Local Luminous Infrared Galaxies. III. Co-evolution of Black Hole Growth and Star Formation Activity?

1.4 MB·English

by  Almudena Alonso-Herrero

Draftversion January18,2013
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
LOCAL LUMINOUS INFRARED GALAXIES. III. CO-EVOLUTION OF BLACK HOLE GROWTH AND STAR
FORMATION ACTIVITY?
Almudena Alonso-Herrero1,2, Miguel Pereira-Santaella3, George H. Rieke4, Aleksandar M. Diamond-Stanic5,
Yiping Wang6, Antonio Herna´n-Caballero1 and Dimitra Rigopoulou7
Draft version January 18, 2013
ABSTRACT
3
Local luminous infrared (IR) galaxies (LIRGs) have both high star formation rates (SFR) and a
1
highAGN (Seyfert and AGN/starburstcomposite)incidence. Therefore,they are idealcandidates to
0
explore the co-evolution of black hole (BH) growth and star formation (SF) activity, not necessarily
2
associated with major mergers. Here, we use Spitzer/IRS spectroscopy of a complete volume-limited

n sample of local LIRGs (distances of < 78Mpc). We estimate typical BH masses of 3 × 107M⊙
a using [Neiii]15.56 µmand optical [Oiii]λ5007 gas velocity dispersions and literature stellar velocity
J
dispersions. We find that in a large fraction of local LIRGs the current SFR is taking place not only

7 in the inner nuclear ∼ 1.5kpc region, as estimated from the nuclear 11.3µm PAH luminosities, but
1 also in the host galaxy. We next use the ratios between the SFRs and BH accretion rates (BHAR)
to study whether the SF activity and BH growth are contemporaneous in local LIRGs. On average,
] localLIRGshaveSFRtoBHARratioshigherthanthoseofopticallyselectedSeyfertsofsimilarAGN
O
luminosities. However, the majority of the IR-bright galaxies in the RSA Seyfert sample behave like
C local LIRGs. Moreover, the AGN incidence tends to be higher in local LIRGs with the lowest SFRs.
. AllthissuggeststhatinlocalLIRGsthereisadistinctIR-brightstarformingphasetakingplaceprior
h
to the bulk of the current BH growth (i.e., AGN phase). The latter is reflected first as a composite
p
and then as a Seyfert, and later as a non-LIRG optically identified Seyfert nucleus with moderate SF
-
o in its host galaxy.
r Subject headings: galaxies: nuclei — galaxies: Seyfert — infrared: galaxies
t
s
a
[ 1. INTRODUCTION trigger a high rate of star formation leading to a lumi-

nous infrared (IR)-active phase (Sanders et al. 1988),
1 One of the most fundamental relations in extragalac-
either as a luminous IR galaxy(LIRG) or ultraluminous
v tic astronomy is that, at least in the local universe, the
5 masses of supermassive black holes (BH, with masses IR galaxy (ULIRG). These are defined as having an IR
01 MBH > 106M⊙) correlate with the stellar mass and 8re−sp1e0c0t0ivµemlyl(usemeinSoasnidtyerLsI&R >Mi1r0a1b1eLl⊙19o9r6L,IfRor>a1r0ev12ieLw⊙).,
velocity dispersion of the bulges of their host galax-
4 As an alternative to this sequence, Kormendy et al.
ies (see e.g. Magorrianet al. 1998; Gebhardt et al.
. (2011) proposed that BHs in bulgeless galaxies and in
1 2000;Marconi & Hunt 2003;Ha¨ring & Rix 2004). This
galaxies with pseudobulges grow through phases of low-
0 seems to imply that bulges and supermassive BH evolve
level Seyfert-like activity. Such growth is believed to
3 together and regulate each other.
be driven stochastically by local processes (secular pro-
1 Mergers of gas-rich galaxies are efficient in both pro-
: ducing elevated star formation rates (SFR) and trans- cesses), and thus it would not have a global impact on
v thehostgalaxystructure(seealsoHopkins et al. 2008).
portinggastothenuclearregiontoallowforBHgrowth.
Xi Ifsufficientmatterbecomesavailableveryclosetothenu- Moreover, there is now evidence of two fundamentally
differentmodes ofBHgrowthatworkinearly-andlate-
r clear BH and is accreted, the nucleus of the galaxy will
a shine with enormous power as an active galactic nucleus type galaxies in the local universe (Schawinski et al.
2010). TheroleofmoderateluminosityAGNthatreside
(AGN, Lynden-Bell 1969). In parallel, and probably
inlate-typegalaxiesanddonotinvolveanyrecentmajor
before the fully developed AGN phase, the merger will
mergers is not clear and needs further investigation.
⋆This work is based on observations made with the Spitzer LIRGs are powered by both AGN and star
SpaceTelescope,whichisoperatedbytheJetPropulsionLabo- formation activity (Sanders & Mirabel 1996).
ratory,CaliforniaInstituteofTechnologyunderNASAcontract
Alonso-Herreroet al. (2012) showed that while an
1407
1Instituto de F´ısica de Cantabria, CSIC-Universidad de AGN accompanies star formation activity in a large
Cantabria,39005Santander, Spain proportion of local LIRGs, in most cases the AGNs are
2AugustoGonz´alezLinaresSeniorResearchFellow not energetically important. This is because the AGNs
3Istituto di Astrofisica e Planetologia Spaziali, INAF-IAPS,
hosted in local LIRGs have Seyfert-like luminosities
00133Rome,Italy
4Steward Observatory, University of Arizona, Tucson, AZ and thus their bolometric contribution to the total
85721, USA IR luminosity is small (typically 5%). Therefore, the
5Center for Astrophysics and Space Sciences, University of IR luminosities of LIRGs imply SFRs in the range
Ca6liNfoartniioan,aSlanAsDtireognoo,mLiacaJlolOlab,sCerAva9t2o0ri9e3s,,UCShAaoyang District 11−110M⊙yr−1 using the Kennicutt (1998) prescrip-
Beijing100012,China tionconvertedtoaKroupa (2012)InitialMassFunction
7AstrophysicsDepartment,UniversityofOxford,OxfordOX1 (IMF).
3RH,UK
2 Alonso-Herrero et al.
TABLE 1
Spitzer/IRS Calibrationsources
Name AOR
PNG043.1+37.7 4109056
PNG011.7-00.6 4109568
PNG025.8-17.9 4110080
PNG265.7+04.1 4111104
PNG206.4-40.5 4111616
PNG285.7-14.9 4112128
PNG054.1-12.1 4112640
PNG009.4-05.0 4114176
PNG342.1+10.8 4115200
PNG316.1+08.4 4116224
Calwav-22B-NGC7027 15343104
IRSS-SEPN-N7293-0001 15752960
It is reasonable to assume that the same gas that is Fig.1.—Distributionof theobserved (not corrected forinstru-
used to form stars in the host galaxy can also be used mental resolution) FWHMs of the [Neiii]15.56µm line of local
to feed the AGN provided that there is a mechanism LIRGscomparedwiththoseofthecalibrationsources. Thelatter
are assumed to provide the instrumental resolution at the wave-
able to transport the gas to the inner region (on scales
length of the line. The dot represents the average value of the
of less than 0.1pc) of the galaxy (see e.g. the review of FWHMofthecalibrationsourceswiththecorresponding1σerror.
Alexander & Hickox 2012). Indeed, it is now apparent We marked those LIRG nuclei classified as AGN (Seyfert and/or
thatthereisarelationbetweenstarformationactivityon [Nev] emitters) and composites. The rest are Hii-like or have
noclassification. Thethicklinehistogramrepresentstheobserved
sub-kiloparsecandbulgescalesandtheBHaccretionrate
FWHMofthosegalaxiesdeemedtohavespectroscopicallyresolved
(BHAR), as shown by observations and numerical sim- lines.
ulations (Heckman et al. 2004; Hopkins & Quataert
2010; Diamond-Stanic & Rieke 2012). This relation is
also predicted to be present, although with a lower detail by Alonso-Herrero et al. (2006, 2012). We drew
significance, for the integrated SFR of the galaxy thesamplefromtheIRASRevisedBrightGalaxySample
(Hopkins & Quataert 2010). Therefore, LIRGs show (RBGS, Sanders et al. 2003) to include all the sources
propitious conditions to study the co-evolution of star with log(LIR/L⊙) ≥ 11.05 and vhel = 2750 – 5300 km
formation activity and BH growth as they show both s−1. The sample is composed of 45 IRAS systems, with
highintegratedandnuclearSFRs(Alonso-Herrero et al. 8 containing multiple nuclei. Therefore the sample in-
2006) and a high occurrence of AGN and composite nu- cludes 53 individual galaxies. For the assumed cosmol-
clei (60−70%, Yuan et al. 2010; Alonso-Herrero et al. ogythe distancesareinthe range≃40−78Mpc,witha
2012). medianvalueof65Mpc. TheIRluminositiesoftheindi-
In this paper we study the co-evolution of the SFR vidual galaxies9 are in the range log(LIR/L⊙)=10.64−
activity and the BH growth in the complete volume- 11.67, with a median value of log(LIR/L⊙)= 11.12. All
limited sample of local LIRGs of Alonso-Herrero et al. the relevant information for the sample can be found in
(2006,2012). WeuseSpitzerInfraredSpectrograph(IRS, table 1 of Alonso-Herrero et al. (2012).
Houck et al. 2004) observations to estimate the nuclear Combininganumberofopticalandmid-infrared(mid-
(∼1.5kpc)SFRs andcomparethem withthe integrated IR) indicators, Alonso-Herreroet al. (2012) derived an
values from the IR luminosities. We look for spectro- AGN detection rate of ∼62% for this complete volume-
scopicallyresolved[Neiii]15.56 µm lines. Assumingthat limited sample of local LIRGs. The derived AGN bolo-
this line is produced in the narrow line region (NLR) of metric luminosities, from the mid-IR spectral decom-
the AGN, wemeasureits velocitydispersionanduse the position and/or X-ray observations, are in the range
Dasyra et al. (2011) relations to obtain the masses of L (AGN) = (0.4 − 50)× 1043ergs−1 with a median
bol
the BH hosted in local LIRGs. We complement these of ∼ 1.4 × 1043ergs−1 (Pereira-Santaellaet al. 2011;
with observations of the optical line [Oiii]λ5007 and lit- Alonso-Herreroet al. 2012). However, these AGN are
erature values of the stellar velocity dispersion σ∗ to ob- overall only responsible for ∼ 5% of the IR luminosity
tain further estimates of the BH masses in local LIRGs. emittedby localLIRGs, althoughthe AGN contribution
All this information allows us to determine the ratios varies from source to source (see Imanishi et al. 2010;
betweenthe SFR(both, nuclearandintegrated)andthe Alonso-Herreroet al. 2012). This is in good agreement
BHAR in local LIRGs and compare them with those of with Petric et al. (2011) for The Great Observatories
optically selected Seyfert galaxies. This comparison lets All-Sky LIRG Survey (GOALS; Armus et al. 2009).
us explore the role of the LIRG-phase in the growth of
BH in the local universe. Throughout this paper we as- 2.2. Spitzer/IRS Observations
sume the following cosmology H = 70 km s−1Mpc−1,
0 2.2.1. Observations
Ω =0.3 and Ω =0.7.
M Λ
We retrieved Spitzer/IRS (Houck et al. 2004) spec-
2. SAMPLE,OBSERVATIONS,ANDDATAREDUCTION troscopy of the sample of LIRGs taken with the high
2.1. The sample of local LIRGs
9 Galaxiessignificantlybelowtheselectionluminosityaremem-
Forthisworkweusethecompletevolume-limitedsam- bers of multiple systems with total luminosities that satisfy the
ple of local LIRGs that was presented and discussed in selectioncriterion.
Local Luminous Infrared Galaxies. III. Co-evolution of Black Hole Growth and Star Formation Activity? 3
resolution (R ∼ 600) short-high (SH) and long-high
TABLE 2
(LH) modules that cover the 9.9−19.6µm and 18.7− Fluxes andEWof the nuclear 11.3µm PAH feature
37.2µmspectralranges,respectively. Detailsonthepro-
grammeIDs,the observations,anddatareductionofthe Galaxy Flux EW
galaxies, including those from GOALS (Armus et al. NGC23 430.4±8.5 0.78±0.05
2009), are given in Alonso-Herrero et al. (2012) and MCG+12-02-001 396.8±7.9 0.42±0.01
Pereira-Santaellaet al. (2010a,b). Wenotethatforvar- NGC633 166.1±4.0 0.61±0.04
ESO297-G012 128.4±3.5 0.60±0.05
ious reasons two galaxies in this volume-limited sample
UGC01845 347.5±5.2 0.74±0.03
were not observed with the Spitzer/IRS high spectral
UGC02982 379±35 0.84±0.15
resolution modules. We extracted the nuclear spectra CGCG468-002NED01 59.6±5.6 0.24±0.03
assuming a point source calibration. CGCG468-002NED02 78.4±7.7 0.46±0.07
We also retrieved SH and LH spectroscopy from the UGC03351 275.5±8.3 0.78±0.06
Spitzer archive of 12 calibration sources and planetary NGC2369 357.1±3.9 0.58±0.02
nebulae to obtain an accurate estimate of the spectral NGC2388 428±14 0.53±0.04
resolution using fine structure lines (see Section 2.2.3). MCG+02-20-003 149.5±4.1 0.70±0.06
ThenamesandcorrespondingAstronomicalObservation NGC3110 299.0±4.0 0.80±0.03
Request(AOR)numbersaregiveinTable1. Wereduced NGC3256 1547.2±4.8 0.556±0.004
the data as for the sample of local LIRGs. ESO264-G057 165±13 0.76±0.11
IC694 446±16 0.38±0.02
2.2.2. Line Fitting NGC3690 308±25 0.09±0.01
ESO320-G030 450.3±4.6 0.78±0.03
For the fine-structure lines we used Gaussians to mea- MCG-02-33-098W 99.0±4.2 0.27±0.02
surethe lineflux, equivalentwidth(EW),andfull width MCG-02-33-098E 102.3±6.8 0.62±0.12
half maximum (FWHM) and a first order polynomial IC860 43.2±2.2 0.44±0.04
to fit the local continuum (see Pereira-Santaellaet al. MCG-03-34-064 23.0±3.2 0.013±0.002
2010a, for more details). The fluxes of the NGC5135 376.9±4.0 0.46±0.01
[Neii]12.81 µm, [Neiii]15.56 µm, and [Oiv]25.89 µm ESO173-G015 434±31 0.53±0.06
lines can be found in Pereira-Santaellaet al. (2010b) IC4280 249.4±7.0 0.80±0.04
andAlonso-Herrero et al. (2012). Forthe11.3µmpoly- UGC08739 146±41 0.84±0.38
cyclic aromatic hydrocarbon (PAH) feature first we fit- ESO221-IG010 259.9±9.4 0.63±0.05
NGC5653 385±10 0.81±0.07
ted the local continuum using a linear fit between 10.6
NGC5734 402±11 0.80±0.06
and 11.8µm. After subtracting the continuum, we in-
NGC5743 251±14 0.75±0.12
tegrated the flux in the wavelength range between 10.8 IC4518E 64.6±1.4 0.62±0.05
and 11.6µm. We list the fluxes, EW, and corresponding IC4518W 62.6±5.9 0.16±0.02
errors of the 11.3µm PAH feature in Table 2. Zw049.057 62.3±4.5 0.96±0.23
NGC5936 240.0±6.6 0.67±0.05
2.2.3. Spectrally resolved [Neiii]15.56µm lines NGC5990 243±56 0.29±0.09
The velocity dispersion of the ionized gas in the NLR NGC6156 141±16 0.22±0.03
allowsmeasuringtheBHmass(M )inanAGNsupple- IRAS17138-1017 307.6±4.3 0.53±0.02
mentingreverberationmappingteBchHniques. Asdiscussed IRAS17578-0400 177±10 0.76±0.09
IC4687 478.2±4.2 0.70±0.02
byGreene & Ho (2005),theNLRissufficientlycompact
IC4734 195.1±3.8 0.59±0.03
to be illuminated by the AGN, while large enough to
NGC6701 242.2±4.2 0.67±0.03
feel the gravitational potential of the bulge. Therefore, MCG+04-48-002 394±33 0.73±0.11
thereisarelativelygoodcorrelationbetweenthegasand NGC7130 230.5±3.7 0.36±0.01
the stellar velocity dispersion around local AGN. In the IC5179 391.9±3.6 0.68±0.02
mid-IR Dasyra et al. (2008, 2011) showed that the ve- NGC7469 583.5±7.7 0.237±0.005
locitydispersionofthefinestructurelines[Siv]10.51 µm, NGC7591 142±12 0.36±0.06
[Neiii]15.56 µm, [Nev]14.32 µm, and [Oiv]25.89 µm of NGC7679 363±16 0.61±0.05
AGN are well correlated with M . NGC7769 85±13 0.57±0.21
BH
For our Spitzer/IRS spectra we focus on the NGC7770 131±18 0.47±0.13
[Neiii]15.56 µm line to look for spectrally resolved line NGC7771 359±16 0.65±0.07
profiles. In AGN the flux and luminosity of this fine- Notes. Thefluxesareinunitsof10−14ergcm−2s−1 andthe
EWareinunitsofµm.
structurelinearefoundtocorrelatewellwiththoseofthe
mid-IR [Nev] lines at 14.3 and 24.3µm (Gorjian et al. mid-IR spectrum for accurate measurements.
2007; Pereira-Santaellaet al. 2010b). The high ioniza- Using the calibration sources (Section 2.2.1) we ob-
tion potential of the [Nev] lines indicates that they are tained twelve independent measurements of the unre-
mostly excited by AGN and thus, the [Neiii]15.56 µm solvedwidth ofthe [Neiii]15.56 µm line and inferredan
emission is likely to be as well. When compared with instrumentalwidthofFWHM =0.02618±0.0007µm.
inst
othermid-IRfinestructurelineswithrelativelyhighion- This is equivalent to a spectral resolution of R =
ization potentials (e.g., [Siv]10.51 µm, [Oiv]25.89 µm, λ/FWHM ∼ 595 ± 16 and an instrumental veloc-
inst
and the [Nev] lines), the [Neiii]15.56 µm line is a good ity dispersion of σ = 215 ± 6kms−1 at the wave-
inst
compromise between ionization potential, critical den-
sity10, intensity of the line, and a clean region of the ionizationpotential thantheoptical[Oiii]λ5007emissionlinebut
theopticallinehasaslightlyhighercriticaldensity(seefigure6of
10 For reference the [Neiii]15.56µm line has a slightly higher Dasyraetal. 2011).
4 Alonso-Herrero et al.
Fig.2.—Observedprofiles(solidlinesandfilleddots)ofthespectrallyresolved[Neiii]15.56µm linesoflocalLIRGs thatareclassified
as a Seyfert and/or are [Nev] emitters. The shaded area shows a Spitzer/IRS SH unresolved profile represented as a Gaussian with
FWHM=0.02618µm,asdeterminedfromSpitzer/IRScalibrationsources(Section2.2.1).
Fig.3.—Toppanel: Observedprofiles(solidlinesandfilleddots)ofthe[Oiii]λ5007linefittedwithtwoGaussiancomponents(dashed
lines) as explained in Section 2.3. Bottom panel: Same as top panel but for galaxies whose profiles are fitted with one
Gaussian. The plotted errors for the 6dF data (NGC 2369, MGC −03-34-064, NGC 5135, and NGC 7130) are those
computed from the rms of the continuum adjacent to the line.
length of this line. While the derived spectral resolution (σ −σ )>3×pǫ2 +ǫ2 , whereσ andǫ are
obs inst m inst inst inst
is very similar to that from Dasyra et al. (2011), the the instrumental resolution and its standard deviation,
standard deviation of our measurements is much lower. and σ and ǫ are the measured values for the LIRGs.
obs m
Dasyra et al. (2011) obtained σ as the average of Thetotalerroroftheobservationsǫ includesboththe
inst obs
the values of the high excitation fine structure lines of error in the measurement ǫ and that associated with
m
a sample of AGN that were deemed not to be resolved the instrumental value, that is, ǫ =(ǫ2 +ǫ2 )1/2.
obs m inst
by Spitzer/IRS with their method. By doing so, they Using the criterion described above we found that
probably also included measurements of barely resolved 19 LIRGs in our sample show resolved [Neiii]15.56 µm
lines. This wouldtherefore explainthe higher dispersion lines. Most are LIRGs classified as Seyfert (Figure 1)
of their instrumental value. and/orare[Nev]emitters. ThosenucleiclassifiedasHii,
For the LIRGs we need a criterion to assess whether on the other hand, tend to show FWHM close to the in-
the lines are resolved or not since a significant num- strumental resolution. Composite nuclei have FWHM
ber of sourceshave measuredwidths slightly largerthan between those of unresolved lines and the clearly re-
the instrumental resolution. Figure 1 shows the distri- solved[Neiii]15.56 µm lines. ForthoseLIRGsdeemedto
bution of the measured FWHM of the [Neiii]15.56 µm haveunresolved[Neiii]15.56 µm lines, the averagemea-
line for the sample ofLIRGs comparedwith those ofthe sured FWMH of the [Neiii]15.56 µm line is 0.02776±
calibration sources. To avoid analyzing barely resolved 0.0017µm. That is, the observed values for these lines
linesweadoptthefollowingcriterionforthevelocitydis- are on average 2σ above the instrumental resolution.
persion to determine whether a line is clearly resolved: Table 3 lists the intrinsic values of the velocity
Local Luminous Infrared Galaxies. III. Co-evolution of Black Hole Growth and Star Formation Activity? 5
TABLE 3
Velocity dispersions andBHmasses
Galaxy Class σ([NeIII]) σ([OIII])n σ([OIII])b σ∗ Ref logMBH
km s−1 km s−1 km s−1 km s−1 M⊙
Seyfert nucleiand [NeV]emitters
CGCG 468-002-NED01 Sy2∗ 135±12 218±22 ··· ··· ··· 8.06
NGC3690 Sy2 167±10 ··· ··· 144±11 1 7.52
MCG −03-34-064 Sy1 368±18 249±10 656±26 155 2 7.65
NGC5135 Sy2 138±9 127±9 305±21 124±6 3 7.24
IC4518W Sy2 156±12 ··· ··· ··· ··· 7.48
NGC5990 Sy2 264±11 ··· ··· ··· ··· <8.45
NGC6156 [Nev] 160±13 ··· ··· ··· ··· 7.52
MCG +04-48-002 [Nev] 158±7 ··· ··· ··· ··· 7.50
NGC7130 Sy2: 259±16 130±9 463±32 147±5 3 7.55
NGC7469 Sy1 182±13 145±7 336±17 152±16 4 7.61
NGC7591 Sy2: 169±17 ··· ··· ··· ··· 7.62
NGC7679 Sy1/Sy2 102±7 176∗ ··· 96 2 6.77
Composite and Hii nuclei
NGC1614 Composite 157±10 ··· ··· 164±8 5 7.75
NGC2369 Composite 133±8 146±15 ··· ··· ··· ···
IC694 LINER <70 ··· ··· 141±17 5 7.48
NGC3256 Hii <112 ··· ··· 100±6 3 6.84
MCG −02-33-098W Composite 103±9 ··· ··· ··· ··· ···
ESO 173−G015 ··· 102±7 ··· ··· ··· ··· ···
IC4280 Hii 128±8 ··· ··· ··· ··· ···
IRAS17138−1017N Composite <80 ··· ··· 72±5 6 6.24
IRAS17578−0400 ··· 119±8 ··· ··· ··· ··· ···
NGC6701 Composite <159 ··· ··· 150±20 7 7.59
NGC7771 Composite 178±17 ··· ··· ··· ··· ···
Notes.— The references for thespectral class of the nucleiare listed in Alonso-Herrero et al. (2012) except for
CGCG 468-002-NED01, which is from thenewly analyzed optical spectrum (see Section 2.3).
∗The [Oiii]velocity dispersion for NGC7679 is also from reference 2.
References for σ∗: 1. Ho et al. (2009). 2. Gu et al. (2006). 3. Garcia-Rissmann et al. (2005). 4. Onken et al. (2004). 5.
Hinz& Rieke (2006). 6. Shier& Fisher (1998). 7. M´arquez et al. (1996).
dispersion for the LIRGs with resolved [Neiii]15.56 µm [Oiii]emissionlinesisσ ∼125kms−1. Long-slitspec-
inst
lines as well as their spectroscopic classes from troscopy was available for CGCG 468-002-NED01 and
optical spectroscopy and/or mid-IR indicators NGC 7469 taken with FAST on the Mt. Hopkins Till-
(Alonso-Herrero et al. 2012). The velocity disper- inghast 60-inch telescope as part of different observing
sions(correctedforinstrumentalresolution)arebetween programs. Theslitwidthwas3′′ andthespectralresolu-
102 and 368kms−1. For galaxies in our sample with tion was σinst ∼75kms−1. None of the spectra are flux
literature values of the stellar velocity dispersions whose calibrated,but they can be used to measure the velocity
[Neiii]15.56 µm lines are deemed to be unresolved, we dispersion of the emission lines as well as their ratios.
provide in this table upper limits to the gas velocity WefirstmeasuredtheopticallineratiosofCGCG468-
dispersion. In Figure 2 we show the normalized profiles 002-NED01 since we did not have an optical clas-
of the twelve Sy/[Nev] emitter LIRGs with spectrally sification for this galaxy. We classify this galaxy
resolved [Neiii]15.56 µm lines compared with those of as a Seyfert 2. In Pereira-Santaellaet al. (2010b)
anunresolvedprofilerepresentedasaGaussianfunction. and Alonso-Herrero et al. (2012) we detected the mid-
IR [Nev] lines in this galaxy and measured a high
2.3. Optical [Oiii]λ5007 measurements [Oiv]25.89 µm/[Neii]12.81 µm lineratio. Weusedboth
as evidence of the presence of an AGN in this nucleus.
We obtained archival optical spectra of six LIRGs
The optical classification confirms this.
classified as AGN or Composite from the six-degree
For all the galaxies we modeled the [Oiii] emission
Field (6dF) Galaxy Survey (6dfGS, Jones et al. 2004,
lines at 5007˚A and 4959˚A by fitting a Gaussian profile
2009)andfromobservationswiththeFASTspectrograph
to each line. The relative position of the Gaussians was
(Tokarz & Roll 1997; Fabricant et al. 1998) from the
fixed according to the rest-frame wavelengths of the two
Astrophysics (CfA) telescope data center.
NGC 2369, MCG −03-34-06, NGC 5135, and lines. Likewise, the relative intensity was also fixed to
the value determined by the atomic parameters. We
NGC7130wereobservedwiththe6dFmulti-objectfibre
(angular diameter of 6′.′7) spectrograph on the United found that the fit with one component was only satis-
factory for CGCG 468-002-NED01 and NGC 2369. For
Kingdom Schmidt Telescope (UKST). At the typical
the rest we had to add a broad component to the [Oiii]
distance of these LIRGs the spectra cover the central
∼2kpc. The instrument spectral resolution around the lines. Thewavelengthsandintensitiesofthenarrow(also
6 Alonso-Herrero et al.
called core) component and the broad component were
considered independent. To estimate the errors of the
velocity dispersions we used the rms of the continuum log(MBH/M⊙)=−1.78+4.24 log(σOIII/kms−1) (2)
adjacenttothe line forthe 6dFspectraandthe errorsof
For both relations Dasyra et al. (2011) fixed the
the spectra for the FAST data. The typical errorsof the
slope to the value derived from the correlation between
velocity dispersions of the [Oiii] lines are less than 10%,
the stellar velocity dispersion and the BH mass by
and are listed for each galaxy in Table 3.
Gu¨ltekin et al. (2009):
The top panel of Figure 3 shows the galaxies for
whichthe [Oiii]λ5007line fits requiredtwocomponents,
whereasthebottompanelshowstheprofilesofNGC2369 log(MBH/M⊙)=8.12+4.24 log(σ∗/200kms−1) (3)
and CGCG 468-002-NED01. Table 3 lists the velocity
dispersion of the core of the line and the broad com- The typical uncertainties on the derived MBH can be as
ponent, if present, corrected for instrumental resolution. high as 0.8dex if using the gas σ based on the rms of
The velocity dispersions of the core components mea- the relations (Dasyra et al. 2011) and 0.4dex if using
sured in the four galaxies with double components are σ∗ (Gu¨ltekin et al. 2009).
in good agreement with literature σ∗ values, except for Before we compute the BH masses, we compare
MCG −03-34-064. the gas velocity dispersions for the LIRGs with both
ForthefourLIRGswithdoublecomponentsthebroad [Neiii]15.56 µm and [Oiii]λ5007 measurements. As we
component is blueshifted typically by 50−300kms−1. saw in Section 2.3, in some cases the optical line is fit-
The presence of such asymmetric or broad blue wings ted with two components. For AGN there is generally a
in the [Oiii]λ5007 lines is common in AGN (see e.g., good agreement between σ∗ and the gas velocity disper-
Greene & Ho 2005). The total [Oiii] emission is dom- sionfromthe narrowcomponentof[Oiii](Onken et al.
inated by the broad component in MCG-03-34-064 and 2004; Greene & Ho 2005). Therefore, the core of the
in NGC 7130, whereas in NGC 5135 and NGC 7469 the [Oiii]λ5007 line probes the gravitational movements in
core component dominates the emission. In NGC 5135 theNLRandthus canbe usedtoestimateMBH. Onthe
Bedregal et al. (2009) based on both the relatively otherhand,thebroadcomponentstendtobeblueshifted,
broad [Sivi]1.96µm coronal line and its spatial extent maybe moreinfluenced by the AGN, andare likely pro-
suggested the existence of different kinematical compo- duced by outflows. As can be seen from Table 3, the ve-
nents and the possible presence of AGN induced out- locity dispersions from the [Neiii]15.56 µm line appear
flows. Gonz´alez Delgado et al. (1998)foundblueshifted tobeintermediatebetweenthoseofthetwoopticalcom-
components in the Lyα and [Oiii]λ5007 lines of both ponents. Spectrallyresolvedblueshifted[Neiii]15.56 µm
NGC7130andNGC5135. Thevelocitydispersionofthe lines (as wellas the [Nev]lines) havealso been reported
broadcomponentofthe [Oiii]λ5007line ofNGC 7130is for local ULIRGs (Spoon & Holt 2009), and have been
similar to the Hα broad component (σ ∼ 400kms−1, interpreted as the result of outflows. Thus, it is possible
thatinsomecaseswemightoverestimatetheBHmasses
Bellocchi et al. 2012). In NGC 7469 Wilson et al.
when using the [Neiii]15.56 µm line if there are several
(1986) also detected two components in the [Oiii]λ5007
components. The most suspect galaxy in our sample
line.
would be NGC 5990, for which σ predicts the most
NeIII
massiveBHbyfarinoursample. Wenote,however,that
the calibrationof Dasyra et al. (2011)should take these
3. RESULTS effects into consideration.
3.1. Black Hole Masses In Table 3 we list MBH as calculated from, in this
order of preference, the stellar velocity dispersion, the
InTable3welist,inadditiontothegasvelocitydisper-
velocity dispersion of the core of the [Oiii] line, and
sions from the [Neiii]15.56 µm and optical [Oiii]λ5007
the [Neiii]15.56 µm line velocity dispersion. For those
lines,thestellarvelocitydispersionσ∗valuesfromthelit- LIRGs with no definitive evidence of AGN activity
erature,whenavailable. Wealsoincludedinthistableσ∗ (i.e., classified as composites, Hii or unknown), we
values fromthe literaturefor other LIRGs in our sample
only list M in the second part of the table if we
(IC 694,NGC 3256,IRAS 17138−1017,and NGC 6701) BH
have a value of σ∗. This is because we cannot be
whose[Neiii]15.56 µm linesappearunresolvedattheSH
surewhetherthe[Neiii]15.56 µm emissioncomesmostly
spectral resolution. We note, however, that in the case
from the NLR or whether it is produced by star forma-
of on-goingmajor mergers the stellar velocity dispersion
tion (Pereira-Santaellaet al. 2010b).
vs. BH mass relation might not be applicable.
ForthoseLIRGshostinganactivelyaccretingBH(that
Dasyra et al. (2008, 2011) showed that mid-IR re-
is,withclearsignsofAGNactivity,firstpartofTable3)
solvedfine-structure lines can be used to study the NLR
of AGN. In particular they demonstrated that the mass we find typical BH masses of 3×107M⊙, ranging from
of the BH correlates with both the velocity dispersion 6×106M⊙ to 3×108M⊙. The estimated BH masses
for those LIRGs classified as composite or Hii are also
and the luminosity of the NLR region for a sample of
inthisrange. Themassesofthe LIRGBHsappeartobe
optically selected AGN. The relation is:
similar to those of the currently growing BH in the lo-
caluniversehostedinlate-typegalaxies(Heckman et al.
log(MBH/M⊙)=−1.82+4.24 log(σNeIII/kms−1) (1) 2004; Schawinski et al. 2010).
The typical Eddington ratios (L (AGN)/L ) for
bol Edd
Dasyra et al. (2011) fitted a similar relation for the the LIRGs with an estimate of M are 2×10−2, and
BH
optical [Oiii]λ5007 line: range between 5×10−4 and 0.1. This means that the
Local Luminous Infrared Galaxies. III. Co-evolution of Black Hole Growth and Star Formation Activity? 7
BH in local LIRGs are accreting at a lower efficiency tends typically 1.5kpc. We use the following recipes
than those in local ULIRG (typical values of 0.08−0.4 putforwardbyDiamond-Stanic & Rieke (2012)tocom-
depending on the method used for estimating the BH pute the nuclear SFRs based onthe [Neii]12.81 µm line
mass, Dasyra et al. 2006; Veilleux et al. 2009). 11.3µm PAH feature luminosities:
ThemajorityofLIRGswithspectroscopicallyresolved
[Neiii]15.56 µm lines are classified as Seyfert galaxies.
However,in local LIRGs there is a high fraction of com- SFR(M⊙ yr−1)=9.6×10−9 L(11.3µm,L⊙) (5)
posite objects (AGN/starburst). Moreover,this fraction
remainsapproximatelyconstantasafunctionofIRlumi-
nosity from galaxies with LIR ∼ 1010L⊙ up to ULIRGs SFR(M⊙ yr−1)=8.9×10−8 L([NeII],L⊙). (6)
(Yuan et al. 2010). As canbe seenfrom Figure 1, com-
positenucleiinlocalLIRGshavevaluesoftheFWHMof Diamond-Stanic & Rieke (2012) calibrated Equa-
[Neiii]15.56 µm on average lower than those of Seyfert tions 5 and 6 for galaxies with LIR < 1011L⊙ using
nuclei but higher than those of nuclei classified as Hii the Rieke et al. (2009) templates and a Kroupa IMF.
and the instrumental resolution. Assuming that these We chose to use these calibrations rather than those for
composite nuclei do indeed host an AGN and that the higher IR luminosities because the median value of the
[Neiii]15.56 µm line is probing their NLR, then their IR luminosity of the individual galaxies in our sample is
BH masses oughtto be less than ∼2−3×107M⊙. The 1.3×1011L⊙. In addition, the width of the IRS SH slit
estimated typicalAGN bolometric luminosities from the only probes the nuclear regions of our sample of LIRGs
mid-IRspectraldecompositionare1043ergs−1 orless,so and thus lower IR luminosities.
we cannot constrain their Eddington ratios. We also note that Equation 5 assumed that the flux
TheBHoflocalLIRGsareonlymarginallylessmassive of the 11.3 µm PAH feature was measured using pah-
thanthoseoflocalULIRGs. TheBHmassesofthelatter fit (Smith et al. 2007). The spectral coverage (∼
areintherange∼107−5×108M⊙ (Tacconi et al. 2002; 10−18µm) and spectral resolution of the SH data used
Dasyra et al. 2006;Veilleux et al. 2009). Samplesoflo- here are not adequate to use pahfit. We measured
calULIRGsaremostlydominatedbygas-richinteracting the PAH fluxes using a local continuum (Section 2 and
galaxies and major mergers (Sanders & Mirabel 1996), Table 2). Following Smith et al. (2007) we applied a
withcoalescedULIRGshavingslightlylargerBHmasses multiplicative factor of two to the PAH fluxes measured
thanpre-coalescenceULIRGs(Dasyra et al. 2006). Our with a local continuum to make a proper comparison
sample is both flux and volume-limited andthus is com- with pahfit fluxes. We also corrected the 11.3µm PAH
posed mostly of LIRGs with LIR ∼1−2×1011L⊙ (in- fluxes for extinction using the nuclear strength of the
dividual galaxies), which also tend to be spiral galaxies, 10µmsilicatefeature(fromAlonso-Herrero et al. 2012)
minormergers,andgalaxiesingroups. Thatis,morpho- and the extinction law given in Smith et al. (2007).
logically local LIRGs are not dominated in numbers by For most of our local LIRGs this correction is small,
majormergers(seee.g.Sanders & Ishida 2004;Kaviraj as the silicate absorptions are moderate (see e.g. fig-
2009; Pereira-Santaella 2012) and their BH masses are ure 6 in Alonso-Herreroet al. 2012) as compared with
similartothoseofpre-coalescentULIRGs(Dasyra et al. the more deeply embedded population of local ULIRGs
2006). (Spoon et al. 2007).
While the [Neii]12.81 µm emission in low luminos-
3.2. Star Formation Rates ity and moderate luminosity AGN is mostly produced
by star formation, in high luminosity AGN this emis-
We computed the SFRs for our sample of LIRGs
sion line can also have an important contribution from
using a number of IR-based indicators, including
the AGN (Pereira-Santaellaet al. 2010b). To cor-
the fine-structure [Neii]12.81 µm line (Roche et al.
rect the [Neii]12.81 µm line flux for any possible AGN
1991; Ho & Keto 2007), the 11.3µm PAH feature
contribution, we followed the method put forward by
(Brandl et al. 2006; Diamond-Stanic & Rieke 2012),
Veilleux et al. (2009). This method reproduces the ob-
and the total 8 − 1000µm IR luminosity (Kennicutt
served values of the [Oiv]25.89 µm/[Neii]12.81 µm line
1998). Although it is not clear whether the presence
ratio (given in Alonso-Herrero et al. 2012) as a frac-
of an AGN could destroy the PAH carriers or not, at
tional combination of the typical line ratio of an AGN
the typical luminosities of the AGN hosted in local
(∼ 4) and that of a starburst galaxy (∼ 0.01), where
LIRGs the 11.3µm PAHs do not seem to be affected
the AGN contribution is the free parameter. Using this
(Diamond-Stanic & Rieke 2010).
methodwefindthattheAGNcontributiontothenuclear
For the integratedSFRs we usedL ofthe individual
IR [Neii]12.81 µm emission of local LIRGs is small. These
galaxies as given in Alonso-Herrero et al. (2012, their
correctionsareonlynecessaryfor10LIRGsandtheAGN
table 1) after subtracting the AGN contribution. We
contributiontothe nuclear[Neii]12.81 µm fluxesranges
used the Kennicutt (1998) relation in terms of the IR
from 3% to 53% (average ∼18%).
luminosity converted to a Kroupa IMF:
Figure 4 (left panel) compares the nuclear SFRs as
SFR(M⊙ yr−1)=1.14×10−10 (LIR,L⊙). (4) tdheerivAeGdNfrocomnttrhibeut[iNone)ii]a1n2d.8f1roµmmthliene11(.a3fµtemr PreAmHovfienag-
WeobtainedtotalSFRsfortheindividualgalaxiesinour ture. We have color coded symbols by the AGN bolo-
sample of between 1 and 60M⊙yr−1. metric contribution to IR luminosity of the galaxies (see
For the nuclear SFR we used the Spitzer/IRS spec- Alonso-Herreroet al. 2012). As can be seen from this
troscopy. Atthe mediandistance of65Mpc foroursam- figure, the majority of LIRGs with small or no (<25%)
ple of LIRGs, the IRS SH slit width (4.7arcsec) sub- AGN contribution show a good agreement between the
8 Alonso-Herrero et al.
Fig. 4.—Leftpanel: comparisonbetweenthenuclearSFRsfromthe[Neii]12.81µm line(correctedforAGNemission)andthe11.3µm
PAHfeatureforthesampleoflocalLIRGs. ThecolorsdenotetheAGNbolometriccontributiontoLIRoftheindividualgalaxies: <1%and
noAGNcontribution(blue),1−7%(yellow),7−25%(orange),and>25%(red). ThehighlydiscrepantpointwithnoAGNcontribution
totheleftofthestraightlineisIC860. Thestraightsolidlineisa1:1lineandistheaveragevalueoftheratiobetweenthenuclearSFRs
computed with the two estimators (see the text). Right panel: comparison between the nuclear SFR from the 11.3µm PAH feature and
the integrated SFR from the IR luminosity after subtracting the AGN contribution as estimated by Alonso-Herreroetal. (2012). The
straight solid line shows an average ratio between the nuclear and the total SFRs of 0.5 (see text), typical of local LIRGs, whereas the
dashedlinerepresentsa1:1relation.
derived nuclear SFRs using the two indicators. Indeed, and the total SFR from L would also be on average
IR
we find that the ratio between the nuclear SFR esti- 1.3. This gives us confidence that we used the appropri-
mated from the 11.3µm PAH feature and that from the ate SFR calibrations for the nuclear rates.
[Neii]12.81 µm line is on average 1.0±0.3.
In a few cases with a significant AGN bolomet- 3.3. Relation between BH accretion rates and SFR in
LIRGs
ric contribution the nuclear SFR derived from the
[Neii]12.81 µm lineishigherthanthatfromthe11.3µm The presence of an AGN is an unambiguous signpost
PAHfeature,evenaftercorrectingfortheAGNcontribu- ofaperiodofBHgrowth. TheAGNluminositycanthen
tion. GiventhegoodagreementbetweenthenuclearSFR be expressed in terms of the BHAR m˙ and the mass-
BH
computedwiththetwomethodsforLIRGswithsmallor energy conversion efficiency ǫ (see Alexander & Hickox
no AGN contribution and to avoid AGN contamination 2012, and references therein):
issues from now on for the nuclear SFR we use those
derived from the 11.3µm PAH feature. Thus, the indi-
vidual nuclei of the sample of local LIRGs show values m˙ BH(M⊙yr−1)=0.15(0.1/ǫ)(Lbol/1045ergs−1) (7)
of the nuclear SFR of between ∼0.8 and ∼20M⊙yr−1. We used this relation with the typical value of ǫ =
In Figure 4 (right) we compare the total SFR against 0.1 in the local universe (Marconi et al. 2004) and
the nuclear (typically on scales of 1.5kpc) SFR for the L (AGN)fromthe mid-IRspectraldecomposition(see
bol
individual galaxies of the LIRG sample. It is clear Alonso-Herreroet al. 2012). We obtained BHAR be-
that a significant fraction of the individual galaxies in tween 0.0007 and 0.08M⊙yr−1 for local LIRGs. The
our sample have more than half of their total SFRs uncertainties of the estimates of the BHAR are domi-
taking place outside the nuclear (1 − 2kpc) regions. nated by the uncertainties of the AGN bolometric lu-
For the individual galaxies of the sample the average minosities which are typically less than 0.4dex (see
value of the nuclear SFR over the total SFR ratio is Alonso-Herreroet al. 2012, for details).
0.5 ± 0.3. This agrees with findings using other indi- Having calculated the BHAR we can now compare
cators(Hattori et al. 2004;Alonso-Herrero et al. 2006; them with the nuclear and total SFR for the sample of
D´ıaz-Santos et al. 2010;Rodr´ıguez-Zaur´ınet al. 2011). LIRGs. The distribution of SFR/m˙ can provide clues
BH
Wealsonote,however,thatinapproximately20%ofour asto whether AGNandon-goingstarformationactivity
volume-limitedsampleofLIRGsthenuclearemissionac- are contemporaneous in local LIRGs. Recent numerical
counts for most of the total SFR measured in the indi- simulations by Hopkins (2012) predict a time offset be-
vidualgalaxies,as is the case for a largefractionof local tween the peaks of these activities which is thought to
ULIRGs. depend on the physical scale within the galaxy and the
Incidentally,thesamecomparisonforthenuclearSFRs dynamical time of the galaxy, among other parameters.
(Figure 4, left) but using the SFR equations calibrated Figure 5 shows the distribution of SFR/m˙ for the
BH
for LIR > 1011L⊙ (see Diamond-Stanic & Rieke 2012), sampleoflocalLIRGsforthenuclear(∼1.5kpc)andthe
results in nuclear SFRs from the 11.3µm PAH feature integratedSFRs, only for those LIRGs with an estimate
which are on average a factor of 1.3 higher than those of L (AGN). The nuclear and integrated SFR/m˙
bol BH
from the [Neii]12.81 µm line. Moreover, the ratio be- areintherange∼12to∼2×104,withmedianvaluesof
tween the nuclear SFR from the 11.3µm PAH feature thenuclearandintegratedlog(SFR/m˙ )of3.1and3.4,
BH
Local Luminous Infrared Galaxies. III. Co-evolution of Black Hole Growth and Star Formation Activity? 9
Fig.5.—DistributionsoftheSFR/m˙BH ratiosfornuclear(typi-
calscalesof1.5kpc)SFRs(lowerpanel)andintegratedSFRs(up- Fig.6.—DistributionsoftheratiobetweenthenuclearSFRand
per panel) for the sample of local LIRGs with an estimate of the theBHAR(bottompanel)andtheintegratedSFRandtheBHAR
AGNluminosity. ThenuclearSFRarefromthe11.3µmPAHfea- (upper panel) for the RSA sample of AGN (blue). In orange we
tureandtheintegratedSFRfromLIR aftersubtractingtheAGN markthose RSA AGN that are alsoIR-bright galaxies (see text).
component (seethetextfordetails). WeonlyincludedRSAgalaxieswithm˙BH >10−4M⊙yr−1.
respectively. Mostofthe nucleiclassifiedasSeyfertfrom
Inthis sectionwecomparethe propertiesofthe AGNs
optical spectroscopyshow nuclear log(SFR/m˙ )<3.0.
BH identified in local LIRGs with the optically selected
The nuclei classified as composite tend to show larger
Seyfert galaxies in the revised-Shapley-Ames catalog
values of the nuclear SFR to BHAR ratio,as expected if
(RSA, Sandage & Tammann 1987; Maiolino & Rieke
they were powered by both SB and AGN activity.
1995). Thisopticalsamplewasspectroscopicallyselected
For the entire sample of local LIRGs we obtain
from the original galaxy-magnitude limited sample of
PSFR/Pm˙ BH ≥ 2900. This is a lower limit because RSA galaxies and is believed not to be biased against
for half of the sample we did not detect an AGN com-
low-luminosity Seyfert galaxies. Also, as for the sample
ponent and therefore we are only using an upper limit
of local LIRGs studied in this work, the Seyfert activity
for their BHAR. If we repeat this only for those LIRGs
inthe RSAsampledoes notappearto be drivenprimar-
with an estimate of their AGN bolometric luminosities
ily by mergers for the majority of the galaxies. Finally,
we obtain a time-averaged value of SFR/m˙ ≃ 2500,
BH theAGNbolometricluminositiesoftheRSASeyfertsare
if we assume that all LIRGs go through an AGN phase,
similar to those of the local LIRGs.
that is, a ∼ 50% duty cycle. The two values are con-
sistent with each other taking into account the typi- The median values of SFR/m˙ BH for our LIRGs are
cal uncertainties in calculating L (AGN) (∼ 0.4dex, on average higher than those of local Seyfert galax-
bol
ies, especially the nuclear values (see Figure 5 and
Alonso-Herrero et al. 2012). This ratio is a few times
Diamond-Stanic & Rieke 2012). Of course, this is not
higher than that measured for the local population of
bulge-dominatedgalaxies(∼103, Heckman et al. 2004) completely unexpected as local LIRGs, including those
hostinganAGN,areessentiallyselectedbytheirstarfor-
andthe localnormalizationofthe BHmassversusbulge
mationactivity,whereastheRSASeyfertsareselectedby
mass relation (Marconi & Hunt 2003; Ha¨ring & Rix
their host galaxy magnitudes. The latter is more likely
2004). Eventhoughthe SFRs ofLIRGs aremuchhigher
probing the stellar mass, rather than the star formation
than those of local bulge galaxies, the time-averaged
activity.
SFR/m˙ valueoflocalLIRGsissimilartothatoflocal
BH
Wecanmakeamoremeaningfulcomparisonifweiden-
bulge galaxies during the first 0.3Gyr of their on-going
tify the IR-bright Seyferts in the RSA sample. We set
starburst (Wild et al. 2010, their figure 9). This sug-
gests the existence of similar time delays between the thelimit tolog(LIR/L⊙)=10.8tomatchapproximately
thelowerlimitcoveredbytheIRluminositiesoftheindi-
peakofstarformationandBHgrowthinlocalLIRGs as
vidualgalaxiesof our sample of localLIRGs (see table 1
well.
in Alonso-Herrero et al. 2012). We also included from
4. DISCUSSION our sample the newly identified Seyfert nuclei with an
estimate of L (AGN). The values of SFR/m˙ for the
4.1. Comparison between the AGNs in LIRGs and the bol BH
RSA Seyferts are taken from Diamond-Stanic & Rieke
optically identified RSA Seyferts
(2012) except for those already included in our sam-
10 Alonso-Herrero et al.
ple of LIRGs. Diamond-Stanic & Rieke (2012) com-
puted the nuclear SFRs from the 11.3µm PAH fea-
ture and the MIPS 24µm photometry, and L (AGN)
bol
from the [Oiv]25.89 µm line. Finally, for the few RSA
Seyferts deemed to have their [Oiv]25.89 µm luminosi-
ties strongly contaminated by star formation, we esti-
mated L (AGN) (and thus BHAR) from their hard
bol
X-ray luminosities and using a bolometric correction.
In this comparison we only included RSA Seyferts with
m˙ BH > 10−4M⊙yr−1, to match the approximate AGN
detection threshold in the LIRG sample. With this
limit the IR-bright galaxy fraction in the RSA sample
of Seyfert galaxies is then ∼22%±5%.
Fig.7.—DistributionsoftheintegratedSFRsofthelocalLIRGs
Figure 6 shows the distributions of the nuclear and (greyhistograms)andofthosewithaSeyfertclassificationand/or
total SFR/m˙ for the all the RSA Seyferts indicating [Nev] emitters (solid line histograms). The data points represent
BH
whichonesarealsoIR-bright. Clearly,thoseRSASeyfert the AGN fraction (scale on the right hand vertical axis) in SFR
intervalswith1σ errorbarsfortheAGNfraction.
classified as IR-bright have ratios similar to, although
slightlysmallerthan, thoseofthe completesampleoflo-
cal LIRGs. Given the fact that the AGNs in the RSA trated in the central regions for IR luminosities above
sample are not very different from those in the LIRGs, 1011.8L⊙, as shown by D´ıaz-Santos et al. (2010). We
the results above would suggest that the bright AGN conclude that the higher AGN incidence at low SFRs in
phase comes after and is somewhat distinct from the local LIRGs provides further evidence for a time delay
LIRG star forming phase. This interpretation would be betweenthe peaksofthe starformationandBHgrowth.
consistent with observational works showing a delay be-
4.3. Key differences between merger and non-merger
tweentheonsetofthestarformationactivityandthelat-
local LIRGs
ter feeding of the AGN with the consequent BH growth
(Davies et al. 2007; Wild et al. 2010) as well as with Before we explore the role of the LIRG phase in the
predictions from numerical simulations (e.g., Hopkins contextofstarformationactivityandBHgrowthofmas-
2012). sive late-type galaxies (next section), it is important to
point out the differences in the star formation histories
4.2. SFRs versus AGN Detection in local LIRGs of merger and non-merger LIRGs in the local universe.
It is well known that the AGN fraction increases The triggering of the activity in local ULIRGs and
with increasing IR luminosity from LIRGs to ULIRGs, likely also in the most luminous local LIRGs is driven
and at the highest IR luminosities the AGN might by major mergers. As a consequence, ULIRGs tend to
dominate bolometrically the luminosity of the sys- host more luminous AGN (quasar or nearly quasar-like
tem (Veilleux et al. 1995, 2009; Yuan et al. 2010; luminosities), slightly more massive BH and have higher
Nardini et al. 2010; Alonso-Herreroet al. 2012). We AGNbolometriccontributionsthatlocalLIRGs(seeSec-
now examine further the possibility that the IR-bright tion 3.1 and Veilleux et al. 2009; Nardini et al. 2010;
AGN phase comes after the distinct IR-bright star- Alonso-Herreroet al. 2012). Local merger LIRGs do
forming phase. If that were the case, we would expect showevidenceforarecentandintenseperiodofstarfor-
the AGN fraction in LIRGs to be higher at the lowest mation that consumed the gas faster, that is, a bursty
SFR. The AGN fraction here is computed only for those SFR (see e.g., Alonso-Herrero et al. 2000, 2001) as in
LIRGs with a secure AGN detection, that is, a Seyfert local ULIRGs (Rodr´ıguez-Zaur´ınet al. 2010). In this
classification and/or the presence of mid-IR [Nev] lines. respect,mostofthelocalmerger-LIRGscouldbeconsid-
As in the previous section, we computed the integrated eredassub-ULIRGsandthey mightgothrougha future
SFRoftheindividualgalaxiesfromL aftersubtracting ULIRG phase or may already have experienced such a
IR
the AGN contribution. phase (Murphy et al. 2001).
Figure 7 shows the AGN fraction as a function of Major mergers, however, do not dominate in num-
the total SFR of the galaxy. There is a tendency for bers the population of local LIRGs, especially at LIR <
LIRGs in lowest SFR bin to have a higher AGN in- 3 × 1011L⊙, as inferred from morphological studies
cidence (36%+14%) than those in the highest SFR bin (see e.g., Sanders & Ishida 2004; Alonso-Herreroet al.
−12% 2006; Kaviraj 2009). Therefore, in most local LIRGs
(17%+20%). Alternatively, this could be explained if it the current episodes of star formation activity and BH
−11%
is harder to detect AGN in LIRGs with very high SFRs, growth are likely the result of a less violent process,
especially if it takes place mostly in the nuclear regions. e.g., via minor mergers, fly-by companions, and/or sec-
Conversely, it may be easier to identify an AGN for low ular evolution. Poggianti& Wu (2000) showed that
nuclearSFRs andextendedstar formation. The firstex- mostisolatedgalaxiesintheirsampleofIR-brightgalax-
planation is not likely however,as most of the LIRGs in ies showed on average more moderate optical Balmer
our sample in the high SFR bin also tend to have rather absorption features than the strongly interacting sys-
extendedSF.Inotherwords,fromFigure4(rightpanel) tems in their sample. This suggests that in non-merger
there is no tendency for the fraction of nuclear SFR to LIRGsthereisnotastrongcontributionfromadominant
increase with the total SFR of the galaxy. In fact, in post-starburst stellar population that would be associ-
local LIRGs the mid-IR emission (both from AGN and ated with a recent bursty star formation history. This
star formation activity) only starts to be highly concen- agrees with the finding that most of the mass in non-