A Steep Faint-End Slope of the UV LF at z~2-3: Implications for the Missing Stellar Problem

C. Steidel ( CaltechC. Steidel ( Caltech))

Naveen Reddy (Hubble Fellow, NOAO)

Galaxies in Real Life and Simulations, Leiden, Netherlands, 16 September 2008

Are There Problems with Stellar Mass/ Star Formation Rate Measurements?

• SFR / stellar mass calibrations

• IMF evolution?

• Dust extinction

• Incompleteness

Luminosity/Redshift Evolution of Dust;

Incompleteness of Stellar Mass Density Measurements

Reddy & Steidel (2008b)

Quantifying Number Densities via the UV Luminosity Function

Advantages of UV:

- direct tracer of massive star formation, modulo the effects of dust

- deepest observations up to 2000 times more sensitive than those at IR and longer wavelengths

- accessibility over almost the entire age of the Universe

Need for Spectroscopy:

- assess contamination

- quantify perturbing effects of line emission and consequence for sample completeness

- color corrections for IGM opacity

- physical state of the ISM (stellar population; outflows)

Going Beyond the Spectroscopic Limit

Reddy et al 2008a

?Use spectroscopic trends as a zeropoint for determining how galaxy properties depend on UV luminosity

Preferential Scattering due to Lya:

- contribution of high Ly EW systems at UV-faint magnitudes; ~10% (>50 A) at M ~ -17

Change in mean dust attenuation:

- no variance with UV luminosity?

- evolution of dust with UV luminosity; UV-faint galaxies less dusty than UV-bright ones?

Upshot: these systematics modulate the inferred faint-end slope of the UV LF by up to ~10%, with a tendency to steepen

Advantages of our analysis

- > 2000 spectroscopic redshifts at the bright-end

- modeling of systematic effects

- maximum-likelihood constraints on LF that are robust to non-uniform sources of scatter

- 31000 LBGs in 31 independent fields

Results on the UV LF at z~2-3

Reddy et al 2008b

Steep faint-end slope of ~ -1.73, similar to that measured at z~4-6

0.1L*0.07L*

• z=2:

N(0.07L*<L<L*)~0.98N(>0.07L*)

(0.07L*<L<L*)~0.84(>0.07L*)

z=3:

N(0.1L*<L<L*)~0.97N(>0.1L*)

(0.1L*<L<L*)~0.82(>0.1L*)

Make use of UGR+JK+IRAC photometry for several hundred galaxies

- spectroscopic redshifts remove a key degeneracy from this analysis!

- independent monochromatic indicators constrain SFRs and reddening

Stellar Population Modeling

R<25.5

Reddy et al 2008b

Trends between UV luminosity and SFR with Stellar Mass

Sawicki et al. 2007

Dust Corrections as a Function of UV Luminosity

~4-5 dust correction

~2 dust correction

Mean Correction of ~2-3 for Total UV

Luminosity Density

For >0.083L* galaxies

Estimate of the Stellar Mass Function at z~2

Reddy et al 2008b

Integral of the Star Formation History

z

L>0.083L*(z>2)

[L>0.083L*(z=2)]

Integrated to L=0.083L*(z=6)

What about contribution from massive galaxies?

Contributions to the Stellar Mass Density at z~2

For >0.083L* galaxies

(R<25.5;<10^11 M*) ~ 3.7 +/- 0.2 crit

(R>25.5;<10^11 M*) ~

2.0 +/- 0.2 crit

(>10^11 M*) ~

1.6 +/- 0.4 crit

Comparisons with the Integrated Star Formation History

>0.083L* R<25.5; M<10^11 Msun

R>25.5; M<10^11 Msun

M>10^11 Msun

WRONG!

Redshift Evolution of the Faint-End Slope

Reddy et al 2008b

Slope roughly constant at z>2, with ~ -1.73

- but, strong evolution in UV LF implies sub-L* galaxies at z~6 are different from sub-L* galaxies at z~2

- evolution of faint-end slope may have less to do with delayed feedback and more to do with the availability of low mass halos with cold gas below z~2

Conclusions

Constraints on the faint-end slope of the UV LF:

- UV LF evolves strongly between z~6 and z~2

- very steep faint-end slope of the UV LF of ~ -1.73 at z~2 and z~3, remarkably similar to those derived at higher redshifts (z~4-6)

-but, faint-end population evolves between z~6 and z~2, so evolution of faint-end slope may have less to do with feedback and more to do with the availability of low mass halos with cold gas below z~2

- dust corrections depend on how far one integrates to obtain the UV LD

Combining the stellar masses of star-forming galaxies at z~2-3 with the luminosity function:

- significant stellar mass density in UV-faint galaxies (R>25.5) as in UV-bright ones

- appears to resolve the supposed discrepancy between stellar mass density estimates and the integrated star formation history up to z~2

ADDITIONAL SLIDES

Infrared Luminosity Function at z~2

Total IR LD ~ 1.3e09 Lo/Mpc3

Total IR LD (Caputi et al. 2006) ~ 6.6e08 Lo/Mpc3

Limit of MIPS observations without prior information