Radio continuum properties of luminous infrared galaxies. Identifying the presence of an AGN in the radio
E. Vardoulaki, V. Charmandaris, E. J. Murphy, T. Diaz-Santos, L. Armus, A. Evans, J. M. Mazzarella, G. C. Privon, S. Stierwalt, L. Barcos-Munoz
AAstronomy & Astrophysics manuscript no. VardoulakiE_etal2014_astroph c (cid:13)
ESO 2018October 15, 2018
Radio continuum properties of luminous infrared galaxies
Identifying the presence of an AGN in the radio
E. Vardoulaki (cid:63) , V. Charmandaris , E. J. Murphy , T. Diaz-Santos , , L. Armus , A. S. Evans , J. Mazzarella , G.C. Privon , S. Stierwalt , L. Barcos-Muñoz Department of Physics, University of Crete, GR-71003, Heraklion, Greece Institute for Astronomy, Astrophysics, Space Applications & Remote Sensing, National Observatory of Athens, GR-15236, Athens,Greece Chercheur Associé, Observatoire de Paris, F-75014, Paris, France Infrared Processing & Analysis Center, MS 100-22, California Institute of Technology, Pasadena, CA, USA Nucleo de Astronomia de la Facultad de Ingenieria, Universidad Diego Portales, Av. Ejercito Libertador 441, Santiago, Chile Department of Astronomy, University of Virginia, Charlottesville, VA, USAReceived ; accepted
ABSTRACT
Context.
Luminous infrared galaxies (LIRGs) are systems enshrouded in dust, which absorbs most of their optical / UV emission andradiates it again in the mid- and far-infrared. Radio observations are largely una ff ected by dust obscuration, enabling us to study thecentral regions of LIRGs in an unbiased manner. Aims.
The main goal of this project is to examine how the radio properties of local LIRGs relate to their infrared spectral charac-teristics. Here we present an analysis of the radio continuum properties of a subset of the Great Observatories All-sky LIRG Survey(GOALS), which consists of 202 nearby systems ( z < Methods.
We used the archival data (cid:63)(cid:63) at 1.49 and 8.44 GHz to create radio-spectral-index maps using the standard relation betweenflux density S ν and frequency ν , S ν ∼ ν − α , where α is the radio spectral index. By studying the spatial variations in α , we classifiedthe objects as radio-AGN, radio-SB, and AGN / SB (a mixture). We identified the presence of an active nucleus using the radiomorphology, deviations from the radio / infrared correlation, and spatially resolved spectral index maps, and then correlated this tothe usual mid-infrared ([NeV] / [NeII] and [OIV] / [NeII] line ratios and equivalent width of the 6.2 µ m PAH feature) and optical (BPTdiagram) AGN diagnostics. Results.
We find that 21 out of the 46 objects in our sample ( ∼ ∼ ∼ / SB. After comparing to other AGN diagnostics we find 3 objects out ofthe 46 ( ∼ Key words.
Galaxies: active – Galaxies: starburst – Galaxies: nuclei – Radio continuum: galaxies – Infrared: galaxies
1. Introduction
The all-sky survey of the Infrared Astronomical Satellite (IRASNeugebauer et al. 1984) revealed the existence in the local uni-verse of the so-called Luminous and Ultraluminous InfraredGalaxies (U / LIRGs). This is a class of galaxies with an infraredluminosity (L IR ; 8 µ m < λ < µ m) between 10 L (cid:12) and10 L (cid:12) for the LIRGs and over 10 L (cid:12) for the ULIRGs. De-tailed analysis of the properties of these systems has revealedthat they contain large quantities of dust, which absorbs over90% of their optical / UV emission (for the most luminous sys-tems) and re-radiates it in the infrared (Armus et al. 2009). Re-cent advances in space infrared surveys by ISO and Spitzer havealso demonstrated that LIRGs compose a significant fraction ( ≥ (cid:63) email: [email protected] (cid:63)(cid:63) The VLA images used in this study are available in electronic format the CDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) orvia http: // cdsweb.u-strasbg.fr / cgi-bin / qcat?J / A + A / formation activity at z ∼ z ∼ ∼
18% of LIRGs, compared to ∼
90% of ULIRGs, are associated with late type mergers based onanalysis of the Spitzer / IRAC images. These numbers change to ∼
19% in LIRGs and ∼
81% in ULIRGs, based on HST classi-fication of merger stages (Haan et al. 2011, not all objects have
Article number, page 1 of 32 a r X i v : . [ a s t r o - ph . GA ] A ug n associated HST merger stage). In total, ∼
68% of LIRGs and ∼
95% of ULIRGs are merging systems based on IRAC mergerstages (Stierwalt et al. 2013), while from HST merger stages weget ∼
91% of LIRGs and ∼
93% of ULIRGs associated with amerger (Haan et al. 2011).The di ff erence in physical size of the area responsible for theenergy production results in observable di ff erences in their spec-tral energy distributions, especially in the mid-IR. Spectroscopicstudies, in particular in the infrared range, which is less a ff ectedby extinction, reveal a great diversity in the 5-30 µ m mid-IRspectral features (Diaz-Santos et al. 2010b, 2011). The emissionfrom polycyclic aromatic hydrocarbons (PAH), tracers of pho-todissociation regions and the gas heating via star formation, isclearly seen in LIRGs and has been used extensively to estimatetheir star formation rate in the local universe (i.e. Calzetti 2010).However, as the infrared luminosity of the systems increases andreaches the ULIRG range, PAH emission becomes weaker (De-sai et al. 2007). We also see that in LIRGs (Stierwalt et al. 2014).In addition, in LIRGs the average size of the area producing themid-IR continuum emission is larger than in ULIRGs, and it ap-pears that there is a correlation between the compactness of themid-IR continuum emission and the IRAS far-IR colours of thesource (e.g. Diaz-Santos et al. 2010b). The typical size of themid-IR emitting region in most LIRGs is of the order of 500pc (FWHM), while the estimated upper limit for the core sizein ULIRGs in the mid-IR is < ∼
20% of LIRGs host an AGNdetected in the mid-IR, and its presence becomes more prevalentas the IR luminosity increases (e.g. Armus et al. 2007; Petricet al. 2011; Inami et al. 2013; Stierwalt et al. 2013). The ac-tual contribution of this AGN to the bolometric luminosity of agalaxy is a challenging problem. Detailed knowledge of eachsource and observations across the large portion of its spectrumare often required (Veilleux et al. 2009). In some very dust-obscured sources, the very central regions are hidden from di-rect view at wavelengths shorter than ∼ µ m (e.g. Laurent etal. 2000; Spoon et al. 2004). At radio wavelengths one can over-come the e ff ects of dust obscuration and study the centres ofthese objects in great detail. Thus we can identify signs of AGNactivity, such as weak jets (e.g. Lonsdale et al. 2003), or observemore extended structures not associated with AGN but typical ofstar-forming regions (Condon et al. 1991).In this paper we use a radio-analysis technique to the radiocontinuum observations of Condon et al. (1990, 1991) in orderto identify the presence of AGN. As discussed earlier, LIRG nu-clei are enshrouded in dust, making them optically thick in theUV / optical, and often even in the mid-IR. Even though X-rayphotons originating in an AGN penetrate the dust e ffi ciently, nu-clei with hydrogen column densities N(H) > cm − becomeoptically thick to X-ray emission. On the other hand, radio ob-servations are largely una ff ected by dust obscuration, enablingus to study the central regions of LIRGs in an unbiased man-ner. The high spatial resolution ( ∼ arcsec) of VLA radio obser-vations already in hand allows us to investigate the presence ofan active nucleus in U / LIRGs via the identification of weak jets (e.g. Lonsdale et al. 2003) or spatial variation of the radio spectraindex.Even though radio observations of LIRGs have been in theliterature for more than 20 years (Condon et al. 1990, 1991),with our analysis we aim to obtain additional information aboutthe nature of these objects. Condon et al. (1991) presented a clas-sification of these objects based on the integrated radio spectralindex, calculated between 1.4 and 8.44 GHz. But using the in-tegrated radio spectral index of an object to distinguish betweenan AGN and a nuclear SB can be challenging, since both exhibitflat and steep radio spectral indices. Here we present a methodthat takes advantage of the data available at 1.49 and 8.44 GHz.By studying the variations in the radio spectral index spatially,we aim to distinguish between the contribution from the AGNand the SB in the radio (Sec. 3).Furthermore, we compare our radio findings to known mid-IR and optical AGN diagnostics. Current AGN diagnostics in-clude a combination of emission-line diagnostics such as [OIII] / H β and [Ne III] / [O II] ratios (BPT diagram Baldwin et al.1981; Kewley et al. 2006), mid-IR colours and lines (e.g. Gen-zel et al. 1998; Laurent et al. 2000; Lacy et al. 2004) and hardX-ray emission. Diagnostics based on Spitzer mid-IR observa-tions of the GOALS sample reveal that ∼
20% host an AGN, butonly in 10% the AGN dominates the mid-IR emission from thegalaxy (Petric et al. 2011). These estimates rely either on thepresence of dust emission heated to near sublimation tempera-tures ( ∼ ff ects these photons have onthe interstellar gas. More specifically, PAH emission from AGNis known to be suppressed (see Wu et al. 2009), while emis-sion lines such as [Ne V]14.3 / µ m and [O IV]25.9 µ m, withionisation potential of 97 and 55 eV respectively, are observedunder the same circumstances (Tommasin et al. 2010). We wishto examine how these diagnostic-tools vary in LIRGs for whichevidence of an AGN is established via their radio emission.We use J2000.0 positions and the convention for all spectralindices, α , that flux density of a radio source S ν ∼ ν − α , where ν is the observing frequency. In Sec. 2 we present the sample usedin the analysis (Sec. 3). In Sec. 4 we present our results of theradio analysis and our discussion on the comparison to mid-IRand optical AGN diagnostics. We also give an upper limit for thesize of the compact starburst in the radio. In Sec. 5 we presentour conclusions. In the Appendix we give an object-by-objectanalysis and the radio-spectral-index maps and histograms forall the objects of our sample.
2. Sample
This study is part of the Great Observatories All-sky LIRG Sur-vey (GOALS; Armus et al. 2009), a local ( z < µ m, covering the entire sky surveyed by IRAS at Galac-tic latitudes | b | > ◦ . The GOALS sample is ideal not onlyin its completeness and sample size, but also in the proximityand brightness of the galaxies. The galaxies span the full rangeof nuclear spectral types (type-1 and type-2 AGN, LINERs andstarbursts) and interaction stages (major mergers, minor mergersand isolated galaxies).The multi-wavelength coverage of the GOALS sample in-cludes Spitzer / IRAC imaging at 3.6, 4.5, 5.8 and 8.0 µ m as wellas Spitzer / MIPS at 24, 70 and 160 µ m. Spitzer / IRS 5-37 µ m Article number, page 2 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs spectra are available for the nuclei of all 293 galaxies. High res-olution optical and near-IR Hubble Space Telescope imaging isavailable for 88 galaxies with L IR > . L (cid:12) in addition to far-UV data for 25 objects, making GOALS one of the most com-plete local sample imaged by Hubble. Complete tables showingall available data can be found at http: // goals.ipac.caltech.edu / .Here we present a new analysis of archival radio continuummaps available for the GOALS galaxies. For 151 LIRG systems,radio maps were obtained with the Very Large Array (VLA) at1.4 GHz (Condon et al. 1990). In particular, 57 objects were ob-served in A configuration (1 (cid:48)(cid:48) .
5, 1 (cid:48)(cid:48) .
8, 2 (cid:48)(cid:48) . (cid:48)(cid:48) . (cid:48)(cid:48) , 6 (cid:48)(cid:48) , 7 (cid:48)(cid:48) and 8 (cid:48)(cid:48) resolution), 103 in C con-figuration (15 (cid:48)(cid:48) , 18 (cid:48)(cid:48) , 21 (cid:48)(cid:48) and 24 (cid:48)(cid:48) resolution) and 6 in D con-figuration (48 (cid:48)(cid:48) , 54 (cid:48)(cid:48) and 60 (cid:48)(cid:48) resolution). Objects that containdi ff use features, such as extended disks, require low-resolutionconfigurations in order to recover that faint di ff use emission,while more luminous (L IR > L (cid:12) ) infrared sources are as-sociated with smaller ( ∼ L IR / L (cid:12) ) ≥
3. Analysis of radio data
Using the 1.49 and 8.44 GHz images we construct radio spectralindex maps ( α -maps) to probe the origin of the emission aris-ing from the central kpc of our LIRG sample. The availabledata at 1.49 and 8.44 GHz have di ff erent resolutions. In order tobe compared, they need to be convolved to the same resolution.Thus we change the resolution of high-resolution 8.44-GHz mapto the one of the lower-resolution 1.49-GHz map, and convolvethe 8.44 GHz map to the Point Spread Function (PSF) of the 1.49GHz map, after deconvolving with the PSF of the 8.44 GHz map.We finally make sure that the flux in the maps is conserved afterthe image manipulation.To ensure that there are no misalignments in the final imageafter convolution, we checked that the maximum-flux pixel in theoriginal and final image of each object have the same R.A. andDec. We also checked whether the original 1.49- and 8.44-GHzimages have the maximum-flux pixel at the same coordinates. Inmost cases the di ff erence is negligible, and can be ignored. Ina couple of cases though (NGC 0034 S, IC 1623A / B), the dif-ference is larger than the pixel size of the 1.49 GHz map ( > ff ect,and could be attributed to the existence of an AGN. Studies (e.g.Lobanov 1998) have shown that the peak of core emission getscloser to the black-hole (BH) as we observe at higher frequen- ( T b / K) (Condon+91)23456 l og ( T b / K ) ( m ea s u r e d ) Fig. 1.
Brightness temperature T b (in K) as measured from the 8.44GHz maps, versus brightness temperature from Condon et al. (1991).Not all of our objects have a measurement in Condon et al. (1991), thusare omitted from the plot. Symbols: blue diamonds denote the commonobjects, filled blue diamonds denote objects classified as radio-AGN(see Table 4), and the black squares denote objects that are classifiedas mid-IR AGN based on Petric et al. (2011). The dotted line is theone-to-one relation. The relative mean dispersion is 2.5%. cies. This is obvious in NGC 0034 (see α -map, Fig. 2), wherethe 8.44 GHz maximum-flux position is matching the bulge po-sition, whereas the 1.49 GHz maximum-flux position is furtherout (see notes on the objects in Appendix A for further discus-sion).After assuring the maps are aligned and have the same res-olution, we divide them based on the relation between flux den-sity and frequency given above, to obtain the radio spectral indexmaps. Only flux densities above 3 σ in both images are used inthe calculation of α . The resulting α -maps are shown in Fig. 2.The errors in α were calculated using standard error propagation.Since the synthesised beam-size of the 8.44 GHz observations ismuch less than the one at 1.49 GHz, the area observed at 8.44GHz is much smaller than at 1.49 GHz. As a result, even afterconvolving the 8.44 GHz to the resolution of the 1.49 GHz maps,the resulting α -maps can be smaller than the size of the sourceat 1.49 GHz due to lack of information from the smoothed 8.44GHz images (e.g. IC 1623A / B).Radio and infrared properties of our LIRGs are presentedin Table 1. Fluxes at 1.49 and 8.44 GHz were measured fromthe original maps by applying a 3 σ flux cut. We wish to men-tion that at 8.44 GHz there is an obvious di ff erence between ourmeasurements and the ones presented in Condon et al. (1991).The possible explanation for that is the 3 σ flux cut we apply,that could be excluding some di ff use emission from the source,while in some cases Condon et al. (1991) only report the peakflux density at 8.44 GHz.As part of the radio analysis of our objects, we also classifythem based on their radio structure at both frequencies available(1.49 & 8.44 GHz) following the scheme: compact (COM) forobjects whose contours are circular around the maximum-fluxposition; extended (EXT) for objects whose contours are not cir- Article number, page 3 of 32 ular around the maximum-flux position and extend outwards;di ff use (DF) for objects whose radio structure appears di ff use;double (DB) for objects that show a double radio structure; andcomplex (CMX) for objects that show multiple radio compo-nents. These classifications are presented in Table 1, and aidin our understanding of whether the object contains an AGN ornot.By studying the variation in the value of α in the radio-spectral-index maps we aim to identify which objects have anAGN component (see Sec. 3.1). Furthermore, by comparing ourradio-based result with those from known mid-IR AGN diag-nostic tools (see Sec. 4.2), we are looking for evidence of buriedAGN identified in the radio that are not detected in the mid-IR.We also construct histograms ( α -hists), shown in Fig. 2- Right . The α -hists are based on the α -maps and give the valuesof α at each pixel ( top panel). The middle panel is the α + errorhistogram, where the α -map and error map (Fig. 2- Middle ) wereadded. The bottom panel shows the α -error histogram, producedafter subtracting the error map (Fig. 2- Middle ) from the α -map.These histograms are necessary in the interpretation of the anal-ysis and in order to classify the objects in our sample. The distri-bution shown reflects the number of values of the radio spectralindex, from flat to steep, within the area occupied by the sourcein the α -maps. Whenever we refer to distance from the centre wemean the distance from the position in the map that correspondsto the maximum flux density at 1.49 GHz, as given in Table 1.Condon et al. (1991) used the brightness temperature T b todistinguish between SB and AGN, where an AGN would have T b ∼ K, but they do not provide brightness temperatures forall of the objects in our sample. Here we provide measurementsof the brightness temperature T b for all individual objects inour sample, using the relation given in Condon et al. (1991),log[( S / Ω )(mJy / arcsec )] ≈ log[ T b (K)] - 1.3, where S is the fluxdensity and Ω = π × a × b / (8 × ln(2)), where a is the majoraxis and b the minor axis both in arcsec. We use the functionGAUSS2DFIT in IDL to fit an ellipse to each radio componentat 8.44 GHz and calculate the deconvolved major and minor axesand the corresponding position angle PA. All values are given inTable 2. The di ff erences from the values of T b reported in Con-don et al. (1991) are due to the di ff erence in the flux densities wemeasured at 8.44 GHz from the ones in literature (see Fig. 1):the relative mean dispersion between measured values and thosefrom the literature is 2.5%. One of the goals of this study is to unveil buried AGN in the cen-tre of local LIRGs by identifying variations in the radio-spectral-index maps (see Fig. 2) created using the radio continuum mapsat 1.49 and at 8.44 GHz. AGN or SB classification based on in-tegrated values of the radio spectral index is inconclusive, sinceboth can display similar values.In the case of radio-loud radio sources that display radio jets,separating SB and AGN is more straight-forward, since thesestructures can only be created by powerful AGN. In the case ofradio-quiet radio sources though, separating the two contribu-tions is challenging. The LIRGs in our sample have been knownto contain a nuclear SB that is identified in radio via synchrotronemission and low brightness temperatures (Condon et al. 1991);only powerful AGN have T b (cid:38) -10 K (see also Clemens etal. 2008). At the same time, some of them are known to havean AGN contribution (see Vega et al. 2008; Petric et al. 2011;Alonso-Herrero et al. 2013, for the AGN percentage). Still, AGNand SB can both give a steep radio spectral index ( ∼ α ∼ α -maps,created from data below this wavelength range, will help us todiscern between these two cases.Due to the unified AGN scheme (e.g. Antonucci 1993), wehave to also take into account orientation e ff ects. In radio-loudAGN, depending on the viewing angle we can get a compactradio source or a more extended radio structure (FRI or FRII orin-between, e.g. Fanaro ff & Riley 1974). In the former case, thiscan occur when the jet is pointing towards the observer and thereis constant flow of material causing a relativistic e ff ect known asDoppler boosting. The resulting integrated radio spectral indexfrom this source would be inverted, .i.e we get negative valuesof α . In the case of FRI or FRII radio sources, where one candistinguish the jet and lobes, especially in the case of FRII, thepicture changes. Material gets transferred through the jet to thelobes further out form the core. The hot-spots created displayflat radio spectral indices ( α < α to get steeper than ∼ ± σ uncertainty based onthis range ( α = α -histograms (see Fig. 2- Right ) we find thatthe objects in study fall in the following categories, after takinginto account the uncertainties:1. flat α values with steep tail (29 / α values with small dispersion (7 / α values (5 / α values in the centre (4 / α values, from flat to steep (3 / / α -maps, the radio struc-ture at both 1.49 & 8.44 GHz, the α -histograms, and the uncer-tainties in α : Article number, page 4 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs T a b l e . R a d i o a nd I n fr a r e dd a t a f o r t h e L I R G s y s t e m s o f ou r s a m p l e . N a m e R . A . D ec . S . GH z S . GH z α . GH z . GH z l og L . GH z l og L I R q I R I R R a d i o S t r u c t u r e A lt e r n a ti v e ( J )( J )( m J y )( m J y )( W H z − )( L (cid:12) ) . GH z . GH z N a m e ( )( )( )( )( )( )( )( )( )( )( )( )( ) ( ) NG C ( S ) h m s . − ◦ m s . ± . . ± . . ± . . . . . C O M ? E X T ? NG C ( )I C A / B h m s . − ◦ m s . ± . . ± . . ± . . . . . C M X C M XA r p236 ( ) C G C G - h m s . ◦ m s . ± . . ± . . ± . . . . . C O M ? C M X M C G + - - ( )I R A SF - h m s . − ◦ m s . ± . . ± . . ± . . . . . C O M ? C O M ? M A S X J - ( )I R A SF + h m s . ◦ m s . ± . . ± . . ± . . . . . C O M C O M III Z w ( ) UG C ( S ) h m s . ◦ m s . ± . . ± . . ± . . . . . C M X C M X ( )I R A SF + h m s . ◦ m s . ± . . ± . . ± . . . . . E X T C M X ( ) E S O -I G ( N ) h m s . − ◦ m s . ± . . ± . . ± . . . . — C O M D B M C G - - - ( ) E S O -I G ( S ) h m s . − ◦ m s . ± . . ± . . ± . . . . — C O M D F M C G - - - ( )I R A SF - h m s . − ◦ m s . ± . . ± . . ± . . . . . C O M E X T ? M A S X J - ( ) NG C h m s . ◦ m s . ± . . ± . . ± . . . . . C O M C O M ? A r p243 ( )I R A SF + h m s . ◦ m s . ± . . ± . . ± . . . . . C O M C O M ( ) UG C ( N E ) h m s . ◦ m s . ± . . ± . . ± . . . . . C O M ? C O M ? M C G + - - ( ) UG C ( S W ) h m s . ◦ m s . ± . . ± . . ± . . . . . D F D F A r p055 ( ) UG C h m s . ◦ m s . ± . . ± . . ± . . . . . C O M ? E X T M C G + - - ( )I R A SF + h m s . ◦ m s . ± . . ± . . ± . . . . . C O M C O M M A S X J + ( )I R A SF + h m s . ◦ m s . ± . . ± . . ± . . . . . C O M E X T M A S X J + ( ) M C G + - - h m s . ◦ m s . ± . . ± . . ± . . . . . E X TE X T A r p148 ( ) UG C ( N W ) h m s . ◦ m s . ± . . ± . . ± . . . . . C O M ? C O M ? I R A SF + ( ) UG C ( S E ) h m s . ◦ m s . ± . . ± . . ± . . . . C O M ? C O M ? I C ( ) NG C ( E ) h m s . ◦ m s . ± . . ± . . ± . . . . . C M X C M X I C ( ) NG C ( W ) h m s . ◦ m s . ± . . ± . . ± . . . . . C M X C M X I C N o t e s . C o l u m n s : ( ) N a m e o f t h e g a l a xy ; s y s t e m s a r e m a r k e d w it h t h e r e l a ti v e po s iti on t o eac ho t h e r( e . g . N , S , E , W ) ; t h e nu m b er s b e f o re t h e o b j ec t n a m e s a re u s e d i n F i g . . ( ) & ( ) R i gh t A s ce n s i on ( J ) & D ec li n a ti on ( J ) , r e s p ec ti v e l y , a tt h e r a d i opo s iti on w it h m a x i m u m fl uxd e n s it y a t . GH z . ( ) & ( )I n t e g r a t e d fl uxd e n s it y a t . & . GH z , r e s p ec ti v e l y , o f t h e g a l a xy w h e r e fl ux a bov e σ i s b e i ng c on s i d e r e d i n t h eca l c u l a ti on . W ea l s og i v e t h ee rr o r on t h e m ea s u r e m e n t . ( ) R a d i o s p ec t r a li nd e x ca l c u l a t e db e t w ee n1 . a nd8 . GH z . ( ) T h e r a d i o l u m i no s it y a t . GH zca l c u l a t e du s i ng t h e s t a nd a r d r e l a ti onb e t w ee n fl ux a nd l u m i no s it y ;t h e v a l u e s o f l u m i no s it yd i s t a n cea r e t a k e n fr o m D i az - S a n t o s e t a l . ( ) . ( ) T h e t o t a li n fr a r e d l u m i no s it y ca l c u l a t e d fr o m t o1000 µ m f o r eac hg a l a xy ( s ee D i az - S a n t o s e t a l . ) . ( ) T h e p a r a m e t e r q I R = l og [( S I R ( − µ m ) / . × W m − ) / ( S . GH z / W m − H z − )]( H e l ou e t a l . ) . ( ) T h e r a ti oo f t o t a l I R l u m i no s it y t o t h e l u m i no s it y a t µ m (I R ; e . g . E l b aze t a l . ) . ( ) & ( ) T h e r a d i o s t r u c t u r ea t . & . GH z , r e s p ec ti v e l y : C O M f o r c o m p ac t; E X T f o r e x t e nd e d ; C M X f o r c o m p l e x , D B f o r doub l e ; D B D B f o r doub l e doub l e ; a nd D F f o r d i ff u s e ; a ’ ? ’ i nd i ca t e s un ce r t a i n t y i n t h ec l a ss i fi ca ti on . ( ) G i v e s a n a lt e r n a ti v e n a m e f o r t h e g a l a xy . Article number, page 5 of 32 a b l e . ( C on ti nu e d ) N a m e R . A . D ec . S . GH z S . GH z α . GH z . GH z l og L . GH z l og L I R q I R I R R a d i o S t r u c t u r e A lt e r n a ti v e ( J )( J )( m J y )( m J y )( W H z − )( L (cid:12) ) . GH z . GH z N a m e ( )( )( )( )( )( )( )( )( )( )( )( )( ) ( )I R A SF + ( N E ) h m s . ◦ m s . ± . . ± . . ± . . . . . t C O M C O M ( )I R A SF + ( S W ) h m s . ◦ m s . ± . . ± . . ± . . . . . t C O M C O M ( ) UG C h m s . ◦ m s . ± . . ± . - . ± . . . . . C O M C O M ? M r k231 ( ) VV a ( N W ) h m s . ◦ m s . ± . . ± . . ± . . . . . C O M C M XUG C ( W ) ( ) VV a ( S E ) h m s . ◦ m s . ± . . ± . . ± . . . . . C O M C M XUG C ( E ) ( ) UG C h m s . ◦ m s . ± . . ± . . ± . . . . . C O M ? C M XA r p193 ( ) NG C ( N E ) h m s . ◦ m s . ± . . ± . . ± . . . . . C O M C O MM r k266 ( N E ) ( ) NG C ( S W ) h m s . ◦ m s . ± . . ± . . ± . . . . — E X TE X T M r k266 ( S W ) ( ) NG C ( C ) h m s . ◦ m s . ± . . ± . . ± . . ——— E X TE X T M r k266 ( C ) ( ) NG C ( N W ) h m s . ◦ m s . ± . . ± . . ± . . . . . C M X C M XA r p240 ( ) UG C h m s . ◦ m s . ± . . ± . . ± . . . . . C O M ? C M X M r k273 ( )I R A SF - ( N E ) h m s . − ◦ m s . ± . . ± . . ± . . . . . t C O M C O M ( )I R A SF - ( S W ) h m s . − ◦ m s . ± . . ± . . ± . . . . . t C O M C O M ( ) VV a ( N ) h m s . ◦ m s . ± . . ± . . ± . . . . . D B D B D B UG C ( ) VV ( N ) h m s . ◦ m s . ± . . ± . . ± . . . . . C O M ? C O M ? I Z W ( ) VV ( S ) h m s . ◦ m s . ± . . ± . . ± . . . . C O M ? C O M ? I Z W ( )I R A SF + h m s . ◦ m s . ± . . ± . . ± . . . . . C O M C O M ( ) UG C h m s . ◦ m s . ± . . ± . . ± . . . . . C O M ? D B A r p220 ( ) NG C h m s . ◦ m s . ± . . ± . . ± . . . . . D B D B I Z W ( )I R A SF + ( N E ) h m s . ◦ m s . ± . . ± . . ± . . . . . t E X T D F ( )I R A SF + ( S W ) h m s . ◦ m s . ± . . ± . . ± . . . . . t C O M C O M ( )I R A SF - h m s . − ◦ m s . ± . . ± . . ± . . . . . C O M ? C O M ? ( )I C h m s . ◦ m s . ± . . ± . . ± . . . . . C O M C M X Z W . ( ) M r k0331 ( N ) h m s . ◦ m s . ± . . ± . . ± . . . . . C O M ? C M XUG C ( E ) N o t e s . T h e b a s e - l og a r it h m o f t h e t o t a li n fr a r e d l u m i no s it yo fI R A SF - i s . . H e r e w e g i v e t h e fr ac ti on a l c on t r i bu ti ono f t h e t w o c o m pon e n t s t o t h e t o t a li n fr a r e d l u m i no s it y . T h e fr ac ti on a l c on t r i bu ti on w a s ca l c u l a t e dbydo i ngpho t o m e t r yon t h e a nd24 µ mm a p s o f t h e ob j ec t . S i m il a r l y f o rI R A SF + l og ( L I R ) = . a nd f o rI R A SF + l og ( L I R ) = . . A ( t ) n e x tt o t h e v a l u e s o fI R s t a nd s f o r t o t a l m ea s u r e m e n t f o r t h e s y s t e m . Article number, page 6 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs radio-AGN when i ) α has negative values (inverted) due tosynchrotron self-absorption from a jet pointing towards theobserver; ii ) α is flat in the centre, displaying a steep tailsuggesting synchrotron ageing, and the values of α get largerthan 1.1; iii ) the α -map has steep α values that exceed 1.1; iv ) there is a shift in the core radio emission between 1.49 &8.44 GHz (see NGC 0034 N & IC 1623A / B in Appendix A).2. radio-SB when i ) α is flat across the distribution, and thebrightness temperature is less than 10 -10 K; ii ) α is flatin the centre, displaying a steep tail suggesting synchrotronageing due to CR electrons, but α does not exceed the valueof 1.1; iii ) α has a steep distribution ( ∼ radio-AGN / SB when we cannot distinguish between thetwo, and when due to the uncertainties in α it is not clearthat the steepening in the radio spectral index exceeds thetheoretical value of 1.1. This category can contain an AGN,or a SB or a mixture of both. Table 2.
Brightness Temperature measurements name log a b P.A.( T b / K) (”) (”) (deg.)(1) (2) (3) (4) (5)NGC 0034 (S) 3.15 0.39 0.32 12.2IC 1623A / B 3.75 0.21 0.13 -4.8CGCG 436-030 3.07 0.43 0.31 -12.4IRAS F01364-1042 3.69 0.14 0.14 -14.3IRAS F01417 + + + + + + + + + + + Notes.
The brightness temperature was measured directly from the 8.44GHz radio maps using the relation stated in Condon et al. (1991),log[(S /Ω )(mJy / arcsec )] ≈ log[T b (K)] - 1.3, where Ω = π × a × b / (8 × ln(2)) and S is the flux density, a the major axis and b the minoraxis in arcsec. The values are given in Column 2. We used the func-tion GAUSS2DFIT of IDL to fit ellipses to the radio components. Thisgives us the deconvolved major and minor axes, shown in Columns 3 &4 respectively, and the position angle PA of the ellipse from the x-axisin degrees, counter-clockwise, shown in Column 5.Article number, page 7 of 32 . Results and Disscussion The radio classification, which is based on the interpretation ofour radio analysis (see Sec. 3.1) is presented in Table 4. Fora better understanding of our results, notes on individual ob-jects are presented in the Appendix A, where we explain indetail the characteristics of each object and the way we classi-fied them. Furthermore, all the α -maps are available online athttp: // users.physics.uoc.gr / ∼ eleniv / radiogoals.html.Based on the above classification and the radio propertiespresented in Table 1, we find that from the 46 individual LIRGs,21 are radio-AGN, 9 are radio-SB and 16 are radio-AGN / SB.
Luminous infrared galaxies are known for their tight rela-tion between radio and infrared emission (e.g. Helou et al.1985; Bell 2003; Murphy 2009; Ivison et al. 2010). This isexpressed by the parameter q IR = log [( F IR(8 − µ m) / . × W m − ) / ( S . / W m − Hz − )], a measure of the star-formation-rate of a galaxy, where F IR(8 − µ m) is the total fluxin the infrared band, from 8 to 1000 µ m in W m − and S . isthe flux density at 1.4 GHz in W m − Hz − .In Fig. 3 we explore this relation for the 46 individual LIRGsin our sample. On the left we show the parameter q IR as afunction of the total infrared luminosity from 8-1000 µ m, andon the right with respect to the radio luminosity at 1.49 GHz.Higher resolution mid-IR observations (see Diaz-Santos et al.2010b, 2011; Stierwalt et al. 2013), enabled us to measure to-tal infrared luminosities for each galaxy of the system separately(Diaz-Santos et al. 2013), and calculate the value of q IR for indi-vidual galaxies, which could not be done by Condon et al. (1991)due to lack of available observations. We note that the mean q IR = ± ± q IR less than 2 σ from the mean haveexcess radio emission making them radio-louder than the rest.From Fig. 3 we see that only one galaxy (VV 705 S) lies justbelow the 2 σ uncertainty line. VV 705 S is classified as a radio-AGN based on our radio analysis, while based on the mid-IRand optical diagnostics presented in Sec. 4.2, it is a SB (see Ta-ble 4). This diagram thus, presents additional evidence for anobject containing an AGN. On the other hand, the rest of the ob-jects in our sample are located above the 2 σ uncertainty line, thebottom line in the figure, and as a result there is no strong evi-dence from this figure for the existence of an AGN due to excessradio emission above what expected from a SB. LIRGs have been explored in detail in the mid-IR and one ofthe AGN diagnostics used extensively is the ratio of high to lowionisation lines (e.g. [Ne V] / [Ne II] and [O IV] / [Ne II]), as itis a powerful tool to identify AGN (see e.g. Armus et al. 2007;Petric et al. 2011). The presence of the [NeV] line more, and toless extend that of the [O IV] line, in the mid-infrared spectrumof an object indicates an AGN (e.g. Petric et al. 2011), since theionisation potential of the [Ne V] line is too high to be producedby stars; for [O IV] the ionisation potential can be similar tothat coming from OB stars. Furthermore, the presence of anAGN in the centre of the LIRG may cause the PAH features to disappear, albeit on sub-kpc scales, due to strong radiation fields.As a result the PAH features are weak or absent in the centralregions of AGN (e.g. Genzel et al. 1998; Armus et al. 2007).Empirically (see Stierwalt et al. 2013; Murphy et al. 2013), anequivalent width (EQW) less than 0.27 µ m in the 6.2 µ m PAHfeature suggests that an AGN is contributing significantly to theMIR emission of the LIRG.Following Armus et al. (2007) and Petric et al. (2011) weplot the [Ne V] / [Ne II] and [O IV] / [Ne II] line ratios (Inami etal. 2013) to the 6.2 µ m PAH equivalent width (EQW; Stierwaltet al. 2013) in Fig. 4. Our sample of LIRGs is shown in bluediamonds, where objects classified as radio-AGN based on ouranalysis are marked with filled blue diamonds, and objects withAGN fraction (the contribution of the AGN to the total bolomet-ric luminosity) above zero (Petric et al. 2011) are marked withblack squares. For comparison, we overplot a sample of knownAGN shown as open red triangles, and a sample of known SBshown as open green circles; the data are taken from Armus etal. (2007). As it was discussed in detail in Section 3.3 of Ar-mus et al. (2007), average values from a set of "classical AGN"from Weedman et al. (2005) as well as well known nearly "purestarbust" systems from Brandl et al. (2006) and Bernard-Salaset al. (2009) were used to define the extreme high and low val-ues of the [NeV] / [NeII] flux ratio and the 6.2 µ m PAH EQW.Following that, it was assumed that the 100% contribution ofstarburst is associated with an EQW of ∼ µ m for the 6.2 µ mPAH feature. This was decreasing linearly to the lowest possiblemeasured value of 0.006 µ m which marked the 0% contributionand it was associated with the strongest AGN. Similar logic wasused for the [NeV] / [NeII] and [OIV] / [NeII] ratios presented inthe Y-axes of Fig. 4. Following Armus et al. (2007), the horizon-tal and vertical lines in Fig. 4 mark the fractional AGN and SBcontributions to the [NeV] / [NeII] or [OIV] / [NeII] and 6.2 µ mPAH EQW, respectively: the 100% level is taken as the averagevalue of the [Ne V] / [Ne II] or the [OI V] / [Ne II] ratio and 6.2 µ m PAH EQW among the AGNs and SBs used for comparison,respectively. The 50%, 25% and 10% levels are also shown.None of our objects has a strong AGN contribution based onthe [Ne V] detection, but there are some objects that have morethan 10% AGN contribution in their infrared luminosity. Wefurther see that some objects (22 out of the 46), although thereare not identified as AGN based on this diagnostic, are classifiedas radio-AGN based on the α -maps (e.g. IC 1623A / B).From the [Ne V] data, 8 LIRGs have a [NeV] detection andthus have an AGN contribution, 27 LIRGs have upper limits, and7 don’t have an entry in Table 4, since there is no high resolutionspectrum available. It is known that the [OIV] 25.89 µ m emis-sion line can be produced both by a weak AGN as well as fromshocks near massive stars (i.e. Lutz et al. 1998; Bernard-Salas etal. 2009). However, we do include it for reasons of complete-ness.Based on the EQW of the 6.2 µ m PAH feature, 12 of ourLIRGs are mid-IR AGN, i.e. have EQW less than 0.27 µ m.Cross-correlating the results from [Ne V] detection and 6.2 µ mEQW we find 4 LIRGs that are identified as AGN based on thosetwo mid-IR diagnostics (NGC 2623, UGC 05101, UGC 08696and IC 5298), which are also radio-AGN. The AGNs used for comparison are I Zw 1, NGC 1275, Mrk 3, PG0804 + + + + +
099 (Weedman et al.2005). The comparison starburst galaxies are NGC 660, NGC 1222, IC342, NGC 1614, NGC 2146, NGC 3256, NGC 3310, NGC 4088, NGC4385, NGC 4676, NGC 4818, NGC 7252, and NGC 7714 (Brandl et al.2006; Bernard-Salas et al. 2009).Article number, page 8 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs
Fig. 2.
Each row corresponds to an object from our sample. (
Le f t ) Radio-spectral-index maps ( α -maps) are represented with the colour-map;the colour bar on the right shows the radio spectral index values from low (flat α ) to high values (steep α ). Overlaid, the radio contours at 1.49GHz (black dashed) and at 8.44 GHz (white solid), where the contour levels scale as log of the flux density from the lowest flux value (3 σ ) tothe highest. ( Middle ) The error map for the radio spectral index per pixel value. The colour bar on the right shows the range of values. Contoursas on the
Le f t . In both plots, on the bottom left corner we give the beam of the 1.49 GHz map as black circle, and the beam of the 8.44 GHz mapas grey circle / ellipse. On the bottom right we give a scale-bar of 2 arcsec, as visual aid. ( Right ) Histograms of the α -maps: ( T op ) Histograms ofthe radio spectral index, based on Fig. 2-
Le f t , that give the values of α at each pixel. ( Middle ) Histograms where the errors on α (Fig. 2- Middle )have been added. (
Bottom ) Histograms where the errors on α have been subtracted. The rest of the figure is located in the Appendix. The 6.2 µ m EQW diagnostic gives us 12 AGN, 13 SB and12 AGN / SB, where objects below 0.27 µ m are AGN, objectsabove 0.54 µ m are SB and objects with EQW in-between thosevalues are AGN / SB. From the rest 9 objects without a EQWentry in Table 4, one system (ESO 550-IG025 N & S) wasnot observed, and for the rest the archival SL staring mode ofthe observations was not centred on the galaxy nucleus (IRAS F01417 + + + µ m EQW, 2 have a [Ne V] detection (UGC 08387 and VV340aN; see notes on the objects in Appendix A), and thus are mid-IRAGN based on [NeV]. Furthermore, 6 of these 13 SB are classi-fied as radio-AGN: UGC 02369 S, UGC 04881 SW, MCG + Article number, page 9 of 32 ( L IR / L O • )2.02.53.03.5 q I R IC1623A/BUGC05101NGC5256NEVV705S VV340a ( L / W Hz -1 )2.02.53.03.5 q I R IC1623A/BUGC05101VV705SVV340aNGC5256NE
Fig. 3. ( Le f t ) q IR vs infrared luminosity L IR / − µ m in units of solar luminosity. ( Right ) q IR versus radio luminosity at 1.49 GHz (see Table 1).In both figures, the blue diamonds represent the 46 individual objects in our sample, where objects identified as radio-AGN based on the α -mapsare shown as filled blue diamonds (see Table 4). Black squares denote objects with mid-IR fraction above zero (Petric et al. 2011), thus they havea contribution to their mid-IR flux from an AGN. Red filled circles show the mean value of q IR for each luminosity bin and the error bars arethe standard deviation; note that the median luminosity is calculated instead of the mean to avoid edge e ff ects due to bin selection. The dashedhorizontal line shows the mean q IR for the sample with value of 2.75 ± σ error is shown by the horizontal dotted lines. Objects with q IR values below the 1 σ error lines are marked with their name. + / IPAC Extragalactic Database (NED; 10out of 46 objects), or from Vega et al. (2008) and Mazzarella etal. (2012) for 30 out of 46 objects, or by performing an anal-ysis of our objects using the BPT diagram. For the latter weuse the SDSS DR7 (Abazajian et al. 2009) catalogue. Aftercross-correlating it to our radio RA and Dec, we obtained fluxesfrom optical spectroscopy for 22 out of 46 objects. We thenused the standard BPT diagram (Kau ff mann et al. 2003), [OIII] / H β versus [N II] / H α to separate star-forming (SF) objects,from composite (COMP) and AGN, and the [O III] / H β versus[S II] / H α and [O III] / H β versus [OI] / H α diagrams to separateSeyferts from LINERs (Kewley et al. 2006). We present thesediagrams in Fig. 5. From the 22 objects, 7 are SF objects, 7COMP, 2 Seyfert and 4 LINER, while two (IRASF01364-1042and IRASF08572 + µ m PAH we get 12 AGN, 13 SBand 12 composite objects; c) while from the optical diagnosticswe get 12 Seyferts, 7 LINERs, 6 SF, 16 composites and 3 HII re-gions. We conclude that 3 objects out of the 46 in our sample ( ∼ + + ∼ ∼ L IR / L (cid:12) ) ≥ ff er from blending from a circum-nuclear SB and an AGN. In the case of enhanced star-formationin the circumnuclear region detecting the destruction of PAHsdue to the X-ray hard-UV photos from an AGN is challengingsince the PAH EQW from the surrounding regions will be large.Based on the 6.2 µ m PAH EQW we have ∼
26% of objects con-
Article number, page 10 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs taining an AGN, while based on the [Ne V] detection this dropsto ∼ Based on Condon et al. (1991) and Murphy et al. (2013), thevalue of the radio spectral index calculated between 1.49 and8.44 GHz is connected to the compactness of the starburst, i.e.the flatter the radio spectral index the more compact the emit-ting area. If a flat radio spectral index is associated with com-pactness, then by calculating the area where the radio spectralindex is flat we can estimate the size of the compact starburst.Here we calculate the flat area in kpc from the α -maps. Thiscorresponds to the area where the radio spectral index is ≤ α -maps. Objects that have unresolved (COM) radio structuresat both 1.48 and 8.44 GHz (see Table 1) are placed in brack-ets. These are IRAS F01417 + + + + + + α ≤ .The mid-IR sizes are calculated from the 13.2 µ m continuum inthe IRS spectrum of each object (Diaz-Santos et al. 2010b). Thespatial profiles of each source along the slit were extracted andfitted with a Gaussian. The stellar profiles were also fitted witha Gaussian and scaled to the galaxy profile. Finally the stellarPSF was subtracted to give the intrinsic size of emission. Inorder to compare to the flat area, we assume a circular mid-IRarea in kpc (Table 5). Ideally we want to compare the flat areadetermined above, with the mid-IR area as measured from mid-IR imaging, but these measurements are not available at the timeof writing. Since the mid-IR and flat areas, as given in Table 5,were not calculated in a similar manner, we cannot perform aone-to-one comparison.In total, areas where the radio spectral index is flat range be-tween ∼ , with a mean of 45 ±
24 kpc . Excludingunresolved and barely resolved objects at 1.49 GHz the range is ∼ . The mid-IR areas for the sample presented hererange between ∼ with a mean of 9 ±
11 kpc . In gen-eral the mid-IR areas are much smaller, but this may be partlyattributed to the di ff erent methods of calculating the mid-IR andradio areas. Radio-area measurements are based on a 2D image,while mid-IR-area measurements are based on a 1D cut extrap-olated to 2D assuming azimuthal symmetry (Diaz-Santos et al.2010b); the latter would give roughly the same measurement foran object that is symmetric. Also, we have to note that mostmid-IR areas are upper limits (see Table 5). Thus we will refrainfrom drawing any strong conclusions from this analysis.In order to compare the objects in our sample we group themby radio classification, as shown in Table 3. These results sug-gest that the areas where the radio spectral index is flat are notstatistically di ff erent for radio-AGN and radio-SB. This also ap-plies to the mid-IR areas, as both radio-AGN and radio-SB dis- Table 3.
Range of flat-alpha radio area and mid-IR area per radio class radio-class radio area mid-IR area(kpc ) (kpc )(1) (2) (3)AGN 5-100 (44 ±
26) 0.6-44 (11 ± ±
25) 0.6-45 (13 ± / SB 16-65 (46 ±
15) 0.5-26 (5 ± Notes.
Range of area in the radio where the radio spectral index is flatand of mid-IR area in kpc grouped by radio class, excluding unresolvedsources. Values in parenthesis give the mean and standard deviation. play comparable ranges. Still, the radio area that correspondsto flat radio spectral index is larger than the mid-IR area whencomparing objects grouped by their radio classification, as in Ta-ble 3.In the radio, the area that the radio spectral index is flat cor-responds to the area over which the SB is very compact and theopacity is > top ), and the mid-IR size versus the radio spectral index( bottom ). We see a decrease of radio size with increasing ra-dio spectral index meaning the steeper the radio spectral index,the smaller the size of the flat area, which is expected. The ob-ject with the smallest flat- α size is VV705 (S), and is an upperlimit. We calculate the Spearman rank correlation coe ffi cient us-ing censoring for the upper limits with the available packages inIRAF. We find a negative correlation of -0.255 with 10% proba-bility that a correlation is not present. Similarly, we find a nega-tive correlation in Fig. 6- bottom , with ρ = -0.189 and 21% prob-ability that a correlation is not present. We see that the mid-IRarea decreases as the integrated radio spectral index steepens.There are three objects that do not follow this trend (NGC 5257NW, VV 340a N and NGC 6090), as they have large mid-IR sizeand very steep radio spectral indices. NGC 6090 is very di ff useat 8.44 GHz with low flux density, giving a high radio spectralindex between 1.49 and 8.44 GHz. Our results are limited due tothe small number statistics and the large amount of upper limitsin the mid-IR areas. We present this figure as a general compar-ison to the flat areas, but we will refrain from making a generalstatement about it, also because of the way the mid-IR areas werecalculated. Using our sample we explore the ratio of the total infrared lumi-nosity L IR to the rest-frame luminosity at 8 microns L µ m (IR8),which is a star-formation compactness indicator (Daddi et al.2010a; Elbaz et al. 2011). Star-forming galaxies follow the in-frared main sequence, defined by IR8 ≈ σ ≈ + Article number, page 11 of 32 .e. IR8 = µ m luminosity can be the result of a dust-obscuredAGN, hidden by the SB. To reveal a possible contribution from adusty AGN we correct for starburstiness R SB (Elbaz et al. 2011),which is the ratio of the specific star-formation-rate (sSFR) tothe specific star-formation-rate of a galaxy of the main sequenceat the same redshift as our object (sSFR MS ). We calculate thesSFR using the relation sSFR [yr − ] = L IR / M (cid:63) × × − (Kennicutt 1998), where L IR the infrared luminosity as in Ta-ble 1 and M (cid:63) the stellar mass (Howell et al. 2010; Diaz-Santoset al. 2013), and the sSFR MS [Gyr − ] = × t − . (Elbaz et al.2011), where t is the cosmic time elapsed since the Big Bang inGyr calculated using at the redshift of each object.We correct the IR8 values of the object in our sample for star-burstiness and find that 2 objects fall below 1 σ from the mean,i.e. IR8 / R SB = α Here we present the ratio of the flat area to the total area ( f f ) asgiven in the α -maps, with the resolution of the 1.49 GHz maps,for fluxes above 3 σ (Fig. 7). We see a decrease in the fraction f f as the radio spectral index steepens. This is expected as the areawhere the radio spectral index is flat gets smaller as the medianradio spectral index in the α -map gets steeper (larger values).Unresolved sources that display low median values of α havelarge fractions of f f where we believe the AGN dominates forobjects where median α < α and the limit above where we do notexpect to see a steepening of the radio spectral index in compactstarbursts ( α ∼ α -histograms we observe that in sev-eral objects, the radio spectral index reaches values above α = α shown by the error bars in Fig. 7. Thusexamining the histogram by eye and not depending solely on theresults of Fig. 7 is important for the interpretation of the data.We attribute these long steep tails to the existence of an AGN, asdiscussed in Sec. 3.1.We also compare the median values of α to the flat area andthe brightness temperature Tb, but we do not find an obviouscorrelation. Just a trend for higher brightness temperatures andflat areas to correspond to flatter mean α .
5. Conclusions
We present results from the study of radio continuum observa-tions of 35 LIRG systems, or 46 individual galaxies, from theGOALS sample, observed at both 1.49 & 8.44 GHz with theVLA (Condon et al. 1990, 1991). Our objective is to identify theAGN component in these systems using radio observations thatare largely una ff ected by dust obscuration. We use the highestresolution maps available at the time of the study, down to 1 (cid:48)(cid:48) . (cid:48)(cid:48) .
25 resolution at 8.44 GHz, in orderto probe the central kpc of these LIRGs. These objects are se-lected by Condon et al. (1991) to be brighter in the infrared thanlog( L IR / L (cid:12) ) ≥ q IR values,i.e. brighter at radio than at IR wavelengths, contain an AGNbased on the radio-spectral-index maps and compare our resultsto optical and the following mid-IR AGN diagnostic tools: (a)the [Ne V] emission line and (b) the equivalent width of the 6.2 µ m PAH feature. Finally, we give the size of the area correspond-ing to flat radio spectral index, which is connected to compact-ness, and thus gives the size of the compact starburst at low-GHzradio wavelengths.In summary, our results are:1. Out of the 46 individual objects, we find 21 ( ∼ ∼ ∼ / SB.2. 8 out of the 46 ( ∼ ∼ .In order to separate the AGN and SB components more ac-curately than presented here, one needs higher-resolution obser-vations, e.g. with the EVLA, in order to resolve in greater detailthe radio structures associated with the centre of these LIRGs. Acknowledgements.
We would like to thank the anonymous referee whose care-ful reading and detailed comments greatly improved the manuscript. EV isfunded by the Action ”Supporting Postdoctoral Researchers” of the OperationalProgram ”Education and Lifelong Learning” (Action’s Beneficiary: GeneralSecretariat for Research and Technology), and is co-financed by the EuropeanSocial Fund (ESF) and the Greek State. EV would like to acknowledge HanaeInami and Jim Condon for most useful conversations, and Grigoris Maravelias,Thomas Robitaille and Laure Ciesla for their most helpful assistance in Python.VC would like to acknowledge partial support from the EU FP7 Grant PIRSES-GA-2012-316788.
References
Abazajian, K. N., Adelman-McCarthy, J. K., Agueros, M. A., et al. 2009, ApJS,182, 543Alonso-Herrero, A., Pereira-Santaella, M., Rieke, G. H., et al. 2013, ApJ, 765,78Antonucci, R., 1993, ARA&A, 31, 473Appleton, P. N., Fadda, D. T., Marleau, F. R., et al. 2004, ApJS, 154, 147Armus, L., Heckman, T., Miley, G. 1987, AJ, 94, 831
Article number, page 12 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs
Armus, L., Charmandaris, V., Bernard-Salas, J., et al. 2007, ApJ, 656, 148Armus, L., Mazzarella, J. M., Evans, A. S., et al. 2009, PASP, 121, 559Baldwin, J. A., Phillips, M. M., Terlevich, R. 1981, PASP, 93, 5Bell, E. F. 2003, ApJ, 586, 794Bernard-Salas, J., Spoon, H. W. W., Charmandaris, V., et al. 2009, ApJS, 184,230Brandl, B. R., Bernard-Salas, J., Spoon, H. W. W, et al. 2006, ApJ, 653, 1129Blundell, K. M., Rawlings, S., 2001 ASPC, 250, 363Calzetti, D., 2010, arXiv:1010.4996Capetti, A., Zamfir, S., Rossi, P., et al. 2002, A&A, 394, 39Caputi, K. I., Lagache, G., Yan, Lin et al. 2007, ApJ, 660, 97Carilli, C. L., Barthel, P. D., 1996, A&ARv, 7, 1Clemens, M. S., Vega, O., Bressan, A., et al. 2008, A&A, 477, 95Clemens, M. S., Scaife, A., Vega, O., et al. 2010, MNRAS, 405, 887Condon, J. J., Helou, G., Sanders, D. B., et al. 1990, ApJS, 73, 359Condon, J. J., Huang, Z.-P., Yin, Q. F., et al. 1991, ApJ, 378, 65Condon, J. J. 1992, ARA&A, 30, 575Daddi, E., Elbaz, D., Walter, F., et al. 2010a, ApJ, 714, L118Desai, V., Armus, L., Spoon, H. W. W., et al. 2007, ApJ, 669, 810Diaz-Santos, T., Armus, L., Charmandaris, V., et al. 2013, ApJ, 774, 68Diaz-Santos, T., Charmandaris, V., Armus, L., et al. 2011, ApJ, 741, 32Diaz-Santos, T., Charmandaris, V., Armus, L., et al. 2010b, ApJ, 723, 993Elbaz, D., Dickinson, M., Hwang, H. S., et al., 2011, A&A, 533, 119Elbaz, D., Cesarsky, C. J., Chanial, P., et al. 2002, A&A, 384, 848Evans, I. N., Primini, F. A., Glotfelty, K. J., et al. 2010, ApJS, 189, 37Fanaro ff , B. L., Riley, J. M., MNRAS, 167 ,31Genzel, R., Lutz, D., Sturm, E., et al. 1998, ApJ, 498, 579Haan, S. , Surace, J. A., Armus, L., et al. 2011, ApJ, 141, 100Hasinger, G., Miyaji, T., Schmidt, M. 2005, A&A, 441, 417Helou, G., Soifer, B. T., Rowan-Robinson, M. 1985, ApJ, 298, 7Howell, J. H., Armus, L., Mazzarella, J. M., et al. 2010, ApJ, 715, 572Huynh, M. T., Gawiser, E.; Marchesini, D., et al. 2010, ApJ, 723, 1110Inami, H., Armus, L., Charmandaris, V., et al. 2013, ApJ, 777, 156Inami, H., Armus, L., Surace, J. A., et al. 2010, AJ, 140, 63Ivison, R. J., Swinbank, A. M., Swinyard, B., et al. 2010, A&A, 518, 35Iwasawa, K., Sanders, D. B., Teng, S. H., et al. 2011, A&A, 529, 106Kau ff mann, G., Heckman, T. M., Tremonti, C., et al. 2003, MNRAS, 346, 1055Kennicutt, R. C. 1998, ARAA, 36, 189,Kewley, L. J., Dopita, M. A., Sutherland, R. S., et al., 2001, ApJ, 556, 121Kewley, L. J., Groves, B., Kau ff mann, G., et al. 2006, MNRAS, 372, 961Kim, D.-C.; Sanders, D. B.; Veilleux, S., et al. 1995, ApJS, 98, 129Krolik, J. H., 1999, ’Active galactic nuclei : from the central black hole to thegalactic environment’, Princeton University PressKukula, M. J., Dunlop, J. S., Hughes, D. H., et al. 1998, MNRAS, 297, 366Lacy, M., Storrie-Lombardi, L. J., Sajina, A., et al. 2004, ApJS, 154, 166Laurent, O., Mirabel, I. F., Charmandaris, V., et al. 2000, A&A, 359, 887Le Floc’h, E., Papovich, C., Dole, H., et al. 2005, ApJ, 632, 169Leipski, C., Falcke, H., Bennert, N., et al. 2006, A&A, 455, 161Leroy, A. K., Evans, A. S., Momjian, E., et al. 2011, ApJ, 739, 25Lobanov, A. P., 1998, A&A, 330, 79Longair, M. S., 2011, High Energy Astrophysics, Cambridge, UK, CambridgeUniversity PressLonsdale, C. J., Lonsdale, C. J., Smith, H. E., et al. 2003, ApJ, 592, 804Lutz, D., Kunze, D., Spoon, H. W. W., et al. 1998, A&A, 333, 75Magdis, G. E., Rigopoulou, D., Helou, G., et al. 2013, A&A, 558, 136Magnelli, B., Elbaz, D., Chary, R. R., et al. 2009, A&A, 496, 57Mao, M. Y., Norris, R. P., Emonts, B., 2014, MNRAS, 440, 31Mazzarella, J. M., Iwasawa, K., Vavilkin, T., et al. 2012, AJ, 144, 125Murphy, E. J., 2013, ApJ, 777, 58Murphy, E. J., Stierwalt, S., Armus, L., et al. 2013, ApJ, 768, 2Murphy, E. J., Condon, J. J., Schinnerer, E., et al. 2011, ApJ, 737, 67Murphy, E. J., 2009, ApJ, 706, 482Neugebauer, G., Habing, H. J., van Duinen, R., et al. 1984, ApJ, 278, 1Niklas, S., Klein, U., Wielebinski, R., 1997, A&A, 322, 19O’Dea, C. P., 1998, PASP, 110, 493Paragi, Z., Frey, S., Fejes, I, et al. 2000, PASJ, 52, 983Petric, A. O., Armus, L., Howell, J., et al. 2011, ApJ, 730, 28Prouton, O. R., Bressan, A., Clemens, M., et al. 2004, A&A, 421, 115Rich, J. A., Kewley, L. J., Dopita, M. A., 2011, ApJ, 734, 87Sanders, D. B., Soifer, B. T., Elias, J. H., et al. 1988, ApJ, 325, 74Shanks, T., Georgantopoulos, I., Stewart, G. C., et al. 1991, Nature, 353, 315Soifer, B. T.; Neugebauer, G.; Matthews, K., et al. 2000, AJ, 119, 509Spoon, H. W. W., Moorwood, A. F. M., Lutz, D., et al. 2004, A&A, 414, 873Stierwalt, S., Armus, L., Charmandaris, V., et al. 2014, ApJ, 790, 124Stierwalt, S., Armus, L., Surace, J. A., et al. 2013, ApJS, 206, 1Tommasin, S., Spinoglio, L., Malkan, M. A., et al. 2010, ApJ, 709, 1257Vega, O., Clemens, M. S., Bressan, A. et al. 2008, A&A, 484, 631Veilleux, S., Rupke, D. S. N., Kim, D.-C, et al. 2009, ApJS, 182, 628Veilleux, S., Kim, D.-C., Sanders, D. B., 1995, ApJS, 98, 171Weedman, D. W., Hao, L., Higdon, S. J. U., et al. 2005,Wu, Y., Charmandaris, V., Huang, J., et al. 2009, ApJ, 701, 658 µ m PAH EQW ( µ m)0.0010.0100.1001.00010.000 [ N e V ] / [ N e II]
AGN100%50%25%10% SB100%50%25%10% µ m PAH EQW ( µ m)0.010.101.0010.00 [ O I V ] / [ N e II]
AGN100%50%25%10% SB100%50%25%10%
Fig. 4. [NeV] / [NeII] ( top ) and [OIV] / [NeII] ( bottom ) line ratios (In-ami et al. 2013) for the 46 objects in our sample (blue diamonds) versusthe equivalent width (EQW) of the 6.2 µ m PAH feature (EQW valuesfrom Stierwalt et al. 2013); filled blue diamonds show objects identi-fied as radio-AGN based on our radio diagnostic method, while blacksquares denote objects with mid-IR fraction above zero (Petric et al.2011). For comparison we show a sample of AGN (red open trian-gles) and a sample of SB (including LIRGs; green open circles) takenfrom Armus et al. (2007). The horizontal line gives the SB contribu-tion, where the 100% is the mean value of the 6.2 µ m PAH EQW. Thevertical line shows the AGN contribution, where as 100% we take themean value of the [NeV] / [NeII] ratio. Article number, page 13 of 32 able 4. Comparison of radio and mid-IR AGN diagnostics
Name radio [Ne V] / [O IV] / µ m PAH EQW ( µ m) opticalclassification [Ne II] [Ne II] (and classification) classificationNGC 0034 (S) AGN (6) < < / SB) Sey2
V08 / Sey
BPT(r) / Sey2
VX95
IC 1623A / B AGN (6) < / SB) SF
BPT(r) / HII
VX95
CGCG 436-030 AGN / SB (1) < / SB) COMP
BPT(s) / LINER, HII
V08
IRAS F01364-1042 AGN / SB (1) < < / SB) LINER
V08 , NED
IRAS F01417 + V08
UGC 02369 (S) AGN (5) < < V08 (t)
IRAS F03359 + / SB (1) < < BPT(r) / HII
VX95
ESO 550-IG025 (N) AGN / SB (1) — — — COMP
BPT(r) / LINER
VX95
ESO 550-IG025 (S) AGN (3) — — — COMP
BPT(r) / LINER
VX95
IRAS F05189-2524 SB (2) 0.71 1.46 0.03 (AGN) Sey1
NED / Sey2
V08
NGC 2623 AGN (1) 0.06 0.16 0.27 (AGN) COMP
BPT(s) / LINER, Sey2
V08
IRAS F08572 + < < V08
UGC 04881 (NE) AGN / SB (1) < < / SB) COMP
BPT(s) / HII
VX95 / HII
V08(t)
UGC 04881 (SW) AGN (3) — — 0.61 (SB) COMP
BPT(r) / HII
VX95 / HII
V08 (t)
UGC 05101 AGN (1) 0.08 0.20 0.13 (AGN) SF
BPT(s) / Sey1
NED / LINER, Sey1.5
V08
IRAS F10173 + < < / SB) COMP
BPT(s)
IRAS F10565 + < < / SB) COMP
BPT(r) / HII
V08
MCG + < < V08
UGC 06436 (NW) AGN / SB (1) < < BPT(s) / LINER, HII
V08(t) / SB NED(t)
UGC 06436 (SE) SB (1) — — — SF
BPT(s) / LINER,HII
V08(t) / SB NED(t)
NGC 3690 (E) AGN / SB (5) < / SB) COMP
BPT(s)
NGC 3690 (W) AGN / SB (5) < BPT(s)
IRAS F12112 + < / SB) LINER
BPT(s) / HII
V08
IRAS F12112 + / SB (4) < / SB) SF
BPT(s) / HII
V08
UGC 08058 AGN (1) < < V08 , NED
VV 250a (NW) AGN / SB (1) < BPT(r) / HII
VX95
VV 250a (SE) SB (2) < < BPT(s)
UGC 08387 AGN (1) 0.01 0.06 0.62 (SB) LINER
BPT(s) / LINER, HII
V08
NGC 5256 (NE) SB (1) — — — SF
BPT(s) / LINER
M12
NGC 5256 (SW) AGN (1) 0.15 — 0.44 (AGN / SB) LINER
BPT(s) / Sey2
M12 , NED
NGC 5256 (C) AGN / SB (1) — 0.98 — —NGC 5257 (NW) SB (2) < BPT(r) / HII
VX95
UGC 08696 AGN (1) 0.23 1.07 0.12 (AGN) SF
BPT(s) / Sey2
V08 , NED / LINER
V08
IRAS F14348-1447 (NE) AGN / SB (1) — < V08
IRAS F14348-1447 (SW) AGN / SB (1) < V08
VV 340a (N) AGN (3) 0.05 0.43 0.58 (SB) COMP
BPT(r) / LINER
VX95
VV 705 (N) AGN / SB (1) < < BPT(s) / HII
V08 (t)
VV 705 (S) AGN (3) — — 0.75 (SB) HII
V08 (t)
IRAS F15250 + < < BPT(s) / LINER
V08
UGC 09913 AGN / SB (1) < < NED / SF BPT(s) / HII, LINER
V08
NGC 6090 AGN (3) < BPT(s)
IRAS F17132 + < < BPT(r) / HII
VX95
IRAS F17132 + / SB(1) < < / SB) COMP
BPT(r) / HII
VX95
IC 5298 AGN (1) 0.35 0.41 0.12 (AGN) Sey2
V08 , NED / LINER, HII
V08
Mrk 0331 (N) AGN (1) < < V08
Notes.
Columns: (1) Name of the galaxy; systems are marked with the relative position to each other (e.g. N, S, E, W). (2) Classification based onradio spectral index maps and histograms. In parenthesis we give the histogram category. (3) Ratio of [Ne V] to the [Ne II] emission line: if the[Ne V] line is detected then the galaxy contains an AGN; if detection is an upper limit then it is a possible AGN; and if there is no high-resolutionspectrum for the object, then we mark with ’—’. (4) Ratio of the [O IV] to the [Ne II] emission line; if there is no high-resolution spectrum forthe object, then we mark with ’—’. (5) EQW in microns of the 6.2 µ m PAH feature. AGN or SB classification are based on empirical values asfollows: values below 0.27 µ m suggest the presence of an AGN; values above 0.54 µ m denote a SB; and values in-between suggest the presenceof both AGN and SB (see Stierwalt et al. 2013; Murphy et al. 2013). (6) Optical classification based on optical spectroscopy; reference is given ineach case. For classification based on the BPT diagram (s stands for SLOAN data and r for RBGS data; (Kim et al. 1995)), SF stands for objectin the star-forming region, COMP stands for objects between star-forming and AGN regions in the BPT (e.g. Kewley et al. 2001; Kau ff mann etal. 2003; Kewley et al. 2006); Sey stands for Seyfert with subcategories 1, 1.5 and 2. References: ’NED’ for objects drawn from the NASA / IPACExtragalactic Database (http: // ned.ipac.caltech.edu); ’BPT’ from the BPT diagram; ’M12’ from Mazzarella et al. (2012); ’V08’ from Vega et al.(2008); ’VX95’ from Veilleux et al. (1995); (t) stands for reference for the galaxy system.Article number, page 14 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs Table 5.
Size of area corresponding to flat radio spectral index, it’s fraction to total, and mid-IR area
Name flat area fraction of flat area mid-IRto total area area(kpc ) in α -map (kpc )NGC 0034 (S) 4.84 + . − . / B 7.27 + . − . + . − . + . − . + + . − . + + . − . + . − . + . − . + . − . + + . − . + . − . + + . − . + + . − . + + . − . + . − . + . − . − . + . − . + + − . + . − . + . − . + . − . + . − . + . − . − . + . − . + . − . + . − . + . − . + . + + . − . + + + . − . + . − . + . − . + . − . Notes.
Columns: (1) Name of the galaxy; systems are marked with the relative position to each other (e.g. N, S, E, W). (2) Square kpc of the areacorresponding to flat radio spectral index ( α ≤ α -maps. The errors were calculated by taking into account the error per pixel as shown inthe error α -maps. Values in parenthesis are upper limits, while square brackets denote unresolved sources. (3) Fraction of the flat area to the totalarea marked by the α -map (Fig. 2). (4) Size of area of mid-IR emission at 13.2 µ m in kpc ; the calculation method is described in Diaz-Santos etal. (2010b); values in parenthesis denote upper limits, meaning the measured size is <
10% larger than the instrumental size (PSF) so an intrinsicsize for the emitting region couldn’t be calculated. The reported upper limits are on the measured size (not intrinsic), i.e. 1.1 × PSF size. For objectsmarked with ’—’ there is no available measurement. Note: IRAS F12112 + + α -map). Here we give the flat area of the SW radio component which is associatedwith the most luminous infrared source. The total flat area of NGC 5256 (SW) and (C) is 47.19 + . − . kpc . We put a limit on NGC5256 (C) becausethe area is partly overlapping with the southernmost component. Article number, page 15 of 32 ([NII]/H α )−2−101 L og ([ O III] / H β )
315 2324 4127 1421 221629 3033 393720 1911 2840
SF COMP AGN
127 9814172632 364244 −2 −1 0 1Log ([SII]/H α )−1.0−0.50.00.51.0 L og ([ O III] / H β )
315 232441272122 1629 3033 39372019 11 2840 127 14172632364244
SF LINERSey −2.5 −2.0 −1.5 −1.0 −0.5 0.0 0.5Log ([OI]/H α )−1.0−0.50.00.51.0 L og ([ O III] / H β ) SF LINERSey
Fig. 5.
BPT diagrams showing emission line ratios [OIII] / H β versus [NII] / H α ( le f t ), [SII] / H α ( middle ) and [OI] / H α ( right ). In all figures, theblue diamonds represent the 46 individual objects in our sample, where objects identified as radio-AGN based on the α -maps are shown as filledblue diamonds (see Table 4). Numbers next to the symbols represent the objects in our sample as given in Table 1. On the le f t graph, the solidand dotted lines separate the SF galaxies from the AGN and mark the area where composites (COMP) are located (Kau ff mann et al. 2003). Thesolid lines on the middle and right graphs are from Kewley et al. (2006) and separate the SF from Seyfert (Sey) and LINERs.Article number, page 16 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs Appendix A: Notes on the objects
NGC 0034
This is a galaxy pair, but only the south companionhas been observed at both frequencies used in our analysis. Thecentre of radio emission at 1.49 GHz does not coincide with theHST bulge position (Haan et al. 2011), but is ∼ ff set is of the order of ∼ α -map, where we see a transitionof the values of α from flat to steep as we go from SE to NW.Furthermore, there is no indication from the study of Inami etal. (2013) that there is shocked line emission based on IRS spec-troscopy of this object. Prouton et al. (2004) estimate a 40%contribution from the AGN to the total infrared luminosity, andbased on the optical diagnostics presented in this paper this ob-jects is associated with a Seyfert 2 (Table 4). We classify thisobject as a radio-AGN because of the shift, and because the ra-dio spectral index in the α -map exceeds out theoretical value of1.1. IC 1623A / B Condon et al. (1991) does not give a measurementfor the flux density at 8.44 GHz of this source, but we provide aflux density above 3 σ in Table 1 measured from the map. Thesource is large (24 arcsec long) at 1.49 GHz, but due to the smallangular size of the 8.44 GHz map, only a part of the object isobserved, hence the α -map is small. The radio spectral indexhas a wide range of values. The maximum flux pixel in the 1.49-GHz map is not at the same R.A. and Dec as the maximum fluxpixel in the 8.44-GHz map. This o ff set is of the order of ∼ α f l a t a r ea ( kp c ) IC1623A/BVV705SVV705NGC5256C α m i d -I R a r ea ( kp c ) UGC04881SWNGC5257NW VV340aNGC6090
Fig. 6. ( T op ) Size of area corresponding to flat radio spectral index(in kpc ) versus integrated radio spectral index calculated between 1.49& 8.44 GHz. The dotted horizontal line marks the flat to steep divisionin radio spectral index. The Spearman ρ = -0.255 with 10% probabilitythat the correlation is not present. ( Bottom ) Mid-IR area in kpc of thecore of mid-IR emission, assuming emission from a circular area (seeSection 4.3), versus integrated radio spectral index (see Table 1). TheSpearman ρ = -0.189 with 21% probability that the correlation is notpresent. In both plots, the dotted vertical line marks the flat to steepdivision in the radio spectral index. Symbols as in Fig. 3; unresolvedobjects are marked with circles instead of diamonds. the shift to be real. As a result, the flat radio-spectral-index areais o ff set from the radio centre at 1.49 GHz, but matches the peakbrightness at 8.44 GHz. Rich et al. (2011) find shocked emis-sion and outflow associated with the east galaxy, as a result ofthe merging process in this late type merging galaxy system. Onthe contrary, there is no indication from the study of Inami et al.(2013) that there is shocked line emission based on IRS spec-troscopy of this object, and the FeII line present in the spectrum, Article number, page 17 of 32 α fr ac ti on o f f l a t - a r ea t o t o t a l Fig. 7.
The ratio of size of the area corresponding to flat radio spec-tral index (in kpc ) as in Table 5, to the total radio size as in the α -mapsversus median radio spectral index calculated from the α -maps. The dis-persion (standard deviation) around the median is shown with horizon-tal dotter error-bars. Symbols: blue open diamonds denote the objectsof our sample, blue filled diamonds denote objects that are classifiedas radio-AGN based on the α -maps, and black squares denote objectsthat have an AGN contribution in their mid-IR luminosity (Petric et al.2011); unresolved objects are marked with circles instead of diamonds.The vertical dotted lines at 0.5 and 1.1 mark the flat-steep division andthe limit above where we don’t expect steepening from compact star-burst, respectively. a good indicator of shocked emission, is probably not dominatedby shocks. We classify this object as a radio-AGN because of theshift, and because the radio spectral index in the α -map exceedsout theoretical value of 1.1. CGCG 436-020
The α -map and the α -histogram show a flatrange of values ( ∼ α barely exceed our theoretical cut value toseparate AGN from SB of α = / SB in the radio. We note,there is a HST bulge associated with the centre of radio emission(Haan et al. 2011), as additional evidence an AGN could precent.
IRAS F01364-1042
From the α -map of this object we see thatthe centre of radio emission is associated with a very flat indexvalue, which is slightly steepening outwards. Based on our clas-sification, this steepening might be due to the AGN, but since itdoes not exceed much, within the uncertainties, our theoreticallimit it can also be due to a SB located in an HII region. Thenthe CR electrons as they escape the SB are losing energy whichis shown as steepening in the radio spectral index. We classifythis object as AGN / SB.
IRAS F01417 + This is a COM radio source with flat rangeof α values and we classify it as a SB. To be an AGN, the bright-ness temperature should be above 10 -10 K (e.g. Clemens et al.2008; Condon et al. 1991), but this object has a value of ∼ × K. Note: the inverted values of α are located at the edges of the α -maps, and because the errors are of the order of 0.2, we don’ttake them into consideration. UGC 02369 (S)
Only the south companion has been observed at8.44 GHz. From the radio structure at 8.44 GHz we see a core with di ff use radio emission around it. There is no indication fora weak jet from the radio structure. There is no evidence that thisis an AGN based on mid-IR AGN diagnostics. Yet, the α -mapgives very steep values, above 1.1, suggesting an AGN. Thus weclassify this object as radio-AGN. IRAS F03359 + This radio source is associated with theeast companion of a galaxy pair, while the HST bulge is asso-ciated with the west companion. It has an extended radio sourceat 1.49 GHz, that presents a core-jet-like feature at 8.44 GHz.The radio spectral index at the centre of the source is flat, steep-ening outwards with values of α exceeding 1.1, but due to theuncertainties we will classify this as AGN / SB.
ESO 550-IG025 (N)
This is the north radio component of thegalaxy pair. This object is brighter in the radio than the IR, un-like the south companion where we observe the opposite. Thisgives a low value of q IR , but not lower than the 1 σ uncertainty(see Fig. 3). The north component presents a flat radio-spectral-index distribution steepening outwards. We have to note thoughthat the brightness temperature is low for this object to have apowerful AGN ( ∼ K; Table 2). Due to the uncertainties inthe steep values of α we classify it as AGN / SB.
ESO 550-IG025 (S)
This object is associated with the southcompanion of the galaxy pair, and the α -map exhibits steep val-ues that indicate synchrotron emission ( α ∼ α > IRAS F05189-2524
This object has a [NeV] detection makingit a mid-IR AGN. Similarly the 6.2 µ m PAH EQW suggests anAGN. The optical classification suggests a Seyfert galaxy, prob-ably of type 2. Based on the α -map and the α -histogram, this isa radio-SB, since the α -map gives values of α that are flat. TheSB is also suggested by it’s brightness temperature ( ∼ × K; Table 2). The α -map in this case does not help us identify theAGN in this ULIRG, although it could be a young AGN and wecould be seeing emission from a jet that is being formed. Evi-dence for this could be the slightly elongated towards N-S radiostructure at 8.44 GHz, but it can also be an artefact of the radiomap. NGC 2623
The mid-IR diagnostics classify this objects as anAGN. In the optical it is a composite, having both AGN and SBcontributions. The α -map shows a flat distribution, steepeningoutwards with values of α > IRAS F08572 + This object shows a compact core at 8.44GHz with several di ff use point-like structures in the surroundingarea. Vega et al. (2008) show that this object contains an AGNthat dominates the bolometric luminosity of the galaxy. Fromthe α -map we see that the core emission gives an inverted α ,getting flatter outwards. Because values of α at the core of radioemission are inverted, suggesting emission from a face-on jet,we classify this object as a radio-AGN. UGC 04881 (NE)
The north-east radio component of a galaxypair presents flat values of α that get steeper outwards, as seenby the α -histogram (Fig. 2- Right ), but values of α barely exceedthe value of 1.1, within the uncertainties, thus we classify it asAGN / SB. The mid-IR diagnostics suggest an AGN / SB, and inthe optical the BPT diagram gives a composite including bothAGN and SB. We see that the α -map and α -hist does not help usdistinguish between the AGN and the SB. UGC 04881 (SW)
The α -map of the SW radio component showssteep values of α ∼ ∼ . × K; Article number, page 18 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs
Table 2). Based on the 6.2 µ m EQW diagnostic, this is a SB,while based on the BPT diagram it is a composite. We classifyit as a radio-AGN. UGC 05101
The α -map shows a flat distribution in the centre,steepening outwards with values of α > IRAS F10173 + The α values are negative suggesting self-absorption as expected in a face-on AGN with the jet pointingtowards the observer. The brightness temperature is ∼ × K(Table 2), close to the value for AGN. We classify it as a radio-AGN.
IRAS F10565 + This is a compact radio source at 1.49GHz, but at 8.44 GHz it appears more structured. Based on thedistribution of values of the radio spectral index, and the his-togram, this is a radio-SB. The inverted values of α located at theedges of the α -map suggest optically thick region at a distanceof 2 arcsec from the centre of the radio source at 1.49 GHz. Hav-ing an optically thick region so far from the nucleus is rare, buthas been observed before (e.g. Inami et al. 2010). After check-ing the HST image of this galaxy we do not see any feature orexcess emission at the position where the radio spectral indexin inverted. We believe this is an area where the obscuration iseither too high even for radio observations to penetrate or theinverted values are caused by the very di ff use radio emission at8.44 GHz at the edges of this radio source. Since the errors areof the order of 0.2, we don’t take these values into consideration.
MCG + This is an extended radio source at 1.49 GHz,and at 8.44 it presents a jet-like structure (similar to FRI; Fa-naro ff & Riley 1974), whose centre is not associated with thecentre of IR emission at 8 µ m (map not shown here). The α values are flat in the centre of radio emission at 1.49 GHz, steep-ening outwards and exceeding the value of 1.1, which is alsoseen in the histogram. Based on our classification this is a radio-AGN. UGC 06436 (NW)
The north-west companion of this system ex-hibits flat values of α , steepening outwards, as seen in the α -mapand histogram exceeding the value of 1.1, but due to the uncer-tainties we classify it as AGN / SB. We note the low brightnesstemperature ( ∼ × K; Table 2) that does not favour a pow-erful AGN.
UGC 06436 (SE)
The α -map of the south-east radio companionof this galaxy pair has flat values in the centre steepening slightlyoutwards, not exceeding the value of 1.1. We classify this galaxyas a radio-SB. NGC 3690 (E)
The east radio component of this interacting sys-tem has a complex radio structure, and the α -map shows fourdiscrete areas. The central radio component, the one associ-ated with the bulge, has flat values of α in the centre steepen-ing rapidly outwards and getting higher than 1.1 suggesting aradio-AGN. The other areas have the characteristics of SB re-gions. The wide range of α values in the histogram are due tothe fact the radio source is composed of smaller radio compo-nents at 8.44 GHz. By adding up these separate α -maps createsa histogram with wide range of values of α . We classify thisobject as AGN / SB in the sense that it contains both AGN andSB.
NGC 3690 (W)
The west radio companion of this galaxy mergeralso has a complex radio structure, and in the α -map we see threediscrete areas. The NE area displays steep values of α gettinglarger than 1.1 going outwards, suggesting an AGN. The othertwo regions show a flat centre, steepening outwards slightly,probably from a SB. The wide range of α values in the histogram are due to the fact the radio source is composed of smaller radiocomponents at 8.44 GHz. By adding up these separate α -mapscreates a histogram with wide range of values of α . We classifythe west radio companion of the galaxy pair as AGN / SB in thesense that it contains both AGN and SB.
IRAS F12112 + This is a double radio source in agalaxy merger, where the NE component is associated with theNE merging galaxy. The distribution of α values goes from in-verted to flat, but the spatial distribution is particular in the sensethat from SW to NE α gets from inverted to flat values. Also,the centre of the NE radio component has highest values of α than the surroundings. We see that the NE radio component fallsmainly in category 2 of histograms. The inverted values of α arelocated at the edges of the radio source, and due to the uncer-tainty we do not take them under consideration. We classify it asa radio-SB. For the NE radio component we have a [OIV] mea-surement, which does not on it’s own confirm the presence ofan AGN. In order to obtain measurements for the total infraredluminosity, we found the fractional contribution of the two com-ponents at 8 µ m and applied this to the L IR of the system. Thevalues are given in Table 1. IRAS F12112 + The SW radio component of thisgalaxy pair is associated with the SW merging galaxy, and has acategory 2 histogram with some inverted values at the edges ofthe source. Due to the uncertainty we will classify it as AGN / SB.
UGC 08058
The inverted radio spectral index and the range ofvariations suggest emission coming from a face-on AGN withthe jets pointing towards the observer. We classify it as a radio-AGN.
VV 250a (NW)
This is the north-west companion of an in-teracting galaxy-system. The flat values of α in the centre ofthis source, steepening outwards are highly uncertain around thevalue of 1.1. Thus we cannot distinguish between an AGN or aSB. We classify it as AGN / SB.
VV 250a (SE)
Taking into account the errors in α and after look-ing at the histogram for this source, we classify it as category 2and thus a SB due to the flat values of α . UGC 08387
The map at 8.44 GHz reveals a jet-like radio struc-ture, FRI-like (Fanaro ff & Riley 1974). The centre of radioemission matches the HST bulge position. The α -map shows adistribution of flat values which are steepening outwards, gettinghigher than 1.1. We classify this as a radio-AGN. Although the6.2 µ m PAH EQW suggests this is a SB, we get a [NeV] detec-tion telling us this object contains an AGN. The question is, howis it possible for an object to have strong 6.2 µ m PAH featurethat was not destroyed by the AGN. One explanation could bethat the PDR is located in a circumnuclear region that does notget influenced by the AGN, thus the PAH feature is not beinga ff ected. NGC 5256 (NE)
This is the north-east companion of a galaxysystem, a compact radio source. Based on the distribution of val-ues of the radio spectral index and the histogram, and the steep-ening in the values of alpha at the edges of the source ( α <
NGC 5256 (SW) & (C) At 1.49 GHz the SW and C (for cen-tre) radio components are barely separated. The C radio com-ponent at 13:38:17.5 + µ msource (map not shown here). The source is in-between the two Article number, page 19 of 32 alaxies of this interacting system, is believed to be a very fastand short-lived shock that e ff ectively destroys most of the dustgrains, sublimates the molecular gas, and inhibits star formation(Mazzarella et al. 2012). Based on our radio classification, andthe steepening of α ( < / SB, in thesense that is associated with an AGN since it has the signature ofa radio-AGN. The SW radio component shows flat distributionof values steepening outwards rapidly and exceeding the valueof α = / SB can be explained by a circumnuclearstar-forming region that does not get a ff ected by the AGN. NGC 5257 (NW)
Only the north-west companion of this inter-acting galaxy-system has been observed at 8.44 GHz. This radiosource has a complex radio structure where the radio at 1.49 GHztraces the gas in the spiral arms of the NW galaxy. Only the nu-cleus was observed at 8.44 GHz due to the smaller field of viewof the 8.44 GHz observations compared to the one at 1.49 GHz.In Fig. A.1 we give just the radio source associated with the nu-cleus for presentation issues. The α -map and histogram show aflat α distribution, and thus we classify it as a radio-SB. We notethat there are some steep values at the edges of the central radiosource and the explanation is cooling of CR electrons; yet theuncertainty is of the order of 0.2 for these values for α . UGC 08696
At 8.44 GHz we see a complex system. There is adouble radio source associated with the IR source at 8 µ m (mapnot shown here), which has jet-like structure and has two addi-tional features, the brightest of which is located north-east of theradio centre. The α -map shows flattening in the centre, follow-ing the 8.44-GHz structure. The values get steeper towards theSE of the centre. Based on the α -map and the histogram this is aradio-AGN, due to the steepening of α > IRAS F14348-1447 (NE)
This is the NE radio component of adouble radio source, associated with a merging galaxy-system.The radio centre at 1.49 GHz is associated with the HST bulge.The radio sources are not completely separated, so the his-tograms created are imperfect. We see that the NE componentmatches category 1 of our classification, with α steepening above1.1 but due to the uncertainties we classify it as AGN / SB. It isalso a mid-IR AGN based on the PAH feature, and a LINER inthe optical. In order to obtain measurements for the total infraredluminosity, we found the fractional contribution of the two com-ponents at 8 µ m and applied this to the L IR of the system. Thevalues are given in Table 1. IRAS F14348-1447 (SW)
This is the SW radio component of adouble radio source, associated with a merging galaxy-system.The radio centre of the south component is not matching the oneat 8 µ m (map not shown here). The radio centre at 1.49 GHz isassociated with the HST bulge. The radio sources are not com-pletely separated, so the histograms created are imperfect. TheSW radio component matches category 1, with steepening above α = / SB.
VV 340a (N)
Only the north companion of the interacting sys-tem has been observed at both 1.49 & 8.44 GHz. The x-shaped radio structure at 1.49 GHz suggests a double-double ra-dio source or an X-shaped radio source (e.g. Capetti et al. 2002),i.e. a restarted radio source associated with an AGN. We do nothave further observations at 8.44 GHz to cover the whole areaseen at 1.49 GHz. Based on our classification and the steep val-ues of α ( > µ m EQW classifies it as a SB. Probably thereis a circumnuclear PDR region that does not get a ff ected by theAGN. Based on the BPT this is a SF object. VV 705 (N)
This is the north radio component of an interact-ing galaxy-system, showing flat α in the centre steepening out-wards with α > / SB. This object is classified as a SB based on the 6.2 µ m EQW. The small radio contour between the north and southradio components has a steep distribution of values suggestingsynchrotron emission, maybe as a result of the interaction be-tween the galaxies. VV 705 (S)
The south companion of this galaxy system showsvalues of α associated with synchrotron emission ( α ∼ α = µ m EQW. IRAS F15250 + The α -map of this compact radio sourceshows a negative distribution of values suggesting synchrotronself-absorption. A jet from an AGN pointing towards the ob-server is a possible explanation. The brightness temperature isof the order of 10 K (Table 2) which makes us confident to clas-sify it as a radio-AGN.
UGC 09913
At 8.44 GHz we see a double radio structure, withdi ff use radio emission around it. The centre of radio emission isassociated with inverted values of α suggesting synchrotron self-absorption, before emission becomes optically thin. The steep-ening of the radio spectral index from the centre outwards, asseen in the α -map and histogram, can be either from an AGNor due to CR electrons in an HII region. Thus we classify it asAGN / SB.
NGC 6090
This is a beautiful galaxy merger that displays a com-plex double radio structure at 1.49 GHz. Only the NE radio com-ponent has been observed at 8.44 GHz, and shows di ff use radioemission. The α -map gives a distribution of steep values sug-gesting a radio-AGN. In the mid-IR it is classified as a SB, andas a Seyfert based on the BPT diagram. IRAS F17132 + This is the north-east radio compo-nents of a galaxy pair that shows a distribution of steep valuesof α . We believe the NE component is a radio-AGN, due to theintense steepening of the radio spectral index further out fromthe core, suggesting synchrotron ageing from an AGN. We have[Ne V] and [O IV] measurement only for the NE component. Inorder to obtain measurements for the total infrared luminosity,we found the fractional contribution of the two components at8 µ m and applied this to the L IR of the system. The values aregiven in Table 1. Condon et al. (1991) do not give brightnesstemperature for the NE radio component. IRAS F17132 + The south-west radio componentshows flat values of α suggesting a compact SB. We do not havemid-IR nor optical classification for this object. IRAS F22491-1808
The HST bulge position agrees with the ra-dio centre at 8.44 GHz but not at 1.49 GHz, but the distancebetween the radio centres is negligible. The α -map gives a flatdistribution of values, steepening outwards, but we cannot distin-guish between AGN or SB due to the uncertainties. We classifythis is as AGN / SB.
IC 5298
This radio source resembles an FRI radio source (Fa-naro ff & Riley 1974) at 8.44 GHz, and has a tail at 1.49 GHz.The distribution of α values, flat in the centre steepening out-wards ( α > α , in particular the α − d α value is 1.1, but we willkeep the radio-AGN classification because there is also an indi-cation from the radio structure. This object is also classified as Article number, page 20 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs an AGN based on the mid-IR diagnostics, and is a Seyfert 2 orLINER in the optical.
Mrk 0331
This is the north companion of a galaxy system. At8.44 GHz we see a ring-structure. The α -map and histogramshow a flat centre steepening outwards with values of α gettingmuch higher than 1.1. We classify it as a radio-AGN. We notethat in the mid-IR this is a SB, and a Seyfert 2 in the optical. Article number, page 21 of 32 ig. A.1.
Each row corresponds to one object or system from our sample. (
Le f t ) Radio-spectral-index maps ( α -maps) are represented withthe colour-map; the colour bar on the right shows the radio spectral index values from low (flat α ) to high values (steep α ). Overlaid, the radiocontours at 1.49 GHz (black dashed) and at 8.44 GHz (white solid; original unsmoothed maps), where the contour levels scale as log of the fluxdensity from the lowest flux value (3 σ ) to the highest. On the bottom right we give a scale-bar of 2 arcsec, as visual aid. ( Middle ) The error mapfor the radio spectral index per pixel value. The colour bar on the right shows the range of values. Contours as on the
Le f t . In both plots, on thebottom left corner we give the beam of the 1.49 GHz map as black circle, and the beam of the 8.44 GHz map as grey circle / ellipse. On the bottomright we give a scale-bar of 2 arcsec, as visual aid. ( Right ) Histograms of the α -maps: ( T op ) Histograms of the radio spectral index, based onFig. A.1-
Le f t . (
Middle ) Histograms where the errors on α (Fig. A.1- Middle ) have been added. (
Bottom ) Histograms where the errors on α havebeen subtracted.Article number, page 22 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs Fig. A.1. (Continued) Article number, page 23 of 32 ig. A.1. (Continued)Article number, page 24 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs
Fig. A.1. (Continued) Article number, page 25 of 32 ig. A.1. (Continued)Article number, page 26 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs
Fig. A.1. (Continued) Article number, page 27 of 32 ig. A.1. (Continued)Article number, page 28 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs
Fig. A.1. (Continued) Article number, page 29 of 32 ig. A.1. (Continued)Article number, page 30 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs
Fig. A.1. (Continued) Article number, page 31 of 32 ig. A.1. (Continued)Article number, page 32 of 32ardoulaki et al.: Radio Continuum Properties of LIRGs