Mapping of the molecular gas inflow towards the Seyfert nucleus of NGC4051 using Gemini NIFS

0.76 MB·English

by  Rogemar A. Riffel

Mon.Not.R.Astron.Soc.000,1–17(2002) Printed2February2008 (MNLATEXstylefilev2.2)
Mapping of molecular gas inflow towards the Seyfert
nucleus of NGC 4051 using Gemini NIFS
Rogemar A. Riffel1⋆, Thaisa Storchi-Bergmann1, Cl´audia Winge2,
8
Peter J. McGregor3, Tracy Beck4 and Henrique Schmitt5
0
0 1Universidade Federal do RioGrande do Sul, IF, CP 15051, Porto Alegre 91501-970, RS, Brazil.
2 2Gemini Observatory,c/o AURA Inc., Casilla 603, La Serena, Chile.
3Research School of Astronomy and Astrophysics, Australian National University,Cotter Road, Weston Creek, ACT 2611, Australia.
n
4Gemini Observatory,Northern Operations, Hilo, HI, USA.
a
5Remote Sensing Division, Naval Research Laboratory, 4555 Overlook Avenue, SW, Washington, DC 20375, USA;
J
Interferometric Inc., 13454 Sunrise Valley, Suite 240, Herndon, VA 20171.
1
3

Accepted 1988December 15.Received1988December14;inoriginalform1988October11
]
h
p
- ABSTRACT
o
We present two-dimensional (2D) stellar and gaseous kinematics of the inner
r
t ∼130×180 pc2 of the Narrow Line Seyfert 1 galaxy NGC4051 at a sampling of
s
a 4.5pc,fromnear-infraredK-bandspectroscopicobservationsobtainedwiththe Gem-
[ ini’s Near-infrared Integral Field Spectrograph (NIFS) operating with the ALTAIR

adaptive optics module. We have used the CO absorption bandheads around 2.3µm
2
to obtainthe stellarkinematics whichshow the turnoverofthe rotationcurve atonly
v
≈55pc from the nucleus, revealing a highly concentrated gravitationalpotential. The
8
stellarvelocitydispersionofthebulgeis≈60kms−1–implyingonanuclearblackhole
8
4 massof≈ 106M⊙ –withinwhichpatchesoflowervelocitydispersionsuggestthepres-
1 enceofregionsofmorerecentstarformation.Frommeasurementsoftheemission-line
. profileswehaveconstructedtwo-dimensionalmapsforthefluxdistributitions,linera-
1
0 tios,radialvelocitiesandgasvelocitydispersionsfortheH2,Hiand[Caviii]emitting
gas. Each emission line samples a distinct kinematics. The Brγ emission-line shows
8
0 no rotationas well as no blueshifts or redshifts in excess of 30kms−1, and is thus not
: restricted to the galaxy plane. The [Caviii] coronal region is compact but resolved,
v extendingoverthe inner75pc.Itshowsthehighestblueshifts –ofupto −250kms−1,
i
X and the highest velocity dispersions, interpreted as due to outflows from the active
nucleus, supporting an origin close to the nucleus. Subtraction of the stellar velocity
r
a field from the gaseous velocity field has allowed us to isolate non-circular motions
observed in the H2 emitting gas. The most conspicuous kinematic structures are two
nuclear spiral arms – one observed in blueshift in the far side of the galaxy (to the
NE), and the other observed in redshift in the near side of the galaxy (to the SW).
We interpret these structures as inflows towards the nucleus, a result similar to those
ofpreviousstudies in whichwe havefoundstreamingmotionsalongnuclearspiralsin
ionized gas using optical IFU observations. We have calculated the mass inflow rate
alongthenuclearspiralarms,obtainingM˙H2 ≈4×10−5M⊙yr−1,avalue∼100times
smallerthantheaccretionratenecessarytopowerthe activenucleus.Thiscanbeun-
derstood as due to the fact that we are only seeing the hot “skin” (the H2 emitting
gas)ofthetotalmassinflowrate,whichisprobablydominatedbycoldmoleculargas.
From the H2 emission-line ratios we conclude that X-ray heating can account for the
observed emission, but the H2λ2.1218µm/Brγ line ratio suggests some contribution
from shocks in localized regions close to the compact radio jet.
Key words: galaxies:Seyfert – infrared: galaxies – galaxies: NGC4051 (individual)
– galaxies: kinematics
1 INTRODUCTION
⋆ E-mail:[email protected] The presence of supermassive black holes (SMBHs) at
the centres of all galaxies which have stellar bulges is
2 Rogemar A. Riffel et al.
nowadays widely accepted by the astronomy comunity 9.3Mpc(Barbosa et al. 2006), such that 1′.′0 corresponds
∼
(Gebhardt et al. 2000; Ferrarese & Merritt 2000). Accord- to 45pc at the galaxy. It harbors one of the closest AGN,
ing to this scenario the energy emitted by an Active classified as a Narrow-Line Seyfert 1 (NLS1). We have se-
Galactic Nucleus (AGN) is due to the accretion of ma- lected NGC4051 for thisstudypartially on thebasis of the
terial onto the SMBH and implies the presence of a gas recentwork ofBarbosa et al.(2006),whoobtained 2Dstel-
reservoir close to the AGN. Sim˜oes Lopes et al. (2007) us- lar kinematics using the IFU of the Gemini Multi-Object
ing archival Hubble Space Telescope (HST) optical im- Spectrograph(GMOS)toobservethestellarabsorptionlines
ages for a large sample of early-type galaxies with and of the Calcium triplet around 8500˚A. These authors have
without AGNs, found that all AGN hosts have circumnu- shownthat theturnoverofthestellarrotation curveoccurs
clear gas and dust, while this is observed in only 26% of at onlyR 50pcfrom thenucleus,indicatingthat thestel-
∼
a pair-matched sample of inactive galaxies. A gas reser- larmotionsaredominatedbyahighlyconcentratedgravita-
voir close to the AGN is also supported by the presence tionalpotential.AsNIFSwiththeALTAIRadaptiveoptics
of recent star formation in the circumnuclear region of ac- module provides a much better image sampling and resolu-
tive galaxies (Schmitt,Storchi-Bergmann & Cid Fernandes tionthantheGMOS-IFU,wedecidedtofurtherinvestigate
1999; Boisson et al. 2000; Storchi-Bergmann et al. 2000; the kinematics of the nuclear region of NGC4051 in order
Cid Fernandeset al. 2001, 2005; Storchi-Bergmann et al. tobetterconstrainthegravitational potentialandthemass
2005).However,thestrongestsignaturesthatfeedingtothe oftheSMBH.Inadition,thisgalaxyshowsstrongH emis-
2
SMBHisoccurringincludetheobservationofstreamingmo- sion (Riffel, Rodr´ıguez-Ardila & Pastoriza 2006), making it
tionsinionizedgasalongnuclearspiralstowardsthenucleus a good candidate for a studyof its kinematics.
of nearby active galaxies using two-dimensional (2D) opti- Previous studies of NGC4051 include HST narrow-
calspectroscopy(Fathiet al.2006;Storchi-Bergmann et al. band [Oiii] images which show an unresolved nuclear
2007). source and faint extended emission (by 1′.′2) along the po-
Streaming motions as feeding signatures to the nu- sition angle PA=100◦ (Schmitt & Kinney 1996), the ap-
clear region have been previously observed in radio wave- proximate direction of the alignment of two radio compo-
lengths.Adler & Westpfahl(1996),forexample,havefound nents at 6cm separated by 0′.′4 (Ulvestad & Wilson 1984).
streamingmotions towardsthecenteralongthespiral arms Veilleux (1991) reported that the profiles of optical forbid-
of M81 in Hi, while Mundell& Shone (1999) have found den emission lines present blue wings reaching velocities of
similar streaming motions towards the nucleus along the up to 800kms−1, and proposed a model for the narrow-
weak bar of NGC4151. Closer to the center, most of the line region (hereafter NLR) with outflows and obscuring
gas is in the molecular phase, and CO observations have dust. Evidence for outflows have also been observed by
been used to map the gas kinematics and inflows (e.g. Christopoulou et al. (1997) for the [Oiii] emission to 1′.′5
Garc´ıa-Burillo et al. 2003; Kripset al. 2005; Boone et al. NE of the nucleus. Nagao, Taniguchi & Murayama (2000)
2007). usingthe[FeX]λ6374 emission linereport ahigh ionization
Molecular hydrogen emission lines are also relatively region extending to 3′.′0 SE of the nucleus. Lawrence et al.
strong in the near-IR K-band spectra of active galax- (1985) found a X-ray variability on time scales of the or-
ies, and previous studies suggest that its distribution derof1hourandSalvati et al.(1993)reportedfluxchanges
and kinematics is distinct from that observed in the by a factor of 2 in 6 months observed at 2.2µm and sug-
other emission lines, which are usually dominated by out- gestthatemission bydustreprocessing oftheUVradiation
flows (e.g. Crenshaw & Kraemer 2000; Das et al. 2005). from the nucleus is an acceptable explanation. Ponti et al.
In the Seyfert galaxies NGC2110 and Circinus, for ex- (2006)modelledthenuclearemissionbytwopowerlawcom-
ample, Storchi-Bergmann et al. (1999) have found broader ponents,oneduetotheAGNandotherduetoreflectionby
emission-line profiles for [Feii] and Paβ than for the theaccretion disk.
H λ2.1218µm emission-line in a study using near-IR long- This paper is organized as follows: in section 2 we de-
2
slit observations, suggesting a stronger influence of nuclear scribe the observations and data reduction. In section 3 we
outflows on the former emission lines and a different origin presenttheresultsforthestellarkinematics.Insection4we
fortheH emittinggas,consistentwiththecolderkinemat- present the emission-line flux distributions and in section 5
2
icsofthegalaxy disk.More recently,usingtwo-dimensional we present the results for the gas kinematics. In section 6
(hereafter 2D) near-IR spectroscopy of the Seyfert galaxy we discuss the results and in section 7 we present the con-
ESO428-G14, we (Riffelet al. 2006) have found that the clusions of this work.
H emission distribution was mostly restricted to theplane
2
ofthegalaxyandwaslessaffectedbytheAGNoutflowthan
the[Feii]and Paβ emission lines.
2 OBSERVATIONS AND DATA REDUCTION
In this paperwe use adaptiveoptics IFU spectroscopic
dataobtainedwiththeGemini’sNear-infraredIntegralField TheIFUspectroscopic datawereobtainedwithNIFSoper-
Spectrograph(NIFS- McGregor et al.2003),inthenear-IR ating with the ALTAIRadaptiveoptics module on the8-m
K-bandatasamplingof0′.′1 0′.′1,toinvestigatethestellar Gemini North telescope in January 2006 under the instru-
and gaseous kinematics of th×e inner 3 4arcsec2 of the ment science verification program GN-2006A-SV-123. The
nearby activegalaxy NGC4051. The∼stella×r kinematics will IFU has a square field of view of 3′.′0 3′.′0, divided into
be used to constrain the galaxy potential and mass of the 29 slices with an angular samplin≈g of 0×′.′1 0′.′04. The ob-
×
SMBH,butourmaingoalistolookforsignaturesoffeeding serving procedures followed the standard Object-Sky-Sky-
mechanisms at parsec scales through theH2 kinematics. Object dither sequence, with off-source sky positions since
NGC4051 is a SABbc galaxy at a distance of only thetargetisextended,andindividualexposuretimesof750s
Molecular gas inflow towards the Seyfert nucleus of NGC4051 3
tatedtothesameorientationoftheIFUobservations.Inthe
top-rightpanelwepresentanimageobtainedfromtheNIFS
data cube for the 2.12µm continuum emission, obtained
byinterpolation of thecontinuumundertheH λ2.1218µm
2
emission line. In thebottom panels we present threechara-
teristicIFUspectra:thenuclearspectrum(positionNinthe
continuummap),aspectrumfromalocationat1′.′2Eofthe
nucleus (position A) and another from 0′.′75 W of the nu-
cleus (position B). Both spectra correspond to an aperture
of 0′.′3 0′.′3. The emission lines are identified in the nuclear
×
spectrum:theH linesatλ=2.0338, 2.1218 and2.2235µm,
2
the Hi Brγ at 2.1661µm and the [Caviii]coronal emission
line at 2.3211µm. The 12CO and 13CO absorption band-
headsused to obtain thestellar kinematics are identifiedin
thespectrum from position A.
3 LINE OF SIGHT VELOCITY
DISTRIBUTION
In order to obtain the line-of-sight velocity distribution
(LOSVD)wehaveusedthepenalizedPixel-Fitting(pPXF)
method of Cappellari & Emsellem (2004) to fit the stellar
Figure1.Topleft:WHTK-bandlargescaleimageofNGC4051 absorptions present in the K-band spectra. The algorithm
from Knapenetal. (2003). The image has been rotated to the findsthebestfittoagalaxyspectrumbyconvolvingatem-
same orientation of the NIFS observations. The box represents platestellarspectrumwiththecorrespondingLOSVD.This
the IFU field of view. Top right: 2.12µm continuum map from proceduregivesasoutputtheradialvelocity,velocitydisper-
IFU spectroscopy. Bottom: Spectra at the positons N, A and B sion and higher-order Gauss-Hermite moments. The pPXF
marked at the top-right panel with the emission lines and CO
method allows the use of several template stellar spectra
absorptionbandheads identified.
and to vary the weights of the contribution of the differ-
ent templates to obtain the best fit, minimizing the tem-
plate mismatch problem. However, the use of pPXF re-
centered at λ = 2.2499µm. Two set of observations, each
quires templates which match closely the galaxy spectrum.
with three individual exposures, were obtained at different
spatial positions; the first one centered at the position 0′.′4 Emsellem et al. (2004)presentaextensivediscussion about
from the nucleus along the PA= 74◦ and the second at the use of the pPXF method and the templates mismatch
a position offset 0′.′5 along PA= 1−06◦. The longest extent problem and Silge & Gebhardt (2003) present a discussion
abouttemplatemismatchforkinematicfittingusingtheCO
coveredbytheIFUobservationswas orientedalong thepo-
sitionanglePA=106◦,whichcorrespondstotheorientation absorption bandheadsin thenear-IR.
A high signal-to-noise ratio isrequired for reliable stel-
ofthelineofnodesderivedbymodellingofthestellarveloc-
lar kinematic measurements using the pPXF method. In
ityfieldwitharotatingdisk(Barbosa et al.2006).Wehave
previous works the data are usually binned to give S/N ra-
usedtheK−G5605gratingandthefilterHK−G0603,which
tio between 40 and 60 over the whole field-of-view. In our
resulted in an arc lamp line full width at half maximuum
(FWHM) of 3.2˚A. data, ratios smaller than these are observed only very close
to the borders of the field (S/N 35 measured blueward
The data reduction was accomplished using tasks con-
≈
tainedinthenifspackagewhichispartofgeminiirafpack- of the first CO band-head). Our typical S/N ratios are 80
age as well as generic iraf tasks. The reduction procedure with maximuum values reaching 120 close to the nucleus,
thusallowingreliablekinematicmeasurementswithoutspa-
included trimming of the images, flat-fielding, sky subtrac-
tial binning of thedata.
tion,wavelengthands-distortioncalibrations. Wehavealso
removed the telluric bands and flux calibrated the frames
by interpolating a black body function to the spectrum of
3.1 The stellar templates
thetelluricstandardstar.ThefinalIFUdatacubecontains
1160 spectra, each spectrum corresponding to an angular In this study we have selected as template spectra, those
coverage of 0′.′1 0′.′1, which translates into 4.5 4.5pc2 at of the spectroscopic library of late spectral type stars
× ×
the galaxy and covering the spectral region from 1.991µm observed with the Gemini Near Infrared Spectrograph
to 2.425µm. The total observed field of view 2′.′9 4′.′0 (ob- (GNIRS) IFU using the grating 111 l/mm in the K-band
×
tained by mosaicing the two set of observations) thus cor- (Winge, Riffel & Storchi-Bergmann 2007). This library is
respondstoaregion of projected dimensions130pc 180pc composed of spectra of 29 objects, which include dwarf, gi-
at thegalaxy. × ant and sub-giant stars with spectral types from F7iii to
In the top-left panel of Fig.1 we present a large scale M3iii,observed in thespectral range from 2.24 to2.42µm.
K-band image from Knapen et al. (2003) obtained at the Of these, 23 stars were also observed on a second set-
William Hershel Telescope (WHT), The central rectangle ting extending the wavelength coverage down to 2.15µm.
shows the IFU field of view. The large scale image was ro- The spectra of the templates have signal-to-noise ratios
4 Rogemar A. Riffel et al.
larger than 50 blueward of the CO2-0 first-overtone band-
head. Detailed discussion about this library is presented by
Winge, Riffel & Storchi-Bergmann(2007).Thespectralres-
olution of the stellar spectra is 3.3˚A – very close to the
spectralresolutionoftheNIFSdata,thusthelibrarycanbe
usedtofitthestellarkinematicswithoutanyresolutioncor-
rection. We have verified this in some tests using available
NIFS spectra of a few stars. We opted to use the GNIRS
librarybecausetherearetoofewavailabletemplatespectra
with NIFS.Wehavechosen 16 stars from thelibrary which
bettermatch thestellar features of NGC4051.
In order to investigate the influence of the stellar tem-
plates on the velocity dispersions obtained we have fitted
the stellar kinematics using individual stars as templates.
We observed that the large scale structures in σ maps are
similarforallstars,butthemeanσvaluesvarysignificantly
– higher EWs in the templates result in lower σ values for
the galaxy. This result evidences the importance of using a
large library of stellar templates in order to obtain reliable
velocity dispersion measurements.
3.2 The effect of the [Caviii] coronal line
Thenuclearspectrumof NGC4051 (Fig.1) shows thecoro-
nal emission line of [Caviii] at 2.3211µm. Davies et al. Figure 4. One dimensional stellar rotation curve for a pseudo-
slit oriented along the major axis of the galaxy. The points are
(2006) haveshown that thisline affects thekinematic mea-
the observed radial velocities and the full line is the modeled
surementsobtainedfromtheCObandheads.Inordertoin-
velocities.
vestigate the influenceof this line in our measurements, we
havechosentwospectralregionstofitthestellarkinematics:
thefirstone(2.258-2.372µm)includesthe[Caviii]lineand 40kms−1 withtheturnoveroccurringat 55pcfromthe
≈ ≈
the second (2.258-2.314µm and 2.346-2.372µm) excludes nucleus in good agreement with the velocity field obtained
thisline(togetherwiththe“contaminated”CObandhead). by Barbosa et al. (2006) from optical IFU data. The mean
We observed that the radial velocities and velocity disper- uncertaintiesinthevelocitiesare 10kms−1.Theturnover
≈
sionsderivedusingthefirstspectralrangearehigherinthe and amplitude of the rotation curvecan be more easily ob-
nuclear region, where the [Caviii] emission line is present, servedintheonedimensionalcutofthestellarvelocityfield
thanthoseobtainedbyusingthesecondspectralregion.We shown in Fig.4. The one dimensional rotation curve was
thus decided to exclude the contaminated 12CO3-1 band- obtained by the averare of the velocities within a pseudo-
head from the stellar kinematics fitting in theregion where slit with1arcsec widthorientedalongthemajor axisofthe
the[Caviii]emissionlineispresent(thecentral0′.′8radius). galaxy.
In regions away from the nucleus we used the first spectral Thestellarvelocitydispersionmapisshowninthetop-
range which includes the12CO3-1 bandhead. rightpanelofFig.3.Theσmappresentsvaluesrangingfrom
35 to 90kms−1, with mean uncertainties of 8kms−1.
≈ ≈ ≈
The bottom panels show the higher order Gauss-Hermite
3.3 The stellar kinematics moments h3 (left) and h4 (right). These moments measure
devationsofthelineprofilefrom aGaussian:theparameter
In Fig.2 we show the fits of the stellar templates to the
h measuresasymmetricdeviationsandtheh measuresym-
3 4
galaxy spectra with the program pPXF for four different
metricdeviations(van derMarel & Franx1993).Thevalues
positions: the nucleus; 1′.′2 SE and 1′.′2NW of the nucleus,
h and h vary from -0.15 to 0.15 with mean uncertainties
3 4
where we observe the turnoversof the rotation curveand a
of 0.03. The highest values of h are observed to SE of the
locationat1′.′5Eofthenucleus,almostattheborderofthe 3
nucleusandthelowestvaluestoNWofthenucleus.Theh
4
NIFSfield.Weobservethatthestellartemplatesfitverywell
valuesare nearly zero overmost of theIFU field.
thegalaxy spectraatmostpositions,includingregionsnear
the border of the IFU field, where the signal-to-noise ratio
is smaller. The results for the nucleus should nevertheless 3.4 Kinematic Modelling
be considered with caution, as the fitting may have been
Since the stellar velocity field is dominated by rotation, it
affected by dust emission and emission lines present in the
was fitted with a velocity model produced by a Plummer
galaxy spectrum butabsent in thestellar templates.
potentialinordertoobtainthesystemicvelocity,orientation
InFig.3wepresenttheresultingstellarkinematics.The
of the line of nodes, bulge mass and the position of the
black regions in this figure are masker regions where the
kinematical center. The Plummer potential is given by:
signal to noise ratio in the spectra was too low to provide
a reliable fit. In the top-left panel we show the stellar ve- GM
Φ= , (1)
locity field, which shows a velocity range from 40 to −√r2+a2
≈ −
Molecular gas inflow towards the Seyfert nucleus of NGC4051 5
Figure2.SamplefitsofthestellarkinematicsofthenuclearregionofNGC4051usingpPXF.Topleft:fitofthenuclearspectrum;top
right:fitofthespectrum at1′.′5Eofthenucleus;bottom left:fitofthespectrumatthelocationcorrespondingtotheblueturnoverof
the rotation curve at 1′.′2SE of the nucleus; bottom-right: fit of the spectrum at the positionof the redturnover of the rotation curve
at 1′.′2 NW of the nucleus. The observed spectra are shown in black, the fits in red, the residuals in green, while in blue we show the
spectralrangenotincludedinthefitofthecentralregion.
whereaisascalelength,ristheradialdistanceintheplane for the free parameters. As the inclination of the disk is
ofthegalaxy,M isthemassinsider andGistheNewton’s tightly coupled with M as V2 Msin(i), it cannot be left
gravitationalconstant.Definingthecoordinatesofthekine- asafreeparameter.Wehaverad∝optedthevalueofi=41.4◦,
maticalcenterofthesystemas(X ,Y ),theobservedradial anestimateobtainedfromcos(i)= b,wherebandaarethe
0 0 a
velocity at position (R,Ψ), where R is the projected radial semi-minor and semi-major axis of the large scale disk as
distance from the nucleus in the plane of the sky and Ψ is quoted in the NASA/IPAC Extragalactic Database (NED)
thecorrespondingpositionangle,isgivenby(Barbosa et al. for this galaxy.
2006): The parameters derived from the fit are: the systemic
velocitycorrected bytheobservatorymotionrelativetothe
Vr =Vs+r(R2R+2GAM2)3/2 cos2(Ψsi−n(Ψi)c0o)s+(Ψsi−nc2o(ΨsΨ20−()iΨ)0) 3/4(2) Mlkoicnae=lm7sat.7tai±ncda0la.6crde×not1ef0rr7eiMsstv⊙eVrsayn=cdl7o1sAe6t±=o13t19h.ek7mp±esa2−k.71o,pfΨct.0hTe=hc1oe2n0dt◦ien±ruivu1em◦d,
(cid:16) (cid:17)
emission,withX =4.9 1.4pcandY =3.1 1.2pc,whereX
0 0 0
where V is the systemic velocity, i is the inclination of the ± ±
s andY aremeasuredin relation tothelocation correspond-
0
disk (i = 0 for face on disk) and Ψ is the position an-
0 ing to thepeak of thecontinuum.
gle of the line of nodes. The relations between r and R,
In Fig.5 we present the derived rotation model in the
and between a and A are: r = αR and a = αA, where
left panel and the residuals of the stellar velocity field (ob-
α=qcos2(Ψ−Ψ0)+ sin2c(oΨs(−i)Ψ0).Theequationabovecon- servedminusmodelled)intherightpanel.Weconcludethat
tains six free parameters, including the kinematical center, the stellar velocity field is well described by the Plummer
which can bedetermined by fittingthe model to theobser- potential – the residuals are close to zero over most of the
vations. This was done using a Levenberg-Marquardt least- IFU field.The highest residuals ( 20kms−1) are observed
squares fitting algorithm, in which initial guesses are given within 0′.′3 from the nucleus, w∼here the stellar kinemat-
∼
6 Rogemar A. Riffel et al.
Figure 3.StellarkinematicmapsobtainedfromthepPXFfit.Top:radialvelocity(left)andvelocitydispersion(right)maps.Bottom:
h3 and h4 Gauss-Hermite moments. The mean uncertainties are10 kms−1 for radialvelocity, 8kms−1 for σ, and 0.03 for h3 and h4.
Thedashedlinesshowthepositionofthelineofnodes.
Molecular gas inflow towards the Seyfert nucleus of NGC4051 7
Figure5.RotatingdiskmodelforthestellarkinematicsofNGC4051(left)andresidualmap–observedvelocityfieldlessmodel(right).
Thedashedlinesmarksthepositionofthelineofnodes.
ics fitting may have been affected by emission by dust and the top 1′′ of the observed field was not high enough to
∼
emission lines. allowthemeasurementofthislineandthisiswhytheregion
appears black in Fig.6. The mean flux uncertainty for Brγ
is24%.TheBrγ fluxdistributionhasthehighestfluxvalues
at the nucleus and is approximately symmetric around the
4 EMISSION-LINE FLUX DISTRIBUTIONS
nucleus, not showing any elongation as observed in the H
2
WehavefittedGaussianstotheobservedemission-line pro- emitting gas.
files in order to obtain the integrated flux, radial velocity
The flux map of the [Caviii]λ2.3211µm coronal emis-
(from the central wavelength of the line) and velocity dis-
sion line is presented in the bottom middle panel, where
persion(fromthewidthoftheline).Thecorrespondingflux
themean flux uncertainties are 15%. The [Caviii]emission
maps are shown in Fig.6. The H λ =2.0338, 2.1218 and
2 peaks at the nucleus and is resolved, extending up to 0′.′8
2.2235µm flux distributions are presented in the top left,
fromthenucleus,whichcorrespondstoaprojecteddistance
middleandright panels, respectively,with mean uncertain-
of 36pcat thegalaxy.
tiesof 16%,for thefirst one,5% for thesecond oneand9%
forthethird.Blackregionsidentifylocationswheretheline ThelineratioH λ2.1218µm/Brγ (onlynarraocompo-
2
fitting failed due to low signal to noise ratios. The molecu- nent for Brγ) can be used as a diagnostic for the excita-
larhydrogenemissionisextendedovermostoftheobserved tion mechanism of the molecular hydrogen emission lines
field.Thehighestfluxvaluesareobservedatthenucleus,de- (eg. Riffelet al. 2006), with higher H /Brγ ratios being
2
finedasthelocation ofthepeakofthecontinuumemission. interpreted as a larger contribution from shocks or X-ray
The H distribution is extended towards the NE, between to the H excitation. We present this line ratio map in
2 2
the direction of the radio axis (adopted as the one which the bottom-right panel of Fig.6. The lowest values are
connectsthetworadiopeaks)andthelineofnodes.TheH H λ2.1218µm/Brγ 1 observed at the nucleus and to the
2 2
fluxdistributionshowsalsoagoodagreementwiththe[Oiii] N.Thehighestvalue≈sreachH λ2.1218µm/Brγ 8andare
2
≈
narrow band image from Schmitt & Kinney (1996), whose observed predominantly in two regions, one aproximately
contours are overplotted in green on the top right panel of 1′.′0 W of the nucleus and another at 0′.′8 E of the nucleus.
Fig.6. We note that these two regions are close to the tips of the
The flux map in the Brγ narrow component is shown compact 3.6cm radio structure (black contours in the top
in the bottom left panel of Fig.6. In the central region middle panel). Nevertheless, such a direct interpretation of
(r < 25pc) we have fitted two Gaussian components in or- thismapshouldbeconsideredwithcaution,asthefluxdis-
dertoseparatethenarrowandbroademissionlinecontribu- tributions and kinematics (see next section) are obviously
tions,whileinregionsfurtherawayfromthenucleusasingle different for the H and Brγ emission lines, implying that
2
Gaussianwasenoughforthefit.Thesignal-to-noiseratioin may originate in different regions of the galaxy.
8 Rogemar A. Riffel et al.
Figure 6. Top) From left to right: H2λ=2.0335, 2.1218 and 2.2235µm flux maps; Bottom) left: Brγ flux distributions; middle:
[Caviii]λ2.3211µm flux distributions and right: H2λ2.1218/Brγ line ratio. The thick black contours overlaid to the H2λ2.1218µm
intensitymaparefromtheVLAradio3.6cmcontinuum image,thinblacklinesareisointensitiescontours foreachpanel andthegreen
contoursarefroman[Oiii]narrow-bandimagefromHST.Thespatialscaleandorientationshownatthetop-leftpanelarethesamefor
all panels. The dashed white line represent the lineof nodes of the stellar velocity field and the full white linerepresent the PA which
connects thetworadioemissionpeaks.
5 GAS KINEMATICS other important kinematic components, evidenced by large
deviationsfrom simplerotation.Particularly conspicuousis
IntheleftpanelsofFig.7wepresenttheradialvelocityfield
a blueshifted region to the NE, showing velocity values up
obtainedfromthecentralwavelengthsoftheH2λ2.1218µm to 100kms−1,extendingby 1′′ fromthenucleus.The
andBrγ emission lines,withmeanuncertaintiesof4kms−1 ≈− ∼
Brγ velocity field shows no rotation and the total velocity
and9kms−1,respectively.WechosetheH2λ2.1218µmline range is only 50kms−1.
torepresenttheH velocityasitisstrongerandthuspresent ≈
2
smalleruncertaintiesinthemeasurementsthantheotherH2 In the right panels of Fig.7 we present the velocity
emission lines. The systemic velocity of the galaxy, derived dispersion (σ) maps obtained from measurements of the
from the stellar kinematics modelling, has been subtracted FWHM of the emission lines, such that σ = FWHM. The σ
2.355
from all theemission line velocity plots. values were corrected for the instrumental broadening and
TheH velocityfieldshowsa“rotationpattern”similar themeanuncertaintiesare6%forH and22%forBrγ.The
2 2
tothatofthestars,withtheNWsidereceedingandtheSE H σ map has values in the range 40–100kms−1. A par-
2
side approaching, although it is quite clear that there are tial ring of low velocity dispersion∼values (σ 45kms−1)
≈
Molecular gas inflow towards the Seyfert nucleus of NGC4051 9
is observed surrounding the nucleus, while higher values byahighlyconcentratedgravitationalpotential.Thisisalso
(80 6 σ 6 100kms−1) are observed over most of the re- supported by the small value obtained for the scale length
maining field. The Brγ σ map presents the highest values (A = 39.7pc) from the modelling of the velocity field. The
of up to 100kms−1 at the nucleus – but there are un- derived parameters are in approximate agreement with the
≈
certainties in these σ values due to the broad component ones derived by Barbosa et al. (2006) from a similar mod-
contribution – and at 0.8arcsec to the SW, close to the elling using optical IFU data for a field-of-view of 7′′ 5′′
western tip ofthe3.6cm≈radiostructure.Atpositions away (315 225pc2). The exception is our value of Ψ ≈= 1×20◦,
0
from the nucleustheBrγ σ is lower than 40kms−1. whic×his 13◦largerthantheirs.Ontheotherhand,ourΨ
0
is 13◦s∼mallerthantheonederivedbyDumas et al.(2007)
∼
from stellar kinematics obtained overthemuchlarger field-
5.1 Gas “Tomography” of-view of 50′′ 40′′ (2250 1800pc2). We atribute these
≈ × ×
differencestothedifferencesinfield-of-view,consideringalso
The high spectral resolution of the data has allowed us
thatourfield-of-view, whichcorrespondsto130 180pc2 at
to slice the emission-line profiles of H2λ2.1218µm and thegalaxy,isonlysamplingthestellarkinematic×sveryclose
[Caviii]λ2.3211µm into a sequence of velocity bins. With
to thenucleus.
this“tomography”technique,wecansamplethekinematics
For an orientation of the line of nodes of Ψ = 120◦,
along the whole emission-line profile, including the wings. 0
andassumingthatthespiral armsofNGC4051 aretrailing
In order to obtain the “tomography” images, we resampled
(orientationofthearmsisshowninFig.1),weconcludethat
the spectra into bins of 1˚A with the scombine iraf task
the NE is the far side and the SW is the near side of the
andthencombinedthetwosetsofobservationsintoasingle
galaxy.
data cube using the tasks scombine and imcombine. The
Thestellarvelocitydispersionvaluesshowpatcheswith
velocitysliceswereobtainedaftersubtractionofthecontin-
lower values ( 40-50kms−1) on top of a common back-
uumdeterminedasaveragesofthefluxesfrombothsidesof ground of valu≈es ranging from 60 to 70kms−1. A possi-
the emission line. The slices correspond to velocity bins of
ble interpretation is that these patches are colder regions
42kms−1 (3˚A) and are shown in Figs.8 and 9. In these
≈ with more recent star formation than theunderlyingbulge.
figures, each panel presents flux levels in logarithmic units
This is supported by optical spectra of the nuclear re-
for thevelocity slice shown.The zerovelocity is adopted as
gion which show clear signatures of intermediate age stars
the systemic velocity of the stars obtained from the stellar
(Cid Fernandeset al.2002).Averysmallσdropisobserved
kinematic modelling.
right at the nucleus, but we take this result with caution
The slices trace the gas from negative (blueshifts) to
because of the obvious contamination of the spectra of the
positive (redshifts) velocities relative to the systemic veloc-
nuclear region by emission from dust and broad emission
ity of the stars. For H , the highest blueshifts, which reach
190kms−1,areobs2ervedpredominantlytotheNEalong lines.
a≈c−urved elongated structure similar to the one observed WecanestimatethemassoftheSMBH(MBH)fromthe
in the radial velocity map (top-left panel of Fig.7). This bulge stellar velocity dispersion (σ∗) as log(MBH/M⊙) =
satrcuuc6rt0vuekrdem–asr−wm1h.o–AsetdmovmeolorinpcaihttoieelsosgtcyhloeisneeFmtoiigs.ssy8ioscnteamnfriobcme, tdh≈eesce−rmi1bi9es0dsioatnos α6a0n+dkmσβ0slo−=g1(2σa0∗s0/kσrme0p),sre−ws1ehn(eTtraerteiαmvea=ionfe8.te1ht3ea±lb.0u2.l00g06e2,)(β.toA=pdor4pi.g0thi2ntg±pσa0∗n.3≈e2l
i≈s −approximately symmetric and dominated by a region of of Fig.3), we obtain MBH = 1.1 ± 0.3 × 106M⊙. This
valueis ingood agreement with those obtainedbyprevious
1′′ radiuscenteredatthenucleus.Astheslicesreachpos-
≈ authors from reverberation mapping and scaling relations
itive velocities, the dominant structure is another curved
(Shemmeret al.2003;Kaspi et al.2000).Forthismass,the
arm extendingto 2′′ totheWofthenucleus.Thehighest
redshifts reach 1≈60kms−1. radius of influence of the SMBH is ≈ 1.3pc, thus not re-
For[Caviii≈],weshowinFig.9thehighestvelocitybins solved at the spatial resolution of ourdata.
andexcludeafewlowvelocitybinswherethekinematicsare
similar. The highest blueshifts reach -250kms−1, which 6.2 Gas Kinematics
≈
arehigherthanthoseobservedfortheH emittinggas,while
2
thehighestredshiftsreachvelocitiesof170kms−1similarto The simultaneous observation of the stellar and gaseous
kinematics allowed us to construct residual maps for the
theones observed for H .
2
gaseous kinematics relative to the stellar kinematics model
We opted not to show slices in the Brγ emission-line
described in section3.4. The residual map for the H emit-
profile due to the fact that its narrow component is indeed 2
tinggasispresentedinFig.10.Themostconspicousfeature
narrow and quite symmetric, and close to the nucleus it is
in this map is the elongated structure to the NE, which
hard to deblend it from the broad line profile, which intro-
shows blueshifts of up to 100kms−1, also seen in the
ducestoo much uncertaintyin thederived kinematics. ≈ −
correspondingvelocityslicesinFig.8.Theseblueshiftscould
beeitherduetoaninflow,ifthegasislocated intheplane,
as the NE is the far side of the galaxy; or to an outflow, if
6 DISCUSSION thegas is extendedtohigh latitudes (abovetheplane),e.g.
inaconicalstructureorientedtowardsus.Wediscussbelow
6.1 Stellar Kinematics
thesetwo possibilities.
Asobservedinthetop-leftpanelofFig.3and inFig.4the The outflow interpretation is supported by the follow-
turnoveroftherotationcurveoccursatonly 55pcfromthe ing facts: (1) that the elongated NE structure is approxi-
≈
nucleus, suggesting that the stellar motions are dominated mately oriented along the radio jet; (2) a similar structure
10 Rogemar A. Riffel et al.
Figure7.Left:VelocityfieldfortheH2λ2.1218µm(top)andBrγ(bottom)emissionlines.Right:Velocitydispersionmapsforthesame
emissionlines.ThecontoursandlinesareasdescribedinFig.6,aswellasthespatialscale.