1. The micro- and nano-Jy sky
Paolo Padovani, ESO, Germany
The sub-mJy source population
Which type of sources will SKA see at micro-Jy and nano-Jy flux densities? How many of them and how faint?
Which multi-wavelength properties will they have? How will we identify them?
Will take some time to explain the limitations of our modelling but the talk should also be interesting for the “experts”!
March 24, 2010
P. Padovani SKA 2010 Meeting

2. Radio (1.4 GHz) number counts
strong evolution of qso
and radio galaxies
N(>S) = (0.091±0.014) (S/200Jy)-1.10±0.13
[arcmin-2]
flattening below mJy
March 24, 2010
P. Padovani SKA 2010 Meeting
Kellerman et al. 2008

3. The sub-mJy population: a “hot” topic
The composition of the sub-mJy population has been debated over the past ∼ 25 years
More than 200 refereed papers in the ADS including: Mitchell & Condon 1985; Windhorst et al. 1985; Danese et al. 1987; Benn et al. 1993; Hopkins et al. 1998; Gruppioni et al. 1999; Richards 2000; Bondi et al. 2003; Ciliegi et al. 2003; Hopkins et al. 2003; Jackson 2004; Seymour et al. 2004, 2008; Huynh et al. 2005, 2008; Prandoni et al. 2006; Fomalont et al. 2006; Simpson et al. 2006; Bondi et al. 2007; Ivison et al. 2007; Bondi et al. 2008; Owen & Morrison 2008; Smolčić et al. 2008; Wilman et al. 2008; Padovani et al. 2009; Massardi et al. 2010
Most accepted paradigm until recently: star-forming galaxies
March 24, 2010
P. Padovani SKA 2010 Meeting

4.   Sub-mJy populations Padovani et al. 2009 Sub-mJy sources:
57±8% SFG
41±6% AGN
19±5% RQ AGN (46% of all AGN)
100% 
64% 0%
≈1/2 ≈1/2 SFG AGN
36% 
24%
10%
≈1/ ≈1/2
RL RQ
(low-power!)
RQ AGN counts scaled down version by factors 3-4 of SFG
SFG take over
Broadly consistent with, e.g., Smolčić et al and Seymour et al. 2008
March 24, 2010
P. Padovani SKA 2010 Meeting
Padovani et al. 2009

5. AGN and star formation in the Universe
VLA Observations
AGN and star formation in the Universe
Merloni et al. 2006
A complex relationship
accreting gas feeding black hole (AGN) might trigger starbursts
black hole feeds energy back to its surroundings through winds and jets:
winds might accelerate star formation by compressing gas
winds might blow away all gas, stopping accretion and star formation
In any case, AGN are now thought to play a major role in galaxy evolution  AGN Feedback
(e.g., Cattaneo et al. 2009)
Star formation rate evolution
AGN evolution
cosmic down-sizing: activity stops earlier and more rapidly in more massive objects
Franceschini et al. 1999
March 24, 2010
P. Padovani SKA 2010 Meeting

6. Going fainter: a simple approach
Smallest flux density flim of a population depends on:
minimum radio power at redshift ≈ 0
maximum redshift [zmax]
luminosity evolution
Largest surface density of a population depends on:
number density at redshift ≈ 0 [NT]
density evolution; if none, then
N(>flim)/sr = NT x Volume(zmax)/4π
(1+z)3
March 24, 2010
P. Padovani SKA 2010 Meeting
z=4

7. March 24, 2010 P. Padovani SKA 2010 Meeting
How did I put these numbers together?
Z=6 assumed; for z=20 N(>f) does not change much (~ 2; RQ AGN & SFG ) while flux limits get smaller
~ 10 x (RQ AGN & SFG)
March 24, 2010
P. Padovani SKA 2010 Meeting

8. Going fainter: a more complex approach
Assume/derive luminosity function for the relevant populations
Assume/derive evolution
Assume cosmology
Calculate number counts:
Previous modelling work reaching flux densities < 1 μJy include Hopkins et al. (2000), Windhorst (2003), Jackson (2004), Jarvis & Rawlings (2004), and Wilman et al. (2008) [SKA Design Study]
March 24, 2010
P. Padovani SKA 2010 Meeting

9. star-forming galaxies
~ 87%
radio-quiet AGN
Fanaroff-Riley Is
star-forming galaxies
March 24, 2010
P. Padovani SKA 2010 Meeting
Wilman et al. 2008

10. Open Issues Extrapolations; samples used typically have:
flux densities much larger than those of interest (e.g., equiv. f1.4GHz ≈ 0.1 Jy for radio-loud AGN)
lower redshifts and higher powers
different frequencies (requires spectral assumptions)
Evolution is often not included self-consistently
Radio-quiet AGN predictions (excluded by older calculations!) rely on X-ray – radio power correlation
MHz  GHz for alpha_r ~ 0.7
March 24, 2010
P. Padovani SKA 2010 Meeting

11. A complementary approach
Visibility Function
Use a sample selected at lower flux densities (closer to those we are interested in): VLA-CDFS
Adopt evolution as derived from the sample
Calculate radio-quiet AGN counts from a radio-selected sample (no extrapolation from the X-ray band)
Avoid any assumptions (when possible)
 Simpler but closer to the data
We know that SFGs and RQ AGN will reach AT LEAST nanoJy fluxes! So what’s wrong?
Condon 1989
March 24, 2010
P. Padovani SKA 2010 Meeting

12. March 24, 2010 P. Padovani SKA 2010 Meeting
Give short summary: VLA-CDFS area small, therefore LFs do not go faint enough;
Need to complement them; Sadler et al. used NVSS
March 24, 2010
P. Padovani SKA 2010 Meeting

13. star-forming galaxies ~ 97%
~ 94%
March 24, 2010
P. Padovani SKA 2010 Meeting

14. (more) Caveats
Missing populations? (c.f. radio-quiet AGN until recently!):
Dwarf galaxies are very faint and very numerous: P ≤ W/Hz and Plim < W/Hz < [Leroy et al. 2005] and number densities ≈ 6 x those of non-dwarfs [Blanton et al. 2005]
P < W/Hz: (bright) ellipticals can reach P ≤ 1019 W/Hz (e.g., Sadler et al. 1989; Miller et al. 2009; Capetti et al. 2009), well below the AGN LF limits (both observed and used in modelling)
Recent radio background results (Fixsen et al. 2009, submitted): CDFS-based number counts make up only ≈ 1/3 of it (see also De Zotti et al and Massardi et al. 2010)
Faint-end of the LF (star-forming galaxies and AGN)
SFG LF (expected)
Stacked non-detections:
P ~ W/Hz
AGN expected LF
First part: 25’
Wilman et al. (2008) limit
VLA-CDFS limit
March 24, 2010
P. Padovani SKA 2010 Meeting
Miller et al. 2009

15. Source Identification
VLA-CDFS sample
median R = 22.8
(and 9 “empty fields”)
Mignoli et al. (2009)
Star-forming galaxy (blue)
~ 50% of our sources have spectroscopic redshift, ~ 80% spectro or photoz
Quiescent elliptical (red)
March 24, 2010
P. Padovani SKA 2010 Meeting

16. Selecting star-forming galaxies (SFG) candidates
SFGs (at z ≈ 1) are typically NOT hosted by ellipticals or lenticulars  optical morphology
SFGs have relatively low radio powers [locally Pr < 1024 W/Hz]: Pr < W/Hz  spectra or photometric redshifts
SFGs have relatively low radio-to-optical luminosity ratios [locally R = log(Pr/Lopt) < 1.4]: R < 1.7  optical data
SFGs have relatively low X-ray powers: Lx < 1042 erg/s  X-ray data (vital to separate SFGs from AGN)
X-rays vital to separate SFG from AGN
March 24, 2010
P. Padovani SKA 2010 Meeting

17. Selecting star-forming galaxies (SFG) candidates
AGN
The identification of faint radio sources requires a wealth of multi-wavelength data!
star forming galaxies
“old” galaxies
PAH: polycyclic aromatic hydrocarbons
Complete sample: SFG: +5, -6; AGN: +6, - 5
Sajina et al. 2005
(also: Lacy et al. 2004,
Stern et al. 2005, and
others)
March 24, 2010
P. Padovani SKA 2010 Meeting

18. The multiwavelength properties of micro-Jy and nano-Jy sources: X-rays
Same idea: f_band vs. f_r with expected fluxes for faint radio sources
Assume low-power E’  E’s and dwarf irr.  SFGs (dwarf E’s  E’s?)
And AGN are going to be only a small fraction (~ 1-2%) of total number of sources at micro-Jy levels;
Other way around: Most of the 10 million sources in WFXT Wide Survey will have an SKA counterpart!
March 24, 2010
P. Padovani SKA 2010 Meeting

19. The multiwavelength properties of micro-Jy and nano-Jy sources: optical band
E-ELT: extended source; UDF t_exp in z band ~ 4 days; JWST: point source (!) but 2.8h only; 1 micro-Jy “all-sky survey”: 2 10^9 sources
~ 60% SFG should be det. by LSST 10yr survey at 1 micro Jy; ~ 1/3 at 100 nJy
Most LSST sources, apart from the most radio-quiet ones, will have an SKA counterpart
March 24, 2010
P. Padovani SKA 2010 Meeting

20. The multiwavelength properties of micro-Jy and nano-Jy sources: spectra
~ 55% of the SFG with f > 1 micro Jy will have R > 26 (65% R > 25)  < 50% of the SFG will be within reach of 8/10 m telescopes
~ 80% of the sources with f > 100 nJy will have R > 26 (90% R > 25)  < 20% of the SFG will be …
March 24, 2010
P. Padovani SKA 2010 Meeting

21. The multiwavelength properties of micro-Jy and nano-Jy sources: infrared band
SPICA: SPace Infrared telescope for Cosmology and Astrophysics
Diffraction limited: angular resolution ~ 2” at 24 micron
March 24, 2010
P. Padovani SKA 2010 Meeting

22. Summary
Q. Which type of sources will SKA see at micro-Jy and nano-Jy flux densities?
A. SFGs, FR Is, and radio-quiet AGN; plus dwarf galaxies and low-power (P < 1021 W/Hz) ellipticals, which are not included in current models
Q. How many of them and how faint?
A. At least 105/deg2 above 1 μJy and 7 105/deg2 above 100 nJy. Dwarfs and SFGs should be the most numerous classes, followed by low-power ellipticals (reaching, with radio-quiet AGN, below 0.1 nJy)
March 24, 2010
P. Padovani SKA 2010 Meeting

23. Summary (continued)
Q. Which multi-wavelength properties will they have?
X-ray: very faint!
Optical imaging: most > 1 μJy sources should be detected by LSST (“half-sky” survey); (some) fainter sources should be detectable with long exposures by JWST/E-ELT (small areas)
Optical spectra/redshifts: most > 1 μJy sources will be beyond the capabilities of 8/10 m telescopes (photo-z’s with LSST?); some of them will be within reach of the E-ELT
Near Infrared: most > 1 μJy sources will be below the limits of all-sky infrared surveys; SFGs will be detectable with JWST and especially with SPICA (down to 100 nJy)
The bulk of radio-quiet AGN above 1 μJy will be detected by IXO (small
area) [but small fraction of total population!]; no hope for SFGs (dwarfs?),
FR Is (low-power E’s?), and fainter radio-quiet AGN
imaging: above 1 μJy more than 50% of SFGs (dwarfs?) and radio-quiet AGN
should be detected by LSST (“half-sky” survey); the rest, with FR Is (low-power
E’s) should be detectable with long exposures by JWST/E-ELT (small areas).
Below 1 μJy optical detection will be increasingly harder
> 50% (80%) of the > 1 μJy (> 100 nJy) population will be beyond thecapabilities
of 8/10 m telescopes (photo-z’s with LSST?); some of them will be within
reach of the E-ELT
March 24, 2010
P. Padovani SKA 2010 Meeting

24. Summary (continued) Q. How will we identify them? A.
On large areas of the sky SKA will be quite “alone” in the multi-wavelength arena (with the exception of the optical band and down to ≈ 1 μJy). However, most all-sky survey sources will have an SKA counterpart!
JWST, the E-ELT, and SPICA will match the SKA but only above ≈ μJy and on small areas. At the 10 nano-Jy level it’s going to be pretty lonely …
 need to use as much radio information (HI redshifts, size, morphology, spectral index, etc.) as possible (but I think it won’t be enough …)!
March 24, 2010
P. Padovani SKA 2010 Meeting