AGN AND THE CHARACTERISTICS OF THEIR HOST GALAXIES

by

William McAlpine A Dissertation Submitted to the Graduate Faculty of George Mason University In Partial fulfillment of The Requirements for the Degree of Doctor of Philosophy Physics

Committee:

Dr. Shobita Satyapal, Dissertation Director

Dr. Jessica L. Rosenberg, Committee Member

Dr. Mario Gliozzi, Committee Member

Dr. Rebecca Goldin, Committee Member

Dr. Michael Summers, Director, School of Physics, Astronomy, and Computational Sciences

Dr. Richard Diecchio, Interim Associate Dean for Student and Academic Affairs, College of Science

Dr. Peggy Agouris, Interim Dean, College of Science

Date: Fall Semester 2013 George Mason University Fairfax, VA AGN and the Characteristics of their Host Galaxies

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy at George Mason University

By

William McAlpine Bachelor of Science George Mason University, 2009

Director: Dr. Shobita Satyapal, Professor School of Physics, Astronomy, and Computational Sciences

Fall Semester 2013 George Mason University Fairfax, VA Acknowledgments

First I would like to thank my advisor, Shobita Satyapal, for her encouragement and sup- port throughout my graduate studies. Her excitement for science is contagious and inspired me to complete this work. I would also like to thank the other members of my disserta- tion committee, Jessica Rosenberg, Mario Gliozzi, and Rebecca Goldin for their time and encouragement.

ii Table of Contents

Page List of Tables ...... v List of Figures ...... vi Abstract ...... vii 1 Introduction ...... 1 2 Black Holes in Bulgeless Galaxies ...... 3 2.1 Introduction ...... 3 2.2 Observations and Data Reduction Procedure ...... 6 2.3 Results ...... 7 2.3.1 X-Ray Spectral Analysis ...... 7 2.3.2 X-Ray Temporal Analysis ...... 9 2.3.3 UV - Optical Data ...... 10 2.4 Previous Observations of NGC 3367 and NGC 4536 ...... 11 2.5 Contamination by X-Ray Binaries ...... 14 2.6 Bolometric Luminosities ...... 16 2.7 Black Hole Mass Estimation Methods ...... 17 2.8 Summary ...... 19 3 Optically Hidden AGNs in non-U/LIRG Interacting Galaxies ...... 21 3.1 Introduction ...... 21 3.2 The Sample ...... 24 3.2.1 The non-U/LIRG pairs sample ...... 24 3.2.2 Comparison Samples: The SINGS and ULIRG Sample ...... 26 3.2.3 Sample Properties and Potential Selection Biases ...... 31 3.3 Observations and Data Reduction ...... 31 3.4 Incidence of AGNs in Galaxy Pairs ...... 34 3.4.1 Comparison with Isolated and ULIRG Samples ...... 34 3.5 Which Galaxies Host -Identified AGNs? ...... 36 3.6 What are the AGN Properties and how do they Depend on Merger Stage? . 37 3.6.1 How do Mid-IR Identified AGNs Compare to Optically Identified AGNs? 39

iii 3.7 Mid-IR Identified Dual AGN ...... 42 3.8 Summary and Conclusions ...... 44 4 WISE Study of AGN in Galaxy Pairs ...... 45 4.1 Introduction ...... 45 4.2 Sample Selection ...... 50 4.3 Results ...... 51 4.3.1 Selection of AGN by WISE ...... 51 4.3.2 Optical Selection of AGN by SDSS ...... 52 4.3.3 The Effect of AGN Luminosity on AGN Selection by WISE . . . . . 54 4.3.4 AGN Fraction in Galaxy Zoo Mergers ...... 55 4.4 Other Causes of Red WISE Colors ...... 58 4.5 Reliability of Photometry in Close Pairs ...... 58 4.6 Conclusion ...... 58 Bibliography ...... 60

iv List of Tables

Table Page 2.1 Observation Log ...... 7 2.2 Best Fit Model Results for PN Data ...... 7 2.3 Black Hole Mass Estimates ...... 17 3.1 Sample Properties ...... 29 3.2 Fluxes for Full Sample ...... 32 4.1 WISE AGN fraction by Galaxy Zoo merger vote fraction ...... 57 4.2 Comparison between average W1-W2 color and angular separation for the projected pairs sample ...... 59

v List of Figures

Figure Page 2.1 Images of NGC 4536 and NGC 3367 ...... 5 2.2 Best fit models and data-to-model ratios ...... 9 2.3 XMM-Newton EPIC light curves for NGC 3367 and NGC 4536 ...... 10

2.4 αOX plotted versus 2500 A˚ monochromatic luminosity ...... 12 3.1 DSS images of the non-U/LIRG pairs ...... 27 3.2 The properties of the non-U/LIRGs pair sample, the isolated SINGS sample, and the ULIRG sample ...... 28 3.3 IRS spectra showing the [NeV] 14.3 µm line ...... 35 3.4 Distribution of 3σ limiting [NeV] line sensitivities for our pairs sample . . . 36 3.5 Properties of galaxies with [NeV] detections ...... 38 3.6 Distribution of merger stages for galaxies with [NeV] detections ...... 39 3.7 Distribution of [NeV] luminosities ...... 40 3.8 [NeV] luminosity vs. pair separation ...... 40 3.9 Comparison of the properties of mid-IR identified AGN and optically classi- fied Seyferts with [NeV] detections...... 41 4.1 Composite SEDs with varying AGN contribution ...... 48 4.2 WISE colors of AGN as a function of with increasing amounts of contamination from the host galaxy ...... 49 4.3 Fraction of galaxies that have a color of W1-W2≥0.5 ...... 52 4.4 BPT diagram of the galaxy pair sample with the demarcation line of AGN and star-forming galaxies ...... 53 4.5 Fraction of galaxies that are classified as an optical AGN according to the classification scheme of Kewley et al. (2001)...... 53 4.6 WISE color-color plot of hard X-ray selected BAT AGN ...... 55 4.7 [OIII] luminosities of optically selected AGN in the pairs sample ...... 56

vi Abstract

AGN AND THE CHARACTERISTICS OF THEIR HOST GALAXIES William McAlpine, PhD George Mason University, 2013 Dissertation Director: Dr. Shobita Satyapal

The vast majority of optically identified active galactic nuclei (AGNs) in the local uni- verse reside in host galaxies with prominent bulges, supporting the hypothesis that black hole formation and growth is fundamentally connected to the buildup of galaxy bulges. However, recent mid-infrared spectroscopic studies with Spitzer of a sample of bulgeless galaxies reveal strong evidence for AGNs in these galaxies. We present follow-up X-ray ob- servations recently obtained with XMM-Newton of two such sources, the late-type optically normal galaxies NGC 3367 and NGC 4536. Detailed spectral analysis reveals that for both galaxies, the 2-10 keV emission is consistent with low-luminosity AGNs. These observa- tions therefore add to the growing evidence that the fraction of late-type galaxies hosting AGNs is significantly underestimated using optical observations alone. A comparison of the mid-infrared [NeV] luminosity and the X-ray luminosities suggests the presence of a highly absorbed X-ray source in both galaxies. Next, a mid-infrared spectroscopic study was conducted on a sample of 30 pairs of interacting galaxies with archival Spitzer high-resolution spectroscopic observations. Pre- vious studies of galaxy interactions have focused primarily on luminous galaxies, or were conducted in the optical or the UV where emission from the AGN may be hidden. Based on the detection of the [NeV] 14.3 µm emission line we find an AGN detection rate of 18%, including 4 optically unidentified AGN. Our study therefore reveals that optical studies miss AGNs even in non-U/LIRG interactions. Furthermore, we find that these AGNs are found in both disturbed and undisturbed hosts with a wide range of pair separations and

Hubble types. Combining our study with previously published studies of U/LIRGs, we find that the AGN detection rate is highest in ULIRGs. In addition, we find tentative evidence that the incidence of AGNs is bimodal, with merger stage, with the highest incidence found in the earliest and latest merger stages. Finally, we find evidence based on mid-infrared spectroscopy for three possible dual AGNs, suggesting that a significant population of dual AGNs is hidden in the optical. The study of a larger sample of interacting galaxies is important to understand the relation between the incidence of AGN and the interaction stage of merging galaxies. Using infrared observations from the Wide-field Infrared Survey Explorer (WISE) we find that galaxy pairs have a higher AGN fraction than isolated galaxies and the AGN fraction increases with decreasing pair separation. While WISE only detects AGNs that dominate the emission of their host galaxies, it is able to detect obscured AGN. WISE finds a similar trend with AGN fraction and pair separation as optical studies, however WISE selects a different population of AGN since only a third of the optically selected AGN are categorized as AGN by WISE. Chapter 1: Introduction

We now know that supermassive black holes lurk in the centers of most bulge-dominated galaxies in the local Universe and that their black hole masses and the stellar velocity dispersions, σ∗, of their host galaxy’s bulge are strongly correlated (Gebhardt et al. 2000,

Ferrarese & Merritt 2000). This discovery has launched numerous speculations that the formation and evolution of galaxies and supermassive black holes are fundamentally linked, and that perhaps the process that builds galaxy bulges is coupled to the growth of the central black hole. The nature of this connection and the detailed physics behind it is currently unknown and remains at the forefront of extragalactic astrophysics today.

According to the current paradigm of hierarchical structure formation, major mergers are ubiquitous and are a crucial element in the assembly of present-day galaxy bulges

(e.g., Kaufman et al. 1993). Furthermore, numerical simulations predict that gravitational encounters between galaxies induce gravitational torques on the gas that efficiently drives the gas to the nuclear regions of the host galaxies (e.g. Mihos & Hernquist 1996, Cox et al.

2008), where it can fuel both and accretion onto the central black hole. It is therefore natural to assume that major merging between galaxies grows galaxy bulges and simultaneously triggers accretion onto the . Indeed, prior to our work, the vast majority of actively growing black holes - i.e., active galactic nuclei (AGN)

- in the local Universe were found in galaxies with prominent bulges, and only 1 galaxy without an evidence of a bulge in the local Universe was confirmed to host an AGN. On the other hand, extensive optical studies of interacting galaxies show little or no evidence of enhanced AGN activity (e.g. Bushouse et al. 1996, Ellison et al. 2008, Darg et al.

2009). However, these studies were based on optical spectroscopic observations, which can be severely limited in the study of bulgeless or interacting galaxies, where a putative AGN 1 is likely to be both energetically weak and deeply embedded in the center of a dusty and possibly tidally disturbed host. In such systems, the traditional optical emission lines used to identify AGN can be dominated by emission from star formation regions, in addition to being significantly attenuated by dust in the host galaxy. As a result, it is by no means clear what fraction of late-type or interacting galaxies host AGN, and whether the purported connection between merging and black hole growth holds. Therefore, some key fundamental questions on the connection between black holes and galaxy formation and evolution have yet to be answered, such as:

• Is the bulge a necessary ingredient for a black hole to grow?

• Do black holes ever form in purely bulgeless galaxies?

• If so, how does the black hole mass and accretion rate relate to the properties of the

parent galaxy?

• Are they related in any way to the mass of the disk?

• Which forms first in a galaxy, the black hole or the bulge?

• Are interactions necessary for black hole formation and growth?

2 Chapter 2: Black Holes in Bulgeless Galaxies

2.1 Introduction

The discovery that at the heart of virtually all early-type galaxies in the local Universe lies a massive nuclear black hole strongly suggests that black holes play a pivotal role in the formation and evolution of galaxies. The well-known correlation between the black hole mass, MBH, and the host galaxy stellar velocity dispersion σ⋆ (Gebhardt et al. 2000; Ferrarese & Merritt 2000) suggests an intimate connection between black hole growth and the build-up of galaxy bulges, a connection that is reinforced by the fact that the vast majority of optically-identified active galactic nuclei (AGN) are found in early type hosts

(e.g., Heckman 1980; Ho et al. 1997, henceforth H97).

However, a number of recent studies have now shown that determining if AGNs reside in low-bulge galaxies cannot be definitively answered with only optical observations (e.g.,

Satyapal et al. 2007, 2008, 2009; Ghosh et al. 2008; Desroches & Ho 2009). The problem arises because a putative AGN in a galaxy with a minimal bulge is likely to be energetically weak and deeply embedded in the center of a dusty late-type spiral. As a result, the optical emission lines can be dominated by the emission from star forming regions, severely limiting the diagnostic power of optical surveys in determining the incidence of AGNs in low-bulge systems. In a recent mid-infrared spectroscopic study with Spitzer of a sample of optically

“normal” late-type galaxies, we found remarkably the presence of high-ionization [NeV] lines in a significant number of sources, providing strong evidence for AGNs in these galaxies, and suggesting that the AGN detection rate in late-type (Sbc or later) galaxies is possibly more than 4 times larger than what optical spectroscopic studies alone indicate (Satyapal et al.

2008). While the detection of a [NeV] line strongly suggests that these galaxies harbor an

3 AGN, [NeV] emission can originate in shock-heated gas produced by starburst-driven winds

(Contini 1997). X-ray detection would provide corroborating evidence and strengthen the case for the presence of an AGN based on the luminosity and spectrum of the source.

X-ray observations arguably represent one of the most effective means to confirm the existence of an AGN and to investigate the AGN properties since: 1) Unlike the IR emission line features that probe lower density gas at larger distances from the black hole, X-rays are produced (and reprocessed) in the inner hottest nuclear regions where accretion occurs.

2) The penetrating power of (hard) X-rays allows them to carry information from the inner core of the galaxy without being significantly affected by absorption, and to probe the

24 −2 presence of the putative torus (provided that NH < 10 cm ). 3) X-rays are far less affected by the host galaxy contamination than optical radiation. 4) The hard X-rays (2-10 keV) can more robustly constrain the bolometric luminosity of the AGN.

Here we present a pilot study with XMM-Newton observations of NGC 3367 (D=43.6Mpc;

Tully 1988) and NGC 4536 (D=17.7Mpc; Saha et al. 2006) which are two of the late-type galaxies in our Spitzer sample (Satyapal et al. 2008) showing robust evidence for a [NeV] line. The distance of NGC 4536 determined by use of Cepheids, is significantly higher than reported by Tully (1988), yielding larger luminosities in all bands which are more consis- tent with an AGN. Among the galaxies with [NeV] detections, these 2 galaxies have optical spectra in the extreme HII-region range, indicating that there is absolutely no hint of an

AGN based on their optical spectra. Both galaxies are isolated very low bulge systems; deep HST imaging of NGC 4536 for example, shows no evidence for a classical bulge but instead reveals a surface brightness profile consistent with a pseudo-bulge that exhibits spiral structure (Fisher 2006a). The presence of AGNs in these systems is indeed highly surprising.

4 GALEX XMM 0.3-2 keV

3 arcmin 3 arcmin

GALEX XMM 0.3-2 keV

1 arcmin 1 arcmin

Figure 2.1: Images of NGC 4536 (top two) and NGC 3367 (bottom two) from GALEX in the near-ultraviolet (from Gil de Paz et al. 2007) and smoothed XMM in the 0.3-2 keV energy range. Both panels show the smoothed XMM 2-10 keV image overlaid as contours with levels starting at 0.25 counts (smoothed), incremented in 0.25 steps. The smoothing of the XMM images used a Gaussian with kernel of radius = 4.8” (3 pixels).

5 2.2 Observations and Data Reduction Procedure

We observed our 2 targets with XMM-Newton in June, 2008. The nominal durations range between 27 and 34 ks (NGC 3367 - MOS: 34ks, PN: 32ks; NGC 4536 - MOS: 28ks, PN:

27ks). All of the EPIC cameras (Str¨uderet al. 2001; Turner et al. 2001) were operated in full-frame mode with the medium filter because of the presence of bright nearby sources in the field of view. As a precaution, for NGC 3367 and NGC 4536 the MOS cameras were operated in small window mode to prevent photon pile-up. The recorded events were screened to remove known hot pixels and other data flagged as bad; only data with FLAG=0 were used.

The data were processed using the latest CCD gain values. For the temporal and spectral analysis, events corresponding to pattern 0–12 (singles, doubles, triples, and quadruples) in the MOS cameras and 0–4 (singles and doubles only, since the pn pixels are larger) in the pn camera were accepted. Arf and rmf files were created with the XMM-Newton

Science Analysis Software (SAS) 7.1. Investigation of the full–field light curves revealed the presence of several background flares for NGC 3367 and NGC 4536. These time intervals were excluded, reducing the effective total exposures time to the values reported in Table

2.1. The nuclear X-ray sources in NGC 3367 and NGC 4536 are well-separated from other

field X-ray sources in the X-ray images (Figure 2.1) and extraction radii used for source spectra and light curves are 30′′ in both cases. For reference, 10′′ correspond to 2.11 kpc for

NGC 3367, and 0.89 kpc for NGC 4536. The derived X-ray positions for the central sources are 10h46m35s + 13d44m59s and 12h34m27s + 02d11m16s for NGC 3367 and NGC 4536, respectively. These positions are consistent with the positions of the optical nuclei taken from NED based on SDSS images (10h46m34.s954 + 13d45m03.s09 and 12h34m27.s050 +

02d11m17.s29 for NGC 3367 and NGC 4536, respectively). There are also off-nuclear X-ray sources seen in Figure 2.1 which are not investigated here. Background spectra and light curves were extracted from source-free circular regions on the same chip as the source, with extraction radii ∼2 times larger than those used for the source. There are no signs of pile-up

6 Table 2.1: Observation Log

Source Date PN exposure PN rate [dd/mm/yyyy] [ks] [s−1] NGC 3367 06/16/2008 17.8 0.100  0.003 NGC 4536 06/17/2008 14.4 0.008  0.003

Table 2.2: Best Fit Model Results for PN Data

Name Power Law Thermal Intrinsic 0.3-2 keV 2-10 keV Photon Index Component kT NH Luminosity Luminosity (keV) 1021cm−2 1040 ergs s−1 1040 ergs s−1 NGC 3367 2.15  0.08 0.64  0.03 —– 3.3 2.0 NGC 4536 2.3  0.3 0.58  0.03 1.1  0.4 1.7 0.9

in the pn or MOS cameras according to the SAS task epatplot. The Reflection Grating

Spectrometer (RGS) data have signal-to-noise ratio (S/N) that is too low for a meaningful analysis.

The observation log with dates of the observations, EPIC net exposures, and the average count rates are reported in Table 2.1.

The spectral analysis was performed using the XSPEC v.12.3.1 software package (Ar- naud 1996; Dorman & Arnaud 2001). The EPIC data have been re-binned in order to contain at least 20 counts per channel, depending on the brightness of the source. The errors on spectral parameters are at 90% confidence level for one interesting parameter

(∆χ2 = 2.71).

2.3 Results

2.3.1 X-Ray Spectral Analysis

We fitted the EPIC spectral data in the 0.3 - 10 keV range. To analyze the spectral data we first attempted to find the best-fit model for the pn data, which would then be applied

7 to the MOS1 and MOS2 data. For NGC 3367 we fitted the data with a simple power law with galactic absorption which did not provide an acceptable fit. Positive residuals at low energy suggest the addition of a thermal component which provided an acceptable fit with

2 a χred = 1.34 for 88 degrees of freedom. Allowing the column density to vary or adding additional absorption components did not improve the fit. Therefore we determined that the best-fit model shown in Figure 2.2 was produced by a power law with an added thermal component. The model provided a kT value of 0.64  0.03 keV for the thermal component and a photon index value of 2.15  0.08. Applying this model to the pn, MOS1, and MOS2

2 data resulted in a good fit with a χred = 1.25 for 140 degrees of freedom. A summary of the results is shown in Table 2.2.

We then determined the contribution of the components to the overall flux by evaluating each component separately. Removing the power-law component and leaving the thermal component gave fluxes of 6.3 × 10−14 and 1.7 × 10−15 ergs cm−2 s−1 for the 0.3–2 keV and the 2–10 keV ranges respectively. Restoring the power-law component and removing the thermal component gave fluxes of 6.0 × 10−14 and 9.1 × 10−14 ergs cm−2 s−1 for the 0.3–2 keV and the 2–10 keV ranges respectively. From these values it can be determined that the power-law component dominates in the 2–10 keV energy range with 98% of the luminosity.

For NGC 4536 we again fitted the data with a simple power law with galactic absorption which also did not result in an acceptable fit. As before, we then added a thermal com-

2 ponent, which resulted in an acceptable fit with a χred = 1.37 for 70 degrees of freedom. Next we added an absorber at the position of the source, which further improved the fit to

2 a χred = 1.23 for 69 degrees of freedom. The best-fit model included a thermal component with a kT value of 0.58  0.03 keV, a photon index value of 2.30  0.26, and an absorber

21 −2 with NH = (1.14  0.4) × 10 cm . Applying this model to the pn, MOS1, and MOS2

2 data resulted in a reasonably good fit with a χred = 1.24 for 106 degrees of freedom. Lastly we determined the contribution of the different components to the overall flux.

Removing the power-law component and leaving the thermal component gave fluxes of

8 "!# -+)*%%'( "!# -+)*&'%(

# # ! ! *,1; *,1; # # ! "!"# ! "!"#

#" !% #" !% )6:59*.091*.8 )6:59*.091*.8

% % $ $

70936 # 70936 # " " "!&# $ & "!'# $ ' *5172<*.41/ *5172<*.41/

Figure 2.2: Best fit models and data-to-model ratios obtained by fitting the EPIC pn spectra in the 0.3-10 keV energy range. Both models include a best-fit power law plus a thermal component and are absorbed by Galactic NH . NGC 4536 requires additional intrinsic absorption.

5.5 × 10−14 and 1.7 × 10−15 ergs cm−2 s−1 for the 0.3–2 keV and the 2–10 keV ranges respectively. Restoring the power-law component and removing the thermal component gave fluxes of 6.1×10−14 and 6.0×10−14 ergs cm−2 s−1 for the 0.3–2 keV and the 2–10 keV ranges respectively. From these values we again determined that the power-law component dominates in the 2–10 keV range with 98% of the luminosity.

2.3.2 X-Ray Temporal Analysis

Since in general the X-ray variability is one of the defining properties of AGNs, it is impor- tant to investigate the temporal properties of NGC 3367 and NGC 4536. We studied the short-term variability of both sources using EPIC pn data with time-bins of 1000 s. Figure

2.3 shows the EPIC time series in the 0.3-10 keV band. A visual inspection of Figure 2.3 reveals that the average count rate of NGC 4536 is larger than the one of NGC 3367 by a factor of ∼10, and suggests that low-amplitude flux changes might be present in both light curves. However, a formal analysis based on a χ2 test indicates that there is statisti- cally significant variability only in NGC 4536 (Pχ2 ≃ 9%), while no significant variability

(Pχ2 ≃ 83%) is detected in NGC 3367.

9 Figure 2.3: XMM-Newton EPIC light curves for NGC 3367 and NGC 4536. The time bin is 1000 s.

2.3.3 UV - Optical Data

The OM allows us to simultaneously investigate the optical-UV properties of NGC 4536 and NGC 3367. In the case of these galaxies, the AGN is of low luminosity and furthermore obscured in the optical. It is thus likely that the optical-UV emission is dominated by star formation in the host galaxy in addition to being attenuated by dust. Nonetheless, we compare the UV-X-ray properties of NGC 4536 and NGC 3367 to optically identified more powerful AGNs to see if they are similar. In particular, we can compare the broadband spectral index αOX between our galaxies and optically identified AGNs. Using a sample of 333 optically selected Seyferts, Steffen et al. (2006) found that the broadband spectral index αOX are highly correlated with the UV monochromatic luminosity. For our sample, we apply the same approach by computing the spectral index, αOX = log(l /l )/ 2500A˚ 2keV log(ν /ν ) (Tananbaum et al. 1979), where lν is the monochromatic luminosity in 2500A˚ 2keV units of erg s−1 Hz−1. This is then plotted against the UV monochromatic luminosity.

The 2500 A˚ flux is obtained by converting the flux measured in the UVM2 band (2310 A)˚

−0.7 assuming a typical slope of 0.7 (fν ∝ ν ). Different extinction prescriptions were tested to account for the reddening in the UV band; the Small Magellanic Cloud extinction law was finally used for NGC 3367 and NGC 4536 (see Gliozzi et al. 2008 for details). E(B-V)

10 values of 0.029 mag and 0.018 mag were used for NGC 3367 and NGC 4536 respectively.

The observed magnitudes for NGC 3367 and NGC 4536 are 14.3 and 15.2 respectively.

The extinction corrected fluxes in the UVM2 band are 1935 µJy and 770 µJy for NGC 3367 and NGC 4536 respectively. The fluxes at 2500 A˚ are 2046 µJy and 814 µJy for NGC 3367 and NGC 4536 respectively.

The values of αOX for our sample are shown superimposed to the best-fit linear regression found by Steffen et al. (2006) in Figure 2.4. As can be seen from the figure, NGC 3367 and

NGC 4536 are clear outliers to the relation and are underluminous in the X-ray as compared to the sample of Seyferts. A possible explanation for this discrepancy may be that optical and UV emission is dominated by star formation. This is reinforced by the fact that these objects are optically classified as H II galaxies, suggestive of nuclear star formation.

2.4 Previous Observations of NGC 3367 and NGC 4536

NGC 3367 is an isolated face-on Sc barred galaxy (de Vaucouleurs et al. 1976) that is optically classified as an HII galaxy (H97). Although optically there is no hint of an AGN, radio observations reveal a bipolar synchrotron outflow from an unresolved compact nucleus with a diameter <65 pc (Garc´ıa-Barreto et al. 2002), and the possibility of two lobes straddling the nucleus and extending up to 12 kpc (Garc´ıa-Barretoet al. 1998). Assuming a spectral index in the range of, α = 0 − 0.7, we estimate a 5 GHz flux for the nucleus of 1.0 to 1.4 mJy based on the VLA 8.4 GHz measurement of 0.96 mJy at ≲ 0.3′′ resolution by

Garc´ıa-Barretoet al. (2002). The higher spectral index assumed is that measured for the core at matched lower resolution (4.5′′) between 1.4 GHz (as published in Garc´ıa-Barretoet al. 1998) and 8.5 GHz, the latter being newly determined from our analysis of an archival

VLA dataset (program AK550). However this spectral measurement may be dominated by diffuse emission surrounding the core as seen in the maps from Garc´ıa-Barretoet al. (1998).

In a study of M81, another low luminosity AGN, α = 0-0.3 has been observed (Markoff et al. 2008). The central stellar velocity dispersion is 61.2  10.1 km s−1 (Ho et al. 2009).

11 Figure 2.4: αOX plotted versus 2500 A˚ monochromatic luminosity. The solid line corre- sponds to the best-fit relation derived by Steffen et al. (2006) and the dashed lines account for the uncertainties in the fit.

12 In the MIR Spitzer observation of NGC 3367 the [NeV] lines at 14.3 µm and 24.3 µm were detected (Satyapal et al. 2008). The detected fluxes of these lines were 1.2  0.3 and 0.93  0.24 10−22W cm−2 respectively. The aperture from which the [NeV] 14.3 µm line was detected corresponds to a projected size of ∼1 × 2.4 kpc. Photoionization models of mid-infrared fine-structure line fluxes suggests that about 10 to 30 percent of the total bolometric luminosity of the galaxy may be attributed to the AGN.

NGC 4536 is a barred late-type spiral SABbc (de Vaucouleurs et al. 1976) optically classified as an HII galaxy (H97). HST imaging shows no evidence of a classical bulge, but has a surface brightness profile consistent with a pseudo-bulge that appears to exhibit spiral structure (Fisher 2006b). To the best of our knowledge, the only hint of an AGN in the optical is the optical line ratios from Space Telescope Imaging Spectrograph spectroscopy that suggest the presence of a weak AGN (Hughes et al. 2005). There is evidence of star formation in the central ∼ 20′′ × 30′′ region with strong Brγ emission (Puxley et al. 1988) and 10.8 µm emission (Telesco et al. 1993). Also a continuum-subtracted image of Hα

+ [NII] emission (Pogge 1989) is suggestive of nuclear star formation. The radio emission has a diffuse morphology with three separate peaks (Vila et al. 1990; Laine et al. 2006) that may be an annular ring of star formation around the nucleus. The morphology is similar to what is seen at 10.8 µm (Telesco et al. 1993), as well as in the 1-0 S(1) molecular hydrogen line (Davies et al 1997). ROSAT High Resolution Imager data revealed two ultra luminous X-ray sources, one of which may be coincident with the optical nucleus (Liu &

Bregman 2005). The second ULX is 150′′ away from the nucleus of the galaxy, well outside the 30′′ aperture radius of the XMM-Newton observation. The central region of NGC4536 is composed of an extended radio source at arcsecond resolution (Vila et al. 1990). From our reanalysis of the VLA 4.9 GHz map (1.3′′ beam; program AD176) published by Vila et al. (1990), we measure a point source upper limit of <1.7 mJy for the nucleus based on the observed peak surface brightness at the optical center. The central stellar velocity dispersion is 85  1 km s−1 (Batcheldor et al. 2005).

13 The [NeV] 14.3 µm line was detected in this galaxy from an ∼770 × 770 pc region, approximately 10′′ northeast of the optical nucleus (Satyapal et al. 2008). A flux of 0.32

 0.09 × 10−21 W cm−2 was detected. [OIV] 26 µm and [NeIII] 15.5 µm lines were also detected, but were concentrated at a location nearer to the optical nucleus than the [NeV] emission. Photoionization models of mid-infrared fine-structure line fluxes suggests that about 10 to 30 % of the total bolometric luminosity of the galaxy may be attributed to the

AGN.

2.5 Contamination by X-Ray Binaries

The 2-10 keV band X-ray luminosity of 2.0 x 1040 ergs s−1 and 0.9 x 1040 ergs s−1 for

NGC 3367 and NGC 4536 respectively is consistent with low-luminosity AGNs (Ho et al.

2001). A compilation of low-luminosity AGN from Ho et al. (2001), finds X-ray luminosities ranging from ∼ 1037− ∼ 1041 ergs s−2

While the observed X-ray emission is most likely due to the presence of an AGN, there is a possibility of contamination due to X-ray binaries (XRBs). Gilfanov (2004) examined the connection between the X-ray luminosity of low mass X-ray binaries (LMXBs) and the stellar mass of 11 galaxies with no evidence of current star formation. An average ratio of

28 −1 −1 Lx/M⋆ = 8.3 × 10 ergs s M⊙ was found for the 0.3-10 keV luminosity. For NGC 3367 and NGC 4536, LMXBs would account for approximately only 5 percent of the observed

X-ray luminosity according to this ratio.

High mass X-ray binaries (HMXBs) however may account for significantly higher portion of the observed luminosity since both galaxies show emission lines consistent with ongoing star formation. Colbert et al. (2004) used a sample of 32 spiral, elliptical, and irregular galaxies, with 1441 X-ray point sources detected by Chandra to study the relation between the X-ray emission (0.3-8 keV) from these point sources and the properties of the galaxy.

They related the X-ray point source luminosity, assumed to be mostly from black hole

HMXBs and ULXs, to K-band luminosity and the FIR-UV luminosity. Using the K-band 14 luminosity to determine the stellar mass and the FIR-UV luminosity to determine the star formation rate (SFR) they found a linear relationship between X-ray luminosity, stellar mass, and the SFR of a galaxy. Using this relation, we estimated the luminosity expected from X-ray sources other than an AGN. SFRs for NGC 3367 and NGC 4536 were estimated using the extinction-corrected UV luminosities calculated earlier, and using the formula

−1 given in Kennicutt (1998). With SFRs of 0.65 and 0.16 M⊙ yr for NGC 3367 and NGC

4536 respectively, the relation gives an X-ray luminosity from point sources of approximately

10 percent of the observed luminosity from both NGC 4536 and NGC 3367, suggesting that

AGN dominates the X-ray luminosity.

The possible contamination by XRBs may also be estimated based on their expected density within the area observed by the 30′′ beam of the XMM-Newton. Pence et al. (2001) examined the distribution of X-ray point sources within NGC 5457. Using the surface density of X-ray point sources in NGC 5457, a rough estimate of the number of X-ray sources in NGC 3367 and NGC 4536 may be determined. They determine the surface density of

X-ray sources as a function of the distance from the nucleus. Assuming a distance of 6.85

Mpc for NGC 5457 (Saha et al. 2006), the distance from the nucleus may be compared with our galaxies. The XMM-Newton extraction aperture for NGC 3367 corresponds to a radius of 6.3 kpc and is estimated to have a surface density of 0.29 sources per kpc2.

The XMM-Newton extraction aperture for NGC 4536 corresponds to a radius of 2.6 kpc and is estimated to have a surface density of 0.56 sources per kpc2. These values give an expected number of sources of approximately 36 and 12 sources for NGC 3367 and NGC

4536 respectively. Typical XRBs have luminosities of up to 1037 ergs s−1 (White et al.

1995). However, assuming all the X-ray sources in NGC 3367 and NGC 4536 emit at ∼1038 ergs s−1 (the highest luminosities reached by XRBs in NGC 5457, Pence et al. 2001) NGC

3367 would require at least 200 sources and NGC 4536 would require at least 90 sources in order to account for the observed X-ray emission in the 2-10 keV band if these sources were exclusively due to XRBs. While we assumed a similarity between NGC 5457 and our

15 sample that may not exist, the probability of these many sources emitting at an average luminosity of 1038 ergs s−1 within an area of a few kpc2 is extremely unlikely.

In addition to the lower X-ray luminosities, XRBs are expected to be characterized by different SEDs over the UV - optical - X-ray range compared to those from AGNs (Yuan

& Cui 2005). By using the [NeV] emission as a proxy for UV emission, the ratio of the

[NeV] luminosity to the X-ray luminosity can be used as a crude indicator of the SED of the ionizing source. The L[NeV]/Lx ratio for NGC 3367 and NGC 4536 is 0.130 and 0.013 respectively, consistent with optically identified AGN (Gliozzi et al. 2009). The L[NeV]/Lx ratio will of course depend on a number of parameters including the ionization parameter and the fraction of the emission due to star formation. Moreover there is a large scatter in the correlation between [NeV] emission and UV emission. However, the similarity between this ratio in NGC 4536 and NGC 3367 suggests that XRBs alone cannot be responsible for both the X-ray and [NeV] line emission in NGC 3367 and NGC 4536. Lastly, the X-ray variability presented in section 2.3.2 for NGC 4536 strongly suggests that the emission is from a single source.

We have thus shown that the X-ray luminosities of NGC 4536 and NGC 3367 are unlikely to be due to XRBs. Furthermore, the detection of the [NeV] line together with the high

X-ray luminosities, is highly suggestive of an AGN.

2.6 Bolometric Luminosities

An estimation of the bolometric luminosity of the AGN may be found using the X-ray and [NeV] luminosities using their respective bolometric correction factors. If we assume that the X-rays are emitted exclusively by the AGN, the X-ray luminosity can be used to estimate the bolometric luminosity of the AGN. If we assume that the black holes at the center of these galaxies have relatively low masses given their low bulge masses, then the Eddington ratios of these galaxies are likely to be high (Greene & Ho 2007). Lusso et al. (2010) studied 545 X-ray selected type 1 AGNs from the XMM-COSMOS survey

16 Table 2.3: Black Hole Mass Estimates

Galaxy logMEdd logMEdd logMBMC logMσ logMfund [NeV] X-ray NGC 3367 >5.19 >3.93 3.85 - 5.40 5.94 6.85 - 6.92 NGC 4536 >3.91 >3.58 <4.15 6.52 <6.67

with available black hole masses and bolometric luminosities estimated using their spectral energy distribution. Assuming that these AGNs emit at very high rates with Eddington ratios >0.2 we adopt a mean bolometric correction factor of 53 (Lusso et al. 2010). Using this bolometric correction factor, the estimated bolometric luminosities are logLbol(ergs

−1 −1 s ) = 42.0 and logLbol(ergs s ) = 41.7 for NGC 3367 and NGC 4536, respectively. For these galaxies photoionization by a starburst is unlikely to give rise to significant

[NeV] emission (Abel & Satyapal 2008). If we assume that the [NeV] only arises from the AGN, we can use the relation between bolometric luminosity and [NeV] luminosity established in a large sample of more powerful AGNs (Satyapal et al. 2007) to estimate the bolometric luminosity of the AGNs. The bolometric luminosities are estimated to be

−1 −1 logLbol(ergs s ) = 43.3 and logLbol(ergs s ) = 42.0 for NGC 3367 and NGC 4536, respec- tively. These luminosities are a factor of 20 and 2 higher, respectively, than the bolometric luminosities estimated using the X-ray luminosities. This discrepancy may suggest that some of the [NeV] emission originates in shocks. However, given the large scatter in the correlations used to estimate the bolometric luminosities, any definitive conclusions cannot be made about the discrepancy.

2.7 Black Hole Mass Estimation Methods

Eddington Mass Limits — Using the [NeV] and X-ray estimated bolometric luminosities, we can obtain lower limits on the mass of the black hole assuming the Eddington limit. The black hole mass estimates are given in Table 2.3.

17 Bulk Motion Comptonization (BMC) Model — One possible method to determine the mass in BH systems relies on the fact that hard X-rays are produced by the Comptoniza- tion process in both stellar and supermassive black holes. Specifically, by fitting the X-ray spectra of Galactic black holes (GBHs) with the BMC model during their spectral transi- tion, one can derive a universal scalable relationship between the photon index Γ and the normalization of the BMC model NBMC. If one assumes that the physics of a black hole system is the same no matter the scale and that the bulk of the X-ray emission is produced by Comptonization, this technique can be used to compare masses of black holes of differing scales. Using the known mass and properties of a Galactic black hole used as reference, the mass of the black hole in question can be extrapolated. A more detailed description of the application of this method to AGNs is given in Gliozzi et al. (2009, 2010). The mass estimates obtained using this technique vary depending on the uncertainty of the BMC spectral parameters and on the GBH used as a reference giving a range of values. This mass range for NGC3367 and the upper limit mass for NGC4536 is compatible with the lower mass limit derived using the Eddington limit as can be seen in Table 2.3.

M-σ Relation — The mass of the central black holes can be estimated using the M-

σ relation, assuming that the relation extends to lower mass ranges. Using the velocity dispersion values of 61.2  10.1 km s−1 (Ho et al. 2009) for NGC 3367 and 85  1 km s−1

(Batcheldor et al. 2005) for NGC 4536 and the M-σ relation from G¨ultekinet al. (2009b), the masses are found to be consistent with the lower mass limits derived using the Eddington limit, but significantly higher than those derived using the BMC model.

Fundamental Plane — This method of estimating black hole mass uses an apparent correlation between X-ray luminosity, 5GHz radio luminosity, and black hole mass. G¨ultekin et al. (2009a) recently examined this correlation, using dynamically-determined black hole masses, archival data from Chandra, and 5GHz luminosities from the literature for 18 galaxies. Using the X-ray luminosity and the 5GHz core radio flux estimated earlier, mass estimates for our galaxies may be determined. The masses estimated with this method are consistent with masses implied by the M-σ relation as can be seen in Table 2.3. The

18 relation however comes with in intrinsic scatter 0.77 dex, larger than the scatter of the M-σ relation, but like the M-σ relation, the scatter increases as the mass decreases. The relation

7 9 is developed using higher mass black hole on the order of 10 to 10 M⊙ and may not be appropriate for this sample if their central black holes truly have relatively small masses.

G¨ultekinet al. (2009a) explains this scatter with the possibility that the relation may not apply to sources that accrete at high rates. They also explain that the scatter may be skewed by a few outliers.

We note that the last two methods yield significantly larger values for the black hole than the ones obtained using the scaling technique based on the results from the BMC fit.

However, Gliozzi et al. (2010) using very high-quality X-ray spectra of the narrow line

Seyfert 1 galaxy PKS 0558-504, showed that the value of MBH obtained with this technique is fully consistent with the other indirect methods. The low values of MBH derived for NGC 3367 and NGC 4536 using the scaling technique can be reconciled with those derived from the fundamental plane and from the M-σ relation if the measured X-ray luminosity is substantially underestimated (for example because of a significant intrinsic absorption).

This scenario is consistent with the anomalous optical - X-ray spectral index discussed in section 2.3.3.

2.8 Summary

We have analyzed the XMM-Newton observations of two late-type galaxies; NGC 3367 and NGC 4536, which are believed to harbor AGNs based on the previous detection of mid-infrared [NeV] line emission. Our main results are summarized as follows:

1. Detailed spectral analysis of the XMM-Newton data for these galaxies reveals that the

X-ray luminosity is dominated by a power law with 2-10 keV luminosities of 2.0×1040

ergs s−1 for NGC 3367 and 0.9 × 1040 ergs s−1 for NGC 4536, respectively, consistent

with low luminosity AGNs.

19 2. The possibility that significant X-ray emission may be emitted by X-ray binaries was

explored. It was found that XRBs could be responsible for only a small fraction

of the X-ray luminosity detected, and that the significant [NeV] emission cannot be

explained by the presence of XRBs alone.

3. Low-amplitude flux changes may be present in both galaxies, and NGC 4536 shows

statistically significant variability, providing further evidence of an AGN.

4. A comparison of the black hole mass estimated using the BMC model with other

methods suggests the X-ray source may be absorbed in both galaxies. This hypothesis

is supported by the steep spectral index αOX found in both sources.

5 7 4 6 5. Estimated black hole masses range of 10 - 10 M⊙ for NGC 3367 and 10 - 10 M⊙

for NGC 4536.

We have demonstrated that MIR spectroscopy coupled with X-ray observations can more robustly determine the presence of an AGN and estimate the black hole mass.

20 Chapter 3: Optically Hidden AGNs in non-U/LIRG Interacting Galaxies

3.1 Introduction

Based on the current cold dark matter cosmological framework, it is now well-established that galaxy interactions are ubiquitous and that they play a pivotal role in the formation and evolution of galaxies. From both a theoretical and observational perspective, galaxy interactions are undoubtedly responsible for enhanced nuclear star formation (e.g., Mihos

& Hernquist 1996; Larson & Tinsley 1978; Sanders & Mirabel 1996; Kennicutt et al. 1987;

Woods, Geller, & Barton 2006; Woods & Geller 2007; Ellison et al. 2008; Smith & Struck

2010; Patton et al. 2011; Liu et al. 2012), and the formation of spheroids (e.g., Toomre

1977; Lake & Dressler 1986; Shier & Fischer 1998; Rothberg & Joseph 2006). A natural assumption from the tight correlation between central black hole mass and bulge velocity dispersion (e.g., Gebhardt et al. 2000) is that in addition to bulge growth, interactions trigger accretion onto a central supermassive black hole. However, despite over three decades of extensive research, it is still unclear whether or not there is observational evidence for a causal connection between mergers and Active Galactic Nuclei (AGN), and, if so, how this connection depends on merger and host galaxy parameters. A number of studies have found evidence for mergers in AGN hosts (e.g., Surace et al. 1998; Canalizo & Stockton 2001;

Ramos Almeida et al. 2011). On the other hand, recent large surveys that include carefully constructed control samples find no evidence that AGN hosts are more tidally disturbed than quiescent galaxies (e.g., Cisternas et al. 2011). These studies depend critically on sample size and the choice of control sample. Furthermore, since tidal features can be faint and appear only in the gas instead of the stars (e.g. Kuo et al. 2008), the sensitivity 21 (e.g. Canalizo & Stockton 2001; Ramos Almeida et al. 2011) and the wavelength (e.g.,

Hancock et al. 2007; Boselli et al. 2005) of the observations are critical. To circumvent these challenges, there have been a large number of optical spectroscopic studies aimed at

finding AGN signatures in confirmed samples of interacting pairs. However, these studies also yield contradictory results. Several studies have shown little evidence for an increased incidence of AGNs in pairs compared with a control (e.g., Bushouse 1986;

Darg et al. 2010) in direct contrast with a number of other studies (e.g., Keel et al. 1985;

Dahari 1985; Ellison et al. 2011; Silverman et al. 2011).

Since the centers of interacting galaxies may be more obscured than isolated galaxies, a potential shortcoming of previous optical studies is that they can miss buried AGN. More- over, the onset of accretion activity in an interaction will likely be weak and accompanied by star formation. In such cases, the traditional optical spectroscopic line ratios will be dominated by star formation, hiding the AGN (see Abel & Satyapal 2008). In such cases, observations of mid-IR fine-structure lines are a powerful tool for finding optically uniden- tified AGNs. AGNs show prominent high-excitation fine structure lines that are not seen in normal and starburst galaxies (Genzel et al. 1998; Satyapal et al. 2004, 2007, 2008, 2009).

A key prominent line is the [NeV] 14 µm line with an ionization potential of 97 eV. This line is not seen in starburst galaxies since even the most massive stars produce few photons of this energy (e.g., Sturm et al. 2002; Satyapal et al. 2004). The detection of this line in any galaxy therefore provides strong evidence for an AGN. In fact, mid-IR spectroscopic studies of external galaxies have revealed that the census of AGN activity in the local Universe is significantly underestimated by optical spectroscopic studies alone (e.g., Satyapal et al.

2007, 2008, 2009; Goulding & Alexander 2009), underscoring the importance of conducting mid-IR spectroscopic studies of interacting galaxies.

While there have been a number of mid-IR spectroscopic studies of interacting galax- ies, most previous studies have focused primarily on the most luminous interactions, which target typically advanced mergers between galaxies of comparable mass (“major mergers”).

22 Indeed, virtually all large coordinated studies carried out with ISO (e.g., Genzel et al. 1998), and Spitzer (e.g., Veilleux et al. 2009; Armus et al. 2007, 2009; Farrah et al. 2007; Petric

12 11 et al. 2011) have targeted ultraluminous (L> 10 L⊙) or luminous (L> 10 L⊙) infrared galaxies (U/LIRGs), which are invariably advanced mergers of equal mass progenitors. This is a serious deficiency since only a narrow phase in the merger sequence is observationally studied. Theory predicts that as a merger progresses, gravitational instabilities cause large radial gas inflows toward the nuclear regions. These numerical simulations show that mul- tiple gas inflow epochs along the merger sequence occur, resulting in at least two distinct episodes of enhanced star formation and possibly accretion activity, with the first peak tak- ing place shortly after the first close encounter, when the two galaxies are widely separated

(e.g., Mihos & Hernquist 1996; Cox et al. 2006). Therefore, no observational study of the incidence and properties of AGNs in interacting galaxies is complete without the inclusion of a wider range of merger stages.

To address this deficiency, we present a Spitzer mid-infrared spectroscopic investigation of a strategically-selected sample of non-U/LIRG interacting galaxy pairs, which generally target earlier interaction phases. This is the first systematic mid-infrared spectroscopic investigation of a significant sample of non-U/LIRG interactions. The principal goals of this work are to determine the incidence and intrinsic luminosity of optically hidden AGNs and to explore the dependence on pair separation/merger stage and host galaxy properties.

Together with previously published studies aimed at normal isolated galaxies and U/LIRGs, our proposed database will enable a more comprehensive study of merger evolution from the first close pass through to final coalescence.

This chaper is structured as follows. In Section 3.2, we summarize the properties of the

Spitzer sample presented here in addition to previously published comparison samples. In

Section 3.3, we summarize the observational details and data analysis procedure, followed by a discussion of the incidence of AGNs in section 3.4. In Section 3.5, we explore the dependence of AGN incidence on host galaxy and merger properties, followed by an investi- gation of the dependence of AGN properties on pair separation, and merger stage in section 23 3.6. In section 3.7, we discuss the possible dual AGN found in our sample. A summary of our major conclusions is given in Section 3.8.

3.2 The Sample

3.2.1 The non-U/LIRG pairs sample

Here we present a comprehensive mid-IR spectroscopic survey using Spitzer IRS of 30 pairs of interacting galaxies. Our goal was to assemble a sample of interactions that are less lu- minous than the U/LIRG interacting galaxies, since this population of interacting galaxies have not been systematically investigated in any previous infrared spectroscopic studies.

Our pairs sample is an archival sample and is thereby restricted by the number of interact- ing galaxies that have been observed by the high-resolution IRS modules onboard Spitzer.

The sample was assembled by conducting an exhaustive search of the Spitzer archive using several large compilations of interacting galaxies including the Arp Atlas of Peculiar Galax- ies, which selects galaxies with strong morphological indicators of interactions, the optical spectroscopic sample from Keel et al. (1985), which includes an extensive list of physi- cally associated pairs with no morphological indicators of interactions, and the Galaxy Zoo merger list, which contains over 3000 morphologically identified mergers drawn from the

Sloan Digital Sky Survey (Darg et al. 2010). We required that both nuclei in each pair was observed by both high-resolution IRS modules. In addition, we applied the following criteria to the sample: 1) we eliminated all galaxies with more than one companion within

100 kpc, since our intent is to study the effects of interactions and not local galaxy density,

2) we included only galaxies with two distinct nuclei, eliminating all single nuclei, irregular, and starburst galaxies in the Arp atlas, and 3) we eliminated any U/LIRG (all sources with

11 LFIR > 10 L⊙) interacting galaxies, since these galaxies are part of large programs and the results published.. The resulting sample contains 30 galaxy pairs and includes both mergers with clear morphological signatures of interactions as well as physically associated pairs with no evidence of tidal distortion or structural peculiarity. 90% of the pairs in this 24 sample are part of the optical spectroscopic sample from Keel et al. (1985). We note that the galaxies in our final pairs sample were observed as part of a random collection of guar- anteed and open-time programs with differing scientific goals; the sample should therefore not be viewed as complete in any sense.

In Figure 3.1, we display DSS images of the galaxy pairs from our pairs sample. As can be seen from the figure, the morphologies of the pairs in our pairs sample are var- ied, with some systems showing distorted morphologies and prominent tidal features, and others with large pair separations with no obvious signs of interactions. In Table 3.1, we summarize the basic properties of our pairs sample. The properties of the host galaxies in our pairs sample were assembled by conducting an extensive literature search. Distances were complied using NED, adopting the most recent estimated distance. Through visual inspection of the assembled DSS images, galaxies with tidal features were identified and are listed in Table 3.1. Based on the DSS images, all galaxies in our sample were characterized by an merger stage according to the classification scheme based on numerical simulations

(Mihos & Hernquist 1996) first proposed by Surace et al. (1998) and used by Veilleux et al.

(2002) to classify ULIRGs. We emphasize that the identification of tidal features as well as the adopted classification scheme is dependent on the wavelength and sensitivity of the observations. Indeed, a recent HI study conducted by Kuo et al. (2008) of local Seyferts showed remarkably the presence of tidal features in the HI gas in the vast majority of opti- cally undisturbed hosts. The adopted identifications should therefore be viewed with some caution. Stellar masses were estimated using K-band 20 mag arcsec−2 isophotal elliptical aperture magnitudes from 2MASS by assuming a mass-to-light ratio of 1.32 M⊙/L⊙, taken from Cole et al. (2001) for the Salpeter IMF. Hubble type, nuclear classifications based on optical emission line ratios, and IR luminosities where available were also compiled for our sample and are displayed in Table 3.1.

25 3.2.2 Comparison Samples: The SINGS and ULIRG Sample

Any study of interacting galaxies requires a carefully constructed “control” sample, and a thorough investigation of possible statistical biases arising from observational selection.

Since this study is an archival study, no deliberately designed accompanying control sample exists. Moreover, the number of galaxies selected for a control sample is limited by the number of galaxies observed by the high-resolution module onboard Spitzer. The largest sample of nearby galaxies observed by the high-resolution IRS modules is the SINGS sample of nearby galaxies (Kennicutt et al. 2003; Dale et al. 2005). We constructed a comparison sample selected from the SINGS sample, which contains 35 galaxies that can be considered isolated following the criteria adopted by Smith et al. (2007). Galaxies were eliminated if they had companions whose velocities differ by <1000 km s−1, that have an optical luminosity >1/10 the target galaxy, and are separated from the target galaxy by <10 times the optical diameter of either galaxy. Each of the galaxies in this sample has published high- resolution IRS spectra of the nucleus (Dale et al. 2009), which can be readily compared to our data.

In addition to our constructed isolated nearby sample, we also compared the mid-IR spectral properties of our pairs sample with those of ULIRGs from Veilleux et al. (2009).

The sample of ULIRGs compiled by Veilleux et al. (2009) are a subsample of the 1 Jy sample, chosen to be representative of the full sample in terms of redshift and luminosity. We also compared our results to those from a sample of LIRGs that are part of the GOALS sample

(Great Observatory All-sky LIRG Survey, Armus et al. 2009). Since mid-IR spectroscopic data is reported for only those galaxies with [NeV] detections (Petric et al. 2011), we included in our comparative analysis only those galaxies that have [NeV] detections that have a pair visible in the DSS images. Optical classifications of these galaxies were compiled from NED. We note that since the mid-IR spectroscopic results for the full GOALS sample are not yet published, our comparative analysis with the interacting LIRGs is limited.

26 IC750 IC749 NGC4394 NGC4382

NGC2633 NGC2970 NGC2634 NGC2968 NGC3166 NGC3165

12' x 12' 12' x 12'

NGC4567 NGC5481 NGC5480 NGC3800 NGC3799 NGC5394 NGC5195 NGC5194

NGC5395 NGC4568 12' x 12'

NGC7753 NGC7752 NGC3227 NGC3226 NGC2300 NGC2276 NGC4438 NGC4435 NGC275 NGC274

12' x 12' Binar y 12' x 12'

NGC5614 NGC5615 NGC5545 NGC5544 NGC3448 UGC6016 NGC7682 NGC7679 NGC474 NGC470

Binar y 12' x 12'

NGC4038 NGC3396 NGC3395 NGC5427 NGC935 NGC1253A NGC1253

NGC4035 IC1801 12' x 12' NGC5426 Binar y

NGC7715 NGC7714 IC196 IC195 NGC3788 NGC3786 IC564 NGC1242 NGC1241

IC563

Figure 3.1: DSS images of the non-U/LIRG pairs with IRS SH and LH slits overlaid. Images are 6′ by 6′ unless otherwise noted. As can be seen from the images, our pairs sample includes widely separated pairs with no signs of tidal distortion as well as pairs with clear morphological signatures of interactions.

27 40 35 70 Pairs 30 60 ULIRGs 30 Isolated 25 50 20 40 20 15 30 Number Number Number 10 10 20 5 10 0 0 0 0 100 400 800 1200 8 9 10 11 12 N/A 8 9 10 11 12 13 Distance (Mpc) log(M/M8) log(LIR/L8) 20 40 60 50 15 30 40 10 20 30

Number Number Number 20 5 10 10 0 0 0 E S0 Sa Sb Sc Sd-Sm I I II IIIa IIIb IVa IVb V 0 20 40 60 80 Hubble Type Merger Stage Pair Separation (kpc) 35 30 25 20 15 Number 10 5 0 N/A H L S Optical Type

Figure 3.2: Sample Properties: The properties of our pairs sample, the isolated SINGS sample, and the ULIRG sample. We note that Hubble types and stellar masses for the ULIRGs are not readily available in the literature and are therefore not listed. As can be seen, our pairs sample in general consists of infrared-faint galaxies with significantly wider separations and earlier merger stages that does the ULIRG sample.

28 Table 3.1: Sample Properties

Name D Type Tidal Merger PS log M log LIR Optical (Mpc) Features Stage (kpc) (M⊙) (L⊙) Activity IC750 23 Sab n I 23.2 10.9 - H IC749 23 SABcd n I 23.2 10.0 - H NGC2633 30 SBb n I 69.4 10.8 10.7 H NGC2634 30 E1 n I 69.4 10.6 9.1 - NGC2968 26 I0 n I 37.5 10.8 - - NGC2970 26 E1 n I 37.5 9.7 - - NGC3166 22 SAB0/a n I 29.8 11.2 10.0 L NGC3165 22 SAdm n I 29.8 9.4 <9.1 - NGC4394 17 SBb n I 37.0 10.6 9.2 L NGC4382 17 SA0 n I 37.0 11.4 <8.3 - NGC4567 26 SAbc y II 9.3 10.9 - H NGC4568 26 SAbc y II 9.3 11.2 10.8 H NGC5480 25 SAc n I 23.0 10.4 9.9 H NGC5481 25 cD n I 23.0 10.4 <8.4 - NGC3799 45 SBb y I 17.7 10.3 - H NGC3800 45 SABb y I 17.7 11.0 10.6 H NGC5394 47 SAb y IIIa 26.8 10.7 10.7 H NGC5395 47 SAb y IIIa 26.8 11.3 10.7 S2 NGC5194 8 SAbc y IIIa 9.9 11.0 10.3 S2 NGC5195 8 I0 y IIIa 9.9 10.6 9.4 L NGC7752 70 I0 y IIIa 40.3 10.6 10.9 H NGC7753 70 SABbc y IIIa 40.3 11.6 10.9 L NGC3226 21 E2 y II 13.1 10.6 - L NGC3227 21 SABa y II 13.1 10.9 10.2 S NGC2276 31 SABc n I 57.0 10.7 10.7 H NGC2300 31 SA0 n I 57.0 11.3 <9.6 - NGC4435 14 SB0 n I 18.3 10.8 9.2 H NGC4438 14 SA0/a n I 18.3 10.8 9.5 L NGC0274 22 SAB0 y II 5.2 10.4 - - NGC0275 22 SBcd y II 5.2 10.0 9.9 H NGC5614 53 SAab n II 6.7 11.4 10.3 L NGC5615 53 - n II 6.7 10.2 - H NGC5544 42 SB0/a y II 7.0 10.6 - L NGC5545 42 SAbc y II 7.0 10.5 9.7 L NGC3448 25 I0 n I 28.5 10.3 10.1 H UGC06016 25 Im n I 28.5 8.7 <9.1 - NGC7682 56 SBab n I 73.4 10.8 - S2 NGC7679 56 SB0 n I 73.4 10.9 10.9 S2 NGC474 36 SA0 n I 57.5 11.0 <8.8 L NGC470 36 SAb n I 57.5 10.9 10.5 H Continued on next page

29 Table 3.1 – Continued from previous page

Name D Type Tidal Merger PS log M log LIR Optical (Mpc) Features Stage (kpc) (M⊙) (L⊙) Activity NGC4038 25 SBm y II 8.3 11.2 - H NGC4039 25 SAm y II 8.3 11.1 10.9 H NGC3396 27 IBm y II 9.2 9.8 10.4 H NGC3395 27 SABcd y II 9.2 10.3 10.4 H NGC5426 36 SAc y IIIa 24.2 10.7 10.4 H NGC5427 36 SAc y IIIa 24.2 11.0 10.6 S IC1801 41 SBb y II 12.2 10.4 - H NGC935 41 Scd y II 12.2 10.9 10.3 L NGC1253A 21 SBm n I 23.2 9.4 9.4 - NGC1253 21 SABcd n I 23.2 10.2 9.6 H NGC7715 36 Im y IIIa 21.1 9.3 <9.5 H NGC7714 36 SBb y IIIa 21.1 10.6 10.7 H IC196 50 SBb n I 31.9 10.9 9.8 L IC195 50 SAB0 n I 31.9 10.7 - H NGC3786 37 SABa y I 14.8 10.8 - S NGC3788 37 SABab y I 14.8 10.8 - - IC564 82 SAcd n I 37.3 11.1 10.9 H IC563 82 SBab n I 37.3 10.9 10.8 - NGC1241 58 SBb n I 27.9 11.4 10.7 S2 NGC1242 58 SBc n I 27.9 10.4 - H

Col(1): Common Source Names; Col(2): Distances were complied using NED when avail- able, adopting the most recent estimated distance for each pair, otherwise are derived from −1 −1 redshift assuming H0=73 km s Mpc ; Col(3): Hubble types were taken from de Vau- couleurs et al. (1991); Col(4): Tidal features are identified through inspection of DSS images; Col(5): Merger stages as defined in Veilleux et al. (2002) are identified through inspection of DSS images; Col(6): Projected pair separations are calculated from angular separations between galaxies in each pair; Col(7): Stellar masses were estimated using K- band K-20 magnitudes from 2MASS by assuming a mass-to-luminosity ratio of 1.32; Col(8): IR luminosities were calculated from IRAS fluxes according to the prescription from Sanders & Mirabel (1996). Values in parentheses represent the total luminosity for the pair since the individual galaxies are not resolved; Col(9): Optical activity classifications are compiled from the literature.

30 3.2.3 Sample Properties and Potential Selection Biases

The properties of our pairs sample, the isolated galaxy comparison sample, and ULIRG comparison sample are shown in Figure 3.2. As can be seen, our pairs sample contains more distant galaxies than does the isolated sample. Since the physical scale of the IRS aperture will vary with distance, care must be taken in comparisons between these two samples. Our pairs sample and the isolated sample span the same range of Hubble types and stellar masses, however a Kolmogorov-Smirnov (K-S) test gives a 50% probability that the stellar masses of the isolated SINGS sample and our pairs sample are drawn from the same distribution. We emphasize that there are important selection effects that need to be explored in any comparative study carried out on these samples. As can be seen from

Figure 3.2, our pairs sample in general consists of infrared-faint galaxies with significantly larger separations and earlier merger stages than does the ULIRG sample.

3.3 Observations and Data Reduction

All observations were executed in staring mode using both the short-wavelength (SH,

4.7′′×11.3′′, λ = 9.9-19.6 µm ) and long-wavelength (LH, 11.1′′×22.3′′, λ = 18.7-37.2 µm ) high-resolution modules of the Infrared Spectrograph (IRS; Houck et al. 2004) with a spec- tral resolution of R ∼ 600. Data for the galaxies in our pairs sample were obtained from various observing programs (PIDs, 20140, 159, 30323, 3674, 96, 21, 14, and 3374) from

2003 December and 2009 September. Exposure times varied considerably, ranging from 30 to 840 seconds. Background sky observations located 2 ′ from the source were conducted following each observation in order to enable background subtraction.

On-source observations were centered on the galaxy nuclei. DSS images of all galaxies were carefully inspected in ensure that the nucleus of the galaxy falls within the slit. The slit size for the median distance of 30 Mpc corresponds to a projected extraction aperture of 0.7×1.6 kpc and 1.6×3.2 kpc for the SH and LH modules, respectively.

Prior to download the data were preprocessed by the IRS pipeline (version S18.7.0) at

31 the Spitzer Science Center (SSC). The preprocessing steps include ramp fitting, dark-sky subtraction, droop correction, linearity correction, flat-fielding, and flux calibration1. The resulting Spitzer spectra were then further processed using the SMART 8.1.2 analysis pack- age (Higdon et al. 2004) and the corresponding version of the calibration files (ver 1.7.1), to obtain final line fluxes. Using SMART, the spectra were then individually inspected for remaining bad pixels which were subsequently removed. The fine-structure line fluxes pre- sented here were obtained with Gaussian fits to spectral lines and linear fits to the baseline continuum.

Table 3.2: Fluxes for Full Sample

Name [NeV] 14.32µm [NeV] 24.32µm IC750 7.800.36 4.580.66 IC749 <1.06 <0.43 NGC2633 <3.56 <2.84 NGC2634 <1.15 <0.40 NGC2968 <1.39 <0.39 NGC2970 <0.96 <0.36 NGC3166 <1.22 <0.53 NGC3165 <3.85 <0.36 NGC4394 <2.41 <0.47 NGC4382 <8.35 <0.30 NGC4567 <0.99 <1.25 NGC4568 <1.66 <1.97 NGC5480 <0.86 <3.21 NGC5481 <1.85 <0.30 NGC3799 <0.89 <0.89 NGC3800 <0.87 <0.89 NGC5394 <1.61 <0.83 NGC5395 <2.29 <0.35 NGC5194 2.110.15 0.410.07 NGC5195 <0.97 <0.26 NGC7752 <0.16 <0.46 NGC7753 <1.73 <0.22 NGC3226 <1.03 <1.32 NGC3227 27.451.13 17.542.38 NGC2276 1.660.51 <0.35 Continued on next page 1See IRS Instrument Handbook, http://irsa.ipac.caltech.edu/data/SPITZER/docs/irs 32 Table 3.2 – Continued from previous page

Name [NeV] 14.32µm [NeV] 24.32µm NGC2300 <0.68 <0.29 NGC4435 0.650.18 <2.08 NGC4438 <1.76 <2.90 NGC0274 <0.65 <1.57 NGC0275 <1.20 <0.45 NGC5614 <39.64 <1.20 NGC5615 <34.61 <0.30 NGC5544 <5.26 <1.42 NGC5545 <4.61 <0.33 NGC3448 <1.21 <1.21 UGC 06016 <1.00 <3.45 NGC7682 2.600.40 <1.75 NGC7679 4.580.90 8.192.23 NGC474 <4.34 <0.36 NGC470 <6.03 <0.86 UGC08335E <1.08 <7.08 UGC08335W <0.82 <1.16 NGC5257 <0.82 <2.09 NGC5258 <1.35 <1.45 NGC4038 <4.83 <1.11 NGC4039 <2.07 <1.50 NGC3396 <2.17 <0.97 NGC3395 <2.32 <0.46 NGC5426 <1.12 <1.87 NGC5427 1.480.22 2.820.70 IC1801 <1.00 <0.95 NGC935 <1.12 <0.35 NGC1253A 4.131.05 1.980.25 NGC1253 <2.31 <1.69 NGC7715 <27.98 <3.13 NGC7714 <1.97 <3.56 IC196 <0.60 <1.69 IC195 <1.68 <1.50 NGC6286 <6.56 <2.58 NGC6285 2.210.31 <0.49 NGC3786 2.480.28 4.931.28 NGC3788 <1.45 <3.41 UGC09618N 3.300.51 5.561.67 UGC09618S <0.35 <1.40 IC564 <0.36 <1.00 IC563 <0.33 <0.73 Continued on next page 33 Table 3.2 – Continued from previous page

Name [NeV] 14.32µm [NeV] 24.32µm NGC1241 2.820.66 <1.02 NGC1242 <0.57 <0.52

Col(1): Common Source Names; Col(2) & (3): Fluxes are in units of 10−21 W cm−2. 3 σ upper limits are reported for non-detections.

3.4 Incidence of AGNs in Galaxy Pairs

In our pairs sample, we detected the 14 µm [NeV] line in 18% (11/60) of the galaxies, as seen in Figure 3.3, only 7 of these galaxies are optically classified as Seyferts. The detection of AGN not identified in optical surveys demonstrates that optical studies miss

AGNs even in non-U/LIRG galaxy interactions. We note that the observations of these galaxies have limited and variable signal-to-noise ratios that allows for the possibility that the [NeV] detection rate in our pairs sample would likely be higher if the sensitivity of the observations were higher and more uniform. The distribution of the [NeV] detections as a function of the 3 σ [NeV] 14 µm line sensitivity is presented in Figure 3.4. At sensitivities better than 39 ergs s−1, the fraction of galaxies with [NeV] detections increases as seen in

Figure 3.4, suggesting that if deeper exposures were obtained in all the IRS observations, the [NeV] detection rate would be higher than 20%.

3.4.1 Comparison with Isolated and ULIRG Samples

Of the 32 galaxies in the isolated SINGS comparison subsample, only 1 of the galaxies displays a [NeV] 14 µm detection compared to 11 of 60 for our pairs sample. This galaxy

(NGC 3621) is optically classified as a Seyfert (Barth et al. 2009). No new AGNs were discovered using infrared spectroscopy in the isolated SINGs sample. Since the observations

34 6.5 [NeV] 4.5 [NeV] 2.0 ) IC750 ) NGC5194 ) NGC3227 [NeV] -1 6.0 -1 -1

µm µm µm 1.8 -2 -2 4.0 -2 5.5

W cm W cm W cm 1.6 -19 5.0 -19 3.5 -18

4.5 1.4 3.0 4.0 1.2 Flux Density (10 Flux Density (10 Flux Density (10 3.5 2.5

13.6 13.8 14.0 14.2 14.4 14.6 14.8 15.0 14.1 14.2 14.3 14.4 14.5 14.6 14.0 14.2 14.4 14.6 14.8 Wavelength (µm) Wavelength (µm) Wavelength (µm)

2.0 ) NGC2276 ) 1.4 NGC4435 ) NGC7682 -1 2.5 -1 -1 [NeV]

µm µm µm 1.8 -2 [NeV] -2 1.3 [NeV] -2 2.0 1.6

W cm W cm 1.2 W cm -19 -19 -19 1.4 1.5 1.1 1.2

1.0 1.0 1.0

Flux Density (10 0.5 Flux Density (10 0.9 Flux Density (10 0.8

14.0 14.1 14.2 14.3 14.4 14.5 14.6 14.0 14.2 14.4 14.6 14.8 15.0 14.2 14.4 14.6 14.8 15.0 Wavelength (µm) Wavelength (µm) Wavelength (µm)

4 ) NGC7679 [NeV] ) 2.4 NGC5427 ) NGC1253A [NeV] -1 -1 [NeV] -1 4.5 µm µm 2.2 µm -2 -2 -2 3

4.0 2.0 W cm W cm W cm

-19 -19 1.8 -19 2

3.5 1.6 1 1.4 3.0

Flux Density (10 Flux Density (10 1.2 Flux Density (10 0

14.0 14.2 14.4 14.6 14.8 15.0 14.2 14.3 14.4 14.5 14.6 14.7 14.8 14.1 14.2 14.3 14.4 14.5 14.6 14.7 Wavelength (µm) Wavelength (µm) Wavelength (µm)

) 2.6 NGC3786 [NeV] ) 2.4 NGC1241 [NeV] -1 -1

µm 2.4 µm 2.2 -2 -2

2.2 2.0 W cm W cm -19 -19 2.0 1.8

1.8 1.6

1.6 1.4

Flux Density (10 1.4 Flux Density (10 1.2 14.2 14.4 14.6 14.8 14.0 14.2 14.4 14.6 14.8 15.0 Wavelength (µm) Wavelength (µm)

Figure 3.3: IRS spectra showing the [NeV] 14.3 µm line for the galaxies with detections listed in Table 3.2

35 60 Observations 50 Detections 40 30

Number 20 10 0 37 38 39 40 41 42 [NeV] Line Sensitivity (log (ergs s-1))

Figure 3.4: Distribution of 3σ limiting [NeV] line sensitivities for our pairs sample. Galaxies are included in each bin if they have sensitivities equal to or better than the given sensitivity range. Detections are indicated with darkened histograms

in the SINGS sample have comparable or better sensitivities than the observations in our pairs sample, this result strongly suggests that optically hidden AGNs are more prevalent in interacting pairs compared with isolated galaxies. Confirmation of this result requires a careful statistical analysis using a larger interacting sample together with a carefully constructed control sample, an endeavor that can only be accomplished when future infrared spectroscopic missions become available.

Of the 68 galaxies in the ULIRG sample, 28% (19/68) show [NeV] detections, signif- icantly higher than in our pairs sample. The limiting mid-IR sensitivities of the ULIRG observations are typically an order of magnitude worse than those in our non-U/LIRG sam- ple, suggesting that the [NeV] detection rate in ULIRGs would be even greater if more sensitive observations were available confirming recently pulbished results derived by differ- ent methods by Petric et al. (2011). It is therefore clear from our study that the incidence of AGNs in ULIRGs is significantly higher than in non-U/LIRG galaxy pairs

3.5 Which Galaxies Host Infrared-Identified AGNs?

Using our pairs sample, we explore the dependence of AGN incidence on various host galaxy and merger properties in our sample in Figure 3.5. Mid-IR identified AGNs occur in 36 both disturbed and undisturbed hosts with a wide range of Hubble types and galaxy pair separations.

Several previous studies suggest that the incidence of AGNs increases in the final stages of a merger just prior to coalescence (e.g Veilleux et al. 2009; Yuan et al. 2010). Since our sample targets wide pair separations and likely early merger stages, we combined our pairs sample with the previously published ULIRG (Veilleux et al. 2009) sample to explore the dependence of AGN incidence with merger stage. In Figure 3.6 we show the fraction of mid-IR identified AGN as a function of merger stage, where merger stage is based on optical morphology as defined and discussed in section 3.2. As can be seen, the fraction appears to decrease from stage I to stages II, and III (248% to 66%, and 188% respectively), and then increases to 3913% in stage IV. This trend is limited by the small number statistics but hints at the possibility that the distribution is bimodal, with the fraction of galaxies hosting AGN highest at the earliest and latest merger stages, confirming recently published results by Petric et al. (2011) and Yuan et al. (2010). In the next section, we summarize the published literature on the optically normal galaxies in our sample discovered through our mid-infrared spectroscopy to host AGNs.

3.6 What are the AGN Properties and how do they Depend

on Merger Stage?

Several studies have found that the AGN luminosity is correlated with merger stage or pair separation (e.g. Liu et al. 2012; Koss 2012), where the dominance of the AGN increases at more advanced stages, consistent with predictions from numerical simulations (e.g. Di

Matteo et al. 2005; Hopkins et al. 2006). The [NeV] luminosity is tightly correlated with the bolometric luminosity of the AGN and can therefore be used as indicator of the AGN luminosity (see Satyapal et al. 2007; Secrest et al. 2012). In Figure 3.7, we combine the

[NeV] luminosities of our pairs sample with those from the previously published interacting galaxies in the ULIRG (Veilleux et al. 2009) and LIRG (Petric et al. 2011) samples. We 37 1.0 1.0 tidal features 0.8 no tidal features 0.8

0.6 0.6

0.4 0.4

AGN Fraction 0.2 AGN Fraction 0.2

0.0 0.0 0 20 40 60 80 1 3 5 7 9 Pair Separation (kpc) Mass Ratio 1.0 1.0

0.8 0.8

0.6 0.6

0.4 0.4

AGN Fraction 0.2 AGN Fraction 0.2

0.0 0.0 N/A H L S E S0 Sa Sb Sc Sd-Sm I Optical Type Hubble Type 1.0

0.8

0.6

0.4

AGN Fraction 0.2

0.0 8 9 10 11 12 log(M/Me)

Figure 3.5: Properties of galaxies with [NeV] detections: The AGN fraction represents the fraction of galaxies with [NeV] detections in each bin.

38 1.0

0.8

0.6

0.4

AGN Fraction 0.2

0.0 I II III IV V Merger Stage

Figure 3.6: Distribution of merger stages for galaxies with [NeV] detections. The AGN fraction represents the fraction of galaxies with [NeV] in each bin. The first and fourth stages have the highest fraction of AGN.

find a clear deficit of the highest AGN luminosities in galaxies in the earliest merger stages.

We note that since mid-IR sensitivities for the ULIRGs are lower, we cannot say whether or not there is a lack of low luminosity AGNs at the latest merger stages. There is also a similar correlation between [NeV] luminosities and pair separation as seen in Figure 3.8, with increasing luminosities with decreasing pair separation.

3.6.1 How do Mid-IR Identified AGNs Compare to Optically Identified AGNs?

We studied the difference between the mid-IR-identified AGN (those with [NeV] detections but are not optically classified as Seyferts) and optically classified Seyferts with [NeV] detections. Again, we expanded our sample to include the sample of interacting LIRGs and

ULIRGs with [NeV] detections. In Figures 3.8 (a) and (b), we plot the distributions of pair separations and merger stages for mid-IR identified AGNs and optically classified AGNs. As can be seen in Figures 3.8 (a) and (b), both mid-IR identified AGNs and optically classified

Seyferts have similar distributions of pair separations and merger stages. A K-S test shows no significant difference between the pair separation or merger stage distributions of the mid-IR-identified AGNs and optically classified Seyferts.

39 15 10 Stage I-II 5 0 4 Stage III 2

Number 0 Stage IV-V 4 0 4 5 6 7 8 9 log( L[NeV] /Le )

Figure 3.7: Distribution of [NeV] luminosities grouped by merger stage for the combined sample of the non-U/LIRG pairs, the interacting LIRGs, and the ULIRGs

9

8 ) 8

(L 7 [NeV] 6 log( L 5

0 20 40 60 80 Pair Separation (kpc)

Figure 3.8: [NeV] luminosity vs. pair separation for the combined sample of the non- U/LIRG pairs, the interacting LIRGs, and the ULIRGs.

40 20 15 (a) Mid-IR identified AGN (b) Optically Identified AGN 15 10 10

Number Number 5 5

0 0 0 20 40 60 80 I II III IV V Pair Separation (kpc) Merger Stage 15 15 (c) (d)

10 10

Number 5 Number 5

0 0 4 5 6 7 8 9 -2.0 -1.5 -1.0 -0.5 0.0 0.5 log( L[NeV] /Le ) log([NeV]/[NeII])

Figure 3.9: Comparison of the properties of mid-IR identified AGN and optically classified Seyferts with [NeV] detections.

41 In Figures 3.8 (c) and (d), we plot the distributions of the [NeV] luminosity and the

[NeV] 14 µm / [NeII] 12 µm flux ratio for mid-IR identified AGNs and optically-classified

Seyferts. The [NeV] luminosity can be used as a proxy for the AGN bolometric luminosity

(e.g., Satyapal et al. 2007) and the [NeV] 14 µm / [NeII] 12 µm flux ratio is an indicator of the strength of the AGN relative to star formation within the Spitzer aperture (e.g., Genzel et al. 1998; Sturm et al. 2002; Dale et al. 2006; Dale et al. 2009; Satyapal et al. 2009).

As can be seen from Figures 3.8 (c) and (d), the distributions of [NeV] luminosity and the

[NeV]/[NeII] flux ratio are significantly different. According to K-S tests, the likelihood that the two samples are drawn from the same distribution is 3% and <0.1% respectively for the [NeV] luminosity and the [NeV]/[NeII] flux ratio distributions. The optically-classified

Seyferts are typically more luminous and have larger [NeV]/[NeII] flux ratios than the mid-

IR identified AGNs, suggesting that mid-IR spectroscopy identifies lower luminosity AGNs in hosts that have significant nuclear star formation.

3.7 Mid-IR Identified Dual AGN

Despite decades of searching, and strong theoretical reasons that they should exist, observa- tionally confirmed cases of close dual AGNs are extremely rare. Indeed, only 0.1% of quasars are found in pairs with projected separations of tens to hundreds of kpc (e.g. Hennawi et al. 2010), and only a handful of confirmed close dual AGNs in the local Universe (D<100

Mpc) are currently known (e.g. NGC 6240: Komossa et al. 2003, Mrk 463: Bianchi et al.

2008, Arp 299: Ballo et al. 2004, 3C 75: Owen et al. 1985). Recently more dual AGNs were discovered in follow-up studies of double-peaked [OIII] emitters selected from SDSS (Liu et al. 2010), but the fraction of interacting pairs with two AGNs is still low. There are several possible reasons to explain the scarcity of dual AGNs : 1) SMBHs are rarely active at the same time for the vast majority of merger parameters, 2) the fueling stage for the SMBHs occurs at separations that cannot be resolved by current facilities, or 3) dual SMBHs are optically invisible for a large fraction of the time when they are active. Given the scarcity 42 of observations, and the lack of extensive investigations that are carried out at wavelengths less sensitive to extinction, it is impossible to determine the true frequency of dual AGNs and to answer several key questions such as: What makes some mergers fuel dual SMBHs and other not? How does the onset, duration, and accretion rate of dual AGNs depend on the merger parameters and the properties of the host galaxy such as the bulge mass, star formation rate, and gas fraction?

We investigated the incidence of dual AGNs identified by mid-IR spectroscopy in our pairs sample. Of the galaxies that displayed infrared spectral signatures suggestive of

AGNs, we remarkably found 3 dual AGN candidates (labeled in Figure 3.1), Arp 120

(NGC4435/NGC4438), Arp 279 (NGC1253/NGC1253A), Arp 216 (NGC7679/NGC7682), where both galaxies in the pair show distinct mid-IR signatures of AGNs. One galaxy pair,

Arp 216, shows a [NeV] detection in both galaxies. Two other galaxy pairs likely contain dual AGN based upon the detection of [NeV] in one galaxy and an enhanced f[OIV]/f[NeII] line flux ratio or a low f[SIII]/f[SiII] line flux ratio (see Dale et al. 2009) in the other. As can be seen from Table 3.1, except for Arp216, only one of the galaxies in each pair is optically classified as a Seyfert. None of the others shows evidence for AGNs in the optical.

Most galaxies are disk galaxies; interestingly there are a few extremely late-type galaxies that typically do not harbor optically identified AGNs. The projected pair separation for these dual AGN candidates ranges from 18-74 kpc, significantly larger than the projected separations of the few known nearby dual AGNs NGC 6240, Mrk 463, Arp 299, as well as the SDSS sources from Liu et al. (2010). This may suggest that our pairs sample captures an earlier stage of AGN pairing that is overlooked in studies that target U/LIRGs, which are invariably more advanced mergers. Our observations suggest that a significant frac- tion of dual AGNs are optically unidentified and will be discovered through future infrared spectroscopic studies.

43 3.8 Summary and Conclusions

We have conducted a high-resolution mid-infrared spectroscopic study using Spitzer of an archival sample of 30 non-U/LIRG interacting galaxies, which generally target early merger stages. This is the first systematic mid-infrared spectroscopic investigation of a significant sample of non-U/LIRG interactions. Our main results can be summarized as follows:

1. Optical spectroscopic studies of interacting galaxies miss AGNs even in non-U/LIRG

interactions. Of the 11 AGN identified in the mid-infrared based on the detection of

the [NeV] 14 µm emission line, 4 were not detected optically.

2. Our mid-IR spectroscopy reveals that AGNs in our pairs sample are found in both

disturbed and undisturbed hosts, at a wide range of pair separations and Hubble

types.

3. Combining our study with previously published studies of U/LIRGs, we find tentative

evidence that the incidence of AGNs is bimodal with merger stage, with the highest

incidence found in the earliest and latest merger stages. Based on the [NeV] 14 µm lu-

minosity, a reliable proxy for the AGN bolometric luminosity, we also find that there

is a deficit of luminous AGNs in the earliest merger stages.

4. We find evidence based on mid-infrared spectroscopy for three possible dual AGNs in

our pairs sample, suggesting that a significant population of dual AGNs are hidden

in the optical.

44 Chapter 4: WISE Study of AGN in Galaxy Pairs

4.1 Introduction

Mergers of galaxies are common throughout the universe. The merging of galaxies and the interactions that precede these mergers create gravitational instabilities within the galaxies, often allowing vast quantities of dust and gas to be funneled into center of the galaxy. This excess of gas near the centers of these interacting galaxies often leads to an enhancement of star formation (e.g., Ellison et al. 2008). An example of this is seen in the high infrared luminosity ULIRGs, where extremely high star formation rates are observed in these recent mergers (e.g., Sanders & Mirabel 1996). In addition to enhancing star formation, some of this material may accrete onto a central black hole. A strong correlation between the properties of the bulge of the galaxy and the mass of the super-massive black hole at its center (Ferrarese & Merritt 2000; Gebhardt et al. 2000) may be an effect of galaxy mergers.

Merging galaxies and the inflow of material may allow for the simultaneous growth of both the bulge and the black hole, thereby forming this tight correlation. If this material reaches the center of the galaxy due to galaxy interactions, then active galactic nuclei (AGN) will be much more frequent in interacting galaxies and recent mergers. Many studies have shown strong evidence of an increased fraction of AGN in merging galaxies (e.g., Dahari 1985;

Ellison et al. 2011; Keel et al. 1985). Others have found that most AGN are in host galaxies that show no signs of recent interactions with other galaxies (e.g., Cisternas et al.

2011). The precise fraction of AGN that are triggered by interactions or the importance of the merger to the triggering of an AGN is still under debate. One limitation of many previous studies that are conducted in the optical or UV wavelengths is that obscuration may hide the AGN. Since galaxies in interactions have an excess of gas and dust around their

45 centers, the emission from the AGN may be obscured in the optical or UV. Here we will look for signs of AGN activity in the infrared, where the AGN emission may be observed. By comparing the AGN fraction observed in the infrared between pairs of galaxies and isolated galaxies, the fraction of AGN triggered or enhanced by interactions between galaxies can be estimated.

The mid-infrared can be used to identify AGN due to the infrared emission from dust in the vicinity of the AGN. The dust surrounding the AGN, thought to occur in a toroidal shape around the accreting supermassive black hole, reaches very high temperatures and emits strongly in the infrared. This infrared emission is then able to exit the galaxy without excessive absorption. The dust in the torus around the AGN and excessive amounts of dust and gas disturbed in some galaxy interactions, make the identification of AGN at other wavelengths difficult. To study the incidence of AGN in galaxy pairs we will use data from the Wide-field Infrared Survey Explorer (WISE), an all-sky survey conducted in the infrared.

The WISE all-sky survey allows for the ability to look at large samples of galaxies with a high depth of coverage, to separate AGN from non-AGN. From January 7, 2010 to August

6, 2010, WISE surveyed the entire sky in 4 bands; 3.4µm, 4.6µm, 12µm, and 22µm (referred to as W1, W2, W3, and W4 respectively). The resolution is 6.1′′, 6.4′′, 6.5′′ and 12′′ with estimated sensitivities of 0.08, 0.11, 1 and 6 mJy for these 4 bands respectively (Wright et al. 2010).

The reason the W1 and W2 WISE bands are efficient at separating star forming galaxies and AGN is clearly seen in Figure 4.1. Figure 4.1 displays the SED of a star-forming galaxy,

M82, with varying contributions of a template AGN SED. The SED of M82, representing star forming galaxies, has a dip in the region of the WISE bands, whereas the SED of the

AGN shows an increase in emission over the same region. The two main components of the star forming SED in the infrared is the stellar bump peaking at 1.6µm and thermal dust emission heated to temperatures of 25-50K by star formation that is responsible for longer wavelength emission. The WISE bands observe between these 2 regions which results in a

W1-W2 color near 0. As in the star forming galaxy, there is thermal dust emission present in

46 the AGN SED, however dust heated by the AGN reaches much higher temperatures resulting in significant emission into the WISE bands and filling the dip in the star forming galaxy

SED. Dust nearest the AGN in the torus may reach the dust sublimation temperature of

1000-1500 K. Dust heated to various temperatures depending on the distance to the AGN’s center result in a superposition of blackbodies producing a power law continuum beginning near 1µm. This power law continuum results in a red W1-W2 color. The bottom panel of

Figure 4.1 includes an AGN template with an extinction of Aν=2, demonstrating that the shape of the SED at the location of the WISE bands is largely unaffected by extinction and that WISE will be able to detect these obscured AGN. At shorter wavelengths towards the optical and UV, extinction strongly affects the shape of the SED of the AGN, making the nearly indistinguishable from the shape of a star-forming galaxy SED.

Problems with detecting the presence of an AGN with this method include contamina- tion of the infrared continuum emission by the emission of the galaxy, which may prevent the detection of the AGN. Since a galaxy with an AGN will have a composite SED composed of the SED of the AGN and the SED of the host galaxy, a weak AGN will be missed by

WISE. The variation in the shapes of the SEDs due to different AGN contributions can be seen in Figure 4.1. The effect of this on the W1-W2 color is illustrated in Figure 4.2. Fig- ure 4.2 from Figure 1 of Stern et al. (2012), is based on combinations of spectral template models of pure AGN and an elliptical galaxy from Assef et al. (2013). The W1-W2 color of the galaxy gets progressively bluer as the AGN contribution falls from 100% to 50%, and will eventually drop below 0.5 for weaker AGN. Therefore WISE will only select AGN that dominate the emission of the galaxy. In addition to contamination by host galaxy emission, redshift also has an effect on the WISE color since the WISE bands will sample different regions of the galaxy’s spectrum. This is demonstrated in Figure 4.2 where the WISE color is shown to vary depending on redshift. However, for our sample the redshift is limited to

0.2, so there will be little to no effect on the WISE colors of the galaxies in our sample.

47 Figure 4.1: Composite SEDs constructed from M82(red) and QSO1(purple) templates of Polletta et al. (2008) with AGN contribution increasing from 0% in red to 100% in purple. For the lower panel an extinction of Aν = 2 is applied to the QSO template using the Draine (2003) extinction law. The approximate location of WISE band W1(3.4µm) and W2(4.6µm) are marked on the figure. This figure is adapted from Figure 1 of Donley et al. (2012).

48 Figure 4.2: WISE colors of AGN as a function of redshift with increasing amounts of contamination from the host galaxy using templates from Assef et al. (2010). The templates used are of an unextincted AGN and of an elliptical galaxy. This figure is from Figure 1 of Stern et al. (2012).

49 4.2 Sample Selection

In this study we will use 3 samples of galaxies: a sample of galaxy pairs, a control sample matched to the galaxy pairs sample, and projected pairs sample to identify blending issues.

The Sloan Digital Sky Catalog (SDSS) DR7 will be used to create the samples of galaxies that will then be cross matched with the WISE catalog. The SDSS parent sample selection is similar to what is described in Ellison et al. (2011). We will use SDSS objects classified as a galaxy or quasar (specclass 2 and 3) that have extinction corrected r-band Petrosian magnitudes in the range 14.0 < mr < 17.77. Next a redshift range between 0.01 and 0.2 will be imposed to avoid non-cosmological values for redshift on the low end and to avoid incompleteness at higher . Limiting the redshift to 0.2 also prevents any effects of redshift on the W1-W2 color. Lastly a confidence level for the redshift of at least 70% is for inclusion into the parent SDSS sample. Stellar masses are calculated with mass to light ratios used in the MPA/JHU catalog, but using updated SDSS photometry found in

Simard et al. (2011).

From the parent sample, galaxies will be included in the pairs sample if they have a

−1 companion within 80 h70 kpc, ∆v < 500km/s, and a have stellar mass ratio of less than 10 to restrict the sample to major mergers. This sample is found to be biased to higher masses at small separations, due to high incompleteness for pairs closer than 55′′ caused by fiber collisions in SDSS. To account for this incompletion at small angular separations, a random 67.5% of pairs separated by more than 55′′ are excluded from the pairs sample.

This galaxy pair sample contains 16,556 galaxies. Next the control sample is compiled

−1 using the same parent sample using galaxies with no companion within 80 h70 kpc or 10,000 km/s. For each galaxy in the pair sample 10 control galaxies are matched in redshift and mass. Lastly, we will also include a sample of projected pairs or “fake pairs”, that

−1 have an apparent companion within 80 h70 kpc, but with ∆v>1000km/s to ensure that any results are unaffected by blending.

These SDSS pair, control, and projected pair samples are then cross-matched to the full

50 WISE all-sky survey catalog, matching within 6′′. Only galaxies that are detected in both the W1 and W2 band at > 3σare included in the analysis. These final samples consist of

15,853, 164,601, and 4,392 including 96%, 99%, and 99% of the original SDSS samples for the galaxy pair, control, and projected pair samples respectively.

4.3 Results

4.3.1 Selection of AGN by WISE

To select AGN using color we used W1-W2≥0.5 as a cutoff for selecting AGN. We deter- mined that 0.5 is sufficiently high enough to not include star forming galaxies in the AGN classification according to the galaxy SED templates from Assef et al. (2010) for galaxies in the redshift range of our sample. As can be seen in Figure 1 of Assef et al. (2013), W1-

W2=0.5 is above the range of star-forming galaxies at low redshifts. In the pairs sample of galaxies we found that 301 galaxies, or 2% of the sample have a W1-W2 color of 0.5 or greater, which we will refer to as “WISE AGN”. The fraction of galaxies that have W1-W2

≥ 0.5 does vary dramatically with pair separation as seen in Figure 4.3. The fraction of galaxies with W1-W2 ≥ 0.5 increases with decreasing pair separation. The WISE AGN fraction for the control sample is compared to the pairs sample where projected separation represents the separation of the pairs sample galaxy to which the control galaxy is matched.

As can be seen in Figure 4.3 the control sample has a low WISE AGN fraction over the entire range of projected separations demonstrating that the trend of increasing WISE AGN fraction with decreasing projected separations in the pairs sample is real and not biased with respect to mass or redshift. In the control sample 2,266 or 1% of the galaxies were

WISE AGN. As seen in Figure 4.3, the galaxy pairs of projected separations of up to 10kpc are approximately 4 times more likely to be a WISE AGN than the control galaxies matched to the >10 kpc pairs. By 60kpc the fraction of pairs that are WISE AGN approaches the fraction of WISE AGN in the control sample.

51 ftecnrlsml rud60kpc. infrared fraction around the AGN sample in the control approaching seen then the is separations what of pair to lowest similar the is at separation fractions Figure pair highest in the with shown with trend is the the separation seen, pair with be of can done function As was a 4.5. As as fraction AGN studies. optical optical the than fraction, AGN AGN of population different a a selects as the classified of are AGN most optical by time the found same of is 100 than only fraction 4.4 higher Figure a in sample, seen pairs the in AGN classified optically SNR extinction. an galactic and require absorption We into the stellar from takes for fluxes corrected which line been emission (2001) have use that will database al. We MPA/JHU et galaxies. starburst Kewley extreme by most the proposed sample even our scheme account For classification AGN. [NII] the or the use stars by will on photoionization we based is 1981) radiation ionizing al. of source et dominant (Baldwin diagram BPT a Using λ ratios. line optical on the SDSS compared by We AGN of Selection Optical 4.3.2 54/H / 6584

Fraction 0.00 0.04 0.08 0 λ α 10 iue43 rcino aaista aeaclro W1-W2 of color a have that galaxies of Fraction 4.3: Figure 53ad[OIII] and 6563 20 Pairs SampleW1-W2>0.5 Pair Separation(kpc) WISE 30 ≥ o l msinlnsfrotclcasfiain hr r 642 are There classification. optical for lines emission 4 all for 3 WISE 40 G rcint h rcino pial eetdANbased AGN selected optically of fraction the to fraction AGN 50 λ G r o pial lsie sAN Therefore AGN. as classified optically not are AGN 07/H / 5007 60 70 λ β 80 81ln ais ewl eemn hte the whether determine will we ratios, line 4861 52 Fraction 0.00 0.04 0.08 0 10 Control SampleW1-W2>0.5 20 Pair Separation(kpc) 30 WISE 40 50 G,wiea the at while AGN, ≥ 0.5 60 70 WISE 80 WISE WISE As . iue45 rcino aaista r lsie sa pia G codn othe to according AGN optical an as classified (2001). are al. that et Kewley galaxies of of scheme Fraction classification 4.5: Figure and AGN of line demarcation (2001). the al. with et sample Kewley pair by galaxy used the galaxies of star-forming diagram BPT 4.4: Figure

Fraction 0.00 0.04 0.08 Log([OIII]/Hb) 0 -1 0 1 10 -1.5 20 Pairs SampleOpticalAGN Pair Separation(kpc) 30 W1-W2 >0.5 Galaxy PairsSample 40 SF 50 -1.0 60 70 Log([NII]/H 80 53 Fraction

0.00 0.04 0.08 -0.5 0 10 a ) Control SampleOpticalAGN 20 Pair Separation(kpc) 30 0.0 40 50 60 AGN 70 0.5 80 4.3.3 The Effect of AGN Luminosity on AGN Selection by WISE

Most of the optical AGN in the galaxy pairs and control galaxy sample are not selected by WISE to be an AGN. To investigate the reason why so many optically classified AGN are not identified as by WISE we studied a sample of 576 Seyferts identified through the

Swift BAT 70-Month Hard X-ray Survey. Out of the 555 sources that were matched to a

WISE source within 6′′, 92 were not identified as an AGN by WISE. To determine the effect of luminosity on the classification by WISE as an AGN, we compared the WISE W1-W2 color to the X-ray luminosity of the AGN (see Figure 4.6). At X-ray luminosities greater than 1044 ergs s−1, 98% of the BAT AGN had a W1-W2 color greater than 0.5. However, only 84% of the AGN with X-ray luminosities between 1043 and 1044 ergs s−1 and 47% of the AGN with X-ray luminosities less than 1043 ergs s−1 were identified as AGN by WISE.

Therefore the detection rate of AGN by WISE is strongly correlated with the luminosity of the AGN, and appears to drop off quickly as luminosity decreases. This is consistent with contamination from galactic emission preventing the detection of lower luminosity AGN that do not dominate the emission of the galaxy.

We find a similar trend with WISE AGN fraction and AGN luminosity for the optically classified AGNs in our pairs sample. For the optical AGN in our pairs sample we used the [OIII] λ5007 emission line as an indicator of the luminosity of the AGN. The [OIII]

λ5007 line is found to be well correlated with the AGN bolometric luminosity in optically classified AGNs (Heckman et al. 2004). The [OIII] luminosities were available for all of the optical AGN in the pairs sample and ranged from 3.5 × 1038 to 4.3 × 1042 ergs s−1.

As with the BAT AGN we find a strong correlation between the AGN luminosity and the

WISE AGN fraction. As seen in Figure 4.7 the optical AGN that are WISE AGN tend to have higher [OIII] luminosities than optical AGN that are not WISE AGN. For [OIII] luminosities greater than 1042 ergs s−1 we find that 77% of the optical AGN are WISE

AGN. For luminosities between 1041 and 1042 ergs s−1, the WISE AGN fraction drops to

45% and to 4% for [OIII] luminosities less than 1041 ergs s−1. Based on this analysis, the

54 44 -1 Lx > 10 ergs s

2 43 -1 44 -1 10 ergs s < Lx ²10 ergs s

43 -1 Lx ² 10 ergs s

1 [3.4] - [4.6] (mag)

0

0 2 4 [4.6] - [12] (mag)

Figure 4.6: WISE color-color plot of hard X-ray selected BAT AGN

low luminosities of AGN in our sample is largely responsible for the low WISE AGN fraction since the WISE color cut is sensitive to the most powerful AGNs.

4.3.4 AGN Fraction in Galaxy Zoo Mergers

Next we compared our pairs sample to a sample of more advanced mergers taken from Darg et al. (2010b). The sample was obtained through the Galaxy Zoo project, an online system that allows the public to visually classify nearly a million galaxies using images from SDSS.

Each galaxy is viewed and classified by an average of 50 users as a spiral, elliptical, bad image, or merger. A weighted merger vote fraction (fm) from 0 to 1 is calculated for each galaxy based on the fraction of votes for merger weighted by the quality of the individual user. The quality of the user is determined by the users overall agreement to majority

55 G efidfrtecoetpis( pairs closest the for find we AGN tg aaisglx o egrglxe aeamth fte84idvda galaxies individual 814 the are Of of 4% 11% sample, match. only pair a our while have to sample, galaxies matched pairs merger 60% our zoo that to galaxy find matched galaxies we are 3 (2010a) galaxies stage merger al. zoo et galaxy Darg 1 in stage presented of stage merger the other Using each near mergers. too stage are galaxies separate the into that deblended is be this to for was possibility source One one mergers. only zoo pairs, galaxy were the the pairs of 124 merger For zoo sample. galaxy by pairs detected 3003 our the from of pairs 469 galaxy to interactions. matched of signs visible with mergers yls hn5 than separated, 5 less longer than by no more are by that separated with galaxies cores separated includes with are 2 but that Stage pairs for galaxies. includes classifications 1 the Stage stage between space merger (2010a). visible the al. et of Darg use from make pairs also pairs galaxy will these 3003 We of galaxy. sample companion final visible The a user. f particular with that objects by includes classified objects all for opinion are not that are sample that pairs AGN the in selected AGN optically selected of optically and of AGN luminosities [OIII] 4.7: Figure eto hssml n ace tt u apet n u-apeo oeadvanced more of sub-sample a find to sample our to it matched and sample this took We WISE ′′ . oteeaettlo 1 aaismthdbtenorpissml and sample pairs our between matched galaxies 814 of total are there so m

> Number

.,wt pcrlrdhfsbten0.005 between redshifts spectral with 0.4, 0 50 100 150 39 WISE < 0p)o u ar ape hs egr ihvisible with mergers These sample. pairs our of 10kpc) WISE 40 ore.Mthdglx ar edt oti early contain to tend pairs galaxy Matched sources. log(L[OIII] ergs/s) ′′ tg nldsmreswt oe separated cores with mergers includes 3 Stage . G,wihi iia otefato of fraction the to similar is which AGN, 56 41 WISE WISEAGN OpticalAGN 42 AGN 43 < z < .,adhave and 0.1, WISE WISE Table 4.1: WISE AGN fraction by Galaxy Zoo merger vote fraction (fm)

fm N WISE AGN fraction < 0.1 11854 0.014 0.1 - 0.2 1193 0.026 0.2 - 0.4 1191 0.023 0.4 - 0.6 809 0.036 ≥ 0.6 616 0.071

signs of interaction have a higher chance of containing a WISE AGN. However, since our pairs sample favors earlier stage mergers, we find a lower WISE AGN fraction for our galaxy pairs sample.

The majority of galaxies in our pairs sample have merger vote fractions of 0 and 75% have fractions less than 0.1. Therefore most of our pairs have little to no visible signs of interaction, meaning they are in the earliest stages of interaction. Darg et al. (2010b) found that for fm > 0.6 all systems were robust mergers so we will use this value to determine which galaxies are strongly interacting. In our pairs sample we find that 4% have a merger fraction greater than 0.6. We compared the WISE AGN fraction between the undisturbed galaxies with fm < 0.1 and merging galaxies with fm > 0.6. For galaxies with fm < 0.1 we find that only 1% of the galaxies are WISE AGN while for galaxies with fm > 0.6 we find that over 7% of galaxies are WISE AGN. WISE AGN fractions for values between are available in Table 4.1 which shows a steady increase in WISE AGN fraction with larger fm values. However for values less than 0.6 and especially for less than 0.4, misclassification may occur based on projection of nearby stars or galaxies onto the line of sight even if there are no visible signs of interaction (Darg et al. 2010b). Therefore we find that galaxies with visible signs of interaction are more likely to host WISE AGN.

57 4.4 Other Causes of Red WISE Colors

ULIRGs may also produce red W1-W2 colors (Wright et al. 2010) and therefore may be misclassified as an AGN, although many ULIRGs may host obscured AGNs. We matched our pairs sample to the 1 Jy sample, a flux-limited sample of ULIRGs from Kim & Sanders

(1998) and we found only 1 match. Our pairs sample contains only 1 ULIRG due to the deficiency of very close pairs in our pairs sample. A sub-sample of 74 ULIRGs from Veilleux et al. (2009) reveal that over half of the ULIRGs have separations of less than 3 kpc and 86% have separations less than 10kpc. In contrast only 3% of our pairs sample have separations less than 10kpc with none less than 3kpc. Therefore the fraction of WISE AGN in our pairs sample is not affected by the presence of ULIRGs with red W1-W2 colors. The main reason for the lack of ULIRGs is that our galaxy pairs sample does not include the most advanced mergers even in the lowest pair separation bin.

4.5 Reliability of Photometry in Close Pairs

Next we checked for the possibility that the increased fraction of WISE AGN at small separations was affected by blending of the emission from the 2 galaxies. Using the projected pairs sample we compared the AGN fraction by pair separation where we expect little variation since they are not real pairs. We determined the effect of the proximity of the other galaxy on the W1-W2 color for the projected pairs sample. Comparing the color for angular separations from 0 to 100′′, we see very little variation in the mean color based on angular separation. As seen in Table 4.2 the mean W1-W2 color differs by less than 0.02 mag between the closest (<20′′) and furthest (80′′-100′′) separations.

4.6 Conclusion

We have investigated the impact of interactions within galaxy pairs on the incidence of

AGN using WISE data on a large sample of galaxy pairs and a closely matched control

58 Table 4.2: Comparison between average W1-W2 color and angular separation for the pro- jected pairs sample

Separation N Mean W1-W2 Color arcsec mag 0-20 17 0.189 20-40 113 0.189 40-60 255 0.191 60-80 382 0.178 80-100 480 0.174

sample. By using the mid-infrared we are able to detect obscured AGN missed in optical studies. The main results of our study are summarized as follows:

1) We find that galaxy pairs have a significantly higher fraction of AGN in comparison to their controls. Overall we found 2% of the galaxy pairs have WISE selected AGN, while

1% of the control sample galaxies have WISE selected AGN.

2) We find that galaxy pairs with the smallest projected separations are more likely to host AGN. These galaxies have a WISE AGN fraction a factor of 4 higher than their matched controls.

3) WISE detects powerful AGN that dominate the emission of the galaxy. WISE was unable to detect optically identified Seyferts in our pairs sample with [OIII] luminosities less than 1040 ergs s−1.

4) WISE discovers a different population of AGN than optical studies since there is little overlap between the WISE AGN and optical AGN. Only a third of the WISE AGN were optically identified.

5) We find a higher incidence of AGN for more advanced mergers. Using merger fraction values from the Galaxy Zoo project, we find that galaxies with strong visible signs of interaction have a WISE AGN fraction of 7%.

59 Bibliography

Abel, N. P., & Satyapal, S. 2008, ApJ, 678, 686

Armus, L., et al. 2007, ApJ, 656, 148

Armus, L., et al. 2009, PASP, 121, 559

Arnaud, K. A. 1996, in Astronomical Society of the Pacific Conference Series, Vol. 101,

Astronomical Data Analysis Software and Systems V, ed. G. H. Jacoby & J. Barnes, 17

Assef, R. J., et al. 2010, ApJ, 713, 970

Assef, R. J., et al. 2013, ApJ, 772, 26

Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981, PASP, 93, 5

Ballo, L., Braito, V., Della Ceca, R., Maraschi, L., Tavecchio, F., & Dadina, M. 2004, ApJ,

600, 634

Barth, A. J., Strigari, L. E., Bentz, M. C., Greene, J. E., & Ho, L. C. 2009, ApJ, 690, 1031

Batcheldor, D., et al. 2005, ApJS, 160, 76

Bianchi, S., Chiaberge, M., Piconcelli, E., Guainazzi, M., & Matt, G. 2008, MNRAS, 386,

105

Boselli, A., et al. 2005, ApJ, 623, L13

Bushouse, H. A. 1986, AJ, 91, 255

Canalizo, G., & Stockton, A. 2001, ApJ, 555, 719

60 Cisternas, M., et al. 2011, ApJ, 726, 57

Colbert, E. J. M., Heckman, T. M., Ptak, A. F., Strickland, D. K., & Weaver, K. A. 2004,

ApJ, 602, 231

Cole, S., et al. 2001, MNRAS, 326, 255

Contini, M. 1997, A&A, 323, 71

Cox, T. J., Jonsson, P., Primack, J. R., & Somerville, R. S. 2006, MNRAS, 373, 1013

Dahari, O. 1985, AJ, 90, 1772

Dale, D. A., et al. 2005, ApJ, 633, 857

Dale, D. A., et al. 2006, ApJ, 646, 161

Dale, D. A., et al. 2009, ApJ, 693, 1821

Darg, D. W., et al. 2010a, MNRAS, 401, 1043

Darg, D. W., et al. 2010b, MNRAS, 401, 1552

Davies, R. I., Sugai, H., & Ward, M. J. 1997, MNRAs, 291, 314 de Vaucouleurs, G., de Vaucouleurs, A., & Corwin, Jr., H. G. 1976, Second reference cata-

logue of bright galaxies (Austin, TX: Univ. Texas Press) (RC2)

Desroches, L.-B., & Ho, L. C. 2009, ApJ, 690, 267

Di Matteo, T., Springel, V., & Hernquist, L. 2005, Nature, 433, 604

Donley, J. L., et al. 2012, ApJ, 748, 142

Dorman, B., & Arnaud, K. A. 2001, in Astronomical Society of the Pacific Conference Series,

Vol. 238, Astronomical Data Analysis Software and Systems X, ed. F. R. Harnden, Jr.,

F. A. Primini, & H. E. Payne, 415

61 Draine, B. T. 2003, ARA&A, 41, 241

Ellison, S. L., Patton, D. R., Mendel, J. T., & Scudder, J. M. 2011, MNRAS, 418, 2043

Ellison, S. L., Patton, D. R., Simard, L., & McConnachie, A. W. 2008, AJ 135, 1877

Farrah D., et al. 2007, ApJ, 667, 149

Ferrarese, L., & Merritt, D. 2000, ApJ, 539, L9

Fisher, D. B. 2006, ApJ, 642, L17

Fisher, D. B. 2006b, in Astronomical Society of the Pacific Conference Series, Vol. 352, New

Horizons in Astronomy: Frank N. Bash Symposium, ed. S. J. Kannappan, S. Redfield,

J. E. Kessler-Silacci, M. Landriau, & N. Drory, 237

Garc´ıa-Barreto,J. A., Franco, J., & Rudnick, L. 2002, AJ, 123, 1913

Garc´ıa-Barreto,J. A., Rudnick, L., Franco, J., & Martos, M. 1998, AJ, 116, 111

Gebhardt, K., et al., 2000, ApJ, 539, L13

Genzel R., et al. 1998, ApJ, 498, 579

Ghosh, H., Mathur, S., Fiore, F., & Ferrarese, L. 2008, ApJ, 687, 216

Gilfanov, M. 2004, MNRAS, 349, 146

Gliozzi, M., Foschini, L., Sambruna, R. M., & Tavecchio, F. 2008, A&A, 478, 723

Gliozzi, M., Papadakis, I. E., Grupe, D., Brinkmann, W. P., Raeth, C., & Kedziora-

Chudczer, L. 2010, ApJ, 717, 1243

Gliozzi, M., Satyapal, S., Eracleous, M., Titarchuk, L., & Cheung, C. C. 2009, ApJ, 700,

1759

Goulding, A. D., & Alexander, D. M. 2009, MNRAS, 398, 1165

62 Greene, J. E., & Ho, L. C. 2007, ApJ, 670, 92

G¨ultekin,K., Cackett, E. M., Miller, J. M., Di Matteo, T., Markoff, S., & Richstone, D. O.

2009, ApJ, 706, 404

G¨ultekin,K., Richstone, D. O., Gebhardt, K., Lauer, T. R., Tremaine, S., Aller, M. C.,

et al. 2009, ApJ, 698, 198

Hancock, M., Smith, B. J., Struck, C., Giroux, M. L., Appleton, P. N., Charmandaris, V.,

& Reach, W. T. 2007, AJ, 133, 676

Heckman, T. M. 1980, A&A, 88, 311

Hennawi, J. F., et al. 2010, ApJ, 719, 1672

Higdon, S. J. U., et al. 2004, PASP, 116, 975

Ho, L. C., Filippenko, A. V., & Sargent, W. L. W. 1997, ApJS, 112, 315

Ho, L. C., Greene, J. E., Filippenko, A. V., & Sargent, W. L. W. 2009, ApJS, 183, 1

Ho, L. C., et al. 2001, ApJ, 549, L51

Hopkins, P. F., Hernquist, L., Cox, T. J., Di Matteo, T., Robertson, B., & Springel, V.

2006, ApJS, 163, 1

Houck, J. R., et al. 2004, ApJS, 154, 18

Hughes, M. A., et al. 2005, AJ, 130, 73

Keel, W. C., Kennicutt, R. C., Jr., Hummel, E., & van der Hulst, J. M. 1985, AJ 90, 708

Kennicutt, Jr., R. C. 1998, ARA&A, 36, 189

Kennicutt, Jr., R. C., Roettiger, K. A., Keel, W. C., van der Hulst, J. M., & Hummel, E.

1987, AJ, 93, 1011

Kennicutt, Jr., R. C., et al. 2003, PASP, 115, 928 63 Kewley, L. J., Heisler, C. A., Dopita, M. A., & Lumsden, S. 2001, ApJS, 132, 37

Kim, D.-C., & Sanders, D. B. 1998, ApJS, 119, 41

Komossa, S., Burwitz, V., Hasinger, G., Predehl, P., Kaastra, J. S., & Ikebe, Y. 2003, ApJ,

582, L15

Koss, M., Mushotzky, R., Treister, E., Veilleux, S., Vasudevan, R., & Trippe, M. 2012, ApJ,

746, L22

Kuo, C. -Y., Lim, J., Tang, Y. -W., & Ho, P. T. P. 2008, ApJ, 679, 1047

Laine, S., Kotilainen, J. K., Reunanen, J., Ryder, S. D., & Beck, R. 2006, AJ, 131, 701

Lake, G., & Dressler, A. 1986, ApJ, 310, 605

Larson, R. B., & Tinsley, B. M. 1978, ApJ, 219, 46

Liu, J.-F., & Bregman, J. N. 2005, ApJS, 157, 59

Liu, X., Greene, J. E., Shen, Y., & Strauss, M. A. 2010, ApJ, 715, L30

Liu, X., Shen, Y., & Strauss, M. A. 2012, ApJ, 745, 94

Lusso, E., et al. 2010, A&A, 512, A34

Markoff, S., et al. 2008, ApJ, 681, 905

Mihos J., & Hernquist L. 1996, ApJ, 464, 641

Owen, F. N., O’Dea, C. P., Inoue, M., & Eilek, J. A. 1985, ApJ, 294, L85

Patton, D. R., Ellison, S. L., Simard, L., McConnachie, A. W., & Mendel, J. T. 2011,

MNRAS, 412, 591

Pence, W. D., Snowden, S. L., Mukai, K., & Kuntz, K. D. 2001, ApJ, 561, 189

Petric, A. O., et al. 2011, ApJ, 730, 28

64 Pogge, R. W. 1989, ApJS, 71, 433

Polletta, M., Weedman, D., H¨onig,S., Lonsdale, C. J., Smith, H. E., Houck, J. 2008, ApJ,

675, 960

Puxley, P. J., Hawarden, T. G., & Mountain, C. M. 1988, MNRAS, 234, 29

Ramos Almeida, C., Tadhunter, C. N., Inskip, K. J., Morganti, R., Holt, J., & Dicken, D.

2011, MNRAS, 410, 1550

Rothberg, B. & Joseph, R. D. 2006, AJ, 131, 185

Saha, A., Thim, F., Tammann, G. A., Reindl, B., & Sandage, A. 2006, ApJS, 165, 108

Sanders, D. B., & Mirabel, I. F. 1996, ARA&A, 34, 749

Satyapal, S., B¨oker, T., D., McAlpine, W., Gliozzi, M., Abel, N., & Heckman, T. 2009,

ApJ, 704, 439

Satyapal, S., Sambruna, R., & Dudik, R. 2004, A&A, 414, 825

Satyapal, S., Vega, D., Dudik, R., Abel, N., & Heckman, T. 2008, ApJ, 677, 926

Satyapal, S., Vega, D., Heckman, T., O’Halloran, B., & Dudik, R. 2007, ApJ, 663, L9

Secrest, N. J., Satyapal, S., Gliozzi, M., Cheung, C. C., Seth, A. C., & B¨oker, T. 2012, ApJ,

753, 38

Shier, L. M., & Fischer, J. 1998, ApJ, 497, 163

Silverman, J. D., et al. 2011, ApJ, 743, 2

Simard, L., Mendel, J. T., Patton, D. R., Ellison, S. L., & McConnachie, A. W. 2011, ApJS,

196, 11

Smith, B. J., & Struck, C. 2010, AJ, 140, 1975

65 Smith, B. J., Struck, C., Hancock, M., Appleton, P. N., Charmandaris, V., & Reach, W.

T. 2007, AJ, 133, 791

Steffen, A. T., Strateva, I., Brandt, W. N., Alexander, D. M., Koekemoer, A. M., Lehmer,

B. D., Schneider, D. P., & Vignali, C. 2006, AJ, 131, 2826

Stern, D., et al. 2012, ApJ, 753, 30

Str¨uder,L., et al. 2001, A&A, 365, L18

Sturm E., Lutz, D., Verma, A., Netzer, H., Sternberg, A., Moorwood, A. F. M., Oliva, E.,

& Genzel, R. 2002, A&A, 393, 821

Surace, J. A., Sanders, D. B., Vacca, W. D., Veilleux, S., & Mazzarella, J. M. 1998, ApJ,

492, 116

Tananbaum, H., et al. 1979, ApJ, 234, L9

Telesco, C. M., Dressel, L. L., & Wolstencroft, R. D. 1993, ApJ, 414, 120

Toomre. A., 1977, in Tinsley B.M., Larson R.B., eds, The Evolution of Galaxies and Stellar

Populations. Yale University Observatory, New Haven, p. 401

Tully, R. B. 1988, Nearby Galaxies Catalog (Cambridge: Cambridge Univ. Press)

Turner, M. J. L., et al. 2001, A&A, 365, L27

Veilleux, S., Kim, D.-C., & Sanders, D. B. 2002, ApJS, 143, 315

Veilleux, S., et al. 2009, ApJS, 182, 628

Vila, M. B., Pedlar, A., Davies, R. D., Hummel, E., & Axon, D. J. 1990, MNRAS, 242, 379

White, N. E., Nagase, F., & Parmar, A. N. 1995, in X-Ray Binaries, ed. W. H. G. Lewin,

J. van Paradijs, & E. P. J. van den Heuvel (Cambridge: Cambridge Univ. Press), 1

Woods, D. F., & Geller, M. J. 2007, AJ, 134, 527

66 Woods, D. F., Geller, M. J., & Barton, E. J. 2006, AJ, 132, 197

Wright, E. L., et al. 2010, AJ, 140, 1868

Yuan, F., & Cui, W. 2005, ApJ, 629, 408

Yuan, T.-T., Kewley, L. J., & Sanders, D. B. 2010, ApJ, 709, 884

67 Biography

William McAlpine recieved his Bachelor of Science in Physics from George Mason University in 2009. He went on to recieve his Doctor of Philosophy in Physics at George Mason University in 2013.

68