AGN Identification and Host Galaxy

Properties in the MOSDEF Survey

Alison CoilUCSD

Collaborators: James Aird (Cambridge), Mojegan Azadi (UCSD), Gene Leung (UCSD), Alexander Mendez (JHU)

+ Alice Shapley, Naveen Reddy, Mariska Kriek, Brian Siana, Bahram Mobasher (MOSDEF co-PIs)

AGN Selection Biases

All AGN selection techniques have biases:

Need to understand the selection biases for AGN identification at different wavelengths!

Mendez, Coil et al. (2016)

number

stellar mass sSFR

galaxiesX-ray AGNIR AGNradio AGN

PRIMUS survey

IR vs X-ray AGN Properties

• IR-AGN selection relies on the AGN being bright relative to the galaxy • identifies more luminous AGN • in lower luminosity (lower mass,

ie, star forming) galaxies

• X-ray AGN samples span a wide range of specific accretion rate (ie, Eddington ratio), while IR AGN samples are high specific accretion rate only [these are observed distributions, not corrected for depth]

specific accretion rate (LAGN/LEdd)

The Astrophysical Journal, 748:142 (22pp), 2012 April 1 Donley et al.

Figure 1. Composite SEDs constructed using the QSO1 and M82 templates of Polletta et al. (2008), scaled to give 1–10 µm AGN contributions of 0% (red in theonline journal) to 95% (purple in the online journal). The final SEDs have been normalized at 1.6 µm. In the lower panel, we apply an extinction of AV = 2 to theQSO1 SED using the Draine (2003) extinction law. The four IRAC bands at z = 0 are shaded. While luminous unobscured and obscured AGNs have very differentUV-optical SEDs, luminous AGNs should display a red MIR power-law SED regardless of obscuration.(A color version of this figure is available in the online journal.)

Figure 2. Predicted z = 0–3 IRAC colors of AGN/galaxy composite SEDs in Lacy et al. (2004, 2007) color space, where the AGN fraction is defined between1 and 10 µm. The star-forming templates represent the ULIRG IRAS 22491 (square; Polletta et al. 2008), the starburst M82 (star; Polletta et al. 2008), a normalstar-forming spiral galaxy (triangle; Dale & Helou 2002), and an elliptical galaxy (circle; Polletta et al. 2008), where large symbols mark each family of purelystar-forming templates at z = 0. The AGN template is the QSO1 template of Polletta et al. (2008). Additional extinctions of AV = 0–2 and AV = 0–20 are applied tothe star-forming and AGN components, respectively. The wedge is the AGN selection region of Lacy et al. (2007), and the line represents the power-law locus fromα = −0.5 (lower left) to α = −3.0 (upper right). While the templates of purely star-forming galaxies avoid the power-law locus, they enter the current AGN selectionregion at both low and high redshifts. As the AGN’s contribution to the MIR emission increases, the SEDs move inward and redward toward the power-law locus.(A color version of this figure is available in the online journal.)

3

The Astrophysical Journal, 748:142 (22pp), 2012 April 1 Donley et al.

Figure 1. Composite SEDs constructed using the QSO1 and M82 templates of Polletta et al. (2008), scaled to give 1–10 µm AGN contributions of 0% (red in theonline journal) to 95% (purple in the online journal). The final SEDs have been normalized at 1.6 µm. In the lower panel, we apply an extinction of AV = 2 to theQSO1 SED using the Draine (2003) extinction law. The four IRAC bands at z = 0 are shaded. While luminous unobscured and obscured AGNs have very differentUV-optical SEDs, luminous AGNs should display a red MIR power-law SED regardless of obscuration.(A color version of this figure is available in the online journal.)

Figure 2. Predicted z = 0–3 IRAC colors of AGN/galaxy composite SEDs in Lacy et al. (2004, 2007) color space, where the AGN fraction is defined between1 and 10 µm. The star-forming templates represent the ULIRG IRAS 22491 (square; Polletta et al. 2008), the starburst M82 (star; Polletta et al. 2008), a normalstar-forming spiral galaxy (triangle; Dale & Helou 2002), and an elliptical galaxy (circle; Polletta et al. 2008), where large symbols mark each family of purelystar-forming templates at z = 0. The AGN template is the QSO1 template of Polletta et al. (2008). Additional extinctions of AV = 0–2 and AV = 0–20 are applied tothe star-forming and AGN components, respectively. The wedge is the AGN selection region of Lacy et al. (2007), and the line represents the power-law locus fromα = −0.5 (lower left) to α = −3.0 (upper right). While the templates of purely star-forming galaxies avoid the power-law locus, they enter the current AGN selectionregion at both low and high redshifts. As the AGN’s contribution to the MIR emission increases, the SEDs move inward and redward toward the power-law locus.(A color version of this figure is available in the online journal.)

3

100% galaxy

95% AGN

Donley et al. 2012, ApJ

X-ray AGNIR AGN

Mendez, Coil et al. (2013)

AGN Clustering Differences

Mendez, Coil et al. (2016)

AGN selected at different wavelengths have different clustering properties: radio and X-ray AGN are more clustered than IR AGN

radioIRX-ray

PRIMUS survey

AGN Clustering Explained!

Mendez, Coil et al. (2016)This is due entirely to differences in their host galaxy populations -

matching stellar mass and SFR of hosts makes differences disappear.Galaxies of a given stellar mass and SFR have the same clustering

properties whether they host an AGN or not.

radioIRX-ray

MOSDEF survey

Spectroscopic survey at 1.5 < z < 3.5 using MOSFIRE on Keck4 year survey, 48 nights total, finished taking data last monthFull sample has ~1500 galaxies + AGN

Targeting CANDELS fields, H-band selected (depth=24.5)Sample spans a wide range of stellar mass and SFR

Kriek et al. (2015)

Obtain full suite of rest-frame optical emission lines:

Data from first 1/2 of the survey.

We are using the Melendez et al. (2014) line to identify optical AGN.

Less contaminated than Kauffmann, more complete than Kewley.

BPT Diagram at z~2

Azadi, Coil, et al. in prep.

Observed overlap of AGN identified at different wavelengths in MOSDEF

(not corrected for depth!)

Optical / IR / X-ray AGN Properties

λOIII ~ Lbolometric /stellar mass

• Similar LOIII distributions; same OIII flux limit• Bias towards higher stellar mass hosts at all wavelengths; bias against

most massive hosts for IR AGN• Optical and X-ray AGN have somewhat lower specific accretion rates

than IR AGN

Azadi, Coil, et al. in prep.

Optical / IR / X-ray Host Properties

9 10 11 12Log [M*/MΟ •]

-2

0

2

4

Log

SFR

[MΟ • y

r-1]

0.0 0.2 0.4 0.6 0.8 1.0

Median 1.24 1.50 1.08 1.29

X-ray AGNIR AGNOptical AGNGalaxies

• At z~2, AGN hosts span a wide range of SFR at a given stellar mass.• IR and X-ray AGN hosts span the full range of SFR, while optical

AGN hosts tend to lie below the main sequence of star formation.

Azadi, Coil, et al. in prep.

• We don’t find a significant correlation b/w SFR or M* and either LX or LOIII.

Optical / IR / X-ray Host Properties

• UVJ space separates star forming (dusty vs non-dusty) and quiescent galaxies.

• Showing a mass-matched galaxy sample with contours.

• AGN hosts are just as likely to be in dusty vs. non-dusty star forming hosts as inactive galaxies of the same stellar mass.

• Have to do a careful matching of stellar mass to get this result!

Azadi, Coil, et al. in prep.

AGN Selection Biases

IR X-ray opticalAGN

accretion rate

stellar mass

SFR

dust less dust no bias more dust low dust

higher SFR no bias lower SFR lowest SFR

intermediate high high very highmass mass mass mass

higher rate no bias no bias low rate

radio

AGN Selection Biases

IR X-ray opticalAGN

accretion rate

stellar mass

SFR higher SFR no bias lower SFR lowest SFR

intermediate high high very highmass mass mass mass

higher rate no bias no bias low rate

radio

Once these biases are taken into account, within the MOSDEF sample we do not find any significant differences between the properties of galaxies with and without AGN.

z~2 AGN Outflow Identification

blueshifted outflows in

[NII], Ha, [OIII], Hb

Hα S/N

num

ber• The detected AGN outflow rate is 18%

- this is a lower limit on actual outflow rate, as need high S/N spectra and/or fast outflow to detect.

Leung, Coil, et al. in prep.

Outflow Kinematics and Sizes

Spatial extent from 2d spectra:

~1/2 are resolved, FWHM of physical extent is ~3-10 kpc~1/2 are spatially offset from NLR, with a max. offset of ~7 kpc

NLoutflow

• velocity of outflow component ~300-1300 km/s• FWHM ~100-1300 km/s

Leung, Coil, et al. in prep.

Conclusions

• Understanding AGN selection biases is very important!

• Biases are different for selection at different wavelengths.

• X-rays are least biased, but limited to small fields with varying depth.

• Impacts interpretation of host galaxy demographics, AGN accretion rate distributions, AGN clustering and environments, etc.

• z~2 AGN host galaxies have similar properties as inactive galaxies, once selection biases are taken into account.

• Moderate luminosity AGN appear to commonly drive fast, galaxy-wide outflows at z~2.