Table Of ContentMon.Not.R.Astron.Soc.000,1–12(2012) Printed5January2017 (MNLATEXstylefilev2.2)
Gas inflows towards the nucleus of the Seyfert 2 galaxy
NGC 1667
Allan Schnorr-Mu¨ller,1 Thaisa Storchi-Bergmann,1 Fabricio Ferrari2, Neil M. Nagar,3
7
1 1Instituto de F´ısica, Universidade Federal do Rio Grande do Sul, 91501-970, Porto Alegre, RS, Brazil
0 2Instituto de Matem´atica, Estat´ısticae F´ısica, Universidade Federal do RioGrande (FURG), 96201-900, Rio Grande, RS, Brazil
2 3Astronomy Department, Universidad de Concepci´on, Casilla 160-C, Concepci´on, Chile
n
a
J
5January2017
3
]
A ABSTRACT
G
Weuseopticalspectrafromtheinner2×3kpc2 oftheSeyfert2galaxyNGC1667,
.
h obtained with the GMOS integral field spectrograph on the Gemini South telescope
p ataspatialresolutionof≈240pc, toassessthe feeding andfeedbackprocessesinthis
- nearbyAGN. We haveidentified two gaseouskinematicalcomponents in the emission
o lineprofiles:abroadercomponent(σ≈400kms−1)whichisobservedintheinner1–2′′
tr and a narrower component (σ≈200kms−1) which is present over the entire field-of-
s
a view.Weidentifythebroadercomponentasduetoanunresolvednuclearoutflow.The
[ narrowercomponentvelocityfieldshowsstrongisovelocitytwistsrelativetoarotation
pattern, implying the presence of strong non-circular motions. The subtraction of a
1
rotational model reveals that these twists are caused by outflowing gas in the inner
v ≈1′′, and by inflows associated with two spiral arms at larger radii. We calculate an
5
6 ionized gas mass outflow rate of M˙out≈0.16M⊙yr−1. We calculate the net gas mass
8 flow rate across a series of concentric rings, obtaining a maximum mass inflow rate
0 in ionized gas of ≈2.8M⊙year−1 at 800pc from the nucleus, which is two orders of
0 magnitude larger than the accretion rate necessary to power this AGN. However, as
. themassinflowratedecreasesatsmallerradii,mostofthegasprobablywillnotreach
1
0 theAGN,butaccumulateintheinnerfewhundredparsecs.Thiswillcreateareservoir
7 of gas that can trigger the formation of new stars.
1
Key words: Galaxies: individual (NGC1667) – Galaxies: active – Galaxies: Seyfert
:
v – Galaxies: nuclei – Galaxies: kinematics and dynamics
i
X
r
a
1 INTRODUCTION In order to assess if nuclear spirals are capable of
channeling gas inwards to feed the SMBH, our group
has been mapping gas flows in the inner kiloparsec of
It is widely accepted that the radiation emitted by an ac-
nearby AGN using optical and near-infrared integral
tivegalactic nucleus(AGN)isaresultofaccretion ontothe
field spectroscopic observations. So far, we have ob-
central supermassive black hole (hereafter SMBH). How-
served gas inflows along nuclear spirals in NGC1097
ever, understanding how mass is transferred from kilopar-
(Fathiet al. 2006), NGC6951 (Storchi-Bergmann et al.
sec scales down to nuclear scales of tens of parsecs and the
2007), NGC4051 (Riffel et al. 2008), M79 (Riffel et al.
mechanisms involved has been a long-standing problem in
2013), NGC2110, (Schnorr-Mu¨ller et al. 2014) and
thestudyofnuclearactivityingalaxies.Observationalstud-
NGC7213 (Schnorr-Mu¨ller et al. 2014). We also ob-
ies suggest that nuclear dust structures such as spirals, fil-
servedgasinflowsin thegalaxy M81(SchnorrMu¨ller et al.
aments and disks are tracers of the feeding channels to the
2011), where the inflow was mostly traced by dust lanes
AGN, as a strong correlation between nuclear activity and
and in NGC3081 (Schnorr-Mu¨ller et al. 2016), where a
the presence of nuclear dust structures have been found in
nuclear bar is feeding the AGN. Gas inflows have also
early type galaxies (Sim˜oes Lopes et al. 2007). Simulations
been observed by other groups. Near-infrared integral field
showed that, if a central SMBH is present, spiral shocks
spectroscopic observations revealed inflows along nuclear
can extend all the way to the SMBH vicinity and gener-
spiral arms in NGC1097 (Davies et al. 2009), NGC5643
ate gas inflow consistent with the observed accretion rates
(Davieset al.2014)andNGC7743(Davieset al.2014),and
(Maciejewski 2004a,b).
(cid:13)c 2012RAS
2 A. Schnorr-Mu¨ller et al.
gas inflow along a bar in NGC3227 (Davies et al. 2014). 3 RESULTS
Recent ALMA observations releaved streaming motions
In Fig.1 we present in the upper left panel the acquisition
along nuclear spirals in NGC1433 (Combes et al. 2013)
image of NGC1667 and in the upper right panel an im-
andNGC1566(Combes et al.2014).Gasinflowshavebeen
age of the inner 22′′×22′′ of the galaxy obtained with the
observed in CO in NGC2782 (Hunt et al. 2008), NGC3147
WideFieldPlanetaryCamera2(WFPC2)throughthefilter
(Casasola et al. 2008), NGC3627 (Casasola et al. 2011),
F606W aboard the Hubble Space Telescope (HST). In the
NGC4579 (Garc´ıa-Burillo et al. 2009) and NGC6574
middleleftpanelwepresentastructuremapoftheWFPC2
(Lindt-Krieget al. 2008).
HST image (see Sim˜oes Lopes et al. 2007). The rectangle
In the present work, we report results obtained from
in the middle left panel shows the field-of-view (hereafter
optical integral field spectroscopic observations of the nu-
FOV)coveredbytheIFUobservations.Nuclearspiralarms
clear region of NGC1667, aspiral galaxy with Hubbletype
traced by (dark) dust lanes are a prominent feature ob-
SAB(r)c which harbors a Seyfert 2 AGN enclosed by HII
served inside this rectangle. In the middle right panel we
regions (Gonzalez Delgado & Perez 1997). NGC1667 is lo-
present an image from our IFU observations obtained by
catedatadistanceof61.0Mpc(NED1),correspondingtoa
scale of 296pc/′′. integrating the continuum flux within a spectral window
from λ6470˚A to λ6580˚A. The isophotes in this image are
The present paper is organized as follows. In Section 2
extended roughly along the North-South direction, tracing
wedescribetheobservationsandreductions.InSection3we
thepositionofanuclearbar(Laine et al.2002).Inthelower
presenttheproceduresusedfortheanalysisofthedataand
panel we present three spectra of the galaxy corresponding
the subsequent results. In section 4 we discuss our results
to locations marked as A, B and N in the IFU image, se-
and present estimates of the mass inflow rate from two dis-
lected to be representative of the observed spectra. These
tinct methods and in Section 5 we present our conclusions.
spectra were extracted within apertures of 0′.′3×0′.′3.
The spectrum corresponding to the nucleus (marked
as N in Fig.1) is typical of the inner 1′′, showing strong
narrow [OI]λλ6300,6363˚A, [NII]λλ6548,6583˚A, Hα and
2 OBSERVATIONS AND REDUCTIONS [SII]λλ6717,6731˚Aemissionlines.Inthisspectrumthelines
havea broad baseand anarrow core, a shapedistinct from
TheobservationswereobtainedwiththeIntegralFieldUnit
a single Gaussian profile. The spectrum from location A is
of the Gemini Multi Object Spectrograph (GMOS-IFU) at
similar to the nuclear spectrum, and typical of the spectra
the Gemini South telescope on the night of January 26,
of regions between 1′′ to 3′′ from the nucleus. The spec-
2011 (Gemini project GS-2010B-Q-19). The observations
consisted of two adjacent IFU fields (covering 7×5arcsec2 trumfromlocationBistypicalfromthebordersoftheFOV
each) resulting in a total angular coverage of 7×10arcsec2 (≈3′.′5 from the nucleus), where the [NII] and Hα lines are
verynarrowandthe[NII]/HαlineratioistypicalofanHII
around the nucleus. Six exposures of 350 seconds were ob-
region.
tainedforeachfield,slightlyshiftedandditheredinorderto
correctfordetectordefectsaftercombination oftheframes.
Theseeingduringtheobservationwas0′.′8,asmeasuredfrom
the FWHM of a spatial profile of the calibration standard
3.1 Measurements
star. This corresponds to a spatial resolution at the galaxy
of 237pc.The photometric major axis is oriented along the Thegaseouscentroidvelocities,velocitydispersionsandthe
positionangle(PA)165◦(Skrutskieet al.2006,derivedfrom emission-line fluxes were obtained through the fit of Gaus-
aNear-IRimageofthegalaxyfromtheTwoMicronAll-Sky sians to the [NII], Hα, [OI] and [SII] emission lines. In or-
Survey(2MASS)). der to reduce the number of free parameters when fitting
The selected wavelength range was 5600-7000˚A, the[NII]andHαlines,weadopted thefollowing physically
in order to cover the Hα+[NII]λλ6548,6583˚A and motivated constraints:
[SII]λλ6716,6731˚A emission lines. The observations were
(i) Flux /Flux = 2.98, in accor-
obtained with the grating GMOS R400-G5325, set to cen- [NII]λ6583 [NII]λ6548
tralwavelengthofeitherλ6500˚Aorλ6550˚A,ataresolving dance with the ratio of their transition probabilities
power of R≈2000. (Osterbrock & Ferland 2006);
(ii) The Hα, [NII]λ6583 and [NII]λ6548 lines have the
The data reduction was performed using specific tasks
developed for GMOS data in the gemini.gmos package as same centroid velocity and FWHM;
well as generic tasks in iraf2. The reduction process com- Additionaly, we performed an adaptive spatial bin-
prised bias subtraction, flat-fielding, trimming, wavelength ning of our datacube using the Voronoi binning technique
calibration, skysubtraction, relativefluxcalibration, build- (Cappellari & Copin2003)sothatthe[NII]λ6583˚Alinehas
ingofthedatacubesatasamplingof0.1′′×0.1′′,andfinally asignal-to-noiseratioofatleast3ineachpixel.Thisallowed
thealignment and combination of 12 data cubes. ustomeasurethislineintheentireFOVwithoutlosingany
significant spatial information as binning was only neces-
sary near theborders of theFOV and the numberof pixels
in each bin is small. We did not perform an adaptive spa-
1 NASA/IPACextragalacticdatabase
2 IRAF isdistributedbytheNationalOpticalAstronomyObser- tial binning before measuring the [SII] lines as there is no
vatories,whichareoperatedbytheAssociationofUniversitiesfor signal in these lines outside the inner 2′′. Uncertainties in
Research in Astronomy, Inc., under cooperative agreement with thekinematicparameterswereestimatedusingMonteCarlo
theNational ScienceFoundation. simulations based on thebest-fitting absorption spectra.
(cid:13)c 2012RAS,MNRAS000,1–12
Gas inflows towards the nucleus of NGC1667 3
Figure 1. Top left: acquisition image of NGC1667. Top right: WFPC2 image. Middle left: structure map. The rectangle shows the
fieldoftheIFUobservation.Middleright:continuumimagefromtheIFUspectra(fluxinergcm2s−1 perpixel).Thedashedblackline
indicates the position angle of the photometric major axis of the galaxy (PA=165◦), from the 2MASS survey Skrutskieetal. 2006).
Bottom: spectracorrespondingtotheregionsmarkedasN,AandBintheIFUimage.Allimageshavethesameorientation.Thethin
lineintheHSTimageandstructuremapisanartifact.
A visual inspection of the fits to the spectra showed ing the narrow core of the profile, with velocity dispersion
thatin theinner≈2′′ the[NII],Hαand[SII]emission line between 100-180kms−1 and centroid velocities roughly the
profileshaveabroadbaseandanarrowcore,ashapewhich same as those resulting from the single Gaussian fit, and a
couldnotbereproducedbyasingleGaussianprofile.There- secondcomponenttracingthebroadbase,withvelocitydis-
fore, in addition to performing a single Gaussian fit to the persionbetween300-600kms−1andcentroidvelocitiesclose
emission line profiles in the entire datacube, we also per- tosystemic. Wewill refer tothese componentsas the“nar-
formed a two Gaussian fit to the line profiles in the inner rower component”and the“broader component”from here
2′′. The two component fit resulted in a component trac- on.Fits whereoneof thecomponentscontributed less than
(cid:13)c 2012RAS,MNRAS000,1–12
4 A. Schnorr-Mu¨ller et al.
Figure 2. Two Gaussian fits to the [NII] and Hα emission lines from four different regions, labeled as A, B, C and D. The asterisks
correspondtodatapoints,thesolidlinestothefit,crossestotheresidualsandthedashedlinestothenarrowerandbroadercomponents.
10% tothetotal fluxof each emission line wherediscarded. standard deviation matches that expected from the Pois-
In Fig.2 we show four examples of the two Gaussians fits sonian noise of that spaxel. One hundred iterations were
to the[NII] and Hα lines. We followed the same procedure performed and the estimated uncertainty in each parame-
whenfittingthe[SII]lines,butconstrainingthecentroidve- ter - line center, line width, and total flux in the line - was
locity of each component as equal to the value determined derived from the σ of the parameter distributions yielded
fromthetwoGaussianfittothe[NII]lines.Inthenextsec- bythe iterations. In Fig.3 we show theuncertainties in the
tionswepresentanddiscusstheresultsfromboththeSingle measurement of the [NII] emission lines. We note that the
Gaussian and Two Gaussians fits. uncertaintiesdecreaseneartheedgesoftheFOVbecausewe
In order to measure the stellar kinematics, we observestrongemissionfromHIIregionsinthenuclearring
employed the Penalized Pixel Fitting technique there(seespectrum B in Fig1).Uncertainties in thefluxes,
(pPXF, Cappellari & Emsellem 2004) and using the velocities,andvelocitydispersionsoftheHαand[SII]lines
Bruzual & Charlot (2003) stellar population models as aresimilartothoseofthe[NII]line.Uncertaintiesinthegas
templates to fit the stellar continuum from 5700˚A to density are of the order of 35%. Uncertainties in the stellar
6600˚A, where numeroues absorption features are present, velocityandvelocitydispersionareoftheorderof20kms−1
the strongest being the NaI doublet at λλ5890/5896˚A. and 25kms−1 respectively.
Using the Voronoi Binning technique (Cappellari & Copin
2003) we binned the datacube to achieve a signal-to-noise
ratio of at least 5 in the continuum near the NaI dou-
blet. The uncertainties in the kinematic parameters were
estimated using Monte Carlo simulations based on the
best-fittingabsorption spectra. 3.3 Stellar kinematics
Thestellarvelocityfield(leftpanelofFig.4)displaysarota-
tionpatterninwhichtheNsideofthegalaxyisapproaching
3.2 Uncertainties
and the S side is receding. Under the assumption that the
Totesttherobustnessofthefitsandestimatetheuncertain- spiral arms are trailing, it can be concluded that the near
ties in the quantities measured from each spectrum in our sideofthegalaxyistotheE,andthefarsideistotheW.A
datacube, we performed Monte Carlo simulations in which systemic velocity of 4650kms−1 (see section.4.1 for details
Gaussian noise was added to the observed spectrum. For onhowthisvaluewasdetermined)wassubtractedfrom the
each spaxel, the noise added in each Monte Carlo iteration centroidvelocitymaps.Thestellarvelocitydispersion(right
was randomly drawn from a Gaussian distribution whose panel of Fig.4) varies between 100kms−1 and 400kms−1.
(cid:13)c 2012RAS,MNRAS000,1–12
Gas inflows towards the nucleus of NGC1667 5
Figure 3.Uncertainties influx(%),centroidvelocity(kms−1)andvelocitydispersion(kms−1)forthe[NII]emissionline.
locity dispersions are on average 30kms−1 lower and the
“C” shaped region extends further to the northwest and
southeast. The broader component centroid velocity map
(Fig.7)showsvelocitiesclosetosystemic,exceptforasmall
region to the northeast of the nucleus, where velocities of
≈100kms−1 are observed. The velocity dispersion ranges
from 200kms−1 to 600kms−1, the highest velocity disper-
sionbeingobservednorthwestofthenucleusandthelowest
south of thenucleus.
Figure4.Stellarcentroidvelocity(kms−1)andvelocitydisper-
sion(kms−1). 3.5 Line fluxes and excitation of the emitting gas
Weshow the [NII] flux distribution for the single Gaussian
3.4 Gaseous Kinematics fit in the center right panel of Fig.5. The [NII] flux distri-
butionforthenarrowerandbroadercomponentsareshown
3.4.1 Single Gaussian fit
in the center right panels of Fig.6 and Fig.7 respectively.
InFig.5weshowcentroidvelocity(kms−1),velocitydisper- The Single Gaussian and narrower component flux and ex-
sion, flux distribution, and [NII]/Hα ratio maps obtained citation maps present essentially the same structures, thus
from our single Gaussian fit to the [NII] and Hα emis- weonlydiscussthenarrowercomponentmapsfromhereon.
sionlines.Asystemicvelocityof4650kms−1wassubtracted The narrower component [NII] flux distribution shows
from the centroid velocity maps. The gaseous velocity field an elongation along the major axis, a decrease of the emis-
isdominatedbynon-circularmotions;notespeciallythedis- sioninaringat≈2′′ fromthenucleusandanincreasenear
tortions in the isovelocity curves in the inner 2′′, especially thebordersoftheFOV.The[NII]/Hαmapofthenarrower
close to the minor axis. Asymmetries are also visible in the componentpresentsvaluesoftheorderof0.5neartheedges
centroidvelocityfield,forexampleasteepincreaseinveloc- oftheFOV.Thisisconsistentwithphotoionizationbyyoung
ity observed to the northwest, where the centroid velocity stars, which is expected as these regions correspond to the
rises from ≈150kms−1 at 2′.′5 to ≈200kms−1 at 4′′ while inner part of a star forming ring (see Fig.1). The ratio in-
no similar increase is observed to thesouth. creases to 1.5-2.5 in the inner 2′′, values consistent with
Thevelocitydispersion mapsshowlowvalues,between photoionization bya Seyfert nucleus.
60-100kms−1, near the edges of the FOV, with an in- Thebroadercomponent[NII]fluxdistributionshowsan
creaseto130-220kms−1 intheinner2′.′5.Thelargestvalues elongation towards the northeast. The broader component
(≈200kms−1)areobservedina“C”shapedregion,extend- [NII]/Hαratiovariesbetween2-3,higherthanthenarrower
ing from the southeast to thenorthwest of the nucleus. component, which suggests a larger contribution of excita-
tion byshocks in thegas in this component.
In Fig.8 we present the gas density map for the single
3.4.2 Two Gaussians fit
Gaussian fit, narrower component and broader component,
Thenarrowercomponent(Fig.6)velocityfieldisessentially obtained from the [SII]λλ6717/6731˚A line ratio assuming
the same as the one obtained from the single Gaussian fit, anelectronictemperatureof10000K(Osterbrock & Ferland
the differences between the two do not exceed 20kms−1. 2006). The narrower component gas density varies between
The narrower component centroid velocity map shows the 300-800cm−3. The broader component density reaches
same features as thesingle Gaussian map, although theve- ≈1000cm−3.
(cid:13)c 2012RAS,MNRAS000,1–12
6 A. Schnorr-Mu¨ller et al.
Figure5.SingleGaussianfit:gaseouscentroidvelocities(kms−1),velocitydispersion,logarithmofthe[NII]emissionlineflux(inunits
of10−17ergcm2s−1 perspaxel)and[NII]/Hαratio.
Figure 6. Narrower component: gaseous centroid velocities (kms−1), velocity dispersion, logarithm of the [NII] emissionline flux (in
unitsof10−17ergcm2s−1 perspaxel),and[NII]/Hαratio.Thismapwasconstructedcombiningthenarrowercomponentandthesingle
Gaussianmaps(whereonlyonecomponent wasfitted).
4 DISCUSSION The resulting parameters A, c, and p are
380±16kms−1, 7′.′3 ± 0.4 and 1.0 ± 0.1 respectively.
4.1 The stellar kinematics Theposition angleofthelineofnodeswasfixedas165◦±1,
In order to obtain the value of the systemic velocity and which is the position angle of the large scale photometric
therotationvelocityfieldofthediskofthegalaxy,wemod- major axis (Skrutskieet al. 2006). This was done because
eled the stellar velocity field assuming a spherical potential a large range of position angles resulted in good fits
with pure circular motions, with the observed radial veloc- due to the limited number of data points. The systemic
ity at a position (R,ψ) in the plane of the sky given by velocity corrected to the heliocentric reference frame is
(Bertola et al. 1991): 4570±15kms−1 (taking into account both errors in the
measurement and the fit). This value is in good agreement
V =Vs+ {R2[sin2(ψ−AψR0)co+s(cψos−2θψco0s)s2i(nψ(θ−)cψo0s)p]θ+c2cos2θ}p/2 wmietahsuprreemvieonutssd(eSteprrminignoabtioetnsalb.a2se0d05o;nRoapdtiocvailchan&dHRaIf2a1neclmli
where θ is the inclination of the disk (with θ=0 for a face- 1996). The model velocity field is shown in Fig.9.
on disk), ψ is the position angle of the line of nodes, V is
0 s
the systemic velocity, R is the radius, A is the amplitude
of the rotation curve (at large radii), c is a concentration 4.2 The gas kinematics
parameter regulating the compactness of the region with a
4.2.1 The Broader Component
strong velocity gradient and p regulates the inclination of
the flat portion of the velocity curve (at the largest radii). Toelucidatetheoriginofthebroadercomponentgas,wewill
Weassumedthekinematicalcentertobecospatialwiththe compare our results with an analysis of the ionised gas in
peak of the continuum emission. We adopted an inclina- theinner1′′ofNGC1667performedbyFischer et al.(2013)
tion of i=48◦ (Radovich & Rafanelli 1996). A Levenberg- basedonHSTlong-slitspectraobtainedwiththeSpaceTele-
Marquardtleast-squaresminimisationwasperformedtode- scope Imaging Spectrograph (STIS). Fischer et al. (2013)
termine thebest fittingparameters. measured the Hα line and they argued that two Gaussians
(cid:13)c 2012RAS,MNRAS000,1–12
Gas inflows towards the nucleus of NGC1667 7
Figure 7. Broader component: gaseous centroid velocities (kms−1), velocity dispersion, logarithm of the [NII] emission line flux (in
unitsof10−17ergcm2s−1 perspaxel),and[NII]/Hαratio.
Figure 8.Gasdensity(cm−3)obtainedfromthesinglecomponentfitandgasdensityofthenarrowerandbroadercomponents.
were needed to adequately fit the profile. They measured (≈600pc) is dominated bynon-circular motions. Toisolate
the velocities and velocity dispersions of each component thesenon-circularmotionsweneedtosubtracttherotation
and they argued that both are due to a biconical outflow component. The rotational model we obtained for the stel-
with a position angle of the bicone of 55◦, an opening an- larvelocityfieldinSec.4.1isnotanadequatedescriptionof
gle of 58◦ and an inclination in relation to the line of sight the gaseous velocity field, as the stellar component usually
of 18◦. One of the components is narrow, with velocity dis- rotatesmoreslowlythanthegas(van der Kruit & Freeman
persions of ≈100kms−1, while the other is broader, with 1986;Bottema et al.1987;Noordermeer et al.2008).Toac-
velocity dispersions of the order of 250kms−1. They ob- ccount for this difference, we fitted the model described in
served redshifted velocities SW of the nucleus and a mix of Sect.4.1 to thenarrower component velocity field, with the
redshifted and blueshifted velocities to the NE. These re- velocity curve amplitude A as the only free parameter in
sults are mostly consistent with our observations and, con- the fit. The other parameters were fixed as equal to those
sidering that the outflowing velocities of the Hα emitting adopted or derived from thefit of thestellar velocity field.
gas observed by Fischer et al. (2013) drop to zero at a ra-
dius of 0′.′5 and that our observations were obtained with The resulting amplitude of the velocity curve A is
0′.′8 seeing, we argue that the differences from our results 474±7kms−1. Themodel velocity field is shown in Fig.10.
Toaccountforthepossibilityofthegasdiskhavingadiffer-
are likely due to our limited spatial resolution. Thus, we
entorientationthanthestellardisk,weperformedasecond
concludethatthebroadercomponentobservedbyusiscon-
fit, this time with the position angle of the line of nodes
sistent with the model proposed by Fischer et al. (2013) in
as an additional free parameter. This fit resulted in posi-
whichitoriginatesinabiconicaloutflow,whichisunresolved
tion angle of the line of nodes of 150◦, a difference of only
in ourdata.
15◦ compared to the large scale major axis. Additionally,
thestructuresobservedintheresidualmaparepresentand
similarforbothmodels.Wethereforeconcludethatthereis
4.2.2 The Narrower Component
noindicationinourdatathatthenucleargasdiskhasadif-
AsevidencedbythecentroidvelocitymapsshowninFig.6, ferent kinematic position angle from that of the large scale
the narrower component velocity field within the inner 2′′ photometric major axis. Wethusadopt an angle of 165◦ as
(cid:13)c 2012RAS,MNRAS000,1–12
8 A. Schnorr-Mu¨ller et al.
Figure 9.Stellarvelocityfield(kms−1)andmodeledvelocityfield(kms−1).Thedashedwhitelineindicates thepositionangleofthe
majoraxisofthegalaxy.
Figure 10.Fromlefttoright:gaseous centroid velocity(kms−1),structuremap,modelledvelocity fieldandresidualbetween gaseous
centroid velocity and modelled velocity field(kms−1). The dashed white linemarks the position angle of the major axis of the galaxy
and the straightgreylines delimitabiconical outflow alonga positionangle of55◦ and withanopening angle of 58◦. Thethin linein
thetopleftcornerofthestructuremapisanartifact.
thekinematicmajor axis,as the15◦ differencebetween the ingintoaccount thatthesmall spatial scaleand thespatial
best fitting position angle and the large scale photometric distributionofthevelocityresidualsareconsistentwith the
majoraxiscanresultfromabiasinthefitduetothestrong posited biconical outflow of Fischer et al. (2013), we argue
non-circular motions in thegaseous velocity field. the velocity residuals in the inner 1′′ are due to a biconical
outflow.
Abarhasbeenpreviouslyobservedinthenuclearregion
At radii between 1′′and 2′′the residual map reveals a
of NGC1667 (Laine et al. 2002, it is visible in the contin-
two spiral pattern, a redshifted spiral on the near side of
uum image in Fig1 roughly along the N-S direction), and
theresidualvelocitiesobservedintheinner1′′,whichareof the galaxy and a blueshifted spiral on the far side. A com-
theorderof70kms−1,areconsistentwithwhatisexpected parisonwiththestructuremapinFig.10showsthatseveral
dust lanes are observed cospatially to the spiral pattern in
from non-circular motions in a barred potential. However,
the residual map. Assuming the gas is in the plane of the
as previously discussed in Sec.4.2.1, HST long-slit spectra
revealed a biconical outflow along the same orientation in disk, blueshifted residuals on the far side of the galaxy and
the inner 0′.′5 of NGC1667, thus these residuals velocities redshifted residuals on the near side imply we are observ-
ing radial inflow at the nuclear spirals. Residuals cospatial
can also beduetooutflowing gas. Tobetterdistinguish be-
tweenoutflowing,rotating,andothergaskinematics,wede- to the nuclear bar can also be due to non-circular orbits in
the bar. However as the bar seems weak (the dust lanes do
lineatetheposited biconical outflowofFischer et al.(2013)
(PA◦ 55, opening angle 58◦) in the residual map of Fig.10. not trace it), it is not clear that non-circular orbits in the
bar contribute significantly to the observed residuals. The
This makes it clearer that the residuals seen in the inner
1′′ - redshifted residuals observedSW on thefar side of the velocity dispersion increase observed along the spiral arms
(Fig.6) supports the presence of shocks in these regions,
galaxyandblueshiftedresidualsobservedNEofthenucleus
which cause a loss of angular momentum and lead to gas
onthenearside-arealsoconsistentwiththisbiconicalout-
flowing in.
flow. Considering Fischer et al. (2013) also observed both a
broader and a narrower component in their data and tak- Toillustrate thescenario weproposed for thenarrower
(cid:13)c 2012RAS,MNRAS000,1–12
Gas inflows towards the nucleus of NGC1667 9
component,webuiltatoymodelofarotatingdisk,twospi- 4.2.4 Estimating the mass inflow rate
ral arms and a biconical outflow using the software Shape
Inthissectionwedescribeourmethodtoestimatethemass
(Steffen et al. 2011). As the overall contribution of non-
flowratetowards thenucleus:theintegration oftheionised
circularmotionsalongthebartothegaseousvelocityfieldis
gas mass flow rate through concentric rings in the plane of
unclear,wedonotincludethebarinourmodel.Wederived
thegalaxy(aroundthewholeperimeter).Inordertodothis,
theline-of-sight velocity field from a combination of:
we consider that theobserved line-of-sight velocity,v is
LOS
• disk rotation; theresultoftheprojectionofthreevelocitycomponentsina
• rotation and radial inflow at the location of the spiral cylindrical coordinate system at thegalaxy. Thecylindrical
arms; coordinatesare:̟–theradialcoordinate;ϕ–theazimuthal
• a biconical outflow; angle; z – the coordinate perpendicular to the plane. The
galaxy inclination relative to the plane of the sky is i. We
We then convolved this velocity field with a Gaussian assume that gas vertical motions in the disc are negligible,
withaFWHMequaltothatoftheseeingdisk.Acomparison i.e.v =0,andthenweconsidertwopossibilities,asfollows.
z
betweentheobservedvelocityfieldandresidualsandthetoy Wefirstconsiderthattheazimuthalvelocitycomponent
model velocity field and radial velocity is shown in Fig.11. v can be approximated by the rotation model we fitted
ϕ
It is noteworthy that with such simple assumptions thetoy to the velocity field, and subtract it from v in order to
LOS
model provides a good reproduction of theobservations. isolate the radial velocity component. As the de-projection
oftheresultingradialcomponentintotheplaneofthegalaxy
isnotwelldeterminedalongthegalaxy lineofnodesdueto
divisionsbyzero,wehavemaskedoutfromthevelocityfield
aregionofextent0′.′3toeachsideofthelineofnodesinthe
4.2.3 Estimating the mass outflow rate
calculations.
Themassflowingthroughacrosssection ofradiusr can be The gas mass flow rate is given by:
obtained from:
M˙ =ρf v·A, (3)
M˙out = Nevπr2mpf (1) whereρistheionisedgasmassdensity,f isthefillingfactor
(determinedviaEq.2),vistheradialvelocityvectorandA
whereN istheelectrondensity,v isthevelocityofthe
e istheareavectorthroughwhichthegasflows.Sinceweare
inflowing gas , m is the mass of the proton, r is the cross
p interestedintheradial flow,thearea weareinterestedin is
section radius and f is the filling factor. The filling factor perpendiculartotheradialdirection,andthusv·A=v A,
̟
can be estimated from equation (2). Substituting equation
the product between the area crossed by the flow and the
(2)intoequation(4),wehaveforamassflowthroughacone
radial velocity component.
of height h and radius r:
The filling factor f is estimated from:
M˙out = j3m(pTv)LNHαh (2) LHα ∼ fNe2jHα(T)V (4)
Hα e
where j (T)=3.534×10−25ergcm−3s−1
Hα
where L is the Hα luminosity, (Osterbrock & Ferland 2006) and L is the Hα lu-
Hα Hα
J (T)=3.534×10−25ergcm3s−1 for a temperature minosity emitted by a volume V. We consider a volume
Hα
of10000K(Osterbrock & Ferland2006),andhistheheight V =Adx, of a thick ring sector with width dx and cross-
of thebicone (equivalent to the distance from thenucleus). section area A, the same as above. Replacing f obtained
Asthebroadercomponentisunresolvedinourdata,wewill via Eq.2 into Eq.1, we finally have:
base our estimates on the narrower component measure-
m v L
ments. We will adopt the density, outflowing velocity and M˙(̟)= p ̟ Hα . (5)
j (T)N dx
L values obtained from the redshifted region extending Hα e
Hα
fromthenucleusto1′′(296pc)totheSW.Theaverageelec- Integrating Eq. 5 all around the ring of radius ̟, and
trondensityinthisregionis670cm−3(fromthe[SII]ratio). takingintoaccounttheradialandazimuthaldependenceof
ThetotalHαfluxinthisregion is3.6×10−14ergcm−2s−1. v , L and N , we obtain the gas mass flow rate M˙ as a
̟ Hα e
Adopting a distance of 61.0Mpc, we obtain a total lumi- function of the radius ̟. It is worth noting that the mass
nosity for the outflowing gas of L =1.6×1040ergs−1. flowrateascalculated abovedoesnotdependonAanddx,
Hα
Adopting a distance from the nucleus (height of the asthesequantitiescanceloutinEq. 5,whenweuseL as
Hα
cone) of 1′′, we obtain an average projected outflowing thegas luminosity of a volume V =Adx.
velocity of ≈41kms−1. Correcting this velocity by an Wehaveevaluated M˙(̟) throughconcentric Gaussian
inclination of 18◦ (Fischer et al. 2013),weobtain avelocity rings (rings whose radial profile is a normalised Gaussian)
of 132kms−1. We thus obtain an outflow mass rate of fromthegalaxycentreuptothemaximumradiusof1.8kpc
M˙out≈0.16M⊙yr−1 for the narrower component. This (inthegalaxyplane).TheresultisshowninFig.12.Theneg-
value is in good agreement with those from previous ative values mean that there is inflow from ≈1.2kpc down
studies which estimated mass outflow rate in nearby AGNs to400pcfromthenucleus.Withthemassinflowratebeing
(z< 0.1) ranging from ≈0.01M⊙yr−1 to ≈10M⊙yr−1 the largest between 1kpc and 800pc and then decreasing
(Riffel & Storchi-Bergmann 2011a,b; Mu¨ller-S´anchez et al. until it reaches ≈0 at 400 pc from the nucleus. It is worth
2011; Schnorr-Mu¨ller et al. 2014; Sch¨onell et al. 2014; noting, however, that in the inner 400pc (1′.′3) the residu-
Schnorr-Mu¨ller et al. 2016; Karouzos et al. 2016). als are mainly close to the major axis, as can be seen in
(cid:13)c 2012RAS,MNRAS000,1–12
10 A. Schnorr-Mu¨ller et al.
Figure11.Fromlefttoright:Residualvelocitiesaftersubtractionofarotationalmodelfromtheobservedgaseousvelocityfield,radial
componentofthetoymodelvelocityfield(kms−1),observedgaseouscentroidvelocityfield(kms−1)andtoymodelvelocityfield(radial
plusrotational components). Thedashedwhitelinemarksthepositionangleofthemajoraxis.
Figure12.MassflowrateM˙ obtainedfromtheresidualvelocityfield(aftersubtractingtherotationmodel–solidcurve)andfromthe
observedvelocityfield(withoutsubtractingtherotationmodel–dashedcurve).
theresidualmapinFig.9,andaswehavemaskedout from minosity of the Seyfert nucleus of NGC1667, calculated as
the velocity field a region of extent 0′.′3 to each side of the follows:
line of nodes, so a significant part of the residuals are not L
consideredinthemassflowrateestimate.Additionally,itis m˙ = bol
c2η
possible that the gas inflow continues to smaller radii, but
in the case of much stronger emission from the outflowing where η is the efficiency of conversion of the rest mass en-
gas compared to inflowing gas, the inflowing gas would be ergy of the accreted material into radiation. For geomet-
undetectableby our observations. rically thin and optically thick accretion disk , the case
of Seyfert galaxies, η≈0.1 (Frank et al. 2002). The nu-
A second approach is to calculate theinflow rate with-
clear luminosity can be estimated from the X-ray lumi-
outsubtractingtherotationalcomponent.Onemightindeed nosity of L =2.0×1042ergs−1 (Panessa et al. 2006), us-
X
question if subtracting the rotational component is neces-
ing the approximation that the bolometric luminosity is
sary, as its contribution to vLOS may cancel out in the in- L ≈10L .Weusethesevaluestoderiveanaccretion rate
B X
tegration of the mass flow rate around the ring if the gas of m˙ =3.5×10−3M⊙yr−1.
density and filling factor have cylindrical symmetry, as by
Comparing the accretion rate m˙ with the mass inflow
definition,therotationalcomponenthascylindricalsymme-
rate of ionised gas obtained in our analysis above, we find
try.
that in the inner 600pc the inflow rate is 2 orders of mag-
We now compare the estimated inflow rate of ionised nitude larger than the accretion rate, reaching a maximum
gas to the mass accretion rate necessary to produce the lu- mass flow rate of ≈2.8M⊙year−1 at 800pc from the nu-
(cid:13)c 2012RAS,MNRAS000,1–12
