Mid-IR Color-Color Diagrams

Mid-infrared colors have been used to identify the presence of AGN-heated dust in the spectral energy distributions of galaxies. We organize AGN mid-IR diagnostics according to the instrument/bandpasses used below.

Physical Motivation

Donley et al (2012) provide us with a very useful illustration that shows spectral energy distributions of galaxies with increasing contribution from AGN (from 0% to 95%) for both obscured and unobscured quasar components. In the obscured case, a dust attenuation of A_V=2 mag was applied to the QSO template. In both panels of Figure 1 (unobscured and obscured QSO), one can clearly see the excess mid-infrared contribution at 3-10 μm, which can be probed with, e.g., Spitzer/IRAC and WISE observations. In the optically-obscured case, the UV and optical emission is very suppressed and therefore the accretion disk would not be observable directly as a UV excess or bright point source in the optical. Some of these systems may still be picked up through their narrow emission line signatures and/or X-ray emission, while other more extremely buried cases may be elusive at all other bands.

IRAC/MIPS Color Selections

A few methods were developed based on Spitzer/IRAC photometry.

1) The "Lacy wedge" (Lacy et al. 2004, 2007; Sajina at al. 2005)

log(S_5.8/S_3.6) > -0.1

log(S_8.0/S_4.5) > -0.2

log(S_8.0/S_4.5) > 0.8 * log(S_5.8/S_3.6) + 0.5

From Lacy et al (2007): "The log(S_5.8/S_3.6) cut was moved by +0.1 compared to [Lacy et al (2004)]. The exact position of this cut is not critical, and it was felt that shifting this would help remove non-AGN contaminants, particularly as the final 5.8 micron fluxes in the XFLS catalog were increased by 10% in the final version of the XFLS catalog relative to the catalog used [by Lacy et al (2004)]."

2) The "Stern wedge" (Stern et al. 2005)

ch1 = [3.6]

ch2 = [4.5]

ch3 = [5.8]

ch4 = [8.0]

Vega Magnitudes:

(ch3 - ch4) > 0.6

(ch1 - ch2) > 0.2 * (ch3 - ch4) + 0.18

(ch1 - ch2) > 2.5 * (ch3 - ch4) - 3.5

3) IRAC Power-Law (Alonso-Herrero et al. 2006, Donley et al. 2007, 2012)

Alonso-Herrero et al (2006) and Donley et al (2007) both fit a power-law between 3.6 and 8 micron and minimized to select galaxies whose IRAC SEDs followed a power law with spectral index < -0.5. Donley et al. (2007) define a good fit with "a cut in the probability of P>0.1", where "P is the probability that a fit to a power-law distribution would yield a value greater than or equal to the observed ; a probability of 0.5 corresponds to a reduced chi² of 1."

The revised AGN selection criteria from Donley et al. (2012) are defined, for x=log(f5.8/f3.6) and y=log(f8.0/f4.5), as follows:

x > 0.08

y > 0.15

y > (1.12 * x)-0.27

y < (1.12 * x)+0.27

f4.5 > f3.6 && f5.8 > f4.5 && f8.0 > f5.8

4) KI and KIM methods (Messias et al. 2012; also see Messias et al. 2013)

KI = K_s and IRAC bands

The KI criteria are: Ks-[4.5]>0 and [4.5]−[8.0]>0

KIM = K_s, IRAC, MIPS bands

The IM criteria are: [8.0]-[24]>-2.9x([4.5]-[8.0])+2.8 and [8.0]-[24]>0.5

See figures below, and the captions from the original article. These authors also derived color vs. redshift tracks and AGN diagnostics, as detailed below.

5) IRAC color vs. redshift

The advantage of the color-color selections described above is that they require no redshift information. However, when available, the redshift can help break degeneracies as a given SED will make a redshift track that may loop back onto itself within the color-color space. Namely, some high-redshift star-forming galaxies start entering the Lacy and Stern wedges at higher redshifts (especially z>1). In this section, we list some examples of mid-IR color diagnostics that include the redshift information.

5.1) Juneau et al. 2013 (J13)

J13 compared redshift tracks of the IRAC ch1-ch4 color for infrared luminous star-forming galaxies (yellow tracks); dust-free stellar population models (purple tracks) and various AGNs (black lines), including Compton-thick NGC 6240. The have opted to use the most extreme SF galaxy track (IRAS 12112; in cyan) as a dividing line between star-forming galaxies (below) and AGN (above). From the SEDs, this method appears to work well at z>0.5, but may be incomplete with respect to the most extremely obscured cases such as NGC 6240. Also see Figures 2 & 3 from Stern et al 2005 for redshift tracks of various SED types and their IRAC colors.

5.2) Messias et al. 2012 (also see Messias et al. 2013)

When secure redshifts are available, the following colors can be used:

Ks-[4.5]>0 at z<1

[4.5]−[8.0]>0 at 1.1<z<2.5

[8.0]−[24] at z > 2.5–3.

In the absence of a secure redshift, the authors recommend using either the "KI" or "KIM" color-color (see above).

WISE Color Selections

1) Stern et al. (2012; Assef et al. 2013):

"Analysis of Figure 6 suggests that a color cut at W1 – W2 = 0.8 offers an extremely robust AGN sample which is still highly complete. For some uses, a slightly less conservative color cut at W1 – W2 = 0.7 might be preferable, providing a powerful compromise between completeness and reliability for WISE AGN selection."

2) Mateos et al. (2012):

Selection criteria calibrated with the Bright Ultrahard XMM-Newton Survey (BUXS).

y = 0.315 * x

where x ≡ log₁₀(f_12um/f_4.6um) and y ≡ log₁₀(f_4.6um/f_3.4um)

The top and bottom boundaries of the wedge are obtained by adding y-axis intercepts of +0.297 and −0.110, respectively. The MIR power-law α=−0.3 bottom-left limit corresponds to: y = −3.172 * x + 0.436

Completeness and Contamination

Any selection method suffers from incompleteness and contamination, and we usually need to make a trade-off between the two. The emphasis on completeness or purity will depend on the details of the scientific question. We can obtain a good indication of the completeness by measuring the recovery rate of known AGNs. Here, we show one such example work by Mateos et al (2012), where the authors used a sample of Bright Ultrahard X-ray AGNs and computed the recovery through WISE colors separating Type 1 (optically unobscured) and Type 2 (optically obscured) subsets from the parent X-ray sample.