Galaxies 626 Lecture 10 The history of star formation from far infrared and radio observations
Cosmic Star Formation History Various probes of the global SF rate: ρ* (z) M yr 1 comoving Mpc 3 UV continuum (U band surveys, GALEX, LBGs) Hα and [O II] emission in spectroscopic surveys mid IR dust emission 1.4 GHz radio emission
Cosmic SFH: Calibration Kennicutt 1998 Ann Rev A&A 36, 189 (Salpeter IMF) UV continuum (1250 2500 Å) : Pro: Extensive datasets over 0<z<6; know stellar evolutionary tracks & w/imf, # of stars in a given mass interval, so know light produced each λ Con: dust! IMF dependence 2. Line emission (Hα, [O II]) : Pro: Very sensitive probe, available to z~2 (lose in thermal IR) Con: strong IMF dependence ( 3); dust (use Balmer decrement to correct); excitation uncertainties [OII] 3. Far IR emission (10 300 µm) : Pro: Independent method, available for obscured sources to high z: Con: uncertain source of dust heating (AGN/SF?); primarily applicable in starbursts due to detection limits at high z; bolometric FIR flux required
where light emerges depends on where the stars are relative to the clouds
Star Formation UV ν ν IR Dust ν IR Gas Dust V U ν Dust UV ν SN Hα ν Gas e e-ee- - Radio ν IR ν
log Luminosity M82 a Star Forming Galaxy Efstathiou et al. MNRAS submitted 1 Wavelength 10 / µm100
Sllva et al. 1998 The proportion is a strong function of galaxy luminosity ULIG OPT LIG FIR spiral Elliptical
SFR Hα SFR infrared SFR UV Comparison of Star Formation Rate Measures Star formation rates in the radio Solar masses per year Cram et al 1999
For the most luminous galaxies Star formation activity dominated by short lived massive stars Their UV light absorbed by dust and re emitted in FIR At higher redshifts best observed in sub mm Surveys in FIR and sub mm required to trace obscured SFR history
Galaxies 626 Fluxes, luminosities and K corrections
Luminosity Astrophysical objects tend to emit their light energy over a range of different frequencies, ν, it is thus useful to define the Luminosity( Lν) to be the energy per unit time per unit frequency interval ΔE ergs/s Hz 1 L ν= /Δν Δt L ν dν= Lν dν ν de dt Lν dν=l
Luminosities Important to draw a distinction between Bolometric Luminosity LBol, Total luminosity of a galaxy, measured in ergs/s or L (or absolute magnitudes) Line Luminosity e.g. LHα, Total luminosity of an emission line, measured in ergs/s or L In band power e.g. luminosity emitted in a given wavelength interval measured in ergs/s or L (or absolute magnitudes) Luminosity (density) Luminosity per unit frequency, measured in ergs/s Hz 1, often quoted as νl measured in ergs/s
K Correction The emission from a galaxy is observed at a different wavelength from the one that at which it was emitted due to the cosmological redshift In general one wants to compare the emission properties of galaxies at the same (emitted) wavelength The K correction is an additional term in the flux density to luminosity relationship which accounts for this difference
Flux density Flux density ( fν) to be the energy per unit time per unit area per unit frequency interval Watts / metre2 / Hertz, erg/s cm 2 Hz 1 f ν dν= f Jansky: 1Jy = 10 23 erg/s cm 2 Hz 1 Common to see... Flux density ( fλ) the energy per unit time per unit area per unit wavelength interval Both are often simply called flux!
K Correction We want to relate the observed light to the light emitted at the rest frame frequencies f= L 2 1 z 2 4 πr20 S k We can see f ν ν 0 dν 0 = we know = L 4 πd2l L ν νe dνe 4 πd2l νe = 1 z ν0 dνe = 1 z dν0 f ν ν 0 dν0 = [ ] L ν ν 0 1 z 1 z dν0 4 πd 2L
K Correction shift of spectrum f ν ν0 = [ Band pass ] L ν ν 0 1 z 1 z 4 πd 2L
Galaxies at Higher z wavelength If the same object is seen further away Observed band We see the light redshifted ν0 =νe / 1 z νe = 1 z ν0 emitted feature at a fixed λ we see light emitted from bluer parts
galaxies at higher z Observed Band ~ 1013 Hz Emitted band ~ 1014 Hz νe = 1 z ν0 Given rest frame frequency band is observed in a narrower band thus energy per unit frequency increases dνe = 1 z dν0
K Correction in magnitudes If fluxes and luminosities are expressed in magnitudes m= M 5 log Hence name: K correction DL 10 pc K z Shape of spectrum and band pass correction
At 850 micron, as go to higher redshifts, start sampling closer to the peak of the blackbody spectrum so maintain sensitivity; negative K correction
FIR galaxies at high z In Sub-mm galaxies can become easier to detect at high-z
Galaxies 626 Galaxy populations in the submillimeter
Deep SCUBA Imaging at 850 microns Blank & Cluster Lens Fields LH NW (39 hr exposure) Barger et al. 1998 A2390 cluster (22 hr exposure) Cowie, Barger & Kneib 2002
Hughes et al. 1998
Submillimeter Number Counts Barger et al. 1998, 1999 Scott et al. 2002 Webb et al. 2003 Hughes et al. 1998 Blain et al. 1999 Chapman et al. 2002 CBK 2002
Submillimeter Number Counts EBL convergent fit Typical source about 1012 solar L Cowie, Barger, & Kneib 2002
Determining the Redshift Distribution
A370 SCUBA 38 hrs 2. 5' 2. 5'
A370 U /, R, K / 4 '' radius error circles
By spectroscopically observing all possible optical counterparts in the SCUBA error circles, redshifts can be determined for at least 25% of the bright SCUBA population
Some reliable spectroscopic identifications have been made by targeting all nearby optical sources Soucail et al. 1999 Barger et al. 1999
Ivison et al. 1998 z=2.8
However, in most cases the candidate optical/nir counterparts are intrinsically faint and hence difficult to identify
SCUBA sources with spectroscopic ids are biased towards AGN Presence of AGN makes counterpart more optically luminous AGN spectra are easier to identify
Not surprisingly, SCUBA sources with AGN spectra also tend to be the ones with X ray counterparts Thus, searching for counterparts to the SCUBA sources in deep X ray data reveals the presence of AGN
Alternative Approach for Studying the Submillimeter Sources Radio observations are useful for pinpointing the positions of the SCUBA sources (the FIR radio correlation has been empirically determined for both star formers & radio quiet AGNs) Around 60% of submm bright (>5 mjy) sources are 20cm radio sources
HDF N Small circles: 20cm Bigger circles: SCUBA Rectangle: GOODS ACS Large circle: DEIMOS W. H. Wang et al. 2004
Spectroscopy of 20cm Selected Submm Sources redshift desert radio sensitivity limit Chapman et al. (2005) 18 Keck nights ~ 100 submm sources observed ~ 70 identified median redshift = 2.2
Problems
The submm sources that are above SCUBA s blank field detection limit of 2 mjy only contribute ~20% of the total submm background Normalized to Puget et al. 1996 Fixsen et al. 1998 80% 50% 20% 10%
Only 60% of the above 20% can be detected by radio telescopes to 40 microjy at 20 cm 40 µjy ULIRG LIG
Thus, the spectroscopically identified radio selected submm sources only represent ~10% of the total background So, we really need to learn about the faint (<2 mjy) sources A more recent approach that may be much better is a stacking analysis: (1) measure the submm fluxes at the positions of another population (2) determine the mean flux of that population (3) find the contribution to the EBL
For example, we can ask how much of the submillimeter background is produced by the Chandra Deep Field North hard X ray AGN sample?
When we do the stacking analysis, we find that AGN contribute only 15% of the 850µm background Thus, most of the submm EBL is probably due to star formation
Typical Source about 0.7mJy Barger et al. 1998,1999 Scott et al. 2002 Webb et al. 2003 Hughes et al. 1998 Blain et al. 1999 Chapman et al. 2002 CBK 2002 Need another population with at least a surface density of ~25000/deg2 to identify the EBL
Clear that the current 20 cm samples do not have enough sources There are only marginally enough 24 micron sources Thus, the current 20 cm and 24 micron samples are not deep enough to have reached the fainter submm sources
However, NIR samples are deep enough 20 x 20 shown RJK
Band N <S850> Iν (mjy) (Jy Deg 2) ULB 1.6 µm 3094 0.20 ± 0.03 18.3 ± 2.4 IRAC 3.6 µm 5245 0.11 ± 0.0066 19.6 ± 3.4 MIPS 24 µm 493 0.66 ± 0.06 11.4 ± 1.1 VLA 20 cm 101 1.31 ± 0.13 4.0 ± 0.40 1.6 µm + 3.6 µm 1783 24.0 ± 2.0 We can see that 20 cm samples are good at picking out the brighter submm sources, but there aren t enough of them to produce the EBL; NIR does best!
Submm EBL vs Spectral Type Sd? E Sb Intrinsically Red Irr Sc Intrinsically Blue
Submm EBL vs Redshift core Near IR sample
Contrast with the brighter sources: fainter sources peak at lower redshifts core Near IR sample Chapman et al. (2005)
Dusty Star Formation History Compute 20 cm EBL vs z for all the sources contributing to the submillimeter light (the core sample)
Dusty Star Formation History Translate into a bolometric luminosity density using the FIR radio correlation Convert to a star formation rate density [using Kennicutt 1998] A submm analysis gives the same answer at z>1 (where the sources are near ULIRGs) but is subject to assumptions about the dust temperature
Comparison of the FIR determined star formation with the UV determined star formation Maximal corrections for missing EBL, if at z=1 3 Directly measured FIR star formation ρ t
At low redshifts, the correction from the UV is not so large, but at higher redshifts, it is factors of 3 5 (green curves) ρ t
These results are similar to estimates of the extinction correction made from the UV continuum characteristics of UV selected objects However, this is a little misleading, because the samples are somewhat disjoint blue star forming galaxies contribute substantially to the UV star formation but not to the FIR star formation Thus, the extinction corrections must be higher in the galaxies that produce the bulk of the FIR light and lower in the blue star forming galaxies
All of the backgrounds (including 850 µm) have strong contributions from below z=1; however, the UV and 850 µm come from different galaxy populations UV 850 micron 20 cm [OII]
Cumulative star formation history (SFH) shows actual growth of galaxies with time Cosmic baryon density Present day stellar mass density, Cole et al. (2002)
Summary Multiwavelength (optical/uv, NIR, submm) observations have led to a revolution in tracking the history of star formation in the universe Most (60% 80%) of the submm EBL can be detected using the NIR population and comes from intermediate type galaxies at z ~ 1 (not the same sources that dominate the UV) We have a good understanding of the evolution of the co moving density of SF since z~3, which accounts for the observed stellar mass density at z=0. Half the stars we see today were formed by z~1.4
The End