.4

0.2

-9.0 5.5

-4.0

5.0

-4.5

4.5

-5.0

9.2 9.0

-9.5

8.8 8.6

5190

Z= 0.33

0.0

1"

7847

Z= 0.34

1"

3492

Z= 0.35

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

7.0

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

6.0

6.5 6.0 5.5 5.0

-2.0 -2.5 -3.0 -3.5 -4.0 -4.5 -5.0 -5.5 -6.0

-3.0

-10.0

9.4 -9.0 -9.5

-4.0

9.0

-10.5

8.8

-11.0

8.6

-8.5

9.0 8.8

-9.0 8.6

5.0

-4.5

4.5

-5.0

4.0

-5.5

-9.5

8.4

-10.0

8.2

-8.5

1"

9.0

0.4

-4.5 4.5

-9.0

-9.5

Figure 5.2 Continued.

128

0.0

4.0

-5.5

8.8 8.6

-5.0

0.2

4267

9.2

-10.0

-3.5 5.5

8.4

8.4 8.2

High-z sample

Z= 0.35

1"

3268

Z= 0.38

1"

8585

Z= 0.42

1"

1"

2107

Z= 0.67

1"

2607

Z= 0.67

1"

968

Z= 0.67

1"

662

Z= 0.67

6.0

-8.5 -3.5

8.8

5.5 5.0

-4.0

-9.0

8.6 8.4

4.5

-4.5 -9.5

8.2

4.0

1.2

-2.5 9.4

1.0

6.5

-3.0

0.8

6.0

-3.5

0.6

5.5

-4.0

0.4

5.0

-4.5

0.2

4.5

-5.0

7.0

-2.5 -3.0 -3.5 -4.0 -4.5 -5.0 -5.5 -6.0 -6.5

-9.0

9.2 9.0

-9.5 8.8 -10.0

8.6

0.0

900

Z= 0.53

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

1"

355

1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

1.2 1.0 0.8 0.6 0.4 0.2 0.0

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

6.5 6.0 5.5 5.0

-9.5

9.4

-10.0 -10.5 -11.0

9.2 9.0

-11.5 8.8

7.5 -3.0

-9.5 9.4

7.0 6.5

-3.5

-10.0

-4.0

-10.5

9.2

-11.0

9.0

-4.5 -5.0

6.0

7.0 6.5

-5.5

-11.5

-3.0

-9.0

-3.5

-9.5

9.2 9.0

-4.0 6.0

-10.0

-4.5 5.5

-10.5 -5.0

6.5

-2.5

6.0

-3.0

-8.5

-3.5

5.0

-4.0

8.6

9.0 8.8

-9.0

5.5

8.8

8.6 8.4

-9.5

8.2

4.5

6.5 6.0 5.5 5.0

-2.0 -2.5 -3.0 -3.5 -4.0 -4.5 -5.0 -5.5

-8.5

-3.0

-9.0

9.2 -9.0 9.0 -9.5

8.8

-10.0

8.6

-10.5

8.4

-11.0

8.2

9.4 6.5 9.2 6.0 5.5

-3.5

5.0

-4.0

9.0 8.8

-9.5 8.6 -10.0

8.4

Figure 5.2 Continued.

129

Spatial distribution of star-formation

Z= 0.67

1"

53380

Z= 0.73

1"

6933

Z= 0.74

1"

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

9.6 -10.0 7.0

-10.5

-8.5

5.0

-4.0

1.2

7.0

-2.0

1.0

6.5

-2.5

6.0

-3.0

1"

1"

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

0.4 0.0

0.2

797

-9.5 -4.0

8.2

-2.5 -3.0 5.0

8.6 8.4

-3.5

8.2

-4.0

8.0

-1.5

-8.5

9.0

7.0 8.8

-2.0 6.5 6.0

-2.5

-9.0

-3.0 -9.5 -3.5

8.6 8.4 8.2

5.5

8.0

-1.0

7.5

-1.5

7.0

-2.0

6.5

-2.5

6.0

-3.0

5.5

-3.5

9.0

-1.5

1.2

0.4 0.2 0.0

Figure 5.2 Continued.

-9.5

5.5

9.0 8.8

-9.0

6.0

7.0

-2.0

6.5

-2.5

6.0

-3.0

5.5

-3.5

5.0

-4.0

4.5

-4.5

0.6

4394

-8.5

-2.0

0.8

1"

-1.5

6.5

1.0

4445

8.4

9.2 7.0

0.0

1"

8.8 8.6

-3.5 5.5

-8.5 0.6 0.4

130

9.0 -9.0

0.2

Z= 0.67

8.2 -9.5

1.2

0.6

1"

8.6 8.4

0.8

5417

9.0 8.8

-9.0

0.0

1.0

Z= 0.46

-3.0

5.0

3372

Z= 1.31

9.2

-2.5 6.0

0.2

Z= 1.10

9.2

-4.5

-3.5

0.4

Z= 1.00

-11.0

5.5

0.6

9.4

-4.0 6.5

0.8

2525

-3.5

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

6.5

-2.5 -3.0

6.0

-3.5

5.5

-4.0

-9.0

8.8 8.6

-9.5

-8.5

8.4

9.4 9.2

-9.0

9.0 8.8

-9.5

8.6 8.4

-10.0

-8.5 -9.0

9.0 8.8 8.6

-9.5 8.4

-4.5 5.0 -5.0

-10.0

8.2

Low-z sample

Z= 0.76

1"

1.2 1.0

-8.5 6.5

0.6

-2.5 -9.0

0.8

-3.0 6.0 -3.5

-9.5

-4.0

-10.0

0.4 0.2

8275

Z= 3.06

1"

9.5

1.0

9.0

0.8

8.5

0.6

8.0

0.4

7.5

0.2

7.0

1"

1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

9.4

-10.0

9.2

-2.0

-10.5

-2.5

-11.0

-3.0

-11.5

9.0 8.8

6.0

-9.0 -4.0

9.2 -9.5

5.5

9.0 -4.5 -10.0

5.0 4.5

7.0 6.5 6.0 5.5 5.0 4.5

8.8 -5.0

-2.5 -3.0 -3.5 -4.0 -4.5 -5.0 -5.5 -6.0

-10.5

8.6

-9.0

9.2

-9.5

9.0

-10.0

8.8

-10.5

8.6

-11.0

8.4

-8.5

6.5

1"

-9.5 -1.5

6.5

1"

3822

Z= 1.30

1.2

0.0

8015

Z= 0.44

-4.5

0.0

865

Z= 0.28

5.5

9.4 9.2 9.0 8.8 8.6 8.4 8.2

-2.5 8.8 0.2

6.0 -3.0

-3.5 0.0

8.6 8.4

5.5

7705

-9.0

8.2

Figure 5.2 Continued.

population properties derived from the integrated photometry and the resolved photometry for S/N of 10 are shown in Figure 5.3. As we can see from this comparison, there is not a significant bias relative to the uncertainty estimates on the quantities. In the next section, we use the global properties inferred from the integrated stellar population modeling to assemble our low-z sample. The global physical properties of the UDF sample are summarized in Table 5.1.

5.3

Low-z sample

To find a suitable comparison sample at low-z and compare their star formation properties to that of high-z galaxies in a systematic way, we follow the same approach that was taken by Shirazi et al. (2013). Using this procedure we can compare galaxies that have similar star formation activities at high- and lowz. For each UDF galaxy, we select star-forming galaxies from the SDSS DR7 (Abazajian et al., 2009) that have log M∗ and log sSFR within 0.3 dex of that of 131

Spatial distribution of star-formation

Figure 5.3 A comparison between the global stellar population properties derived from the integrated photometry and the ones derived by integrating over the resolved photometry for S/N of 10 are shown for the UDF sample. Top-left panel shows sSFR as a function of stellar mass derived by integrating over resolved values. Top-right panel compares SFR derived using the integrated photometry and SFR derived by integrating over the resolved photometry for S/N of 10. Bottomleft and bottom-right show the same for sSFR and stellar mass, respectively. In all panels shaded regions show 1-σ scatter.

UDF object. Our parent sample is based on the MPA-JHU2 , however we use the imaging from the SDSS DR9. We select star-forming galaxies based on the BPT diagram (Baldwin, Phillips & Terlevich, 1981) and the classification presented by Kauffmann et al. (2003). In order to have the same resolution at high-z and low-z, we select SDSS galaxies that have z < 0.02. The spatial resolution of ACS is 0.1” which covers ∼ 0.61 kpc at z = 0.5. The typical SDSS PSF size is 1.5” which covers the same ∼ 0.61 kpc scale at z = 0.02. In total, we found about 4000 objects as our parent sample but in order to focus on objects that have the highest possible resolution, whenever we have more than 40 low-redshift counterparts we use the 40 lowest redshift ones only, which results in 577 SDSS galaxies.

2 http://www.mpa-garching.mpg.de/SDSS/DR7

132

Low-z sample

Figure 5.4 A comparison between the SDSS sample of 577 galaxies (gray squares) and the UDF sample (colored symbols) in the mass-sSFR diagram is shown. The scale shows the redshift of the UDF sample. We excluded two UDF galaxies that have z > 1.5 from the UDF sample. Purple squares show the UDF objects for which we assembled the SDSS sample. We can see that the UDF objects that have very high sSFRs and high masses (colored symbols without purple squares around them) are offset from low-z analog galaxies. Thus, we can not find any low-z sample for them using our sample selection criteria.

Figure 5.6 The SDSS parent sample (577 low-z galaxies) is shown in the BPT diagram. The color coding shows the sSFR. Note that the sSFR decreases with increasing [N ii] /Hα (metallicity), and for a given metallicity, it increases with increasing [O iii] /Hβ. 133

Spatial distribution of star-formation

Figure 5.5 The median values of the mass and sSFR of the SDSS sample (gray square) selected for each UDF galaxy are compared with that of the UDF galaxies that have local comparison samples (colored symbols). We run SExtractor (Bertin & Arnouts , 1996) on the SDSS r-band images to produce segmentation maps. The default setting for SExtractor was to use a DETECT THRESH = 0.61, DETECT MINAREA = 40, ANALYSIS THRESH = 0.61, DEBLEND MINCONT = 0.03 and DEBLEND NTHRESH = 32. For individual objects this was sometimes adjusted e.g. when a bright nearby star affects the background level. To create the mosaic images we made use of the routines3 developed by Zibetti et al. (2009) to download and align the SDSS images. We compare our low-z and high-z sample of galaxies in the stellar mass - sSFR plane in Figure 5.4. As we can see, the UDF objects that have very high sSFR and mass are offset from the low-z analog sample. Therefore, we cannot find low-z analogs for all 47 UDF objects. Purple squares show the UDF objects for which we could assemble a SDSS comparison sample. In Figure 5.5, we show the median values of the mass and sSFR of the SDSS sample (gray square) selected for each UDF galaxy in comparison with that of the UDF galaxies that have local comparison samples (colored symbols). We show the emission line properties of our SDSS sample based on the BPT diagram in Figure 5.6. This diagram is used extensively to classify galaxies in terms of their main sources of ionization/excitations. Note that our star-forming galaxies are all distributed below the classification lines shown by dashed line derived by Kauffmann et al. (2003). The x-axis shows [N ii] /Hα which is a metallicity tracer and the y-axis show [O iii] /Hβ which is sensitive to ionization properties of the main ionizing sources in emission line galaxies, the color coding in this diagram shows the sSFR. From this figure we can see that the sSFR decreases with increas3 Downloaded from http://www.arcetri.astro.it/~zibetti/Software/SDSSmosaic.html and updated to use SDSS DR9.

134

Low-z sample

Figure 5.7 The procedure of Voronoi binning galaxy images from the SDSS illustrated for six different objects. For each object the u − g, g − r, r − i (binned) and g-band images are shown. The spectroscopic ID and the redshift are also indicated. ing [N ii] /Hα (metallicity), and for a given metallicity, it increases with increasing [O iii] /Hβ in our low-z sample.

5.3.1

Resolved stellar population modeling of low-z sample

We use the same technique used for the UDF data to bin the SDSS color images. This has been done for target S/N of 5, 10, 20 in the u − g color image and S/N of 5 in the g − r and r − i color images. An example of the Voronoi maps is shown in Figure 5.7 for six SDSS galaxies. Each group of panels belongs to a single object and the four panels show the unbinned g-band image, the Voronoi binned u − g, g−r and r −i images (as indicated on the maps). The target S/N for this particular illustration is S/N of 10 in the u − g color.

5.3.2

Fitting models to the color map of the SDSS data

We follow the same approach which was used for the UDF data and fit models to the u, g, r, i and z flux of effective pixels of the SDSS data, derived from the Voronoi binning procedure. An example of the fitting procedure for one SDSS galaxy (PlateID-MJD-FiberID: 2746-54232-104) is shown in Figure 5.8. The r band image is shown in the left-most column where the redshift of the SDSS galaxy is also indicated. Thereafter follows the Voronoi binned u − g color map, 135

Spatial distribution of star-formation the stellar mass map, the SFR map, the sSFR map and the r-band luminosityweighted age map. Each row, from the top to the bottom, shows the result of SED fitting for target S/N of 5, 10, 20 in the u−g color and S/N of 5 in the g−r and r −i color images, respectively. By comparing the SED fitting results that we get using different Voronoi binning, we conclude that generally using u-g binning identifies structures very well, while a binning using redder colors washes out the structures completely. The reason might be because of the lower S/N in the u-band than g or r, so a cut at S/N of 5 for g − r results in a very low S/N in u − g and hence the SED fits will be poor. We can also see (by comparing the first three rows) that a S/N of 5 in u-g is sufficiently good to derive galaxy properties. In the following, we use the results of the SED fitting on maps that were Voronoi binned requiring a target S/N of 5 for the u-g SDSS images. The clumpy structure of the galaxy can be seen in the SFR and the sSFR maps but we note that the mass maps have a much more smoothed distribution. This is consistent with Wuyts et al. (2012) who showed the stellar mass maps derived for high-z galaxies are more smoothed than their color maps.

5.3.3

Integrated properties

In order to test the result of our resolved stellar population modeling, we integrate over the inferred resolved physical properties and compare them with the results that are derived from the integrated photometry. Figure 5.9 shows a comparison between the global stellar population properties derived from the integrated photometry, and those derived by integrating over the resolved photometry for S/N of 5 for the SDSS sample. Top-left panel shows sSFR as a function of stellar mass derived by integrating over resolved values. Top-right panel compares SFR derived using the integrated photometry and SFR derived by integrating over the resolved photometry. Bottom-left and bottom-right show the same for sSFR and stellar mass, respectively. From these comparisons we can see that the integrated and the summed resolved properties for the quantities used to match to the high-z sample, the stellar masses and sSFRs, are in good agreement (within 0.1 dex) for galaxies with M? > 108.5 M .

5.4

Structural parameters

Galaxy morphologies can be parameterized using a variety of indicators such as the concentration index (Abraham et al., 1996; Conselice et al., 2003), the Gini coefficient (0 < Gini < 1, a more concentrated morphology corresponds to a higher Gini, see, e.g., Abraham et al. 2003; Lotz et al. 2004; F¨orster Schreiber et al. 2011a) and the M20 index (M20< 0, a more clumpy morphology corresponds to a higher M20, see Lotz et al. 2004). We use galSVM code4 (Huertas-Company et al., 2008, 2009, 2011) to measure Gini and M20 parameters on the mass, SFR and sSFR maps inferred for the UDF and SDSS objects and show the results in Figure 5.10 and Figure 5.11. 4 http:www.lesia.obspm.fr/~huertas/galsvm.html

136

Structural parameters

r-band Z=0.008

u-g bins 10 "

Log Mass 0.8 0.6

Log SFR 6.0 5.5

0.4 0.2 u-g voronoi sn5 0.0

Z=0.008

10 "

0.8

5.0 4.5

6.0

0.6 5.5 0.4 0.2

5.0

u-g voronoi sn10 0.0

Z=0.008

10 "

0.8

6.0

0.6 0.4

5.5

0.2

5.0

u-g voronoi sn20 0.0

Z=0.008

10 "

0.8 0.6 0.4 0.2

g-r voronoi sn5

Log sSFR -3.5 -4.0 -4.5 -5.0 -5.5 -6.0 -6.5 -3.5 -4.0 -4.5 -5.0 -5.5 -6.0 -6.5 -3.5 -4.0 -4.5 -5.0 -5.5 -6.0 -6.5

6.0 5.5 5.0 4.5 4.0 3.5 3.0

-3.5 -4.0 -4.5 -5.0 -5.5 -6.0 -6.5 -7.0

6.0 5.5 5.0 4.5 4.0 3.5 3.0

-3.5 -4.0 -4.5 -5.0 -5.5 -6.0 -6.5 -7.0

Log age -9.5

9.4

-10.0

9.2

-10.5

9.0

-11.0

8.8

-9.5

9.4

-10.0

9.2

-10.5 -11.0

10 "

0.8 0.6 0.4 0.2

r-i voronoi sn5 0.0

8.8 9.6

-9.5

9.4

-10.0 -10.5 -11.0

9.2 9.0

-11.5

8.8

-9.0

9.4

-9.5 -10.0

9.2 9.0 8.8 8.6

0.0

Z=0.008

9.0

-9.0 -9.5

9.4 9.2 9.0

-10.0

8.8 8.6

Figure 5.8 An example of the fitting procedure for one SDSS galaxy (PlateIDMJD-FiberID: 2746-54232-104). The Voronoi binning has been done for target S/N of 5, 10, 20 in the u − g color image and S/N of 5 in the g − r and r − i color images. The result of SED fitting for each one has been shown in each row from the top to the bottom. The left column shows the r-band image of the object with the redshift indicated. The next columns show the Voronoi binned u − g color map, the stellar mass maps (log M∗ /M ), the SFR maps (log SFR/M yr−1 ), the sSFR maps (log sSFR/yr−1 ) and r-band luminosity-weighted age maps (log age/yr), respectively. By comparing the SED fitting results that we get using different Voronoi binning (shown in each row), we see that binning in u − g is able to identify structures very well, while a binning using redder colors washes out the structures completely. The reason might be because of the lower S/N in the u-band than g or r, so a cut at S/N of 5 for g − r results in a very low S/N in u − g and hence the SED fits will be poor. Using this we conclude that generally using u-g binning provides us with better results. We can also see (by comparing the first three rows) that S/N of 5 in u-g is sufficiently good to derive physical properties. We can see the clumpy structure of the galaxy in the SFR and the sSFR maps. Note that those clumps have younger ages in the age maps. The stellar mass maps show more smoothed structure.

137

Spatial distribution of star-formation

Figure 5.9 A comparison of global stellar population properties derived from the integrated photometry and the resolved photometry inferred for S/N of 10 are shown for the SDSS sample. Top-left panel shows sSFR as a function of stellar mass derived by integrating over resolved values. Top-right panel compares SFR derived using the integrated photometry and SFR derived by integrating over the resolved photometry. Bottom-left and bottom-right show the same for sSFR and stellar mass, respectively. From these comparisons we can see that the derived stellar masses and sSFRs are in good agreement (within 0.1 dex) for galaxies with M? > 108.5 M .

We adopt the stellar mass-weighted centers for all maps on which the structural measurements are measured. The color scales in Figure 5.10 shows the inferred Gini parameters. In all panels we show the sSFR as a function of stellar mass. The left panels show the Gini parameters derived from the mass maps, the middle ones show the Gini parameters derived from the SFR maps and the right panels show the Gini parameters inferred from the sSFR maps. Top panels show the results for the UDF galaxies and the bottom panel show the median values of the Gini parameters derived for the SDSS samples of UDF objects. We can see that using the mass maps we get a higher Gini values than SFR map and sSFR maps. This shows that the mass maps are more concentrated (higher Gini). The SDSS counterparts show less concentrated morphologies in all maps but the SFR map. While there are also regular trends seen in the SDSS plots we caution that the sample here is constructed for differential analysis against the high-z data and is not complete in any statistical 138

Structural parameters sense for making statements about trends in the low-z Universe. The color scales in Figure 5.11 shows the inferred M20 parameter for the UDF galaxies on the top panels and the SDSS galaxies on the bottom panels. In all panels we show the sSFR as a function of mass. The left panels show the M20 parameters derived from the mass maps, the middle one show the M20 parameters derived from the SFR maps and the right panels show the M20 parameters inferred from the sSFR maps. We can see that using the mass maps we get a higher M20 values than SFR map and sSFR maps. This shows that the mass maps are less clumpy or smoother than the other maps. We can also note that SDSS galaxies show less clumpy morphology (lower M20) than UDF objects. To show the comparison between low-z and high-z data more quantitatively, in Figure 5.12 we compare the differences in the Gini parameters that we infer for the UDF objects and that of their SDSS counterparts. The left panels show the ∆Gini (∆Gini = GiniUDF − GiniSDSS ) parameters derived from the mass maps, the middle ones show the ∆Gini parameters derived from the SFR maps and the right panels show the ∆Gini parameters inferred from the sSFR maps. In all top panels on the x-axis we show the integrated sSFRs measured for the UDF galaxies and in all bottom panels on the x-axis we show the integrated stellar masses measured for the UDF galaxies. We plot the median values of the ∆Gini by solid circles and show 1-σ scatter by error bars, the results for the UDF galaxies at z > 0.5 and z < 0.5 are shown by blue and gray colors, respectively. The most striking result in the figure is the positive ∆Ginimass for all sSFRs and masses and on average a negative ∆GiniSFR . However, the figure shows at most a very weak correlation between the ∆Gini parameters and the Log sSFR. A similar picture is presented by the stellar mass but there is a tentative correlation between ∆GinisSFR and stellar mass. This shows that galaxies that are more massive at high-z are more concentrated than their local analogs compared to the one that are less massive at high- and low-z. While the UDF galaxies are more concentrated in their stellar content but their star formation is more extended compared to galaxies with the same global properties in the local Universe. In Figure 5.13 we compare the differences in the M20 parameters that we infer for the UDF objects and that of their SDSS counterparts. The left panels show the ∆M20 (∆M20 = M20UDF − M20SDSS ) parameters derived from the mass maps, the middle ones show the ∆M20 parameters derived from the SFR maps and the right panels show the ∆M20 parameters inferred from the sSFR maps. In all top panels on the x-axis we show the integrated sSFRs measured for the UDF galaxies and in all bottom panels on the x-axis we show the integrated stellar masses measured for the UDF galaxies. We show the median values of the ∆M20 derived for the UDF galaxies at z > 0.5 by solid blue circles and the ones at z < 0.5 by solid gray circles. 1-σ scatter in the ∆M20 is shown by error bars. The figure shows a correlation between the ∆M20mass and the Log sSFR and also stellar mass. This shows more massive galaxies are more clumpier than their local analogs compared to less massive galaxies at high-z and low-z. On average high-z galaxies are more clumpier using ∆M20sSFR maps. We show the correlation between ∆M20sSFR and ∆Ginimass in Figure 5.14, where blue circles show the results for the UDF galaxies at z > 0.5 and gray circles show 139

Spatial distribution of star-formation them for the UDF galaxies at z < 0.5. In general, more concentrated galaxies at high-z are more clumpier. The average ∆M20sSFR for the UDF galaxies with z < 0.5 is 0.06 and for the UDF galaxies with z > 0.5 is 0.12. However, the average ∆Ginimass changes from 0.36 to 0.38 from z < 0.5 to z > 0.5.

5.5

Discussion

Figure 5.13 showed us that galaxies with the same global properties (e.g., stellar mass and sSFR) at high- and low-z show slightly different distribution in their sSFR maps with an evolution in clumpiness at high-z. We also tested this results by measuring the structural parameters on the color maps that are correlated with star formation activity, e.g., B − V for the UDF and u − g for the SDSS. The results agree well with the parameters that we inferred using sSFR maps at high- and low-z. Our high-z galaxies show a median of ∼ 0.3 dex higher Ginimass than their local counterparts. There is also a correlation between Ginimass and stellar mass showing that galaxies that are more massive at high-z are more concentrated than their local analogs compared to the ones that are less massive. This has been shown also for passive galaxies at z & 1.5 which show more concentration compared to local ellipticals with the same stellar mass (e.g., van Dokkum et al., 2008) There is a possibility of getting more concentration if galaxies at high-z have AGN. We know that there is an evolution in X-ray luminosity function of AGN up to redshift ∼ 1.2 (e.g., Aird et al., 2010). However, as we do not use emission lines, we do not expect our SED fitting results are affected by presence of AGNs. As we mentioned above, high-z galaxies that we study here are slightly more clumpy in their star formation distributions than their local analogs. Based on our selection criteria, the energy injection from the massive star populations is expected to be approximately the same in our high- and low-z galaxies. Therefore, in order to have more clumpy morphology at high-z, they need to have more surface density of the disk compared to their local analogs. These higher surface densities at high-z that we would postulate being responsible for the slightly more unstable disks can be caused by cold gas accretion. Then clumps can be formed from gravitational instability within these gas-rich disks at high-z (see e.g., Genzel et al., 2011). However, as cold gas accretion seems unlikely at low-z, this can not be a physical explanation for clumpy morphology in the local Universe. Galaxies that we compare at high- and low-z have the same global properties, including the same star formation activity, but the high-z galaxies show more extended star formation distributions in SFR maps compared to their low-z analogs. It is beyond the scope of this study to discuss what mechanisms can cause very extended star formation at high-z and not at low-z in galaxies with the same global properties. We note that also the low-z data here is constructed for differential analysis against the high-z data and is not complete in any statistical sense for making statements about trends in the low-z Universe. 140

Discussion

Figure 5.10 Top: sSFR-mass map for the UDF objects. Bottom: the same for the median of the SDSS counterparts. Color scales show the Gini parameter measured on the mass map, SFR map and sSFR map from left to right, respectively. We can see that using the mass maps we get a higher Gini values than SFR map and sSFR maps. This shows that the mass maps are more concentrated (higher Gini). The SDSS counterparts show less concentrated morphologies in all maps but SFR map.

141

Spatial distribution of star-formation

Figure 5.11 Top: sSFR-mass map for the UDF objects. Bottom: the same for the median of the SDSS counterparts of UDF galaxies. Color scales show the M20 parameter measured on the mass map, SFR map and sSFR map from left to right, respectively. We can see that using the mass maps we get a higher M20 values than SFR map and sSFR maps. This shows that the mass maps are less clumpier or more smoother than the other maps. We see also SDSS galaxies show less clumpy morphology (lower M20) than UDF objects.

142

Discussion

Figure 5.12 We compare the differences in the Gini parameters that we infer for the UDF objects and their SDSS counterparts as a function of the UDF sSFRs on the top and as a function of the UDF stellar masses on the bottom panels. We plot the median values of the ∆Gini by solid circles and show 1-σ scatter by error bars. Blue color marks the results for the UDF galaxies at z > 0.5 and gray color shows ∆Gini for the UDF galaxies at z < 0.5. Note the positive ∆Ginimass for all sSFRs and masses and a negative ∆GiniSFR . We see also a very weak correlation between the ∆Gini parameters and the Log sSFR. There is a tentative correlation between ∆GinisSFR and stellar mass.

143

Spatial distribution of star-formation

Figure 5.13 We compare the differences in the M20 parameters that we infer for the UDF objects and their SDSS counterparts as a function of the UDF sSFRs on the top and as a function of the UDF stellar masses on the bottom panels. We show the median values of ∆M20 by solid circles and show 1-σ scatter in ∆M20 by error bars. Blue color marks the results for the UDF galaxies at z > 0.5 and gray color shows ∆M20 for the UDF galaxies at z < 0.5. We see a correlation between these ∆M20 and sSFR and stellar mass that shows more massive galaxies (low sSFR) are more clumpier than their local analogs compared to less massive galaxies at high-z and low-z. Note that the UDF galaxies and SDSS counterparts agree better in the M20 parameter inferred from the sSFR maps compared to the M20 inferred from mass or SFR maps. On average high-z galaxies are more clumpier in all maps.

144

Conclusion

Figure 5.14 This plot shows ∆M20sSFR as a function of ∆Ginimass . Blue circles show the results for the UDF galaxies at z > 0.5 and gray circles show them for the UDF galaxies at z < 0.5. Error bars show 1-σ scatter. We see that there is correlation between ∆M20sSFR and ∆Ginimass showing more concentrated galaxies at high-z are more clumpier. The average ∆M20sSFR for the UDF galaxies at z < 0.5 is 0.06 and for the UDF galaxies at z > 0.5 is 0.12. However, the average ∆Ginimass changes from 0.36 to 0.38 from z < 0.5 to z > 0.5.

5.6

Conclusion

We investigated the differences between the spatial distributions of star formation at high-redshift and low-redshift Universe for galaxies with similar global properties. We studied the resolved stellar populations of these galaxies and compared the spatial distributions of star formation and mass by measuring the structural parameters for high-z galaxies and their low-z counterparts. Galaxies at high-z that we study here have more concentrated stellar content but their star formation is more extended compared to galaxies with the same global properties at z∼0. In general, these galaxies are also more clumpier in their star formation distributions than their local analogs. Therefore, in order to have more clumpy morphology, high-z galaxies need to have more surface density of the disk compared to their local analogs.

145

Spatial distribution of star-formation

ID

z spec

Log M∗ [M ]

Log SFR [M yr−1 ]

Log sSFR [yr−1 ]

Log age [yr]

03088 57290 01375 08810 04142 06206 08261 08749 04816 04396 01829 01266 07995 06188 07725 08461 05670 01971 05620 01000 05606 50001 05190 07847 03492 04267 03268 08585 00900 04929 02107 02607 00968 00662 00355 53380 06933 02525 03372 05417 00797 04445 04394 08275 09397 00865 08015 03822 07705

0.13 0.22 0.66 0.74 0.74 0.95 1.09 1.10 1.22 1.22 1.24 1.24 1.29 1.31 1.32 1.43 0.13 0.15 0.21 0.21 0.23 0.24 0.32 0.33 0.34 0.35 0.35 0.38 0.42 0.44 0.53 0.67 0.67 0.67 0.67 0.67 0.73 0.74 1.00 1.10 1.31 0.46 0.67 0.76 1.22 3.06 0.28 0.44 1.30

8.00.0 −0.0 8.50.1 −0.1 9.10.1 −0.1 10.60.0 −0.3 10.30.0 −0.0 0.2 10.8−0.1 0.2 10.5−0.3 10.30.0 −0.0 9.80.5 −0.3 10.30.2 −0.1 9.10.3 −0.2 11.00.0 −0.1 10.70.3 −0.4 0.3 10.7−0.3 0.3 10.5−0.4 10.60.4 −0.1 8.90.0 −0.1 0.1 9.0−0.0 7.50.1 −0.1 7.60.1 −0.1 9.30.0 −0.1 10.00.1 −0.1 8.80.2 −0.0 0.1 10.0−0.1 9.80.0 −0.0 7.80.1 −0.1 9.00.0 −0.0 10.20.4 −0.0 10.40.0 −0.0 10.50.0 −0.0 10.20.1 −0.1 10.70.0 −0.0 0.0 10.2−0.2 0.3 9.5−0.0 9.40.1 −0.0 10.30.1 −0.1 9.40.4 −0.2 0.5 9.8−0.2 10.60.0 −0.0 10.30.0 −0.0 10.80.2 −0.2 10.10.0 −0.0 0.0 10.4−0.2 0.0 10.4−0.3 10.11.0 −0.2 11.20.4 −0.4 0.1 9.4−0.0 0.0 10.7−0.0 8.90.3 −0.3

−2.00.1 −0.1 −0.80.3 −0.3 0.2 0.5−0.2 0.2 1.3−0.0 1.70.0 −0.0 −0.10.6 −0.7 0.3 1.7−2.2 0.1 2.2−0.0 1.40.1 −0.5 −0.10.9 −0.2 0.20.2 −0.1 0.70.3 −0.0 1.20.1 −0.2 −0.10.4 −0.9 0.1 1.0−0.3 0.2 1.2−0.7 −1.40.1 −0.0 −0.50.2 −0.1 −0.70.1 −0.2 −0.70.2 −0.2 −0.60.1 −0.1 −0.80.8 −1.5 0.0 0.3−1.1 0.4 −0.2−0.7 1.30.0 −0.0 −1.00.2 −0.3 0.1 0.6−0.0 0.8 −0.1−0.0 1.10.1 −0.2 1.60.0 −0.0 −0.50.5 −0.3 2.00.0 −0.0 0.0 1.7−0.2 1.20.2 −0.1 0.60.3 −0.3 −0.00.7 −0.5 0.2 1.0−0.1 0.1 1.4−0.2 2.10.0 −0.0 2.00.1 −0.2 1.50.3 −0.0 1.10.0 −0.0 0.0 1.8−0.0 0.2 1.6−0.0 1.80.1 −0.0 1.60.3 −1.0 −0.60.0 −0.1 0.0 1.7−0.0 −0.10.2 −0.1

−10.10.1 −0.2 −9.40.3 −0.3 0.2 −8.7−0.2 −9.40.4 −0.0 −8.60.0 −0.0 −10.90.5 −0.9 −8.90.3 −2.0 0.1 −8.1−0.1 −8.50.3 −0.9 −10.40.7 −0.2 −9.00.3 −0.4 −10.40.4 −0.0 −9.50.4 −0.5 −10.90.3 −0.7 −9.60.4 −0.5 0.4 −9.7−0.5 −10.30.1 −0.0 −9.50.1 −0.1 −8.30.2 −0.3 −8.40.2 −0.3 −10.00.1 −0.1 −10.90.8 −1.5 0.0 −8.5−1.3 −10.30.4 −0.6 −8.70.0 −0.0 0.2 −8.8−0.3 −8.50.0 −0.1 −10.30.4 −0.0 −9.40.2 −0.2 −8.90.0 −0.0 −10.70.4 −0.4 −8.80.0 −0.0 0.1 −8.5−0.0 −8.40.1 −0.1 −8.90.4 −0.3 −10.30.6 −0.4 −8.40.3 −0.4 0.3 −8.5−0.5 −8.60.0 −0.0 −8.30.1 −0.2 −9.40.4 −0.1 −9.10.0 −0.0 −8.60.3 −0.1 −8.90.5 −0.1 −8.40.2 −1.0 −10.00.8 −0.9 −10.10.1 −0.1 0.0 −9.0−0.1 −9.10.4 −0.4

9.20.1 −0.1 8.80.2 −0.2 8.20.2 −0.2 9.40.0 −0.3 8.60.0 −0.0 9.20.3 −0.2 8.40.6 −0.3 7.70.0 −0.1 8.40.8 −0.6 9.40.1 −0.3 8.50.4 −0.3 9.70.0 −0.4 9.20.2 −0.4 9.40.2 −0.2 9.20.4 −0.5 9.30.2 −0.3 9.40.0 −0.1 9.00.1 −0.1 7.90.6 −0.3 8.00.5 −0.3 9.20.1 −0.2 9.50.2 −0.2 8.70.9 −0.0 9.30.2 −0.3 8.80.0 −0.0 8.40.2 −0.2 8.00.1 −0.1 9.70.0 −0.1 8.80.2 −0.1 8.80.0 −0.0 9.40.2 −0.2 8.20.0 −0.0 8.50.0 −0.6 8.00.6 −0.1 8.50.2 −0.4 9.30.2 −0.3 8.30.6 −0.6 8.40.7 −0.6 8.40.0 −0.0 7.90.1 −0.1 9.10.2 −0.3 9.00.0 −0.0 8.70.0 −0.4 8.90.0 −0.5 8.10.9 −0.4 9.00.4 −0.4 9.20.2 −0.1 8.50.0 −0.0 8.60.4 −0.4

Table 5.1 The integrated properties of the UDF sample derived from the SED fitting to the integrated photometry.

146

Conclusion

UDF ID

Log M∗ [M ]

Log SFR [M yr−1 ]

Log sSFR [yr−1 ]

Log age [yr]

03088 57290 01375 08810 04142 06206 08261 08749 04816 04396 01829 01266 07995 06188 07725 08461 05670 01971 05620 01000 05606 50001 05190 07847 03492 04267 03268 08585 00900 04929 02107 02607 00968 00662 00355 53380 06933 02525 03372 05417 00797 04445 04394 08275 09397 00865 08015 03822 07705

8.20.1 −0.1 8.60.1 −0.1 9.00.2 −0.2 10.30.1 −0.1 — 10.70.1 −0.1 — — 9.70.1 −0.1 10.20.1 −0.1 0.1 9.0−0.1 10.80.1 −0.1 10.50.1 −0.1 10.60.1 −0.1 10.30.1 −0.1 10.50.1 −0.1 8.90.1 −0.1 0.1 8.9−0.1 7.60.1 −0.1 7.60.1 −0.1 0.1 9.2−0.1 10.00.1 −0.1 8.70.2 −0.1 10.10.1 −0.1 9.60.1 −0.1 7.80.1 −0.1 9.00.2 −0.2 10.10.1 −0.1 10.20.1 −0.1 — 10.10.1 −0.1 — — 9.70.1 −0.1 9.30.3 −0.1 10.20.1 −0.1 0.1 9.1−0.1 0.3 9.5−0.2 — — — 9.90.1 −0.1 — — — — 9.30.1 −0.1 — 8.90.1 −0.1

−1.60.3 −0.3 −0.90.3 −0.2 0.00.4 −0.1 0.2 0.8−0.1 — −0.10.2 −0.3 — — 0.70.1 −0.3 −0.20.3 −0.2 −0.20.2 −0.2 0.1 0.4−0.2 0.70.1 −0.1 −0.20.3 −0.3 0.60.2 −0.1 0.60.1 −0.1 −1.20.3 −0.3 −0.50.2 −0.2 0.2 −0.7−0.1 0.2 −0.8−0.2 −0.60.2 −0.2 −0.70.3 −0.3 0.3 −0.3−0.2 −0.30.3 −0.3 0.70.1 −0.2 −1.00.3 −0.2 0.30.3 −0.4 −0.10.3 −0.2 0.80.1 −0.1 — −0.40.3 −0.3 — — 0.70.1 −0.3 0.00.3 −0.2 −0.20.3 −0.2 0.1 0.7−0.1 0.1 −0.2−0.0 — — — −1.20.1 −0.1 — — — — −0.60.3 −0.2 — −0.40.3 −0.1

−9.90.3 −0.3 −9.50.2 −0.2 −8.60.3 −0.4 −9.60.2 −0.1 — −10.80.3 −0.3 — — −8.60.2 −0.3 −10.40.3 −0.2 −9.20.3 −0.2 −10.40.3 −0.2 −9.80.1 −0.1 −10.80.3 −0.3 −9.80.2 −0.2 −9.80.1 −0.2 −10.30.3 −0.4 −9.50.3 −0.2 0.2 −8.3−0.1 −8.40.2 −0.2 −9.90.3 −0.3 −10.80.4 −0.3 −8.70.4 −0.3 −10.40.3 −0.3 −8.90.2 −0.2 −8.80.3 −0.2 −8.60.3 −0.3 −10.30.3 −0.3 −9.60.2 −0.1 — −10.60.3 −0.4 — — −8.60.2 −0.3 −9.00.3 −0.4 −10.30.3 −0.2 −8.50.1 −0.1 0.3 −8.8−0.3 — — — −9.20.5 −0.4 — — — — −10.00.3 −0.3 — −9.20.3 −0.3

9.00.2 −0.2 8.90.2 −0.2 8.90.2 −0.4 9.50.1 −0.2 — 9.60.2 −0.2 — — 9.30.2 −0.2 9.40.2 −0.2 8.80.3 −0.3 9.50.1 −0.2 9.60.1 −0.2 9.50.2 −0.2 9.40.2 −0.2 9.50.2 −0.2 9.30.2 −0.2 8.90.2 −0.2 7.90.3 −0.2 8.00.3 −0.2 9.20.2 −0.2 9.50.2 −0.2 8.30.3 −0.3 9.40.2 −0.2 8.80.2 −0.2 8.30.3 −0.3 8.90.2 −0.4 9.40.2 −0.2 9.40.2 −0.1 — 9.50.2 −0.2 — — 9.30.2 −0.2 8.80.3 −0.3 9.40.2 −0.2 8.90.2 −0.2 8.70.4 −0.4 — — — 9.50.0 −0.3 — — — — 9.20.2 −0.2 — 8.70.2 −0.2

Table 5.2 The integrated properties of the analog sample of the UDF galaxies derived from the SED fitting to the integrated photometry.

147

REFERENCES

References Abazajian, K. N., Adelman-McCarthy, J. K., Ag¨ ueros, M. A., et al. 2009, ApJS, 182, 543 Abraham, R. G., van den Bergh, S.,& Nair, P. 2003, ApJ, 588, 218 Abraham, R. G., Ellis, R. S., Fabian, A. C., Tanvir, N. R., & Glazebrook, K. 1999, MNRAS, 303, 641 Abraham, R. G., Tanvir, N. R., Santiago, B. X., Ellis, R. S., Glazebrook, K.,& van den Bergh, S. 1996, MNRAS, 279, 47 Aird, J., Nandra, K., Laird, E. S., et al. 2010, MNRAS, 401, 2531 Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 5 Bertin, E.,& Arnouts, S. 1996, A&AS, 117, 393 Bournaud, F., Daddi, E., Elmegreen, B. G., et al. 2008, A&A, 486, 741 Bouch´e, N., Dekel, A., Genzel, R., et al. 2010, ApJ, 718, 1001 Brinchmann, J., Charlot, S., et al. 2004, MNRAS, 351, 1151 Bruzual, G,& Charlot, S. 2003, MNRAS, 344, 1000 Bundy, K., Fukugita, M., Ellis, R. S., et al. 2009, ApJ, 697, 1369 Cappellari, M.,& Copin, Y. 2003, MNRAS, 342, 345 Coe, D., Ben´ıtez, N., S´ anchez, S. F., et al. 2006, AJ, 132, 926 Conselice, C. J., Yang, C.,& Bluck, A. F. L. 2009, MNRAS, 394, 1956 Conselice, C. J., Bershady, M. A., Dickinson, M., & Papovich, C. 2003, AJ, 126, 1183 Contini T., et al., 2012, A&A, 539, 91 Cowie, L. L., Hu, E. M., & Songaila, A. 1995, AJ, 110, 1576 Daddi, E., Dickinson, M., Morrison, G., et al. 2007, ApJ, 670 156 Dekel, A., Birnboim, Y., Engel, G., et al. 2009, Nature, 457, 451 Dickinson, M., Hanley, C., Elston, R., et al. 2000, ApJ, 531, 624 Diehl, S., & Statler, T. S. 2006, MNRAS, 368, 497 Elbaz, D., Daddi, E., LeBorgne, D., et al. 2007, A&A, 468, 33 Elmegreen, B. G., Elmegreen, D. M., Fernandez, M. X.,& Lemonias, J. J. 2009, ApJ, 692, 12 Elmegreen, M. E., Elmegreen, B. G., Ravindranath, S.,& Coe, D. A. 2007, ApJ, 658, 763 Elmegreen, B. G., & Elmegreen, D. M. 2006, ApJ, 650, 644 Elmegreen, B. G.,& Elmegreen, D. M. 2005, ApJ, 627, 632 Epinat, B., Tasca, L., Amram, P., et al. 2012, A&A, 539, 92 Gallazzi, A., Brinchmann, J., et al., 2008, MNRAS, 383, 1439 Gallazzi, A., Charlot, S., et al., 2005, MNRAS, 362, 41 Griffiths, R. E., Casertano, S., Ratnatunga, K. U., et al. 1994, ApJ, 435, 19 F¨ orster Schreiber, N. M., Shapley, A. E., Erb, D. K., et al. 2011a, ApJ, 731, 65 F¨ orster Schreiber, N. M., Shapley, A. E., Genzel, R., et al. 2011b, ApJ, 739, 45 F¨ orster Schreiber, N. M., Genzel, R., Bouch´e, N., et al. 2009, ApJ, 706, 1364 F¨ orster Schreiber, N. M., Genzel, R., Lehnert, M. D., et al. 2006, ApJ, 645, 1062 F¨ orster Schreiber, N. M., Genzel, R., Lutz, D.,& Sternberg, A. 2003, ApJ, 599, 193 Genzel, R., Newman, S., Jones, T., et al. 2011, ApJ, 733, 101 148

REFERENCES Genzel, R., Tacconi, L. J., Gracia-Carpio, J., et al. 2010, MNRAS, 407, 2091 Genzel, R., Burkert, A., Bouch´e, N., et al. 2008, ApJ, 687, 59 Genzel, R., Tacconi, L. J., Eisenhauer, F., et al. 2006, Nature, 442, 786 Guo, Y., Giavalisco, M., Cassata, P., et al. 2011a, ApJ, 735, 18 Guo, Y., Giavalisco, M., Ferguson, H. C., et al. 2011b, (arXiv1110.3800) Huertas-Company, M., Aguerri, J. A. L., Bernardi, M., Mei, S., & S´anchez Almeida, J. 2011, A&A, 525, A157 Huertas-Company, M., Tasca, L., Rouan, D., et al. 2009, A&A, 497, 743 Huertas-Company, M., Rouan, D., Tasca, L., Soucail, G., & Le F`evre, O. 2008, A&A, 478, 971 Kauffmann, G., Heckman, T.M., et al. 2003, MNRAS, 346, 1055 Keres, D., Katz, N., Fardal, M., Dav´e, R.,& Weinberg, D. H. 2009, MNRAS, 395, 160 Kereˇs, D., Katz, N., Weinberg, D. H., & Dav´e, R. 2005, MNRAS, 363, 2 L´ opez-Sanjuan, C., Le F`evre, O., de Ravel, L., et al. 2009, A&A, 501, 505 Lotz, J. M., Jonsson, P., Cox, T. J., Croton, D., Primack, J. R., Somerville, R. S.,& Stewart, K. 2011a, ApJ, 742, 103 Lotz, J. M., Davis, M., Faber, S. M., et al. 2008, ApJ, 672, 177 Lotz, J. M., Primack, J.,& Madau, P. 2004, AJ, 128, 163 Newman, S. F., Shapiro Griffin, K., Genzel, R., et al. 2012, APJ, 752, 111 Newman, S. F., Genzel, R., F¨ orster-Schreiber, N. M., et al. 2012, APJ, 761, 43 Noeske, K. G., Weiner, B. J., Faber, S. M., et al. 2007, ApJ, 660, 43 Overzier, R. A., Heckman, T. M., Kauffmann, G., et al. 2008, APJ, 677, 37 Overzier, R. A., Heckman, T. M., Schiminovich, D., et al. 2010, APJ, 710, 979 Renzini, A. 2009, MNRAS, 398, 58 Robaina, A. R., Bell, E. F., Skelton, R. E., et al. 2009, ApJ, 704, 324 Rodighiero, G., Daddi, E., Baronchelli, I., et al. 2011, ApJ, 739, 40 Shapiro, K. L., Genzel, R., F¨ orster Schreiber, N. M., et al. 2008, ApJ, 682, 231 Shapley, A. E., Steidel, C. C., Erb, D. K., et al. 2005, APJ, 626, 698 Shirazi, M., Brinchmann, J., & Rahmati, A. 2013, arXiv:1307.4758 Teplitz, H. I., McLean, I. S., Becklin, E. E., et al. 2000, ApJ, 533, L65 Thompson, R. I., Storrie-Lombardi, L. J., Weymann, R. J., et al. 1999, AJ, 117, 17 van Dokkum, P. G., Franx, M., Kriek, M., et al. 2008, ApJL, 677, L5 Welikala, N., Connolly, A. J., Hopkins, A. M., & Scranton, R. 2009, APJ, 701, 994 Williams, R. J., Quadri, R. F.,& Franx, M. 2011, ApJ, 738, L25 Windhorst, R. A., Fomalont, E. B., Kellermann, K. I., et al. 1995, Nature, 375, 471 Wuyts, S., F¨ orster Schreiber, N. M., Genzel, R., et al. 2012, ApJ, 753, 114 Wuyts, S., F¨ orster Schreiber, N. M., Lutz, D., et al. 2011, ApJ, 738, 106 Zibetti, S., Charlot, S., & Rix, H.-W. 2009, MNRAS, 400, 1181

149

Nederlandse samenvatting

Sterrenstelsels in al hun verschijningen huisvesten tijdens hun leven miljarden sterren. Het is de aanwezigheid van deze schijnende sterren, die het mogelijk maakt dat wij ze kunnen waarnemen door de kosmische tijd. Hoewel we sterrenstelsels voornamelijk door middel van sterlicht observeren, kunnen we deze sterren niet individueel oplossen, tenzij ze heel dichtbij staan. Daarom wordt al het licht van de miljarden sterren bij elkaar opgeteld, geanalyseerd met gebruik van sterpopulatiemodellen om informatie over de evolutie van sterrenstelsels eruit af te kunnen leiden. Sterlicht bereikt ons niet zonder door het interstellair medium (ISM) te reizen, dat wolken van gas en stofdeeltjes bevat. Gas en stof kunnen het licht van de sterren absorberen en opnieuw uitzenden, of het richting ons verstrooien, wat het interpreteren van wat we observeren in sterrenstelsels erg gecompliceerd maakt. Ondanks al deze moeilijkheden kunnen we, door alleen het totale licht van sterrenstelsels te analyseren, de globale eigenschappen van sterrenstelsels, zoals stermassa, stervormingssnelheid en leeftijd, binnen een goede marge vaststellen, met behulp van sterpopulatiemodellen. Door de sterpopulatiemodellen en photoionisatiemodellen te combineren, kunnen we het emissielijnspectrum van stervormende sterrenstelsels dat ontstaat in het ge¨ıoniseerde gas rond jonge sterren, verder analyseren, wat ons een schat aan informatie biedt over de eigenschappen van sterrenstelsels op kleine schaal, bijvoorbeeld het ISM. Dit proefschrift is een poging om de relatie tussen deze eigenschappen op kleine schaal en de globale eigenschappen van stervormende sterrenstelsels te begrijpen door de kosmische tijd, door middel van sterpopulatiemodellen en photo-ionisatiemodellen. Stervorming in sterrenstelsels kan niet alleen door middel van sterren direct worden getraceerd, maar ook door het effect van sterren op het omringende gas te bestuderen. Hieronder volgt een vereenvoudigde samenvatting van dit proefschrift. Voor meer informatie verwijs ik graag naar het desbetreffende hoofdstuk.

Dit proefschrift De straling van sterren ioniseert het omringende gas en produceert nebulaire emissielijnen doordat het ge¨ıoniseerde gas recombineert. We kunnen de stervormings151

Nederlandse samenvatting snelheid meten, samen met andere eigenschappen van sterrenstelsels, met behulp van emissielijnen van zowel nabijgelegen als verafgelegen stelsels. Voor dit proefschrift maak ik veel gebruik van nebulaire emissielijnen om de intrinsieke eigenschappen van sterrenstelsels in het nabije en verafgelegen Universum te meten. Ik gebruik ook nebulaire emissielijnen om massieve sterren indirect te traceren en de eigenschappen van het ISM op kleine schaal te onderzoeken. Hoofdstuk 2: De evolutie van massieve sterren is een complex en nog niet goed begrepen proces. Hoewel we bij het observeren van het stellair continu¨ um gelimiteerd zijn door interstellaire absorbtie bij λ < 228 ˚ A, kunnen we de nebulaire He ii emissielijn van λ 4686 gebruiken om waardevolle informatie te krijgen over het hoge-energie gedeelte van stellaire spectrale energie verdelingen. Slechts de meest extreme stervormende sterrenstelsels vertonen nebulaire He ii emissie en het is algemeen gedacht dat Wolf-Rayet (WR) sterren de benodigde ioniserende straling hiervoor produceren. In Hoofdstuk 2 bestuderen we de fysieke eigenschappen van emissielijnstelsels in de SDSS, die He ii vertonen. Wanneer we ons baseren op deze gegevens, vinden we dat de He ii niet geassocieerd is met WR kenmerken in een groot aantal stervormende sterrenstelsels met zodanige emissie, bij lage metalliciteit. Het gebrek aan WR sterren heeft belangrijke implicaties voor de evolutie van de meest massieve sterren bij lage metalliciteit. Niet-homogene sterevolutiemodellen (bijv. Yoon & Langer, 2006) en ruimtelijke offset tussen de WR sterren en de He ii gebieden (bijv. Kehrig et al. 2008) kunnen twee mogelijke verklaringen zijn voor deze discrepantie. We tonen ook dat huidige sterpopulatiemodellen de gevonden ratio tussen He ii en Hβ in gebieden met lage metalliciteit niet kunnen produceren. Dit resultaat heeft gevolg voor het interpreteren van de waarnemingen van stelsels op hoge roodverschuiving, waar de metalliciteit typisch lager wordt verwacht. Een ander belangrijk resultaat in deze studie is het defini¨eren van een nieuw diagnostisch diagram door middel van de He ii /Hβ ratio, welke kan worden gebruikt om de AGN contributie in stervormende sterrenstelsels die He ii emissie vertonen, binnen een marge vast te stellen. Hoofdstuk 3: De gedetailleerde analyse van verafgelegen sterrenstelsels, wordt gelimiteerd door hun kleine schijnbare diameter aan de hemel en hun zwakke schijnbare magnitude. Beide beperkingen kunnen worden overkomen door stelsels te observeren die door zwaartekrachtlenzen optisch zijn vergroot. In Hoofdstuk 3 analyseren we opgeloste data van de zogenaamde ”8 o’clock arc”, een gelensd Lyman break sterrenstelsel, samen met HST beelden van dit stelsel, waarop het lensmodel van dit stelsel is gebaseerd. Met dit lensmodel kunnen we ontrafelen hoe het stelsel eruitziet zonder het lenseffect en de Hβ emissie, snelheid en snelheidsdispersie in kaart brengen. We tonen aan dat een eenvoudig roterend schijfmodel niet in staat is het snelheidsveld van het sterrenstelsel te reproduceren en dat we een complexer snelheidsveld nodig hebben. Het Hβ profiel van het stelsel toont een brede blauwverschoven vleugel, wat een uitstroom van 200 km/s suggereert. De geschatte oppervlaktedichtheid en gasmassa van de ”8 o’clock arc”toont een 152

gasinhoud dat 2.5 tot 7 maal hoger is dan in vergelijkbare stelsels in de SDSS. Hoofdstuk 4: De meeste sterren die ons tegenwoordig omringen zijn enige miljarden jaren geleden gevormd, toen er een piek was in de stervormingsactiviteit van het Universum. De omstandigheden waaronder deze sterren geboren werden is van groot belang, maar het is erg moeilijk deze te bestuderen door de beperkte observationele resolutie van vergelegen objecten. In Hoofdstuk 4 presenteren we een nieuwe benadering om direct de dichtheid in stervormende gebieden van sterrenstelsels die zich dichtbij de piek in stervormingsactiviteit van het Universum bevinden, te vergelijken met die van nabijgelegen stelsels. Om indirect het ISM op hoge roodverschuiving te traceren, gebruiken we emissielijnintensiteiten van vergelegen stelsels. We kalibreren een nieuwe relatie tussen de [O iii]λ5007/[O ii]λ3727 emissielijnratio en de ionisatieparameter om het verschil tussen de ionisatieparameters tussen de samples van hoge en lage roodverschuiving te schatten. We analyseren de ionisatieeigenschappen van een sample van stelsels met hoge roodverschuiving tussen 2.6 en 3.4, waaronder de ”8 o’clock arc¨en vergelijken die met stelsels met vergelijkbare fysieke eigenschappen in het lokale Universum. We tonen aan dat, nadat we rekening hebben gehouden met alle verschillen in eigenschappen op grote schaal, zoals massa en specifieke stervormingssnelheid, de dichtheid in stervormingsgebieden acht maal hoger was in het verleden. Dit impliceert dat de meerderheid van sterren in het Universum zijn ontstaan in gas dat aan heel verschillende schalingsrelaties voldeed, dan wat we in het hedendaagse Universum zien. Dit is een treffend resultaat dat sterke beperkingen oplegt aan de omstandigheden van stervorming in normale sterrenstelsels in het vroege Universum. Hoofdstuk 5: In Hoofdstuk 5 bestuderen we de verschillen tussen de ruimtelijke verdeling van stervorming op hoge en lage roodverschuiving, voor stelsels met vergelijkbare globale eigenschappen (bijv. stermassa en stervormingssnelheid). We gebruiken multi-band imaging gegevens die beschikbaar zijn in de HUDF en vergelijken deze kwantitatief met de gegevens van lage roodverschuiving van de SDSS. Hiermee bestuderen we de fysieke processen die klonterige stervormingsverdelingen veroorzaken in stelsels met vergelijkbare stervormingsactiviteit op zowel hoge als lage roodverschuiving. We vergelijken de opgeloste sterpopulaties van deze stelsels door de structurele parameters van stelsels op hoge en lage roodverschuiving te meten. We tonen aan dat stelsels op hoge roodverschuiving een meer geconcentreerde sterinhoud hebben, maar dat hun stervorming meer wijd verspreid is in vergelijking met stelsels met dezelfde globale eigenschappen in het lokale Universum. We laten zien dat stelsels op hoge roodverschuiving klonteriger zijn qua stervormingsdistributie, dan hun lokale equivalenten. Deze klonterige morfologie wekt de suggestie dat stelsels op hoge roodverschuiving een grotere dichtheid in de schijf hebben, dan hun lokale equivalenten.

153

Publications

1. On the spatial distribution of star formation in distant and nearby galaxies Maryam Shirazi & Jarle Brinchmann 2013, MNRAS, to be submitted. 2. Stars were born in significantly denser regions in the early Universe Maryam Shirazi, Jarle Brinchmann & Alireza Rahmati 2013, ApJ, submitted, arXiv:1307.4758. 3. The physical nature of the 8 o’clock arc based on near-IR IFU spectroscopy with SINFONI Maryam Shirazi, Simona Vegetti, Nicole Nesvadba, Sahar Allam, Jarle Brinchmann, Douglas Tucker 2013, MNRAS, submitted, arXiv:1306.6282. 4. Strongly star forming galaxies in the local Universe with nebular He II 4686 emission Maryam Shirazi & Jarle Brinchmann 2012, MNRAS, 421, 1043. 5. Accelerating universe in brane gravity with a confining potential Maliheh Heydari-Fard, Maryam Shirazi, Shahram Jalalzadeh, Hamid Reza Sepangi 2006, Physics Letters B, 640, 1.

155