Transcript LSST 2013

Connecting 20cm Observations
to Other Views of the Universe
Amy Barger
I will use GOODS-N/CDFN for illustration, focusing on z>1 populations,
where very luminous star formers become common
Comparison of FIR-determined star formation with UV-determined (
how much mass has accumulated by that time)
r˙approximates
t
rt
Amount of star
formation in UV
at early times is
very small
Star formation rate density for a Salpeter intial
mass function to 0.1 M(sun) x cosmic time
Wang et al. 2006
Ultraluminous Infrared Galaxies (ULIRGs)
• Discovered by IRAS in mid 80s
• LFIR > 1012 L >> Loptical (dusty)
• Disturbed morphologies. Powered
by nuclear starbursts and/or
obscured AGN (much debated)
Prototypical ULIRG: Arp 220
Most of its energy output is due to
starburst (180 solar masses/yr)
HST NICMOS Image
FIR-Radio Correlation
Powerful correlation known to exist at low redshifts;
parameterized by q (Helou et al. 1985, 1988; Condon et al. 1991)
æ
ö
æ P1.4GHz ö
LFIR
q = log ç
÷ - log ç
÷
12
è 3.75´10 erg / s ø
è erg / s / Hz ø
I will discuss later the evidence that this continues to hold
at high redshifts, at least for the most luminous galaxies
We can use this to convert between radio power and FIR
luminosity, and hence to star formation rates (SFRs) for
the luminous star-forming galaxies
Owen et al. (2013) radio sample with spectroscopic redshifts
(deepest 1.4 GHz VLA image; 5s=11.5 mJy)
CDFN has deep X-ray data for identifying X-ray AGN
2 Ms X-ray; Alexander et al. 2003
X-ray AGN with LX>1042 erg/s
X-ray quasars with LX>1044 erg/s
Not all of the highest radio power
sources are X-ray AGN!
Optical/NIR faint tail hard to
study, even with photometric
redshifts. A few have CO
measurements, but what are
the others?
They are not X-ray AGN (red)
High-Redshift Radio Sources
• Many of the optical/NIR faint radio sources may lie at high
redshifts
• In fact, there is a strong correlation between K magnitude and
redshift in the radio sample, making K magnitude a crude
redshift estimator [famous old K-z relation]
Black=spectroscopic
Millimetric redshifts
are determined from
the radio to submm
flux ratios, assuming
an Arp 220 SED
At z<1.5, nearly all high radio power sources are AGN
(either X-ray AGN or red and dead galaxies)
How many of the high radio power sources at z>1.5
are star formers?
Separating Star-Forming Galaxies from AGN
(1) High-resolution radio observations:
extended=star forming; compact=AGN
(e.g., Chapman et al. 2004 with MERLIN+VLA; Momjian et al.
2010 with HSA on VLBI; Guidetti et al. 2013 with e-MERLIN)
(2) Time variability (see Brandt talk)
(3) Submillimeter/millimeter surveys to look at the FIR properties of the
radio sources
Relatively inefficient to do this with pointed observations with
ALMA/SMA, as their small fields-of-view require scanning over large
areas
Better to use a single-dish camera system like SCUBA-2 or future
experiments like CCAT
SCUBA-2 area comparable in size to high-sensitivity VLA region (9’ radius)
SMA areas
SMA
detections
GOODS-N HST
Large beam sizes mean multiple sources may contribute to the flux.
SMA or ALMA can identify these multiples
SCUBA-2 12’
radius field
SCUBA-2
positions
(larger circles)
SMA sources
(small circles)
Multiplicity Controversy
• Karim et al. (2013) surveyed with ALMA the submillimeter sources
(SMGs) detected in their LABOCA Extended Chandra Deep Field South
Submillimeter Survey (LESS)
• They found that all of their brightest sources (S870mm>12mJy) were
composed of emission from multiple fainter SMGs, each with
S870mm<9mJy. No ALMA sources had S870mm>9mJy.
• They therefore proposed a natural limit of <103 solar masses per year on
the SFR of SMGs
• In the GOODS-N, we have 3 SMA detections of SCUBA sources with
S860mm>12 mJy and a 4th with S860mm>9mJy, all of which are singles
• LABOCA has a larger beam size than SCUBA, and ALMA has a smaller
field-of-view than the SMA, so multiplicity or non-detections may be
more common in the LABOCA/ALMA observations than in the
SCUBA/SMA observations
Radio data deep enough that most SCUBA-2 sources have counterparts
SMA 24’’
radius field
radio sources
(small circles)
SCUBA-2
position
(large circle)
With
accurate
positions,
one can
reliably use
multi-l data
to construct
spectral
energy
distributions
(SEDs)
Clear that distant SMG SEDs are quite similar to Arp 220
Arp 220
nbBn(T)
b=1
Td=47 K
Here the SMGs are normalized to Arp 220 at rest-frame 100 micron
A considerable fraction of the spectroscopically
identified high-redshift radio sources are SMGs
~20% of the SMGs are X-ray AGN [none are X-ray quasars], where we
may be seeing star formation in the host galaxy
Selection Biases
• The spectroscopically identified radio sources may be biased
against red and dead galaxies and very obscured AGN and be
biased towards
– Seyferts and quasars with high excitation emission-line features
– Star formers with strong UV absorption line features
– SMGs followed up with CO observations
• Thus, we cannot look at this figure and say what fraction of
high radio power galaxies are star formers
• All we can say is that there is a substantial number of high
radio power star-forming galaxies
FIR-Radio Correlation at High Redshifts
• We can use SMGs to test the correlation to high redshifts
– Assume Sn µ n a and a=-0.8 (Condon 1992; Ibar et al. 2010) to
get the radio power
– Compute LFIR over 42.5-122.5 mm
-data fully cover this range (vary Td and b to fit)
-lower redshift analyses used this range
• We can also test how well we can estimate the millimetric
redshifts using the Arp 220 SED (Carilli & Yun 1999; Barger et al.
2000) by comparing with the spectroscopic redshifts
Yun et al. (2001) found a local avg of 2.34+/0.01 and a scatter of 0.26 dex
The SMGs over z = 2 – 4.2 are consistent: avg of 2.36
Millimetric redshifts obtained from the ratio of the 860 mm flux to the
radio flux and the Arp 220 SED agree to within a multiplicative factor
of 1.4 with the available spectroscopic redshifts
Assume Invariance of FIR-Radio Correlation
• We have only shown this invariance for the more luminous
SMGs over z=2-4.2 (where we are able to measure the
Herschel fluxes)
• If we assume this invariance extends down to our 860 mm
flux threshold of 2 mJy and to z>5, then we can compute the
star formation rates (SFRs) for the individual sources from
the 1.4 GHz power (using q to obtain the normalizing
constant)
SFRs range from 600 to 8000 solar masses per year. Inconsistent
with the natural limit of <1000 proposed by Karim et al. (2013) based
on their ALMA observations of the LABOCA CDFS submm survey
A large and relatively invariant fraction of the overall SFR density is
contained in these massively star-forming galaxies: ~30% of the
Hopkins & Beacom (2006) extinction-corrected UV SFRD
Most of the star formation in the dusty, massively star-forming galaxies
may come from sources just below our current sensitivity limits
Slope if equal amounts of SFR in
equal bins
Can probe fainter with a combined submm and radio sample
Implications for LSST
• Can select the optically faint radio sources in the LSST Deep
Drilling Fields and eliminate slightly brighter objects based
on variability
• Survey this restricted sample with pointed ALMA
observations to determine if they are SMGs and what their
CO redshifts are
• Determine the distribution function for the high SFR dusty
galaxies
• Map the star formation history in these objects to the highest
redshifts possible. We now know massively star-forming
galaxies out to z=6.5 (Riechers et al. 2013); how common are
these?