My field of interest is galaxy formation. I am investigating
the early phases of evolution, when protogalaxies
separated from the expanding background and collapsed
under gravity. I am specifically interested in the set of
events that gave rise to the rotating disks of stars and gas
that we find in spiral galaxies today. Theories of galaxy
formation predict somewhat different conditions as
massive clouds of gas collapse to form galaxies in the
early Universe. We look back in time some 10 to 15 billion
years at remote clumps of hydrogen gas and stars with
rather large cosmological redshifts.
Since the starlight emitted by such distant objects is
exceedingly faint, I have focused on the interstellar gas.
To study its composition, temperature and motion, we use
the light of distant quasars located behind the
protogalactic clouds. As the quasar light passes through
the gas, some of it is absorbed by hydrogen molecules
and other elements,leaving an "absorption -line" signature.
Of particular interest is the "damped Lyman-alpha"
absorption, produced by cold, neutral hydrogen gas.
Surveys with large optical telescopes have been used to
search for these damped Lyman-alpha absorption
systems. The results of the surveys show that most of the
known matter in the distant, early universe is contained in
these cold, quiescent layers of neutral hydrogen that are
responsible for the damped Lyman-alpha lines. Moreover,
the co-moving density seen in these clouds is similar to
the density of visible matter in current spiral galaxies.
Therefore, the implication is that we have detected the
component in the early universe that makes up the present
stellar content of galaxies.
We are currently using the Keck Telescope with its
high-resolution spectrograph to study damped Lyman-alpha
and other absorption features that we associate with
protogalactic disks. The kinematics of the clouds provide
information about the newly-formed disks, their rotational
and random motions. We can also learn about the
abundances of elements heavier than hydrogen, which
tells us about the star-formation history in these systems.
The following are recent results arising from research carried out with
(Download .pdf file of recent results
Star Formation Rates in Damped Lyman-Alpha Systems|
I have developed a new technique based in HIRES observations that yields
the heating rate of neutral gas in damped Lyman-alpha systems (DLAs), and
from that the star formation rate per unit comoving volume. Until now,
comoving star formation rates have been obtained for highly luminous
objects such as Lyman-Break Galaxies. My technique measures rates in
objects more representative of the protogalactic mass distribution and
at redshifts currently inaccessible to the Lyman-Break Technique.
The idea- based on the heating mechanism for the ISM- is as follows: Massive
stars that form in the DLA neutral gas emit far UV radiation that illuminates
dust grains in the gas. Some of the incident photon energy goes into photoejected
electrons which then heat the gas. In this case, the heating rate per nucleon,
Figure 1: Star formation rate per unit comoving volume versus z for Einstein
de-Sitter cosmology with h=0.5. Red circles denote galaxies detected in
emission. Green (magenta) points and error bars denote DLAs for the CNM
(WNM) solutions. Blue and cyan curves are fits to the emission and the CNM
and WNM points respectively. The CNM solution is more likely since the WNM
case produces more background radiation than observed.
where D/G is the dust-to-gas ratio and epsilon is
is the heating efficiency. In a plane parallel layer, the mean intensity of far
UV radiation, J, is proportional to psi*, the star formation rate per
unit area. Since epsilon is well determined, we can measure psi*
provided we know D/G and gammad. I measure gammad by
equating it to the cooling rate. This is inferred from the C II*
absorption line which measures the population of the excited P23/2
fine-structure state. By measuring cooling rates in about 30 DLAs and by
equating the dust-to-gas ratio to the metallicity, [Fe/H], I find that the star
formation rate per unit area is similar to that in our galaxy (log psi* =
-2.4 Msolar kpc-2y-1).
I then use DLA statistics to deduce the comoving star formation rate rho*.
Figure 1 shows that rho* for DLAs is similar to rho* for
Lyman-Break Galaxies if the DLA gas is in the cold neutral medium (CNM). The warm neutral
medium (WNM) case is probably ruled out, because the higher star formation rates result in too
much background radiation. I am currently considering the many implications of this result.
One of these is that the value of rho* implies that the mass of metals
accumulated by z ~ 2.5 is higher than observed in DLAs.
Figure 2 summerizes our most recent results on the metallicity evolution of DLAs.
The figure shows Fe abundance ([Fe/H] is the logarithmic abundance of Fe with
respect to H, normalized to solar abundance) versus redshift. The size of each
data point is proportional to logN (HI) to indicate the relative contributions
to <Z>, the column density weighted mean [Fe/H]. We measure
<Z> since it equals the ratio of the comoving density of metals
to that of hydrogen. Figure 2 indicates a very mild evolution of the metallicity
of protogalaxies. Dividing the data into two redshift bins,
Figure 2: [Fe/H] vs z. Red squares are for data points taken with HIRES.
Blue circles are column-density weighted mean, <Z>, for
discussed in text. Green points are data taken with ESI.
[1.5,3] and zhigh = [3,4.5], we find no statistical difference
between <Z(zlow)> and <Z
(zhigh)>. On the other hand, there is a slight evolution
in the unweighted mean [Fe/H]. In any case, the constancy of <Z>
with redshift is at variance with the predictions of most chemical evolution
models which predict stronger evolution in <Z>. Furthermore,
I note that out of the 41 objects in Figure 2, none has [Fe/H] < -2.7.
This is true even though we could have measured [Fe/H] = -3.5 in most DLAs.
Whether this "floor" is a function of population III pre-enrichment is a matter
Kinematics of Damped Lyman-Alpha Systems|
We have studied the kinematics of the neutral and ionized gas in DLAs by
obtaining accurate HIRES velocity profiles of low ions such as Fe II and
high ions such as C IV. The results of our work are summarized in Figure 3.
This compares the low-ion with high-ion velocity profiles. Our statistical
analysis supports the visual impression that the ionized gas and the neutral
gas comprise distinct kinematic subsystems. This is indicated by the misalignment
in velocity space of the narow components that make up the velocity profiles,
and the difference between the widths of the profiles. However, despite their
differences, the kinematic subsystems are interrelated as indicated by a
statistically significant C IV versus low-ion cross-correlation function.
Moreover, the velocity widths of the C IV profiles are greater than or
equal to the low-ion widths in 29 out of the 32 cases. This indicates the
two systems are in the same gravitational potential well. We have used these
results to test the standard scenario in which the high-ion (ionized) gas
undergoes radial infall into dark matter halos containing centrally located
rotating disks comprising the low-ion (neutral) gas. Neither the CDM nor
passive evolution modles predict kinematics that are consistent with all
our kinematic tests. The main problem is that the models fail to reproduce
the significant amplitude if the C IV versus low-ion cross-correlation
function. We are currently testing new models in which the infalling
ionized gas has the same angular momentum per unit mass as the collapsed
neutral gas. In principle, this should increase the correlation
between the velocity fields of the two components.
Figure 3: Velocity profiles of high-ions (dark lines) and low-ions
(light lines) in 32 separate DLAs.(Click on image for large version.)
Survey for Lyman-Break Galaxies Correlated with DLAs|
The z ~ 4 phase of this project is now finished. We have completed
multi-color photometry and multi-slit spectroscopy on 3 fields toward 4 DLAs.
We have doubled the number of know Lyman-Break galaxies (LBGs) at z ~ 4.
Our efficiency in finding LBGs was improved through the use of photometric
redshifts. However, the sample is still too small to measure either the two-point
correlation function for LBGs or the cross-correlation function between
LBGs and DLAs at these redshifts. So far the LBG two-point correlation function
has only been measured at z ~ 3.
Possible Detection of Cosmological Evolution |
of the Fine Structure Constant
Using high-resolution spectra of metal lines obtained with Keck I HIRES spectrograph,
my collaborators and I have search for time variability of the fine structure constant,
alpha. Variations in alpha would lead to detectable shifts in the rest
wavelengths of redshifted UV resonance lines such as those used to deduce kinematics
and metallicities in DLAs (e.g. Ni II, Cr II, Zn II, etc). For relativistic fine structure
splitting in alkalai-type doublets, the separation between the lines is proportional to
alpha2, so that small relative variations in the separation are
proportional to alpha. We used a new technique which is more sensitive than the
alkalai technique, since it is not restricted to transitions with respect to the same
ground state. Rather, it compares transitions relative to different ground states in
different species. Using species with widely differing atomic masses produces an
increase in sensitivity because the difference between ground-state relativistic
correction can be large and of opposite sign. We measure variations in alpha,
by explicitly including these variations in a Gauss-Newton optimization code in which
other parameters such as velocity width and thermal width of the absorption lines are
allowed to vary freely. The results, based on 72 quasar absorption systems, are
summarized in Figure 4. The figure shows possible evidence that (dalpha /
alpha) decreases with z.
This is a potentially exciting result that needs to be checked with independent data
sets. We have searched for possible systematic errors and find that removal of the
most important of these, atmospheric dispersion and isotopic abundance evolution,
would enhance the significance of these results.
Sample at present:|
28 Mg II/Fe II absorption systems toward 17 QSOs (Churchill)|
18 DLA absorption systems (+3 further Mg/Fe systems)|
toward 13 QSOs (Prochaska/Wolfe)
21 Si IV doublets toward 13 QSOs (Prochaska/Wolfe)|
(All Keck spectra, <s/n> ~ 30/pixel, resolution FWHM
~7 km/sec |
for entire dataset)
(d alpha / alpha)|
Fe II/Mg II||
0.5 < z < 1.8
-0.70 +/- 0.23|
Ni II/Cr II/Zn II||
1.8 < z < 3.5
-0.76 +/- 0.28|
2.0 < z < 3.0
-0.5 +/- 1.3|
-0.10 +/- 0.17|
Table 1: Summary of results for 4 independent samples. Values of|
(d alpha/alpha) are weighted means and are in units of
and AD indicate "many-multiplet" and
"alkalai-multiplet". Nobs is
the number of absorption
systems in each sample.
Overall deviation from zero is 4 sigma, (uncorrected for any systematic effects.)
Figure 4: Summary of evidence for variations in fine structure constant
from Webb,et al. (2001)
Thick Disk Abundance Patterns|
We (Prochaska,et al.,2000) have obtained accurate HIRES spectra for 10 thick
disk stars. Our analysis confirms previous studies of O and Mg which show
enhancements of [O/Fe] and [Mg/O] relative to the thin disk. We also find that
the elements Si, Ca, Ti, Mn, Co, V, Zn, Al, and Eu have a chemical history distinct
from the thin disk. Moreover, the thick disk abundance patterns tend toward the
values observed for halo stars with [Fe/H] = -1. This suggests that the thick
disk stars had a chemical enrichment history similar to the metal-rich halo stars.
Furthermore, we find that all 10 stars exhibit enhanced alpha/Fe ratios
with O, Si, and Ca exhibiting trends of decreasing overabundance with increasing
[Fe/H]. If these trends are explained by the onset of Type Ia Sn, then the thick
stars formed over the course of >/= 1Gyr. This slow formation time is
apparently incompatible with dissipational collapse scenarios for the formation
of the thick disk. Models with heating of an initially thin disk are favored
Figure 5: Comparison of abundance patterns of DLAs (green circles)
and thick-disk stars (blue stars) from Prochaska,et al.,2000.
Figure 5 shows the impact of the thick disk abundances on interpretations of
abundances in DLAs. The DLA abundance pattern exhibits a classic example of
halo abundances with enhanced alpha-elements, deficient Mn, and normal
Ni, Cr, and Al. While the [Zn/Fe] ratio of DLAs is high, the overall Type II
Sn pattern is clearly evident. In fact, there is tentative evidence for an
evolutionary sequence in the abundances of DLAs to the thick disk stars.
On the other hand, the abundance pattern is also consistent with element
depletion in the ISM. This would explain the enhanced Si, Zn, and S, but is
inconsistent with the Mn trend. Our guess is that the DLAs consist of a Type II Sn
pattern on top of a small amount of dust.