Subaru Adaptive-optics High-spatial-resolution Infrared K- and L'-band Imaging Search for Deeply Buried Dual AGNs in Merging Galaxies
aa r X i v : . [ a s t r o - ph . GA ] D ec Astrophysical Journal
Preprint typeset using L A TEX style emulateapj v. 11/26/04
SUBARU ADAPTIVE-OPTICS HIGH-SPATIAL-RESOLUTION INFRARED K - AND L ′ -BAND IMAGINGSEARCH FOR DEEPLY BURIED DUAL AGNS IN MERGING GALAXIES Masatoshi Imanishi and Yuriko Saito Subaru Telescope, 650 North A’ohoku Place, Hilo, Hawaii 96720, U.S.A.
Astrophysical Journal
ABSTRACTWe present the results of infrared K - (2.2 µ m) and L ′ -band (3.8 µ m) high-spatial-resolution ( < . ′′ K − L ′ compact sources, which are sensitive indicators of active galactic nuclei(AGNs), including AGNs that are deeply buried in gas and dust. We observed 29 merging systemsand confirmed at least one AGN in all but one system. However, luminous dual AGNs were detectedin only four of the 29 systems ( ∼ K -band stellar emission photometry in individual nuclei.We found that mass accretion rates onto SMBHs are significantly different among multiple SMBHs,such that larger-mass SMBHs generally show higher mass accretion rates when normalized to SMBHmass. Our results suggest that non-synchronous mass accretion onto SMBHs in gas- and dust-richinfrared luminous merging galaxies hampers the observational detection of kiloparsec-scale multipleactive SMBHs. This could explain the significantly smaller detection fraction of kiloparsec-scale dualAGNs when compared with the number expected from simple theoretical predictions. Our resultsalso indicate that mass accretion onto SMBHs is dominated by local conditions, rather than by globalgalaxy properties, reinforcing the importance of observations to our understanding of how multipleSMBHs are activated and acquire mass in gas- and dust-rich merging galaxies. Subject headings: galaxies: active — galaxies: nuclei — galaxies: Seyfert — galaxies: starburst —quasars: supermassive black holes — infrared: galaxies INTRODUCTION
Recent observations have revealed that supermassiveblack holes (SMBHs) are ubiquitously present at thecenter of galaxy spheroidal components, and that themasses of SMBHs and spheroidal stars are correlated(Magorrian et al. 1998; Ferrarese & Merritt 2000; Gul-tekin et al. 2009; McConnell & Ma 2013), suggesting thatSMBHs are an important galaxy ingredient whose forma-tion is closely related to galaxy formation. The widelyaccepted cold dark matter-based galaxy formation the-ories postulate that small gas-rich galaxies merge andgrow into massive galaxies, as seen in the present-dayuniverse (White & Rees 1978). If SMBHs are presentin the progenitor gas-rich small galaxies, then the merg-ing galaxies should have multiple SMBHs. In this case,kiloparsec-scale dual active galactic nuclei (AGNs) areexpected to be common if the mass accretion onto bothSMBHs is sufficiently high to create luminous observableAGNs (Colpi & Dotti 2011).Optical spectroscopic searches for AGNs with double-peaked emission lines (Wang et al. 2009; Liu et al. 2010;Smith et al. 2010; Liu et al. 2011; Pilyugin et al. 2012; Geet al. 2012; Barrows et al. 2013) and subsequent follow-up observations at other wavelengths have been exten-sively performed in the search for kiloparsec-scale dualAGNs (Fu et al. 2011; Rosario et al. 2011; Shen et al.2011; Tingay & Wayth 2011; Comerford et al. 2012; Fu
Electronic address: [email protected] Department of Astronomy, School of Science, Graduate Univer-sity for Advanced Studies (SOKENDAI), Mitaka, Tokyo 181-8588 et al. 2012; Liu et al. 2013). These studies have providedsome examples of kiloparsec-scale dual AGNs, but thedetected fraction of optically identifiable kiloparsec-scaledual AGNs ( < a few %) is significantly smaller than thenumber derived from the simple theoretical predictionthat the majority of galaxy mergers should have mul-tiple SMBHs and become dual AGNs if both SMBHsare sufficiently mass-accreting (Rosario et al. 2011; Yuet al. 2011). Several scenarios have been proposed toreconcile this discrepancy between theory and observa-tions, but it is still unclear which scenario is most likely.First, double-peaked emission is expected only in dual ac-tive SMBHs whose orbital planes are aligned relativelyedge-on along our line of sight. Those systems whoseorbital planes are aligned face-on are overlooked withthese methods (Rosario et al. 2011; Wang et al. 2012).Although this scenario may be able to explain the dis-crepancy by a factor of a few, it is probably difficult toaccount for the difference by more than an order of mag-nitude, provided that the alignment of the orbital planesof two active SMBHs is random in terms of our line-of-sight. A second, more plausible explanation is that mostAGNs in gas-rich galaxy mergers are deeply buried ingas and dust along virtually all directions (Hopkins et al.2005, 2006) and can become optically elusive (Maiolinoet al. 2003; Rosario et al. 2011). Third, it is also possiblethat even though multiple SMBHs are present, only oneof them has sufficient mass accretion to be observation-ally detectable as an AGN (i.e., non-synchronous SMBHmass accretion) during galaxy mergers (van Wassenhoveet al. 2012). If we are to unveil observationally the true Imanishi et al.fraction of kiloparsec-scale dual AGNs in gas- and dust-rich merging galaxies, it is of particular importance toapply observational methods that are sensitive to AGNswhose SMBHs orbital motion is relatively face-on alongour line-of-sight and to deeply buried AGNs.High-spatial-resolution imaging observations at thewavelengths of strong dust penetration are a powerfultool for this purpose because imaging observations can,in principle, preferentially detect AGNs with face-onmultiple SMBH orbit geometry, and buried AGNs aredetectable at such wavelengths. X-rays in the 2–10 keVrange have higher dust-penetrating power than the op-tical in the Galactic interstellar medium (Ryter 1996),and it is well known that an AGN is a much stronger X-ray emitter (relative to the bolometric luminosity) thana starburst (Ranalli et al. 2003; Shang et al. 2011), so X-ray observations are expected to be sensitive to AGNs,including obscured ones. In fact, strong dual AGN candi-dates were discovered from 2–10 keV X-ray observationsof several nearby gas- and dust-rich infrared luminousmerging galaxies (Komossa et al. 2003; Ballo et al. 2004;Bianchi et al. 2008; Piconcelli et al. 2010; Koss et al.2011; Fabbiano et al. 2011). Although the presence ofdual AGNs is strongly suggested, the observed 2–10 keVX-ray emission is, in most cases, only a scattered compo-nent behind Compton-thick (N H > cm − ) obscuringmaterial rather than a directly transmitted component.Thus, the intrinsic AGN X-ray luminosity is difficult toestimate because the scattering efficiency is unknown. Toextend the 2–10 keV X-ray dual AGN survey systemati-cally to gas- and dust-rich merging galaxies, the ChandraX-ray Observatory, with its spatial resolution of ∼ . ′′
5, isa particularly powerful tool for spatially resolving closelyseparated multiple AGNs. In fact, the Chandra X-rayObservatory has been used to find obscured dual AGNsin further gas- and dust-rich merging galaxies (Teng et al.2005; Iwasawa et al. 2011b; Teng et al. 2012; Koss et al.2012; Liu et al. 2013). However, the detected X-ray fluxesare, in many cases, so faint that detailed spectral anal-ysis is hampered. Both the scattered X-ray componentof Compton-thick obscured AGNs and the emission origi-nating from stars can reproduce the observed faint X-rayfluxes, often making it difficult to interpret the detectedX-ray emission as solid evidence for an AGN (Teng et al.2009; Iwasawa et al. 2011b; Liu et al. 2013). X-ray ob-servations at >
10 keV could directly detect transmittedX-ray emission from some mildly Compton-thick AGNs(Itoh et al. 2008), but the spatial resolution at >
10 keV isstill insufficient ( > ′′ ) (Harrison et al. 2013) to spatiallyresolve many interesting closely separated dual AGNs.High-spatial-resolution radio observation using theVLBI technique is another powerful tool to detect closelyseparated dual AGNs, as radio wavelengths are less sus-ceptible to dust extinction (Rodriguez et al. 2006). How-ever, the radio VLBI technique is sensitive only to a smallfraction of radio-loud AGN population, and it is not sen-sitive to the radio-quiet AGNs that comprise the ma-jority of the AGN population (Goldschmidt et al. 1999;White et al. 2000). The very small detectable fraction( < > µ m are potentially anothereffective (even improved) method for investigating buried dual AGNs in gas- and dust-rich merging galaxies. First,compared with the optical, dust extinction is substan-tially reduced ( < × A V ) (Nishiyama et al. 2008,2009). Additionally, AGNs, including both radio-loudand radio-quiet ones, are observationally distinguishablefrom starbursts. Strong polycyclic aromatic hydrocar-bons (PAH) emission features seen at 3–20 µ m are usu-ally observed in starbursts, but not in pure AGNs (Moor-wood 1986; Imanishi & Dudley 2000), due to the destruc-tion of PAHs by strong AGN X-ray emission (Voit 1992).In a pure AGN, a PAH-free continuum due to AGN-heated, large ( ∼ µ m) hot dust grains is observed.Thus, infrared spectroscopy can be a unique means tofind obscured AGNs by separating them from starburstactivity, as demonstrated by the successful detection ofmany buried AGNs in the brightest main nuclei of gas-and dust-rich merging galaxies (Imanishi et al. 2006a;Imanishi 2006; Imanishi et al. 2007; Armus et al. 2007;Imanishi et al. 2008; Nardini et al. 2008; Veilleux et al.2009; Imanishi 2009; Nardini et al. 2009; Imanishi et al.2010a,b; Nardini et al. 2010). Such infrared spectroscopywith high-spatial-resolution is, in principle, useful fordual AGN searches, but in practice, its application tofaint AGNs in secondary galaxy nuclei is limited by spec-troscopic sensitivity.However, these infrared spectroscopic observationshave clearly shown that the infrared 2–5 µ m continuumemission in AGNs is systematically redder than the emis-sion from starbursts (Imanishi et al. 2008; Sani et al.2008; Risaliti et al. 2010; Imanishi et al. 2010b). The ra-diative energy generation efficiency of an AGN (= massaccreting SMBH; 6–42% of Mc ) (Bardeen 1970; Thorne1974) is much higher than that of a starburst (= nu-clear fusion reaction inside stars; ∼ ). Thus,high luminosity can be generated from a very compactregion in an AGN. A larger amount of dust in the closevicinity ( <
10 pc) of an AGN can be heated to high tem-peratures with several 100 K (Barvainis 1987), producingstronger infrared L ′ -band (3.8 µ m) radiation than a star-burst whose infrared 2–5 µ m flux is usually dominatedby stellar photospheric blue emission. Observationally,pure AGNs are known to display strong L ′ -band flux ex-cess relative to the K -band (2.2 µ m) ( K − L ′ > K − L ′ ∼ K − L ′ sources. An im-portant point is that while simple high-spatial-resolutionimaging observations at wavelengths shorter than the K -band often cannot easily differentiate AGNs from com-pact starbursts for spatially compact emission at the dis-tance of interesting gas- and dust-rich merging galaxies(Scoville et al. 2000), high-spatial-resolution imaging atboth K and L ′ can better distinguish between AGNsand compact starbursts by combining morphological andcolor information. More importantly, this infrared imag-ing method is more sensitive than the infrared spectro-scopic method and thus is applicable to a larger numberof fainter secondary nuclei of gas- and dust-rich merginggalaxies, which is crucial for dual AGN studies. Giventhat the ratio of infrared dust extinction to X-ray ab-sorption (dust + gas) toward obscured AGNs is empiri-cally known to be much smaller than the Galactic valuenfrared AO imaging of merging galaxies 3(Alonso-Herrero et al. 1997; Granato et al. 1997; Faddaet al. 1998; Georgantopoulos et al. 2012), the infrared K and L ′ -band observations could be sensitive even toCompton-thick buried AGNs, as was demonstrated insome sources (Imanishi & Dudley 2000; Imanishi et al.2006a; Teng et al. 2005; Imanishi et al. 2008; Teng etal. 2009), making the infrared K - and L ′ -band imagingmethod particularly promising.The recent availability of laser-guide-star adaptive op-tics (LGS-AO) on ground-based 8–10-m telescopes hasenabled us to routinely achieve spatial resolutions of < . ′′ K - and L ′ -band images for the bulkof extra-galactic sources. Theoretical simulations of gas-and dust-rich galaxy mergers (Hopkins et al. 2005, 2006;van Wassenhove et al. 2012) predict that dual AGNs be-come luminous, particularly at the late merging stage,at separations of less than a few kiloparsecs. This phys-ical separation corresponds to < ′′ at z > K - and L ′ -band imaging method hasseveral advantages: (1) it has even better spatial reso-lution ( < . ′′
2) than the Chandra X-ray Observatory has( ∼ . ′′ K - and L ′ -band imaging observations of nearby gas- and dust-richgalaxy mergers using the Subaru 8.2-m telescope atopMauna Kea, Hawaii. This is one of the best sites in theworld for conducting highly sensitive L ′ -band observa-tions, due to high elevation ( ∼ = 71 km s − Mpc − , Ω M = 0.27, and Ω Λ = 0.73 (Komatsu et al.2009). The luminosity distance is calculated using thecosmological calculator created by Wright et al. (2006).In Section 2, we describe our sample population. In Sec-tion 3, we present our observations and data analysis,followed by results in Section 4. We discuss the impli-cations of our results in Section 5, and conclude with asummary of our work in Section 6. TARGET SELECTION
We primarily target luminous infrared galaxies(LIRGs) whose infrared luminosities exceed L IR > L ⊙ because they are a representative sample ofgas- and dust-rich galaxy mergers in the local universe(Sanders & Mirabel 1996). In gas-rich LIRGs, it is ex-pected that many SMBHs are mass accreting and sothe detection rate of multiple SMBHs through luminousAGN searches is higher than in gas-poor galaxy mergers.Previous systematic infrared spectroscopy revealedthat buried AGNs play an increasing energetic role withincreasing galaxy infrared luminosity and become par-ticularly important in ultraluminous infrared galaxies(ULIRGs) with L IR > L ⊙ (Imanishi 2009; Veilleuxet al. 2009; Imanishi et al. 2010a,b; Nardini et al. 2010).Thus, ULIRGs are of particular interest. Among AGN-hosting ULIRGs at the brightest nuclei, those with spa-tially resolved multiple nuclei in seeing-limited images(Kim et al. 2002) are our first targets because investigat- ing the presence of AGNs in the fainter secondary nucleiis a straightforward way to search for dual AGNs. In ad-dition to these, we also include apparently single-nucleusULIRGs in seeing-limited images because it may be pos-sible to detect closely separated dual AGNs in our high-spatial-resolution AO images. Multiple-nuclei mergingULIRGs without strong AGN signatures in the brightestmain nuclei are also added to see whether AGNs exist inthe fainter secondary nuclei.In addition to ULIRGs, we include five LIRGs forwhich previous 2–10 keV X-ray observations have sug-gested the presence of dual AGNs (Komossa et al. 2003;Bianchi et al. 2008; Piconcelli et al. 2010; Koss et al.2011; Fabbiano et al. 2011). Our aim in including thesetargets in our sample is to confirm that dual AGN signa-tures could be found using our infrared K - and L ′ -bandimaging method.In total, 29 merging systems were observed, and theirbasic properties are summarized in Table 1. Our sam-ple is neither homogeneous nor complete in a statisticalsense. Our goal is to determine whether our infraredmethod, sensitive to buried AGNs, results in a substan-tially higher fraction of dual AGN detections than previ-ous optically based methods, and to investigate whetherthe observed dual AGN fraction approaches unity, as pre-dicted by theories, in gas- and dust-rich merging galaxiesin the local universe. OBSERVATION AND DATA ANALYSIS
We used the IRCS infrared camera and spectrograph(Kobayashi et al. 2000) at the Nasmyth focus of theSubaru 8.2-m telescope atop Mauna Kea, Hawaii (Iye etal. 2004), together with the 188-element adaptive optics(AO) system, using laser-guide stars (LGS) or natural-guide stars (NGS) (Hayano et al. 2008, 2010). For theLGS-AO to work properly, we need to find a star orcompact object brighter than 18–19 mag in the opti-cal R -band within ∼ ′′ from the target to use as aguide star for good tip-tilt correction. The AO cor-rection itself is made using a laser spot with an op-tical R -band magnitude of ∼ R -band within ∼ ′′ from the target to serve a guide star for AO correc-tion. The number of external galaxies with such suitableNGS-AO guide stars is limited. Table 2 tabulates ourobservations, including information on guide stars usedto achieve good AO performance and standard stars forphotometric calibration. In summary, NGS-AO was usedduring the observing runs in 2011 June and 2013 May,and LGS-AO was adopted for all the remaining observingruns.During our Subaru AO observations, the sky was clear.The seeing in the K -band (2.2 µ m), measured in imagesbefore starting actual AO exposures, was 0 . ′′ . ′′ < < ∼ K -band, we employed the 52 mas (52.77 maspixel − ) imaging mode to observe simultaneously somewidely separated multiple galaxy nuclei and as manystars as possible to probe the point-spread function,within the field of view of 54 . ′′ × . ′′
04 (1024 pixels × − ) imaging mode to scatterthe bright nuclear glare into many pixels and avoid ar-ray saturation. In the L ′ -band, as Earth’s atmosphericbackground emission is high, we used the 20 mas imag-ing mode, whose field of view is 21 . ′′ × . ′′
06 in thefull-array mode. Even using this 20 mas mode, we hadto use a sub-array mode to avoid saturating the array forthe L ′ -band observations of some sources.For the K -band observations of target merging(U)LIRGs, the exposure times were 0.26–30 sec, and 2–120 coadds were applied. In the L ′ -band, the exposuretimes were 0.076–0.12 sec, with 250–300 coadds. The in-dividual exposure time was set so that signal levels atthe object positions were below the linearity level of theIRCS imaging array ( < K -band data taken in 2011 July, forwhich a five-point dithering pattern was employed.For all observing runs, photometric K - and L ′ -bandsstandard stars (Table 2) were observed, with a meanairmass difference of < . We first inspected the individual frames byeye. A very small fraction of frames showed strange pat-terns compared with the majority of the remaining nor- IRAF is distributed by the National Optical Astronomy Ob-servatories, which are operated by the Association of Universitiesfor Research in Astronomy, Inc. (AURA), under cooperative agree-ment with the National Science Foundation. mal frames. These strange frames were discarded fromour analysis. We created median-combined sky framesto make a sky flat image for each dataset. In this proce-dure, the positions of bright objects and bad pixels weremasked. Individual frames were sky-subtracted and di-vided by the sky flat frames for flat fielding. When theeffects of cosmic ray hits and bad pixels remained for afew pixels at this stage, we made corrections manuallyby replacing the signals of these pixels with the interpo-lated values of the surrounding pixels. Then, we shiftedthe sky-subtracted, flat-fielded images so that the peakposition of each target object landed on the same ar-ray pixel. For images that contained sufficiently brightcompact sources inside the field of view of individualframes, the pixel shift was determined using these brightcompact sources. The K -band frames often containedsuch sources. For a small fraction of the K -band dataand the majority of the L ′ -band data, however, no suchbright compact sources were seen in individual frames. Inthose cases, we used the offset values calculated from thedithering amplitude and pixel scale. We then average-combined these shifted frames to obtain final images.Based on a comparison of the FWHM values of compactobjects between the resulting combined images and in-dividual frames, we found this offset estimate to be veryaccurate for our AO data, due to the good quality of theAO guiding of the Subaru telescope. RESULTS K - and L ′ -band images Figure 1 presents the infrared K - and L ′ -band AO im-ages of the observed (U)LIRG nuclei. The achieved im-age sizes for stars and compact sources are usually 0 . ′′ . ′′ K - and L ′ -bands. Due to thehigher atmospheric background emission, the sensitivityin the L ′ -band is much lower (worse) than that in the K -band. Consequently, the detection significance in the L ′ -band should be much lower than that in the K -band,unless the sources are very red in K − L ′ . In Figure 1,the detection rate of merging (U)LIRG nuclei is smallerin the L ′ -band than in the K -band, and yet a signifi-cant fraction of the observed merging nuclei are clearlydetected in our highly sensitive L ′ -band AO images.In the L ′ -band, the background emission for space-based infrared satellites is much lower than that forground-based telescopes. So, as far as the detection ofa single source is concerned, observations using space-based infrared satellites with small apertures could alsobe useful. WISE (40 cm) and Spitzer (75 cm) infraredsatellites have imaging capabilities at 3.4 µ m (Wright etal. 2010) and 3.6 µ m (Fazio et al. 2004), respectively.These wavelengths are similar to our L ′ -band (3.8 µ m).However, the spatial resolution of WISE at 3.4 µ m andSpitzer at 3.6 µ m are ∼ . ′′ ∼ . ′′ L ′ (Arp 220 and IRAS 16474+3430) with Spitzer IRACcamera images at 3.6 µ m analyzed from archival data.Whereas multiple nuclei are clearly resolved in our AOimages, the nuclei are not resolved in the Spitzer IRACdata. It is clear that our ground-based AO images arenfrared AO imaging of merging galaxies 5better for our scientific purpose. Photometric aperture size K - and L ′ -band emission originating from buriedAGNs are dominated by AGN-heated hot dust locatedin the innermost part ( <
10 pc in physical scale or < . ′′ z > . ′′ . ′′ . ′′ . ′′ < µ m (Minowaet al. 2010, 2012) and varies depending on the brightnessof guide stars used for the LGS-AO tip-tilt correction orNGS-AO correction, their separation from the target ob-jects, and Earth’s atmospheric turbulence at the time ofobservations. Hence, if we extract signal only from theAO-core emission component (0 . ′′ . ′′ L ′ -band using standard stars and a com-pact ULIRG observed with NGS-AO. Figure 4 displaysthe same plot for any compact sources found in the sci-ence target data, observed with LGS-AO. Although AOperformance varies slightly among the different observingruns, 85–93% of total signal is usually recovered using a ∼ . ′′ × − )-radius aperture.Even in the data taken in 2012 October, when seeingwas most unstable among our observing runs in Table2, >
83% of the total signal is included within the se-lected aperture size. Thus, with the ∼ . ′′ L ′ -band LGS- andNGS-AO data.Figure 5 shows the growth of the curves of the encircledsignal in the K -band for standard stars and one ULIRGobserved with NGS-AO. Figure 6 displays the same plotfor any compact sources detected inside the science tar-get frames taken with LGS-AO. In the K -band, Earth’satmospheric turbulence effects are larger than in the L ′ -band. Also, the selected standard stars are generallyfainter in the optical R -band than the L ′ -band standardstars. Thus, poorer growth of curves may be anticipated,and yet 75–90% of signals are usually recovered with the ∼ . ′′ K -band.For L ′ -band data, a standard star was observed withand without NGS-AO on the same night (2011 August22 UT) (Figure 3, upper left). The AO data providehigher encircled signals at small radii than non-AO data,demonstrating the merit of AO for spatially unresolvedcompact source photometry using a small aperture.From the fact that 75–90% and 83–93% of the spa-tially unresolved compact source signals are consistentlyrecovered with a ∼ . ′′ K - and L ′ -band data, respectively, taken on different nights, un-der different Earth’s atmospheric conditions, and withdifferent guide star properties, we can safely concludethat Subaru AO, including both NGS-AO and LGS-AO,can provide such stable data, as long as the AO correc-tion performs sufficiently well. We thus apply the sameaperture size ( ∼ . ′′ K - and L ′ -band, respectively. Table 3summarizes the photometric measurements of individualmerging nuclei using a ∼ . ′′ ∼ . ′′ ∼ . ′′
5- radiusaperture, then the measured photometry could differ by ∼ ∼ . ′′ K -band and L ′ -band, respectively, the derived K − L ′ colorcould differ from the actual color by ∼ ∼ K - and L ′ -bands for the same galaxy nucleus is un-avoidable in our AO photometry of spatially unresolvedcompact emission. In general, since the signal fractionof compact emission within the ∼ . ′′ L ′ - than in the K -band (Figures 3–6), theestimated K − L ′ colors can be at most ∼ K − L ′ colors of compact source emissionin a redder sense is reduced somewhat by the fact thatthe spatially extended stellar contribution to an observedflux (within the ∼ . ′′ K -than in the L ′ -band. This partially compensates for theslightly higher missing flux of compact source emissionin the K - than in the L ′ -band. In summary, the mea-sured K − L ′ colors of spatially compact source emissionshould agree with the true colors within no redder than0.2 mag.As to the absolute flux of the compact source emission,the uncertainty is different from that of the K − L ′ color.While our ∼ . ′′ K - and L ′ -bands, respectively, a ∼ ′′ -radius aperture isbasically used for the photometry of AO-corrected stan-dard stars (point sources with virtually no spatially ex- Imanishi et al.tended emission) to recover as much point source signalas possible ( > ∼ . ′′ ∼ ′′ -radius aperture photometry of standard stars, the K - and L ′ -band compact source fluxes of merging galaxynuclei could be underestimated by as much as ∼ K -band and ∼ L ′ -band. Table 4 com-pares our nuclear 0 . ′′ . ′′ K -band pho-tometry, with 1 . ′′ µ m photometry by Scov-ille et al. (2000) for (U)LIRG nuclei observed by bothgroups. Our photometry tends to be fainter by a fewtenths of a magnitude, most likely because our apertureis slightly smaller, and some fraction of the seeing-sizedAO halo signal of compact source emission is not cov-ered. Hence, our AO photometry should not miss thespatially compact emission from the putative AGNs withmore than ∼ K -band and ∼ L ′ -band, which will not affect our main discussions andconclusions. DISCUSSION
Galaxy nuclei with luminous detectable AGNs
As mentioned in the introduction, AGNs should haveredder K − L ′ colors than starbursts due to the largeramount of hot (several 100 K) dust emission in the for-mer. Although the intrinsic K − L ′ colors could havesome scatter for individual starbursts and AGNs, weadopt the values for intrinsic K − L ′ color = 0.5 magfor starburst activity (Hunt et al. 2002) and K − L ′ =2.0 mag for AGNs (Ivanov et al. 2000; Alonso-Herrero etal. 2003; Videla et al. 2013). The observed K − L ′ colorsvary as a function of the AGN contribution to the ob-served flux, increasing (reddening) with increasing AGNcontribution. Our calculation shows that the K − L ′ colorbecomes > L ′ -band flux exceeds ∼ K − L ′ & K − L ′ & − L ′ -band flux is shown for (U)LIRG nuclei with K − L ′ = 0.5–2.0 mag. (U)LIRG nuclei with K − L ′ < > L ′ -band isvery sensitive to an AGN. Assuming the typical spectralenergy distribution of an AGN and a starburst, the L ′ -band-to-bolometric luminosity ratio of an AGN ( ∼ ∼ L ′ -band fluxcomes from the AGN (no dust extinction case). Even ifthe dust extinction of a buried AGN is larger by A V ∼
35 mag than the surrounding starbursts, 52% (34%) ofthe infrared L ′ -band flux originates in the buried AGN, if we adopt the dust extinction curve derived by Nishiyamaet al. (2008, 2009). In short, we should be able to de-tect moderately luminous buried AGNs even in galaxieswith coexisting strong starbursts. This is likely to be theprimary reason that our method allows the detection ofmany AGNs in the gas- and dust-rich merging (U)LIRGsthat usually accompany strong starburst activity.Next, we discuss an alternative possible mechanism forincreasing K − L ′ colors. Dust extinction can reddenthe colors of obscured starbursts. If the K − L ′ colorsbecome & K − L ′ > K - and L ′ -band wavelength range, with dust extinction at L ′ only 0.5–0.7 times the extinction at K (A L ′ = 0.5–0.7 × A K ) (Nishiyama et al. 2008, 2009; Gao et al. 2009).Adopting A K /A V = 0.062 (Nishiyama et al. 2008), evenin the case of dust extinction as large as A V = 10 mag(foreground screen dust absorption case), the K − L ′ col-ors change by only ∼ K − L ′ & K − L ′ & K − L ′ colors. Giventhat the dust extinction of buried-AGN-heated K - and L ′ -band emitting hot dust in the inner part of the dustyenvelope is most likely to be substantially larger thanthat in the surrounding starburst regions and that aforeground screen dust model is applied to buried AGNs(Imanishi et al. 2007), this effect may not be neglected.We now assume the dust extinction curve of A K /A V = 0.062 and A L ′ /A V = 0.031 derived by Nishiyama etal. (2008, 2009), and consider the two examples of A V = 16 mag and A V =32 mag as dust extinction for theburied-AGN-heated hot dust emission. Flux attenuationof starburst-origin K - and L ′ -band emission by dust ex-tinction is assumed to be negligible here. The buried-AGN-origin K − L ′ colors change from 2.0 mag to 2.5mag and 3.0 mag in the case of A V = 16 mag and 32mag, respectively, while the K − L ′ colors of starburstsremain unchanged as 0.5 mag. In this case, the observed K − L ′ colors become & L ′ -band fluxes are 44% and 41%, respectively, which aresmaller than no dust extinction case for the AGN-heatedhot dust emission ( ∼
50% AGN contribution is requiredto make K − L ′ & V = 16 mag and 32 mag dust extinction toward the buried-AGN-heated hot dust region, the intrinsic AGN-heatedhot dust emission luminosities will increase by a factorof 1.6 and 2.5, respectively, after correction for the fluxattenuation by dust extinction. When this correction isapplied, the fraction of the intrinsic AGN-origin L ′ -bandnfrared AO imaging of merging galaxies 7flux, relative to the observed L ′ -band flux, becomes 55%and 64% for the A V = 16 mag and 32 mag dust extinc-tion case, respectively. These fractions are even largerthan that in the case of no dust extinction for AGN-heated hot dust emission. Therefore, our argument thatimportant AGN contributions are required to reproducethe observed colors of K − L ′ & L ′ -band flux has been estimated (Table 3, column 5; no dustextinction case for the AGN-heated hot dust emission),we can derive the luminosity of the AGN-heated hotdust emission in the close vicinity ( <
10 pc) of an AGN,which dominates the AGN-originated L ′ -band flux. Asour main targets are optically elusive buried AGNs sur-rounded by dust and gas with a covering factor closeto unity, we here assume a simple spherical dust distri-bution. In this geometry, dust has a strong temperaturegradient. Inner (outer) dust has a higher (lower) temper-ature, and dust emission luminosity is conserved at eachtemperature from the hot inner regions to the cool outerregions (Imanishi et al. 2007; Imanishi 2009). The innerhot dust should generate most of the L ′ -band (3.8 µ m)emission, and the intrinsic AGN-origin L ′ -band luminos-ity ( ν L ν or λ L λ ), after subtracting the stellar contamina-tion, should be as large as the intrinsic AGN UV–opticalenergetic radiation luminosity at the very center.The estimated AGN luminosity is tabulated in Table3 (column 6). This AGN luminosity is derived from theobserved AGN-origin L ′ -band luminosity, with no dustextinction correction. Correction for possible dust ex-tinction of the AGN-origin L ′ -band emission will increasethe intrinsic AGN UV–optical energetic radiation lumi-nosity. Additionally, if the dust covering factor aroundan AGN is substantially below unity, the AGN-heatedhot dust emission luminosity underestimates the intrin-sic AGN UV–optical energetic radiation luminosity. Forthese reasons, the derived AGN luminosity in our methodshould be taken as a lower limit. Infrared properties of X-ray dual AGNs
Among the 29 observed infrared luminous merging sys-tems, signatures of dual AGNs with separations of > . ′′ K − L ′ & K − L ′ & K − L ′ color( < ∼ . ′′ K -band image shows a sin-gle nuclear morphology (Comerford et al. 2011). How-ever, no X-ray spectra are shown in this source, and the origin of the detected X-ray emission is less clear thanNGC 3393.X-ray observations are sensitive to an AGN, irrespec-tive of the presence of hot dust in the close vicinity of amass-accreting SMBH, and our infrared imaging methodrequires the presence of hot dust for AGN detection. Ifthe detected X-ray emission from both nuclei originatesin luminous AGNs, one nucleus in Mrk 463, Mrk 739,NGC 3393, and IRAS 20210+1121 could be a hot-dust-deficient AGN, in which case the contribution from AGN-heated hot dust emission to the observed L ′ -band fluxwould be small, and the observed infrared K - and L ′ -band fluxes would be dominated by nuclear stellar-originemission. Fraction of infrared dual AGNs
Among the 29 observed merging (U)LIRGs, only foursystems (Mrk 273, Arp 220, IRAS 16474+3430, and NGC6240) show red ( K − L ′ & ∼
14% (4/29). Because the bot-tom five sources in Table 1 are known dual AGNs fromprevious X-ray observations, their inclusion could biasthe fraction of detectable dual AGNs. However, onlyone out of the five sources (=20%) is an infrared dualAGN, not significantly biasing the total dual AGN frac-tion. The detected dual AGN fraction in our infraredimaging method is apparently slightly higher than thatof previous optical dual AGN surveys with <
5% (Liuet al. 2010; Shen et al. 2011), although the sample sizeis still small, and the sample selection criteria are dif-ferent. However, the detected dual AGN fraction is farbelow unity and is much smaller than the value expectedfrom the simple prediction that the majority of gas- anddust-rich merging galaxies should have multiple activeSMBHs.Because our infrared imaging method is sensitive toburied AGNs, it is unlikely that the small detectabledual AGN fraction is due to the elusiveness of AGNsobscured by dust (see § K -band photometry, including host galaxy emission(Kim et al. 2002; Skrutskie et al. 2006), for spatially re-solved multiple-nuclei merging systems in seeing-limitedimages. These K -band luminosities should reflect thestellar luminosity, and SMBH mass ratios in individualmerging nuclei are expected to be roughly proportional Imanishi et al.to the K -band stellar luminosity ratios (Marconi & Hunt2003; Vika et al. 2012).Figure 7 (Left) compares the K -band stellar luminosityratio (i.e., SMBH mass ratio) and nuclear L ′ -band lumi-nosity ratio (Table 5) between seeing-based spatially re-solved multiple nuclei. If both SMBHs in multiple nucleisystems have similar mass accretion rates when normal-ized to the SMBH mass, then the sources are expectedto be distributed around the solid line. However, mostsources are located far from the solid line, suggesting thatthe mass accretion rates per SMBH mass (= Eddingtonratio) are significantly different between the two nucleiin the majority of the observed multiple-nuclei mergingsystems. In Figure 7 (right), we also compare the K -band stellar luminosity ratio with AGN-origin L ′ -bandluminosity ratio after subtracting stellar contamination.In Figure 7 (left) and (right), some ambiguity remains.In the left panel, stellar emission could contribute to the L ′ -band flux, whereas in the right panel, the AGN sub-traction process could introduce some uncertainty. How-ever, both plots show basically the same behavior: mostsources largely deviate from the solid line, strongly sug-gesting that SMBH activation is non-synchronous. Thebulk of the observed merging (U)LIRGs are distributedalong the upper-left side of the solid line, indicating thatlarger-mass SMBHs generally have higher Eddington ra-tios than smaller-mass SMBHs.In this comparison, we need to include some cautionarystatements. First, the large-aperture K -band photome-try could include emission from AGNs, particularly forluminous AGNs that are weakly obscured by dust. Inthese systems, AGN-origin nuclear emission could con-tribute substantially to the observed K -band flux. Thismight result in an overestimate of the inferred SMBHmass from the observed K -band luminosity. However,since our primary targets are obscured AGNs in gas- anddust-rich merging (U)LIRGs, this effect is not expectedto be severe in most cases. In Figure 7, the bulk of theobserved sources are distributed around the upper-leftside of the solid line, a region in which larger SMBHsare more actively mass accreting (= higher Eddingtonratios). The AGN contribution to the K -band flux canbe high in such larger-mass SMBHs with high Edding-ton ratios if they are less dust obscured. If this effectis present in some sources, then the actual mass of thelarger SMBH is smaller than our estimate. In that case,the true Eddington ratios in larger-mass SMBHs wouldbecome even higher, increasing the non-simultaneity ofthe mass-accretion rates onto multiple SMBHs. Thus,our main conclusion does not change.Second, our method provides higher AGN luminos-ity in the L ′ -band brighter nucleus than in the L ′ -bandfainter nucleus in each merging system because the pos-sible dust extinction of AGN-heated hot dust emissionis not taken into account. Assuming that the Eddingtonratios are similar for SMBHs at two nuclei, if dust extinc-tion of AGN-heated hot dust is generally smaller in thelarger-mass SMBHs than in the smaller-mass SMBHs,then many sources would appear at the upper-left sideof the solid line in Figure 7. The trends observed inFigure 7 could be reproduced without introducing non-synchronous SMBH mass accretion. Similar Eddingtonratios would mean that larger-mass SMBHs have higherabsolute mass accretion rates, requiring a larger amount of fuel in the vicinity of the SMBHs, which could ob-scure the SMBHs. In gas- and dust-rich infrared lumi-nous merging galaxies, more luminous AGNs with higherabsolute mass accretion rates are predicted to be morehighly obscured (Hopkins et al. 2006), making this sec-ond scenario (similar Eddington ratios with less dust ex-tinction in the larger-mass SMBHs) unlikely, and stillsupporting the suggestion that larger-mass SMBHs havegenerally higher Eddington ratios.Third, when we convert K -band stellar luminosity toSMBH mass, we need to mention that younger starformation is likely to have higher K -band luminositythan older star formation for a given galaxy stellar mass(Bell & de Jong 2001). The inferred SMBH mass canbe overestimated for younger star formation. Activelymass-accreting SMBHs indicate the presence of dynam-ically settled nuclear gas, which is likely to cause activeyoung nuclear starbursts (Imanishi 2002; Imanishi et al.2003; Imanishi & Wada 2004; Oi et al. 2010; Imanishiet al. 2011a). Thus, the SMBH mass could be overesti-mated for active SMBHs with higher Eddington ratios.Now, Figure 7 suggests that larger-mass SMBHs havehigher Eddington ratios in general than do smaller-massSMBHs. If the abscissa is changed from K -band stel-lar luminosity to actual SMBH mass, many sources cur-rently located around the upper-left side of the solid linein Figure 7 would move leftward, even strengthening thescenario that larger mass SMBHs have higher Eddingtonratios. Thus, this uncertainty also will not change ourconclusion.Figure 8 plots the comparison of the nuclear L ′ -band togalaxy-wide K -band stellar luminosity ratio (Left) andAGN-origin L ′ -band to K -band stellar luminosity ratio(Right), between two nuclei, as a function of apparentnuclear separation. Since the K -band stellar luminosityis taken as the indicator of SMBH mass (Marconi & Hunt2003; Vika et al. 2012), the ordinate corresponds to theratio of SMBH-mass-normalized mass accretion rates (=Eddington ratio) between the two nuclei. Sources closeto (far away from) the solid horizontal lines mean thatSMBH-mass-normalized mass accretion rates are similar(largely different) between the two nuclei. If dual SMBHactivation preferentially occurs at a later merging stage(van Wassenhove et al. 2012), it is expected that sourceswith small apparent nuclear separation tend to be dis-tributed around the solid horizontal lines. No such trendis seen.Finally, our comparison of SMBH mass and AGN lu-minosity above is limited to (U)LIRGs with spatiallyresolved nuclei in seeing-limited images. Multiple ac-tive closely-separated SMBHs ( << ′′ ), which are spa-tially resolvable only with our AO images (e.g., IRAS16474+3430 in Figure 1, L ′ -band image), are very inter-esting, in terms of the theoretical prediction that suchSMBHs in the late stages of gas- and dust-rich galaxymergers become particularly active and become luminousAGNs (Hopkins et al. 2005, 2006; van Wassenhove etal. 2012). However, although AGN luminosity ratios forclosely separated systems could be derived from our AO-assisted high-spatial-resolution infrared K - and L ′ -bandimaging data (Table 3), SMBH mass ratios are difficultto obtain from stellar emission luminosity in K -band im-ages because host galaxy’s stellar emission strongly over-laps between multiple nuclei. Spatially resolved veloc-nfrared AO imaging of merging galaxies 9ity information obtained with AO-assisted spectroscopywill be useful for inferring SMBH ratios in such small-separation SMBH systems, but such observations are stilllimited to very bright nuclei only (Medling et al. 2011; Uet al. 2013). Investigating the properties of these closelyseparated SMBHs requires additional future work.In summary, our results support the scenario pro-posed by the theory (van Wassenhove et al. 2012)that SMBH mass accretion is not simultaneous amongmultiple SMBHs in gas- and dust-rich merging galax-ies. In general, larger-mass SMBHs are more activelymass accreting (normalized to SMBH mass) in merging(U)LIRGs with multiple nuclei in seeing-limited images.This non-synchronous SMBH activation may reduce thefraction of observable dual AGN, compared to the frac-tion of multiple SMBHs, in merging galaxies. Our resultssuggest that mass accretion onto SMBHs is dominatedby the local environment on the small scale rather thanby global galaxy properties, even in gas- and dust-richinfrared luminous merging galaxies. In this case, it isnot easy to predict SMBH activity through theoreticalsimulations of galaxy mergers. Thus, observations areimportant for understanding how multiple SMBHs areactivated during the gas- and dust-rich galaxy mergerprocess. SUMMARY
We conducted infrared K - and L ′ -band high-spatial-resolution ( < . ′′
2) imaging observations of nearbyinfrared luminous merging galaxies using SubaruLGS/NGS-AO to search for kiloparsec-scale multipleAGNs surrounded by dust through the detection of red K − L ′ compact sources. Given the gas- and dust-rich na-ture of these galaxies, many SMBHs are expected to bemass accreting and hence to become luminous AGNs, butthese AGNs are deeply buried in gas and dust. Using ourinfrared method, which is sensitive to buried AGNs, theobservational detection of multiple SMBHs is expected tobe more feasible than in gas-poor galaxy mergers wheremany SMBHs may not be actively mass accreting due tothe paucity of surrounding gas. We present the followingmain conclusions.1. Among 29 observed merging systems, at least oneAGN was found in all sources except one, demon-strating the effectiveness of our method for the pur-pose of AGN detection in these gas- and dust-richinfrared luminous merging galaxies.2. Kiloparsec-scale dual AGNs were seen in only fourof 29 galaxies, even using our powerful method,which is sensitive to deeply buried AGNs. Thisfraction seems slightly higher than the fraction de-termined by previously published optical spectro-scopic dual AGN searches, despite the small samplesize and the differences in criteria used for sampleselection. However, the fraction is still significantlysmaller than the value derived from the simple the-oretical prediction that most gas- and dust-richmerging galaxies are expected to contain multipleactive SMBHs.3. The AGN luminosity ratios derived from AGN-origin L ′ -band emission between two nuclei are,in most cases, higher than the SMBH mass ratios inferred from large-aperture K -band photometricobservations. When normalized to SMBH mass,larger-mass SMBHs are generally more highly massaccreting than are smaller-mass SMBHs in mostof the observed infrared luminous merging galaxieswith spatially resolved nuclei in seeing-limited im-ages. This trend is independent off the apparentnuclear separation.4. When combined with previous optically based dualAGN searches, our observational results suggestthat the most likely reason for the small ob-served dual AGN fraction in merging galaxies isthat mass accretion onto multiple SMBHs is non-simultaneous rather than the result of the orbitinggeometry of multiple SMBHs or the optical elusive-ness of AGNs deeply buried in gas and dust.5. Our results suggest that in gas- and dust-rich in-frared luminous merging galaxies, mass accretiononto SMBHs is primarily determined by local con-ditions rather than by global galaxy properties.This makes theoretical prediction difficult and ne-cessitates the inclusion of observational constraintswhen attempting to understand what is happeningfor SMBHs in gas- and dust-rich galaxy mergers.We thank the anonymous referee for his/her usefulcomment. We are grateful to Drs. Minowa and Ishiifor their support during our observations at the Sub-aru Telescope and to Sayaka Yamaguchi for her En-glish proofreading. M.I. is supported by a Grant-in-Aid for Scientific Research (23540273). This researchmade use of (1) the SIMBAD database, operated at CDS,Strasbourg, France, and the NASA/IPAC ExtragalacticDatabase (NED) operated by the Jet Propulsion Lab-oratory, California Institute of Technology, under con-tract with the National Aeronautics and Space Admin-istration, (2) data products from the Two Micron AllSky Survey, which is a joint project of the University ofMassachusetts and the Infrared Processing and AnalysisCenter/California Institute of Technology, funded by theNational Aeronautics and Space Administration and theNational Science Foundation, and (3) the NASA/ IPACInfrared Science Archive, which is operated by the JetPropulsion Laboratory, California Institute of Technol-ogy, under contract with the National Aeronautics andSpace Administration.0 Imanishi et al. REFERENCESAlonso-Herrero, A., Quillen, A. C., Rieke, G. H., Ivanov, V., &Efstathiou, A. 2003, AJ, 126, 81Alonso-Herrero, A., Ward, M. J., & Kotilainen, J. K. 1997,MNRAS, 288, 977Armus, L., Charmandaris, V., Bernard-Salas, J., et al. 2007, ApJ,656, 148Balestra, I., Boller, T., Gallo, L., Lutz, D., & Hess, S. 2005, A&A,442, 469Ballo, L., Braito, V., Della Ceca, R., et al. 2004, ApJ, 600, 634Barvainis, R., 1987, ApJ, 320, 537Bardeen, J. M. 1970, Nature, 226, 64Barrows, R., Lacy, C. H. S., Kennefick, J., et al. 2013, ApJ, 769,95Bell, E. F., & de Jong, R. S. 2001, ApJ, 550, 212Beswick, R. J., Pedlar, A., Mundell, C. G., Gallimore, J. F. 2001,MNRAS, 325, 151Bianchi, S., Chiaberge, M., Piconcelli, E., Guainazzi, M., & Matt,G. 2008, MNRAS, 386, 105Braito, V., Della Ceca, R., Piconcelli, E., et al. 2004, A&A, 420,79Burke-Spolaor, S. 2011, MNRAS, 410, 2113Colpi, M., & Dotti, M. 2011, Advanced Science Letters, 4, 181Comerford, J. M., Gerke, B. F., Stern, D., et al. 2012, ApJ, 753,42Comerford, J. M., Pooley, D., Gerke, B. F., & Madejski, G. M.2011, ApJ, 737, L19Dasyra, K. M., Tacconi, L. J., Davies, R. I., et al. 2006a, ApJ, 638,745Dasyra, K. M., Tacconi, L. J., Davies, R. I., et al. 2006b, ApJ, 651,835Diaz, A. I., Prieto, M. A., Wamsteker, W. 1988, A&A, 195, 53Downes, D., & Eckart, A. 2007, A&A, 468, L57Dudley, C. C., & Wynn-Williams, C. G. 1997, ApJ, 488, 720Fabbiano, G., Wang, J., Elvis, M., & Risaliti, G. 2011, Nature, 477,431Fadda, D., Giuricin, G., Granato, G. L., & Vecchies, D. 1998, ApJ,496, 117Farrah, D., Bernard-Salas, J., Spoon, H. W. W., et al. 2007, ApJ,667, 149Fazio, G. G., Hora, J. L., Allen, L. E., et al. 2004, ApJS,154, 10Ferrarese, L., & Merritt, D. 2000, ApJ, 539, L9Forster Schreiber, N. M., Genzel, R., Lutz, D., Kunze, D., &Sternberg, A. 2001, ApJ, 552, 544Fu, H., Myers, A., Djorgovski, S. G., & Yan, L. 2011, ApJ, 733,103Fu, H., Yan, L., Myers, A., et al. 2012, ApJ, 745, 67Gallimore, J, F., & Beswick, R. 2004, AJ, 127, 239Gao, J., Jiang, B. W., & Li, A. 2009, ApJ, 707, 89Ge, J-Q., Hu, C., Wang, J-M., Bai, J-M., & Zhang, S. 2012, ApJS,201, 31Georgantopoulos, I., Dasyra, K. M., Rovilos, E., et al. 2011, A&A,531, 116Goldschmidt, P., Kukula, M. J., Miller, L., & Dunlop, J. S. 1999,ApJ, 511, 612Granato, G. L., Danese, L., & Franceschini, A. 1997, ApJ, 486, 147Gultekin, K., Richstone, D. O., Gebhardt, K., et al. 2009, ApJ,698, 198Harrison, F. A., Craig, W. W., Christensen, F. E., et al. 2013, ApJ,770, 103Hayano, Y., Takami, H., Guyon, O., et al. 2008, SPIE, 7015, 25Hayano, Y., Takami, H., Oya, S., et al. 2010, SPIE, 7736, 21Hopkins, P. F., Hernquist, L., Cox, T. J., et al. 2005, ApJ, 630, 705Hopkins, P. F., Hernquist, L., Cox, T. J., et al. 2006, ApJS, 163, 1Hunt, L. K., Giovanardi, C., & Helou, G. 2002, A&A, 394, 873Hutchings, J. N., & Neff, S. G. 1989, AJ, 97, 1306Imanishi, M. 2002, ApJ, 569, 44Imanishi, M. 2003, ApJ, 599, 918Imanishi, M. 2006, AJ, 131, 2406Imanishi, M. 2009, ApJ, 694, 751Imanishi, M., & Dudley, C. C. 2000, ApJ, 545, 701Imanishi, M., Dudley, C. C., & Maloney, P. R. 2001, ApJ, 558, L93Imanishi, M., Dudley, C. C., & Maloney, P. R. 2006a, ApJ, 637,114 Imanishi, M., Dudley, C. C., Maiolino, R., et al. 2007, ApJS, 171,72Imanishi, M., Ichikawa, K., Takeuchi, T., et al. 2011a, PASJ, 63,S447Imanishi, M., Imase, K., Oi, N., & Ichikawa, K. 2011b, AJ, 141,156Imanishi, M., Maiolino, R., & Nakagawa, T. 2010a, ApJ, 709, 801Imanishi, M., Nakagawa, T., Ohyama, Y., et al. 2008, PASJ, 60,S489Imanishi, M., Nakagawa, T., Shirahata, M., Ohyama, Y., & Onaka,T. 2010b, ApJ, 721, 1233Imanishi, M., & Terashima, Y. 2004, AJ, 127, 758Imanishi, M., Terashima, Y., Anabuki, N., & Nakagawa, T. 2003,ApJ, 596, L167Imanishi, M., & Wada, K. 2004, ApJ, 617, 214Itoh, T., Done, C., Makishima, K., et al. 2008, PASJ, 60, S251Ivanov, V. D., Rieke, G. H., Groppi, C. E., et al. 2000, ApJ, 545,190Iye, M., Karoji, H., Ando, H., et al., 2004, PASJ, 56, 381Iwasawa, K.,Mazzarella, J. M., Surace, J. A., et al. 2011a, A&A,528, 137Iwasawa, K., Sanders, D. B., Teng, S. H., et al. 2011b, A&A, 529,106Kim, D. -C., & Sanders, D. B. 1998, ApJS, 119, 41Kim, D. -C., Veilleux, S., & Sanders, D. B. 2002, ApJS, 143, 277Kobayashi, N., Tokunaga, A. T., Terada, H., et al. 2000, IRCS:Infrared camera and spectrograph for the Subaru Telescope, inProc. SPIE 4008: Optical and IR Telescope Instrumentation andDetectors, eds M. Iye & A. F. Moorwood, 1056Komatsu, E., Dunkley, J., Nolta, M. R., et al. 2009, ApJS, 180,330Komossa, S., Burwitz, V., Hasinger, G., et al. 2003, ApJ, 582, L15Koss, M., Mushotzky, R., Treister, E., et al. 2011, ApJ, 735, L42Koss, M., Mushotzky, R., Treister, E., et al. 2012, ApJ, 746, L22Liu, X., Civano, F., Shen, Y., et al. 2013, ApJ, 762, 110Liu, X., Shen, Y., Strauss, M., Greene, J. E. 2010, ApJ, 708, 427Liu, X., Shen, Y., Strauss, M. A., & Hao, L. 2011, ApJ, 737, 101Magorrian, J., Tremaine, S., Richstone, D., et al. 1998, ApJ, 115,2285Maiolino, R., Comastri, A., Gilli, R., et al. 2003, MNRAS, 344, L59Maloney, P. R., & Reynolds, C. S. 2000, ApJ, 545, L23Marconi, A., & Hunt, L. K. 2003, ApJ, 589, L21Max, C. E., Canalizo, G., & de Vries, W. H. 2007, Science, 316,1877McConnell, N. J. & Ma, C-P. 2013, ApJ, 764, 184McLeod, K. K., Rieke, G. H., Rieke, M. J., & Kelly, D. M. 1993,ApJ, 412, 111Medling, A. M., Ammons, S. M., Max, C. E., et al. 2011, ApJ, 743,32Minowa, Y., Hayano, Y., Oya, S., et al. 2010, SPIE, 7736, 77363NMinowa, Y., Hayano, Y., Terada, H., et al. 2012, SPIE, 8447,84471FMoorwood, A. F. M. 1986, A&A, 166, 4Nardini, E., Risaliti, G., Salvati, M., et al. 2008, MNRAS, 385,L130Nardini, E., Risaliti, G., Salvati, M., et al. 2009, MNRAS, 399,1373Nardini, E., Risaliti, G., Watabe, Y., Salvati, M., & Sani, E. 2010,MNRAS, 405, 2505Nishiyama, S., Nagata, T., Tamura, M., et al. 2008, ApJ, 680, 1174Nishiyama, S., Tamura, M., Hatano, H., et al. 2009, ApJ, 696, 1407Oi, N., Imanishi, M., & Imase, K. 2010, PASJ, 62, 1509Perez, E., Manchado, A., Garcia-Lario, P., & Pottash, R. 1990,A&A, 227, 407Piconcelli, E., Vignali, C., Bianchi, S., et al. 2010, ApJ, 722, L147Pilyugin, L. S., Zinchenko, I. A., Cedres, B., et al. 2012, MNRAS,419, 490Puxley, P. J. 1991, MNRAS, 249, 11pRanalli, P., Comastri, A., & Setti, G. 2003, A&A, 399, 39Risaliti, G., Gilli, R., Maiolino, R., & Salvati, M. 2000, A&A, 357,13Risaliti, G., Imanishi, M., & Sani, E. 2010, MNRAS, 401, 197Rodqiguez, C., Taylor, G. B., Zavala, R. T., et al. 2006, ApJ, 646,49 nfrared AO imaging of merging galaxies 11
Rosario, D. J., McGurk, R. C., Max, C. E., et al. 2011, ApJ, 739,44Ryter, C. E. 1996, ApSS, 236, 285Sanders, D. B., & Mirabel, I. F. 1996, ARA&A, 34, 749Sanders, D. B., Soifer, B. T., Elias, J. H., Neugebauer, G., &Matthews, K. 1988, ApJ, 328, L35Sani, E., Risaliti, G., Salvati, M., et al. 2008, ApJ, 675, 96Scoville, N. Z., Evans, A. S., Thompson, R., et al. 2000, AJ, 119,991Severgnini, P., Risaliti, G., Marconi, A., Maiolino, R., & Salvati,M. 2001, A&A, 368, 44Shang, Z., Brotherton, M. S., Wills, B. J., et al, 2011, ApJ, 196, 2Shen, Y., Liu, X., Greene, J. E., & Strauss, M. A. 2011, ApJ, 735,48Shuder, J. M., & Osterbrock, D. E. 1981, ApJ, 250, 55Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131,1163Smith, K. L., Shields, G. A., Bonning, E. W., et al. 2010, ApJ, 716,866Soifer, B. T., Neugebauer, G., Matthews, K., et al. 2000, AJ, 119,509Teng, S. H., Wilson, A. S., Veilleux, S., et al. 2005, ApJ, 633, 664Teng, S. H., Veilleux, S., Anabuki, N., et al. 2009, ApJ, 691, 261Teng, S. H., Schawinski, K., Urry, C. M., et al. 2012, ApJ, 753, 165Thorne, K. S. 1974, ApJ, 191, 507Tingay, S. J., & Wayth, R. B. 2011, ApJ, 141, 174U., B., Medling, A., Sanders, D. B., et al. 2013, ApJ, 775 115Ueno, S., Koyama, K., Awaki, H., Hayashi, I., & Blanco, P. R. 1996PASJ 48 389 van Wassenhove, S., Volonteri, M., Mayer, L., et al. 2012, ApJ, 748,L7Veilleux, S., Kim, D. -C., & Sanders, D. B. 1999a, ApJ, 522, 113Veilleux, S., Kim, D. -C., Sanders, D. B., Mazzarella, J. M., &Soifer, B. T. 1995, ApJS, 98, 171Veilleux, S., Sanders, D. B., & Kim, D. -C. 1997, ApJ, 484, 92Veilleux, S., Sanders, D. B., & Kim, D. -C. 1999b, ApJ, 522, 139Veilleux, S., Rupke, D. S. N., Kim, D.-C., et al. 2009, ApJS, 182,628Videla, L., Lira, P., Andrews, H., et al. 2013, ApJS, 204, 23Vika, M., Driver, S. P., Cameron, E., Kelvin, L., & Robotham, A.2012, MNRAS, 419, 2264Voit, G. M. 1992, MNRAS, 258, 841Wang, J-M., Chen, Y-M., Hu, C., et al. 2009, ApJ, 705, L76Wang, X. -W., & Zhou, H. -Y. 2012, ApJ, 757, 124White, R. L., Becker, R. H., Gregg, M. D., et al. 2000, ApJS, 126,133White, S. D. M., & Rees, M. J. 1978, MNRAS, 183, 341Wright, E. L. 2006, PASP, 118, 1711Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., et al. 2010,AJ, 140, 1868Xia, X. Y., Xue, S. J., Mao, S., et al. 2002, ApJ, 564, 196Yu, Q., Lu, Y., Mohayaee, R., & Colin, J. 2011, ApJ, 738, 92Yuan, T. -T., Kewley, L. J., Sanders, D. B. 2010, ApJ, 709, 884Zhou, S., Wynn-Williams, C. G., & Sanders, D. B. 1993, ApJ, 409,149
TABLE 1Properties of the Observed Infrared Luminous Merging Galaxies
Object z f f f f log L IR Optical AGN in the(Jy) (Jy) (Jy) (Jy) (L ⊙ ) Class main nucleus ?(1) (2) (3) (4) (5) (6) (7) (8) (9)IRAS 00091 − < a (cp b ) Y , , IRAS 00188 − < a (Sy2 b ) Y , , , , IRAS 05024 − a,b Y , IRAS 05189 − a,b Y , , , , , , , , , , , IRAS 08572+3915 0.058 0.32 1.70 7.43 4.59 12.1 LI a (Sy2 b ) Y , , , , , , , , , IRAS 12127 − < a (HII b ) Y , , , , IRAS 12540+5708 (Mrk 231) 0.042 1.87 8.66 31.99 30.29 12.5 Sy1 a,b Y , , , , , , , IRAS 13335 − < < a (cp b ) Y IRAS 13428+5608 (Mrk 273) 0.038 0.24 2.28 21.74 21.38 12.1 Sy2 a,b Y , , , , , , , , , , , IRAS 13443+0802 0.135 < < a (cp+Sy2 b ) NIRAS 13451+1232 (PKS 1345+12) 0.122 0.14 0.67 1.92 2.06 12.3 Sy2 a,b Y , , , , , , , IRAS 14348 − a (cp b ) Y , , , , , IRAS 15327+2340 (Arp 220) 0.018 0.48 7.92 103.33 112.40 12.1 LI a,b Y , , , , IRAS 16468+5200 0.150 < a (cp b ) Y , , IRAS 16474+3430 0.111 < a (cp b ) Y IRAS 16487+5447 0.104 < a (cp b ) Y , IRAS 17044+6720 0.135 < a (Sy2 b ) Y , , , IRAS 21208 − < < a (cp b ) Y IRAS 23233+2817 0.114 < a,b Y IRAS 23234+0946 0.128 < a (cp b ) Y , IRAS 23327+2913 0.107 < a (Sy2 b ) Y , IRAS 23389+0300 0.145 < < a,b NIRAS 23498+2423 0.212 < a,b Y , , , , UGC 5101 0.040 0.25 1.03 11.54 20.23 12.0 LI c (Sy2 b ) Y , , , , , Mrk 463 0.051 0.51 1.58 2.18 1.92 11.8 Sy2 d,e,f Y , , , , , Mrk 739 0.030 0.16 0.31 1.26 2.41 10.9 Sy1+HII g Y NGC 3393 0.013 0.13 0.75 2.25 3.87 10.4 Sy2 h Y NGC 6240 0.024 0.56 3.42 22.68 27.78 11.8 LI b,c Y , , , IRAS 20210+1121 0.056 0.29 1.40 3.39 2.68 11.9 Sy2+LI i Y Note . — Col.(1): Object name. Col.(2): Redshift. Cols.(3)–(6): f , f , f , and f are IRAS fluxes at 12 µ m, 25 µ m, 60 µ m, and 100 µ m. Forthe first 23 galaxies, the flux is derived from Kim & Sanders (1998). For the last six galaxies, we use the IRAS Faint Source Catalog. Col.(7): Decimallogarithm of infrared (8 − µ m) luminosity in units of solar luminosity (L ⊙ ), calculated with L IR = 2 . × × D(Mpc) × (13.48 × f + 5.16 × f + 2 . × f + f ) [ergs s − ] (Sanders & Mirabel 1996), where we adopt H = 71 km s − Mpc − , Ω M = 0.27, and Ω Λ = 0.73 (Komatsu et al.2009), to estimate the luminosity distance D (Mpc) from the redshift. Col.(8): Optical spectral classification and references. Sy1, Sy2, LI, HII, and cpmean Seyfert 1, Seyfert 2, LINER, HII-region, and starburst+AGN composite type, respectively. a : Veilleux et al. (1999a). b : Yuan et al. (2010). c :Veilleux et al. (1995). d : Shuder & Osterbrock (1981). e : Hutchings & Neff (1989). f : Sanders et al. (1988). g : Koss et al. (2011). h : Diaz et al. (1988). i : Perez et al. (1990). Col.(9): The presence of AGN signatures in the brightest main nucleus (Y = yes, N = no), and several selected representativereferences. : Imanishi et al. (2007). : Veilleux et al. (2009). : Nardini et al. (2010). : Imanishi et al. (2006a). : Imanishi et al. (2010b). : Veilleuxet al. (1999b). : Imanishi & Dudley (2000). : Risaliti et al. (2000). : Soifer et al. (2000). : Severgnini et al. (2001). : Imanishi & Terashima(2004). : Armus et al. (2007). : Farrah et al. (2007). : Imanishi et al. (2008). : Teng et al. (2009). : Dudley & Wynn-Williams (1997). :Imanishi et al. (2011b). : Maloney & Reynolds (2000). : Braito et al. (2004). : Xia et al. (2002). : Balestra et al. (2005). : Iwasawa et al.(2011a). : U et al. (2013). : Veilleux et al. (1997). : Downes & Eckart (2007). : Imanishi et al. (2001). : Imanishi et al. (2003). : Ueno etal. (1996). : Bianchi et al. (2008). : Koss et al. (2011). : Fabbiano et al. (2011). : Komossa et al. (2003). : Piconcelli et al. (2010). nfrared AO imaging of merging galaxies 13 TABLE 2Observation Log
Object Band Date Exposure Standard Star LGS-AO or NGS-AO guide star(UT) (min) Name mag Name R -band separationUSNO mag (arcsec)(1) (2) (3) (4) (5) (6) (7) (8) (9)IRAS 00091 − − − − − a a − a
16 34L’ 2013 May 8 22.5 HD129653 6.9 1458-0231011 a
16 34IRAS 13443+0802 K 2012 May 20 9 S791-C 11.2 0977-0294304 16 57L’ 2012 May 20 18 HD106965 7.3 0977-0294304 16 57PKS 1345+12 K 2012 March 23 4 P272D 11.2 western nucleus 12 0L’ 2012 March 23 9 HD136754 7.2 western nucleus 12 0IRAS 14348 − − a
12 37L’ 2011 June 20 5 HD129655 6.7 0924-0386013 a
12 37IRAS 20210+1121 K 2011 June 20 7.5 FS149 10.1 1015-0589702 a
13 47L’ 2011 June 20 6.3 HD201941 6.6 1015-0589702 a
13 47
Note . — Col.(1): Object name. Col.(2): Observed band. K - or L ′ -band. Col.(3): Observing date in UT. Col.(4): Net on-sourceexposure time in min. Col.(5): Standard star’s name. Col.(6): Standard star’s magnitude in the K - or L ′ -band. Col.(7): Guide star name(USNO number) used for the LGS-AO tip-tilt correction or NGS-AO correction. Col.(8): Guide star’s optical R -band magnitude. Col.(9):Separation between the target object and guide star in arcsec. a NGS-AO guide star
TABLE 3Nuclear Photometry and Estimated AGN Component
Object K (0 . ′′ L ′ (0 . ′′ K − L ′ (0 . ′′
5) AGN L
AGN
WISE(3.4 µ m)(mag) (mag) (mag) (%) (10 ergs s − ) (mag)(1) (2) (3) (4) (5) (6) (7)IRAS 00091 − ± A IRAS 00091 − > < < < A IRAS 00188 − ± − ± A IRAS 05024 − > < A IRAS 05189 − ± B ± > B < < < · · · IRAS 12127 − ± − > < < < · · · Mrk 231 9.2 7.2 C ± − ± A IRAS 13335 − ± A IRAS 13335 − > < < < A Mrk 273 SW 13.6 11.3 D ± A Mrk 273 NE 13.1 11.5 D ± A IRAS 13443+0802 NE 14.6 14.1 0.5 ± ± > < < < ± A PKS 1345+12 E 15.6 > < < < A IRAS 14348 − ± A IRAS 14348 − ± A Arp 220 W 13.0 11.7 1.3 ± A Arp 220 E 13.2 12.1 1.1 ± A IRAS 16468+5200 E 16.5 15.0 1.5 ± A IRAS 16468+5200 W 16.7 > < < < A IRAS 16474+3430 S 14.6 13.6 1.0 ± A IRAS 16474+3430 M E ± A IRAS 16487+5447 SW 15.7 14.6 1.1 ± A IRAS 16487+5447 NE 16.0 > < < < A IRAS 17044+6720 14.5 11.9 2.6 ± − ± − > < · · · IRAS 23233+2817 N 14.6 12.7 1.9 ± A IRAS 23233+2817 S 16.3 > < < < A IRAS 23234+0946 NW 15.2 13.7 1.5 ± A IRAS 23234+0946 SE 16.7 > < < < A IRAS 23327+2913 S 14.5 13.2 1.3 ± > < < < ± A IRAS 23389+0300 S 17.1 > < < < A IRAS 23498+2423 NW 13.7 11.4 2.3 ± A IRAS 23498+2423 SE 17.4 > < < < A UGC 5101 12.0 9.5 2.5 ± ± A Mrk 463 W 14.1 13.3 0.8 ± A Mrk 739 E 12.2 10.2 2.0 ± ± · · · NGC 3393 13.0 11.9 1.1 ± F G ± A NGC 6240 N F G ± A IRAS 20210+1121 S 13.5 11.0 2.5 ± ± · · · Note . — Col.(1): Object name. Col.(2): K -band (2.2 µ m) magnitude within the central ∼ . ′′ − data). For Mrk 231, the 0 . ′′ × − . The possible uncertainty for spatially unresolved compact emission is taken as < § L ′ -band (3.8 µ m) magnitude within the central ∼ . ′′ − data). The possible uncertainty for spatially unresolved compact emission is taken as < § L ′ -band nuclei, the upper limits are derived at the K -band peak position basedon 3 σ of sky fluctuation. Col.(4): K − L ′ color magnitude within the central ∼ . ′′ < § L ′ -band in %. Col.(6): AGN luminosity in 10 ergs s − from AGN-origin ν F ν valueat L ′ after removing the starburst contribution. The possible uncertainty is < µ m photometric magnitude with 6 . ′′ A More than one nucleus combined, due to low-spatial-resolution data. B Zhou et al. (1993) estimated 10.0 mag and > µ m using a 2 . ′′ L ′ -band (3.8 µ m) magnitude, measured with a smaller aperture,is ∼ µ m to 3.8 µ m (Imanishi et al. 2008). C Zhou et al. (1993) estimated 7.5 mag at 3.4 µ m using a 2 . ′′ L ′ with asmaller aperture is 0.3 mag brighter. This galaxy shows a red, rising continuum flux from 3.4 µ m to 3.8 µ m(Imanishi et al. 2010b) and possesses a Seyfert 1 nucleus, in which case, a flux time variation may be presenton a ∼ D Zhou et al. (1993) estimated 10.9 mag at 3.4 µ m using a 2 . ′′ E This nucleus is first spatially-resolved in our AO images, and is different from the northern nucleus definedby Kim et al. (2002). We call as M (= middle nucleus). A third faint emission component may be presentbetween the M and S nuclei in the L ′ -band image in Figure 1. Photometry of this component was notattempted because its signal largely overlaps with the signal of the brighter M nucleus in the chosen ∼ . ′′ F The nuclear separation in the L ′ -band is estimated to be ∼ . ′′ ± . ′′
05, where the uncertainties of bothpeak position coordinate determination and pixel scale (20.57 ± − ) are taken into account. Thisis comparable to the values shown by Max et al. (2007) (1 . ′′ ± . ′′
007 at L ′ ), estimated in the 2–10 keVX-ray (1 . ′′ ± . ′′
2) (Komossa et al. 2003; Max et al. 2007) and in the radio 1.4–5 GHz (6–21 cm) (1 . ′′ . ′′ G Zhou et al. (1993) estimated 10.1 mag and 10.8 mag at 3.4 µ m using a 1 . ′′ nfrared AO imaging of merging galaxies 15 TABLE 4Comparison of our K -band (2.2 µ m) Photometry with 2.2 µ mPhotometry by Scoville et al. (2000) Object 2.2 µ m mag K -mag DifferenceScoville et al. (2000) our data(1) (2) (3) (4)IRAS 05189 − − − Note . — Col.(1): Object name. Col.(2): Nuclear 1 . ′′ µ m derived by Scoville et al. (2000). Col.(3): Nuclear 0 . ′′ . ′′ K -band (2.2 µ m) magnitude based on our data.Col.(4): Difference in the photometry between Scoville et al. (2000) andour data. For all sources, our data tend to show slightly fainter photomet-ric magnitudes than computed by Scoville et al. (2000) due to the smalleraperture size and the signal spread into the seeing-sized halo outside theAO core. TABLE 5Luminosity Ratio and Nuclear Separation in Seeing-based Multiple Nuclei (U)LIRGs
Object K(stellar) ratio L ′ ratio L ′ (AGN) ratio Separation Separation(arcsec) (kpc)(1) (2) (3) (4) (5) (6)IRAS 00091 − < < > >
176 5.5 6.1IRAS 12127 − > >
63 10.4 24.2IRAS 13335 − A > > A > >
38 2.0 4.4IRAS 14348 − A > > > > − > ∞ A > > > >
10 12.5 24.1IRAS 23389+0300 N, S 6.49 (15.08, 17.11) > > > >
38 4.1 13.9Mrk 463 E, W 1.74 (10.49, 11.09) 100 324 3.9 3.8Mrk 739 E, W 3.53 (11.37, 12.74) 28 ∞ Note . — Col.(1): Object name. Col.(2): K -band flux ratio, measured with the same aperture size between twonuclei, as an approximation of stellar emission luminosity ratio. K -band photometric values in individual nucleiare shown in parentheses; the first value is for the nucleus shown first in column 1. The 4-kpc aperture K -bandphotometry by Kim et al. (2002) is basically adopted, but for the last three sources (Mrk 463, Mrk 739, and IRAS20210+1121), photometry from the 2MASS point source catalog (4 ′′ aperture) is employed. Col.(3): Nuclear L ′ -band (3.8 µ m) luminosity ratio based on our photometry (Table 3, column 3). Col.(4): Nuclear AGN-origin L ′ -band (3.8 µ m) luminosity ratio after the subtraction of stellar emission component (Table 3, column 6). InCols. (2)–(4), the luminosity at the nucleus listed first in column 1 is divided by that listed second in column1. Col.(5): Apparent nuclear separation in arcsec calculated from our Subaru LGS/NGS-AO K -band images.Col.(6): Apparent nuclear separation in kpc calculated using H = 71 km s − Mpc − , Ω M = 0.27, and Ω Λ = 0.73(Komatsu et al. 2009). A Based on the velocity dispersion obtained through near-infrared spectroscopy, SMBH mass ratios are computedto be 2.45, 1.72, 0.61, and 3.29 for IRAS 13335 − − Fig. 1.—
Our Subaru AO K - (2.2 µ m) and L ′ -band (3.8 µ m) images of observed (U)LIRG nuclei. The field of view (FOV) is 10 ′′ x10 ′′ . North is up, and east is to the left. For IRAS 13443+0802, 23327+2913, and 20210+1121, two separate images are shown becausethe largest separations of multiple nuclei are > ′′ (Kim et al. 2002; Piconcelli et al. 2010). Fig. 2.—
Comparison of our Subaru AO L ′ -band (3.8 µ m) images with Spitzer IRAC 3.6 µ m images of two ULIRGs (Arp 220 and IRAS16474+3430). The FOV is 10 ′′ × ′′ . North is up, and east is to the left. Fig. 3.—
The growth of the curve of the encircled signal in the L ′ -band for standard stars observed with NGS-AO on various nights(Table 2) using the standard stars themselves as AO guide stars. The growth of the signal of a ULIRG dominated by spatially unresolvedcompact emission (Mrk 231), observed with NGS-AO, is also plotted. For the upper left plot, the growth of the curves of a standard star(HD 1160) observed with NGS-AO (two data sets) and without AO (seven data sets) on the same night (2011 August 22 UT) are shownand compared. NGS-AO data (solid lines) show a higher central concentration of signals than non-AO data (dashed lines), demonstratingthe power of Subaru AO for improved photometry of spatially unresolved compact sources using a small aperture. The horizontal andvertical dotted lines indicate an 80% signal fraction and employed ∼ . ′′ ∼ . ′′ × − ) consistently contains 83–93% of spatially unresolved compact source signals in the Subaru L ′ -bandNGS-AO images. nfrared AO imaging of merging galaxies 21 Fig. 4.—
The growth of the curve of the encircled signal in the L ′ -band observed with LGS-AO. (Left) : Compact merging nuclei with nodiscernible extended emission component (Mrk 739 eastern nucleus and IRAS 08572+3915 north-western nucleus), observed in 2012 April.In the L ′ -band images of merging galaxies, it is harder to find bright stars with high detection significance inside frames than in the K -bandbecause of the smaller field of view and larger background noise. For these reasons, compact merging nuclei are used to estimate the signalgrowth of the curve for LGS-AO at L ′ . As these merging nuclei could contain spatially extended host galaxy emission components, thesignal fraction at each aperture size is a lower limit on the spatially unresolved emission components. (Middle) : Compact merging nucleiobserved in 2012 May (IRAS 12127 − (Right) : A bright star inside the field of view of IRAS 05024 − L ′ -band image, where a significant noise patternis recognizable. The growth of the curve may be affected by this noise. The horizontal and vertical dotted lines indicate the 80% signalfraction and the ∼ . ′′ ∼ . ′′ × − ) in the Subaru L ′ -band LGS-AO images. This signal fraction iscomparable to that of Subaru L ′ -band NGS-AO images (Figure 3). Fig. 5.—
The growth of the curve of the encircled signal in the K -band for standard stars observed with NGS-AO during variousobservation nights (Table 2) using the standard stars themselves as AO guide stars. A plot of a ULIRG dominated by spatially unresolvedcompact emission (Mrk 231), observed with NGS-AO, is also shown. The horizontal and vertical dotted lines indicate the 80% signalfraction and employed ∼ . ′′ ∼ . ′′ × − ) for Subaru K -band NGS-AO data. For Mrk 231, the ∼ . ′′ × − (see § nfrared AO imaging of merging galaxies 23 Fig. 6.—