Star Formation History of the Hubble Ultra Deep Field

Rodger ThompsonSteward ObservatoryUniversity of Arizona

Blameless Collaborators

Mark DickinsonDaniel EisensteinXiaohui FanGarth IllingworthRob KennicuttMarcia Rieke

Topics

The star formation intensity distribution functionStar formation history of the HUDFNature of the star forming galaxiesThe near infrared background

Purpose of the Program

Track the evolution of baryons from gas to starsDetermine the major venues and epochs of star formation What galaxies form the most stars? Episodic

or steady?

Is there a near infrared background? What constraints do the observations place

on Pop. III star formation?

The HUDF and NUDF

The Hubble Ultra Deep Field (HUDF) is a small area (~11 sq. min.) centered on the Chandra Deep Field South with deep ACS images in 4 broad optical bands.The NICMOS Ultra Deep Field (NUDF) is a smaller area (~6 sq. min) near the center of the HUDF with deep images at 1.1 and 1.6 m.All results discussed here come from the NUDF

The F775W Mag. vs Redshift

AGN

Characteristics of the Galaxies

Preponderance of very early SEDs Little blue galaxies

Very few L > L* galaxiesFew z > 6 galaxies Partly due to relatively conservative

source extraction.

Very few luminous galaxies with z > 3 compared to the NHDF

Star Formation Rates

Star formation rate determined from the rest frame 1500 Å flux via the Madau relation.The flux is measured from the selected SED without extinction to produce an extinction corrected SFR.

Star Formation Intensity Distribution

The star formation intensity x is the SFR in M per year per proper square kpc.

The distribution function h(x) is the sum of all proper areas in an x interval, divided by that interval and divided by the comoving volume defined by the field and redshift interval.*Under this definition SFR is the first moment of h(x); SFR = ∫x h(x) dx

* Lanzetta et al. 1999, ASP Conf. Ser. 191, 223

Application of the Distribution

The SFR is calculated for every pixel that is part of a galaxy.Assumes a uniform SED and extinction within a galaxyAssumes that the rest frame 1500 Å light is distributed in the same way as the observed flux in the ACS F775W band.

Star Formation Density

Log Star Formation Intensity x in M per year per kpc2

Log(h(x))

60% complete

95%complete

Redshift = 1

Starburst

About 80% of the stars are formed in a “starburst region”

The Observed h(x)

Correction for Surface Brightness Dimming

The (1+z)-4 photon surface brightness dimming removes galaxies and parts of galaxies from detection.This is the cause of the dip in h(x) at low values of x for z >1.The SFR can be corrected for this effect by matching the h(x) determined for z=1 to the bright end of the h(x) at other z values and integrating over the matched distribution.

Star Formation History of the NUDF

How is the Star Formation Distributed?

90% of the star formation occurs in 10-20% of the galaxies.The percentage increases with redshift, probably because of dimming.Evidence for episodic star formation with a duty cycle of 10%?

Comparison with the NHDF

SFR History of the Universe

Roughly constant from z = 1 to 6Average of z = 1 to 3 higher than 4 to 6Since most of the star formation occurs in ~10% of the galaxies, wider but shallower surveys may map the majority of star formation

Near Infrared Background Excess

Claims of a Near InfraRed Background (NIRBE) of ~70 nW m-2 sr-1, not due to known galaxies, stars or zodiacal light, that peaks at 1.4-1.6 m.Resolved objects in the NUDF and NHDF contribute 6-7 nW m-2 sr-1, a factor of 10 below the claimed background.Fluctuations in deep 2MASS images claimed as evidence for a population of very high redshift (10-15) Pop. III stars. (Kashlinsky et al. 2006)

Implications of the NIRB

Most popular model for the NIRB is the light from the high redshift Pop. III stars that reionized the universe.

Requires that the total number of baryons turned into stars in the first 3% of the age of the universe be greater than or equal to the total number of baryons converted to stars in the remaining 97%.

The metals produced by this conversion must be hidden in black holes.

There must be no x-ray producing accretion onto the black holes.

The NIRB must not interact with TeV emission from distant blazars.

Fluctuation Analysis

Results of the Fluctuation Analysis

The fluctuations observed in the 2MASS field can be completely accounted for by the redshift 0-7 galaxies such as those observed in the NUDFThere is no need for an excess population of high redshift Pop.III stars to account for the fluctuationsFluctuations have been removed as evidence for a NIRBE at 1.6 m

The IRTS NIRBE

Wide field of view spectrometer Aperture almost 17 times the size of the

NUDF

Zodiacal light and contributions from sources determined from modelsAfter subtraction of modeled components, 70 out of 330 nW m-2 sr-

1 remain and is attributed to a NIRBE

The NIRBE According to IRTS

Differences

The zodiacal component determined by medianed images in the NUDF exceeds the IRTS modeled component by 100 nW m-2 sr-

1. Dwek et al. 2006 point out that the IRTS spectrum is better fit by a zodiacal spectrum than a high z Pop.III spectrum.The IRTS NIRBE is most likely due to an under estimate of the zodiacal light component by the model.

Caveats

A NIRBE component that is flat on scales of greater than 100” would be mistaken for zodiacal light in our reduction.

At odds with CMB predictions

A NIRBE component that is clumped on the order of several arc minutes could be missed by our two small fields.

Archival proposal to check other fields

However the light in a NIRB can not be distributed in the same manner as the light from baryonic matter at redshifts of 6 and less.

Conclusions

The star formation history of the universe is roughly constant from z=1-6.The vast majority of star formation occurs in a minority of galaxies at any one time.Fluctuations have been removed as evidence for a NIRBE at 1.6 m.The IRTS NIRBE is probably zodiacal flux.Any NIRBE must be either maximally flat or maximally clumped.