Introduction

High redshift galaxies and the formation of stars during the cosmic history are mostly investigated by unbiased observations in the ultraviolet, optical or near infrared. Using large telescopes and efficient photon detectors, during the last ten years a large number of high redshift galaxies candidates were detected, most of which were later confirmed by high sensitivity optical spectroscopy. As an example Fig. 1 shows the ”Hubble Deep Field“, a ~ 10 day integration optical image taken with the Hubble Space Telescope that reveals galaxies up to redshifts corresponding to the earliest epochs of star formation in the universe. Most galaxies found in such deep optical surveys show modest star formation activity, with rates implied from the rest ultraviolet luminosity of order a few solar masses per year.

Figure 1: The Hubble Deep Field. The left picture shows a region of 150'' × 150'', observed at optical wavelengths for 10 days with the Hubble Space Telescope. The right picture shows the same region seen at 850 m by the Submillimeter Common User Bolometer Array (SCUBA) at the JCMT (Hughes et al. 1998).

That the optical view on early star formation is incomplete is illustrated, e.g., by the existence today of a prominent population of massive elliptical galaxies, which show an exhausted interstellar medium (ISM), and are expected to originate from very violent starbursts in the past. Furthermore, galaxy clusters and groups contain hot, metal enriched plasma that required intense activity of massive stars sometime in the past. Optical observations are so far missing any evidence for objects where the transformation of baryons into stars and heavy elements was taking place at rates high enough to explain these observations.

Hints on the missing link between the active and quiescent universe can be found in the local universe. The IRAS mid-infrared surveys revealed that in a small fraction of local massive galaxies, in so called ultraluminous infrared galaxies (ULIRGs), star formation is taking place at unusually high rates, with over one hundred solar masses per year transformed into stars. The optical spectra of these objects do typically not reveal signatures of such intense star formation activity, which is explained by dust obscuration of the starburst regions. Although the total contribution of ULIRGs to the average rate of star formation in the present universe is small, this may not have been so in the past, and it emphasizes the necessity to extend the high redshift surveys for star formation activity from the optical to the far infrared, to wavelengths where the radiation can escape unhindered even from dust-obscured regions.

It was only during the past five years that new powerful instrumentation allowed a systematic exploration of the distant universe at longer wavelengths. Firstly, NASA’s Cosmic Background Explorer satellite (COBE) discovered a bright isotropic far-infrared (FIR) background radiation. This cosmic infrared background (CIRB) is likely of extragalactic origin and was interpreted as the integrated emission by dust in distant, primeval galaxies.

Later, the Infrared Space Observatory (ISO) allowed sensitive, high spatial resolution observations of mid-infrared (MIR) and FIR extragalactic sources. These deep ISO surveys were able to follow the evolution of the infrared emissivity of galaxies up to redshifts of about unity, showing that the integrated diffuse extragalactic background does indeed consist of individual sources at cosmological distances.

Recent deep imaging observations with the submm Common User Bolometer Array (SCUBA) on the 15-m James Clerk Maxwell telescope (JCMT) and with the Max-Planck Millimeter Bolometer Array (MAMBO) at the IRAM 30m telescope were also able to resolve a substantial fraction of the CIRB at (sub)mm wavelengths into a population of very luminous infrared galaxies at redshifts of unity and beyond. The right image in Figure 1 shows the region of the Hubble Deep Field observed by SCUBA. Surprisingly, the submillimeter emission is concentrated in a few bright sources, which appear to have no obvious correspondence to any optically visible object. This and other such deep submillimeter and millimeter wavelength images established that at least the brightest sources comprising the (sub)millimeter extragalactic background are a distinct high-redshift population that has no correspondence in the optical or near-infrared. Numerous efforts to identify the (sub)mm sources in deep optical/near-IR images and to obtain spectroscopic redshifts remained mostly without results, but did produce enough evidence to locate these sources in the distant universe. The SCUBA and MAMBO surveys suggested that the average star formation rate in the CIRB population dominates star formation at redshifts beyond one. This conjecture has however been challenged by optical observers, who now do correct for extinction in computing average star formation rates in optical, high-redshift galaxies, and thereby derive average star formation rates at least consistent with those derived from FIR to mm surveys (e.g. Adelberger & Steidel). The uncertainties in both camps remain large, and the issue which population of objects dominates the energy (and metallicity) production in the early universe is still open: The SCUBA and MAMBO surveys only resolve the bright end of the (sub)mm background population, so one needs to extrapolate the brightness distribution to invisible objects to compute the average star formation rate at a given epoch. The optical counts on the other hand rely on uncertain extinction corrections (of order five) to trace the star formation history of the universe.

The presented work attempts to model the star formation history of the universe in a way consistent with all related observations. Constrained by the source number counts that millimeter and submillimeter deep imaging surveys provide, and by the source redshift distribution estimated from radio to (sub)mm flux density ratios, I simulate a population of high redshift starburst galaxies. I model the spectral energy distribution (SED) of these objects from MIR to radio wavelengths. The thermal emission from dust heated by starbursts is thereby assumed to account for the bulk of the MIR and FIR luminosity of these galaxies. Based on the functional form of the well-established luminosity distribution of local galaxies, and on the estimated redshift distribution of (sub)mm sources I model the luminosity distribution of high redshift starburst galaxies as a function of redshift. From the bolometric luminosities of the simulated galaxies the average star formation rate per comoving volume is computed and compared with that derived from optical surveys. The spatial distribution of the (sub)mm starburst population is derived from Monte-Carlo simulations on a density field given by numerical simulations of the evolution of dark matter from the Big Bang to the present. Artificial (sub)mm maps are thereby created that can be compared with the observations in order to study the source clustering properties, and to study the effects of confusion and noise on the source number statistics.

Acknowledgments

As some people I have to say ”thank you“ to do not understand English, but all people to be addressed here speak German, I will write the acknowledgments in German. The only two exceptions are Chris Carilli, whom I thank for the helpful communication, and Pierre Cox, who provided me with the spectra of Arp 220 and M82. Thank you!

Danksagung

Ich bedanke mich bei all Jenen, die mir bei der Erstellung dieser Diplomarbeit geholfen haben. Mein besonderer Dank gilt Dr. Frank Bertoldi, der meine Arbeit sehr gut betreut hat. Er hatte auch für die dümmeren meiner Fragen stets ein offenes und geduldiges Ohr. Die Beobachtungsreisen zum IRAM-Teleskop, die er möglich gemacht hat, waren ein schönes und lehrreiches Erlebnis. Seine Unterstützung und Hinweise bei der schriftlichen Ausarbeitung dieser Arbeit und bei meinem Diplom-Kolloquium waren ausgesprochen hilfreich. Herrn Prof. Dr. Karl Menten danke ich für die Ermöglichung dieser Diplomarbeit. Ich wünsche ihm weiterhin gute Genesung von der schweren Krankheit. Herrn Prof. Dr. Johannes Schmid-Burgk und Herrn Prof. Dr. Peter Schneider danke ich für die Übernahme des Referates bzw. Koreferates. Auch Sie hatten stets ein offenes Ohr für meine Fragen und Belange. Großer Dank gebührt meinen Eltern, die mich stets in allem, besonders aber in meinem Studium unterstützt haben, nicht zuletzt auch finanziell. Mein Dank gilt auch der Diplom-Lernrunde, namentlich Mark Hillenbach, Andreas Sunkler und Maja Tiljar, mit denen zusammen die Bewältigung der Diplom-Prüfungen erheblich leichter und angenehmer war. Bedanken möchte ich mich auch bei den ,,Insassen“ des Kinderzimmers, als da wären Matthias Kadler, Jens Kauffmann, Maja Tiljar, Bernd Weferling und Maik Wolleben. Mit Ihnen zusammen ist auch das häßlichste Büro ein Raum in dem man sich gut aufgehoben fühlt. Ihre menschliche, technische und fachliche Hilfe war sehr nützlich. Schließlich möchte ich mich bei Ursula Linden für Ihre liebevolle Unterstützung bedanken.