The main objective of this Ph.D. thesis is to study the evolution of starburst galaxies during the last 8 Gyrs. At high redshift (z>1) galaxies show a large variety of unusual morphologies that are rare to find in the local Universe. The main mechanisms transforming the appearance of high-redshift galaxies are related to star formation processes, with giant clumps of star forming stars altering the otherwise smooth shape of galaxies. This work aims to better understand the physical processes driving the evolution with cosmic time of the star formation in the Universe. Thus, we focus on intermediate redshift galaxies to bridge the gap between high-redshift starburst galaxies and the local Universe.
This Ph.D. thesis is based on the extensive use of the largest survey ever performed with the Hubble Space Telescope (HST), the COSMOS survey. Besides the high spatial resolution and deep imaging from the Advanced Camera for Surveys (ACS), COSMOS has been observed with unprecedented detail using the state-of-the-art ground- and space-based instruments. This new wealth of multiwavelenght information set the basis of the current study and it is thoroughly explained in Chapter 2.
A strong effort has been devoted to efficiently extract the information encoded in the COSMOS multiwavelength photometric dataset. Particular attention has been paid to the sample selection and characterization of a well-defined sample of starbursts galaxies. Two different sets of diagnostic diagrams have been developed to this aim: the first is optimized for targets up to z~0.5 (sampling the Halpha and [OIII] emission lines); and the second that allow us to increase the redshift range up to z~1 (sampling [OIII] and the Balmer break). This accurate search has produced a catalogue with over 1000 starburst galaxies covering 8 Gyrs of evolution and a large range of morphologies (from compact and smooth blue galaxies to clumpy starbursts).
The resulting sample are galaxies with masses spanning 10^6 < M_sun < 10^11. The mass function of starburst galaxies in the COSMOS survey up to z~0.5 is quite similar to that of the entire galaxies sample in COSMOS at this redshift range. In fact, only small differences at the high-mass end are found, which could be attributed to massive red galaxies that does not show up when, as it is our case, are looking for starbursts.
The galaxy sample has been classified (chapter 3 and 4) according to the presence and distribution of starburst knots/clumps within the galaxy. As a result, we classify them as: sknot when it consists of a single knot and mknots when several knots are present. Objects with a single knot surrounded by diffuse emission are classified as sknot + diffuse. An important part of this work is the study of the star-forming regions themselves, characterizing their physical properties such as star forming rate (SFR), masses, etc. These, together with the analysis of the starburst properties in relation with other morphological issues like their sizes and their location and number in their host galaxy, have been discussed in the framework of the more recent galaxy formation and evolutionary scenarios. Our results support a scenario were large and massive clumps at the galaxy centers would be the end product of the coalescence of surviving smaller clumps from the outskirts. Thus, making it unlikely that mergers are the reason behind the observed starburst knots.
To further understand the evolution of starburst galaxies and their constituent knots/clumps, it is fundamental to know the properties of the host galaxies where they are found. The depth of the ACS/HST imaging, reaching surface brightness levels of 27-28 mag arcsec^-2 in the F814W band, allowed us to perform an analysis of the underlying stellar component for most of the galaxies in our low redshift sample (0 < z < 0.5). We have carried out the detailed modeling, by means of two-dimensional surface brightness fitting algorithm (GALFIT), of this extended structure. We have developed new techniques to deal with the variety of starburst morphologies: Sknot, Sknot+diffuse and Mknots galaxies, in order to properly remove the ionized gas contamination of the host galaxy. As result, 170 of the galaxies could be reliably fit through the 2D algorithm (74% of Sknot, 73% of Sknot+diffuse, and 89% of Mknot galaxies). From the fit it results that the 74% of the host have an exponential luminosity profile what corresponds to disc-like structures. The remaining 2% however are spheroidal like. This last case corresponds mostly to Sknot galaxies. This result are presented in Chapter 5, together with results from the analysis of the bursts of these galaxies led us to propose that these are an intermediate phase of the so call low surface brightness galaxies. The rest, Sknot+diffuse light and Mknots are mostly disk systems, which are undergoing star formation now and whose starbursts are more luminous and bright the more to the center of the galaxy they are. These results also support the scenario of galaxies growing through new starburst that by coalescence may migrate to the centers to forming new bulges, or feeding the existing one.
In our quest to understand the star-formation processes leading to the current population of starburst galaxies in the local Universe in Chapter 7 we also performed an analysis of the high-resolution spectra, taken with the Intermediate dispersion Spectrograph and Imaging System (ISIS) at the William Herschel Telescope (WHT) in the Observatorio del Roque de los Muchachos (ORM), of a subsample of our low-redshift starburst galaxies. Our main goal in this study was the analysis of multi-components in the Halpha emission line, to understand the hydrodynamic regimes of the star formation in these galaxies. In particular, we look for the presence of starbursts system in the negative feedback regime, i.e., systems that could be enriched by the fallback of material after the previous generation of massive stars exploded. The entire dataset and results have also been analyzed using universal scale relations like the one that relate the luminosity of the clumps or the emission line with the size. The result confirms the existence of such relations.