UC Santa Cruz

The buildup of galaxies is one of the most fundamental questions in

modern cosmology. The study of this process in the first few Gyr of the

Universe, starting from the first stars, is a challenging endeavor. In

this thesis we have made extensive use of the deepest optical and

infrared images currently available from the Hubble Space

Telescope (HST) and the Spitzer Space

Telescope to study the properties of the stellar populations

and the stellar mass buildup in galaxies in the first 1.5 Gyr after the

Big Bang.

We have studied the spectral energy distributions (SEDs) of

z∼4-8 Lyman break galaxies (LBGs) in the rest-frame

UV and optical and compared them to synthetic stellar population models

to learn about the properties of these galaxies. We have found that the

typical best fit ages for these systems are in the range 300-600 Myr. In

a more general context this is not a very old population but at

z≥4 this represents a large fraction of the cosmic

time, indicating that these galaxies likely started forming stars much

earlier, at z≥10.

The star formation Rates (SFRs) estimated for LBGs at

z≥4 are generally in the range 1-100 M_sun

yr^-1. The stellar mass estimates are most robust for

sources with good Spitzer/IRAC detections,

corresponding to galaxies with stellar masses

≥10^8.5 M_sun at z∼4

(≥10^9.5 M_sun at z∼7).

For sources with lower rest-frame optical luminosities, that, as a

result, are individually undetected in IRAC, their average stellar

masses have been studied in a stacking analysis of a large number of

sources. This enables us to reach stellar masses

∼10^7.8M_sun at z∼4.

The stellar masses show a fairly tight correlation with UV luminosity or

SFR, and the zeropoint of the relation does not seem to evolve strongly

with redshift. This relation is a direct reflection of a correlation

between the UV and optical colors and it favors a typical star formation

history (SFH) at high redshift in which the SFR of a galaxy increases as

a function of time. This is consistent with the observed brightening of

the UV luminosity function (UV LF) and with expectations from numerical

simulations.

We have taken advantage of the UV luminosity vs. stellar mass relation

observed in LBGs at z≥4-7 to derive the stellar mass

function (SMF) of galaxies at these redshifts. The method uses a

combination of the UV LF and the mean UV vs. stellar mass relation

(including the scatter, estimated to be ∼0.5 dex at bright

luminosities at z∼4). This method allows an

analytic estimate of the low mass slope of the SMF. This slope (the

power-law exponent of the SMF at low masses), is estimated to be in the

-1.44--1.55, range which is flatter than the UV LF faint end slope at

these redshifts (≤-1.74). This means that low mass systems

contribute less to the total stellar mass density (SMD) of the Universe

than would have been estimated assuming a constant mass-to-UV-light

ratio. We show that this is also much flatter than the theoretical

predictions from simulations, which generally over-predict the number

density of low mass systems at these redshifts.

The UV luminosity vs. stellar mass relation indicates only a small

variation of the mass-to-light ratio as a function of UV luminosity.

This is confirmed in a stacking analysis of a large number of sources

from the HUDF and the Early Release Science fields (∼400

z∼4, ∼120 z∼5, ∼60 z∼6, 36 at z∼7).

Interestingly, the stacked SEDs at z≥5 in the

rest-frame optical shows a color [3.6]-[4.5]∼0.3 mag. This color is

hard to reproduce by synthetic stellar population models that only

include stellar continua, and it probably indicates the presence of

moderately strong emission lines (Hα EW_rest∼300

Å). The contribution from such emission lines in the IRAC fluxes

indicates that the stellar masses and ages could both be over-estimated

by a factor ∼2.

One of the most interesting results presented in this thesis is the

apparent plateau of the specific SFR (sSFR = SFR / stellar mass). In

early results, the similarity in the SEDs of galaxies at a given UV

luminosity in the z∼4-7 redshift range resulted in

very similar estimates of the SFR and stellar masses of these galaxies.

Furthermore, we find that the reported sSFR estimates at

z∼2 are also very similar to the ones in the

z∼4-7 redshift range (∼2 Gyr^-1

for ∼5×10⁹ M_sun galaxies). A puzzle

arises from the fact that the dark matter accretion rate onto halos

is predicted to decrease monotonically and rather fast as a function

of cosmic time (approximately propotional to

(1+z)^2.5). If gas and star formation

follow the inflow of dark matter, the sSFR at a constant mass should

also decrease monotonically with time, which is contrary to the

indication from these observations. When we include the possible

effects of emission lines, the stellar masses decrease by a factor

∼2× at z≥5. The revised stellar masses

may favor a slowly rising sSFR at z≥2, but the

rise as a function of redshift is still much slower

(sSFR(z) propto

(1+z)^0.6) than that of specific dark

matter accretion rate. This suggests that the stellar mass buildup

is somehow decoupled from the dark matter buildup at early times.

A detailed understanding of the connection between the buildup of galaxy

mass and dark matter is key for models of galaxy formation in the early

Universe. It will be crucial to expand on analyses like the one

presented here, including larger samples and broader stellar mass

ranges, to explore the buildup of galaxies with improved statistics.

Wide-area surveys with newly acquired HST and

Spitzer data, and the upcoming generation of

instruments, will likely provide the opportunity to make such a

connection.