Extended X-ray emission in the IC 2497 - Hanny's Voorwerp system: energy injection in the gas around a fading AGN

2 MB·

by  Lia F. Sartori

MNRAS000,1–8(2016) Preprint29January2016 CompiledusingMNRASLATEXstylefilev3.0
Extended X-ray emission in the IC 2497 - Hanny’s
Voorwerp system: energy injection in the gas around a
fading AGN
Lia F. Sartori1(cid:63), Kevin Schawinski1, Michael Koss1, Ezequiel Treister2,
W. Peter Maksym3, William C. Keel4, C. Megan Urry5, Chris J. Lintott6,
O. Ivy Wong7
6
1 1Institute for Astronomy, Department of Physics, ETH Zu¨rich, Wolfgang-Pauli-Strasse 27, CH-8093 Zu¨rich, Switzerland,
0 2Departamento de Astronom´ıa, Universidad de Concepcio´n, Casilla 160-C, Concepcio´n, Chile,
2 3Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA,
4Department of Physics and Astronomy, University of Alabama, Tuscaloosa, AL 35487, USA,
n
5Yale Center for Astronomy & Astrophysics, Physics Department, P.O. Box 208120, New Haven, CT 06520, USA
a
6Oxford Astrophysics, Denys Wilkinson Building, Keble Road, Oxford OX1 3RH, UK
J
7ICRAR, The University of Western Australia M468, 35 Stirling Highway, Crawley, WA 6009, Australia
7
2

AcceptedXXX.ReceivedYYY;inoriginalformZZZ
]
A
G
ABSTRACT
h. We present deep Chandra X-ray observations of the core of IC 2497, the galaxy as-
p sociated with Hanny’s Voorwerp and hosting a fading AGN. We find extended soft
- X-ray emission from hot gas around the low intrinsic luminosity (unobscured) AGN
o (L ∼ 1042 −1044ergs−1). The temperature structure in the hot gas suggests the
r bol
t presence of a bubble or cavity around the fading AGN (Ebub ∼ 1054 −1055 erg). A
s
possiblescenarioisthatthisbubbleisinflatedbythefadingAGN,whichafterchang-
a
[ ing accretion state is now in a kinetic mode. Other possibilities are that the bubble
has been inflated by the past luminous quasar (Lbol ∼1046ergs−1), or that the tem-
1 perature gradient is an indication of a shock front from a superwind driven by the
v
AGN.WediscussthepossiblescenariosandtheimplicationsfortheAGN-hostgalaxy
0
interaction,aswellasananalogybetweenAGNandX-raybinarieslifecycles.Wecon-
5
clude that the AGN could inject mechanical energy into the host galaxy at the end
5
of its lifecycle, and thus provide a source for mechanical feedback, in a similar way as
7
0 observed for X-ray binaries.
.
1 Keywords: galaxies:active–X-rays:galaxies–quasars:general–quasars:individual
0 (IC2497)
6
1
:
v
i
X 1 INTRODUCTION suggests that the cloud is photoionised by the hard contin-
uum of an active galactic nucleus (AGN) in IC2497 rather
r IC2497 is a massive (M ∼ 1010M ), nearby (z = 0.05)
a (cid:12) thanstarformation(Lintottetal.2009).Moreover,therel-
spiral galaxy associated with Hanny’s Voorwerp (HV here-
atively quiet kinematics (line widths < 100 km/s) and low
after), an extended emission line region (11 x 16 kpc in
electrontemperature(T =13500±1300K)exludethepos-
projected extent) located at ∼ 20 kpc in projection from e
sibility of ionisation from shocks (Lintott et al. 2009).
the core of the galaxy and closely matching its redshift.
HV was discovered by Hanny van Arkel1 a citizen scien-
Largescaleradioobservationsofthesystemrevealedan
tist participating in the Galaxy Zoo project (Lintott et al. extendedHIstructure(M =8.5±2.1×1010M )withirreg-
(cid:12)
2008, 2009). The optical spectrum of HV is dominated by
ular kinematics around the southern part of IC2497, which
[OIII]λλ4959,5007emission,andthepresenceoflinesofhigh
suggests a tidal origin (Jo´zsa et al. 2009). These observa-
ionisationspeciessuchas[NeV]λλ3346,3426andHeIIλ4616
tions also show a kpc-scale structure which may be a jet
coming from the AGN. HV lies where the jet meets the HI
reservoirandcorrespondstoalocaldecrementofHIcolumn
(cid:63) E-mail:[email protected] density which may be due to photoionisation. At smaller
1 Hanny’sVoorwerpmeansHanny’sobjectindutch scales,Rampadarathetal.(2010)foundaradioAGNinthe
(cid:13)c 2016TheAuthors
2 Lia F. Sartori et al.
core of IC2497, and a second nuclear source which may be the galaxy, but also to (partially) resolve its emission. In
a jet hotspot in the large scale jet reported by Jo´zsa et al. Section 4 we discuss the results and possible consequences
(2009). for the AGN-host galaxy interaction as well as the analogy
Inordertoproducesufficientionisingphotonstopower bewteen AGN and X-ray binaries.
theobserved[OIII]emissioninHV,thequasarinIC2497has
to have a bolometric luminosity of at least L = 1046erg
bol
s−1.However,theopticalnuclearspectrumshowsonlyvery
weak emission lines, so that the ionising source is classified
as LINER or low-luminosity Seyfert galaxy (Lintott et al.
2009; Keel et al. 2012a). The two possible explanations to
reconcile observed and expected emission are 1) the quasar
isobscuredonlyalongourlineofsightbutnotinthedirec-
tion of HV, or 2) IC2497 hosts a faded AGN. This second
option would mean that the quasar dropped in luminosity
within the last ∼ 200 kyr, the travel time needed from the
photons to reach the cloud (Keel et al. 2012a), but HV still 2 OBSERVATIONS
remains lit up because of this travel time. Support for the
fading scenario was given by Schawinski et al. (2010) who We observed IC2497 with the Chandra Advanced CCD
analysedtheIRdatafromIRASandX-raydatafromXMM Imaging Spectrometer (ACIS) on 2012 January 8 (ObsID
andSuzaku obtainedforIC2497.First,asalreadyreported 13966, 59.35 ks) and 2012 January 11 (ObsID 14381, 53.4
inLintottetal.(2009)theIRdatashownoevidenceofthe ks; PI Kevin Schawinski, Cycle 13). We took the data with
reprocessed luminosity expected from an obscured strong theS-array(ACIS-S)inveryfaint(VFAINT)time-exposure
quasar. In addition, the XMM spectrum is best fitted with (TE)mode.Weperformedastandarddatareductionstart-
an unobscured power-law with photon index Γ=2.5±0.7, ing from the level 1 event files using the CIAO 4.7 soft-
asexpectedfromanunobscuredAGN,andacomponentfor ware (Fruscione et al. 2006) provided by the Chandra X-
the hot gas in the galaxy. The quality of the fit is not im- ray Center (CXC). We ran the script chandra_repro with
proved if additional absorption and obscuration are taken thelatestCALDB4.6.8setofcalibrationfiles,appliedsub-
intoaccount.Thismodelimpliesabolometricluminosityof pixel position using the Energy Dependent Subpixel Event
the AGN L = 4.2×1042erg s−1, which is ∼ 4 orders of Repositioning algorithm (pix_adj=EDSER; Tsunemi et al.
bol
magnitudelowerthanwhatexpectedfromHV.Inaddition, 2001; Li et al. 2003, 2004), and flagged the background
Suzaku observations show only a marginal detection with events most likely associated with cosmic rays by run-
the 15-30 keV PIN camera (hard X-ray detector). If these ningthetaskacis_process_events(check_vf_pha=yesin
counts are real, this would mean that tha AGN is strongly chandra_repro). We analysed the background light curves
obscuredandweareobservingthereemissioninsteadofthe withtheCIAOroutinelc_sigma_clip()andfoundnointer-
AGNpower-law,buttheobtainedluminositywouldstillbe val of unusually strong background flaring. Because of the
∼2ordersofmagnitudebelowwhatexpectedfromHV.Also low count rate the observations show no evidence of pileup.
the HST observations presented by Keel et al. (2012a) ar- Inordertocorrectfortheoffsetintherelativeastrome-
gueagainstthehypothesisofaluminous,stronglyobscured tryofthetwoobservations,wefirstrebinnedthereprocessed
AGN.WFC3imagesshowacomplexduststructurenearthe event files to 1/4 pixel resolution (0.125 arcsec) using the
nucleusofthegalaxy,buttheviewofthenucleusisnothin- CIAO dmcopy routine with binning factor 0.25. We then ran
dered by absorbing features, so that obscuration cannot be the CIAO dmstat routine on the rebinned images to get the
the explanation of the lack of strong AGN features. More- centroid of the emission from IC2497. In the following we
over, there is no high-ionisaion gas near the nucleus, sug- will consider the centroid as the centre of the emission in
gesting that the current radiative output from the AGN is the respective observation.
low. We extracted the spectra and generated the response
The observations described above suggest that the filesusingtheCIAO toolspecextract(seeSection3formore
quasarinIC2497droppedinluminositybyatleast2orders details about the extraction regions). For the background
ofmagnitudeinthelast∼105 years.Thesystemcomposed estimation we considered three source-free circular regions
byHVanditsgalaxyisthereforeagreatlaboratorytostudy witha5arcsecradius∼20arcsecawayfromthesource.We
AGN variability on previously inaccesible timescales. The then grouped each spectrum with a minimum of 3 counts
analysis of the galaxy allows us to study the present state per bin using the Heasoft tool grppha.
of the AGN and the AGN-host galaxy interaction in this We simulated the PSF images needed for the anal-
“post quasar”phase, while HV provides information about ysis using MARX 5.1 (Davis et al. 2012). First, we cre-
thepaststateoftheAGN.Finally,thedistancebetweenthe ated a model of the emission using Xspec and Sherpa
two provides information about timescales. (see Section 3 for details about the assumed models).
Inthispaperwepresenttheanalysisofnew,deepChan- We then used MARX to run the raytrace simulation and
dra X-ray observations of IC2497. The paper is organised project the ray-tracings onto the ACIS-S detector. Since
as follows. In Section 2 we describe the X-ray observations we are interested on the sub-pixel regime we ran the sim-
and the data reduction. In Section 3 we describe the per- ulations with pixadj=EDSER and AspectBlur=0.19 arcsec
formed imaging and spectroscopic analysis. Because of the (telescope pointing uncertainty). We also included the tele-
good spatial resolution reached by Chandra, this analysis scope dithering (DitherModel=INTERNAL, DitherAmp_Ra=8,
allows us not only to measure the overall X-ray flux from DitherAmp_Dec=8) and corrected for SIM drift and offset2.
MNRAS000,1–8(2016)
Extended X-ray emission in IC 2497 3
14381
5 kpc
5 arcsec
13966
105 ks PSF
HST [OIII] narrow band image
Figure1.Left:Chandra X-rayimagesforObsID14381(53.4ks),ObsID13966(59.35ks),andsimulated105 ksPSF.Thegreenannuli
showtheregionsfromwhichweextractthespectraandcomputethebrightnessprofile.Thespatialscaleisthesameforeachimage.The
colourscaleislinear(0-5countsfortheobservations,0-1800countsforthe105 ksPSF).Right:Chandra X-raycontoursoverplottedon
theHubble[OIII]narrowbandimageofIC2497(top)andHanny’sVoorwerp(bottom;fromKeeletal.2012a).
3 ANALYSIS consider a source-free annulus with internal radius 4 arcsec
and external radius 7 arcsec. The obtained radial profiles,
3.1 Radial profile and source extension
normalizedtothetotalnumberofcountsina4arcsecaper-
Wecomputethe0.5-6.0keVbackground-subtractedsurface ture, are shown in Fig. 2. The error bars correspond to the
brightnessprofilesofthetwoobservationsstartingfromthe 1 σ confidence interval (68.23%). For the annuli with more
reprocessed event files. We use the CIAO tool dmextract to than25counts(N>25)wecomputethe1σerrorofthetotal
extractthecountsinfourconcentricannulicenteredonthe counts (n√ot background subtracted) using Gaussian statis-
centroidofeachexposure,andwithouterradius0.5,1.0,1.5 tic,σ=√N.ForN≤25weusetheGehrelsapproximation
and 4.0 arcsec (Fig. 1, first panel). For the background we σ=1+ N +0.75(Gehrels1986;thisactuallycorresponds
to the 84.13% upper limit for a Poisson distribution). The
profiles of the two observations are consistent within 1 σ (2
σ in the last annulus), showing no evidence of variability.
2 For details on simulating a PSF image using MARX see http:
//cxc.harvard.edu/chart/threads/marx_sim/ The emission from an AGN is expected to be point-
MNRAS000,1–8(2016)
4 Lia F. Sartori et al.
100 ]
nts 1134936861 1keV− 10−2
ou mean 1 −
c PSF s
ot 10-1 s [ 10−3
/ t unt
o
2arcsec10-2 alized c 10−4
s / orm
t n
n
ou10-3 4 10−3
c ×
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
radius [arcsec] 1 2 5
Energy [keV]
Figure 2.NormalizedsurfacebrightnessprofilesoftheChandra
images and of the simulated PSF. ObsID 13966 and 14381 are Figure 3. Chandra 0.5 - 6.0 keV X-ray spectrum of IC2497
showninpurpleandgreen,respectively,whilethebluelineshows (total emission, r = 4 arcsec) for ObsID 14381 (black) and Ob-
the mean between the two observations. The profiles of the two sID 13966 (grey). The spectra are binned so that each bin con-
observations are consistent within 1 σ (2 σ in the last annulus). tainsatleast3counts.Thelinescorrespondtothebestfitmodel
The gray line shows the profile of the simulated PSF. The PSF phabs×(zpow+APEC), where zpow corresponds to the AGN emis-
profileissteepertowardthecenterthanthesource’sprofile.This sion and APEC to the emission from a hot diffuse gas (see text).
confirmsthatthesourceisextended. TheabsorptionisfixedtotheGalacticvalue.Theresidualsofthe
fit are shown in the bottom panel. The dotted and dashed lines
correspond to the phabs×zpow and phabs×APEC models, respec-
like. However, the radial profile suggests that the emission tively,bothnormalisedbyafactor10.
in IC2497 is extended. Moreover, the spectrum extracted
froma4arcsecapertureisbest-fitwithacombinationoftwo
3.2 Spectral fit and temperaure profile
models:apower-lawmodelfortheAGNemission,andacol-
lisionallyioniseddiffusegasmodel(APEC)representingthe For both observations we use specextract to extract the
hotgasinthegalaxy,whichcanbeexpectedtobeextended spectrum of the total X-ray emission in IC2497 from an
(Fig. 3, Table 1). This is true both performing a simulta- aperture with r = 4 arcsec centered on the centroid. We
neous fit of the two observations, or consideringthem sepa- fit the spectra in the 0.5 – 6.0 keV energy range with
rately (the best-fit parameters are always consistent within Xspec using Cash statistic (Gehrels 1986) as fit statistic,
the90%confidencelevel,seemoredetailsaboutspectralfit and Chi-Square statistic as test statistic. Since there is
inSection3.2).Inordertounderstandiftheobservedexten- no variability in the observations and between the obser-
sion is physical, and not only due to the PSF, we compare vations at the 90% level we fit the spectra from the two
theradialprofileoftheobservationswiththeexpectedPSF data sets simultaneously. The source is best fitted as a
profile. We simulate the PSF image as described in Section sum of two components: a power-law model representing
2, starting from the fit to the total emission. We first cre- the AGN emission (zpow model, Γ = 2.33+0.15), and a
−0.16
ate a model of the total emission using Xspec and Sherpa, model for the hot gas emission at the redshift of the host
andthespectralmodeldescribedabove,wherewealsocon- galaxy, with abundance fixed at solar value (APEC model,
sider Galactic absorption (phabs×(zpow+APEC)), and then kT =0.953+0.048 keV)3.Theabsorptioncomponent(phabs)
−0.054
use MARX to run the ray trace simulation and project the is fixed at the Galactic value (NH = 1.31×1020 cm−2, as
ray-tracings onto the ACIS-S detector. In order to get a given by the Colden Calculator with the NRAO dataset4).
goodsamplingofthePSF,andthusunderstanditsanalytic All the best fit parameters are listed in Table 1. The
shape,weincreasetheexposuretimeto105 ks.Wecompute power-law component is consistent with a low luminosity
the PSF brightness profile in the same way as described AGN with L2−10keV =5.8×1040ergs−1.
for the real observations. Since the PSF profile should in
principle depend on the input spectrum and on the source In order to understand the properties of the extended
position on the ACIS-S detector we repeat the simulations emission we attempt a spatially resolved spectral analysis.
for both observations separately. However, due to the simi- For each observation we use specextract to extract the
larityofmodelspectraandpositions(bothobservationsare
nearly on-axis) the normalized PSF brightness profiles are
3 SinceIC2497isamassivegalaxy,theassumptionofsolarabun-
the same.
danceshouldbeappropriate.However,wealsoperformedthefits
Fig. 2 shows the comparison between the normalized
considering metallicities of 0.2 and 0.5 solar, and the differences
brightness profile of the source (two observations and mean
in the measured temperatures are only at the <10% level. This
value) and of the PSF. The PSF profile is steeper toward
istruebothforthetotalemissionandfortheannulidescribedin
thecenterthanthesource’sprofiles.Thisconfirmsthatthe thenextparagraphs
source is extended. 4 http://cxc.harvard.edu/toolkit/colden.jsp
MNRAS000,1–8(2016)
Extended X-ray emission in IC 2497 5
spectrum from a circular aperture of radius 0.5 arcsec cen-
tered on the centroid, and 3 annuli with outer radius 1.0,
1.5 and 4.0 arcsec. Since we will account for the PSF dur- 1.2
ing the fit, as described in details in the following, we set
correctpsf=no. We then fit the spectra in the same way
as described for the total emission. The best fit parameters ]1.0
V
are listed in Table 1, while the fitted spectra are shown in e
k
appendix A. [
TheAGNemissionisexpectedtobepoint-likeandthe kT 0.8
zpowcomponentshouldinprinciplebepresentinthenuclear
spectrumonly.However,becauseofthePSFsomeAGNcon-
tributionispresentalsointheexternalannuli.Wetherefore 0.6
perform the following analysis, where we fix the AGN com-
ponentinthefitsuchthatthenormalizationandthepower-
lawslopearePSF-dependent,andletonlytheAPECmodel 0 1 2 3 4
free to vary. First, we simulate a PSF in the same way as radius [arcsec]
describedinSections2and3butforazpowmodelonly,ex-
tractthespectrafromtheannuliconsideredfortheanalysis,
and fit them using the response files (RMF and ARF files) Figure 4. Projected temperature profile from the spatially re-
obtained for the real spectra5. Since the number of counts solved spectral analysis. Each bin corresponds to one annulus
definedinFig.1.Thepurplepointscorrespondtotherealdata.
inthesimulatedspectraishigh,wegroupthemwithamin-
Thereisahintofatemperaturegradientfromthecentertothe
imum of 25 counts per bin and use Gaussian statistic. In
first two annuli with a statistical significance above 1 σ, which
eachannulusthespectrumisconsistentwithapower-lawat
could indicate the presence of a bubble powered by the central
the redshift of the galaxy (zpow). However, since the PSF
AGN.Thegrayshadedareacorrespondstothetemperaturepro-
alsodependsontheenergy,thepower-lawslope(PhoIndex) filemeasuredinthesimulatedPSF(seeSection3.2formorede-
varies with radius, as described in Table 1. We therefore tails). This temperature does not show the gradient seen in the
fitthe realspectrawiththe phabs×(zpow+APEC)modelde- realdata,whichconfirmsthatthegradientisnotduetothePSF.
scribed above, where for each annulus we fix the power-law
slopeatthevalueobtainedfromthePSFsimulation.Inor-
der to fix the power-law normalisation, we first fit the total
emission(0.0–4.0arcsec),andassumethatineachannulus
thenormalisationscalesinthesamewayasseeninthePSF
simulation: repeat the fit of the real data without fixing the power-law
slopeandnormalisation.Alsointhiscasewefindatempera-
norm PSF turegradientconsistentatthe90%confidencelevelwiththe
norm_pl = i ×norm_pl (1)
i norm PSF tot one showed in Fig. 4. Finally, we investigate the possibility
tot
that the temperature gradient is due to a PSF effect. We
wherenorm_pl andnorm_pl arethenormalisationof
i tot simulate a PSF with the same model as the total emission
thepower-lawcomponentintherealspectraextractedfrom
(phabs×(zpow+APEC)) and measure the temperature in the
thei-teannulusandfromthetotalaperture(0.0-4.0arcsec),
sameannuliasusedfortherealsource.AswecanseeinFig.
whilenorm PSF andnorm PSF arethesamebutforthe
i tot 4 (grey shadow), the measured temperature does not show
PSF simulation.
thegradientseenintherealdata,whatagainconfirmsthat
In Fig. 4 we show the projected temperature profile
the gradient is not due to the PSF.
(from the APEC parameter kT). We find a hint of increas-
The temperature increase showed in Fig. 4 could be
ingtemperaturefromthecentertowardthefirstandsecond
symmetric, or due to a feature in a preferred direction. In
annuli with a statistical significance above 1 σ. We discuss
ordertoinvestigatethesepossibilitieswerepeattheanalysis
possible implications of the found temperature gradient in
dividingthefirstannulus(0.5–1.0arcsec)into4semiannuli
Section 4. In order to understand if this gradient is due to
of90degeach.TheresultsarelistedinTable1.Becauseof
systematicsinthefitsweperformaMonteCarlosimulation:
the lower number of counts in each region, these fits are
first we use the Xspec function fakeit to simulate spectra
lessreliablethantheonetothetotalannulus.However,the
similar to the observed ones (similar power-law slope and
temperaturegradientseemstobepresentineachdirection,
normalisation)withagiventemperatureprofile.Wethenfit
although it is more important in the southern part. This
the simulated spectra using the same procedure as for the
seems to rule out the possibility of a single feature with
realdata.Oursimulationsshowthatwecanrecoverthein-
significantly higher temperature.
put profile within 1 σ, and thus the gradient is not due to
It is important to note though that the low number
systematics.Wealsoinvestigatethepossibilitythatthegra-
of counts in our spectra do not allow us to deproject the
dient is artificially introduced by our fitting procedure. We
emission from the different shells, so that the reported gas
temperaturesandnormalisationsareprojectedvaluesalong
5 inthiswaywetreatthesimulatedspectraassimilaraspossible the line of sight. A deeper ∼ 1 Ms Chandra exposure will
totherealspectra.Werepeatedtheanalysisalsousingsimulated allow us to obtain a minimum of 200 counts in all region
RMF and ARF files, but the fit parameters do not differ signifi- of interest, and thus properly deproject the emission and
cantly. carefully measure the intrinsic gas parameters.
MNRAS000,1–8(2016)
6 Lia F. Sartori et al.
Region NetCounts PLslope PLnorm Temperature APECnorm
[10−7] kT[keV] [10−7]
Totalemission(0.0-4.0arcsec) 620 2.33+0.15 67.486+9.453 0.953+0.048 38.735+6.558
−0.16 −9.162 −0.054 −6.381
Nucleus(0.0-0.5arcsec) 262 2.47 46.708 0.702+0.211 5.193+1.941
−0.212 −1.823
Annulus1(0.5-1.0arcsec) 221 2.31 15.387 1.133+0.100 21.143+4.293
−0.111 −3.974
Annulus2(1.0-1.5arcsec) 52 1.89 2.019 1.003+0.248 5.023+1.077
−0.120 −1.130
Annulus3(1.5-4.0arcsec) 84 1.65 1.213 0.809+0.158 10.013+1.425
−0.080 −1.315
Semi-ann.1(45-135deg) 54 2.28 3.709 1.064+0.249 5.185+2.893
−0.123 −1.277
Semi-ann.2(135-225deg) 56 2.37 3.881 0.920+0.105 5.316+1.343
−0.139 −1.180
Semi-ann.3(225-315deg) 55 2.27 3.709 1.575+und. 7.933+7.583
−0.343 −3.163
Semi-ann.4(315-405deg) 55 2.29 3.886 1.283+0.386 7.401+4.406
−0.185 −2.173
Table 1. Best fit parameters for each region considered in this analysis (nucleus, three annuli, four semi-annuli). The assumed model
isphabs×(zpow+APEC).Inthenucleusallthefitparametersarefree.Thepower-lawslopeandnormalisationarefixedaccordingtothe
PSFsimulation(seetextfordetails).AbundanceisfixedatSolarvalueandabsorptionatGalacticvalue(NH=1.31×1020 cm−2).
4 DISCUSSION 2014). On the other hand, the projected temperature mea-
suredinthecenterandintheexternalannulusareconsistent
We detect extended soft X-ray emission in the nucleus of
withwhatfoundinULIRGs(Grimesetal.2005).Thehigh
IC2497,inadditiontothepoint-likepower-lawemissionex-
projected temperature, especially in the two central annuli,
pected from an AGN. We associate this emission with a
may be an indication that we are observing a shock front,
hot, diffuse gas. The projected temperature profile (Fig. 4)
e.g.fromanAGN-drivensuperwinds.ModelsofAGN-driven
showsaresolvedcoldcenter.Thetemperaturegradientmay
superwinds suggest a temperature of T ≈1010−1011K for
be interpreted as a bubble or cavity in the hot gas which is
theinnerreverseshockandT ≈107Kfortheforwardshock,
kinematically inflated by the AGN. Alternatively, the tem-
and the expanding medium can persist also after the AGN
perature gradient could indicate a shock front, e.g. from an
hasswitchedoff(Zubovas&King2012).Theobservedtem-
AGN-driven superwind. The different hypothesis, and the
peratures are consistent with the forward shock, but too
consequences for the AGN-host galaxy interaction, are de-
low for the inner shock. Is the AGN in IC2497 too faint to
scribed in the following subsections.
producesuchahightemperature?Oristheobservedshock
ThecurrentanalysisalsosupportstheideathatIC2497
front a residual of a wind driven by the AGN during the
hosts a fading AGN, as suggested by Lintott et al. (2009),
past quasar phase? Deeper X-ray observations (leading to
Schawinski et al. (2010) and Keel et al. (2012a). In fact, by
X-ray spectra with more counts) will allow us to deproject
scaling the quasar spectral energy distribution (SED) tem-
the emission and carefully constrain the intrinsic tempera-
platesfromElvisetal.(1994)tomatchthemeasuredX-ray
ture, density and X-ray luminosity of the gas in each shell
luminosity, L =5.8×1040ergs−1, we obtain a bolo-
2−10keV aroundtheAGN.Thiswillallowustoestimatethevelocity
metricluminosityL ∼1042ergs−1,whichis∼4ordersof
bol of the shock and the ratio between X-ray and mechanical
magnitudebelowwhatisneededinordertopowerHanny’s
luminosity(e.g.Chu&MacLow1990),andthereforeprobe
Voorwerp(Lintottetal.2009;Schawinskietal.2010).This
the superwind hypothesis and constrain between the two
result is consistent with what reported by Schawinski et al.
scenarios.
(2010).AsreportedbySchawinskietal.(2010),iftheAGN
is compton thick, this would mean that it is strongly ob-
scured and we are observing the reemission instead of the Another possible interpretation of the observed pro-
AGN power-law, and the bolometric luminosity would be jected temperature gradient is that the AGN is, or was,
L ∼1044ergs−1 (seesection1).Thisisstill∼2ordersof blowing away the hot gas around it by injecting kinetic en-
bol
magnitudebelowwhatisneededinordertopowerHanny’s ergy,thereforecreatingabubbleorcavity.Ifthehypothesis
Voorwerp. Hard X-ray observation from NuSTAR are now ofabubbleinflatedbyafadingAGNisconfirmed,thiswould
crucial in order to understand if the AGN is obscured and give us new insights into the AGN-host galaxy interaction
measure the real magnitude of the shutdown. in the“post-quasar”phase, as described below. It is impor-
tant to note though that the low number of counts in our
spectra do not allow us to deproject the emission from the
different shells, so that the reported gas temperatures and
4.1 Hot gas in IC2497
normalisations are projected values along the line of sight.
TheprojectedtemperatureobservedinthegasinIC2497is A deeper ∼1 Ms Chandra exposure will allow us to obtain
intherangekT ≈0.7−1.1keV(T ≈0.8−1.2×107K).This a minimum of 200 counts in all region of interest, and thus
relativelyhighthemperaturemayindicatethattheemission properly deproject the emission and carefully measure the
from the hot gas is not due to photoionisation, as seen in gas parameters. In this way we will be able to further test
otherSeyfertgalaxies(e.g.Bianchietal.2006,Greeneetal. the bubble hypothesis.
MNRAS000,1–8(2016)
Extended X-ray emission in IC 2497 7
4.2 Energy in the bubble points,withatypicalAGNphaselasting∼105 yrs.Ifwhat
we observe in IC2497 is typical to each burst, i.e. the AGN
In order to understand the effect of the AGN on the sur-
injects ∼ 1055 erg in each cycle, the sum of the energy in-
rounding medium we need to estimate the energy in the
jectedineachcyclecouldstarthavingsomeeffectonthegas
bubble, i.e. the energy required to create it. The uncertain-
reservoire (although alone it would probably be insufficient
tiesinthiscalculationarequitehighduetotheuncertainties
tototallyunbindit).Finally,ifweasumethatthebubbleis
in the spectral fitting and the unknown shape of the bub-
inflated by the fading AGN after the quasar drop in lumi-
ble and hot gas. Therefore, the obtained energy should be
nosity(i.e.inthelast∼2×105 yr),theAGNshouldinject
seenasanorderofmagnitudeestimation.Onepossibleway
atleast∼1042 erg/sinkineticform,andthebubbleshould
to approach the problem is to consider an analogy with the
expand with a velocity v ∼5000 km/s.
cavitiesobservedinclusters(e.g.Bˆırzanetal.2004,Rafferty exp
et al. 2006, McNamara & Nulsen 2012, Vantyghem et al.
2014). In this case the energy needed to create the bubble 4.3 Accretion state changes, AGN feedback and
can be espressed as: analogy with X-ray binaries
γ
E = pV (2) Asexplainedabove,onepossibleinterpretationofthebubble
bub γ−1 bub
isthatitisinflatedbythefadingAGNafterthequasardrops
wherepisthepressureofthegassurroundingthebub- in luminosity, and we are witnessing a state change in the
ble, V is the bubble’s volume, and γ is the heat capacity accretion mode of the quasar. After dropping in accretion
bub
ratio of the gas inside the bubble (γ =4/3 for a relativistic rate,thelow-EddingtonAGNmovedfromaradiativemode
gas,γ =5/3foranon-relativistic,monoatomicgas).Forour to a kinetic mode, and now puts out part of its energy in
estimationweconsiderasphericalbubblewithradiusr=1 kineticform.Inthisway,itcandoafeedbackonthegalaxy,
arcsec, corresponding to r = 987 pc at the redshift of the e.g.byinflatingabubbleinthesurroundinggasorlaunching
source. The pressure p can be expressed as: theradiojets.ThehypothesisofastatechangeinIC2497is
consistent with the analysis done by Keel et al. (2012a). In
fact,theHST STISspectraandimagesindicatethepresence
p=n(kT) (3)
of an outflow from the nucleus which suggests that some of
where n and kT are the number density and temper- theenergyoutputfromtheAGNswitchedfromradiationto
ature of the gas outside the bubble, as derived from the kinetic form.
spectral fit to the third annulus (1.5-4.0 arcsec). For a fully The idea of a bubble inflated by an AGN undergo-
ionised gas with 10% He we can assume n ∼ n and ing a state change fits naturally into the general picture of
e H
n∼2n ,sothatncanbederivedfromtheAPECnormalisa- unifying BH accretion physics from X-ray binaries (XRB)
e
tion: to AGN. Rapid state transitions, on timescales of days to
10−14 (cid:90) weeks, are often observed in XRB, and are connected to a
norm= n n dV (4) change in radiative output (e.g. Fender et al. 2003, Coriat
4πD 2(1+z)2 e H
A etal.2011,Eckersalletal.2015).Theluminoussoftstateis
where we consider the normalisation obtained in the radiativelyefficientandtheaccretioncanbedescribedbyan
third annulus, while the volume corresponds to a napkin opticallythick,geometricallythinaccretiondisc(Shakura&
ring for a sphere with r = 4 arcsec and a cylinder with Sunyaev 1973). On the other hand, the hard state is radia-
r = 1.5 arcsec (i.e. the volume of the external shell which tively inefficient and the accretion is due to an advection
is projected into the third annulus). In this way we obtain dominated (ADAF; Narayan & Yi 1995) or jet dominated
n ≈ 0.02cm−3 and p ≈ 2.5×10−11erg cm−3, so that the acretionflow(JDAF;Falckeetal.2004),sothatmostofthe
energy in the bubble is E ∼1054−1055 erg. energyoutputfromtheaccretingBHisinkineticform.The
bub
Now we will carry out comparisons to test what these hypothesis of a state switch connected to the drop in lumi-
numbers can tell us about the bubble itself and the effect nosityisconsistentalsowithDoneetal.(2007)whichshow
of the AGN on the host galaxy. The energy in the bubble that, as the Eddington ratio decreases, the radiative power
is ∼ 3−4 orders of magnitude higher than what expected ofanXRBdecreaseswhilethemechanical/kineticpowerin-
from a supernova explosion, so that the bubble cannot be creases. Alexander & Hickox (2012) suggested that this is
inflated by supernovae. Keel et al. (2012a) found a ring of true also for AGN. In this framework, low-luminosity AGN
ionised gas on the northern part of the nucleus of IC2497. correspond to the hard XRB, while the quasars correspond
Theringhasaradiusr≈250pcwhichcanbeseeninboth tothesoftstate.DifferentstudiesshowedthattheAGNac-
HST Hα and [OIII] continuum images, and with a kine- cretion discs can indeed change their state from a classical
matic age < 7×105yr (see Figure 5 in Keel et al. 2012a). radiatively-efficient quasar state to a radiatively inefficient
The kinetic energy in this ring is ∼ 3×1051 erg, ∼ 3−4 accretion flow in a similar way as observed in X-ray bina-
orders of magnitude lower than what found in our bubble, ries.However,sincethedynamicalandviscoustimescalesin-
andalsothespatialextentisdifferent.Thisisanindication volvedintheaccretionstatechangesincreasewithBHmass,
thatadditionalcomponentsneedtobeconsidered,probably thetimescalesexpectedforAGNaresignificantlylonger(e.g.
happening at the same time, near the nucleus of IC2497. Narayan & Yi 1995, Maccarone et al. 2003, McHardy et al.
Finally, the binding energy of IC2497 is ∼ 1060 erg, which 2006, K¨ording et al. 2006, Schawinski et al. 2010). As an
means that the present-day AGN is not able to blow away example, the analogous of a state transition lasting several
the whole ISM. However, Schawinski et al. (2015) recently days in an X-ray binary with a ∼10M BH will last ∼104
(cid:12)
suggestedthatAGN“flicker”onandoff100-1000timesalter- yrinanAGNwitha∼107M SMBH.Thesetimescalesare
(cid:12)
nating high-Eddington bursts and lower-Eddington troughs consistent with what is seen in IC 2497.
MNRAS000,1–8(2016)
8 Lia F. Sartori et al.
5 SUMMARY KeelW.C.,etal.,2012a,AJ,144,66
KeelW.C.,etal.,2012b,mnras,420,878
We have presented new, deep Chandra X-ray observa-
KeelW.C.,etal.,2014,preprint, (arXiv:1408.5159)
tions of IC 2497. This galaxy is associated with the ex- K¨ordingE.G.,JesterS.,FenderR.,2006,mnras,372,1366
tended emission line cloud called Hanny’s Voorpwerp and Li J., Kastner J. H., Prigozhin G. Y., Schulz N. S., 2003, ApJ,
hosts a fading AGN. The X-ray data show extended 590,586
hot gas around the very low-luminosity unobscured AGN Li J., Kastner J. H., Prigozhin G. Y., Schulz N. S., Feigelson
(L ∼1040ergs−1).Moreover,thedatasuggestatem- E.D.,GetmanK.V.,2004,ApJ,610,1204
2−10keV
peraturegradientinthehotgaswhichmaybeinterpretedas LintottC.J.,etal.,2008,MNRAS,389,1179
anexpandingbubble(r ∼1kpc,E ∼1054−1055erg). LintottC.J.,etal.,2009,MNRAS,399,129
bub bub
MaccaroneT.J.,GalloE.,FenderR.,2003,MNRAS,345,L19
One hypothesis is that this bubble is inflated by the AGN
McHardyI.M.,KoerdingE.,KniggeC.,UttleyP.,FenderR.P.,
that, after changing accretion state and dropping in lumi-
2006,Nature,444,730
nosity,isnowinakineticmode.Asaconsequence,theright
McNamaraB.R.,NulsenP.E.J.,2012,NewJournalofPhysics,
place to search for and study mechanical AGN feedback 14,055023
could be galaxies hosting fading AGN (i.e. the Voorwerp- NarayanR.,YiI.,1995,ApJ,452,710
jes,Keeletal.2012b,Keeletal.2014).TheideaofanAGN Rafferty D. A., McNamara B. R., Nulsen P. E. J., Wise M. W.,
doing mechanical work at the end of its radiatively efficient 2006,ApJ,652,216
phasealsofitsintothegeneralpictureofunifyingblackhole RampadarathH.,etal.,2010,A&A,517,L8
accretion physics from X-ray binaries to AGN. Follow-up SchawinskiK.,etal.,2010,ApJ,724,L30
SchawinskiK.,KossM.,BerneyS.,SartoriL.F.,2015,MNRAS,
deeper Chandra observations will allow us to confirm the
451,2517
presence of the bubble and better constrain its properties
ShakuraN.I.,SunyaevR.A.,1973,A&A,24,337
and its effect on the host galaxy.
Tsunemi H., Mori K., Miyata E., Baluta C., Burrows D. N.,
GarmireG.P.,ChartasG.,2001,ApJ,554,496
Vantyghem A. N., McNamara B. R., Russell H. R., Main R. A.,
ACKNOWLEDGMENTS Nulsen P. E. J., Wise M. W., Hoekstra H., Gitti M., 2014,
MNRAS,442,3192
LFS, KS and MK gratefully acknowledge support from ZubovasK.,KingA.,2012,ApJ,745,L34
Swiss National Science Foundation Professorship grant
PP00P2 138979/1 and MK support from SNSF Ambizione
grantPZ00P2 154799/1.LSthanksClaudioRicciandHans
APPENDIX A: X-RAY SPECTRA AND XSPEC
Moritz Gu¨nther for useful discussions. OIW acknowledges
FITS
a Super Science Fellowship from the Australian Research
Council.
This work is based on observations with the Chandra ThispaperhasbeentypesetfromaTEX/LATEXfilepreparedby
satelliteandhasmadeuseofsoftwareprovidedbytheChan- theauthor.
dra X-ray Center (CXC) in the application packages CIAO
and MARX. The research has made use of NASA’s Astro-
physics Data System Bibliographic Service.
REFERENCES
AlexanderD.M.,HickoxR.C.,2012,NewAstron.Rev.,56,93
BianchiS.,GuainazziM.,ChiabergeM.,2006,A&A,448,499
BˆırzanL.,RaffertyD.A.,McNamaraB.R.,WiseM.W.,Nulsen
P.E.J.,2004,ApJ,607,800
ChuY.-H.,MacLowM.-M.,1990,ApJ,365,510
CoriatM.,etal.,2011,MNRAS,414,677
Davis J. E., et al., 2012, in Society of Photo-Optical Instru-
mentation Engineers (Proc. SPIE) Conference Series. p. 1,
doi:10.1117/12.926937
DoneC.,Gierlin´skiM.,KubotaA.,2007,A&ARv,15,1
Eckersall A. J., Vaughan S., Wynn G. A., 2015, MNRAS, 450,
3410
ElvisM.,etal.,1994,apjs,95,1
FalckeH.,K¨ordingE.,MarkoffS.,2004,A&A,414,895
FenderR.P.,GalloE.,JonkerP.G.,2003,MNRAS,343,L99
Fruscione A., et al., 2006, in Proc. SPIE. p. 1,
doi:10.1117/12.671760
GehrelsN.,1986,apj,303,336
GreeneJ.E.,PooleyD.,ZakamskaN.L.,ComerfordJ.M.,Sun
A.-L.,2014,preprint, (arXiv:1404.4875)
Grimes J. P., Heckman T., Strickland D., Ptak A., 2005, ApJ,
628,187
J´ozsaG.I.G.,etal.,2009,A&A,500,L33
MNRAS000,1–8(2016)
Extended X-ray emission in IC 2497 9
Figure A1. Chandra X-ray spectra extracted from the nuclear region and from the three annuli, for ObsID 14381 (black) and Ob-
sID 13966 (grey). The spectra are binned so that each bin contains at least 3 counts. The lines correspond to the best fit model
phabs×(zpow+APEC), where zpow corresponds to the AGN emission and APEC to the emission from a hot diffuse gas (see text). The
absorptionisfixedtotheGalacticvalue.Theresidualsofthefitareshowninthebottompanel.
MNRAS000,1–8(2016)